+ All Categories
Home > Documents > Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing...

Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing...

Date post: 14-Oct-2020
Category:
Upload: others
View: 1 times
Download: 0 times
Share this document with a friend
199
Child Pedestrian-Motor Vehicle Collisions and Walking to School in the City of Toronto: The Role of the Built Environment By Linda May Rothman A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy The Institute of Medical Science University of Toronto © Copyright by Linda Rothman, 2014
Transcript
Page 1: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

Child Pedestrian-Motor Vehicle Collisions and Walking to

School in the City of Toronto:

The Role of the Built Environment

By

Linda May Rothman

A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy

The Institute of Medical Science

University of Toronto

© Copyright by Linda Rothman, 2014

Page 2: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

ii

Child Pedestrian-Motor Vehicle Collisions and

Walking to School in the City of Toronto:

The Role of the Built Environment

Linda May Rothman

Doctor of Philosophy

The Institute of Medical Science

University of Toronto

2014

Abstract

Introduction: Child pedestrian-motor vehicle collisions are a major population health issue

worldwide. Although there are numerous benefits of active transportation, walking to school

could potentially increase collision risk. The built environment has been associated with self-

reported walking to school and with child pedestrian motor-vehicle collisions. It is important

to determine if there are built environment features related to more walking but which also

create safe walking environments. The thesis objective was to examine the relationships

between observed walking to school, child pedestrian-motor vehicle collisions and the role of

the built environment in Toronto, Canada.

Methods: Literature related to children walking for transportation, pedestrian-motor vehicle

collisions and the built environment was systematically reviewed. Observational counts of

school travel mode were conducted at 118 elementary schools in 2011 and mapped onto

school attendance boundaries together with police-reported child pedestrian-motor vehicle

collisions (2002-2011) and built and social environment data. The relationship between

walking proportions and collision rates was examined controlling for the environment.

Results: There was a mean collision rate of 7.1/10,000/year within school boundaries. The

mean proportion of observed walking was 67%. Several built environment features were

Page 3: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

iii

related to more walking; however, school crossing guards reduced the influence of other

features on walking. Walking to school was unrelated to collision rates once built

environment features were controlled for. Higher multi-family dwelling density was related to

lower collision rates; whereas higher one way street and traffic calming densities, lower traffic

light density, school crossing guards and lower school socioeconomic status were related to

higher collision rates. Significant features were generally related to road crossing.

Conclusions: This is the first large observational study examining walking to school, collision

risk and the environment. Results suggest that safety is concerned with built environment

features primarily related to road crossing, and not the numbers walking. The associations

between school crossing guards and traffic calming with higher collision rates were

unexpected. Mechanisms for mitigating road crossings for children are not well understood

and controlled research designs are needed. Future policy to increase children’s active

transportation should be developed from strong evidence that addresses child pedestrian

safety.

Page 4: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

iv

Acknowledgments

I owe my sincerest gratitude to the following people without whom this research would not

have been possible:

Dr. Andrew Howard, for his continued guidance, support, and mentorship over the years. He

has been a true mentor; encouraging me to pursue this doctoral research, and providing me

with both the professional and personal support to find my own path as a researcher.

Dr. Teresa To, for her mentorship and valued methodological guidance. Her calm and caring

support was invaluable throughout the process.

Dr. Colin Macarthur, for his keen interest, and attention to detail which was instrumental in

ensuring the scientific quality of this work. His clarity of vision and communication was

instrumental in helping me to clarify my directions and my goals.

Dr. Ron Buliung, for his invaluable and unique contributions. Most notably, he broadened my

perspective by introducing me to a whole new way of thinking as related to geography and

urban planning, and helped me integrate this knowledge into the area of child injury

prevention.

My peers and friends at the Hospital for Sick Children and at IMS, particularly Morgan Slater,

Maricar Aruta, Sarah Richmond and Joanne Goldman, for their encouragement and support

throughout the process.

My parents, in-laws and my sister for their encouragement throughout. And my brother,

Lorne, for his patient delivery of statistical support whenever needed.

I owe my deepest gratitude to my husband, Gary, and my children, Zev, Kobi and Gil. After

many years of deliberation regarding the pursuit of a doctoral degree, their encouragement,

patience, love and never-ending energy are responsible for making the completion of this

thesis possible. This thesis is dedicated to my family.

Page 5: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

v

I would also like to acknowledge the following scholarship/fellowship programs: Canadian

Institutes of Health Research (CIHR) Doctoral Research Award Program and Frederick

Banting and Charles Best Canada Graduate Scholarship, The Hospital for Sick Children,

Research Training Program (Restracomp) and the Ontario Neurotrauma Foundation, Summer

Internship Program in Injury Prevention.

Page 6: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

vi

Contributions

Linda Rothman (author) solely prepared this thesis. She conducted the systematic literature

review with the assistance of the research librarian and a second reviewer. She developed the

field survey, hired the student observers and organized and supervised the data collection and

data entry of the primary data. She procured the secondary datasets and processed and

conducted spatial and traditional statistical analyses. She was responsible for the writing of

the thesis and all resulting publications. .

The following contributions by other individuals are acknowledged:

Dr. Andrew Howard (Primary Supervisor) – mentorship; guidance and assistance in planning,

execution, and statistical analysis as well as manuscript/thesis preparation.

Dr. Teresa To (Co-supervisor) – mentorship; guidance and assistance in planning, execution,

and statistical analysis as well as manuscript/thesis preparation.

Dr. Colin Macarthur (Thesis Committee Member) –guidance and assistance in planning,

execution, and statistical analysis as well as manuscript/thesis preparation.

Dr. Ron Buliung –guidance and assistance in planning, execution, and spatial analysis as well

as manuscript/thesis preparation.

Elizabeth Uleryk (Research Librarian, Hospital for Sick Children) -assistance in developing

the literature search strategy (Chapter 3).

Andi Camden -assistance in reviewing the articles (Chapter 3).

Page 7: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

vii

Table of Contents

Acknowledgments ........................................................................................................................ iv Contributions ................................................................................................................................ vi List of Tables .............................................................................................................................. xii List of Figures ............................................................................................................................ xiv List of Appendices ...................................................................................................................... xv

List of Abbreviations ................................................................................................................. xvi 1 Introduction .............................................................................................................................. 1

1.1 Rationale ........................................................................................................................... 1 1.2 Overall Objective .............................................................................................................. 2

1.3 Specific Objectives ........................................................................................................... 2 1.4 Thesis Organization .......................................................................................................... 3

2 Background .............................................................................................................................. 4 2.1 Burden of Motor Vehicle Collisions ................................................................................. 5

2.1.1 Pedestrian-Motor Vehicle Collisions .................................................................... 5

2.1.2 Children ................................................................................................................. 6 2.1.3 Injury and Disability ............................................................................................. 7

2.2 Health Outcomes of Walking for Transportation ............................................................. 8 2.2.1 Child Pedestrian-Motor Vehicle Collisions .......................................................... 8 2.2.2 Prevention of Chronic Conditions ...................................................................... 10

2.3 Measurement ................................................................................................................... 12 2.3.1 Child Pedestrian-Motor Vehicle Collisions ........................................................ 12

2.3.2 Walking to School ............................................................................................... 13

2.4 Conceptual Frameworks ................................................................................................. 14

2.4.1 Child Pedestrian-Motor Vehicle Collisions ........................................................ 15 2.4.2 Walking to School ............................................................................................... 16

2.5 Correlates and Interventions ........................................................................................... 20 2.5.1 Child Pedestrian-Motor Vehicle Collisions ........................................................ 20

2.5.1.1 Correlates .............................................................................................. 20

2.5.1.2 Interventions to Decrease Child Pedestrian-Motor

Vehicle Collisions ................................................................................ 21

2.5.2 Walking to School ............................................................................................... 24 2.5.2.1 Correlates .............................................................................................. 24 2.5.2.2 Interventions to Increase Walking to School ........................................ 26

2.6 Geographic Information Systems (GIS) ......................................................................... 27

2.6.1 GIS and Child Pedestrian-Motor Vehicle Collisions .......................................... 28 2.6.2 GIS and Walking to School ................................................................................ 30

2.7 The Setting - The City of Toronto .................................................................................. 32

2.8 Policy ............................................................................................................................. 33 2.8.1 Child Injury Prevention ....................................................................................... 34

2.8.1.1 National ................................................................................................ 34 2.8.1.2 Provincial .............................................................................................. 35

2.8.2 Walking to School ............................................................................................... 35 2.8.2.1 National ................................................................................................ 35

Page 8: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

viii

2.8.2.2 Provincial .............................................................................................. 36

2.8.2.3 Municipal .............................................................................................. 37 2.9 Gaps in Knowledge Regarding Child Pedestrian-Motor Vehicle Collisions, Walking

to School and the Built Environment .............................................................................. 38

3 Walkable but Unsafe? A Systematic Review of Built Environment Correlates of Walking

and Child Pedestrian Injury .................................................................................................... 39 3.1 Preface ............................................................................................................................. 39 3.2 Abstract ........................................................................................................................... 40

3.2.1 Objectives ........................................................................................................... 40

3.2.2 Methods ............................................................................................................... 40 3.2.3 Results ................................................................................................................. 40 3.2.4 Conclusions ......................................................................................................... 40

3.3 Introduction ..................................................................................................................... 41

3.4 Methods ........................................................................................................................... 43 3.4.1 Eligibility ............................................................................................................ 44

3.4.2 Data Extraction ................................................................................................... 45 3.4.3 Quality Assessment ............................................................................................. 46

3.4.4 Analysis ............................................................................................................... 46 3.5 Results ............................................................................................................................. 47

3.5.1 Walking ............................................................................................................... 47

3.5.2 Child Pedestrian Injury ....................................................................................... 48 3.5.3 Quality Assessment ............................................................................................. 49

3.5.3.1 Walking ................................................................................................ 49 3.5.3.2 Child Pedestrian Injury ......................................................................... 49

3.5.4 Safety and Walking ............................................................................................. 52

3.5.4.1 Less Injury (Safer) and Walking Correlates ......................................... 52

3.5.4.2 More Injury (Less Safe) and Walking Correlates ................................. 52 3.5.4.3 Inconsistent/Untested Correlates of Injury and Walking ..................... 53

3.6 Discussion ....................................................................................................................... 54

3.7 Conclusions ..................................................................................................................... 57 3.8 Supplementary/Supporting Analysis .............................................................................. 58

3.8.1 Walking to School ............................................................................................... 58 3.9 Supplementary Tables ..................................................................................................... 59

4 Influence of Social and Built Environment Features on Children’s Walking to School:

An Observational Study. ........................................................................................................ 60 4.1 Preface ............................................................................................................................. 60 4.2 Abstract ........................................................................................................................... 61

4.2.1 Objectives ........................................................................................................... 61

4.2.2 Methods ............................................................................................................... 61 4.2.3 Results ................................................................................................................. 61

4.2.4 Conclusions ......................................................................................................... 61 4.3 Introduction ..................................................................................................................... 62 4.4 Methods ........................................................................................................................... 63

4.4.1 Study Design, Setting and Population ................................................................ 63 4.4.2 Outcome Variable ............................................................................................... 63 4.4.3 Independent Variables ........................................................................................ 63

4.4.3.1 Built Environment ................................................................................ 64

Page 9: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

ix

4.4.3.1.1 Density ....................................................................................... 64

4.4.3.1.2 Diversity..................................................................................... 64 4.4.3.1.3 Design ........................................................................................ 65

4.4.3.2 Social Environment .............................................................................. 65

4.4.4 Statistical Analysis .............................................................................................. 65 4.5 Results ............................................................................................................................. 66 4.6 Discussion ....................................................................................................................... 72

4.6.1 Limitations .......................................................................................................... 73 4.6.2 Strengths ............................................................................................................. 74

4.7 Conclusion ...................................................................................................................... 75 4.8 Supplementary/Supporting Analyses .............................................................................. 76

4.8.1 Principal Component Analysis ........................................................................... 76 4.8.2 Proportion Observed Walking ............................................................................ 77

4.8.3 Network Analysis ................................................................................................ 77 4.8.4 Predicted Values ................................................................................................. 77

4.8.5 Sensitivity Analysis ............................................................................................ 78 4.8.5.1 Trimming of Variables ......................................................................... 78

4.8.5.2 Residual Diagnostics ............................................................................ 78 4.8.5.3 Alternative Modeling Strategies ........................................................... 79

4.9 Supplementary Tables ..................................................................................................... 80

4.10 Supplementary Figures .................................................................................................. 84 5 Motor Vehicle-Pedestrian Collisions and Walking to School: The Role of the Built

Environment ........................................................................................................................... 86 5.1 Preface ............................................................................................................................. 86 5.2 Abstract ........................................................................................................................... 87

5.2.1 Objectives ........................................................................................................... 87

5.2.2 Methods ............................................................................................................... 87 5.2.3 Results ................................................................................................................. 87 5.2.4 Conclusions ......................................................................................................... 87

5.3 Introduction ..................................................................................................................... 88 5.4 Methods ........................................................................................................................... 88

5.4.1 Study Design, Setting and Population ................................................................ 88 5.4.2 Outcome .............................................................................................................. 89

5.4.3 Exposure ............................................................................................................. 89 5.4.4 Potential Covariates ............................................................................................ 89 5.4.5 Data Sources ....................................................................................................... 90

5.4.5.1 Canadian Census .................................................................................. 90 5.4.5.2 Municipal Property Assessment Corporation (MPAC) ........................ 90

5.4.5.3 Site Audits ............................................................................................ 90 5.4.5.4 City of Toronto ..................................................................................... 92

5.4.5.5 Toronto District School Board ............................................................. 92 5.4.6 Statistical Analysis .............................................................................................. 92

5.5 Results ............................................................................................................................. 93 5.6 Discussion ....................................................................................................................... 94

5.6.1 Comparisons of Findings to Previous Studies .................................................... 97 5.6.2 Confounders ........................................................................................................ 97 5.6.3 Effect Modifiers .................................................................................................. 98

Page 10: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

x

5.6.4 Unexpected Results ............................................................................................. 99

5.6.5 Strengths and Limitations ................................................................................. 100 5.6.6 Future Research ................................................................................................ 100

5.7 Conclusions and Policy Implications ............................................................................ 101

5.8 Supplementary/Supporting Analyses ............................................................................ 102 5.8.1 Collision Rates .................................................................................................. 102 5.8.2 Pedestrian Action During Collision .................................................................. 102 5.8.3 Predicted Values ............................................................................................... 102 5.8.4 Sensitivity Analysis .......................................................................................... 103

5.8.4.1 Residual Diagnostics .......................................................................... 103 5.8.4.2 School Travel Time Collisions ........................................................... 103 5.8.4.3 Alternative Collision Data Years ........................................................ 104 5.8.4.4 Alternative Outcome........................................................................... 104

5.9 Supplementary Tables ................................................................................................... 105 5.10 Supplementary Figures................................................................................................. 109

5.11 Detailed Methods ......................................................................................................... 111 5.11.1 Data Sources ..................................................................................................... 111

5.11.1.1 Observational Study ........................................................................... 111 5.11.1.2 Site Survey .......................................................................................... 112 5.11.1.3 Canadian Census ................................................................................ 113

5.11.1.4 City of Toronto ................................................................................... 113 5.11.1.5 Toronto Police Services ...................................................................... 114

5.11.1.6 Toronto District School Board (TDSB)/Toronto Catholic District

School Board (TCDSB) ...................................................................... 114 5.11.1.7 Municipal Property Assessment Corporation (MPAC) ...................... 114

5.11.1.8 Teranet (via licensing from the University of Toronto) ................... 114

5.11.2 Mapping ............................................................................................................ 115 5.11.2.1 Spatial Analysis .................................................................................. 115

5.11.2.1.1 Area Interpolation- Polygon in Polygon Areal Weighting .... 115

5.11.2.1.2 Buffer Analysis ...................................................................... 116 5.11.2.1.3 Network Analysis .................................................................. 116

5.11.3 Statistical Analysis ............................................................................................ 116 5.11.3.1 Negative Binomial Regression ........................................................... 116

5.11.3.2 Forward Stepwise Manual Regression ............................................... 117 5.11.3.3 Confounding ....................................................................................... 117 5.11.3.4 Effect Modification (Interactions) ...................................................... 118

6 General Discussion ............................................................................................................... 119 6.1 Summary ....................................................................................................................... 119

6.2 Unifying Discussion ...................................................................................................... 122 6.2.1 Density: ............................................................................................................. 124

6.2.2 Diversity ............................................................................................................ 124 6.2.3 Design ............................................................................................................... 126

6.2.3.1 Distance to School .............................................................................. 126 6.2.3.2 Design Features with No Significant Associations with Child

Pedestrian-Motor Vehicle Collisions.................................................. 128 6.2.3.3 Design Features with Significant Positive Associations with Child

Pedestrian-Motor Vehicle Collisions.................................................. 129

Page 11: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

xi

6.3 Strengths and Limitations ............................................................................................. 130

6.3.1 Strengths ........................................................................................................... 130 6.3.2 Limitations ........................................................................................................ 132

6.4 Policy Implications ....................................................................................................... 136

6.4.1 Integration of Walking to School and Child Pedestrian-Motor Vehicle

Policies .............................................................................................................. 136 6.4.2 Identification of Evidence-Based Targets ......................................................... 137

6.4.2.1 Walking to School .............................................................................. 137 6.4.2.2 Child Pedestrian-Motor Vehicle Collision Targets ............................ 137

6.4.3 Appropriate Outcome Measurement ................................................................. 138 6.4.4 Evidence-Based Built Environment Strategies ................................................. 139

6.4.4.1 Distance and School Boundaries ........................................................ 140 6.4.4.2 Short-term Versus Long-term Built Environment Strategies ............. 141

6.5 Knowledge Translation Activities ................................................................................ 142 6.6 Future Research ............................................................................................................ 145

6.6.1 Further Analysis from the Present Study .......................................................... 145 6.6.1.1 Specific Built Environment Design Features and Collisions ............. 145

6.6.1.2 Parent- Perceived Traffic Danger and the Built Environment ........... 146 6.6.1.3 Observed versus Self-Reported Walking ........................................... 146

6.6.2 Methodological Approaches for Future Studies ............................................... 147

6.6.2.1 Randomized Controlled Trials (RCT) ................................................ 147 6.6.2.2 Longitudinal Cohort ........................................................................... 148

6.6.2.3 Case Control ....................................................................................... 148 6.6.2.4 Quasi Experimental, Pre-Post Design ................................................ 148 6.6.2.5 Cross Sectional Studies in Other Settings .......................................... 149

6.7 Conclusions ................................................................................................................... 149

References ................................................................................................................................. 151 Appendices ................................................................................................................................ 172

Page 12: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

xii

List of Tables

Table 2-1: Haddon's Matrix. Pedestrian-motor vehicle collision example ............................. 15

Table 2-2: Built environment variables most associated with travel demand. ........................ 19

Table 3-1: Quality assessment using EAI: Number of studies (%). ....................................... 50

Table 3S-1: Correlates of walking to school and child pedestrian injury ............................... 59

Table 4-1: Descriptive statistics of candidate variables for multivariate modeling. ............... 68

Table 4-2: Unadjusted Incident Rate Ratios (95% CI) for candidate variables (p<.2) for

multivariate modeling ............................................................................................................... 70

Table 4-3: Correlates of walking to school in adjusted analysis (IRR = incident rate ratios

(IRR, 95% CI = confidence interval). ....................................................................................... 71

Table 4-4: Correlates of walking to school in adjusted analysis stratified by presence of

school crossing guard (IRR = incident rate ratios, 95% CI = confidence interval). ................. 71

Table 4S-1: Data sources and variable type .............................................................................. 80

Table 4S-2: Built environment factor loadings from principal component analysis. ............... 82

Table 4S-3: Results of negative binomial regression excluding 3 outlier schools.. ................ 83

Table 5-1: Variables according to conceptual component, level of measurement and data

source. ....................................................................................................................................... 91

Table 5-2: Descriptive statistics and significant unadjusted incident rate ratios

(p <.20, IRR = incident rate ratio, 95% CI= 95% confidence interval). ................................... 95

Table 5-3: Correlates of child pedestrian collisions in adjusted analyses

(IRR = incident rate ratio, 95% CI= 95% confidence interval). ............................................... 96

Table 5-4: Incidence rate ratios of collisions stratified by traffic light density tertiles ............ 96

Table 5S-1: Correlates of child pedestrian collisions in adjusted analysis for all schools

and excluding 7 outlier schools .............................................................................................. 105

Table 5S-2: Correlates of child pedestrian collisions in unadjusted and adjusted models

for all collisions and those restricted to school travel times ................................................... 106

Table 5S-3: Correlates of child pedestrian collisions in unadjusted and adjusted models

for 10 years, 7 years and 5 years of collision data. ................................................................. 107

Table 5S-4: Correlates of child pedestrian collisions and walking to school in adjusted

analysis using school populations as alternative denominator. .............................................. 108

Page 13: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

xiii

Table 6-1: Summary table of built environment variables associated with walking to school

and child pedestrian-motor vehicle collision from the literature and from the

study analyses. ........................................................................................................................ 123

Table 6-2: Individualized school report knowledge users ..................................................... 142

Table 6-3: Actions taken attributed to individualized school reports by school principals ... 143

Page 14: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

xiv

List of Figures

Figure 2-1: The causal model for injuries. ................................................................................ 16

Figure 2-2: Conceptual framework of an elementary-aged child's travel behavior. ................ 17

Figure 2-3: A conceptual framework for the environmental determinants of active travel in

children. .................................................................................................................................... 18

Figure 2-4: A behavioral model of school transportation. ....................................................... 18

Figure 2-5: Distance to school. ................................................................................................. 24

Figure 2-6: Child pedestrian-motor vehicle collisions and roadway design features. ............. 28

Figure 2-7: Child pedestrian-vehicular collisions in school zones. ......................................... 29

Figure 2-8: Six former municipality boundaries prior to 1998. ............................................... 32

Figure 2-9: Pre-World War II grid street patterns in downtown Toronto ................................ 33

Figure 2-10: Post-World War II street patterns in inner suburbs (Scarborough) .................... 33

Figure 3-1: The Preferred Reporting Items for Systematic Reviews and Meta-Analyses

(PRISMA) flow diagram. .......................................................................................................... 43

Figure 3-2: Correlates of walking and child pedestrian injury. ............................................... 52

Figure 4-1: Flowchart of school participation. ......................................................................... 67

Figure 4S-1: Distribution of walking proportion across 118 study schools ............................ 84

Figure 4S-2: Distribution of the proportion of roads in 118 study school boundaries within

1.6 km of schools ...................................................................................................................... 84

Figure 4S-3: Predicted walking rates by intersection density .................................................. 85

Figure 5-1: Multivariate relationships between walking to school, child pedestrian injury

and the built environment. ........................................................................................................ 98

Figure 5S-1: Distribution of collision rates/10,000/year within 118 study

school boundaries .................................................................................................................... 109

Figure 5S-2: Top 5 pedestrian actions at time of collisions .................................................... 109

Figure 5S-3: Predicted collision rate/10,000/year by multi-family dwelling density ............. 110

Page 15: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

xv

List of Appendices

Appendix A: Search Strategies .............................................................................................. 172

Appendix B: Summary of walking publications ..................................................................... 174

Appendix C: Summary of child pedestrian-motor vehicle collision publications ................. 177

Appendix D: Elementary school boundaries (TDSB) and pre-amalgamated City

of Toronto ............................................................................................................................... 179

Appendix E: Observational counts data collection form ........................................................ 180

Appendix F: Site survey ......................................................................................................... 181

Appendix G: Vehicle speed data collection form ................................................................... 183

Page 16: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

xvi

List of Abbreviations

AIC Akaike information criteria

AST Active school transportation

ATLICO After tax, low income cut-offs

CI Confidence interval

DA Dissemination areas

DALY Disability adjusted life years

GIS Geographic information systems

GTA Greater Toronto Area

EAI Epidemiological Appraisal Instrument

HMC High motorized countries

IRR Incidence rate ratio

IKT Integrated knowledge translation

JK Junior kindergarten

LOI Learning opportunities index

LMC Low motorized countries

MPAC Municipal Property Assessment Corporation

PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses

SD Standard deviation

SES Socioeconomic status

SRTS Safe routes to school

STP School travel planning

TDSB Toronto District School Board

VIF Variance inflation factors

YLD Years Lived with Disability

Page 17: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

1

1 Introduction

In this chapter, the rationale for conducting the research is presented along with the overall and

specific thesis objectives. The thesis organization is then described by chapter and terminology

used throughout the thesis is clarified.

1.1 Rationale

Walking as a form of active transportation has numerous benefits at the individual and

population level. The benefits for the individual include prevention of obesity, hypertension,

osteoporosis and other chronic conditions; and for the population include reduced traffic

congestion, better air quality, and improved quality of life. There is evidence that children’s

physical activity is related to physical activity in adulthood and therefore, the promotion of

walking in children could potentially help prevent adult onset of chronic conditions. There are

risks however, associated with walking near roadways. Road traffic injuries are the leading

cause of death for school age children in Canada. Much of children’s exposure to walking and to

traffic is during their travel to school. It is important that the relationship between the rates of

children walking to school and pedestrian-motor vehicle collisions be established given the

recent popularity of programs to increase walking to school. There is evidence that increased

walking in a community may be associated with less pedestrian-motor vehicle collisions because

of a “safety in numbers” effect. There is also evidence; however, that increased traffic exposure

when walking to school is associated with more pedestrian-motor vehicle collisions. These

conflicting findings may be due to differences in the built environment, in that some

environments may provide safer walking conditions than others.

Specific features of the built environment have been associated with self-reported walking to

school as well as with child pedestrian-motor vehicle collisions. However, no studies to date

have used objective observational exposure data. Studies of the built environment and child

Page 18: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

2

pedestrian-motor vehicle collisions have also generally not taken into account road traffic

exposure in the context of proportion of children walking to school. It is important to determine

if there are features of the built environment which are positively associated with increased

walking and create a safer environment (i.e., less injury) to provide optimal environments for

walking to school. It is equally important to determine if there are features of the built

environment that are associated with increased walking but with an increased risk of injury.

1.2 Overall Objective

To examine the relationships between observed walking to school, child pedestrian-motor

vehicle collisions in the City of Toronto, and the role of the built environment.

1.3 Specific Objectives

1. To systematically review the literature on the relationships between the built environment,

walking to school and child pedestrian-motor vehicle collision rates.

2. To estimate the proportion of observed children walking to school in the City of Toronto

(kindergarten to grade 6).

3. To determine the association between the built environment and proportions of children

walking to school.

4. To estimate child pedestrian-motor vehicle collision rates in the areas surrounding elementary

schools in the City of Toronto.

5. To determine how features of the built environment influence the relationship between

proportion of children walking to school and child pedestrian-motor vehicle collisions.

Page 19: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

3

1.4 Thesis Organization

Chapter 1 provides a brief rationale along with the overall objective of the research and specific

research objectives. Chapter 2 presents the context of the research through a detailed review of

literature pertaining to child pedestrian-motor vehicle collisions and walking to school (with a

focus on Canada). Excepts of this literature review have been included in a book chapter that has

been accepted for publication (Buliung R, Larsen K, Falkner G, Rothman L, Fusco C. Driven to

School: Social Fears and Traffic Environment. In: Walks A. ed. Driving Cities, Driving

Inequality, Driving Politics: The Political Economy and Ecology of Automobility. Winnipeg,

MB: Routledge). The literature review describes the burden of pedestrian-motor vehicle

collisions focusing on children, and examines potential health outcomes of walking to school.

The measurement of pedestrian-motor vehicle collisions and walking to school is discussed, and

conceptual frameworks related to both outcomes are presented. Correlates and interventions

related to both child pedestrian-motor vehicle collisions and walking to school are reviewed with

a focus on the built environment. The use of Geographic Information Systems (GIS) to study

both child pedestrian-motor vehicle collisions and walking to school is presented, and the City of

Toronto, as the setting of the study, is described. The current status of Canadian policy related to

pedestrian injury prevention and walking to school is summarized. Finally, the gaps in

knowledge regarding walking to school, child pedestrian-motor vehicle collisions and the built

environment are identified.

Chapters 3-5 are reformatted versions of manuscripts that have been published or are currently

under review. Supplementary/supportive analyses that were not included in the published

versions are appended at the end of the chapters. Chapter 3 presents a systematic literature

review addressing Objective #1 identified above. The material in this chapter has been published

in Injury Prevention (Rothman, L., Macarthur, C., Buliung, R., To, T., & Howard, A. Walkable

but unsafe? a systematic review of built environment correlates of walking and child pedestrian

injury. Injury Prevention, 18 (Suppl 1) 2012: A223-A223.) Chapter 4 addresses Objectives 2

and 3, and has been published in Preventive Medicine. (Rothman L, To T, Buliung R, Macarthur

C, To T, Howard A. Influence of social and built environment features on children’s walking to

school: an observational study. Prev Med. 60; 2013:10-15).

Page 20: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

4

Chapter 5 addresses Objectives 4 and 5 and has been published online in Pediatrics (Rothman L,

To T, Buliung R, Macarthur C, To T, Howard A Pedestrian-motor vehicle collisions and walking

to school: the role of the built environment. Pediatrics published online: 2014(doi:

10.1542/peds.2013-2317). Chapter 5 includes a detailed methods section further describing the

data collection and analyses pertinent to both Chapters 4 and 5, which was not included in the

published papers. In Chapter 6, the major findings of the thesis are summarized and the

strengths and limitations are discussed, followed by a description of policy implications,

knowledge translation activities and an outline of proposed future studies.

Several terms are used interchangeably in the literature and in this thesis and require

clarification. The terms “child pedestrian injuries” and “child pedestrian-motor vehicle

collisions” were used interchangeably. Although child pedestrian injury implies a measure of

severity, many studies use this terminology to reflect the collision occurrence. The terms “active

school transportation (AST)” and “walking to school” were also used interchangeably. AST

consists of not only walking to school, but also cycling and other active means (e.g. scooters).

Since the numbers of children using active means other than walking in Toronto are extremely

small, AST generally reflects walking to school in this location. Finally, built environment is

also referred to as the “physical environment” or “urban form” in many of the referenced papers.

The built environment refers to the man-made physical environment that provides the setting for

human activities. It includes urban form, physical road infrastructure, land use patterns and

transportation systems.1

2 Background

The purpose of this chapter is to provide background information and context for this thesis.

This section discusses the global burden of motor vehicle collisions with a focus on pedestrian

collisions and children. This section also describes the injury and disability burden of motor

vehicle collisions and collisions involving pedestrians.

Page 21: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

5

2.1 Burden of Motor Vehicle Collisions

Road safety is an international health policy imperative, given the devastating burden of road

traffic injuries. Road safety is also a priority for global sustainable development policy, directed

at increasing the safety and accessibility of non-motorized transportation to reduce air pollution

and traffic congestion.2,3

In March 2010, The United Nations General Assembly declared 2011-

2020 the Decade of Action for Road Safety with the aim of reducing road traffic injuries and

fatalities.2 In 2010, road traffic crashes resulted in approximately 1.24 million people killed and

another 20-50 million non-fatal injuries worldwide.2 According to the Global Burden of Disease

Study, in 2010, road injury ranked eighth for global death rates with a 47% increase since 1990.4

Road injury also ranked seventeenth for Years Lived with Disability (YLDs) with a 30%

increase since 1990.4 It is predicted that road injury will rise in ranking to the fifth leading cause

of death globally and the seventh leading cause of Disability Adjusted Life Years (DALYs) lost

by 2030.5-7

The burden of road traffic collision is higher in low and middle income countries which have

higher annual road traffic fatality rates (18.3 and 20.1 per 100 000 population, respectively)

compared to high-income countries (8.7 per 100 000).2 Middle income countries have seen rapid

motorization over the last 20 years, with similar trends in lower income countries.8,9

Unfortunately, this rapid motorization has not been accompanied by investment in road design

planning and safety strategies, such as law enforcement and education.2

Pedestrian-Motor Vehicle Collisions 2.1.1

The UN General Assembly dedicated the Second UN Global Road Safety Week in May 2013 to

pedestrian safety, in the context of the Decade of Action for Road Safety 2011-2020. Pedestrian

fatalities represent approximately 22% of road traffic deaths worldwide.2 In low income

countries, the proportion is as high as 35% as more people use walking as their main mode of

transportation.2 Some countries report more than 75% of their road traffic fatalities occur in

pedestrian/cyclists.2 Higher fatalities among vulnerable road users in middle and low income

Page 22: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

6

countries is the result of increased motorization and the traffic mix, where there is less developed

traffic safety infrastructure.

Pedestrian-motor vehicle collision rates in high income countries have been declining over the

past 20 years. Declining trends have been noted in the United States, Canada, Europe and New

Zealand.10-15

In Toronto, Canada, there was a 24% decline in collision rates from 2000-2009.

Despite declining pedestrian-motor vehicle collision rates from 1995-2009, pedestrians

accounted for approximately 50% of all road traffic fatalities in Toronto.16

Although pedestrians

accounted for only 7% of the transportation mode share, they represented 52% of all fatalities

and 11% of all collisions involving motor vehicles.16

The high proportion of pedestrian fatalities

is markedly different than that of the rest of Canada, where pedestrians account for 13% of road

traffic fatalities.17

This indicates that pedestrian-motor vehicle fatalities are a serious problem in

urban environments. In Toronto, 39% of pedestrian-motor vehicle collisions led to hospital

visits, 8.8% resulted in hospitalization and 1.4% resulted in fatality in 2009.16

The estimated

annual cost of pedestrian -motor vehicle collisions in Toronto, which include discounted future

earnings, direct medical costs and other direct costs, totals $53,606,465.16

Children 2.1.2

Children are especially vulnerable to road traffic injuries because of their small stature and

developing physical and cognitive skills. In 2010, road injuries were the 5th leading cause of

death for children ages 5-9 years, the 4th leading cause for ages 10-14 years and the leading

cause for young people 15-24 years of age world-wide.4,18

Road traffic crashes result in more

than 260,000 child fatalities each year and approximately 10 million non-fatal injuries; leaving

one million children with long-term disabilities.19

Child pedestrians represent 5-10% of children

of those with road traffic injuries in high-income countries, whereas they represent between 30-

40% in low and middle-income countries.20

In Cape Town Safe Africa, 75% of children

admitted to a hospital trauma unit in 2011 as a result of road traffic collisions, were pedestrians.21

In these countries, child pedestrians share the roads with many types of motorized transport.20

Although there has been a downward trend in road traffic injuries in high income countries, they

continue to be a leading cause of child death.5,10,22

In high-income European countries, 1 in 5

Page 23: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

7

childhood injury deaths are the result of road traffic injuries.5 In the United States, motor vehicle

crashes were the leading cause of death for children under the age of 14 years in 2011, with 3

child fatalities and 469 injuries on average, every day.23

In Canada, unintentional injuries are the

leading cause of death in children, with motor vehicle fatalities occurring almost six times more

often than any other unintentional injury group.24

Hospitalization due to motor vehicle injuries

ranks 2nd

(after falls) for all injury admissions in young people in Canada.24

In 2010, there were

295 fatalities, approximately 2000 serious injuries and almost 30,000 reported injuries due to

traffic crashes in Canadian children ages 0-19, caused by vehicle occupant trauma, pedestrian

injury, and cycling collisions.17

Beyond the injuries and the burden it places on emergency and

rehabilitation systems, road traffic injuries in children 0-19 years of age cost the Canadian health

care system $1 billion annually.25

There was a 50% decline in hospitalizations and deaths due to child pedestrian-motor vehicle

collisions in Canada from 1994-2004.10,26

Despite these declines, the burden remains high. In

children under age 14 in 2001, the proportion of all road user fatalities that were pedestrian

related is 25% as opposed to 13% in adults.27

Approximately 56 child pedestrians die and 780

are hospitalized with serious injuries in Canada every year.10

In children ages 5-9 years,

pedestrian-motor vehicle collisions are tied with motor vehicle collisions at 18% as the primary

cause of unintentional injury death in this age group in Canada.10

Injury and Disability 2.1.3

Head injuries are the leading cause of traffic-related injuries and fatalities, especially in

children.28

Injuries to limbs such as fractures, abrasions and contusions are also common in road

traffic collisions, particularly for those injured as pedestrians.20

Child pedestrian-motor vehicle

collisions are more severe than collisions involving motor vehicle occupants because of the lack

of physical protection separating them from the colliding force. Because of a child’s short

stature, a child’s head is frequently the first point of contact with the bumper of a colliding

vehicle. From 1983-1990, pedestrian injuries accounted for two thirds of all severe/fatal traffic

injuries in children under 17 years of age in the Northern Manhattan Injury Surveillance System,

with 45% sustaining head trauma.29

In children presenting to the emergency department after a

Page 24: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

8

road traffic collision in Great Britain, approximately 2/3 of pedestrians had injuries above the

neck, with 33% sustaining severe head injuries.30

Brison et al. found that head and neck injuries

were the primary cause of pedestrian-related death in children under five years in Washington.31

Head injuries accounted for over 20% of pedestrian hospital admissions in Canada in those < 20

years in 2002,32

and 19% presenting to emergency departments in 2008/2009.10

Long-term disability is common among those who survive motor vehicle collisions, with road

traffic injuries ranking 9th

in 2002 for DALYS, representing an estimated 38 million DALYs.8 In

Toronto, Canada, 72% of child pedestrians and 59% of motor vehicle occupants who were

seriously injured and admitted to hospital, required assistance with daily activities when they

returned home after six months.33

Younger age and a primary diagnosis of a central nervous

system injury were associated with requiring assistance.

2.2 Health Outcomes of Walking for Transportation

Child pedestrian-motor vehicle collisions and the prevention of chronic conditions are potential

health outcomes of walking for transportation. This section reviews the trends in walking to

school and child pedestrian-motor vehicle collisions over the last 20 years and discusses the

potential relationship between pedestrian volume and collisions. The benefits of walking as a

means of physical activity are discussed and the relationships to health outcomes such as obesity

are described.

Child Pedestrian-Motor Vehicle Collisions 2.2.1

Many believe the downward trend in child pedestrian-motor vehicle collisions over the last 20

years in higher income countries, is because of children walking less, thereby reducing their

exposure to the risk of collisions with a motor vehicle.12,14,34

In the United States, 41% of

schoolchildren walked or biked to school in 1969, and this had dropped to 13% in 2001.35

In the

2004 Canadian National Transportation Survey, 50% of children reported never walking to

Page 25: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

9

school.36

In an analysis of the Transportation Tomorrow Survey (TTS), Buliung et al. found that

walking mode share for trips to school in 11-13 year olds in the Greater Toronto Area (GTA)

decreased from 53% to 42.5% from 1986-2006.37

The 2013 Active Healthy Kids Canada Report

Card on Physical Activity for Children and Youth indicated that parents report that 24% of

Canadian 5-17 year olds use only active transportation to and from school and 14% use a

combination of active and inactive modes of transportation.38

There has been an increase in

those who only report inactive modes of transportation to/from school from 51% to 62% from

2000-2010.38

Despite the decreasing numbers of children walking to school, almost 50% of pedestrian-motor

vehicle collisions involving children <17 years in Toronto, occurred during school travel times

and months. Warsh et al. used Geographic Information Systems (GIS) to assess the distance of

police-reported collisions in school age children related to school location.39

More than 1/3 of

collisions were within 300 m of a school, with the highest density of collisions among children

occurring within 150m of a school. Yiannakoulias et al. analyzed emergency department

surveillance data from all hospitals in Edmonton. Peak times of child pedestrian-motor vehicle

collisions were in the morning (7:00-9:00) and afternoon (15:00-18:00) which corresponded with

school start and finish times and peak times of traffic volume.40

Unfortunately, road traffic exposure is poorly understood and there exists conflicting evidence

related to pedestrian volume and collisions. Jacobsen found a ‘safety in numbers’ effect in 3

large population datasets conducted in Europe and the U.S. Pedestrian volume, as measured by

journey to work share, distance or trip/day/capita was associated with decreased collisions.

Jacobsen calculated that an individual pedestrian’s collision risk decreased to 66% in

communities where there is twice as much walking.41

Studies which address pedestrian-motor

vehicle collisions and specifically walking to school, have found the reverse, namely that

positive associations exist between walking exposure and child pedestrian-motor vehicle

collisions.42-44

Macpherson et al. conducted a survey in 2,501 grades 1 and 4 students in 43

elementary schools in Montreal, Quebec. A strong positive correlation was found between

numbers of parent-reported road crossings on a school day and child pedestrian-motor vehicle

injuries according to police records (correlation coefficient = 0.78).42

Rao et al. conducted

surveys in 804 grades 1 and 4 students in 26 schools in Baltimore, Maryland. They found a

significant inverse correlation between the proportion of children driven home from school as

Page 26: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

10

reported by parents and students, and the rate of police-reported pedestrian-motor vehicle

collisions (r = -0.79, p<.01).43

In the Canadian Health Behavior in School-Aged Children

Survey, a weighted sample of 20,076, 11-15 year old students completed self-report surveys in

419 schools regarding their use of active transportation and active transportation injuries.45

Gropp et al. reported a 1.5 increase in the odds of active transportation injury in the year

previous to the survey for those who engaged in active transportation over longer distances (>15

minutes), after adjusting for age and urban/rural status.44

There was evidence of a dose response

relationship between longer travel distances and injury. Therefore, depending on the mix of

walking and driving and the environmental conditions present, walking promotion could either

increase or decrease the risk of injury per trip. Additionally, environmental conditions that

ensure safe walking may be different for children compared to adults. Optimal conditions for

safe walking to school must be defined, because if planned poorly, increased walking has the

potential to increase injury risk in children.

Prevention of Chronic Conditions 2.2.2

The promotion of physical activity in children is important to encourage healthy lifelong

lifestyles and to reduce the prevalence of obesity and associated impact on health. Obesity is on

the rise in Canada. Almost 9 % of children ages 6-17 in Canada are obese and this has increased

2.5 times from 1978/1979 to 2004.46

A systematic review of the literature by Singh et al. found

that children who are overweight or obese are at an increased risk of becoming an overweight

adult.47

Another review of the literature by Ball et al. found that childhood obesity contributes to

the early development of cardiovascular diseases and type 2 diabetes.48

The proportion of deaths

attributed to being overweight or obese in adults has been estimated to have increased from 5.1%

in 1985 to 9.3% in 2000 in Canada.49

The Canadian physical activity guidelines for children 5-17 years recommend a minimum of 60

minutes of moderate-to vigorous-intensity activity per day.50

It was estimated in the Canadian

Physical Activity Levels Among Youth study, that 88% of children and youth do not meet the

recommended physical activity guidelines.51

Focus is turning more towards lifestyle activities to

increase physical activity such as walking, biking and taking stairs which can be done on a daily

Page 27: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

11

basis and over the lifespan. Walking is an especially accessible means of physical activity for

most people, as there is no special equipment or facilities required and it can be incorporated into

the daily trips to work or to school. In addition to the numerous health benefits of active

transportation, there are other transportation benefits including less traffic congestion, less fuel

costs, shorter and more reliable travel times, and fewer road traffic collisions, and societal

benefits including less air and noise pollution, less crime, community cohesion and higher real

estate value.16

In adults, active transportation has been shown to be associated with less obesity.16

Gordon-

Larsen et al. found that men who walked or cycled to work were half as likely to be obese.52

Frank et al. found a 6% increase in the likelihood of obesity with every additional hour spent in a

car every day.53

A systematic literature review by Faulkner et al. found that children who

actively commute to school reported significantly higher levels of physical activity.54

Although

the cross-sectional design of the majority of papers prevent inferences of causality between AST

and physical activity, it is possible to conclude that children who engage in AST are more

physically active. There was little evidence of a relationship between active transportation to

school and healthier BMI, probably due to the short walking distances to school. The physical

benefits of physical activity in children have not been well established, perhaps because of the

low frequency of morbidity due to sedentary behaviours in children.55

The benefits of active commuting in childhood may not be apparent until years later, assuming

the active commuting habits are maintained.54

There is evidence that children’s physical activity

is related to physical activity in adulthood. In a review of the literature, Malina found a

correlation between participation in physical activity during childhood and youth into

adulthood.56

In a study of children and adolescence, pre or early-pubescent boys classified as

sedentary based on measurements of TV viewing and video game playing were 2.2 times more

likely than their peers to be classified as sedentary adolescents, five years later.57

In a 21-year

tracking study using data from the Cardiovascular Risk in Young Finns Study, Telama et al.

found that high levels of physical activity at ages 9 to 18 years increased the odds that the

individuals would be highly active adults, with the probability being even higher if the physical

activity had lasted for several years in youth.58

The evidence supports the promotion of walking

and active transportation in children as a form of physical activity, which may continue into

adulthood to prevent obesity and the development of chronic conditions.

Page 28: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

12

2.3 Measurement

Accurate measure of outcomes is essential to the validity of the research process. In this section,

the strengths and limitations of different methods used to measure child pedestrian-motor vehicle

collisions are reviewed. The issues around the inconsistency in measurement of walking to

school and the effect on prevalence estimates are discussed.

Child Pedestrian-Motor Vehicle Collisions 2.3.1

To effectively study the relationship between walking to school and collisions, the validity of

data sources and outcome measurement must be established. The main sources for pedestrian-

motor vehicle collision data are hospital/trauma surveillance databases, death registries,

coroner’s reports and police-reported data. Standardized Emergency Medical Services (EMS)

clinical databases in Canada and in the United States may also be potential sources for collision

data. Although death registries, coroner’s reports, EMS service reports and health/trauma

databases can potentially be rich sources of information regarding the specifics of the injury and

health outcome, the patient population represents the most severe end of the pedestrian-motor

vehicle collision spectrum and results are not generalizable to all collisions. Police-reported

collision data also have limitations, as they have been found to underreport child pedestrian-

motor vehicle collisions.59-61

In a study of pedestrian-motor vehicle collisions in those under 15

years old, comparing emergency department records and the coroner’s logbook to a police-

reported database in Orange County, California, Agran et al. found that 20% of hospital

admissions in children under 15 years of age were not reported to police.61

Generally,

unreported collisions were in very young children (0-4 years) and were non-traffic, such as

backing up collisions and those occurring off-road (e.g. on sidewalk). Unreported pedestrian-

motor vehicle collisions may also be due to the perception of these types of collisions as injury

events rather than a reportable motor vehicle collision.61

In the U.S., some jurisdictions are also

not required to report collisions occurring on private property, where most non-traffic incidents

Page 29: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

13

occur.61

Police data are also less likely to capture less severe collisions. In Ontario, the

Highway Traffic Act indicates that collisions must be reported if it results in personal injury or in

property damage exceeding $1,000.62

Therefore, police-reported collision databases in Ontario

would likely not include collisions where there was no or very minimal injury.

Collision reporting is very different in lower income countries as reporting of collisions to police

is not always mandatory. In Uganda, by Lee et al. found police-reported child pedestrian injury

rates were approximately the same as in a hospital-based trauma registry, but were 14 times

lower than those found in a community-based survey, and 35 times lower than those reported by

teachers.63

Underestimation by the police may be because of failure of the police to record the

incident or failure to report to the police.64

The limitations of using police-reported data in lower

income countries must be recognized as collision rates may be severely underestimated.

Police-reported collision data are routinely collected in high income countries. These data are

population-based and therefore have greater generalizability compared to data restricted to a

particular hospital or trauma registry system. These data also include detailed on-scene

information regarding location and circumstances, and geographic coordinates of collision

locations. Although it is recognized that collisions involving no/little injury may be

underrepresented in police-reported collision databases, these databases provide the most useful

data compared to other sources when investigate environmental conditions associated with

pedestrian-motor vehicle collisions.

Walking to School 2.3.2

The methods of measurement of walking to school are inconsistent between studies. There are a

differences in how the outcome is measured (e.g. usual trip, numbers of trips per week), recall

time frames (last week, today), and age ranges from study to study.65,66

Self-reported methods

are generally used to measure walking to school, including parent or student written

questionnaire, online or telephone surveys or travel diaries.65,66

Self-report or proxy-report measures of walking to school have not been well-validated, which

could lead to error due to selection and social desirability bias, recall error and low response

Page 30: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

14

rates.65,67

Rossen et al. reported only moderate agreement between parent versus child-reported

walking to school during face-to-face interviews (kappa= 48.7%).68

An older study reported by

Routledge et al. in 1974, used a ‘moving observer’ technique to validate child-reported exposure.

One hundred and forty two children were followed home from school, and were then interviewed

regarding road crossings the next day.69

A statistically significant difference was found in

number of road crossings with children slightly underreporting crossings. Stevenson et al. also

examined the validity of children’s reported estimates of usual walking activities during the

course of a typical week (i.e., habitual exposure”) using several different techniques including

the moving observer technique and pedestrian diaries.70

An interview was conducted with the

child daily for a week and asked questions regarding the regular walking journey each day

(habitual exposure). The moving observer also recorded the characteristics of the journey

described above. A total of 52 observations were made for 13 children. A high concordance was

found between reported and observed habitual exposure. However, higher mean values were

reported in diaries than at the interview for 3/5 habitual exposure questions.

More consistent and objective measures of walking would improve accuracy of prevalence

estimates of walking to school.65,71

To date, only one study by Sirard et al. used direct

observational counts of children’s mode of transport to school to examine prevalence and

correlates of active transportation.72

In their study, two to three observers visited 8 schools to

identify travel behavior in the morning and afternoon on 5 consecutive school days in the fall.

The study sample was small, and correlates examined only included school SES level, school

urbanization level, weather conditions and temperature. The study results were limited by

minimal geographic diversity.

2.4 Conceptual Frameworks

Conceptual frameworks help frame the multitude of factors affecting child pedestrian-motor

vehicle collisions and walking to school. This section presents conceptual frameworks related to

both these outcomes, with a focus on those which incorporate the built environment. The

influence of the built environment has been well recognized in both the injury prevention and the

Page 31: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

15

active transportation fields. Conceptual frameworks related to pedestrian-motor vehicle

collisions have evolved over the last 40 years, whereas, the development of frameworks related

to school travel have been more recent; generally occurring over the last 15 years as interest has

grown regarding the promotion of active transportation. In all models related to AST, the

outcome of interest has been school travel mode; no models have been extended to illustrate the

impact of school travel mode on pedestrian-motor vehicle collisions as an outcome. . Similarly,

none of the models focused on pedestrian-motor vehicle collisions have incorporated walking

exposure. Therefore, the intention of this research was to build on the models presented below,

to explore how the built environment influences the relationship between walking to school and

child pedestrian-motor vehicle collision outcomes.

Child Pedestrian-Motor Vehicle Collisions 2.4.1

The most prominent conceptual model originally developed to describe motor vehicle collisions

and later extended to all types of injury, is known as “Haddon’s matrix”.73

Table 2-1: Haddon's Matrix. Pedestrian-motor vehicle collision example

(adapted from SafeKids Canada, 2004, Copyright Parachute 2013, permission granted to

reproduce).74

Host Agent/Vehicle Environment

Physical Social

Pre-event

(before the

child is hit)

Road

crossing

behaviour

Adult

supervision

Child’s age

Child’s

gender

Risk taking

Speed

Driver

behaviour

Driver

knowledge

Driver

experience

Vehicle design

Road design

Presence/

condition of

sidewalks

Pedestrian

proximity to traffic

Signage

Crosswalks

Type of housing

Weather

Daylight

Value placed on

pedestrian safety

Policy/promotion of

pedestrian safety

measures

Law enforcement

Neighbourhood socio-

economic conditions

Event

(during

collision)

Head striking

vehicle

Vehicle

impacting

pedestrian

Availability of

phone for

emergency call

Person available to

notify emergency

personnel

Post-event

(after child

is injured)

Post injury

care

Severity of

injuries

Distance to trauma

center

Family and social

support

Trauma center training

Page 32: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

16

This model by William Haddon Jr. was instrumental in increasing the understanding of the

different factors contributing to both the occurrence and the severity of road traffic crashes. In

the first dimension of the 9-cell model matrix, there are 3 phases of hazardous events during

which countermeasures can be taken: the pre-event stage, the event, and the post event stage.

The 2nd dimension of the model includes 3 factors: Host, agent, and environment. Table 2-1

presents an example of the Haddon matrix completed for child pedestrian-motor vehicle

collisions. In this example, the host is the child at risk of injury and the agent is the energy

transferred to the host by a vehicle. The environment refers to the physical environment,

including the characteristics of the setting in which the injury event takes place (the roadways),

and the social environment, which refers to the social and legal norms and practices (e.g. child

supervision, speed limits).75

By examining the 3 factors during the different crash phases, it is

possible to identify risk/protective factors and develop preventive strategies.76

In another causal model for injury by Peek-Asa, the environment influences the transfer of

kinetic energy (i.e., agent) through a vehicle (motor vehicle) to the human host (Figure 2-1).77

She describes the environment, as physical (either natural or man-made), social, economic

cultural and demographic. Peek-Asa

emphasizes that modification to the

physical environment is the most effective

approach to preventing injuries, as it is

passive (i.e., does not require anything

from the host to be effective), and it

affects populations rather than just

individuals.

Walking to School 2.4.2

The first conceptual framework relating to an elementary children’s travel behaviour was

developed by McMillan (Figure 2-2).78

In this model, school travel mode is a result of parental

decision processes. Urban form has an indirect relationship with a parent’s decision regarding

the child’s mode of transport to school. Elements of the urban form have to be processed

Figure 2-1: The causal model for injuries

(permission granted to reproduce).

Page 33: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

17

Urban Form

Mediating factors-Neighborhood safety (real/perceived)-Traffic safety (real/perceived)-Household transportation options

Children’s travel behavior (trip to school)

Moderating Factors-Social-cultural norms-Parental attitudes-Sociodemographic

Parental decision-making

through factors such as social/cultural norms, sociodemographic characteristics, household

transportation options and real/perceived neighborhood and traffic safety, which are then linked

directly to the parent’s decision of transportation model.

Figure 2-2: Conceptual framework of an elementary-aged child's travel behavior

(permission granted to reproduce).

More detail regarding the concept of the built environment is provided in models by Panter et

al.79

and Mitra.80

Panter et al. took an ecological approach to understanding travel behavior in

their conceptual framework of the environmental determinants of active travel in children

(Figure 2-3).79

The framework describes four domains of influence on choice of active travel

modes; individual/household (i.e., attitudes, characteristics, and perceptions), external factors

(e.g. weather), the main moderators (age, sex and distance), and finally, physical environmental

factors, including characteristics of the neighbourhood, destination and route environment.79

This framework does not, however, account for the underlying behavioural processes involved in

choosing modes of transportation.

More recently, Mitra developed a behavioural model of school transportation using a social-

ecological framework, which draws on ecological theories of human behaviour such as described

by Bronfenbrenner and Sallis et al. (Figure 2-4).80-82

These theories emphasize the influence of

the environment on behaviour. Mitra’s model hypothesizes multiple levels of influence of mode

choice for school transportation and independent mobility: the urban environment, household,

characteristics of a child/youth, and other external factors.80

The urban environment is an

important component of this model and influences travel by its spatial structure (i.e., distribution

of residences, employment and other facilities), its built environment (i.e., land use mix,

transportation network and urban design features) and its social environment.80

Page 34: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

18

Figure 2-3: A conceptual framework for the environmental determinants of active travel in

children (permission granted to reproduce). .

Figure2-4: A behavioral model of school transportation (permission granted to reproduce).

Page 35: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

19

A broad range of many intercorrelated built environment factors have been studied in relation to

walking to school, necessitating a model to organize these factors. A landmark study from the

urban design literature by Cervero and Kockelman described a model that proposed that built

environment variables related to travel demand can be organized and described along 3 principal

dimensions, referred to as the 3Ds: density, diversity and design (Table 2-2).83

Table 2-2: Built environment variables most associated with travel demand (permission

granted to reproduce).

3 D’s Built Environment Variables

Density Population per developed acre

Employment per developed acre

Accessibility to all jobs

Diversity Dissimilarity index (proportion dissimilar land use)

Mean entropy (land use mix index)

Per developed acre rates of:

-retail stores

-activity centers

-parks and recreational sites

Proportion of commercial-retail parcels that are vertically mixed

(more than one land-use on site)

Proportion of residential acres within ¼ mile of convenience or

retail store

Design Proportion of intersections that are four-way

Proportion of blocks with:

-sidewalks

-planting strips

-overhead lights

-flat terrain (< 5% slope)

-quadrilateral shape

Block face length

Sidewalk width

Distance between overhead lights

Proportion of commercial parcels with:

-paid parking

-side or front lot, on-street parking

According to this model, those living in higher-density neighbourhoods that have more land–use

diversity and more pedestrian-oriented designs (e.g. street trees, sidewalks ) are more likely to

walk or bike for transportation.83

This 3D model has continued to be used and adapted to include

other “D”s in the literature (e.g. destination accessibility and distance to transit) to organize built

environment factors.84

This model was originally developed to study adult walking behaviour,

Page 36: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

20

but recently has also been used in the children’s school transport literature. Wong et al. used the

model to organize literature in a systematic review of the literature related to GIS measurement

of built environment correlates of active school transport.85

Lin et al. in their analysis of built

environment effects on children’s school travel in Taipei also organized the explanatory

variables according to the 3Ds.86

The usefulness of using the 3 Ds paradigm to classify built

environment has been well demonstrated, and it will be subsequently used in this thesis.

2.5 Correlates and Interventions

This section reviews the correlates of child pedestrian-motor vehicle collisions and interventions

designed to reduce collisions with the focus on the built environment. The correlates of walking

to school and interventions to increase walking are also discussed along with the issues related to

inconsistencies in the association between walking and the built environment.

Child Pedestrian-Motor Vehicle Collisions 2.5.1

2.5.1.1 Correlates

Behavioural, social, cultural and built environmental factors all play a role in child pedestrian-

motor vehicle collisions. However, increased emphasis is being placed on factors related to the

built environment which are felt to be the most modifiable. Risk factors of child pedestrian-

motor vehicle collisions were examined in a systematic review of Medline literature by Wazana

et al. in 1997.87

Eighteen analytic studies were reviewed and risk factors were classified into the

following groups 1) child 2) social/cultural 3) physical environment and 4) driver. The child risk

factors were identified in descending order of impact: younger age, behaviour (e.g. impulsivity),

non-white and male. Social risk and cultural risk factors were: lower income, more children

living in home, less parent preventive behaviours, mother working, lower maternal education,

history of mother being hospitalized and illness in family. The physical environment risk factors

for child pedestrian-motor vehicle collisions or greater severity of injury were: higher traffic

Page 37: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

21

volume, higher speed limit and vehicle speed, absence of play area, being on road (versus off-

road), streets with predominantly rental units and multi-family dwellings, higher proportion of

curb side parking, shared driveway, major roadways, after 3 pm, rainy weather, and darkness.

The driver risk factors for increased injury severity were lack of avoidance behaviour and higher

speed driving.

Wazana defined directly modifiable factors as risk factors which can be directly affected by an

intervention. He indicated that the most directly modifiable risk factors were those related to the

physical environment and that other than age and SES, the physical environment risk factors had

the greatest magnitude of risk associated with them. Wazana described how the focus on the

modification of environmental risk factors in Sweden and Denmark may explain their success in

decreasing child pedestrian mortality rates.

A recent paper by Dimaggio and Li, systematically reviewed the literature focused on pediatric

pedestrian injury risk and the built roadway environment.88

A meta-analysis using Bayesian

techniques was conducted to synthesize the evidence on the association of roadway

characteristics with pediatric pedestrian injury risk. Ten databases were searched and 26

quantitative articles were selected for inclusion. The synthesized effect estimate for the

association of roadway characteristics with injury risk was OR = 2.5 (95% CI: 1.8, 3.2) for

pediatric populations. Although this analysis did not specifically identify which roadway

characteristics are most amenable to intervention, the analysis suggested that built environment

interventions directed at the roadway may result in meaningful reductions in pediatric pedestrian

injury risk.

2.5.1.2 Interventions to Decrease Child Pedestrian-Motor Vehicle Collisions

Interventions to reduce child pedestrian-motor vehicle collisions have traditionally been directed

at traffic safety education. Educational interventions when used in isolation, however; have not

led to a reduction in deaths and serious injuries from road traffic collisions.89

Although these

interventions can change behaviour, effectiveness has not been shown in terms of reducing rates

of road traffic crashes.8 A systematic review of randomized trials of road safety educational

interventions to reduce pedestrian-motor vehicle collisions found some programs did improve

Page 38: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

22

safety knowledge and road crossing behavior but methodological quality was poor and none of

the studies reviewed linked changes in these behaviours to injury.90

Pedestrian-motor vehicle

collision prevention programs are felt to have limited value, as they have not been shown to have

substantial effects on injury rates.12,91,92

Roberts has suggested that scarce resources be redirected instead to environmental approaches

which have substantial evidence supporting their efficacy.91

Modification of the built

environment removes the responsibility for traffic safety solely from the individual, and benefits

the community as a whole which is a more promising and efficient approach. A cost

effectiveness analysis demonstrated that 18 child pedestrian deaths could be prevented each year

in New Zealand if funds were redirected from pedestrian education to traffic calming.91

In many countries, roads have been built focused on motor vehicle users, with less consideration

for pedestrian safety. High speeds road have been built in residential areas and there have not

been adequate safe play and walking areas integrated into the planning of communities.20

Many

environmental modification strategies have focused on speed reduction. Speed is the major risk

factor for crashes, and directly influences injury severity.8,93

The World Health Organization

reports that pedestrians have a 90% chance of surviving collisions at < 30 km/h or below, but

less than a 50% chance of surviving at collisions at >45 km/hr.8,94

Since 2002, speeding has

been a factor in approximately 1/3 of motor vehicle crash deaths in the United States.95

In

Sweden, a power model estimating the relationship between speed and safety found a 5%

increase in mean speed led to approximately a 10% increase in all injury and a 20% increase in

fatal collisions.96

Changes in speed limit laws and reducing speed limits from 30 mph to 20 mph

was associated with an estimated reduction in child pedestrian-motor vehicle collisions by 67%

in the United Kingdom.97

In Zurich, reduction of speed from 60 to 50 km/hr was associated with

a decrease in collisions by 16%, injured pedestrians by 20%, and fatalities by 25%.98

Other interventions aimed at environmental modification have shown effectiveness in reducing

collisions and injury. Retting et al. in a review of traffic engineering literature and pedestrian-

motor vehicle collisions divided countermeasures into 3 categories: speed control, separation of

pedestrians from vehicles and measures that increase the visibility of pedestrians.99

Speed

control measures included traffic calming devices such as speed humps and lane narrowing and

multiway stop-sign controls. Separation of pedestrians from vehicles included devices to

Page 39: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

23

separated pedestrians by time (i.e., exclusive pedestrian signal phases) or by place (i.e., barriers

and sidewalks). Visibility interventions included lighting, crosswalk markings and adaptations to

parking. Measures that were found to be highly effective were single-lane roundabouts,

pedestrian islands and pedestrian signal phasing and increased roadway lighting. There were

other promising measures which had only limited evaluation. The review concluded that

pedestrian-motor vehicle collisions could be reduced by 50% to 75% in specific locations and

25% area-wide.

Sixteen controlled before-after studies in high income countries addressing area-wide traffic

calming strategies such as those to slow down traffic (e.g. speed humps), visual changes

(lighting), redistribution of traffic (e.g. one-way streets) and changes to road environments (e.g.

trees) were reviewed by Bunn et al.100

This review found evidence for a 37% reduction in fatal

outcomes and 11% reduction in severe outcomes using area wide traffic calming. Other

systematic reviews focus specifically on the effectiveness of red light cameras,101

speed

enforcement detection devices,102

and street lighting in reducing crashes.103

All reviews reported

that these types of interventions are effective in reducing the number of crashes causing

injury/fatality.

A study was recently published by Dimaggio and Li, which examined the effectiveness of

environment safety improvements in reducing pedestrian injuries in school-age children in New

York City as part of the Safe Routes to School Program (SRTS).104

Improvements at 124

schools included speed reduction devices (e.g. speed bumps, speed boards), high visibility

crosswalks, and exclusive pedestrian signals. The annual rate of pedestrian injury decreased

33% in school-age children (44% during school-travel hours), and 14% in other age groups in

census tracts with SRTS interventions. The rates remained unchanged in areas without SRTS

interventions. This study highlighted the need for evaluation of programs designed to increase

walking to school to focus on pedestrian injury outcomes, as well as on walking rates.

Page 40: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

24

Walking to School 2.5.2

2.5.2.1 Correlates

Several literature reviews have investigated the correlates of active travel by children. The

strongest and most consistent correlate with walking to school is distance to school. Wong et al.

conducted a systematic review examining GIS measured built environment correlates of AST,

and found that measured distance to school was negatively associated with AST.85,105-108

In an

Australian study, the odds of walking to school was 5 times greater for school trips < 800 m

compared to trips > 800, for 5-6 year olds, and 10.2 times greater for 10-12 year olds.108

Mitra et

al. in a study of 11-12 year olds in Toronto found that a 1 km decrease in GIS measured travel

distance increased the odds of walking by

0.71 to 0.72 times.105

Reported distance to school is also

strongly associated with school travel

mode.78,109-116

In a study by Mcmillan, the

probability of AST increased if reported

distance from home to school was less

than a mile (i.e., < 1.6 km).117

In an

analysis of the US Department of

Transportation’s 2001 National Household

Travel Survey for ages 5-13 years, Mcdonald

found that travel time (i.e., distance to school)

had the strongest effect on the decision to walk to school with a 10% increase in walk travel time

leading to a 7.5% decrease in walk mode share.111

She created scenarios based on findings from

her study (Figure 2-5): for example, if all children lived 0.8 km from their school the model

estimates that 34% would walk. If students lived 1.6 km (1 mile) from their school, 19% would

walk. In a model by Salmon, 47% of those living within a 15-minute walk to school (estimated

to be approximately 1.6 km) usually walked compared with 4% of those living further away.112

Other correlates of active travel in children are more difficult to define. Many studies have

examined the correlates of walking in adults, with some consensus that walking in adults

12

43

34

19

51

0

10

20

30

40

50

%

Figure 2-5: Distance to school.

Page 41: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

25

is related to density, mixed land use, pedestrian facilities (including sidewalks, trails,

crosswalks), high connectivity grid network, short block lengths, many intersections with few

cul-de-sacs/dead ends and accessibility (proximity to multiple destinations).67,118,119

Studies of

walking in elementary school-age children are less common and results are inconsistent with

what has been found in adults.

In children, walking to school has been examined as their primary walking destination. In a

systematic review of the literature which analyzed walking to school correlates using

multivariate analysis, Sirard et al, distinguished between factors at the policy, neighbourhood

and parent/family level that influenced AST.65

Correlates identified to have a positive association

with AST included: 1) Policy level: physical education classes 2) Neighborhood level

(objectively measured): shorter distance to school, urban area, street/intersection density,

windows facing the street, complete sidewalk system, mixed land use, area-level SES (higher

SES), residential and/or workplace density, and population density 3) Parent/family level

(reported): urban area, stores/facilities close by, walking and bike facilities, land-use mix,

aesthetics, socialization, family approval of walking, sidewalks on most streets (child report),

male gender, single-parent family, number of children and stay-at-home parent.

Associations with built environment variables tend to be inconsistent. In the Sirard review,

many built environment correlates had positive associations with walking in some studies, and

null associations in other.65

The related concepts of route directness and street connectivity have

been reported to be both positively associated,112,120

and negatively associated with

walking.108,121

Residential/population density, mixed land use, sidewalks, crosswalks, trails,

traffic lights, parks/recreational facilities, lower road class, lower traffic volume and less street

connectivity (including dead- ends and cul-des sacs), have all been found to be associated with

walking to school in some studies, 35,105,107,108,110-112,116,117,120-135

with other studies reporting null

associations.105,107,117,120,121,124,127,130 ,136,137

Traffic calming and less speed were consistently

associated with more walking.117,121,125,130

There are several possible explanations for the inconsistent results. Built environment correlates

vary by walking purpose in children.130

Age ranges of targeted populations of children vary

from study to study. There are differences in how walking outcome is measured (parent versus

child-reported) and how walking is conceptualized, with the reporting of usual trip, trip

Page 42: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

26

per/week, or with different recall time frames. Differences in how the correlates are measured

can also affect results, as some studies measured the built environment using perceived measures

and others using objective measures such as field surveys and databases using GIS. There have

also been methodological challenges identified when using GIS measurement of correlates. In a

systematic review of 14 studies examining GIS measured environmental correlates of AST by

Wong et al., inconsistencies were identified between studies regarding how data were geocoded,

the different buffer size and shapes used, the quality of environmental data and difficulties with

the estimation of the school routes.85

As a result, conclusions were limited; with distance to

school being the only consistent significant negative correlate with AST identified. Measurement

standards are required to establish a definitive list of factors that influence walking in children.

2.5.2.2 Interventions to Increase Walking to School

Few studies examine the effectiveness of interventions directed at increasing AST. A systematic

review conducted in 2011 by Chillon et al. found 14 studies focused on interventions to increase

walking to school in children and adolescents.138

The intervention design for each study was

examined using the Active Living by Design Community Action Model.139

This model is a

framework with multi-level strategies to increase physical activity and has been used in other

AST studies. There are five strategies outlined in this model: 1) Preparation: deliberate process

of getting ready for and reinforcing action 2) Promotion: education and encouraging opinion

leaders and the public 3) Programs: organized activities directed at increasing physical activities

4) Policies: rules or standards that affect physical activities and 5) Physical projects: removing

barriers to physical activity and create opportunities by directly changing the built environment.

Chillon reviewed 13 studies, of which 2 studies included all 5 strategies. Only 3 studies

integrated physical projects. Only the studies by Boarnet et al. focused on physical projects

directed at the built environment in terms of infrastructure projects in the community.128,140

Boarnet et al. found that children who passed infrastructure projects completed as part of Safe

Routes to School programs, including sidewalk, crossing and traffic control projects in California

at 10 schools were more likely to show increases in active school travel than those who did not

(15% vs. 4%).128

Projects with evidence of success in increasing AST were related to

Page 43: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

27

replacement of 4-way stop signs with traffic signals, and sidewalk gap closures.140

Generally,

there was evidence of promising yet small effectiveness of interventions. All interventions

evaluated were heterogeneous in nature and it was therefore difficult to determine which aspects

of the interventions were most effective. All studies were also rated weak in the global rating in

the quality assessment. The review emphasized the need for higher quality studies examining

interventions directed at increasing AST, including experimental study designs, appropriate

statistical analysis (taking into account confounders) and reliable and valid data collection

methods.

2.6 Geographic Information Systems (GIS)

This section discusses the use of Geographic Information Systems (GIS) in public health and in

collision research. Studies that have used GIS to examine the spatial distribution of child

pedestrian-motor vehicle collisions and walking to school are reviewed. The benefits of the use

of GIS for this type of research are described.

When studying the influence of the built environment on transportation and health outcomes at a

population level, data are most effectively organized using GIS. GIS are tools that can be used to

organize data from existing sources that incorporate a spatial framework. GIS are commonly

described as “computer information platforms designed to collect, manage, store, and analyze

spatial and non-spatial data.141,142

Although the use of GIS in geographic research is well

established, the use of GIS for public health issues is relatively new.143,144

The development of

desktop GIS software in the last 20 years has enabled health researchers to examine data spatially

and to create maps.143,144

Since then, the use of GIS has proliferated into many areas of public

health worldwide and in Canada. GIS databases are comprised of geometric data (street

addresses, postal codes, cities, coordinates) and attribute data (socioeconomic data, census

data.).143

With these data, GIS can be used to create large datasets based on geography to

identify relationships among variables that influence collisions according to a range of

aggregations, such as census geographic areas or other types of administrative boundaries.145

For

Page 44: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

28

example, child pedestrian-motor vehicle collisions can be mapped onto school attendance

boundaries along with roadway features (Figure 2-6).

GIS and Child Pedestrian-Motor Vehicle Collisions 2.6.1

Child pedestrian-motor vehicle collision studies incorporating GIS methodology have emerged

in the past several years primarily from the geography, engineering and environmental and life

sciences disciplines, with only a few studies from the public health sector. GIS methodology is

particularly useful to study child pedestrian-motor vehicle collisions as these collisions have a

strong geographical component. GIS can be used to describe geographically–based high risk

areas and populations, and to identify potential correlates of collisions in these locations. Many

GIS studies have focused on identifying locations and cluster/hot spots of child pedestrian-motor

vehicle collisions,137,146-148

with several also investigating temporal aspects.137,147,149,150

For

example, one of the earliest studies using GIS to examine child pedestrian-motor vehicle

collisions was conducted by Braddock et al. who mapped police-reported child pedestrian-motor

vehicle collisions and the child’s residence to identify high-occurrence areas in Connecticut.146

They found two high occurrence areas and compared characteristics between the two areas.

Weiner et al. used emergency department and trauma registry to locate and examine child

pedestrian-motor vehicle collisions in Jacksonville, Florida and found a high density of collision

in the urban core of northwest Jacksonville.149

Figure 2-6: Child pedestrian-motor vehicle collisions and roadway design features.

Page 45: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

29

Several studies have investigated child pedestrian-motor vehicle collisions related specifically to

school locations, which are unique in that these locations are characterized by fluctuating periods

of high intensity car and pedestrian traffic.39,147,148,151,152

Warsh et al. used GIS to investigate

child pedestrian-motor vehicle collisions within and outside school zones in Toronto based on

time of day and school months.39

They found that the highest density of collisions occurred

within 150m of schools (Figure 2-7).

In Montreal Canada, Cloutier et al. investigated the association between social and built

environment variables and child pedestrian-motor vehicle collisions near schools, as identified

by the Quebec Automobile Insurance Corporation (SAAQ).151

Positive associations were found

between child pedestrian injury risk and school crossing guards, land use diversity, residential

density, deprivation and child population density. Yiannakoulias et al. used emergency

department surveillance systems in Edmonton Alberta, to identify peak collisions times and

location of high collision incidence, specifically related to school travel.40

GIS methodologies have also been used to examine both environmental and individual correlates

of child pedestrian-motor vehicle collisions.148,150-155

Lightstone et al. used GIS to map child

Figure 2-7: Child pedestrian-vehicular collisions in school zones (permission granted

to reproduce).

.

Page 46: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

30

pedestrian-motor vehicle collisions in Long Beach California and found that collisions were

more frequent in census tracts with higher population density.153

The study also found that

children less than 5 years of age were more likely to be hit at a midblock; whereas, those ages 5-

14 years were more likely to be hit at intersection locations. Intersection collisions were more

likely to occur on major arterials and local streets and further from children’s homes whereas

most midblock collisions occurred within 0.1 miles of the child’s home. Dissanayake et al.

investigated the associations between land use and child pedestrian-motor vehicle collisions in

Newcastle upon Tyne in Great Britain, at the ward administrative level.150

Secondary retail,

educational sites, and primary retail were positively associated; whereas high density residential

areas and junction density were negatively associated with child pedestrian-motor vehicle

collisions. In the Edmonton study by Yiannakoulias et al. correlation analysis was conducted to

investigate the relationship between traffic volume and collisions, which found collisions most

frequently occurred during peak periods of traffic flow and in areas of high traffic volume.40

GIS and Walking to School 2.6.2

Traditionally, environmental correlates of walking in children have been measured in terms of

parent and/or child reports via face-to-face or telephone interviews, paper questionnaires,

telephone survey, or computerized questionnaire.66

Objective methods of built environment

measurement are however, becoming more commonly utilized, including field surveys, pre-

existing data and GIS methodologies.66

In a review of 24 studies on the environmental

determinants of active travel in youth, 11 measured environment variables using objective

measures including field audits and computer mapping, 10 used self-reported assessment of the

environment and 3 used a combination of both.79

Similar to child pedestrian research, studies using GIS methodologies related to active

transportation are not as prominent in the public health literature, but are more common in the

geography, urban and transportation planning fields. Several Canadian studies used GIS to

examine correlates of walking to school. In Toronto Ontario, Mitra, et al. used GIS and spatial

analysis to examine the spatial clustering and temporal patterns of walking to school trips as

reported by adult-proxies for 11-13 year old children in The Transportation Toronto Survey.137

Walking was found to cluster in areas with low household income and within the urban and

Page 47: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

31

inner-suburban Greater Toronto Area. Larsen et al. used GIS to link survey questions regarding

active transportation to school from 11-13 year old students in London Ontario, with databases

containing social and physical environment variables from the City of London Planning

Department, Statistics Canada and environmental audits. Students living within 1 mile from 21

schools were surveyed.156

AST was positively related to shorter trips, being male, presence of

street trees and higher land use mix. Active transportation on the school to home journey was

also associated with lower neighborhood incomes and lower residential densities. In a study by

Gropp et al. individual, school and neighborhood level correlates of walking to school <1 km

buffer surrounding the school were examined using GIS. Age between 11-13 years (versus 14-

15), and urban (versus rural) status were positively correlated with increased walking.157

Objective measures of the built environment have been compared to self-reported measures of

the same features. In a study by Lin and Moudon, 200 objectively measured variables

representing the environment around the residences of 608 participants in King County,

Washington were analyzed using GIS.158

Twelve objectively measured variables were found to

be significantly related to reported walking, of which 3 also had self-reported measured values

from a survey. After re-running the model using the self-reported measured variables, objective

measures of the built environment were found to have stronger associations with walking and

produce stronger models than self-reported methods of the same features.

Although the value of collecting objective information regarding the built environment has been

well demonstrated, parental perceptions of the environment must also be considered given that

parents are the primary decision-makers regarding their children’s travel to school.

Recommendations have been made to incorporate objective measures together with parent and

student perceptions of the built environment when investigating associations with

walking.66,71,85,158

Pont et al. noted that collecting information on both the objective and

perceived environment, their effects on AST can be targeted more effectively.66

The feasibility and value of using GIS to organize and analyze data related to child pedestrian-

motor vehicle collisions and walking promotion research has been shown in the published

literature. Built environment characteristics specific to a setting can be processed and analyzed

using GIS along with collision and walking outcomes. More focus on the use of GIS by those in

Page 48: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

32

the public health fields would be beneficial, as it would help direct interventions and health

policy related to injury prevention, to the specific location where it is most needed.

2.7 The Setting - The City of Toronto

When conducting spatially-based analyses of data, it is important to understand the setting in

which study takes place. This section describes the City of Toronto and the characteristic road

design features in its neighbourhoods.

Toronto is the largest city in Canada and according to the 2011 census is home to 2.6 million

residents. Toronto is the fourth largest city in North America after Mexico City, New York City,

and Los Angeles. The city was first incorporated in

1834, and continued to grow over the next century

through a process of annexation of villages and

neighbourhoods. The most recent of these

annexations occurred in 1998, with the

amalgamation of the regional government of

Metropolitan Toronto with 5 inner ring suburban

municipalities; North York, Scarborough,

Etobicoke, East York and York (Figure 2-8).159

This amalgamation formed a city with

neighbourhoods dating from the 19th

century in an

older urban pre-World War II core, which is

surrounded by post-world War II inner suburbs.160-162

The older core of the city is characterized by straight grid street patterns which allow traffic to

circulate everywhere. An example of the grid-based street network is provided in Figure 2-9. In

these neighbourhoods, elementary schools were planned as central features of a neighbourhood

with residential development occurring within walking distance (e.g. a 400 m radial distance).105

Figure 2-8: Six former municipality

boundaries prior to 1998.

Page 49: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

33

After World War II, there was a move towards implementing the concept of neighbourhood

units, with residential areas surrounding a local school and quiet long winding streets and cul-

de-sacs for use only within the neighbourhood (Figure 2-10).161

Large collector streets and

arterials surrounded the neighbourhoods to allow traffic to move across the city. Reverse-lots

were used in subdivisions where the houses faced in on local streets. The idea was to buffer

residential areas from major roadways. Hess has described how suburban segregated land use

patterns, and street systems with loops and cul-de-sacs increase walking distances between

housing and services, which has a negative impact on the use of walking and cycling for

transport.161

Toronto is among the few North American cities which have vibrant residential areas in all parts

of the city. People not only live in the inner suburbs but also in the downtown core. Toronto

therefore, provides a rich and varied landscape within which to study the effects of environment

on health outcomes.

2.8 Policy

It is important to understand the policy climate related to pedestrian-motor vehicle collisions and

walking to school when conducting research related to safe walking, in order to potentially affect

Figure 2-9: Pre-World War II

grid street patterns in downtown

Toronto.

Figure 2-10: Post-World War II street

patterns in inner suburbs (Scarborough).

Page 50: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

34

policy change. Current policies at the national, provincial and municipal levels are described in

this section.

Child Injury Prevention 2.8.1

2.8.1.1 National

Canada’s vision is to have “the safest roads in the world”.163

This vision was initially developed

by the Canadian Council of Motor Transport Administrators (CCMTA) in 1976, along with other

key Canadian stakeholders. The Road Safety Vision 2001 was the first national road safety

vision, and under this plan fatalities decreased by 10% and serious injuries declined by 16%.163

Under the subsequent Road Safety Vision 2010, fatalities were 6% lower than baseline and

serious injuries were almost 15% lower.163

The next Road Safety Strategy 2015, will determine

success by achieving yearly downward trending in fatalities and serious injuries, and will be rate

–based with the targeting rate of 5 fatalities per 100,000 population.163

The strategy provides

jurisdictions with a framework of key target groups and a framework of best practice initiatives

to adopt or modify.163

Vulnerable road users are one of the key target groups. Eighty nine

initiatives are directed towards vulnerable road users, 11 of which target children.163

Effective

initiatives identified include: education and training, speed reduction (reducing speed limits),

separation of traffic from pedestrians (traffic islands), visibility, traffic calming, safe walking

routes and train crossings.163

Eight of the initiatives specifically mention school zones, and

include: education/training, increased driver penalties, school/parent patrol programs, speed

reader boards, traffic signage for child-friendly routes, and crosswalk treatments.163

Charitable organizations play a large part in influencing policy related to injury prevention in

Canada. Parachute is a national charitable organization, which focuses on preventable injury. It

was formed in 2012 by uniting several injury organizations; Safe Communities Canada, Safe

Kids Canada, SMARTRISK and ThinkFirst Canada. One of Parachute’s mandates is to advocate

for changes to policy, standards and legislation at all levels of government to keep Canadians

safe. Their current pedestrian safety policy focuses on speed reduction stating that vehicular

speed greater than 30-40 km/h presents a greater risk to pedestrians.164

Page 51: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

35

2.8.1.2 Provincial

The province of Ontario has several injury prevention policy initiatives, although none

specifically related to road traffic and pedestrian-motor vehicle collisions. The Ontario Injury

Prevention Strategy was developed in 2007 by The Ministry of Health Promotion in

collaboration with other government ministries and agencies, public health professions and

injury prevention experts.165

In 2009, the Ontario Public Health Association formed a

workgroup to focus on implementing the plan and advocating for coordinated injury prevention

policy at different government levels.166

Public Health Ontario, funded by The Ministry of

Health and Long-Term Care, provides expert scientific and technical advice and support related

to several areas affecting health in Ontario including injury prevention to local public health

units, health care providers and institutions and government.167

Walking to School 2.8.2

2.8.2.1 National

In the US, the Safe Routes to School Program (SRTS) was implemented in 2005 in response to

federal transportation legislation.168

The program dedicated $612 million towards SRTS from

2005 – 2009. As of 2012, the Federal program provided nearly $1.15 billion to states,

benefitting more than 14,000 schools. Funds were used for both infrastructure and non-

infrastructure projects. Evaluation has been conducted both in terms of the impact of its state-

wide policy as well as on a more micro-level in specific locations. An evaluation of STRTS-

related state laws such as minimum bussing distance, sidewalk construction, crossing guards,

traffic control measures and speed zones, by Chirqui et al. found that such laws were associated

with more AST.169

They concluded that the existence of SRTS-related state laws were effective

in reducing barriers to and facilitating AST.

There is no national plan to fund walking promotion programs in Canada. SRTS programs are

conducted at a grassroots/activist level.170

The Canadian Active & Safe Routes to School

Program, is a community-based initiative which is an umbrella group for non-profit community-

based organizations.171

The Public Health Agency of Canada and Transport Canada has

Page 52: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

36

provided some funding to pilot the school travel planning framework across Canada. However,

no national strategy to increase AST exists, and funding has been on a project-by-project basis.

A national charitable organization, Active Healthy Kids Canada, was established in 1994 to

provide information on physical activity among children and youth, and to build better programs,

campaigns and policies.172

The organization released an annual report card on Physical Activity

for Children and Youth, which is a comprehensive assessment of the current state of physical

activity among Canadian children and youth. In the 2013 report card, the federal government

was given a rating of C- due to a lack of progress on public strategies and the lack of a national

physical activity plan.38

The provincial/territorial governments had a C rating, because of

variability in investment and progress on public policy. Finally, non-government institutions

achieved a B + rating, due to their leadership and commitment in developing strategies and

allocating resources; however, there was a need for greater coordination between non-

government and different levels of government to sustain progress.

2.8.2.2 Provincial

At the provincial level, the Ontario government’s Metrolinx Agency initiated a Regional

Transportation Plan in 2008 entitled “The Big Move” in the Greater Toronto and Hamilton area.

This plan intends to spend $200 million over 20 years towards active transportation infrastructure

and research.173

The transportation plan includes the “Stepping It Up” school travel planning

program (STP), which generally consists of education, activities and events, capital

improvement, and enforcement.174

In 2008, an STP pilot conducted in 12 schools in 4

Canadian provinces found modest increases of just over 2% in rates of active transportation.170

It was noted that the focus on interventions was on the lower-cost educational strategies rather

than investment in capital improvement (environmental modifications) and enforcement

strategies to enhance effectiveness in increasing AST rates.170

At the conclusion of the final pilot

in 2012 in the Greater Toronto and Hamilton area, results were modest but promising, with an

overall average decrease in school car trips of 7% in the morning and 3% in the afternoon with a

similar increase in pedestrian trips.174

Other evaluations of STP programs elsewhere have had

mixed results with some finding significant increases in walking to school175

and others finding

Page 53: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

37

no effect.176

Continued evaluation of programs is necessary to determine the reasons for

variability in outcomes, particularly focusing on most effective components of the programs.

2.8.2.3 Municipal

A recent collaboration was formed between Toronto Public Health and the City of Toronto,

Transportation Services, as it was recognized that there was the need for transportation and

public health to work together to create healthy cities. The state of active transportation in

Toronto along with the health benefits and the collision risks were reviewed in a 2012 report

entitled ‘Road to Health: Improving Walking and Cycling in Toronto’.16

Program advisors for

this report included individuals from the Ontario Medication Association, Heart and Stroke

Foundation of Ontario, the YMCA of Greater Toronto, and the Toronto Centre for Active

Transportation. The report described the need to make active transportation infrastructure

funding more of a priority. The report reviewed how to make active transportation safer citing

specifically; reduction in vehicle speed limits, traffic calming,177

separation of traffic,178

re-

allocating space from motor vehicles to active transportation and safer intersections (e.g. traffic

signal phases, physical interventions , roadway and intersection markings). Several methods of

facilitating effective action in Toronto included goal setting, developing plans, policies and

standards, collecting better data and enhancing partnerships across different levels of government

and between public health planners and transportation engineers and city planners.

Although policy initiatives related to pedestrian-motor vehicle collisions and children walking do

exist in Canada, the majority of these initiatives have been developed separately and under

different organizations and levels of government. Policies are needed that simultaneously seek

to decrease pedestrian-motor vehicle collisions and increase walking to school. In the past,

evaluations of policies designed to increase walking to school have focused on walking outcomes

but have not addressed the impact of increased walking on child pedestrian-motor vehicle

collisions. It is essential that implications of the impact of AST promotion programs be properly

understood in terms of this potentially increased risk of road traffic injury. In the U.S., SRTS

programs which traditionally only measured active transportation outcomes, are now required to

be based on a data driven process to reduce fatalities and serious injuries when applying for

Page 54: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

38

federal Highway Safety and Infrastructure Program (HSIP) funding.179

The recent collaboration

between Toronto Public Health and the City of Toronto, Transportation Services, highlights the

need to consider both active transportation and health outcomes such as injury when investing in

interventions and has been identified as a starting point for future action to achieve this goal.

2.9 Gaps in Knowledge Regarding Child Pedestrian-Motor Vehicle

Collisions, Walking to School and the Built Environment

This section summarizes the gaps in knowledge related to child pedestrian-motor vehicle

collisions, walking to school and the built environment and relates them to the thesis objectives.

Walking in children and collision risk have rarely been considered together. Programs and

policies to increase walking to school have focused on walking outcomes, with little

consideration of outcomes related to safety. The influence of the built environment on the

relationship between walking and collision risk is also not well understood. In this thesis,

Objective 1 will use the systematic review process to determine current knowledge and identify

gaps regarding the correlation of the built environment with walking to school and child

pedestrian-motor vehicle collision rates. Objective measures of walking to school have rarely

been used, but rather have relied on parent and child report. This thesis describes a large scale

study conducted to collect objective observational data to better estimate proportions of children

walking in elementary schools in the City of Toronto (Objective 2). Objective measures of the

built environment will be incorporated into the analyses from field surveys and existing

databases using GIS, to determine how built environment features are related to children walking

to school (Objective 3). Child pedestrian-motor vehicle collision rates in the areas surrounding

elementary schools will be estimated using police-reported data (Objective 4). Finally, this

thesis aims to determine the role of specific features of the built environment on the relationship

between child pedestrian-motor vehicle collisions and walking, in order to clarify the built

environment conditions which make walking to school safer (Objective 5). Conclusions will be

drawn based on the findings and recommendations made for appropriate policy action to promote

safe active transportation to school.

Page 55: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

39

3 Walkable but Unsafe? A Systematic Review of Built

Environment Correlates of Walking and Child Pedestrian

Injury

3.1 Preface

This chapter contributes to the overall objective of examining the role of the built environment

on the relationship between walking to school and child pedestrian motor-vehicle collisions, by

providing a synthesis of the literature related to built environment correlates of both walking and

child pedestrian injury together. It identifies an important gap in the literature, in that there are

no studies that consider both children walking and pedestrian injury risk. The primary

contribution of this chapter is to identify features of the built environment that are related to both

increased walking and increased safety for children and emphasizes the need to incorporate

pedestrian safety into walking promotion.

This chapter is reformatted from the following manuscript:

Rothman, L., Macarthur, C., Buliung, R., To, T., & Howard, A. Walkable but unsafe? a

systematic review of built environment correlates of walking and child pedestrian injury. Injury

Prevention, 18 (Suppl 1) 2012: A223-A223.

Reprint rights have been granted to Linda Rothman by BMJ Publishing Group, Ltd.

Page 56: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

40

3.2 Abstract

Objectives 3.2.1

The child active transportation literature has focused on walking, with little attention to risk

associated with increased traffic exposure. This paper reviews the literature related to built

environment correlates of walking and pedestrian injury in children together, to broaden the

current conceptualization of walkability to include injury prevention.

Methods 3.2.2

Two independent searches were conducted focused on walking in children and child pedestrian

injury within nine electronic databases until March, 2012. Studies were included which: 1) were

quantitative 2) set in motorized countries 3) were either urban or suburban 4) investigated

specific built environment risk factors 5) had outcomes of either walking in children and/or child

pedestrian roadway collisions (ages 0-12). Built environment features were categorized

according to those related to density, land use diversity or roadway design. Results were cross-

tabulated to identify how built environment features associate with walking and injury.

Results 3.2.3

Fifty walking and 35 child pedestrian injury studies were identified. Only traffic calming and

presence of playgrounds/recreation areas were consistently associated with more walking and

less pedestrian injury. Several built environment features were associated with more walking,

but with increased injury. Many features had inconsistent results or had not been investigated for

either outcome.

Conclusions 3.2.4

The findings emphasize the importance of incorporating safety into the conversation about

creating more walkable cities.

Page 57: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

41

3.3 Introduction

Child pedestrian injuries are a leading cause of injury-related death for Canadian children

younger than 14 years.10

In children ages 5-9, pedestrian collisions are tied with motor vehicle

collisions as the primary cause of unintentional injury death (18%).10

Every year, approximately

56 child pedestrians die and 780 are hospitalized with serious injuries in Canada.10

While the

burden of child pedestrian injuries and fatalities is high, there has been a decline of over 50% in

Canadian hospitalization and deaths from 1994-2003.10

Declining trends are also evident in the

U.S., Europe and New Zealand.11-15

This reduction may not be due to safer traffic environments,

but rather because children are walking less often, thus reducing the exposure to risk of injury

from collision with a motor vehicle.12,14,34,180

While children’s walking trips to all destinations

have fallen, this decline is most apparent for school trips. From 1986-2006, AST (i.e., walking,

biking) declined from 53%-43% in Canadian children age 11-13 years.37

Declines have also

been noted in the U.S, Great Britain and Australia.34,35,180

Initiatives to increase walking in children have been developed to promote healthy active living

and are focused primarily on school trips.128,181,182

The Safe Routes to School (SRTS) concept

began in Denmark in the 1970s, with programs developing in Europe, Australia, New Zealand,

Canada and the United States.168

In the U.S., a national SRTS program was passed in 2005 as

part of the U.S. federal surface transportation bill with 11,000 schools funded by 2011.117,168,181

In Canada, SRTS programs are conducted at a grassroots/activist level with some pilot funding

from a provincial government agency.170

When planning interventions to increase walking to school, the potential effects of increased

walking exposure on pedestrian injury rates should be considered. Gropp et al. recently found a

dose-response relationship between longer school travel distances and injury related to AST in a

Canadian national survey.44

In Toronto, Canada, almost 50% of child pedestrian collisions were

found to occur during school transportation times and the highest density of collisions occurred

within 150 m of a school.39

Page 58: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

42

Since the 1970s, research from the transportation and urban planning fields has investigated the

‘walkability’ of the environment. More recently, public health researchers have become

interested in the effects of the built environment and walkability on physical activity and obesity.

The definition of walkability is problematic, as it varies by discipline and there is no standard set

of factors describing a walkable environment. Walkability has been defined as, “the extent to

which the built environment supports and encourages walking by providing for pedestrian

comfort and safety…”.183

This conceptual recognition of safety in walking has not been well

addressed in the built environment active travel literature. Focus has been on increasing

walking, with little attention paid to the risks associated with increased traffic exposure. Some

researchers have acknowledged the importance of linking road safety indicators to active school

commuting.65,85,119

Reviews related to walkability and children have investigated the correlates of walking, which

encompass many characteristics of the household, behaviours and material and social

environments.65,71,79,85,117,184

There are few systematic reviews; however, that link features of the

built environment to child pedestrian injury. Wazana et al. found that risk factors for child

pedestrian injury were related to the physical environment.87

Their review was limited to

Medline articles from 1985- 1995. Built environment roadway characteristics have been

statistically linked with child pedestrian injury risk (OR = 2.5); however, effects of specific built

environment features have not been examined.88

The purpose of this review is to use the published literature to develop an understanding of how

specific features of the built environment relate to both walking in elementary school children

and child pedestrian injury to direct further research. As this review incorporated a variety of

papers drawn from a wide array of disciplines that use different reporting standards, traditional

systematic review was challenging. However, The Preferred Reporting Items for Systematic

Reviews and Meta-Analyses (PRISMA) guidelines were adhered to as closely as possible.185

Page 59: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

43

Medline

2010

Embase

2962

Transport

4562

Dissertations

and Theses

2956

SafetyLit

55

Web of

Science

1492

CINHAL

335

Scopus

421

Total articles retrieved

15182

Duplicates removed

12865

Hand search selected journals

Google search

19

PyscINFO

389

Unique articles

12884

Initial Screening

Included

433

Included

195

WalkingChild Pedestrian

Injury

Included

134

Included

107

Screened

reference

lists

26

Screened

reference

lists

6

Full-text retrieved

Eligibility

Assessment (2)

Eligibility

Assessment (1)

Included

160

Included

113

Included

50

Included

35

Full-text

retrieved

Data Extraction

Included 628

Excluded 12, 256

Excluded 78

-Qualitative

-Rural

-Determinants

non-specific

-Age

-Results in

another

publication

Excluded 110

Qualitative

-Rural

-Outcome

non-specific

-Determinants

non-specific

-Age

-Results in

another

publication

3.4 Methods

The search strategy was developed in consultation with a research librarian at the Hospital for

Sick Children, Toronto, Canada. As the research question crossed many disciplines, nine

electronic databases were searched until March 1, 2012: Medline (1980-2012), Embase (1980-

2012), Transport (1980-2012), Dissertations and Theses (1980-2012), Web of Science (1980-

2012), Scopus (2004-2012), PyscInfo (1980-2012), CINAHL (1985-2012) and SafetyLit (1995-

2012). Search strategies were broad, given the variety of discipline-specific terminologies

(Appendix A). Two sets of searches were conducted on each database, one for child pedestrian

injury and another for walking in children. Search results are illustrated using a PRISMA flow

diagram (Figure 3-1).185

Figure 3-1: The Preferred Reporting Items for Systematic Reviews and

Meta-Analyses (PRISMA) flow diagram.

Page 60: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

44

Hand searches were done for references from systematic reviews and select journals between the

years 2006-2011: Accident Analysis and Prevention, Injury Prevention, Traffic Injury

Prevention, Transportation Research part D-Transport and the Environment, and Health and

Place. Google searches were used to identify articles/reports/grey literature from websites using

the search terms: pedestrian, child pedestrian, built environment and pedestrian injury,

walkability, active school transportation and active transportation.

Eligibility 3.4.1

Two reviewers from the research team independently reviewed titles and abstracts initially and

then assessed full-text versions using standardized checklists. There were no language

restrictions. Only literature from highly motorized countries (i.e., Australia, Japan, New

Zealand, North America and Western Europe) was considered. This classification of high

(HMCs) versus low motorized countries (LMCs) was developed by the Transport Research

Laboratory and used by the World Health Organization to describe road fatality trends.28,186

Motorization level is measured by number of motor vehicles/1000 population and LMCs have

typically much less motorization levels (<400 MVC/1000 population) than HMCs.186

Data were

not included from LMCs, as the traffic environment is very different, with less developed traffic

safety structures including traffic separation/calming, safety education and law enforcement.14

Inclusion and exclusion criteria were as follows:

Inclusion criteria:

Outcomes included measures of either one or both of:

-Child pedestrian roadway collisions (incidents or severity). Studies with samples

composed of both child pedestrian and cyclists casualties where proportions of each were

not indicated, were included in the review if located in Australia/New Zealand/North

America or Europe (excluding The Netherlands and Denmark). Nationally, the

proportion of pedestrians is greater than cyclists in those locations except in the

Netherlands and Denmark, where cycling rates are greater than walking rates.187-189

-Walking in children (or inversely, being driven)

Page 61: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

45

Specifically identified built environment features (not summary measures, e.g. ”walkability”

or “perceived unsafe environment” where traffic features were not specifically identified)

Full published papers, online reports, or dissertations

Quantitative analysis with statistical significance testing (bivariate, multivariate), or where

possible to calculate statistical significance

Highly motorized countries

>50 percent sample composed of children ages 4-12 years or separate models for children

Exclusion criteria

Specific disease or at-risk groups (e.g. obesity)

Conference abstracts

Qualitative studies

Focus groups

Literature reviews

Descriptive studies

Stratified outcomes (e.g. comparing midblock versus intersection collisions)

Unspecific outcomes (e.g. physical activity not disaggregated by travel mode)

Discrepancies in eligibility were resolved through inter-reviewer discussion and in consultation

with a third researcher as required.

Data Extraction 3.4.2

Data extracted included; publication year, location, data source year(s), study design, authors’

disciplines, journal, population, ages, outcomes, exposures (irrespective of statistical

significance), measurement techniques, statistical methods and covariate adjustment. Significant

associations with the outcome were identified, where p < .05 for correlations/comparisons of

means, or where 95% CIs of odds ratios OR/relative risks excluded 1.0. Built environment

variables were categorized into the 3 D groups, Density Diversity and Design, using the

framework described by Cervero et al.83

Density variables were related to population density,

diversity variables reflected different land uses and design variables were related to roadway

characteristics.83

Page 62: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

46

Studies were identified which considered demographic and socioeconomic covariates, either by

having restricted samples (e.g. specific age group), stratified analysis, testing for inclusion in

multivariate models or matching in case-control studies. Socioeconomic status (SES) covariates

included: income, home ownership vehicle ownership, one parent families, employment, free

school meals, crowding, public assistance, immigrant, education, dwelling size and overall

deprivation indices. Other demographic covariates included: age, sex, race/ethnicity, and

household size.

Quality Assessment 3.4.3

Quality assessment was conducted using the Epidemiological Appraisal Instrument (EAI).190

The

EAI builds on the Downs and Black checklist commonly used for epidemiological studies.191,192

The EAI revision improves the validity and reliability and adapts the checklist for different study

designs. EAI has 43 items with item-specific instructions that specify applicability of each item

to different study designs. Criterion and construct validity have been tested, with an inter-rater

reliability of 90% and a weighted Kappa in the excellent range (.80-1.00). No EAI summary

scores were calculated, instead, the instrument was used as a descriptive assessment of study

components according to guidelines in the PRISMA statement.185

Analysis 3.4.4

Features with any statistically significant negative associations with incidence of child pedestrian

injury or injury severity were categorized as “Less Injury (Safer)” whereas those with

statistically significant positive associations with injury were categorized “More Injury (Less

Safe)”. Features with statistically significant positive associations with walking were

categorized “More Walking” and those with statistically significant negative associations were

categorized “Less Walking”. Correlates were grouped together where appropriate (e.g.

playgrounds and parks) and were categorised into the 3 D categories. Those with inconsistent

findings or had not been tested for either outcome were identified. Null findings have not been

included in this analysis.

Page 63: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

47

3.5 Results

A total of 12,884 unique papers were originally screened, including 19 unpublished papers found

through Google searches (Figure 3-1). Of these, 12,256 papers were excluded, leaving 628

papers; 433 related to children walking and 195 related to child pedestrian injury. No studies

were identified that addressed both correlates of children walking and pedestrian injury.

Eligibility assessment of titles and abstracts produced 134 walking papers (plus 26 from

reference lists) and 107 child pedestrian injury papers (plus 6 papers from reference lists). Full-

text screening resulted in 50 walking and 35 pedestrian injury papers, from which data were

extracted. This represented 65% of the original articles identified. The average inter-rater

agreement was κ = .80 for pedestrian and κ = .84 for walkability papers.

Walking 3.5.1

The walking papers were from a wide variety of disciplines, with 48 from scientific journals, 1

report and 1 meeting paper (Appendix B). The papers were from: U.S. (22, 44%), Australia/New

Zealand (12, 24%), Canada (6, 12 %), UK (5, 10%), and one each from Germany, Holland,

Norway, Sweden, and Switzerland (1, 2%). No papers were published before 1996 that fulfilled

inclusion criteria, and more than half had been published since 2008.

All studies were observational; 94% (47/50) were cross-sectional, one was longitudinal and 2

were time series studies (Appendix B). The conceptualization of the walking outcome varied

(Column 4, Appendix B) with 47 (94%) of studies measuring walking as a prevalence proportion

and the remaining 3 studies as a rate/week. Forty-two studies (84%) measured walking to

school, whereas the remaining 8 measured walking in general or to leisure activities. Walking

was measured in 29 (58%) papers via parent report (questionnaires, face-to-face interviews and

telephone interviews), 9 (18%) via child report including questionnaires, face-to-face interviews,

hand counts), 6 (12%) via both child/parent report, and 6 (12%) via travel diaries. Built

environment correlates of walking were measured by routinely collected administrative data

processed using GIS (23, 46%), and parent reports (22, 44%). Several studies used field surveys

Page 64: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

48

(4, 8%) or child/parent reports (4, 8 %). Most studies (41/50=82%) accounted for both SES and

other demographic covariates.

Child Pedestrian Injury 3.5.2

Child pedestrian injury papers were drawn primarily from health-related fields, and then

transportation research/planning, civil engineering, urban planning and geography (Appendix C).

Most were published scientific papers (33, 94%) from: U.S. (12, 32%), Australia/New Zealand

(7, 18.9%), Canada (9, 24%), the UK (7, 19%), Ireland (1, 3%) and Germany (1, 3%). Fifty

percent of the papers were published between 1990 and 1998. Only seven (19%) were published

since 2008.

All child pedestrian injury studies were observational, with the majority using retrospective data

situated within discrete time intervals, referred to as a cross-sectional retrospective design

(21/35, 60%, Appendix C). There were 10 case-control (29%), 4 time series (11%), and 1 case-

crossover study (1, 3%). Eighteen were ecological studies (51%). Injury was conceptualized

either as an injury or fatality incident, or injury severity. Thirty-four of the cross-sectional

studies measured injury rates per spatial unit or per population (92%), with the remaining 2

studies measuring incident cases. All case control, time series and case-crossover studies

measured incident injury cases. Child pedestrian injuries were measured using police-report

databases (54.1%), hospital surveillance (13, 35.1%), coroner surveillance (4, 10.8%), police

surveillance, trauma databases and other databases (each 3, 8%). Built environment correlates of

child pedestrian injury were measured using databases/GIS (26, 70.3%), field surveys (10,

27.0%), child/parent reports (3, 8.1%) and parent report (2, 5.4%). Thirteen of the 35 child

pedestrian injury papers (37%) controlled for both SES and other demographic variables,

whereas fourteen controlled for one or the other (40%). Eight studies did not control for SES or

demographic these variables (23%).

Page 65: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

49

Quality Assessment 3.5.3

3.5.3.1 Walking

Over 85% of walking studies had clearly stated hypotheses/objectives, described outcomes and

statistical methods, reported main findings, provided estimates of statistical parameters, and

adequately adjusted for covariates/confounders (Table 3-1). Less than 50% explicitly stated the

study design, however, the design of 52% could easily be inferred (“Partial”). Most studies

lacked clear identification of exposure (84%), as typically these papers investigated a group of

covariates.

Assessing external validity was problematic. Sixty eight percent of studies reported participation

rates; participant characteristics were described/partially described in 70% and only 8%

accounted for subject loss/unavailable records (Table 3-1). No sample size calculations were

evident in any of the studies. Only 12 (24%) provided exposure measurement reliability and 9

(18%) provided validity information. Only 8 (16%) provided information regarding reliability of

outcome measures and 5 (10%) reported validity information.

3.5.3.2 Child Pedestrian Injury

Over 90% of cross sectional/longitudinal/time series injury studies had clearly reported

hypothesis/objectives, outcome population source/sampling frame, eligibility criteria, statistical

methods used, and main findings (Table 3-1). Only 20% of studies had explicitly stated the

study design; the design of 76% was implicit or stated in the abstract (“Partial”). Most of these

studies did not identify an exposure (68%). There were no sample size calculations or reported

exposure or outcome reliability and validity. Only 8 (32%) studies performed adequate

individual covariate adjustment, but 17 (68%) performed adequate environmental adjustment.

Estimates of random variability were missing in over 60% of papers.

Page 66: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

50

Table 3-1: Quality assessment using EAI: Number of studies (%).

Walking CS

Injury CS/Long/TS

(n = 25)

Injury CC/CCROSS

( n = 11)

No

(0)

Partial

(1)

Yes

(2)

N/A

/UTD No (0)

Partial

(1)

Yes

(2)

N/A/

UTD

No

(0)

Partial

(1)

Yes

(2)

N/A/

UTD

1. Description

Hypothesis/ Objective(s)

1(2)

4(8)

45(90)

0

0(0)

1(4)

24(96)

0

0

3(27)

8(73)

0

Exposures clear 0 1(2) 7(14) 42(84) 0(0) 0(0) 8(32) 17(68) 0 0 2(18) 9(82)

Outcome clear 0 4(8) 46(92) 0 0(0) 0(0) 25(100) 0 0 0 11(100) 0

Study design * 0 24(48) 26(52) 0 1(4) 19(76) 5(20) 0 1(9) 2(18) 8(73) 0

Population source/

sampling frame

2(4) 7(14) 41(82) 0 0(0) 0(0) 25(100) 0 0 0 11(100) 0

Eligibility criteria 5(10) 10(20) 35(70) 0 0(0) 0(0) 25(100) 0 0 0 11(100) 0

Participation rate/record

availability

15(30) 1(2) 34(68) 0 0(0) 0(0) 25(100) 0 0 0 6(55) 5(45)

Participant characteristics 12(24) 0 37(74) 1(2) 6(24) 2(8) 17(68) 0 0 0 11(100 0

Non-participant

characteristics

42(84) 0 8(16) 0 0 0 0 25(100) 0 2(18) 0 9(82)

Covariates-

individual variables

1(2) 5(10) 40(80) 0 0(0) 0(0) 8(32) 17(68) 1(9) 1(9) 9(82) 0

Covariates –environment

Variables

1(2) 2(4) 40(80) 7(14) 0(0) 4(16) 14(56) 7(28) 2(18) 0 9(82) 0

Statistical methods 1(2) 3(6) 46(92) 0 1(4) 0(0) 24(96) 0 0 2(18) 9(82) 0

Main findings 0 1(2) 49(98) 0 0(0) 0(0) 25

(100)

0 0 0 11(100) 0

Estimates of random

variability

12(24) 0 38(76) 0 16(64) 0(0) 9(36) 0 2(18) 0 9(82) 0

Estimates of statistical

parameters

0 0 50(100 0 2(8) 0(0) 23(92) 0 0 0 11(100) 0

Sample size calculations 49(98) 0 0 0 0 0 25(100) 0 0 2(18) 9(82)

2. Methods Quality

2.1 Subject Selection

Controls comparable to cases 0 0 11(100) 0

Participation rate/record

availability adequacy

14(28) 15(30) 6(12) 15(30) 0 0 0 25(100) 0 1(9) 4(36) 6(55)

Cases/controls recruited over

same period of time 0 0 6(55) 5(45)

Newly incident cases accounted

for 0 0 0 11(100)

Subject losses/unavailable

records

42(84) 2(4) 2(4) 4(8) 0 0 1(4) 24(96) 0 0 0 11(100)

2.2 Measurement

Exposure/

covariate measure reliability

2(4) 9(18) 3(6) 35(70) 0 0 0 25(100) 0 0 0 11(100)

Exposure/

covariate measure validity

0 7(14) 2(4) 41(82) 0 0 0 25(100) 0 0 0 11(100)

Page 67: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

51

Exposure assessment similar for

case/controls 0 0 11(100) 0

Exposure prior to disease 0 0 0 11(100)

Observers blinding 0 0 2(18) 9(82)

Subject blinding 0 0 0 11(100)

Outcome measures reliable 1(2) 4(8) 4(8) 41(82) 0 0 0 25(100) 0 0 0 11(100)

Outcome measures valid 0 5(10) 0 45(90) 1(4) 0 0 24(96) 0 0 0 11(100)

Standard methods of assessing

outcome both case/control 0 0 11(100) 0

Observations over same time for

case/controls 0 0 4 7(64)

2.3 Data Analysis

Prior history of disease

accounted for

0 0 0 11(100)

Adequate covariate adjustment

-individual

0 2(4) 44(88) 4(8) 0 0 8(32) 17(68) 0 1(9) 9(82) 1(9)

Adequate covariate adjustment

–environment

0 0 42(84) 8(16) 0 0 17(68) 8(32) 2(18) 0 9(82) 0

Time between exposure &

outcome same cases/controls 0 0 0 11(100)

Data reported by subgroups 15(30) 0 35(70) 0 7(28) 0 18(72) 0 3(27) 0 8(73) 0

2.4 Generalization

To eligible population

13(26) 16(32) 6(12) 15(30) 0 0 0 25(100) 0 1(4) 4(36) 6(55)

To other relevant populations 13(26) 16(32) 6(12) 15(30) 0 0 0 25(100) 0 1(4) 4(36) 6(55)

CS = Cross Sectional, Long = Longitudinal, TS = Time Series, CC = Case Control, CCROSS = Case Crossover

N/A/UTD= Not applicable/Unable to determine Shading = not applicable for study design

All case control/crossover studies had clearly described outcomes, population sources, eligibility

criteria, participant characteristics, main findings, estimates of statistical parameters, and

measured exposure and outcome similarly for cases and/or controls. In approximately 25% of

papers, the hypothesis and objectives were unclear and lacked a clear statement of study design.

Over 80% of papers provided estimates of random variability, and statistical methods were

described clearly with confounders accounted for. Details regarding participation rates were

frequently lacking. Only 2 (18%) of studies described observer blinding, and only 55%

described the timing of case/control recruitment. There were no reports of covariate/outcome

reliability or validity.

Page 68: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

52

DESIGN

•Traffic calming (e.g. roundabouts,

speed humps)

DIVERSITY (LAND USE)

•Playground, recreation/park, open

space

DESIGN

•Higher road density/length

•Crosswalks

DENSITY

•Higher child/all pedestrian volume density

•Higher child/all population density

DIVERSITY (LAND USE)

•Land use mix

•Proximity to services/facilities

•Schools

•Urban

DIVERSITY (LAND USE)

•Home/school greater distance from

city centre

DESIGN

•Higher traffic speed/posted speed

•Crossing busy/major roads

+ Collision Incidence/Severity

(Less Safe)

Walking

- Collision Incidence/Severity

(Safer)

+

-

Safety and Walking 3.5.4

3.5.4.1 Less Injury (Safer) and Walking Correlates

Correlates with significant associations with both increased walking and decreased injury were

Design and Diversity features; traffic calming (e.g. roundabouts, speed humps), and proximity

to/presence of playgrounds/recreation areas/parks/open space (Figure 3-2). Indicators of overall

traffic calming measures were generally examined with only 2 papers focused on specific

features; roundabouts and speed humps.100,121,130,193,194

The positive association between

recreation areas/playgrounds and safe walking stresses the importance of incorporating these

land use features and facilities into neighbourhoods.

3.5.4.2 More Injury (Less Safe) and Walking Correlates

Correlates associated with a less safe traffic environment and with increased walking were from

each of the 3D’s. Design features were: higher road density/length, and numbers of crosswalks.

Density features were: higher pedestrian volume and higher population density. Diversity

Figure 3-2: Correlates of walking and child pedestrian injury.

Page 69: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

53

features included: number of schools, land use mix and proximity of services (Figure 3-2).

These environmental features may be indicative of locations of greater exposure to traffic, where

there are more child pedestrians/vehicles and higher traffic speeds. Features such as crosswalks

may act as confounders related to increased exposure (as their presence may indicate more

children walking), and/or to increased child pedestrian injury outcomes (due to inadequate

design/use). Marked crosswalks have been associated with an unadjusted twofold elevation of

risk of child pedestrian injury.195

3.5.4.3 Inconsistent/Untested Correlates of Injury and Walking

Many correlates had either positive or negative associations with injury but had inconsistent

results or were not tested for walking. Traffic control mechanisms such as lights, were protective

against injury but the relationship with walking was inconsistent.108,112,121,128,196

Sidewalks were

associated with increased child pedestrian injury but the relationship with walking was

inconsistent. Sidewalks, similar to crosswalks, might be a confounder related to both increased

exposure and to increased child pedestrian injury outcomes. Sidewalks were however,

associated with injury after controlling for vehicle speed and volume.195,197-199

Children may treat

sidewalks as extended play areas, or they may be more cautious when walking along roads where

sidewalks are absent.198

One-way streets and school crossing guards were associated with

increased child pedestrian injury, but had not been tested for walking. Wazana et al. found that

child pedestrian injuries occurred 2.5 times more frequently per kilometer of one-way street than

on two-way streets. More children walking might explain this effect, less driver attention or

children not looking for traffic in the appropriate direction.200

Cloutier et al. found the numbers

of crossing guards was positively related to pedestrian risk around schools.151

This may be a

result of crossing guards being placed in locations that are particularly dangerous for child

pedestrians, with their safety effect not being enough to overcome the excess danger.151

Other factors such as street parking were related to more walking, but had an inconsistent

relationship with injury. Street parking slows traffic down and provides a barrier between

vehicles and pedestrians.201,202

However, street parking can also be a visual obstruction for

pedestrian crossings and contributes to traffic congestion, which may increase potential for

Page 70: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

54

collisions.203,204

Other correlates of walking related to the design of the built environment which

warrant further investigation into their association with injury are the presence of trails,

perceptions of safety, cul-de sacs and dead ends, public transit, and walking network

connectivity.

3.6 Discussion

Traffic calming devices and the presence of playgrounds and recreation areas were the only

factors consistently associated with more walking and less injury. Higher pedestrian volume,

population and road density, schools, urban location, land use mix, proximity to

services/facilities and crosswalks were associated with more walking, but with less safety. The

majority of built environment factors either had inconsistent associations with either walking or

injury, or had not been tested for either one of the outcomes. Quality assessment, using a valid

and reliable tool, revealed noteworthy inconsistencies of method, analysis and reporting in both

the walking and the child pedestrian injury literature.

Many of the built environment correlates associated with pedestrian injury, such as higher

pedestrian volume, urban environment and schools, may not be inherently dangerous, but rather

may be markers for increased exposure to traffic in general and higher speed traffic in certain

environments. The World Health Organization has identified speed as the principal risk factor

for pedestrian-motor vehicle collisions and fatality.28

Well designed studies are needed to study

the effects of interventions directed towards separating child pedestrians from high speed traffic

by space (e.g. playgrounds), by time (traffic lights), and the optimal use of traffic calming

measures to slow vehicle down in areas where there are many child pedestrians.

Several potential explanations exist for inconsistent findings. There was age heterogeneity

across studies, outcome variability by destination, and differences in conceptualization of

outcomes. Study locations were varied, and it has been noted that associations between built

environment features and school travel-mode choice can vary across studies set in different

locations.205

A variety of methods of both outcome and exposure measurement were used, each

Page 71: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

55

with different biases, and few studies reported measurement reliability and validity. More

objective measures of walking should be explored such as using observational counts. Although

databases may be a more objective source of built environment data, there are limitations

depending on how and when the data were collected. When trying to explain walking behavior

in children, it may be most appropriate to model parent or child perceptions of the built

environment, as ultimately, parents and children make the decision of whether or not the child

walks to school.78,108,120

However, child and parent perceptions of the traffic environment differ

and these likely differ from perceptions of a trained individual conducting a field survey.108,206,207

Several studies recommended that objective measurements be combined with perceptions of the

environment.66,85

Correlates of AST have been examined using both objective GIS

measurements and parental perceptions of the built environment, however, only one study,

directly compared them using a GIS-based overall walkability index.108,117,120

Another study

found significant correlations between parent perceptions of overall walkability with walkability

assessed via field audit.208

Limitations of this review included a focus on quantitative studies, which may have resulted in

omission of environmental features. Qualitative work in this exists; however, the majority of the

published work has used quantitative methods. Also, all walking outcomes were considered, but

it is possible that built environment correlates could vary by destination. However, the majority

of papers studied walking to school (84%). All significant correlates were included irrespective

of whether there was control for confounding. The majority of the papers reviewed did control

for SES and demographic factors (75/85= 88%), but other factors such as vehicle ownership,

weather and crime were not consistently controlled for in many of the studies. Meta-analysis

was not conducted due to the heterogeneity of analytic techniques. This is problematic with

meta-analysis of observational studies in general, as the diversity of study designs and

populations result in summary statistics that are difficult to formulate and interpret.209

The

purpose of this review was not however, to assess the magnitude of effects but rather to develop

a list of correlates identified in different bodies of literature, associated with more walking and

less child pedestrian injuries.

Publication bias was difficult to assess as studies were observational and no registry exists.

However, no language restrictions were imposed and Internet searches were conducted to include

unpublished reports and articles. Although difficult to test empirically, studies missing due to

Page 72: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

56

non-significant findings were not likely, as studies of built environment features incorporated

many different exposures, of which there was generally always at least one significant

association. Studies reported both non-significant and significant associations for all exposures

of interest. The analysis also indicated “any” association with the outcome, and did not quantify

the association or the number of studies where significant associations were found. Therefore,

any omission of studies with completely null results, would not have affected the findings. The

PRISMA guidelines state that if publication bias exists, smaller studies would show larger

estimates of the effects of the intervention.185

Comparison of effect size by study size was not

possible using for example, funnel plots, as consistent measurements of exposures and outcomes

between studies were lacking.

The majority of studies reviewed were cross-sectional, and well-designed controlled studies that

examine the built environment, walking and child pedestrian injury were lacking. Therefore,

inferences could only be made regarding associations and not causality. Randomized trials are

difficult to design and implement for traffic interventions due to issues including high costs and

lack of denominator exposure data. Other more feasible study design options should continue to

be explored. Several of the pedestrian injury studies used case-control and case-crossover

methods, which with further refinement, would have better validity than cross-sectional

studies.193,195,197,203,210

Quasi-experimental designs are also feasible and produce more valid

results when studying traffic injury and could be extended to studying walking. These studies

are needed to investigate the effectiveness of design features such as specific traffic calming

devices, crosswalks and sidewalks for children. Further work is also required for features with

either inconsistent associations, or had not been studied for the walking or pedestrian outcome.

The feasibility of modifying built environment features and the time-frame required are

important to consider when designing traffic intervention studies.87

Diversity and density

features may be less easily modified in existing neighbourhoods, but should be considered when

planning new neighbourhoods. Targeted interventions addressed at road environment design

features, such as traffic calming, may be more feasible in an established neighbourhood,

compared to dealing with the more general issues related to higher population densities and land

use mix.

Page 73: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

57

The results of the current review will be disseminated to the City of Toronto, Transportation

Services Department, and to the Green Communities Canada, SRTS Program, with whom

working relationships are already established. Transportation Services is responsible for

designing traffic environments and also conduct retrospective evaluation of the effectiveness of

these environments in terms of injury and pedestrian activity. There must be close coordination

between scientific investigators and traffic planners to implement well-designed prospective

traffic intervention studies. Green Communities Canada currently conducts and evaluates

programs to increase walking to school in Canada.170,171

Researchers and staff can work together

to inform future SRTS evaluations to incorporate injury prevention.

3.7 Conclusions

This review described the current knowledge regarding the relationship between specific built

environment features and both walking and child pedestrian injury. Built environment features

that either slow traffic down (traffic calming) or separate children in space from traffic

(playgrounds), were associated with both increased walking and less pedestrian injury. Many

built environment factors associated with more walking were also associated with greater risk of

injury. Walkability assessment and evaluations of walking promotion interventions should

include a pedestrian injury component to ensure that increased walking does not have detrimental

effects on child pedestrian safety. Likewise, evaluation of traffic safety interventions should also

address the effectiveness of the intervention in promoting walking. An interdisciplinary

approach, including city planners, community organizations and health and planning scholars, is

essential to evaluate and design appropriate interventions to increase walking while ensuring

safety.

Page 74: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

58

3.8 Supplementary/Supporting Analysis

The following analysis was not included in the published manuscript but supports the study

findings.

Walking to School 3.8.1

Built environment correlates were also compiled just for walking to school and excluding other

destinations. There were only two changes where correlates fell within Table 3-S1 when the

destination was limited to school travel. Traffic control devices were associated with more

walking to school (as opposed to inconsistent association with walking to all destinations) and

less injury (i.e., cell 1). Road classification was associated with less walking to school (as

opposed to inconsistent association with walking). Also included in this table were built

environment features with inconsistent results or where they hadn’t been tested for either

walking to school or injury.

Page 75: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

59

3.9 Supplementary Tables

Table 3-S1: Correlates of walking to school and child pedestrian injury.

Injury Incidence/Severity (Safer) Inconsistent Injury Results Not Tested for Injury Injury Incidence/Severity (Less

Safe)

More Walking Traffic calming (e.g. roundabouts,

speed humps)

Playground, recreation/park, open

space

Traffic control (lights or crossing)

Street parking Perceptions of safe

crossing

School size

Sidewalk quality

Windows facing street

Trails

Higher child/all pedestrian volume

density

Higher child/all population density

Schools

Urban

Higher road density/length

Land use mix

Proximity to services/facilities

Crosswalks

Inconsistent

Walking

Results

Higher residential/housing

density/land use

Parking lot

Traffic concerns

Street connectivity

/Route directness

Commercial/retail

intersection/block density

Higher traffic volume

Sidewalks

Not Tested for

Walking Domestic yard

Newer housing

Other people observed crossing the

road at intersections/crosswalks

Higher traffic volume in front of

house-younger children

One way streets

Midblock/uncontrolled midblock

School crossing guard

Visual obstacles

Community buildings

Industrial

Multi-family dwellings

Lives on a through street

Less Walking Home/school greater distance from

city centre

Higher road class en route/in

area (DB)

Altitude

Cul-de-sac

Dead end density

Destination distance

Public transit

School parking issue

Steep hills

Vacant lot

Walking network

connectivity

Higher traffic speed/posted speed

Crossing busy/major

roads

Page 76: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

60

4 Influence of Social and Built Environment Features on

Children’s Walking to School: An Observational Study

4.1 Preface

This chapter contributes to the overall objective of examining the role of the built environment

on the relationship between walking to school and child pedestrian injury, by examining the

influence of social and built environment features on children’s walking to school using

observational outcomes. Results are compared with previous studies which have used reported

walking outcomes. This chapter also provides supplementary detail regarding the methods in

terms data sources and analytic strategies used for both Chapters 4 and 5. The primary

contribution of this chapter is that it is the first large study to correlate direct observational

counts of walking to school with objective built environment data from city databases and field

surveys.

This chapter is reformatted from the following manuscript:

Rothman L, To T, Buliung R, Macarthur C, Howard A. Influence of social and built

environment features on children’s walking to school: an observational study. Prev Med. 60;

2014:10-15.

This article is open access and reprint rights have been granted to Linda Rothman by Elsevier.

Page 77: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

61

4.2 Abstract

Objectives 4.2.1

To estimate the proportion of children living within walking distance who walk to school in

Toronto, Canada and identify built and social environmental correlates of walking.

Methods 4.2.2

Observational counts of school travel mode were done in 2011, at 118 elementary schools. Built

environment data were obtained from municipal sources and school field audits and mapped onto

school attendance boundaries. The influence of social and built environmental features on

walking counts was analyzed using negative binomial regression.

Results 4.2.3

The mean proportion observed walking was 67% (Standard Deviation = 14.0). Child population

(Incidence Rate Ratio (IRR) 1.36), pedestrian crossover (IRR 1.32), traffic light (IRR 1.19), and

intersection densities (IRR 1.03), school crossing guard (IRR 1.14) and primary language other

than English (IRR 1.20), were positively correlated with walking. Crossing guard presence

reduced the influence of other features on walking.

Conclusions 4.2.4

This is the first large observational study examining school travel mode and the environment.

Walking proportions were higher than previously reported in Toronto, with large variability.

Associations between population density and several roadway design features and walking were

confirmed. School crossing guards may override the influence of roadway features on walking.

Results have important implications for policies regarding walking promotion.

Page 78: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

62

4.3 Introduction

The effect of the built environment on physical activity is a topical issue in public health.119

Interventions directed at the “walkability” of the built environment have been promoted to

encourage healthy active living. Walkability is a complex concept, and definitions are varied as

are approaches to operationalizing the concept using modeling techniques. The concept of

walkability will continue to be context-specific until there is a validated and consistent list of

environmental correlates of walking.

Many studies have examined the correlates of adult walking, with some consensus that adult

walking is related to density, mixed land use, pedestrian infrastructure (e.g. sidewalks,

crosswalks) high connectivity (grid network, short block lengths, many intersections, few cul-

de-sacs/dead ends) and accessibility to multiple destinations.67,118,119

Walkability studies for

elementary school children generally focus on walking to school, which has consistently been

negatively associated with distance. 65,66,85

and positively associated with population

density.35,85,120,122-125

Associations with land use, pedestrian infrastructure and connectivity have

been inconsistent and often contradictory to findings in adult studies.66,85

Environmental features

correlated with adult walking may be different than those for children because of differing

destinations and purposes for walking.

Varied methods of measurement for both built environment and walking outcomes may

contribute to inconsistent results.65-67,72,85

Walking outcome has generally been measured

through parent/child report using different outcome definitions (e.g. usual trip, trip per/week),

time frames, and targeted age ranges. To date, only one study incorporated direct observational

counts of children walking to school; however, that study was limited by small sample size and

little geographic diversity.72

The purpose of this study was to 1) estimate the proportion of children living within walking

distance to school who walk to school in a Canadian city and 2) correlate built and social

environment features (with a focus on roadway design), with observational counts of children

walking to school.

Page 79: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

63

4.4 Methods

Study Design, Setting and Population 4.4.1

A prospective observational study was conducted in the spring, 2011, involving junior

kindergarten (JK) to grade 6 elementary schools in Toronto, Canada. Toronto consists of an

older urban core characterized by pre-World War II traditional neighbourhoods, and 5 inner

suburb municipalities, representing newer, car-oriented post-World War II neighbourhoods.159

Exclusion criteria were schools with 1) other grade combinations 2) special programs, which

accept children from outside the school’s attendance boundaries (e.g. French immersion) and 3)

involvement in other walking studies. Children arriving by school bus were excluded as they

don’t live within walking distance to the school. The Toronto District School Board’s (TDSB)

transportation policy states that children grades JK-5 who live >1.6 km and those grades 5+ who

live > 3.2 km from their school are eligible for school bus transportation.211

Ethics approval was

obtained from the Hospital for Sick Children Research Ethics Board and the TDSB.

Outcome Variable 4.4.2

Trained observers counted children either arriving to school walking, by other active means (i.e.,

bicycle or scooter) or by private motorized vehicles. Observations were repeated at 10% of the

schools, one week apart to determine test-retest reliability. The proportion of children walking to

school was calculated from the total number of children observed and excluded those arriving by

school bus.

Independent Variables 4.4.3

Built environment features were identified from a literature review. All variables were mapped

onto school attendance boundaries provided by the TDSB. Features were classified according to

Cervero and Kockelman’s 3D’s; Density, Diversity and Design; originally developed to study

Page 80: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

64

adult walking behaviour but which has since been applied to children’s school transport.83,85,86

Focus of the analysis was on roadway design features, as these are most feasible to change in

existing neighbourhoods compared to those related to density and diversity. Table 4-1 presents

the variables considered for multivariate modeling.

4.4.3.1 Built Environment

4.4.3.1.1 Density

Population density variables were obtained from the 2006 Canadian census by Dissemination

Area (DA). DAs are the smallest standard geographic area for which all census data are

disseminated with approximately 400-700 residents. DAs were mapped onto school boundaries

and area-weighted proportionate analysis was used to estimate the census variables for each

boundary.123,212

4.4.3.1.2 Diversity

Diversity variables reflect different land uses. Recreational facilities and parks data were

obtained from the City of Toronto and parcel level data by land use category was obtained from

The Municipal Property Assessment Corporation (MPAC). Individual land uses were calculated

as percentage of the school boundary. The mix of residential, commercial, industrial,

institutional, and vacant land use (including parks and walkways) within school boundaries was

measured using an entropy index:

Land use mix = Ʃu (pu x ln (pu)/ln n

u = land use classification, p = proportion with specific land use, n = total number

classifications. Scores of 0 = single land use, 1 = equal distribution of all classifications.53,213

Page 81: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

65

4.4.3.1.3 Design

Roadway design variables were obtained at the school level from school site audits conducted by

two trained observers. Presence of adult school guards employed by Toronto Police Services

was recorded. Vehicle speed and volume were measured using manual short-based methods by a

third observer along a roadway within 150m of the school.214,215

Design variables at the school boundary level were obtained from the City of Toronto and

densities were calculated per school boundary area or linear km of roadway. The school was

designated urban if over 50% of the attendance boundary fell within the inner urban area.

4.4.3.2 Social Environment

Student socioeconomic status (SES) was measured using the TDSB’s learning opportunities

index (LOI) which is a composite index including parental education, income, housing and

immigration.216

Scores range from 0-1, with 1 indicating lower SES. The proportion of

households in the school’s DA which fell below after tax, low income cut-offs (ATLICO) was

obtained from the Canadian census as a measure of the SES of the area surrounding the school.

The low income cut-off is an income threshold below which a family devotes a larger share of its

income than the average family, on necessities i.e., food, shelter and clothing.217

The proportion

of children at the school whose primary language was other than English was included as

provided on the TDSB website.

Statistical Analysis 4.4.4

The unit of analysis was the school attendance boundaries, with all features processed and

mapped onto boundaries using ArcMap (ArcMap, version 10). Road network distance buffers

were created around the schools to assess the proportion of roadways within the boundaries

within 1.6 km walking distance of the school.

Page 82: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

66

Statistical analysis was conducted using SAS (SAS, version 9.3). Multicollinearity of variables

was identified by Variance Inflation Factors (VIF) >10. Where pairs of variables were highly

correlated, the variable with the higher standardized unadjusted beta coefficient was retained.

Descriptive statistics were calculated for all independent variables. Mean values and standard

deviations were calculated for continuous variables, and numbers with percentages were

calculated for dichotomous variables.

The proportion of children walking to school was modelled as the dependent variable using

negative binomial regression due to over dispersion of the count data. Features with p < 0.2 in

the unadjusted analysis were included in a forward manual stepwise regression with the entry

order determined by the magnitude of standardized betas. A p value < 0.2 in the unadjusted

analysis was used to screen for inclusion in the multivariate models, as using lower p values may

miss important correlates once other variables are taken into account.218

At each stage of the

modeling, the variables included were re-examined and dropped if not significantly related to the

outcome.219

Model fit was assessed using the Akaike Information Criteria (AIC).220

Poor

weather during observations was retained in the model regardless of significance level. As there

were 42 potential independent variables, a Bonferroni adjusted significance level of <.001

(.05/42), was used.

Stratified analyses by tertiles were conducted for design features to assess for differential impact

on walking outcome. Results of the negative binomial models were presented as incident rate

ratios (IRR) with 95% confidence intervals (CI). Pearson product-moment correlation

coefficients were used to determine test-retest reliability.

4.5 Results

Of 436 elementary schools, 318 schools were excluded, primarily due to ineligible grade

combinations (Figure 4-1). The analysis included 118 schools. The mean observed walking

proportion was 67% (range = 28- 98, standard deviation (SD) = 14.5). Only 1.7% arrived using

other active transportation. High test-retest reliability was noted in 10% (n = 12) of the schools

Page 83: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

67

(Pearson’s r = .96). School attendance boundaries were small, with 75% having an area less

than 1.3 km2. The mean proportion of roads within the boundaries and within 1.6 km of the

school along the road network was 95% (SD .10). A total of 34,099 students lived within the

attendance boundaries, and of these, only 424 who attended regular programs, lived >1.6 km

from the school and traveled by school bus. The descriptive statistics of all variables considered

for multivariate modeling are provided in Table 4-1.

Several built environment design variables had very low densities (i.e., less than .1/km roads),

including flashing lights, minor roads, one way streets, missing sidewalks and traffic calming.

Variables associated with the walking to school in the unadjusted analyses are presented in Table

4-2. Densities of old housing, multi-family dwellings, male children, residential land use, roads

and local roads were dropped from further analyses because of multicollinearity. The final main

effects multivariable model indicated significant positive associations between walking to school

and density and design built environment variables (Table 4-3). Child population (IRR=1.36,

95% CI= 1.21, 1.53) , pedestrian crossovers (IRR =1.32, 95% CI=1.01, 1.72), traffic lights

IRR=1.19, 95% CI=1.07, 1.32), and intersection densities (IRR=1.03, 95% CI= 1.01, 1.05),

presence of a school crossing guard (IRR=1.14, 95% CI=1.07, 1.21) and primary language other

than English (IRR=1.20, 95% CI=1.05, 1.36) were associated with more walking. Child

population density, traffic lights and school crossing guards exhibited the most significant

associations.

Figure 4-1: Flowchart of school participation.

Page 84: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

68

Table 4-1: Descriptive statistics of candidate variables for multivariate modeling.

Variable Description Mean (SD)/

N (%)

Outcome

Proportion walking to schoola,1

67.3% (14.50)

NATURAL ENVIRONMENT

Poor weather (rain or cold) c,5

35 (29.66)

BUILT ENVIRONMENT

Density

School boundary level Child population (#)/1000m

2 b,2

0.54 (0.26)

Total population (#)/1000m2 b,2

6.09 (3.57)

Multi-dwelling (apartments, duplexes) (#)/1000m2 b,2

1.43 (1.30)

Diversity

School boundary level Recreational facilities (#)/km

2 b,3

1.78 (1.58)

Park land area/boundarya,3

7.60 % (6.85)

Entropy (mixed land use) b,4

0.61 (0.13)

Commercial land use area/boundary a,4

6.49 % (7.32)

Residential land use area/boundary a,4

44.2% (2.23)

Industrial land use area/boundarya,4

6.31% (8.96)

Institutional land use area/boundarya,4

8.65% (6.67)

Vacant land area/boundarya,4

8.75% (7.03)

Design

School level

School crossing guard observedc,5

45 (38.14%)

Cars appear to be driving fast near schoolc,5

56 (47.46%)

Traffic congestion seen around school during drop off c,5

76 (64.41%)

Dangerous midblock crossing near school c,5

70 (59.32%)

Dangerous intersection near school c,5

40 (33.9%)

Drop offs opposite side of road c,5

83 (70.30%)

Double parkingc,5

54 (45.80%)

Cars blocking viewc,5

73 (61.90%)

Mean speed > 5 km over speed limitc,5

16 (13.56%)

School traffic/minute b,5

2.14 (1.00)

School boundary level

Other schools within school boundary (#) c,6,7

39 (33.1)

Old houses (pre 1946) (#)/1000m2 b,2

0.57 (0.82)

Collector roads km/km roadsb,3

0.15 (0.09)

Crossing guard (#)/km roadsb,3

0.12 (0.10)

Dead end (#)/km roads b,3

0.16 (0.20)

Flashing lights (#)/km roads b,3

0.07 (0.09)

Intersection (#)/km roadsb,3

5.63 (1.75)

Page 85: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

69

Route connectivity (intersections /dead ends)b 1.16 (0.20)

Local road km/km roads b,3

0.61 (0.15)

Major roads km/km roads b,3

0.16 (.10)

Minor roads km/km roads b,3

0.08 (.07)

One way streets km/km roads b,3

0.07 (.12)

Pedestrian crossover (#)/km roadsb,3

0.10 (0.12)

Roads km/km2 b,3

12.53 (4.99)

Sidewalks (one) missing km /km roadsb,3

0.08 (.09)

Sidewalks (both) missing km/km roads b,3

0.04 (.09)

Traffic calming segment km/km roads (e.g. speed bumps) b,3

0.05 (.07)

Traffic light #/km roads b,3

0.53 (.29)

Trails km/km road b,3

0.51 (0.67)

Urban area c,3

39 (33.05%)

SOCIAL ENVIRONMENT

School level Total school population

b,6

309.67

(143.94)

Males at school (#)a,6

51.64 (31.61)

New immigrants (< 5 years)a,6

11.57 (8.73)

Primary language other than Englisha,6

47.99 (24.98)

Children grades 4 to 6a,6

32.75 (4.56)

School LOI b,6

0.50 (0.28)

AT below LICO cut off (school DA) a,2

13.76 (10.88) Data type:

a proportion,

b continuous

c dichotomous

Data source: 1 Observational counts,

2 Canadian Census,

3City of Toronto,

4MPAC,

5 Site Survey,

6Toronto District

School Board, 7Toronto Catholic District School Board

Page 86: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

70

Table 4-2: Unadjusted Incident Rate Ratios (95% CI) for candidate variables (p<.2) for

multivariate modeling

Variable Description Unadjusted IRRs

(95% CI)

Outcome

Proportion walking to schoola,1

-

NATURAL ENVIRONMENT

Poor weather (rain or cold) c,5

0.93 (0.85, 1.02)

BUILT ENVIRONMENT

Density

School boundary level Child population (#)/1000m

2 b,2

1.46 (1.29, 1.65)

Total population (#)/1000m2 b,2

1.03 (1.02, 1.04)

Diversity

School boundary level

Recreational facilities (#)/km2 b,3

1.03 (1.00, 1.05)

Commercial land use area/boundary a,4

1.81 (1.07, 3.05)

Industrial land use area/boundary a,4

0.65 (0.41, 1.03)

Institutional land use area/boundary a,4

1.73 (0.96, 3.11)

Design

School level

School crossing guard observedc,5

1.12 (1.03, 1.21)

Double parkingc,5

0.94 (0.87, 1.02)

School boundary level

Other schools within school boundary (#) c,6,7

1.05 (0.98, 1.13)

Crossing guard (#)/km roadsb,3

2.03 (1.39, 2.97)

Intersection (#)/km roadsb,3

1.04 (1.02, 1.06)

Pedestrian crossover (#)/km roadsb,3

1.88 (1.36, 2.59)

Traffic light (#)/km roads b,3

1.28 (1.13, 1.46)

Urban areac,3

1.12 (1.03, 1.22)

SOCIAL ENVIRONMENT

School level Total school population

b,6

1.03 (1.01, 1.06)

New immigrants (< 5 years)a,6

2.14 (1.39, 3.29)

Primary language other than Englisha,6

1.20 (1.03, 1.40)

School LOI b,6

1.26 (1.10, 1.45)

AT below LICO cut off (school DA) a,2

1.72 (1.21, 2.45)

Data type: a proportion,

b continuous

c dichotomous

Data source: 1

Observational counts, 2

Canadian Census, 3 City of Toronto,

4 MPAC,

5 Site

Survey, 6

Toronto District School Board, 7 Toronto Catholic District School Board

Page 87: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

71

Table 4-3: Correlates of walking to school in adjusted analysis (IRR = incident rate ratios

(IRR, 95% CI = confidence interval).

Environmental Component Variable Adjusted IRR

(95% CI)

BUILT ENVIRONMENT

Density

Child population (#)/1000m2

1.36 (1.21, 1.53)

Design Pedestrian crossover (#)/km roads 1.32 (1.01, 1.72)

Design Traffic light (#)/km roads 1.19 (1.07, 1.32)

Design School crossing guard present 1.14 (1.07, 1.21)

Design Intersection density (#)/km roads 1.03 (1.01, 1.05)

SOCIAL ENVIRONMENT Primary language not English 1.20 (1.05, 1.36)

NATURAL

ENVIRONMENT

Poor weather 0.93 (0.87, 0.99)

Effect modification was evident only for school crossing guard (Table 4-4). With no crossing

guard present, walking proportions were positively associated with environmental variables and

negatively associated with poor weather. Lower IRRs were evident where crossing guards were

present, except for child population density.

Table 4-4: Correlates of walking to school in adjusted analysis stratified by presence of

school crossing guard (IRR = incident rate ratios, 95% CI = confidence interval).

Environmental

Component Variable Adjusted IRR (95% CI)

School Crossing

Guard not present

School Crossing

Guard present

( n =73) (n = 45)

BUILT ENVIRONMENT

Density Child population (#)/1000m

2 1.29 (1.10, 1.52) 1.41 (1.23, 1.61)

Design Pedestrian crossover (#)/km roads 1.42 (0.98, 2.06) 1.21 (0.89, 1.66)

Design Traffic light (#)/km roads 1.29 (1.11, 1.51) 1.06 (0.95, 1.19)

Design Intersection density (#)/km roads 1.04 (1.02, 1.06) 1.00 (0.98, 1.02)

SOCIAL

ENVIRONMENT

Primary language not English 1.27 (1.05, 1.53) 1.13 (0.98, 1.30)

NATURAL

ENVIRONMENT

Poor weather 0.87 (0.80, 0.95) 1.06 (0.97, 1.16)

Page 88: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

72

4.6 Discussion

This is the first large study to correlate direct observational counts of walking to school with

objective built environment data. The mean proportion of observed walking was high at 67%;

with large variability between schools. The mean proportion of other active modes (i.e. cycling

and scootering) was 1.7%. On average, 31% of children arrived by car. Previous population-

based national and local Canadian surveys reported 50-55% of children walking to school.36,37

The higher proportions in this study were likely due to sampling children within 1.6 km of

schools, whereas previous estimates were not restricted to children living within walking

distance. Observed proportions were also higher than in Australia and the U.S, where

approximately 48% of children living within walking distance reported walking to school.112,221

Strong associations with walking were found for child population density and traffic lights,

which validated previous findings.108,112,122-124

In addition to the strong positive association

found between walking and school crossing guards, there was evidence of crossing guards acting

as an effect modifier between the environment and walking which has not been previously

reported. With a school crossing guard present, other built and social environmental factors had

less impact on walking which has important implications for potential interventions. Although

road design features may be more easily modified in existing neighbourhoods than those related

to population density and land use, roadway modification can be a highly contested, politicized

process. The process to install crossing guards is much simpler in Toronto, and involves a

reported need by the community to the Toronto Police, followed by an assessment of the

location. If presence of school crossing guards overrides other negative effects of the built and

social environment on walking, adding crossing guards may a feasible and effective method to

increase walking proportions.

Although results were less significant for pedestrian crossovers and intersection design features,

the effect size of pedestrian crossover was high with an IRR = 1.32 (95% CI: 1.01, 1.72). This

feature requires further investigation as it has rarely been addressed and generally is combined

with other crossing features.130

Several other studies have also reported a positive relationship

between intersections and walking, either alone or when combined with low traffic

volume.105,106,120,129,132,134

Page 89: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

73

Null results were found for several design and land use diversity features and observed walking.

Although higher road classification,108,121,132

traffic volume112,125,129,134

and speed117,125

have

been associated with less reported walking, other studies using reported outcomes have also

reported null results.105,124

No association was found with traffic calming which has been

associated with more reported walking.121,130

Parks and recreation facilities were not associated

with observed walking; however, positive associations with reported walking have been

identified in the literature.126,131

Finally, although some studies have reported similar null results

between land use diversity and walking to school,107,121,132,136,137

others have reported positive

associations.117,120,127

Further validation of these relationships is required using observational

data.

The proportion whose primary language was other than English, had a strong association with

walking. Although several studies have found small independent effects of ethnicity on

walking,106,111,126

there is little research investigating cultural associations with active school

transportation. Mixed findings have been reported regarding walking to school and SES.65,71

Neither the student level nor the school geographic level SES variables were significant in this

analysis.

Limitations 4.6.1

This was an ecological study and individual level information was unavailable. Car ownership

and distance to school, two important walking correlates, were not included.66,109

Distance was

unlikely to have had a large influence on results, as children included in the walking proportions

likely lived within walking distance of the school, as defined by TDSB transportation policy.211

Child population density and intersection density (an indicator of route directness) were also

included as proxies for distance, similar to other studies.111,123,208

The lack of individual-level

data also prohibited analysis of family characteristics which may affect choices regarding school

transportation. For example, more active families may choose to live in more walkable

neighbourhoods, which may be reflected in their modes of school transportation.

Walking was assessed at the school level, whereas built environment features were quantified at

the school attendance boundary level. School attendance boundaries were selected as the unit of

Page 90: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

74

analysis, as these are most relevant to policy makers at TDSB. The application of school

walking proportions to the whole school boundary was relevant, as attendance boundaries

generally were within 1.6 km walking distance of the school.

This study only looked at travel to school; however in Toronto, more 11-13 year old children

have been reported to walk home from school in the afternoon than walk to school in the

morning.37

Therefore, the estimated walking proportions may beconservative. Different built

environment characteristics are also relevant at the home, route and school level and on the trip

to and from school85,121,122,137

Individual home and route characteristics could not be assessed

given the ecological nature of the data. Results generally confirmed previous null findings of the

effect of school level characteristics and walking,121

with the only significant characteristic

being the presence of a school crossing guard.

In this study, only objectively measured built environment features were assessed. Parent and

child perceptions of the built environment are also important in the decision regarding school

transportation mode.78,108,120

The use of both objective measurements of the environment

together with measures of perceptions of the traffic environment has been recommended, as

these measures can differ.66,85

Future work is planned to incorporate parent perceptions of the

built environment and traffic danger along with the objective measures presented in this analysis.

Strengths 4.6.2

This study was the first to implement a large scale collection of objective observational counts of

walking to school, together with objective built environment data from city databases and field

surveys. The strengths of this study included the objective observational outcome data and the

generalizability of results. The large sample represented virtually all regular program JK-6

schools in Toronto and results are likely generalizable to other regular program elementary

schools in Toronto. Finally, this was the first time objective parcel level land use data were used

in a study of children’s active transportation to school in Toronto.

Page 91: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

75

4.7 Conclusion

To summarize, average walking proportions to school in Toronto were high, with large

variability between schools. Direct observation confirmed associations between walking and

child population density, and with several specific roadway design features. No association was

found between walking to school and land use diversity, indicating that land use, while important

for adult walking, may not be as important for children. Of particular interest was the

association between school crossing guards and walking, and their modifying effect on reducing

the influence of other roadway features on walking. The addition of school crossing guards may

be a feasible and effective method of increasing walking proportions. These results may have

important implications for policies regarding walking promotion around schools.

Page 92: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

76

4.8 Supplementary/Supporting Analyses

The following supplementary analyses were not included in the published manuscript but support

the study findings.

Principal Component Analysis 4.8.1

Principal component analysis (PCA) with a varimax rotation (a type of orthogonal rotation) was

conducted to explore whether grouping of the built environment variables significant in the

univariate analysis corresponded to the density, diversity and design paradigm as developed by

Cervero and Kockelman.83

Orthogonal rotation was used to produce factors that were

uncorrelated, and results replicated. Only variables with factor loadings > .35 were included in

the final extraction. Principal component analysis resulted in 4 components which accounted for

67% of the variability in the 15 built environment variables entered into the final extraction

(Table 4-S2). The first factor, which accounted for 27% of the total variation, represented

design. The highest factor loading was for urban location, followed by school age, one way

streets and traffic calming. The second factor explained 17% of the total variation and

represented land use diversity in terms of urban features. The highest factor loading was for

traffic lights, followed by major roads and commercial and then residential land use. The link

between roads and other transportation features and land use has been well discussed in the

literature.226

Although in this analysis, traffic lights factor into the land use component due to

their correlation with major roads, it has been considered a road design feature in these analyses.

The third factor explained 14% of the variation represented land use in terms of “green” features,

with the highest loading for vacant land, followed by parks and trails. Finally, the fourth factor

explained 8% of the variation, and represented density, with the highest factor loading for child

population density, followed by total population.

It can be concluded therefore, that data in the City of Toronto, do correspond to Cervero and

Kockelman’s 3D model of density, diversity and design,83

with the one difference being that

there were two components identified for diversity features; one related more to urban land use

Page 93: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

77

and other related to “green” land use. These two land use components will be combined for

subsequent use in these analyses. The 3D components will be used to organize built

environment variables in this dissertation.

Proportion Observed Walking 4.8.2

Figure 4S-1 portrays the distribution of the proportion of observed walking at the 118 study

schools. The observed walking proportion was over 60% in 83 (70%) of study schools. Of the

11 schools with < 50% walking, 6/11 (55%), observations were on a rainy day, versus 13/107

(12%) of schools with >50%. Only 1/11 (9%) of these schools had school crossing guards,

whereas 44/107 (41%) of the schools with >50% of walking had school crossing guards.

Observer comments indicated that one school boundary was bisected by train tracks, and several

were bisected by major arterials, which required crossing by some children. Several schools also

had roads nearby with missing sidewalks nearby, and one school had nearby road construction

which was diverting a large amount of fast traffic near the school.

Network Analysis 4.8.3

Although the mean proportion of roads within the school boundaries and within the 1.6 km

walking distance of the school along the road network was 95%, there was substantial variability

(SD 10%). The proportion ranged from 47% to 100% with 5 schools having less than 70% of the

roads within walking distance (Figure 4S-2). The school boundaries of these five schools had

either industrial complexes, parkland, railway tracks, hydro fields or a split attendance boundary

(i.e., areas nonadjacent).

Predicted Values 4.8.4

Estimated predicted collision rates at different levels of each of the covariates are useful to

illustrate the relationships in the adjusted model that may be more easily interpreted than

Page 94: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

78

incidence rate ratios. An example illustrating predicted collision rates at different levels of

intersection density is presented in Figure 4S-6. Intersection density was a continuous variable

in the model, but for ease of interpretation it was set at defined levels within the range observed

in the data as identified in the box plot. In the City of Toronto, intersection density ranged from

3/km road to 10/km road, with the median at 5/1000m2.

For every additional 2 intersections/km road, the walking proportion increased by 6%. This is

equivalent to an IRR of 1.03 as portrayed in Table 4-2 (i.e., 1.03^2 = 1.06) for 2 units. For

example, the predicted walking proportion was 70.7% with 4 intersections/km road and 74.6%

with 6 intersections/km road when a school crossing guard was present and the weather was

good. The rate increased a total of approximately 12% from the lower range of intersection

density observed (4 intersections/km), to the maximum observed (10 intersections/km) The

proportion walking to school was less with no crossing guard and when the weather was poor.

Sensitivity Analysis 4.8.5

The robustness of the final main effects model was tested using a variety of techniques described

below.

4.8.5.1 Trimming of Variables

Visual inspection of histograms was conducted to identify outliers in the data. Trimming of

variables was done for cases that exhibited outliers more than 3 interquartile ranges from the

75th percentile according to box plots. When the final model was rerun with the trimmed

variables, the effect sizes were similar and in the same direction.

4.8.5.2 Residual Diagnostics

Residual diagnostics were conducted with final models to identify outliers using Cook’s d,

leverage and Pearson’s standardized betas. A comparison of the final model and a model

Page 95: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

79

excluding 3 schools which exhibited high cook’s d, are presented in Table 4S-3. IRRs were

similar for all variables, with just poor weather becoming non-significant after the outliers were

excluded.

4.8.5.3 Alternative Modeling Strategies

Alternative modeling strategies were tested using three different walking outcomes. Walking

proportion outcome was analyzed as 1) a continuous variable, using Ordinary Least Squares

Regression (OLS). OLS regression can be used when the distribution of count data approaches

normal 227

2) as a logit transformed response using OLS. Logit transformation addresses the

issue of using OLS with proportional data which represents a closed scale228,229

3) as a binary

response using grouped logistic regression. Logistic regression for grouped data incorporates the

response as the number of “successes”/number of cases, and is also suitable to use for

proportional data.220

When using 3 different modeling strategies, all variables remained

significant with the coefficients going in the correct direction, with the exception of poor

weather, which became non-significant with the OLS, logit transformation and grouped logistic

regression modeling. These analyses indicated that the results of the final model using negative

binomial regression were robust.

Page 96: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

80

4.9 Supplementary Tables

Table 4S- 1: Data sources and variable type.

Data sources and variable name Shape features Created

Variable Type Source/

Retrieval Year

OBSERVATIONAL DATA

Walking to school Assigned to school point Proportion May-June 2011

SITE SURVEY

School crossing guard observed Assigned to school point Dichotomous May-June 2011

Cars appear to be driving fast near

school

Assigned to school point Dichotomous May-June 2011

Traffic congestion seen during drop off Assigned to school point Dichotomous May-June 2011

Dangerous midblock crossing near

school

Assigned to school point Dichotomous May-June 2011

Dangerous intersection near school Assigned to school point Dichotomous May-June 2011

Drop offs opposite side of road Assigned to school point Dichotomous May-June 2011

Double parking Assigned to school point Dichotomous May-June 2011

Cars blocking view Assigned to school point Dichotomous May-June 2011

Mean speed > 5 km over speed limit Assigned to school point Dichotomous May-June 2011

School traffic/minute Assigned to school point Continuous May-June 2011

Poor weather (rain or cold) Assigned to school point Dichotomous May-June 2011

STATISTICS CANADA, CANADIAN CENSUS

Dissemination areas (DA) Polygon Continuous 2006

Child population (#)/1000m2 Assigned to DA polygon Continuous 2006, 2011, 2001

Total population (#)/1000m2 Assigned to DA polygon Continuous 2006, 2011, 2001

Multi-family dwelling (#)/1000m2 Assigned to DA polygon Continuous 2006, 2011, 2001

Old houses (pre 1946 (#)/1000m2 Assigned to DA polygon Continuous 2006, 2011, 2001

AT below LICO cut off (school DA) Assigned to DA polygon Proportion 2006, 2011, 2001

Central city (>50% school boundary in

pre-amalgamated City of Toronto) by

Census Sub Divisions

Polygon Dichotomous 1996

CITY OF TORONTO

Transportation Services

Collision Rates Point Continuous 2002-2011

Sidewalks (one) missing km/km roads Line Continuous Jan 2011

Sidewalks (both) missing km/km road Line Continuous Jan 2011

Traffic calming km/km roads Line Continuous June 2012

Open Data

Flashing lights (#)/km roads Point Continuous June 2011

Page 97: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

81

Park land area / boundary Polygon Proportion Dec 2011

Pedestrian crossover (#) /km roads Point Continuous June 2011

Recreational facilities (#)/km2 Point Continuous May 2012

Traffic light (#)/km roads Point Continuous June 2011

Open Data – Toronto Centreline

Dead end (#)/km roads Point Continuous Dec 2011

Intersection (#)/km roads Point Continuous Dec 2011

Local road km/km roads Line Continuous Dec 2011

Collector roads km/km roads Line Continuous Dec 2011

Major roads km/km roads Line Continuous Dec 2011

Minor roads km/km roads Line Continuous Dec 2011

One way streets km/km roads Line Continuous Dec. 2011

Road density km Line Continuous Dec 2011

Walkways/Trails km/km roads Line Continuous Dec 2011

Route connectivity

( # intersections/dead end )

Derived Continuous Dec 2011

Toronto Police Services

Crossing guard density (#)/km roads Point Continuous 2010/2011

TORONTO DISTRICT SCHOOL BOARD (*CATHOLIC)

Strategy and Planning School Boundary km

2 Polygon Continuous Feb 2011

Schools Point

Total school population Assigned to school point Continuous 2010/2011

Children grades 4 to 6 Assigned to school point Proportion 2010/2011

Other TDSB/Catholic schools within

school boundary*

Point Dichotomous Feb 2011

TDSB Website

Males at school Assigned to school point Proportion Spring 2011

New immigrants (< 5 years) Assigned to school point Proportion Spring 2011

English not primary language Assigned to school point Proportion Spring 2011

Learning opportunities index (LOI) Assigned to school point Continuous 2011

MUNICIPAL PROPERTIES ASSESSMENT CORPORATION (MPAC)

Entropy (mixed land use) Derived Continuous 2011

Commercial land use boundary Polygon Proportion 2011

Industrial land use/boundary Polygon Proportion 2011

Institutional land use area/boundary Polygon Proportion 2011

Vacant land area/boundary Polygon Proportion 2011

TERANET (VIA UNIVERSITY OF TORONTO)

Teranet ownership parcel data Polygon n/a 2012

Page 98: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

82

Table 4S- 2: Built environment factor loadings from principal component analysis.

Components

Factors Design Land Use

Diversity

Land Use

Diversity Density

Urban 90

Old houses 89

Road density 83

One way streets 80

Traffic calming 76

Age of school 73

Intersections 73

Flashing lights 57

Entropy 82

Traffic lights 67

Industrial 62

Commercial 61

Parks 84

Vacant land 78

Trails 72

AT LICO 93

Child population 90

Page 99: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

83

Table 4S- 3: Results of negative binomial regression excluding 3 outlier schools

(IRR, 95% CI).

COMPONENT All Schools

(n = 118)

Excluding 3 outliers

(n = 115)

BUILT ENVIRONMENT

Child population (#)/1000 m2

1.36 (1.21, 1.53)

1.42 (1.27, 1.58)

Pedestrian cross over density (#)/km roads 1.32 (1.01, 1.72) 1.43 (1.12, 1.82)

Traffic light (#)/km roads 1.19 (1.07, 1.32) 1.15 (1.04, 1.27)

School crossing guard present 1.14 (1.07, 1.21) 1.14 (1.07, 1.20)

Intersection density (#)/km roads 1.03 (1.01, 1.05) 1.02 (1.01, 1.04)

SOCIAL ENVIRONMENT

Proportion English not primary language (school)

1.20 (1.05, 1.36)

1.16 (1.03, 1.31)

NATURAL ENVIRONMENT

Poor weather

0.93 (0.87, 0.99)

0.95 (0.90, 1.02)

Page 100: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

84

4.10 Supplementary Figures

6 5

23

29 29

18

7

0

5

10

15

20

25

30

35

Number Of Schools

% Walking

1 4 5 7

101

0

20

40

60

80

100

<59 60-69 70-79 80-89 90-100

Number of School

Boundaries

% Roads within 1.6 km of school

Figure 4S-1: Distribution of walking proportion across 118 study schools. The median

walking proportion was 68.6% (range 27.9% to 98.2%). The walking proportion was >60% in

83 (70%) of the schools.

Figure 4S-2: Distribution of the proportion of roads in 118 study school boundaries within

1.6 km of schools. The median proportion was 98.7% (range 47% to 100%). One hundred and

one schools (86%) had >90% of roads in the school boundary within 1.6 km of the school.

Page 101: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

85

Figure 4S-3: Predicted walking rates by intersection density. In the City of Toronto, intersection

density ranged from 2.9/km road to 10.0/km road, with the median at 5.1 /1000m2. For every

additional 2 intersections/km road, the walking proportion increased by 6%. The proportion walking

to school was less where there was no crossing guard present and when the weather was poor.

75th percentile

Median (5.0)

25th percentile

Maximum (10.0)

Minimum (3.0)

Page 102: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

86

5 Motor Vehicle-Pedestrian Collisions and Walking to

School: The Role of the Built Environment

5.1 Preface

This purpose of this chapter is to further build on the analysis presented in Chapter 4, to address

the overall objective of examining the role of the built environment on the relationship between

walking to school and child pedestrian-motor vehicle collisions. In this analysis, child

pedestrian-motor vehicle collision is identified as the outcome, and walking to school is the

exposure. The primary contribution of this chapter is that it greatly clarifies the relationship

between walking to school and child pedestrian-motor vehicle collisions, and it identifies specific

built environment confounders influencing this relationship.

A manuscript based on this chapter has been published in Pediatrics.

Rothman L, Macarthur C , To T, Buliung R, Howard A. Motor vehicle-pedestrian collisions

and walking to school: the role of the built environment. Pediatrics published online: 2014 (doi:

10.1542/peds.2013-2317).

Reprint rights have been granted to Linda Rothman by the American Academy of Pediatrics.

Page 103: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

87

5.2 Abstract

Objectives 5.2.1

Initiatives to increase active school transportation are popular. However, increased walking to

school could increase collision risk. The built environment is related to both pedestrian collision

risk, and walking to school. We examined the influence of the built environment on walking to

school and child pedestrian collisions in Toronto, Canada.

Methods 5.2.2

Police-reported pedestrian collision data from 2002-2011 for children ages 4-12, proportion of

children walking to school, and built environment data were mapped onto school attendance

boundaries. Collision rates were calculated using 2006 census populations and modelled using

negative binomial regression.

Results 5.2.3

There were 481 collisions with a mean collision rate of 7.4/10, 000 children per year. The

relationship between walking proportion and collision rate was not statistically significant after

adjusting for population density and roadway design variables including; multi-family dwelling

density, traffic light, traffic calming and one-way street density, school crossing guard presence

and school socioeconomic status.

Conclusions 5.2.4

Pedestrian collisions are more strongly associated with built environment features than with

proportions walking. Road design features were related to higher collision rates and warrant

further examination for their safety effects for children. Future policy designed to increase

children’s active transportation should be developed from evidence that more clearly addresses

child pedestrian safety.

Page 104: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

88

5.3 Introduction

Road traffic injuries are the leading cause of child death in most developed countries.22,24,28

In

2010, 61 children died and over 9000 were injured on Canada’s roads.17

Pedestrian collisions

account for approximately 25% of children’s road traffic fatalities.27

Much of children’s

exposure to traffic as pedestrians is during school travel.130,230,231

Although walking to school

rates have declined in Canada, approximately 50% of Ontario children walk to school.36-38

In

Toronto, Canada almost 50% of child pedestrian collisions occurred during school travel times

with more than 1/3 occurring within 300 meters of a school.39

Initiatives to increase active school transportation (AST) are popular. In 2005, the US Safe

Routes to School (SRTS) program received over $1 billion of federal funding.168,170

In Canada,

SRTS programs have developed at a grassroots level, with limited government funding.170

Increased walking to school might benefit overall health, but also increase collision risk. A

recent Canadian study reported a dose-response relationship between longer school travel

distances and injury.44

Many conceptual frameworks describing correlates of walking to school focus on the built

environment (BE). The built environment is defined as “the human-made space in which people

live, work, and recreate on a day-to-day basis.”183

No framework considers child pedestrian

injury, and there has been little research examining pedestrian collision as an AST outcome. The

purpose of this study was to examine the effect of the built environment on the relationship

between observed walking to school and child pedestrian collisions.

5.4 Methods

Study Design, Setting and Population 5.4.1

A cross-sectional study examined child pedestrian collisions from 2002-2011. Walking exposure

was measured in an observational study in the spring, 2011 in Toronto Schools were excluded if

Page 105: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

89

they were participating in another active transportation study or had special programs (e.g.

French immersion), which serve large areas. Further methodological details were reported

previously.232

Ethics approval was obtained from the Toronto District School Board (TDSB) and

the Hospital for Sick Children Research Ethics Boards.

Outcome 5.4.2

Collision data were extracted from Toronto Police Service motor vehicle collision reports from

2002-2011 for children ages 4-12 years. Police-reported collisions include those resulting in no

injury, minimal, minor (seen in the emergency department), major (admitted to hospital), and

fatal injuries. Collision rates by study school boundary were calculated over a 10-year period

and reported as an annualized rate per 10,000 children. Child population numbers were obtained

for Dissemination Areas (DAs, 400-700 residents) using the 2006 Canadian Census. DAs were

mapped onto school boundaries, and population estimates calculated using area-weighted

proportionate estimation.123,212

Exposure 5.4.3

Two trained observers measured walking exposure by counting children either arriving to school

walking, by car or other active means (bicycle or scooter) on a single day. Observations were

repeated in 22 schools (19%) where count accuracy was questioned and were repeated one week

later in another 12 schools (10%) to examine test-retest reliability. Walking proportion was

calculated from the total number observed. The majority of children counted lived within

walking distance of the school (i.e., < 1.6 km), as defined by TDSB transportation policy.

Potential Covariates 5.4.4

Built environment variables were identified from a literature review and conceptually organized

using Cervero and Kockelman’s 3D’s; density of population, diversity of land use, and design of

Page 106: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

90

the roadway environment.83,85,86

Socioeconomic status (SES) variables were included due to

previously reported correlations with child pedestrian injury.233,234

Table 5-1 presents each

variable according to its conceptual category, level of measurement, and data source. Variables

were measured at the school and school attendance boundary level.

Data Sources 5.4.5

5.4.5.1 Canadian Census

DA data were obtained from the 2006 Canadian census. Social environment variables included

the proportion of households falling below the After Tax, Low Income Cut-Offs (ATLICO) in

each school’s DA, as a proxy measure for the school neighbourhood SES.

5.4.5.2 Municipal Property Assessment Corporation (MPAC)

Land use diversity variables were derived using 2011 parcel level data from MPAC, which

classifies and assesses properties in Ontario. Mixed land use was calculated using an entropy

index which ranges from 0 (single land use) to 1 (equal distribution of residential, commercial,

industrial, institutional, and vacant land classifications).53,213

5.4.5.3 Site Audits

School level design variables were obtained from site audits conducted by the observers during

school drop-off time. Only adults employed by Toronto Police Services surrounding the school

were identified as school crossing guards. Vehicle speed and volume were measured along a

road within 150 meters of the school using manual short-based methods.214,215

Page 107: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

91

Table 5-1: Variables according to conceptual component, level of measurement and data source.

Domain Level Data Source Variables

Outcome School boundary Toronto Police Collision Rate/10, 000

Exposure School Site Survey Proportion walking to school

BUILT ENVIRONMENT

Density

School boundary

Census

Multi-dwelling (e.g. apartments)

(#)/1000m2

Diversity School boundary MPAC Commercial land use area/boundary

Derived Entropy Index

MPAC Industrial land use area/boundary

MPAC Institutional land use area/boundary

MPAC Park land area/boundary

City of Toronto Recreational facilities (#)/km2

MPAC Vacant land area/boundary

Design School Site Survey Cars appear fast near school

Site Survey Cars blocking view

Site Survey Dangerous intersection near school

Site Survey Dangerous midblock crossing near

school

Site Survey Double parking

Site Survey Drop offs opposite side of road

Site Survey Mean speed > 5 km over speed limit

Site Survey School crossing guard observed

Site Survey School traffic/minute

Site Survey Traffic congestion during drop off

School boundary City of Toronto Collector roads km/km road

City of Toronto Crossing guard (#)/ 10 km road

City of Toronto Dead end (#)/km road

City of Toronto Flashing lights density (#)10/km road

City of Toronto Intersection (#)/km road

City of Toronto Major roads km/km road

City of Toronto Minor roads km/10 km road

City of Toronto One way streets km/ 10 km road

TDSB,TCDSB Other schools within school

boundary(#)

City of Toronto Pedestrian crossover (#)/km road

City of Toronto Sidewalks (both) missing km/km road

City of Toronto Sidewalks (one) missing km/km road

City of Toronto Traffic calming segment km/10 km

road

City of Toronto Traffic light #/km road

City of Toronto Trails density km/km road (km)

Census Central city

SOCIAL ENVIRONMENT School level TDSB Children grades 4 to 6

TDSB English not primary language

TDSB Males at school

TDSB New immigrants (< 5 years)

TDSB Total school population

TDSB School learning opportunities index

DA level Census AT below LICO cut off (school DA)

Page 108: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

92

5.4.5.4 City of Toronto

Collision outcome data, school boundary level design variables and recreational facility land use

were obtained from The City of Toronto (Transportation Services and the Open Data website).235

Densities were calculated either per school boundary area or per km of roadway. Toronto

consists of an older central city with many pre-World War II traditional neighbourhoods and an

inner ring representing newer, car-oriented post-World War II neighbourhoods.159

Central city

status was assigned if >50% of the school boundary overlapped with the central city area.

5.4.5.5 Toronto District School Board

The TDSB provided social environment variables and school attendance boundary data. The

2011 Learning Opportunities Index (LOI) is a composite measure of children’s SES attending a

school reflecting parental education, income, housing and immigration, and ranges from 0 (high)

to 1 (low SES).

Statistical Analysis 5.4.6

School attendance boundaries were the unit of analysis. All features were mapped onto

boundaries using ArcMap v.10.225

Statistical analysis was conducted using SAS, v.9.3.236

Mutlicollinearity was assessed using Variance Inflation Factors (VIF). If two correlated

variables had VIFs > 10, the variable with the higher standardized unadjusted beta was

retained.237

Child pedestrian collision rates were modeled using negative binomial regression. Variables

with p < 0.2 in the unadjusted analysis were entered using forward manual stepwise regression,

according to the magnitude of standardized betas.218

Variables were retained if significantly

associated with the outcome at p <.05, with confounding identified if the variable also changed

walking estimates by >10%.218,238

Stratified analyses were conducted by tertiles for design

features.

Page 109: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

93

Model fit was assessed using The Akaike Information Criteria.219,220

Incident rate ratios (IRR)

were calculated through exponentiation of betas from the regression models and presented with

95% confidence intervals (CI). Sensitivity analysis was conducted using 10, 7, and 5 years of

collision data. A sub-analysis was done of collisions occurring only during school travel times

(7:30-9:00 am, 11:30 am-1:00 pm, 3:00-5:00 pm, weekdays, September-June). Both before and

after school collisions were included, as numbers of elementary school-age children walking to

and from school were found to be comparable in another observational study.72

5.5 Results

Among 245 JK-grade 6 schools, 126 met inclusion criteria, 8 refused and 118 schools

participated in the study. A total of 481 collisions occurred within 105 school boundaries; the

remaining 13 schools had no collisions. There were 24 collisions resulting in no injury, 191

minimal, 236 minor, 30 major injuries and 1 fatality. The average collision rate was 7.4/10,000

per year (range 0 -27/10,000, SD = 6.7). Two outlier schools with extreme collision rates (>3

interquartile ranges above the 75th

percentile) were excluded. These were inner city schools with

very small attendance boundaries and few resident children contributing to low rate

denominators. The mean proportion of children walking to school was 67.0% (range 27.9-

98.2%, SD = 14.4%) with high test-retest reliability of walking counts (Pearson’s r = .96).

In the unadjusted analysis, increased walking proportions were associated with higher pedestrian

collision rates (IRR=3.47, 95% CI=1.15, 10.47, Table 5-2). This was equivalent to a 13%

increase in collision rate with every 10% increase in walking.

Older housing, residential, road and local road density were dropped from further analyses due to

multicollinearity.

In the adjusted collision model walking proportions were no longer associated with collision

rates (Table 5-3). Collisions were less frequent in areas with higher multi-family dwelling

density (IRR=0.84, 95% CI=0.73, 0.96). Design variables, including higher densities of traffic

lights (IRR=3.20, 95% CI=1.89, 5.41), traffic calming (IRR=1.31, 95% CI=1.06, 1.63), one-way

Page 110: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

94

streets, (IRR=1.19, 95% CI=1.03, 1.36) school crossing guards (IRR=1.45, 95% CI=1.09, 1.91)

and low school SES (IRR=2.36, 95% CI=1.39, 3.99) were positively associated with collisions

and changed the walking exposure by >10%. Significant design variables were generally related

to road crossing. Traffic light density and SES exhibited the strongest associations.

The association between walking and collisions differed by levels of traffic light density,

indicating evidence of effect modification; walking was positively associated with collisions in

low traffic light density areas, with no association in medium or higher density areas (Table 5-4).

Analysis using 5 and 7 year collision data revealed very similar models. School travel time

collisions represented 44% (n = 214) of total collisions within 83 school boundary areas.

Although there were reductions in the magnitude of effects, effect direction was similar to the

full model with less precise estimates due to the smaller sample size.

5.6 Discussion

A significant positive univariate association between walking and pedestrian collisions

disappeared in adjusted models which controlled for population density, design features

(primarily related to road crossings) and SES. Land use diversity was unrelated to collisions.

These findings are encouraging in that it suggests that modification of the built environment may

both promote walking and make it safer. Although causality could not be definitively

determined because of the cross-sectional and ecological study design, the results suggest a

strong influence of the built environment on walking and collisions.

Page 111: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

95

Table 5-2: Descriptive statistics and significant unadjusted incident rate ratios

(p <.20, IRR = incident rate ratio, 95% CI= 95% confidence interval).

Component Mean (SD)

N (%)

Unadjusted IRRs

(95% CI)

OUTCOME

Collision Rate/10, 000a,1

7.41 (6.73) N/A

EXPOSURE

Walking to school proportionb,2

67.0% (14.5) 3.47 (1.15, 10.47)

BUILT ENVIRONMENT

Density

Multi-family dwelling (apartments, duplexes)

(#)/1000m2a

1.41 (1.29)

1.12 (1.00, 1.26)

Diversity

Commercial land use area/boundaryb

6.35 (7.20)

30.04 (4.72, 191.04)

Entropy Indexa 0.61 (0.13) 3.67 (1.11, 12.14)

Recreational facilities (#)/km2a

1.70 (1.46) 1.12 (1.01, 1.24)

Park land area/boundaryb 7.56 (6.80) 0.14 (0.02, 1.35)

Design

School crossing guard observedc

45 (38.79%)

1.51 (1.12, 2.02)

Dangerous intersection near schoolc 39 (33.62%) 1.22 (0.90, 1.67)

Double parkingc 54 (46.55%) 0.82 (0.60, 1.10)

Traffic congestion seen around school 76 (65.52%) 0.81 (0.60, 1.11)

Design

Traffic light #/km roada

0.53 (0.29)

2.61 (1.60, 4.24)

Pedestrian crossover (#)/km roada 0.10 (0.12) 2.41 (0.65, 8.94)

Central cityc 37 (31.90%) 1.70 (1.26, 2.28)

Other schools within school boundary (#)c 38 (32.8) 1.52 (1.13, 2.05)

Minor roads density km/10 km roada 0.77 (0.72) 1.43 (1.16, 1.76)

Traffic calming segment km/10 km road a 0.44 (.70) 1.31 (1.06, 1.61)

One way streets km/10 km roada 0.70 (1.16) 1.29 (1.15, 1.46)

Flashing lights (#)/10 km roada 0.68(0.92) 1.25 (1.08, 1.46)

Intersection (#)/km roada 5.56 (1.70) 1.19 (1.09, 1.29)

Crossing guard (#)/10 km roada 1.15 (0.98) 1.15 (0.98, 1.34)

Collector roads km/km roada 0.15 (0.09) 0.26 (0.04, 1.47)

Sidewalks (both) missing km/km roada 0.05 (.09) 0.08 (0.01, 0.59)

Trails density km/km roada 0.50 (0.74) 0.80 (0.64, 0.99)

Sidewalks (one) missing km/km roada 0.08 (.09) NS

SOCIAL ENVIRONMENT School learning opportunities index

a

0.50 (0.28)

1.75 (1.00, 3.05)

Total school population/100a 3.11 (1.44) 0.93 (0.85, 1.02)

Children grades 4 to 6b 32.70 (45.63) 0.06 (0.00, 1.62)

Data type: a continuous

b proportion

c dichotomous

IRR, Incident rate ratios; 95% CI, 95% Confidence Interval; LOI Learning opportunities index

Page 112: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

96

Table 5-3: Correlates of child pedestrian collisions in adjusted analyses

(IRR = incident rate ratio, 95% CI= 95% confidence interval).

Component Variable IRR (95% CI) P Value

EXPOSURE Walking to school proportion 0.84

(0.29, 2.46)

.747

BUILT

ENVIRONMENT

Density

Multi-family dwelling (#)/1000m2

0.84

(0.73, 0.96)

.009

Design Traffic light #/km roads 3.20

(1.89, 5.41)

<.0001

Design School crossing guard present 1.45

(1.09, 1.91)

.010

Design Traffic calming km/10 km road 1.31

(1.06, 1.63)

.014

Design One way streets km/10 km road 1.19

(1.03, 1.36)

.015

SOCIAL

ENVIRONMENT

School learning opportunities index 2.36

(1.39, 3.99)

.001

Table 5-4: Incidence rate ratios of collisions stratified by traffic light density tertiles

(IRR = incident rate ratio, 95% CI= 95% confidence interval).

Component Variable Low

(n = 38)

Medium

(n = 39)

High

(n = 39)

IRR

(95% CI)

IRR

(95% CI)

IRR

(95% CI)

EXPOSURE Walking to school

proportion

23.33

(1.21, 415.10)

0.36

(0.04, 2.96)

0.97

(0.26, 3.59)

BUILT

ENVIRONMENT

Density

Multi-family dwelling

(#)/1000m2

0.54

(0.28, 1.03)

1.00

(0.82, 1.23)

0.74

(0.63, 0.87)

Design School crossing guard

present

0.94

(0.48, 1.86)

1.55

(0.92, 2.61)

1.90

(1.29, 2.82)

Design Traffic calming km/10

km road

1,89

(1.11, 3.10)

1.48

(1.12, 1.96)

0.70

(0.49, 1.01)

Design One way streets km/10

km road

1.29

(0.98, 1.70)

1.01

(0.84, 1.22)

1.71

(1.37, 2.13)

SOCIAL

ENVIRONMENT

School learning

opportunities index

(LOI)

2.62

(0.82, 8.40)

3.15

(1.56, 6.37)

1.15

(0.50, 2.67)

Page 113: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

97

Comparisons of Findings to Previous Studies 5.6.1

Prior studies of the association between pedestrian collisions and walking have shown

inconsistent results. In population studies, increased pedestrian volume was associated with

decreased collisions leading to a “safety in numbers” effect.41,240

Other studies found positive

relationships between walking exposure and child pedestrian collisions.42-44

Different

environmental conditions and a lack of adequate control for confounders may contribute to these

contradictory findings.

Built environment features have been previously correlated with both walking to school and

pedestrian collisions.65,66,78,91,99,241

Major roads, urban location, street/intersection density,

sidewalks, school crossing guards, population density, distance and traffic controls have been

associated with walking to school.65,241

A reduction in pedestrian injures has been associated

with built environment road design interventions by 50-75% in specific locations.99

Traffic

calming has been associated with a 37% reduction in fatal pedestrian outcomes 242

and a 53%

reduction in child pedestrian collision risk.193

The built environment influences both the walking

exposure and risk, so must be considered as a confounder in child pedestrian safety studies.

Confounders 5.6.2

Multi-family dwelling density, low SES, one-way streets, traffic calming and school crossing

guards were confounders of the relationship between walking to school and child pedestrian

collisions as they changed the walking exposure estimates by >10%. Higher multi-family

dwelling density was associated with lower collision rates, which may indicate a safer walking

environment and supports the “safety in numbers” concept (Figure 5-1). Multi-family dwelling

has been associated with higher pedestrian collision risk at the individual level, unadjusted for

walking exposure.195,202

At the population level, higher population density may reflect shorter

distances to school with fewer road crossings, resulting in less traffic exposure and fewer

collisions. Lower SES has been consistently associated with higher child pedestrian collisions,

233,234 with higher rates attributed to lack of safe play areas and higher walking exposures due to

lower car ownership.43,243-246

More collisions on one-way streets have also been found in

Hamilton Ontario, with a 2.5 times higher injury rate on one-way compared to two-way

Page 114: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

98

Child

Pedestrian

Collisions

Walking

to School

Effect Modifier

Traffic light

Confounders

Multi-Family Dwelling -

Low SES +

One Way Streets +

Traffic Calming +

School Crossing Guards +

streets.200

Traffic calming devices were positively associated with collisions, whereas other

studies reported a negative association; however, no adjustment was made previously for traffic

exposure.100,193,194

Finally, the presence of school crossing guards were positively associated

with collisions, which has similarly been found in Montreal, Quebec.151

Features considered to be confounders in this analysis may fall on the causal pathway

(particularly school crossing guards) between walking to school and pedestrian collisions.

However, the relationships are complicated in that they may be bi-directional with the possibility

of reverse causality. The cross-sectional nature of the data restricted the ability to determine

temporality and directionality of the relationships. Causal pathways need to be established to

determine if any of these features act as mediators between walking to school and child

pedestrian collisions.

Effect Modifiers 5.6.3

Overall, traffic light density was positively related to both walking and collisions. Stratified

analysis revealed that traffic light density was an effect modifier, as walking was only positively

associated with collision rates where traffic light densities were low. Traffic light density has

been previously associated with less pedestrian collisions.247,248

Figure 5-1: Multivariate relationships between walking to school, child

pedestrian injury and the built environment.

Page 115: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

99

Unexpected Results 5.6.4

The positive associations between collision rates and both school crossing guards and traffic

calming were unexpected, as these features are intended to be protective. Several potential

hypotheses exist to explain these relationships. These features may be indicators for more

hazardous locations with the feature’s safety effect insufficient to reduce injury risk.151,195

These

features could also potentially create hazardous traffic situations for children. Although evidence

exists of reduced effectiveness of specific road design features for older adults, this has not been

well studied in children.249

The ecological data may also have contributed to these unexpected findings. Collisions occurred

at a specific location, whereas the built and social environment data were measured area-wide

and may not represent the environment at the collision location. Area-level data would also not

identify if the presence of a built environment feature either displaced collisions or was a marker

for a more dangerous traffic environment elsewhere within the school boundary. The use of

boundaries other than school attendance boundaries could potentially have resulted in different

observed relationships, which is known as the modifiable areal unit problem.105

The school

attendance boundary was selected as the unit of analysis, as it locally relevant for school

transportation policies. Despite these limitations, ecological analysis was most appropriate, as an

understanding of both the physical and social environment measured on a geographic level is

essential when examining pedestrian injury.250

The cross-sectional design necessitated some prior assumptions which may have influenced the

results. The first assumption was that walking exposure measured in 2011 was representative of

exposure throughout the 10 year collision data period (2002-2011). This was reasonable, as

there is evidence of stabilization of active school travel prevalence from 2001 onwards, after an

earlier period of sharp decline.35,37,251

The built environment was also assumed to remain

unchanged over the collision data period. Substantial changes would not be expected in the built

environment generally, as the study neighbourhoods were well-established. However, this may

not have been the case for features more easily implemented, such as the installation of traffic

calming or school crossing guards. The possibility of reverse causality exists, with collisions

occurring prior to the installation of these features.

Page 116: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

100

Strengths and Limitations 5.6.5

This was a large, population-based study which directly measured walking to school as a proxy

for general walking exposure. The only previous study that used direct observation of walking to

school to investigate associations with built environment was limited by a small sample and

little geographic diversity.72

Multivariable modelling was used to test the association between walking and pedestrian

collisions while controlling for the built environment. The model identified built environment

features acting as confounders and effect modifiers of the relationship between walking exposure

and collision outcome. One other recent study modeled walking school trips and active

transportation injuries examined limited numbers of potential built environment confounders.44

However, none of the roadway design features significant in our study were included. The

measurement of walking exposure had some limitations. Walking exposure was only measured

on route to school; however, non-school travel time collisions were included. The study’s

intention was to examine factors related to child pedestrian collisions in Toronto, and not only

those occurring during school travel. As school is children’s most common walking destination,

walking to school is considered the best proxy for their general walking activity.130,230,231

Consistent evidence also exists that children who actively commute to school, walk more in

general.54

The inclusion of non-school travel time collisions when school crossing guards were

off-duty could potentially have produced an artificial positive association between guards and

collisions. However, when restricting the data to school times, the directions of effects were

maintained. Despite the limitations of using walking to school as a proxy for children’s general

walking, it is currently the most feasible measurement of children’s walking exposure. Walking

exposure has been poorly dealt with in the past, and creative methods of measurement are needed

to most accurately evaluate pedestrian risk.

Future Research 5.6.6

Specific built environment features were identified in this analysis that require more rigorous

study to better ascertain the safety effects for children. Randomized control trials would only be

possible where large-scale installation of new road design features is planned, and prospective

Page 117: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

101

longitudinal studies would require prohibitive lengthy follow-up time given the rare collision

outcome. The most feasible design may be a pre-post installation quasi-experimental design.

This design is more rigorous than a cross-sectional study and has been used previously to

examine the effects of pedestrian countdown timers on pedestrian collisions.252

Further spatial

analysis is also required to provide greater insight into collision locations.

5.7 Conclusions and Policy Implications

The study findings are particularly relevant, as active school travel policy is currently undergoing

changes in Canada and the U.S. In Ontario, Canada, the provincial government recently initiated

the “Stepping It Up” school travel planning program under its regional transportation plan,

intending to spend $200 million on active transportation infrastructure and research to improve

safety and achieve AST >60% for all schools.34,36

In the U.S., recent changes to the federal

transportation bill have eliminated SRTS as a separate funding program. Alternative funding

through the federal Highway Safety and Infrastructure Program now requires SRTS projects to

show both evidence of increasing active transportation and a reduction in collisions.179

These

changes provide an opportunity to incorporate safety evaluation into new policies.

Several important conclusions and implications have emerged from the study. Firstly, the

positive relationship between walking and collision rates was no longer significant after

controlling for the built environment. These results are encouraging for walking promotion

programs, suggesting that safety issues are concerned primarily with the built environment and

not the numbers walking. Secondly, design features related to road crossing exhibited the most

influential effects. To increase walking safety in children, focus should be on minimizing or

mitigating road crossings, as opposed to changing other factors such as land use, which may be

more applicable to adults. Finally, the mechanisms of how to mitigate road crossings for

children are not well understand, and well controlled research designs must be integrated into

SRTS program evaluation. Future policy designed to increase children’s active transportation

should be developed from strong evidence that addresses child pedestrian safety.

Page 118: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

102

5.8 Supplementary/Supporting Analyses

The following supplementary analyses were not included in the published manuscript but support

the study findings.

Collision Rates 5.8.1

Figure 5S-1 portrays the distribution of collision rates at the 118 study schools. Ninety schools

(73%) had <10 collisions/10,000/year. There were two outlier schools with 59.1 and 78.2

collisions/10,000/year. The median collision rate excluding outliers was 5.6/10,000/years (range

0 to 27.2 collisions/10,000/year). Thirteen schools (11%) had no collisions.

Pedestrian Action During Collision 5.8.2

The majority of collisions in this age group occurred when the child was crossing the road,

including when at a pedestrian crossover (53.2%), followed by when running into the road (18.6

%, Figure 5S-2). Six percent of collisions occurred when a child was crossing at a pedestrian

crossover and 6% when coming from behind a parked vehicle. Finally, 4% occurred on a

sidewalk or shoulder of a roadway.

Predicted Values 5.8.3

An example illustrating predicted collision rates at different levels of multi-family dwelling is

presented in Figure 5S-3. Multi-family dwelling density was a continuous variable in the model,

but for ease of interpretation, it was set at defined levels within the range observed in the data as

identified in the box plot. In the City of Toronto, multi-family dwelling density ranged from

0.04/1000m2 to 5.6/1000m

2, with the median at approximately 1/1000m

2. Only 7 schools had

values that were 1.5 times above the interquartile range (IQR). Estimated predicted collision rate

Page 119: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

103

values are presented at the different levels of multi-family dwelling density, with each of the

other continuous covariates held at its mean, stratified by school crossing guard (a dichotomous

variable).

With each additional multi-family dwelling per 1000m2, the predicted collision rates decreased

by approximately 16%. For example, the predicted collision rate with 1 multi-family dwelling

was 7.8/10,000/year which decreased to 6.5 /10,000/year with an additional multi-family

dwelling/1000m2 when a school crossing guard was present. Rates were lower where no school

crossing guard was present, with 5.4/10,000/year collisions predicted with 1l multi-family

dwelling. The graph illustrates that at a multi-family dwelling density of 4/1000m2 which is at

the higher end observed, predicted collision rates were almost half than was predicted at

1/1000m2 multi-family dwelling (4.6 versus 7.8/10,000/year), with a school crossing guard. This

illustrates the strong association between fewer child pedestrian-motor vehicle collisions and

areas of higher density housing.

Sensitivity Analysis 5.8.4

5.8.4.1 Residual Diagnostics

Residual diagnostics were conducted with final models to identify outliers using Cook’s d,

leverage and Pearson’s standardized betas. A comparison of the final model and a model

excluding an additional 5 schools in addition to the 2 already excluded (n = 111), are presented

in Table 5S-1. IRRs were similar for all variables and effect direction was the same, with the

exception of the walking exposure. Walking exposure switched from a negative to a positive

association with collisions; however, the association was not significant.

5.8.4.2 School Travel Time Collisions

An analysis was done of collisions restricted to those occurring only during school travel times

(7:30-9:00 am, 11:30 am-1:00 pm, 3:00-5:00 pm, weekdays, September-June, Table 5S-2).

Page 120: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

104

Although there were reductions in the magnitude of effects, effect direction was similar to the

full model with less precise estimates due to the smaller sample size.

5.8.4.3 Alternative Collision Data Years

Sensitivity analysis was conducted using 5 years (2007-2011) and 7 years (2005-2011) of

collision data (Table 5S-3). Similar models resulted to that obtained using the full 10 years of

data. Although there were reductions in the magnitude of effects, effect direction was similar to

the full model with less precise estimates due to the smaller sample size.

5.8.4.4 Alternative Outcome

An alternative collision rate was calculated and expressed by collision per total school population

as opposed to per population of 4-12 year olds living in the school boundary. A comparison of

the results of the original outcome and this alternative are presented in Table 5S-4. Very similar

effects were found, with the exception of LOI, which showed a stronger positive association with

collision with the alternative collision rate per school population.

Page 121: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

105

5.9 Supplementary Tables

Table 5S- 1: Correlates of child pedestrian collisions in adjusted analysis for all schools

and excluding 7 outlier schools (IRR, 95% CI).

Component Variable All Schools

(n = 116)

Excluding outliers

(n = 111)

EXPOSURE Walking to school rate 0.84 (0.29, 2.46) 1.20 (0.44, 3.23)

BUILT

ENVIRONMENT

Density

Multi-family dwelling

(#)/1000m2

0.84 (0.73, 0.96)

0.79 (0.69, 0.92)

Design Traffic light (#)/km roads 3.20 (1.89, 5.41) 4.16 (2.56, 6.75)

Design School crossing guard

present

1.45 (1.09, 1.91) 1.31 (1.01, 1.70)

Design Traffic calming km/10 km

road 1.31 (1.06, 1.63)

1.34 (1.10, 1.63)

Design One way streets km/10 km

road

1.19 (1.03, 1.36) 1.19 (1.05, 1.35)

SOCIAL

ENVIRONMENT

School learning

opportunities index (LOI)

2.36 (1.39, 3.99) 2.37 (1.45, 3.86)

Page 122: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

106

Table 5S- 2: Correlates of child pedestrian collisions in unadjusted and adjusted models

for all collisions and those restricted to school travel times.

Unadjusted Model:

All

(456 collisions)

103 schools

School travel time

collisions

(214 collisions)

83 schools

COMPONENT VARIABLE IRR

(95% CI)

IRR

95% (CI)

OUTCOME Child pedestrian collisions

EXPOSURE Walking to school rate 3.47

(1.15, 10.47)

3.14

(0.82, 12.01)

Adjusted model:

OUTCOME Child pedestrian

collisions

EXPOSURE Walking to school rate 0.84

(0.29, 2.46)

1.73

(0.40, 7.43)

BUILT

ENVIRONMENT

Density

Multi-family dwelling

(#)/1000m2

0.84

(0.73,0.96)

0.73

(0.60, 0.89)

Design Traffic light #/km roads 3.20

(1.89,5.41)

3.17

(1.56, 6.48)

Design School crossing guard

present

1.45

(1.09,1.91)

1.22

(0.83, 1.79)

Design Traffic calming km/10 km

road

1.31

(1.06, 1.63)

1.11

(0.82, 1.51)

Design One way streets km/10 km

road

1.19

(1.03, 1.36)

1.15

(0.94, 1.40)

SOCIAL

ENVIRONMENT

School learning

opportunities index

2.36

(1.39, 3.99)

1.85

(0.92, 3.76)

Page 123: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

107

Table 5S- 3: Correlates of child pedestrian collisions in unadjusted and adjusted models

for 10 years, 7 years and 5 years of collision data.

Unadjusted Model:

10 years

(2002-2011)

(481

collisions)

105 schools

7 years

(2005-2011)

(291

collisions)

98 schools

5 years

(2007-2011)

(n = 183)

85 schools

IRR

(95% CI)

IRR

(95% CI)

IRR

(95% CI)

Outcome Child pedestrian

collisions

Exposure Walking to school rate 3.47

(1.15, 10.47)

2.27

(1.99, 2.58)

1.59

(1.40, 1.81)

Adjusted Model:

Outcome Child pedestrian

collisions

Exposure Walking to school rate 0.84

(0.29, 2.46)

1.01

(0.38, 3.63)

0.66

(0.15, 2.92)

BUILT

ENVIRONMENT

Density

Multi-family dwelling

(#)/1000m2

0.84

(0.73, 0.96)

0.84

(0.72, 0.99)

0.80

(0.66, 0.96)

Design Traffic light #/km roads 3.20

(1.89, 5.41)

2.72

(1.45, 5.07)

2.87

(1.38, 5.10)

Design School crossing guard

present

1.45

(1.09, 1.91)

1.36

(0.97, 1.90)

1.34

(0.91, 1.97)

Design Traffic calming km/10 km

road 1.31

(1.06, 1.63)

1.14

(0.87, 1.47)

1.14

(0.84, 1.54)

Design One way streets km/10 km

road

1.19

(1.03, 1.36)

1.19

(1.01, 1.40)

1.17

(0.96, 1.42)

SOCIAL

ENVIRONMENT

School learning

opportunities index

2.36

(1.39, 3.99)

1.87

(1.01, 3.49)

2.03

(1.00, 4.13)

Page 124: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

108

Table 5S- 4: Correlates of child pedestrian collisions and walking to school in adjusted

analysis using school populations as alternative denominator (IRR, 95% CI).

Component Variable Denominator:

Ages 4-12 living

in school

boundary

Denominator:

School

population

Exposure Walking to school rate 0.84

(0.29, 2.46)

0.84

(0.25, 2.82)

BUILT

ENVIRONMENT

Density

Multi-family dwelling

(#)/1000m2

0.84

(0.73, 0.96)

0.88

(0.76, 1.02)

Design Traffic light #/km roads 3.20

(1.89, 5.41)

2.92

(1.61, 5.29)

Design School crossing guard

present

1.45

(1.09, 1.91)

1.54

(1.12, 2.12)

Design Traffic calming km/10 km

road 1.31

(1.06, 1.63)

1.47

(1.14, 1.88)

Design One way streets km/10 km

road

1.19

(1.03, 1.36)

1.12

(0.96, 1.31)

SOCIAL

ENVIRONMENT

School learning

opportunities index (LOI)

2.36

(1.39, 3.99)

4.28

(2.36, 7.77)

Page 125: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

109

5.10 Supplementary Figures

Figure 5S-1: Distribution of collision rates/10,000/year within 118 study school

boundaries. Ninety schools (73%) had <10 collisions/10,000/year. There were two outlier

schools with 59.1 and 78.2 collisions/10,000/year. The median collision rate excluding outliers

was 5.6/10,000/years (range 0 to 27.2 collisions/10,000/year).

Figure 5S-2: Top 5 pedestrian actions at time of collision (n =481). More than half

of the collision occurred when crossing the road both with/without a pedestrian

crossover (53%).

Page 126: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

110

75th percentile (2.0)

Median (1.0)

25th percentile

Maximum (5.6)

Minimum (0.04)

Figure 5S-3:5Predicted collision rate/10,000/year by multi-family dwelling density. In the City of Toronto, multi-family dwelling density ranged from 0.04/1000m

2 to

5.6/1000m2, with the median at approximately 1/1000m

2. With each additional multi-

family dwelling/1000m2 the predicted collision rates decreased by approximately 16%.

Collision rates were lower when no crossing guard was present. Collision rates were

almost half at the higher end of multi-family dwelling (i.e., 4/1000m2) compared to the

median.

Multi-family dwelling density/1000m

2

Page 127: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

111

5.11 Detailed Methods

The following detailed methods were not included in the published manuscripts due to space

limitations.

Data Sources 5.11.1

All variables are classified according to their data source, variable type and source year in Table

4-S1.

5.11.1.1 Observational Study

A pilot study was conducted in May, June 2010 in 22 schools, to determine the feasibility of

observational counts outside of Toronto elementary schools. Schools were randomly selected,

stratified by census tract socio-economic status. Two trained observers per school counted

transport mode to school (Appendix E). Observers rated their confidence of the counts. Five

schools had counts redone on a different day to assess test-retest reliability. The Pearson’s

correlation coefficient was .99 indicating high correlation between the tests. The mean

confidence in the counts was 90% (SD 10%). The mean proportion of children observed

walking at the schools was 66%, range=30% - 95% (SD 20.1). It was concluded that direct

observation was a feasible method of measuring walking to school.

For the main study, 12 observers were trained to count transport mode to school excluding

school bus outside of 118 JK to grade 6 Toronto elementary schools. Observers were instructed

to stand on public property. Data was collected over 40 school days in May and June 2011. Four

teams of observers covered different areas of the city; with two observers doing the observational

counts per school. Each team was assigned a team leader. Training by the supervisor (i.e., PhD

defendant) included a 3 hour staff orientation and a site visit by the supervisor in the first week

for all teams. Each team leader was required to do a daily check-in with the supervisor either by

Page 128: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

112

telephone or email, and each team member was required to do weekly check-ins with supervisor.

Double data entry was done of all observational counts (primary by team leader, and secondary

by research assistant). A total number of 23,157 students were counted. All data entry was

reviewed on a daily basis by the supervisor.

5.11.1.2 Site Survey

The site survey was a checklist adapted from the School Site Audits, from the Delaware

Department of Transportation,222

after an extensive literature review of other checklists. The

Delaware site audits were developed to assess the school environment to enable the

establishment of Safe Routes to School (SRTS) programs. The survey was designed to examine

the traffic environment directly around each school during morning drop off time (Appendix F).

Items were selected based on an extensive literature review.

A pilot study of the site survey was conducted in May, June 2010 in 22 schools, to determine the

interrater reliability of items. Two trained observers conducted the surveys at each school. Items

that had a high agreement between raters (>80%) were retained. Feedback from observers

contributed to new/adapted items included in the checklist used for the full study.

For the full study, school site audits were conducted by two observers at each of the 118 schools

during school drop-off time. Observers were trained and monitored as described in Section

4.8.1. Observers were requested to fill in aerial view maps of the school sites with specified

features and cross-checked with the site surveys. Site audits items related to the traffic situation

around the school during school drop-off times, and whether or not there was a school crossing

guard present. Only adult guards employed by Toronto Police Services immediately surrounding

the school were identified as school crossing guards. Vehicle speed and traffic volume were

measured by a third observer along a 25m stretch of road within 150m of the school that had

been identified by the school as a roadway that many children use for walking/cycling en route to

the school with higher speed traffic. Average vehicle speed and traffic volume was measured

over a 20 minute period as children arrived at the school using manual short-based methods.

(Appendix G)214,215

Page 129: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

113

Double data entry was done of all site surveys (primary by team leader, and secondary by

research assistant). All data entry was reviewed on a daily basis by the supervisor. Maps were

checked to ensure they corresponded with the data obtained via the site survey.

5.11.1.3 Canadian Census

Data were obtained from the 2006 Canadian census by DA. Where no 2006 data was available

for the DA due to high non-response (as per Statistics Canada), data were obtained either from

the 2011 (preferred, if available), or from the 2001 census. If unavailable from any of the

censuses, averages of neighbouring DAs (i.e., shared a border) were applied to the missing DA.

5.11.1.4 City of Toronto

Child pedestrian collisions, ages 4-12, were extracted from motor vehicle collision reports filed

by the Toronto Police Service from 2002-2011 and obtained from City of Toronto transportation

department. The file included the geographical coordinates of the collision.

Built environment data came either from Transportations Services, or from the Open Data

website, and were generally in the form of ESRI Shapefiles (please see mapping section below

for details), with the exception of recreational facilities. A flat file of addresses of recreational

facilities was obtained and geocoded.

Toronto Centreline data was obtained from the City of Toronto’s Open Data website, and was

used to create all road network features. Centreline data consists of linear features representing

streets, highways, walkways, rivers, railways, highways and administrative boundaries within the

City of Toronto. For the purposes of these analyses, the Centreline data was restricted to just

roadways and laneways, from which the individual road type variables were created.

Intersection nodes were created in ArcMap. Walkways and trails were extracted and combined

into an independent feature.

Page 130: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

114

5.11.1.5 Toronto Police Services

A flat file of intersections where crossing guards were located was obtained for the school year

2010/2011 from the School Safety Patrol Program Co-ordinator, Traffic services, Toronto Police

Services.

5.11.1.6 Toronto District School Board (TDSB)/Toronto Catholic District School

Board (TCDSB)

School boundary ESRI Shapefiles and information regarding school population were obtained

from the TDSB, Planning Division. School boundary ESRI Shapefiles for the Catholic schools

were obtained from the TCDSB, Student Transportation Services. Other variables related to

school demographics were obtained from the TDSB website.

5.11.1.7 Municipal Property Assessment Corporation (MPAC)

The Municipal Property Assessment Corporation classifies and assesses properties in Ontario. It

is a non-for-profit corporation funded by all 444 municipalities in Ontario.223

An inventory of

2011 assessment parcels was obtained for the municipal boundaries studied in this analysis (i.e.,

Etobicoke, York, North York, Toronto, East York, and Scarborough). Each parcel has a unique

15 digit assessment role number. MPAC data includes 3 digit property codes at the assessment

parcel level reflecting specific land uses. There are six overall categories which were used for

this analysis, excluding farms (200 series) which were not relevant in the City of Toronto. The

coding is as follows: 100’s- Vacant Land, 300’s – Residential, 400”s – Commercial, 500’s –

Industrial, 600’s = Special and Exempt (i.e., institutional).

5.11.1.8 Teranet (via licensing from the University of Toronto)

A Teranet Shapefile containing digital property ownership parcel polygons was obtained from

The University of Toronto, Map and Data Library. Teranet owns and operates Ontario’s

Page 131: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

115

Electronic Land Registration System (ELRS) and is the exclusive provider of online property

search and registration in Ontario.224

Each parcel has a 15 digit unique assessment role number.

The Teranet Shapefile was linked with the MPAC data by role number to enable mapping of land

use.

Mapping 5.11.2

All mapping was conducted using Arcmap software10.0.225

The majority of the data came in the

form of ESRI Shapefiles. An ESRI Shape file refers to a set of several files, which is a format

for storing geometric location and associated attribute information.225

The format supports point,

line and area (polygon) shape features. Each map layer representing different feature Shapefiles

must be projected to share the same map projection. Map projections are required to transform

the three-dimensional earth’s surface to create a flat map. The City of Toronto map was created

for this analysis utilizing a NAD_1983_UTM_17N projection (North American Datum).

Data that were not in ESRI Shapefile format were collision reports, MPAC data, crossing guards,

recreational facilities, school and school demographic information. Collision reports were

mapped using the longitudinal and latitudinal coordinates provided in the data. The MPAC data

was linked via the 15 digit assessment role number and mapped using the Teranet Shapefile.

Crossing guard intersections, recreational facilities addresses and school address locations were

geocoded and mapped using Arcmap and the North American Address Locator (Can_streets,

Can_StreetName). School demographic information was linked to the school.

5.11.2.1 Spatial Analysis

5.11.2.1.1 Area Interpolation- Polygon in Polygon Areal Weighting

Polygon in polygon areal weighting (also known as area-weighted proportionate measurement)

was used in these analyses to estimate unknown Canadian census variable values in the school

boundaries, using the known values in the census dissemination areas (DAs). DAs were mapped

onto school boundaries and the fraction of the dissemination area falling within the school

Page 132: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

116

attendance boundary identified using an intersect analysis, was calculated. The variable in

question (e.g. population), assumed to be evenly distributed throughout the DA, was multiplied

by the fraction of the DA falling within the school boundary. The population estimates from

these partially contained DAs were added to the populations of the fully contained DAs to obtain

the estimates within each school boundary. Similar techniques were employed by Braza et al.

and Falb et al. to assign census tract/blocks populations to buffers surrounding schools based on

either circular buffers or pedestrian catchment areas along the street network.123,212

5.11.2.1.2 Buffer Analysis

Point and line features frequently fell on the boundaries between two study schools. Also, some

point coordinates were slightly off the road network likely due to slightly inaccurate geocoding

of the feature. Twenty-five meter buffers were drawn around all school attendance boundaries to

capture point and line features (e.g. collisions, traffic lights, traffic calming). Features that fell

within the 25m buffers that overlapped with another school boundary were assigned to the

closest school, according to Euclidean straight line distance.

5.11.2.1.3 Network Analysis

Networks represent possible routes from one location to another and consist of a system of

interconnected elements; for example lines and points. The network dataset was built using the

reduced Centreline dataset described above. Buffers were created of 1.6 km street network

distances around the schools using the ArcMap network analysis tool, to assess the proportion of

the school boundaries within walking distance of the school.225

Statistical Analysis 5.11.3

5.11.3.1 Negative Binomial Regression

Poisson regression modeling was initially conducted of the rate of children walking to school as

measured by observed counts of children with an offset of numbers of children observed. The

resulting deviance of the model was examined for evidence of over-dispersion of the response

variable. If correction is not done for overdispersion, standard error of the parameter estimate is

Page 133: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

117

overestimated and can result in something seeming significant when it’s not.220

Over-dispersion

frequently occurs in count data. Deviance has an approximate chi-square distribution with n-p

degrees of freedom, where n is the number of observations and p is the number of predictor

variables (including the intercept), and the expected value of a chi-square random variable is

equal to the degrees of freedom (DF). If the model fits well the deviance/DF should be

approximately 1. Initial examination of the model fit with one predictor found the ratio of

deviance to DF was 4.48. The deviance/DF ratio was consistently over 3 with the addition of

predictors. Negative binomial regression modeling was then used with a resulting deviance/DF

ratio close to 1 and never above 1.3. Further support of the use of negative binomial modeling to

correct for over dispersion was a significant dispersion value over 0 in the resulting models (i.e.,

variance>mean). There were no 0 values in the response data, so zero-inflated models were not

further investigated.

5.11.3.2 Forward Stepwise Manual Regression

A p value of < 0.2 in the univariate analysis identified variables significantly associated with the

outcome to include in forward manual stepwise regression as described by Hosmer and

Lemeshow.218

At each stage of the modeling, the variables included were re-examined, and

dropped if not significantly related to the outcome. With each addition of a new variable into the

model, variables previously found to be negative were retested.219

Hosmer and Lemeshow

recommend the use of forward stepwise techniques to obtain the most parsimonious model as

inclusion of greater numbers of variables increases instability, and the results are more dependent

on the observed data and less generalizable.218

5.11.3.3 Confounding

Confounders were identified using change in estimate criterion. If the introduction of the

variable resulted in a >10% change in the estimate of the exposure, the variable was considered a

confounder of the relationship between the exposure and the outcome.218,238

Page 134: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

118

5.11.3.4 Effect Modification (Interactions)

Interactions between road traffic design variables were assessed in these analyses, as there has

been previous evidence of interactions between different road design features related to walking

to school outcomes. Giles-Corti et al., found a highly significant interaction between pedestrian

network connectivity and traffic volume, with children less likely to walk to school regularly in

areas with high pedestrian network connectivity and high road volume, compared to those with

low connectivity and low volume.134

Effect modification was initially assessed by stratifying the

final models by each of the road design variables and identifying whether the Incidence Rate

Ratios (IRRs) differed by strata. Stratification was done by tertiles for continuous variables. The

potential interactions were subsequently confirmed by including the interaction term into the

final model and determining that the p value was < .05. As interaction terms are difficult to

interpret, and their use does not permit further interpretation of main effects, it was decided

instead to present the models stratified by the variables identified as effect modifiers.

Page 135: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

119

6 General Discussion

6.1 Summary

Child pedestrian-motor vehicle collisions continue to be a major health issue worldwide.

Declining trends in high income countries are thought to result from fewer children using

walking as a means of transportation. Numerous benefits from walking have been identified

including those related to physical and social health, and the environment. Policies and

programs to increase children walking to school are being developed in many high-income

countries. Insufficient attention has been paid to the effects of increased walking to school on

child pedestrian-motor vehicle collisions, and to determine what constitutes a safe walking

environment for children. Although it has been established that features of the built environment

can influence both walking to school and child pedestrian-motor vehicle collisions, there has

been little research on the role of the built environment on the relationship between walking to

school and pedestrian-motor vehicle collisions. The three studies described in Chapters 3, 4 and

5 of this dissertation addressed these gaps in knowledge and present evidence emphasizing the

importance of considering child pedestrian-motor vehicle collisions together with the built

environment in future policies and evaluations of walking initiatives.

The main findings of each of the chapters are presented below according to the specific

objectives presented in Section 1.3

1. To systematically review the literature on the relationship between the built environment,

walking to school and child pedestrian-motor vehicle collision rates (Chapter 3).

Chapter 3 presented the results of the systematic review of the literature. The main findings of

this chapter were as follows:

No studies addressed both walking and child pedestrian-motor vehicle collisions in

relation to the built environment.

Page 136: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

120

Only traffic calming and presence of playgrounds/recreation areas and traffic controls

were consistently associated with more walking to school and fewer pedestrian-motor vehicle

collisions in the literature.

Higher pedestrian volume, population and road density, schools, urban location, land use

mix, proximity to services/facilities and crosswalks were associated with more walking, but with

less safety.

The majority of built environment factors either had inconsistent associations (e.g., some

studies showing null and some showing positive associations) with walking or injury or had not

been tested for either one or the other of the outcomes.

All studies were observational and the majority had cross-sectional designs.

The methodological quality of the studies was generally low. There were many

inconsistencies in methodology (e.g., different methods of both outcome and exposure

measurement), analysis (e.g., adjusted and unadjusted) in both the walking and the child

pedestrian-motor vehicle collision literature.

Most studies used parent or child-reported walking rates which have inherent biases that

may affect study results. Many studies also used reported built environment measures rather

than more objective measures.

2. To estimate the proportion of observed children walking to school in the City of Toronto

(kindergarten to grade 6, Chapter 4).

The results of a large observational study of children walking to elementary schools in the City

of Toronto were presented in Chapter 4.

The average proportion of children observed walking to school, was higher than expected at

67%, with a high degree of variability (range 28-98%).

3. To determine how built environment features are related to the proportions of children

walking (Chapter 4).

Page 137: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

121

Higher child population density, pedestrian crossover density, traffic light density,

intersection density, presence of school crossing guards and primary language other than English

were positively correlated with walking.

Presence of a school crossing guard was an effect modifier and reduced the influence of

other features on walking

4. To estimate child pedestrian-motor vehicle collision rates in the areas surrounding

elementary schools in the City of Toronto (Chapter 5).

Chapter 5 extended the observational study to examine child pedestrian-motor vehicle collision

as the outcome, and the observed walking proportion as the primary exposure.

There were 481 collisions with a mean collision rate of 7.4/10, 000 children per year (SD

6.7). Rates ranged from 0 to 27/10,000/year across 118 elementary schools.

5. To determine how features of the built environment influence the relationship between

proportion of children walking to school and child pedestrian-motor vehicle collisions

(Chapter 5).

In the unadjusted analysis, higher proportions of children walking to school were

associated with higher rates of pedestrian-motor vehicle collisions.

In the analysis adjusted for built and social environment features, higher proportion of

children walking to school were no longer associated with higher rates of collisions.

Higher densities of multi-family dwellings were associated with lower collision rates.

Higher densities of traffic lights, traffic calming, one way streets, presence of school

crossing guard and lower school SES were associated with higher collision rates.

Densities of multi-family dwelling density, traffic calming and one way streets, school

crossing guards and school SES were confounders of the relationship between walking to school

and child pedestrian-motor vehicle collisions.

Page 138: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

122

Traffic light density was an effect modifier for walking exposure and collisions.

Increased walking was associated with higher collision rates when traffic lights densities were

low and was no longer associated with collisions where there were medium or higher densities of

traffic lights.

Over 50% of collisions occurred when children were crossing the road. Several of the

built environment factors associated with higher collision rates were related to locations where

children cross roads (e.g., traffic lights, school crossing guards).

6.2 Unifying Discussion

The built environment influences both walking to school and child pedestrian-motor vehicle

collisions, which has important implications for both walking promotion and injury prevention.

When changes to the built environment are made, it is necessary to assess both walking and

collisions outcomes in order to adequately determine the impact of the changes. It is not

sufficient to simply assess walking rate outcomes as has been commonly done for walking

promotion programs. The results suggest that more children walking to school does not

necessarily lead to increased collisions. These findings are promising for the implementation of

walking to school programs, as they suggest that collision rates may not be the direct result of

increased exposure to traffic, but that they are strongly influenced by the environment in which

walking occurs.

Many built environment variables were examined for their association with walking to school

and child pedestrian-motor vehicle collisions in this thesis. Table 6-1 presents a summary of

built environment variables tested and displays associations with walking to school and child

pedestrian-motor vehicle collisions as reported in the literature as the findings from this thesis.

Page 139: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

123

Table 6-1: Summary table of built environment variables associated with walking to school

and child pedestrian-motor vehicle collision from the literature and from the study

analyses.

Built EnvironmentVariables Walking to school Child pedestrian motor

vehicle collisions

Literature Study Literature Study

Density

Child population density +/NS + + NT

Multi-family dwellings NT NT + -

Diversity

Mixed land use +/NS NS (+/NS) NS

Design

Distance to school - NT (+) NT

Road class, traffic volume, traffic

speed

-/NS NS + NS

Parks/recreational facilities +/NS NS - NS

Pedestrian crossovers + + + NS

Sidewalks +/NS (-) NS + NS

Trails +/NS NS NT NS

Street connectivity /route directness - (+) NS NT NS

Intersection/blocks +/NS (-) + + +

One way streets NT NS + +

Traffic calming + NS - +

Traffic lights + + - +

School crossing guard NS + + +

NS = not significant; NT = not tested

Brackets ( ) indicate limited number of studies

Page 140: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

124

Density: 6.2.1

In the analyses reported in Chapters 4 and 5, total population, child population and multi-family

dwellings were highly correlated. Therefore, only one density variable was included in each of

the walking to school and the child pedestrian-motor vehicle collisions models. In general,

population density was correlated to walking to school and lower pedestrian-motor vehicle

collision rates. Previous studies have found that both child population density and multi-family

dwellings were associated with increased collisions.152,195,202,253-255

Child population density was

not tested in our study as population numbers were used to calculate collision rates. The positive

associations between multi-family dwellings and risk of pedestrian-motor vehicle were found in

two case control studies.195,202

In these studies, dwelling density was measured at the individual

level with no adjustment for walking exposure. When measured at the population level, as was

done in our study, higher population density areas surrounding schools may act as a proxy for

shorter distances to school with fewer road crossings, resulting in less traffic exposure and fewer

collisions.

Diversity 6.2.2

Land use mix has been associated with walking to school in the literature in some studies,

117,120,121,125,127 whereas our study and other previous studies have found no

association.107,127,132,136

In the literature, land use has been measured via parent report,120,127

field

survey117

and using databases and GIS.107,120,121,125,132,136

One issue in the studies using GIS to

measure land use (and other built environment variables) is that all measured land use within

different size buffer zones. For example, Kweon et al. used a 2 mile walk zone,125

Mcmillan

and Yarlagadda used a ¼ mile radius of the child’s home,117,136

and Panter used three different

buffer zones; within the neighbourhood (within 800m around child’s home), on the route to

school (within a 100m radius of the shortest route) and area immediately surrounding the

school.121

In this thesis, land use was measured within various sized school attendance

boundaries. Significant associations were found between land use mix and walking to school in

the Kweon, Mcmillan studies and Panter studies, but only for the area immediately surrounding

the school.117,125,136

Conversely, Panter found no significant associations within the

Page 141: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

125

neighbourhood and on route to school and Yarlagadda found no association within .25 miles of

home; 121,136

likewise, no significant association was found within the school attendance

boundaries in this thesis. The differences in these results emphasize the importance of

consistency in the geographic area for measurement of built environment variables to enable

comparison of results, as introduced in Section 2.6.1 and further described with respect to the

modifiable areal unit problem in the limitation section of 6.3.2.

The association between mixed land use and child pedestrian-motor vehicle collisions has not

been well studied. Only one study by Cloutier et al. reported a positive association between

mixed land use and child pedestrian-motor vehicle collisions as measured using GIS and an

entropy index within school catchment areas based on network distances. Clifton measured land

use within a census block group and found no association with collisions in school age

children.253

No association was found between the land entropy index used in Chapter 5 and

child pedestrian-motor vehicle collision.

Several explanations may explain these inconsistencies. Definitions and calculations of land use

mix vary between studies which may influence results. These include: ratio of commercial

properties to total area,253

calculation of an entropy index 125

and the proportion of street

segments with land use mix.117

Results may also be affected by the land use categories used. It

may be that child pedestrian-motor vehicle collisions and walking to school are sensitive to more

specific categories of land use than the 5 used in this thesis: residential, commercial, industrial,

institutional and vacant land. Cloutier et al. calculated an entropy index using 16 categories

which further broke down residential (high, med and low), commercial (small retail, shopping

center; office space), and vacant land (quarry, landfill site, green space, golf course, cemetery,

rural, vacant space) and institutional land uses (community service, public utility).151

Panter

used 17 specific categories of land use which specified more rural uses including; farmland,

woodland, grassland, uncultivated land, other urban, beach, marshland, sea, small settlement,

private gardens, parks, residential, commercial, multiple-use buildings, other buildings,

unclassified buildings, and roads.121

The variability of these categories also may indicate that the

association between land use mix with walking to school and collisions is highly context

specific. For example, the study by Cloutier et al. was located in Montreal, Canada which is a

city of approximately 1.5 million, located on an island.151

The Panter et al. study was located in

the county of Norfolk, United Kingdom, which has approximately 900,000 people, and is largely

Page 142: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

126

rural with a low density population.121

The land use categories used by these two studies portray

the different land uses that characterize these two geographic locations. Our study was

conducted in Toronto, Canada; a much newer and larger city than Montreal, with a population of

2.8 million, which encompasses many densely populated areas. The types of land use that are

important for this large urban center may be substantially different than those in Montreal and

Norfolk which may ultimately affect the relationship between mix and walking to school and

child pedestrian-motor vehicle collisions.

Finally, it was also difficult to determine the validity of the data sources used, as only the study

by Cloutier et al. specifically stated the source of the land use data (City of Montreal Geomatic

Division).151

In our study, MPAC data was used, which is the most valid source of land use data

available for the City of Toronto, and which has not been previously used in studies of either

walking to school or child pedestrian-motor vehicle collisions. The mixed results reported

between child pedestrian-motor vehicle collisions, walking to school and land use mix, as well as

the paucity of evidence for child pedestrian-motor vehicle collisions, emphasizes the need to

examine these relationships further, based on valid and reliable data.

Design 6.2.3

6.2.3.1 Distance to School

Distance to school has been identified as the strongest determinant of walking to school in the

literature. It is therefore important to control for distance when estimating rates of children

walking to school. As mentioned in Chapter 4, reported proportions of children walking to

school can be very different if distance not considered or controlled for. For example, estimates

of proportions of school age children walking in Canada when distance was not considered were

substantially less than what found in this thesis. In the 2004 Canadian National Transport

Survey, 50% of school age children reported having walked to school36

and in the 2006

Transportation Tomorrow Survey, 48% of Toronto children age 11-13 years reported walking to

school.37

In a study by Gropp et al. which controlled for distance by including only respondents

that lived in an urban setting and reported living within 1 mile (1.6 km) of their school, it was

Page 143: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

127

estimated that 63% of youths used AST, which was closer to the 67% measured in this

analysis.157

Distance was not entered as an individual variable in our study analyses, as individual level data

were not available. Distance to school was controlled for by counting only children arriving

either by walking, other active means or by motorized vehicle excluding those arriving by bus.

The majority of children included in the counts lived within a 1.6 km walking distance to school,

as the majority of elementary school-age children living further away are eligible for the school

bus.211

The school boundaries in the City of Toronto currently are generally the right size for

walking to school, with 75% being less than 1.3 km2. On average, 95% of the roads within the

boundaries are within 1.6 km walking distance from the school along the road network. Schools

that had < 70% of their road networks within 1.6 km walking distance either contained an

industrial complex, hydro fields, a large park, railway tracks or had a split boundaries (i.e., two

or more attendance areas not adjacent to one another, usually in the case of a large apartment

building).

Despite the inclusion of only children who lived within walking distance of the school, there

were many children not walking to school and large variability in proportions walking between

schools. The findings indicated that there were built environment factors other than distance

which may have influenced walking to school. Other factors that were not measured which were

reported by school principals and field observers to influence walking included: train tracks and

major arterials bisecting the school boundaries, nearby road construction diverting a large

amount of fast traffic near the school, and split boundaries.

There has been little attention paid to the relationship between the distance walked to school and

child pedestrian-motor vehicle collisions. A Canadian study reported evidence of a significant

dose-response relationship between (self-reported) longer school travel distances (i.e., >15

minutes walking or >5 minutes cycling) and active transportation injury, which included all types

of injuries (not restricted to collisions) occurring in the street/road/parking lot or while biking or

walking/running.44

However, when considering walking only, there was no significant change in

active transportation injury with longer school travel distances. Further investigation is required

to determine whether there is an association between distance walked and specifically, child

pedestrian-motor vehicle collisions.

Page 144: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

128

6.2.3.2 Design Features with No Significant Associations with Child

Pedestrian-Motor Vehicle Collisions

Several design features had no significant associations with child pedestrian-motor vehicle

collisions, which have had correlations previously reported in the literature. No significant

associations were found between proportions walking to school and child pedestrian-motor

vehicle collision rates and road type density (at the school boundary level), or vehicle speed and

traffic volume (on a roadway near the school). These factors have been established as important

negative correlates of children walking to school and as positive correlates with pedestrian-motor

vehicle collisions according to the literature.108,117,121,125,132,151,195,197,202,210,255

Several other design features also had null results, contrary to what was expected from the

literature. Parks/playgrounds/recreational space were not significantly related to either walking

or child pedestrian-motor vehicle collision outcomes in our analyses, whereas parks were

identified in Chapter 3 as one of the only variables in the literature to be positively associated

with self-reported walking and associated with less collisions.126,127,131,135,195,247,255,256

Pedestrian

crossovers were not associated with collisions , even though they were positively associated with

walking to school. Other studies have reported positive associations of pedestrian crossovers

with both walking and collisions.130,195

No association was found between sidewalks and

collisions, although there is some evidence that sidewalks are associated with increased walking

and pedestrian-motor vehicle collisions.31,84,112,116,125,126,196,197

Although there was a reported

positive association between trails and walking to school,120,133

there was no association found

with walking or pedestrian-motor vehicle collision rates in these analyses. Finally, no

association was found between collisions and street connectivity/route directness which is not

surprising as there have been no definitive relationships established with either child pedestrian-

motor vehicle collisions or walking to school. In previous studies, GIS measures of street

connectivity/direct route generally appear to be related to less walking,120,121,129

whereas

measures by parent report are correlated with more walking.112

Null correlations with collisions where significant correlations may be expected based on the

literature may be a result of several factors. Observed walking was used in our analyses rather

than reported walking. In addition, the collision models were all corrected for walking exposure

which may have affected the associations. Finally, traffic speed and volume were only measured

Page 145: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

129

on one road close to the school, which may have not been representative of the speed of traffic

throughout the school boundaries from where the children were walking and where the collisions

occurred. Although road type densities within the school boundary were also used as a proxy for

traffic speed and volume, this measure may not have been sensitive enough to pick up issues

related to children walking and pedestrian-motor vehicle collisions.

6.2.3.3 Design Features with Significant Positive Associations with Child

Pedestrian-Motor Vehicle Collisions

One way streets, traffic lights, school crossing guards, and traffic calming all were significantly

related to collision rates even after adjusting for walking exposure. This research confirmed the

negative effect of one way streets on pedestrian motor-vehicle collisions that was previously

reported in Hamilton Ontario, with a 2.5 higher pedestrian injury rate compared to two-way

streets.200

Traffic lights were related to more walking, similar to previous studies,121,128

and they

were significantly associated with collisions only in areas where there are low densities of traffic

lights. Previous research reported negative relationships between traffic lights and child

pedestrian-motor vehicle collisions; however, no adjustment was made for walking

exposure.246,247

The positive associations between collisions and school crossing guards and traffic calming were

surprising, as these features are designed to reduce traffic speed and/or protect pedestrians.

School crossing guards were positively associated with walking to school and reduced the effects

of other built environment features on walking. The relationship between school crossing guards

and walking to school has not been previously investigated. School crossing guards have been

similarly positively related to child pedestrian-motor vehicle collisions in a previous study by

Cloutier et al.151

The findings for traffic calming contrasted previous studies which have

reported positive associations with walking,121,130

but negative association with

collisions;100,193,194

however, again, no adjustment was made for walking exposure.

Page 146: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

130

Several hypotheses were presented in Chapter 5 to account for the counterintuitive positive

relationships found between these roadway design features and collisions. These hypotheses are

further expanded upon in the limitations section (6.3.2):

Features were indicators of more dangerous locations with higher volume traffic, and the

safety effect of the feature may be insufficient to reduce collision risk.

Road design features could potentially create hazardous traffic situations for children.

Reduced effectiveness of some road design features such as pedestrian crosswalks have

been shown for older adults.249

The effectiveness of these features specifically for

children has not been well studied.

There was poor sensitivity of ecological data, as the area level analysis used would not be

able to discern whether area-level environmental features accurately represented the

collision location or whether collisions were displaced.

There were temporal issues related to the cross-sectional nature of the data and the

potential for reverse causality.

More rigorous study of the effectiveness of these built environment design features is required to

better ascertain the safety effects for children.

6.3 Strengths and Limitations

Strengths 6.3.1

The findings of this thesis addressed major gaps in the literature regarding the relationship

between walking to school and child pedestrian-motor vehicle collisions. This has significant

public health importance as there has been much recent emphasis on increasing safe active

transportation in children. One of the major strengths this work was the use of objective

observational outcome walking exposure data. The importance of using objective measures of

walking as described in Chapter 3 was to reduce bias in outcome measurement. Accurate

Page 147: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

131

measurement of exposure is necessary to properly evaluate collision risk. This is the only study

(known to date), to examine pedestrian-motor vehicle collision risk of school children using

objectively measured observed walking exposure while controlling for built environment

correlates. A significant contribution of this work was also related to the examination of built

environment correlates of child pedestrian-motor vehicle collisions, while controlling for the

objectively measured walking exposure.

Data obtained from a wide variety of data sources were collated and mapped together using GIS

techniques, many of which have never been accessed previously for public health research in

Toronto. The resulting map represented a data-rich resource representing a wide variety of built

and social environment factors which could easily be updated and used for future research.

Although analysis for this thesis was conducted at the school attendance boundary level, other

geographic boundaries could easily be applied to the map. Research relationships have been

established, with the Toronto District School Board, the City of Toronto and Toronto Police

Services so that continued access to data is feasible.

Of particular note, was the establishment of a process to procure parcel level land use data from

the Municipal Properties Assessment Corporation (MPAC), which has not been previously used

to examine active transportation in Toronto. MPAC land use data provides up-to-date, accurate

and specific land-use classification that is not possible to obtain elsewhere. It is a valid and

reliable data source to use to analyze associations of land use with many health outcomes.

The large sample used in this study represented virtually all regular program JK-6 schools in

Toronto. The benefits of being able to access data available in the public domain (through

anonymous observational counts conducted on public property) are enormous, as all schools

could be included, and results of this study can be universally applied to all JK-6 schools in

Toronto. Recommendations made to the TDSB and the City, therefore, would have no

associated limitations related to potential bias of the sample.

Page 148: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

132

Limitations 6.3.2

Although one of the major strengths of the study was the use of objective observational walking

exposure data, this must also be recognized as a limitation for future studies. Collection of

observational data is time consuming and costly. Small studies have been conducted to

determine how well student reports of walking behaviour correspond to observed rates; however,

there is a need to examine this issue on a larger scale.69,70

Observer counts of walking to school can provide a more objective and accurate measure of

walking free from selection and social desirability bias and recall error of self-reported walking.

There are also challenges however, in determining the best location where accurate walking

counts can be conducted. Walking counts were generally only done in two locations surrounding

the school by two observers. Observers were placed at locations where the majority of children

arrived at school, and these locations were corroborated by the school principals whenever

possible. Although it was occasionally difficult to determine whether children walking to school

were dropped off further down the road and out of sight of the observers by a car, these children

were considered pedestrians as they were crossing roads to get to the school. If observers rated

low confidence in their counts, a second count was conducted one week later, with the addition

of a third observer as necessary. It must be noted, that the proportion of children walking to

school was the outcome of interest, and not the absolute numbers of walking. With these

accommodations, it was felt that the proportion of children arriving to school walking was well

represented.

Other limitations were related to when the observational walking data was collected. Data were

only collected generally on one day at each school. It may be that the particular day was unusual

for the school, in that there was a special event which affected transportation modes or the

number of children observed (e.g., an early morning track and field meet prior to the observation

period). However, attempts were made to avoid this by consulting with principals prior to the

observation day and reliability testing on a 10% subsample of the schools indicated that walking

rate estimates were reliable. Data were also only collected for walking to school but not the trip

home and there is evidence that different built environment characteristics are relevant on the trip

to and from school.122,137

Furthermore, although walking exposure was only measured on route

to school, non-school travel time collisions were included. The study’s intention was to examine

Page 149: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

133

factors related to child pedestrian collisions in Toronto, and not only those occurring during

school travel. As school is children’s most common walking destination walking to school is

considered the best proxy for their general walking activity.130,230,231

Consistent evidence also

exists that children who actively commute to school, walk more in general.54

The inclusion of

non-school travel time collisions when school crossing guards were off-duty could potentially

have produced an artificial positive association between guards and collisions. However, when

restricting the data to school times, the directions of effects were maintained. Despite the

limitations of using walking to school as a proxy for children’s general walking, it is currently

the most feasible measurement of children’s walking exposure. Walking exposure has been

poorly dealt with in the past, and creative methods of measurement are needed to most accurately

evaluate pedestrian risk.

Traffic volume and speed are important factors related to both walking and collisions; however,

neither was significant in these analyses. This is likely due to limitations related to

measurement, as traffic volume and speed data is lacking in Toronto, particularly in residential

neighbourhoods around schools. Although there are some traffic and speed counts available,

these are limited to major roadways. In this study, volume and speed were measured only along

one road close to the school, which was identified by the principal and the observer to have

higher speed traffic and along which many children walk to school. As mentioned previously, it

may be that the traffic along this road was not representative of the traffic throughout the school

boundary, and along roadways where children walk to school. Road type density throughout the

school boundary was also used as a proxy of traffic volume and speed; however, it may be that

this measure was not sensitive enough to pick up issues along children’s specific walking routes

to school. Despite this, the majority of features found to be positively correlated with walking in

these analyses (i.e., traffic lights, school crossing guards, pedestrian crossovers and intersection

density) and collisions (traffic lights, school crossing guards, traffic calming and one way

streets), are likely correlated with higher traffic volume or speed. Many are either features of

busier roads or have been put in place in response to busier or faster traffic.

Only objective built environment data were used for this analysis. Parent and child perceptions

of the built environment and injury risk are important to consider in studies of walking behaviour

in children, as the decision to walk to school are based on perceptions of the risks and benefits of

walking.66,71,85

It has been suggested that studies include both objective and self-reported

Page 150: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

134

measures of the built environment, from the parent and child perspective.66,71,85

Similarly, only

quantitative studies were reviewed for the papers included in the systematic review in Chapter 3.

This may have resulted in the omission of papers examining perceived issues regarding the built

environment. However, the majority of published work done is this area has been quantitative in

nature.

The ecological nature of the data also imposed some limitations on the results and

interpretations. There could be an issue of the modifiable areal unit problem (MAUP), which is

a source of bias in spatial analysis with respect to scaling and zoning issues. Scale effects refer

to differences in results due to the size of spatial units used for measurement.105

For example,

results at the smaller dissemination area level may be less stable than those at the census tract

level. Zonal effects refer to the differences in how space is partitioned, even if the space is the

same scale.105

Inconsistencies in the relationship between the built environment and school

travel literature may be the result of different spatial units used in different studies.105

It is

possible that the results of the analyses presented in this thesis do not always correspond to those

found in other studies due to the issues of MAUP, as different environmental features may

influence school travel mode choice at different scales.105

The importance of a strong

behavioural justification of the selection of spatial units has been emphasized, which considers

the interaction between human behavior and the surrounding physical environment.105

The

school attendance boundary was selected as the unit of analysis for the present work, as it

generally represents the street network within walking distance of the school, as portrayed in

Chapter 3, and is the scale relevant for policies directed at interventions by principals, school

boards and trustees, and governments.

There were also several other potential issues related to the ecological nature of the data.

Collisions occurred at a specific location, whereas the built and social environment data were

measured area-wide and may not have represented the environment at the collision location.

Area-level data also would not identify if the presence of a built environment feature either

displaced collisions or was a marker for a more dangerous traffic environment elsewhere within

the school boundary. The school attendance boundary unit of analysis would not be sensitive

enough to pick up these issues. Individual level car ownership, gender and distance to school

which have all been found to be correlated with walking to school, also were not accessible

using ecological data. However, it was described in Chapter 4 that distance was unlikely to have

Page 151: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

135

had a large influence on results, as children included in the proportions walking likely lived

within walking distance of the school, as defined by TDSB transportation policy.211

It was felt

that ecological analysis of these data were most appropriate to answer the research questions, as

the focus was on the behaviour of a population, the surrounding environment around which the

behaviour occurred and the collision outcomes of the population as a whole. Results of these

analyses were ultimately intended to be applied on a policy rather than an individual level.

The cross-sectional design necessitated some prior assumptions which may also have influenced

the results. Data was combined from different time periods necessitating prior assumptions

which may have affected results. The aggregation of several years of collision data from 2002-

2011 was necessary due to the rarity of collision events. The first assumption was that walking

exposure measured in 2011 was representative of exposure throughout the 10 year collision data

period (2002-2011). This was reasonable, as there is evidence of stabilization of active school

travel prevalence from 2001 onwards, after an earlier period of sharp decline.35,37,251

The built

environment was also assumed to remain unchanged over the collision data period. Substantial

changes would not be expected in the built environment generally, as the study neighbourhoods

were well-established. However, this may not have been the case for features more easily

implemented, such as the installation of traffic calming or school crossing guards. The

possibility of reverse causality exists, with collisions occurring prior to the installation of these

features.

Finally, the generalizability of results is difficult to ascertain. The results would be applicable to

all regular program Toronto JK-6 elementary schools, as the vast majority of the schools were

included in the study. It might be expected that the results would be generalizable to other

schools encompassing similar age ranges in Toronto, excluding those who did not live in walking

distance to the school. Although specific findings related to the built environment may be

similar in other urban settings, there is much variability regarding attitudes and perceptions to

walking and safety from community to community. Even perceptions regarding how far is

considered an acceptable distance to walk may vary, especially in different climates and different

cultural groups.

Page 152: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

136

6.4 Policy Implications

The results of this work have important implications for Canadian policy initiatives designed to

increase walking to school and decrease pedestrian-motor vehicle collisions at the municipal,

school board and national levels.

Integration of Walking to School and Child Pedestrian-Motor Vehicle 6.4.1

Policies

This systematic review of the literature in Chapter 3 emphasized the importance of considering

both walking and pedestrian-motor vehicle collisions together, which has not traditionally been

done when developing programs to increase AST. The creation of The Road to Health document

by The City of Toronto Traffic Services, together with Toronto Public Health, represents a

developing awareness in Toronto of the value of integrating information regarding child

pedestrian-motor vehicle collisions and active transportation from a variety of different

sources.16

This report is a call to action, and should be considered a starting point. Follow-up is

necessary on the report’s recommendations regarding methods to facilitate effective action in

Toronto, and the need to make safe active transportation infrastructure more of a priority for

funding. The Road to Health report emphasizes that continued coordination and collaboration of

a variety of stakeholders is necessary in order to achieve safe active transportation to school.

This thesis responds to the message presented by the Road to Health, as it is the result of input

from a variety of different stakeholders to investigate safe active transportation in Toronto. The

results of this work will be shared with Toronto Public Health and the City of Toronto, Traffic

Services, to use as a baseline for planning future collaborative interventions.

Page 153: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

137

Identification of Evidence-Based Targets 6.4.2

6.4.2.1 Walking to School

When designing and evaluating policy effectiveness, it is essential that appropriate targets be

identified with specific consideration as to how they are to be measured. This thesis has

provided evidence to base appropriate targets for walking to school. The target AST rates

envisioned by Metrolinx’s The Big Move, was above 60% for all schools.173

However, there

was no indication in the report regarding how this target was set. In the Road to Health

document produced by Toronto Public Health and the City of Toronto, one of the methods

described to facilitate effective action was to develop quantitative goals, and examples of those

in other jurisdictions were provided.16

Although the importance of establishing goals was

emphasized, these were also not provided in the Road to Health document.

The results of these analyses provide evidence-based targets for walking to school in the City of

Toronto. In JK-grade 6 schools in the City of Toronto, 67% of children walked to school on

average, but the proportion was highly variable, ranging from 28-98%. Seventy percent of study

schools had walking proportions > 60%. The vision of 60% walking of children walking to

school as set by The Big Move, does appear to be realistic in this subset of schools. This rate

was specific to elementary school age children up to grade 6 who lived within walking distance

of the school, as defined by TDSB transportation policy. Policy objectives must define the

targeted school types and the age ranges, as walking proportions would be very different for

schools with older children with larger catchment areas and for schools with alternative

programming where more children are driven. As evident by the walking proportions obtained

in these analyses, it appears to be most appropriate to direct the goal of 60% or higher to

elementary schools with regular programming. Other targets need to be established for different

age groups and types of schools.

6.4.2.2 Child Pedestrian-Motor Vehicle Collision Targets

The Big Move transportation plan identified pedestrian safety as was another of its objectives,

but child pedestrian-motor vehicle collisions targets were not identified.173

Although the

Page 154: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

138

National Road Safety Strategy specified a collision target of 5 fatalities per 100,000, this target

was aimed at the general population.163

Other countries, such as the United Kingdom have a

general target as part of their Road Safety Strategies, but also a specific target directed at

children.257

In the UK, targets for 2010 were a 40% reduction in the number of people killed and

a 50% reduction in the number of children killed or seriously injured in road accidents, compared

to the baseline average of 1994-98. The UK Road Safety Strategy specified a separate target for

children because of the high rates of child pedestrian-motor vehicle collisions, the recognition of

the differential effects of some measures on adults and children, as well as the incorporation of a

range of child safety specific policies. The burden of child pedestrian-motor vehicle collisions in

Canada continues to be high, and both Canadian and municipal policies to protect pedestrians

should follow this example by setting targets for children to address the specific issues related to

this age group.

Thirteen schools (11%) in the analysis presented had no child pedestrian-motor vehicle collisions

over the 10 year period. Five of these schools had higher than average proportions walking.

These results indicate that it is possible in a large urban environment to have school locations

where children can walk to school and not be injured in car traffic. Zero tolerance should

therefore be the target of all policies to reduce child pedestrian-motor vehicle collisions.

The need for safer active transportation for children has been established at the municipal and the

provincial level. Next steps are definitive plans with specified walking and pedestrian goals.

Based on evidence from this thesis, a goal of at least 60% walking to all Toronto JK-grade 6

elementary schools, with no collisions involving school age children occurring within school

boundaries over a 10 year period should be the target.

Appropriate Outcome Measurement 6.4.3

None of the existing policy initiatives in Canada related to walking to school and child

pedestrian-motor vehicle collisions have specified the methods of outcome measurement.

Consistent methods of measurement of targeted outcomes are necessary to ensure the validity

and comparability of results. It is important to specify the method of measurement as walking

rates may be different if reported by parents or children or obtained by observational

Page 155: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

139

measurements. This research demonstrated that observational measurement outside schools is a

feasible and reliable method to measure walking outcomes and could be conducted periodically

to measure policy effectiveness. Similarly, collision rates are best measured using police-

reported data, as they are most generalizable to the city overall.

Evidence-Based Built Environment Strategies 6.4.4

Once specific policy targets have been set, identification of evidence-based interventions is

necessary to achieve these targets. This thesis identified the built environment as an important

correlate of both walking to school and child pedestrian-motor vehicle collisions. Results also

showed that features of the built environment acted as confounders and effect modifiers of the

relationship between walking to school and child pedestrian-motor vehicle collisions. Policies

to increase walking to school and to reduce child pedestrian-motor vehicle collisions should be

based on strong evidence of effectiveness with a focus on the built environment. Distance is an

important correlate of walking to school and must be considered in school board policies related

to school locations. Also, it is important to recognize that different built environment strategies

require different time frames to implement, especially in already established neighbourhoods,

thus making some strategies potentially more suited to newer communities.

The systematic review in Chapter 4 revealed that the quality of studies related to the built

environment was greatly varied and there was a need for more controlled studies. In the Road

Safety Strategy, proposed interventions were supported by references, many of which pertained

to the built environment.163

The references were, however, from a wide variety of sources and

included reports, websites, and published scientific studies with no indication of the quality of

the studies. In addition, many of the references cited were directed at pedestrians in general, and

did not focus on children. As discussed in Chapter 5, it is important that interventions be

evaluated specifically for children, as the effectiveness of road safety features for adults and

children may differ. Evidence-based interventions related to both child pedestrian-motor vehicle

collisions and walking to school are necessary to develop appropriate policy.

Page 156: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

140

6.4.4.1 Distance and School Boundaries

Distance is the most strongly established built environment correlate of walking to school. One

of the Big Move’s supporting policies to expand and enhance active transportation, was to design

school campuses and define school catchment areas to maximize walking and cycling as the

primary means of school travel.173

In our analyses, school attendance boundaries were generally

the appropriate size for walking to school, as roads within school attendance boundaries were

generally within walking distance of the schools. These small school catchment areas; however,

may soon be a thing of the past. In Ontario, many schools have closed in the past 10 years due to

financial constraints, resulting in amalgamations into larger schools with longer travel distances.

Between 1998-2007, there was a reported 192 school closures, with 122 more pending.258

This

economic rationalization of schools also has occurred in other high income countries.105,106,129

Unless there are concerted efforts to maintain smaller community schools, it will be difficult to

increase AST due to prohibitive travel distances.

When delineating school attendance boundaries, land use structures within the school boundaries

must also be identified which might deter walking and may increase walking distances. For

example, several of the school boundaries which had low proportions of roads within the 1.6 km

walking distance of the school, encompassed industrial parks, railway tracks, and hydro fields.

Barriers must be considered, as they can separate residential areas from schools, resulting in

much further walking distances.

Government funding should be available to all schools in the TDSB to develop school travel

plans to guide children along safe and feasible routes to school from all residential locations

within the school attendance boundary. School travel plans must acknowledge distance as well

as potential barriers to walking (e.g., railway crossings, construction sites, major arterial

roadways, missing sidewalks). Results of our analyses showed that children are most frequently

involved in collision when crossing the road and that roadway features associated with increased

collision rates were related to where children cross roads (e.g., school crossing guards, traffic

lights). School travel plans should indicate routes that both minimize distance and avoid road

crossing, by providing safe off-road options wherever possible. School board policy should be

developed around traffic injury prevention, which would include the careful monitoring of

Page 157: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

141

collisions within each of the school boundaries through collaboration with the City of Toronto, in

order that locally relevant strategies can be implemented to reduce risk.

6.4.4.2 Short-term Versus Long-term Built Environment Strategies

Policy initiatives directed at the built environment can be considered either shorter or longer

term. Shorter term initiatives may be more feasible and easier to implement in already existing

neighborhoods, such as, increasing the number of school crossing guards or eliminating one way

streets around schools. Longer term initiatives may be more difficult to implement in older

established neighbourhoods, and may be more appropriate when designing a new community.

Neighbourhoods could potentially be designed with higher population density residential areas,

while ensuring residences are all within 1.6 km of their assigned schools. Routes to school from

all residences could also be designed to minimize, wherever possible, the numbers of road

crossings for children and installing walkways.

An excellent example of this type of planning has occurred recently in Brampton, Ontario, just

outside Toronto. The local public health unit recognized the importance of the built environment

and healthy neighbourhood development to reduce the adverse effects of the built environment

on public health.259

One of the health unit recommendations was to encourage planning and

transportation professionals to consider themselves as enablers of public health and to create

partnerships between departments of public health and planning. A health development

framework was created and a new tool called the Peel Health Development Index was developed

together with Dunn et al. from the Centre for Research on Inner City Health at St. Michael’s

Hospital in Toronto which measured features of the built environment found to be related to

health outcomes including: Density, Service Proximity, Land Use Mix, Street Connectivity,

Road Network & Sidewalk Characteristics, Parking, and Aesthetics & Human Scale.260

New

communities are currently being built in Brampton using this framework, which focus on

pedestrian friendly street designs. Therefore, with the commitment and collaboration of a variety

of disciplines it is possible to create physical environments which focus on increased active

transportation and safety in new communities.

Page 158: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

142

6.5 Knowledge Translation Activities

Knowledge translation activities during the course of this thesis work have been largely in the

form of “end of grant knowledge translation” where knowledge has been disseminated after the

research has been completed.261

The results have been presented to many stakeholders in a

variety of forums, and reflect the attempt to synthesize results for AST and injury prevention.

Stakeholders have included parents, schools, the Toronto District School Board, the City of

Toronto, Transportation Services, The Toronto Police Service, the Canadian Active & Safe

Routes to School program, and the scientific community. Individual reports regarding the

observed proportion of children walking to school, the traffic situation around the schools

observed during the field audits and parent surveys regarding perceptions of traffic danger were

sent to school principals at schools to disseminate within their school communities. Feedback

was received from 47% of the school principals regarding the knowledge uses with whom the

report was shared (Table 6.2), and the actions taken in response to the report (Table 6.3).

Table 6-2: Individualized school report knowledge users

Knowledge users

• Parent council

• School staff

• School advisory council

• Crossing guard

• Community liaise officer

• School crossing guard

• Caretaker

• School superintendent

• Caring and safe schools committee

• Toronto Police Services

• School newsletter

• Toronto Public Health

Page 159: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

143

Table 6-3: Actions taken attributed to individualized school reports by school principals

Actions taken

• Developed a pedestrian/parking safety committee

• New crosswalk installed

• Used info for establishment of Kiss’ N Ride

• Used for proposal to City of Toronto for new crossing guard

• Walking school bus implemented

• Contacted police re: excessive speeding

• Assigned more staff to monitor drop off

• “No stopping, buses only” signs posted along curb

• Started Walking Wednesdays

• Purchased bike rack

• Planned 3 walk to school days

• Registered on the Safe Routes to School website

• Changed bus loading, legal parking and drop-off zones

• Investigated changes to speed limit and signage (e.g. curve ahead)

• Invited Manager of Traffic Operations for City of Toronto to do student talk

about traffic safety

• Traffic safety incorporated into health class discussions

• Established walking goals for school

• New lines painted on driveway

Several presentations and interviews were conducted with a general public audience. These

included participation as a panelist at; a Canadian Institutes for Health Research (CIHR) Café

Scientific on traumatic brain injury for the Hospital for Sick Children and an insurance company

on school zone safety. A poster was presented at a public forum organized by the Hospital for

Sick Children, on brain injury at the Toronto Central Library. Two interviews were conducted

on school traffic safety; one for CBC radio, and one for a parenting magazine.

Results were presented to both the injury prevention scientific community and the general

research community at Canadian and International Injury conferences, a scientific retreat for the

Research Institute and research rounds at Child Health and Evaluative Science at the Hospital for

Sick Children. The scientific audience is soon to be expanded to include transportation

researchers, as results will be presented at the Transportation Research Board meeting in 2014.

An important outcome of this research was the successful collaboration with several different

partners. Grant applications and papers were written with staff from City of Toronto,

Transportation Services, Toronto District School Board, Parachute and Active and Safe Routes

to School as collaborators. A grant application was also written together with investigators from

the University of Toronto, Faculty of Kinesiology & Physical Education. A research partnership

Page 160: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

144

has been established with the University of Toronto, Department of Geography & Planning. The

City of Toronto, Traffic Services, has been instrumental in providing built environment and

collisions data. The Toronto Police Service, School Safety Patrol Program Co-ordinator has

been actively involved in procuring data regarding school crossing guards as well as designing

research questions for future studies. Collaboration with Parachute was initiated to further

investigate school crossing guard effectiveness. Continued input from different stakeholders will

be instrumental to further the goal of integrating walking to school and child pedestrian injury

agendas.

Future knowledge translation activities are planned to continue the dissemination of the results of

this thesis. Reports will be sent and results presented to the TDSB, the Toronto Police Service

and the City of Toronto, Transportation Services (as requested). Results are also to be

disseminated through the Active & Safe Routes to School program. Specific policy objectives

will be formulated based on results and audiences which affect policy will be approached, such

as Parachute, Toronto Public Health, and Public Health Ontario. Continued publication of

future research papers associated with this thesis is planned in scientific journals and other

media. Also, it is planned to continue to present at forums not only directed at the injury

prevention community, but also the active transportation and transportation/urban design

communities, to further promote the integration of injury prevention into these fields. Evaluation

studies will be designed to evaluate the effectiveness of these knowledge translation activities.

Indicators of effectiveness are outlined below:

To disseminate knowledge to various audiences in the scientific community.

Indicators: scientific publications: types/impact factors of journals, numbers of

publications, number of downloads. Presentations: audience type/forums, numbers of

presentations.

To disseminate knowledge to policy makers and affect policy.

Indicators: reports, policy statements (government, TDSB), school documentation,

school crossing guard installation.

To disseminate knowledge to the public.

Page 161: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

145

Indicators: media exposure, invited presentations/interviews, webinars: numbers, types.

To continue to establish partnerships/collaboration between injury prevention

researchers and those involved in AST, obesity, urban planning and transportation

research.

Indicators: numbers/types of partnerships/collaboration, numbers of collaborating

grants, numbers of collaborating reports/papers.

6.6 Future Research

Further Analysis from the Present Study 6.6.1

This was the first large observational study to examine the relationships between observed

walking to school, child pedestrian-motor vehicle collisions and the role of objectively measured

features of the built environment. Several important related studies have emerged from the

results of the present study as described below.

6.6.1.1 Specific Built Environment Design Features and Collisions

The positive associations found between traffic calming, school crossing guards and child

pedestrian-motor vehicle collisions were unexpected, and must be further examined. Controlled

studies are required to assess whether these features create hazardous traffic situations for

children, or if there were spatial or temporal effects contributing to the positive associations.

Further study is essential to better ascertain the safety effects of these features for children. Of

particular interest is the association of school crossing guards and child pedestrian-motor vehicle

collisions, as these guards are intended to specifically address the needs of children. It is

essential to clarify the association between school crossing guards and child pedestrian-motor

vehicle collisions, as the installation of crossing guards are a directed intervention to ‘safely’

increase walking to school. School crossing guards were associated with increased walking and

Page 162: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

146

modified the effect of poor weather, ethnicity and built environment features (such as pedestrian

crossovers, traffic light and intersection density) on walking. The installation of additional

school crossing guards is a much easier and more feasible process in the City of Toronto to

increase AST, as compared to changing other roadway design features. Although it is assumed

that the presence of a school crossing guard increases safety for children walking to school, no

other studies have assessed the effect of school crossing guards on either children’s crossing

behaviour or injury outcomes.

6.6.1.2 Parent-Perceived Traffic Danger and the Built Environment

Analysis is planned of a parent survey conducted in a subsample of 20 schools during the data

collection phase of this project, to examine parental perceptions of traffic danger and the built

environment. The associations between measured and parent-perceived traffic danger, as well as

parent-perceived traffic danger and road design features should be further examined, as parents

ultimately make the decision of whether or not their school age child walks to school. In light of

the results of these analyses, it would be particularly interesting to examine how parent

perception of traffic danger is related to the density of traffic controls and the presence of school

crossing guards, which were features associated with increased walking and increased collisions.

6.6.1.3 Observed versus Self-Reported Walking

Very few studies have tried to validate different techniques to measure walking exposure in

children. Small studies have been conducted which show correlations between child-reported

exposure, travel diaries and observed walking when followed home from school; however, larger

studies are required.69,70

Estimates of parent reported walking to school were collected in the

parent survey and will be compared to the rates observed in a future analysis. Assuming there

were differences in reported versus observed walking, further validation of the results of this

study is required using observational data. It is important that study results be replicated, and

methods extended to include the trip home from school.

Page 163: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

147

Methodological Approaches for Future Studies 6.6.2

Throughout this thesis, the importance of collaboration between multiple stakeholders has been

emphasized. It therefore follows, that future research related to safe active transportation to

school, should employ an integrated knowledge translation (IKT) approach, which involves

engaging the knowledge users into the research process. IKT draws on the principals of

participatory action research where both the researchers and the end knowledge users act as a

team to develop a research question and tools, collecting, analyzing and interpret the data,

develop conclusions and a dissemination strategy and disseminate results.261

Safe AST would be

an ideal focus of future IKT studies, as it is concrete and problem-based and there are many

parent groups and community-based groups who are passionate about safe walking to school.

Possible design methodologies to consider for future research projects designed with an IKT

approach are discussed below.

6.6.2.1 Randomized Controlled Trials (RCT)

Randomized controlled trials to evaluate road design interventions would provide the strongest

evidence to inform policy, but they are difficult to implement due to the rarity of collision

outcomes and the expense of the interventions. They are, however, feasible when a need has

been identified for large-scale installations of new road design features, with a strong

commitment from multiple stakeholders. For example, an RCT could potentially be designed

related to the placement of school crossing guards, as many schools were found not to have

crossing guards in Toronto. The effects of school crossing guards on both walking outcomes and

collisions could be measured. Other more common outcomes could also be explored as proxies

for collisions, such as for example, near-misses. With careful planning and collaborations

between scientists, TDSB, the Toronto Police and the City of Toronto, RCTs of specific traffic

interventions would be possible to inform policy.

Page 164: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

148

6.6.2.2 Longitudinal Cohort

Prospective longitudinal studies are not ideal for rare outcomes such as pedestrian-motor vehicle

collisions due to the lengthy follow up time required. However, studies with this design would

be easier to implement than randomized controlled trials as they would not necessitate

randomization. These methodologies would be most feasible in newly designed communities.

6.6.2.3 Case Control

Case control studies are efficient when outcomes are rare, such as in the case of child pedestrian-

motor vehicle collisions. Several case control studies have compared the home and

neighbourhood characteristics of child pedestrian-motor vehicle collision cases and

controls.193,195,197,198,201-203,247,262

A case control study could be designed in the City of Toronto,

to investigate either the association of collisions with the presence of a school crossing guard or

traffic calming. In the example of the school crossing guard, children involved in pedestrian-

motor vehicle collisions could be extracted from the police database which occurred within a

specified distance of a school, during school travel times when a crossing guard would (if

present) be on duty. An age/sex matched control could be identified from school lists in

Toronto, who would also be matched on the use of walking as a mode of transportation to/from

school. Data regarding other potential important built environment confounders would be

collected including population density, intersection density, and speed limits within the specified

boundary around the school. This type of study would provide stronger causal evidence

regarding the association between school crossing guards/traffic calming and child pedestrian-

motor vehicle collisions as opposed to the traditional cross-sectional study design.

6.6.2.4 Quasi Experimental, Pre-Post Design

Quasi-experimental, controlled pre-post designs are the most feasible when examining the effects

of specific road design interventions such as school crossing guards and traffic calming, on child

pedestrian-motor vehicle collisions. Installation dates are available for both school crossing

Page 165: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

149

guards and traffic calming features, and collision rates will be analyzed before and after

installation while controlling for time and seasonal trends. Although the controlled pre-post

study design is not as rigorous as a randomized trial, it addresses issues of temporality and the

possibility of reverse causation, and provides stronger evidence of causality than a cross-

sectional design. This type of pre-post design has been used previously in Toronto to examine

the effects of pedestrian countdown timers on pedestrian-motor vehicle collisions.252

More

detailed spatial analysis will also be conducted this pre-post study to better understand the spatial

patterns of the built environment features and collisions.

6.6.2.5 Cross Sectional Studies in Other Settings

It has not been established if built environment correlates of child pedestrian-motor vehicle

collisions and walking to school are generalizable to different settings or if findings are location

and culture-specific. This is particularly important when considering the transfer of knowledge

obtained in a large urban centre in a high-income country, to lower income and newly motorized

countries. Replication of this study should be done in North American cities with similar urban

design features, followed by replication in other North American or European cities where

similar data sources are available but have different urban design features.

6.7 Conclusions

This thesis has provided a detailed examination of the relationship between walking to school,

child pedestrian-motor vehicle collisions and the role of the built environment in Toronto. The

research was unique in that it was the first time that literature was reviewed linking the related

concepts of walking to school and child pedestrian-motor vehicle collisions. This was also the

first large observational study to use collision data together with observed walking to school

rates. The usefulness of GIS and spatial analysis techniques to examine walking to school and

child pedestrian-motor vehicle collisions was demonstrated. The research objectives were all

met; the current knowledge of built environment correlates with walking to school and child

Page 166: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

150

pedestrian-motor vehicle collision rates was reviewed; walking to school and child pedestrian-

motor vehicle collision rates were estimated for the City of Toronto; and the influence of the

built environment on walking and child pedestrian-motor vehicle collisions was examined.

Higher variability than expected was found for the proportions walking at elementary schools in

Toronto, even when children lived within walking distance to the school. Therefore, there are

important factors other than distance that influence the decision to walk to school. Increased

proportions walking were unrelated to increased collision rates once the confounding effects of

the built environment were controlled. These results are encouraging for walking promotion

programs, suggesting that child pedestrian safety issues are related primarily to the built

environment and not the numbers walking.

Higher population densities were related to higher walking proportions and lower pedestrian-

motor vehicle collision rates. Therefore, higher density areas may provide the environment most

suited for safe walking to school. Several built environment features designed to protect

pedestrians were associated with increased collision rates, including traffic calming devices and

school crossing guards. It is important to determine whether these built environment features are

markers for dangerous traffic environments or if there are issues regarding efficacy for child

pedestrians. Clarification of the safety effects of school crossing guards is particularly important,

as the presence of school crossing guards modified the influence of other features of the built

environment on walking. School crossing guards may be a feasible intervention to increase

walking to school, assuming their effectiveness in ensuring pedestrian safety is established.

Child pedestrian-motor vehicle collisions continue to be an important health problem both in

Canada and world-wide. Encouraging walking to school in a safe environment is important to

ensure the health of children, both in terms of physical activity, and reduced pedestrian-motor

vehicle collisions. The usefulness of GIS and spatial analysis techniques in the study of child

pedestrian-motor vehicle collisions has been demonstrated. Features of a safe walking

environment for children have been examined and priorities for future research have been

established. Understanding the generalizability of the results in different environments

including those in lower-income countries is essential. The goal of this thesis was to provide

evidence that can support policies and programs that increase walking to school for children in a

safe environment.

Page 167: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

151

References

1. Broberg A, Salminen S, Kyttä M. Physical environmental characteristics promoting

independent and active transport to children's meaningful places. Appl Geogr.

2013;38:43-52.

2. World Health Organization. Global status report on road safety 2013: supporting a

decade of action. Geneva, Switzerland, 2013. Available at:

www.who.int/iris/bitstream/10665/78256/1/9789241564564_eng.pdf. Accessed July,

2013.

3. Commision for Global Road Safety. Safe Roads for All: A post-2015 agenda for health

and development, 2013. Available at:

http://www.makeroadssafe.org/Documents/mrs_safe_roads_for_all.pdf. Accessed

September, 2013.

4. Lozano R, Naghavi M, Foreman K, et al. Global and regional mortality from 235 causes

of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden

of Disease Study 2010. The Lancet. 2013;380:2095-2128.

5. World Health Organization. Overview Fact Sheet: children and road traffic injury. 2008;

http://www.who.int/violence_injury_prevention/child/injury/world_report/Road_traffic_i

njuries_english.pdf. Accessed September, 2013.

6. World Health Organization. World Health Statistics. Geneva, Switzerland, 2008.

Available at: http://www.who.int/whosis/whostat/2008/en/. Accessed September, 2013.

7. Mathers C, Loncar D. Updated projections of global mortality and burden of disease,

2002-2030: data sources, methods and results. Geneva, Switzerland, 2005. Available at:

http://www.who.int/healthinfo/statistics/bodprojectionspaper.pdf. Accessed September,

2013.

8. World Health Organization. World report on road traffic injury prevention. Geneva,

Switzerland, 2004. Available at: whqlibdoc.who.int/publications/2004/9241562609.pdf.

Accessed September, 2013.

9. Bartlett SN. The problem of children’s injuries in low-income countries: a review. Health

Policy Plan. 2002;17:1-13.

10. Safe Kids Canada. Child and Youth Unintentional Injury: 1994-2004, 10 Years in

Review. Toronto, Canada, 2007. Available at:

http://www.mhp.gov.on.ca/en/prevention/injury-prevention/skc_injuries.pdf. Accessed

November, 2013.

11. National Highway Traffic Safety Administration (NHTSA). Traffic Safety Facts, 2009

Data: Children. Washington, DC, 2009. Available at: http://www-

nrd.nhtsa.dot.gov/Pubs/811387.pdf. Accessed September, 2013.

Page 168: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

152

12. Roberts I. Why have child pedestrian death rates fallen? BMJ. 1993;306:1737-1739.

13. Organisation for Economic Co-operation and Development (OECD). Safety of

Vulnerable Road Users, 1998. Available at:

http://ntl.bts.gov/lib/24000/24400/24472/2103492.pdf. Accessed September, 2013.

14. Unicef. A league table of child deaths by injury in rich nations. Innocenti Research

Centre, Florence, 2001. Available at: http://www.unicef-

irc.org/publications/pdf/repcard2e.pdf. Accessed September, 2013.

15. Transport for London. Pedestrian casualties in Greater London. 2010; Topic Factsheet

2010-3:http://www.tfl.gov.uk/assets/downloads/pedestrian-casualties-in-greater-london-

dec2010.pdf. Accessed July, 2012

16. Toronto Public Health. Road to health: Improving walking and cycling in Toronto.

Toronto, Canada, 2012. Available at:

http://www.toronto.ca/health/hphe/pdf/roadtohealth.pdf. Accessed September 2013.

17. Transport Canada. Canadian Motor Vehicle Traffic Collision Statistics, 2010. Available

at: http://www.tc.gc.ca/eng/roadsafety/tp-1317.htm#3. Accessed September, 2013.

18. Institute for Health Metrics and Evaluation. Global Burden of Disease Cause Patterns,

2013. Available at: http://www.healthmetricsandevaluation.org/gbd/visualizations/gbd-

cause-patterns. Accessed September, 2013.

19. Commision for Global Road Safety. Make Roads Safe: A decade of action for road

safety, 2009. Available at:

http://www.makeroadssafe.org/publications/Documents/decade_of_action_report_lr.pdf.

Accessed September, 2013.

20. World Health Organization. Youth and road safety. Geneva, Switzerland, 2007.

Available at: http://whqlibdoc.who.int/publications/2007/9241595116_eng.pdf.

Accessed September, 2013.

21. Red Cross War Memorial Children's Hospital. Red Cross War Memorial Children's

Hospital Trauma Unit Statistics. Cape Town, South Africa, Statistics. Available at:

http://www.childsafe.org.za/downloads/redcross_trauma_unit_stats.pdf. Accessed

September, 2013.

22. Leitch K. Reaching for the Top: A report by the advisor on Health children and youth. ,

2007. Available at: http://www.hc-sc.gc.ca/hl-vs/pubs/child-enfant/advisor-

conseillere/index-eng.php. Accessed September, 2013.

23. National Highway Traffic Safety Administration (NHTSA). Traffic Safety Facts, 2011

Data: Children. Washington DC: U.S Department of Transportation; 2013.

24. Public Health Agency of Canada. Injury in Review, 2012 Edition: Spotlight on Road and

Transport Safety. Ottawa, 2012.

Page 169: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

153

25. SMARTRISK. The economic burden of injury in Canada. Toronto, ON, 2009.

Available at: http://www.parachutecanada.org/downloads/research/reports/EBI2009-Eng-

Final.pdf. Accessed September, 2013.

26. Canadian Institute for Health Information (CIHI). National Trauma Registry: 2006

National Trauma Registry Injury Hospitalizations Highlights Report 2008. Available at:

https://secure.cihi.ca/free_products/ntr_highlights_2007_en.pdf. Accessed September,

2013.

27. Transport Canada. Pedestrian fatalities and injuries, 1992-2001, 2004. Available at:

http://www.tc.gc.ca/eng/roadsafety/tp-tp2436-rs200401-menu-165.htm. Accessed

September, 2013.

28. World Health Organization. World report on child injury prevention. Geneva,

Switzerland, 2008. Available at:

http://www.who.int/violence_injury_prevention/child/injury/world_report/en/index.html.

Accessed September, 2013.

29. Durkin MS, Laraque D, Lubman I, Barlow B. Epidemiology and prevention of traffic

injuries to urban children and adolescents. Pediatrics. 1999;103:e74-e74.

30. Illingworth C. 227 road accidents to children. Acta Paediatrica. 1979;68:869-873.

31. Brison RJ, Wicklund K, Mueller BA. Fatal pedestrian injuries to young children: a

different pattern of injury. Am J Public Health. 1988;78:793-795.

32. Canadian Institute for Health Information (CIHI). National Trauma Registry Analystic

Bulletin: Pedestrian Injury Hospitalizations in Canada, 2001-2002. Toronto, ON, 2004.

Available at: https://secure.cihi.ca/free_products/ntr_pih_may2004_e.pdf. Accessed

September, 2013.

33. Macpherson AK, Rothman L, McKeag AM, Howard A. Mechanism of injury affects 6-

month functional outcome in children hospitalized because of severe injuries. J Trauma

Acute Care Surg. 2003;55:454-458.

34. DiGuiseppi C, Roberts I, Li L. Influence of changing travel patterns on child death rates

from injury: trend analysis. BMJ. 1997;314:710.

35. McDonald NC. Active transportation to school: trends among U.S. schoolchildren, 1969-

2001. Am J Prev Med. Jun 2007;32:509-516.

36. Cragg S, C C, Craig C. 2004 National Transportation Survey. Ottawa, ON: , 2006.

Available at:

http://www.cflri.ca/media/node/593/files/National%20Transportation%20survey%20final

%20report%202006%20formatted%20june%202.pdf. Accessed September, 2013.

37. Buliung RN, Mitra R, Faulkner G. Active school transportation in the Greater Toronto

Area, Canada: an exploration of trends in space and time (1986-2006). Prev Med.

2009;48:507-512.

Page 170: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

154

38. Active Healthy Kids Canada. Are we driving our kids to unhealthy habits, 2013.

Available at: http://www.activehealthykids.ca/ReportCard/2013ReportCard.aspx.

Accessed September, 2013.

39. Warsh J, Rothman L, Slater M, Steverango C, Howard A. Are school zones effective? An

examination of motor vehicle versus child pedestrian crashes near schools. Inj Prev.

2009;15:226-229.

40. Yiannakoulias N, Smoyer-Tomic KE, Hodgson J, Spady DW, Rowe BH, Voaklander

DC. The spatial and temporal dimensions of child pedestrian injury in Edmonton. Can J

Public Health. 2002;93:447-451.

41. Jacobsen PL. Safety in numbers: more walkers and bicyclists, safer walking and

bicycling. Inj Prev. 2003;9:205-209.

42. Macpherson A, Roberts I, Pless IB. Children's exposure to traffic and pedestrian injuries.

Am J Public Health. 1998;88:1840-1843.

43. Rao R, Hawkins M, Guyer B. Children's exposure to traffic and risk of pedestrian injury

in an urban setting. Bull N.Y. Acad Med. 1997;74:65-80.

44. Gropp K, Janssen I, Pickett W. Active transportation to school in Canadian youth: should

injury be a concern? Inj Prev. 2013;19:64-67.

45. Currie C, Griebler R, Inchley J, et al. Health behaviour in school-aged children (HBSC)

study protocol: Background, methodology and mandatory items for the 2009/10 survey

Edinburgh: CAHRU Vienna: LBIHPR2010.

46. Public Health Agency of Canada, The Canadian Insitute for Health Information. Obesity

in Canada, 2011. Available at: http://www.phac-aspc.gc.ca/hp-ps/hl-mvs/oic-

oac/assets/pdf/oic-oac-eng.pdf. Accessed September, 2013.

47. Singh AS, Mulder C, Twisk JW, Van Mechelen W, Chinapaw MJ. Tracking of childhood

overweight into adulthood: a systematic review of the literature. Obes Rev. 2008;9:474-

488.

48. Ball GD, McCargar LJ. Childhood obesity in Canada: a review of prevalence estimates

and risk factors for cardiovascular diseases and type 2 diabetes. Canadian Journal of

Applied Physiology. 2003;28:117-140.

49. Katzmarzyk PT, Ardern CI. Overweight and obesity mortality trends in Canada, 1985-

2000. Can J Public Health. 2003;95:16-20.

50. Canadian Society for Exercise Physiology. Canadian Physical Activity Guidelines for

Childen and Youth 2011. Available at: http://www.csep.ca/english/view.asp?x=949.

Accessed September, 2013.

51. Canadian Fitness and Lifestyle Research Insitute. Kids Can Play! Physical activity levels

of Canadian children and youth. Ottawa, ON, 2009. Available at:

Page 171: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

155

http://www.cflri.ca/media/node/972/files/CANPLAY%20Bulletin%202%20PA%20Level

s%20EN.pdf. Accessed September, 2013.

52. Gordon-Larsen P, Boone-Heinonen J, Sidney S, Sternfeld B, Jacobs Jr DR, Lewis CE.

Active commuting and cardiovascular disease risk: the CARDIA study. Arch Intern Med.

2009;169:1216.

53. Frank LD, Andresen MA, Schmid TL. Obesity relationships with community design,

physical activity, and time spent in cars. Am J Prev Med. 2004;27:87-96.

54. Faulkner GEJ, Buliung RN, Flora PK, Fusco C. Active school transport, physical activity

levels and body weight of children and youth: a systematic review. Prev Med. 2009;48:3-

8.

55. Livingstone M, Robson P, Wallace J, McKinley M. How active are we? Levels of routine

physical activity in children and adults. Proc Nutr Soc. 2003;62:681-702.

56. Malina RM. Tracking of physical activity and physical fitness across the lifespan. Res Q

Exerc Sport. 1996;67:S-48-S-57.

57. Janz KF, Dawson JD, Mahoney LT. Tracking physical fitness and physical activity from

childhood to adolescence: the Muscatine study. Med Sci Sports Exerc. 2000;32:1250-

1257.

58. Telama R, Yang X, Viikari J, Välimäki I, Wanne O, Raitakari O. Physical activity from

childhood to adulthood: a 21-year tracking study. Am J Prev Med. 2005;28:267-273.

59. Maas MW, Harris S. Police recording of road accident in-patients: investigation into the

completeness, representativity and reliability of police records of hospitalized traffic

victims. Accid Anal Prev. 1984;16:167-184.

60. Barancik JI, Fife D. Discrepancies in vehicular crash injury reporting: Northeastern Ohio

trauma study IV. Accid Anal Prev. 1985;17:147-154.

61. Agran PF, Castillo DN, Winn DG. Limitations of data compiled from police reports on

pediatric pedestrian and bicycle motor vehicle events. Accid Anal Prev. 1990;22:361-370.

62. Government of Ontario. Highway Traffic Act, 2012. Available at: http://www.e-

laws.gov.on.ca/html/statutes/english/elaws_statutes_90h08_e.htm#BK296. Accessed

September, 2013.

63. Lee P, Mihailovic A, Rothman L, Mutto M, Nakitto M, Howard A. Pedestrian injuries in

school-attending children: a comparison of injury data sources in a low-income setting.

Inj Prev. 2009;15:100-104.

64. Tercero F, Andersson R. Measuring transport injuries in a developing country: an

application of the capture–recapture method. Accid Anal Prev. 2004;36:13-20.

Page 172: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

156

65. Sirard JR, Slater ME. Walking and bicycling to school: a review. Am J Lifestyle Med.

2008;2:372-396.

66. Pont K, Ziviani J, Wadley D, Bennett S, Abbott r. Environmental correlates of children's

active transportation: A systematic literature review. Health Place. 2009;15:849-862.

67. Saelens BE, Handy SL. Built environment correlates of walking: a review. Med Sci

Sports Exerc. 2008;40:S550-566.

68. Rossen LM, Pollack KM, Curriero FC, et al. Neighborhood incivilities, perceived

neighborhood safety, and walking to school among urban-dwelling children. J Phys Act

Health. 2011;8:262-271.

69. Routledge DRWRHC. The Exposure of Young Children to Accident Risk as Pedestrians.

1974;17:457-480.

70. Stevenson MR. The Validity of Children's Self-Reported Exposure to Traffic. Accid Anal

Prev. 1996;28:599-605.

71. Davison KK, Werder JL, Lawson CT. Children's active commuting to school: current

knowledge and future directions. Prev Chronic Dis. 2008;5:A100.

72. Sirard JR, Ainsworth BE, McIver KL, Pate RR. Prevalence of active commuting at urban

and suburban elementary schools in Columbia, SC. Am J Public Health. 2005;95:236-

237.

73. Haddon Jr W. The changing approach to the epidemiology, prevention, and amelioration

of trauma: the transition to approaches etiologically rather than descriptively based. Am J

Public Health Nations Health. 1968;58:1431-1438.

74. Safe Kids Canada. Making It Happen: Pedestrian Safety. 2004;

www.parachutecanada.org/downloads/research/reports/PedestrianSafety-

CommunityGuide-Eng.pdf. Accessed September, 2013.

75. Runyan CW. Using the Haddon matrix: introducing the third dimension. Inj Prev.

1998;4:302-307.

76. Haddon Jr W. Advances in the epidemiology of injuries as a basis for public policy.

Public Health Rep. 1980;95:411.

77. Peek-Asa C, Zwerling C. Role of environmental interventions in injury control and

prevention. Epidemiol Rev. 2003;25:77-89.

78. McMillan TE. Urban form and a child’s trip to school: the current literature and a

framework for future research. J. Plan. Lit. 2005;19:440-456.

79. Panter JR, Jones AP, Van Sluijs EM. Environmental determinants of active travel in

youth: a review and framework for future research. Int J Behav Nutr Phys Act.

2008;5:34.

Page 173: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

157

80. Mitra R. Independent mobility and mode choice for school transportation: a review and

framework for future research. Transport Reviews. 2013;33:21-43.

81. Bronfenbrenner U. Ecological systems theory. Annals of Child Development. 1989;6:187-

249.

82. Sallis JF, Owen N, Fisher EB. Ecological models of health behavior. Health behavior and

health education: Theory, research, and practice. 2008;4:465-486.

83. Cervero R, Kockelman K. Travel demand and the 3Ds: density, diversity, and design.

Transp Res Part D: Transport and Environment. 1997;2:199-219.

84. Ewing R, Cervero R. Travel and the built environment. J Am Plann Assoc. 2010;76:265-

294.

85. Wong BY-M, Faulkner G, Buliung R. GIS measured environmental correlates of active

school transport: a systematic review of 14 studies. Int J Behav Nutr Phys Act. 2011;8:39.

86. Lin J-J, Chang H-T. Built environment effects on children’s school travel in Taipai:

independence and travel mode. Urban Stud. 2010;47:867-889.

87. Wazana A, Krueger P, Raina P, Chambers L. A review of risk factors for child pedestrian

injuries: are they modifiable? Inj Prev. 1997;3:295-304.

88. DiMaggio C, Li G. Roadway characteristics and pediatric pedestrian injury. Epidemiol

Rev. 2012;34:46-56.

89. O'Neill B, Mohan D, Breen J, et al. The World Bank's global road safety partnership.

Traffic Inj Prev. 2002;3:190-194.

90. Duperrex O, Bunn F, Roberts I. Safety education of pedestrians for injury prevention: a

systematic review of randomised controlled trials. BMJ. 2002;324:1129.

91. Roberts I, Ashton T, Dunn R, Leejoe T. Preventing child pedestrian injury: pedestrian

education or traffic calming. Aust J Publ Health. 1994;18:209-212.

92. Schieber RA, Vegega ME. Reducing childhood pedestrian injuries. Inj Prev. 2002;8

Suppl 1:i1-i10.

93. Organisation for Economic Co-operation and Development (OECD). Speed management.

France, 2006. Available at:

http://www.internationaltransportforum.org/Pub/pdf/06Speed.pdf. Accessed September,

2013.

94. Pasanen E. Driving speeds and pedestrian safety. Helsinki, Finland, 1990. Available at:

http://www.walk21.com/papers/San%20Sebastian%2002%20Pasanen%20Driving%20Sp

eeds%20and%20Pedestrian%20Safet(1).pdf. Accessed September, 2013.

Page 174: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

158

95. Insurance Institute for Highway Safety (IIHS). Fatality facts 2011. 2011;

http://www.iihs.org/research/fatality.aspx?topicName=Overview-of-Fatality-Facts.

Accessed July 8, 2013.

96. Nilsson G. Traffic Safety Dimensions and the Power Model to Describe the Effect of

Speed on Safety. Sweden, 2004. Available at:

ttp://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=21612&fileOId=16933

53. Accessed September, 2013.

97. Pilkington P. Reducing the speed limit to 20 mph in urban areas: child deaths and injuries

would be decreased. BMJ. 2000;320:1160.

98. Walz F, Hoefliger M, Fehlmann W. Speed limit reduction from 60 to 50 km/h and

pedestrian injuries. Paper presented at: Twenty-Seventh Stapp Car Crash Conference

Proceedings with International Research Council on Biokenitcs of Impacts

(IRCOBI)1983; Warrendale, PA.

99. Retting RA, Ferguson SA, McCartt AT. A review of evidence-based traffic engineering

measures designed to reduce pedestrian–motor vehicle crashes. Am J Public Health.

2003;93.

100. Bunn F, Collier T, Frost C, Ker K, Roberts I, Wentz R. Traffic calming for the prevention

of road traffic injuries: systematic review and meta-analysis. Inj Prev. 2003;9:200-204.

101. Aeron-Thomas A, Hess S. Red-light cameras for the prevention of road traffic crashes.

Cochrane Database of Syst Rev. 2005;2.

102. Wilson C, Willis C, Hendrikz JK, Bellamy N. Speed enforcement detection devices for

preventing road traffic injuries. Cochrane Database of Syst Rev. 2006;2.

103. Beyer FR, Ker K. Street lighting for preventing road traffic injuries. Cochrane Database

of Syst Rev. 2009;1.

104. DiMaggio C, Li G. Effectiveness of a safe routes to school program in preventing school-

aged pedestrian injury. Pediatrics. 2013;131:290-296.

105. Mitra R, Buliung R. Built environment correlates of active school transportation:

neighborhood and the modifiable areal unit problem. J Transp Geogr. 2012;20:51-61.

106. Schlossberg M, Greene J, Phillips PP, Johnson B, Parker B. School trips: effects of urban

form and distance on travel mode. J Am Plann Assoc. 2006;72:337-346.

107. Ewing R, Schroeer W, Greene W. School location and student travel: Analysis of factors

affecting mode choice. Transp Res Rec. 2004;2004:55-63.

108. Timperio A, Ball K, Salmon J, et al. Personal, family, social, and environmental

correlates of active commuting to school. Am J Prev Med. 2006;30:45-51.

Page 175: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

159

109. DiGuiseppi C, Roberts I, Li L, Allen D. Determinants of car travel on daily journeys to

school: cross sectional survey of primary school children. BMJ. 1998;316:1426-1428.

110. Fyhri A, Hjorthol R. Children's independent mobility to school, friends and leisure

activities. J Transp Geogr. 2009;17:377-384.

111. McDonald NC. Children’s mode choice for the school trip: the role of distance and

school location in walking to school. Transportation. 2008;35:23-35.

112. Salmon J, Salmon L, Crawford DA, Hume C, Timperio A. Associations among

individual, social, and environmental barriers and children's walking or cycling to school.

Am J Health Promot. 2007;22:107-113.

113. Wen LM, Fry D, Merom D, Rissel C, Dirkis H, Balafas A. Increasing active travel to

school: are we on the right track? A cluster randomised controlled trial from Sydney,

Australia. Prev Med. 2008;47:612-618.

114. Yelavich S, Towns C, Burt R, et al. Walking to school: frequency and predictors among

primary school children in Dunedin, New Zealand. NZ Med J. 2008;121:51-58.

115. Yeung J, Wearing S, Hills AP. Child transport practices and perceived barriers in active

commuting to school. Transp Res Part A: Policy Pract. 2008;42:895-900.

116. Zhu X, Arch B, Lee C. Personal, social, and environmental correlates of walking to

school behaviors: case study in Austin, Texas. The Scientific World J. 2008;8:859-872.

117. McMillan TE. The Relative Influence of Urban Form on a Child's Travel Mode to

School. Transp Res Part A: Policy Pract. 2007;41:69-79.

118. Saelens BE, Sallis JF, Frank LD. Environmental correlates of walking and cycling:

findings from the transportation, urban design, and planning literatures. Ann Behav Med.

2003;25:80-91.

119. Shay E, Spoon S, Khattak A. Walkable environments and walking activity: Final Report.

Chapel Hill, NC, 2003. Available at:

http://www.stc.utk.edu/STCresearch/completed/PDFs/walkfinal.pdf. Accessed

September, 2013.

120. Kerr J, Rosenberg D, Sallis JF, Saelens BE, Frank LD, Conway TL. Active commuting to

school: associations with environment and parental concerns. Med Sci Sports Exerc.

2006;38:787-794.

121. Panter JR, Jones AP, Van Sluijs EMF, Griffin SJ. Neighborhood, route, and school

environments and children's active commuting. Am J Prev Med. Mar 2010;38:268-278.

122. Mitra R, Buliung R, Roorda MJ. The built environment and school travel mode choice in

Toronto, Canada. Transp Res Rec. 2010;2156:150-159.

Page 176: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

160

123. Braza M, Shoemaker W, Seeley A. Neighborhood design and rates of walking and biking

to elementary school in 34 California communities. Am J Health Promot. 2004;19:128-

136.

124. Bringolf-Isler B, Grize L, Mäder U, Ruch N, Sennhauser FH, Braun-Fahrländer C.

Personal and environmental factors associated with active commuting to school in

Switzerland. Prev Med. 2008;46:67-73.

125. Kweon BS, Woo- Hwa S, Folzenlogen R, Jun-Hyun K. Children and Transportation:

Identifying Environments that Foster Walking and Biking to School. College Station,

Texas, 2006. Available at:

http://d2dtl5nnlpfr0r.cloudfront.net/swutc.tamu.edu/publications/technicalreports/473700

-00031-1.pdf. Accessed September, 2013.

126. Kerr J, Frank L, Sallis J, Chapman J. Urban form correlates of pedestrian travel in youth:

Differences by gender, race-ethnicity and household attributes. Transp Res Part D:

Transport and Environment. 2007;12:177-182.

127. Rosenberg D, Ding D, Sallis JF, et al. Neighborhood Environment Walkability Scale for

Youth (NEWS-Y): reliability and relationship with physical activity. Prev Med.

2009;49:213-218.

128. Boarnet MG, Anderson CL, Day K, McMillan T, Alfonzo M. Evaluation of the

California Safe Routes to School legislation: urban form changes and children’s active

transportation to school. Am J Prev Med. 2005;28:134-140.

129. Trapp GSA, Giles-Corti B, Christian HE, et al. Increasing children's physical activity:

individual, social, and environmental factors associated with walking to and from school.

Health Educ Behav. 2012;39.2:172-182.

130. de Vries SI, Hopman-Rock M, Bakker I, Hirasing RA, van Mechelen W. Built

environmental correlates of walking and cycling in Dutch urban children: results from the

SPACE study. Int J Environ Res Public Health. 2010;7:2309-2324.

131. Zhu X, Lee C, Kwok OM, Varni JW. Context-specific correlates of walking behaviors to

and from school: do they vary across neighborhoods and populations? J Phys Act Health.

2011;8 S59-71.

132. Greene J, Daniel D. The Built Environment, Neighborhood Safety, and Physical Activity

Among Low Income Children. Portland, Oregan, 2009. OTREC-RR-09-06 Available at:

http://ntl.bts.gov/lib/33000/33800/33839/OTREC-RR-09-

06_Greene_NeighborhoodSafety_FinalReport.pdf. Accessed September, 2013.

133. Evenson KR, Birnbaum AS, Bedimo-Rung AL, et al. Girls' perception of physical

environmental factors and transportation: reliability and association with physical activity

and active transport to school. Int J Behav Nutr Phys Act. 2006;3.

Page 177: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

161

134. Giles-Corti B, Wood G, Pikora T, et al. School site and the potential to walk to school:

the impact of street connectivity and traffic exposure in school neighborhoods. Health

Place. 2011;17:545-550.

135. Leslie E, Kremer P, Toumbourou JW, Williams JW. Gender differences in personal,

social and environmental influences on active travel to and from school for Australian

adolescents. J Sci Med Sport. 2010;13:597-601.

136. Yarlagadda AK, Srinivasan S. Modeling children’s school travel mode and parental

escort decisions. Transportation. 2008;35:201-218.

137. Mitra R, Buliung R, Faulkner G. Spatial clustering and the temporal mobility of walking

school trips in the Greater Toronto Area, Canada. Health Place. 2010;16:646-655.

138. Chillon P, Evenson KR, Vaughn A, Ward DS. A systematic reivew of interventions for

promoting active transportation to school. Int J Behav Nutr Phys Act. 2011;8:10.

139. Bors P, Dessauer M, Bell R, Wilkerson R, Lee J, Strunk SL. The Active Living by

Design national program: community initiatives and lessons learned. Am J Prev Med.

2009;37:S313-S321.

140. Boarnet MG, Day K, Anderson C, McMillan T, Alfonzo M. California's Safe Routes to

School program: impacts on walking, bicycling, and pedestrian safety. J Am Plann Assoc.

2005;71:301-317.

141. Rushton G. Public health, GIS, and spatial analytic tools. Ann Rev Public Health.

2003;24:43-56.

142. Schuurman N. GIS: A Short Introduction. Malden, Massachusetts: Wiley-Blackwell;

2004.

143. Cusimano MD, Chipman M, Glazier RH, Rinner C, Marshall SP. Geomatics in injury

prevention: the science, the potential and the limitations. Inj Prev. 2007;13:51-56.

144. Briggs DJ, Elliott P. The use of geographical information systems in studies on

environment and health. World Health Stat Q. 1994;48:85-94.

145. Nykiforuk CI, Flaman LM. Geographic information systems (GIS) for health promotion

and public health: a review. Health Promot Pract. 2011;12:63-73.

146. Braddock M, Lapidus G, Cromley E, Cromley R, Burke G, Banco L. Using a geographic

information system to understand child pedestrian injury. Am J Public Health.

1994;84:1158-1161.

147. Kingham S, Sabel CE, Bartie P. The impact of the ‘school run’on road traffic accidents:

A spatio-temporal analysis. J Transp Geog. 2011;19:705-711.

148. Abdel-Aty M, Chundi SS, Lee C. Geo-spatial and log-linear analysis of pedestrian and

bicyclist crashes involving school-aged children. J Safety Res. 2007;38:571-579.

Page 178: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

162

149. Weiner EJ, Tepas III JJ. Application of electronic surveillance and global information

system mapping to track the epidemiology of pediatric pedestrian injury. J Trauma Acute

Care Surg. 2009;66:S10-S16.

150. Dissanayake D, Aryaija J, Wedagama DMP. Modelling the effects of land use and

temporal factors on child pedestrian casualties. Accid Anal Prev. 2009;41:1016-1024.

151. Cloutier MS, Apparicio P. Does the neighbouring environment around schools influence

child pedestrian accidents risk in Montreal? The contribution of geographically weighted

Poisson regression. Revue de géographie et d'aménagement. 2008;1:25-38.

152. LaScala EA, Gruenewald PJ, Johnson FW. An ecological study of the locations of

schools and child pedestrian injury collisions. Accid Anal Prev. 2004;36:569-576.

153. Lightstone AS, Dhillon PK, Peek-Asa C, Kraus JF. A geographic analysis of motor

vehicle collisions with child pedestrians in Long Beach, California: comparing

intersection and midblock incident locations. Inj Prev. 2001;7:155-160.

154. Petch RO, Henson RR. Child road safety in the urban environment. J Transp Geogr.

2000;8:197-211.

155. Blazquez CA, Celis MS. A spatial and temporal analysis of child pedestrian crashes in

Santiago, Chile. Accid Anal Prev. 2013;50:304-311.

156. Larsen K, Gilliland J, Hess PM. Route-based analysis to capture the environmental

influences on a child's mode of travel between home and school. Ann Assoc Am Geogr.

2012;102:1348-1365.

157. Gropp KM, Pickett W, Janssen I. Multi-level examination of correlates of active

transportation to school among youth living within 1 mile of their school. Int J Behav

Nutr Phys Act. 2012;9:124-124.

158. Lin L, Moudon AV. Objective versus subjective measures of the built environment,

which are most effective in capturing associations with walking? Health Place.

2010;16:339-348.

159. City of Toronto. Building the New City of Toronto: Final Three Year Status Report on

Amalgamation, January 1998-December 2000. Toronto, Canada, 2001. Available at:

http://www.toronto.ca/city_manager/pdf/amalgationreport.pdf. Accessed September,

2013.

160. Hess PM. Avenues or arterials: the struggle to change street building practices in

Toronto, Canada. J Urban Des. 2009;14:1-28.

161. Hess P, Moore Milroy B. Making Toronto's Streets. Toronto, 2006. Available at:

http://faculty.geog.utoronto.ca/Hess/Downloads/Hess%20Milroy%20Making%20Toronto

s%20Streets%20report%20cuhi.pdf. Accessed September, 2013.

Page 179: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

163

162. Sewell J. The shape of the city: Toronto struggles with modern planning. Toronto,

Canada: University of Toronto Press; 1993.

163. Canadian Council of Motor Transport Administrators. Canada's Road Safety Stratey

2015, 2010. Available at: http://www.ccmta.ca/crss-2015/index.php. Accessed

September, 2013.

164. Parachute. Pedestrian safety, 2013. Available at:

http://www.parachutecanada.org/policy/item/252. Accessed September, 2013.

165. Ontario Ministry of Health Promotion. Ontario's injury prevention strategy: working

together for a safer, healthier Ontario, 2007. Available at:

http://www.mhp.gov.on.ca/en/prevention/injury-prevention/strategy.pdf Accessed

August, 2013.

166. Ontario Public Health Association (OPHEA). Resolution #5: OPHA Injury Prevention

Workgroup, 2008. Available at:

http://www.opha.on.ca/What%20We%20Do/Position%20Papers/papers/2008-05_res.pdf.

Accessed August, 2013.

167. Public Health Ontario (PHO). Our Organization, 2013. Available at:

http://www.publichealthontario.ca/en/About/Pages/Organization.aspx#.UhZ0oCqF_dg.

Accessed September, 2013.

168. National Center for Safe Routes to School. History of SRTS, 2012. Available at:

http://www.saferoutesinfo.org/about-us/mission-and-history. Accessed September, 2013.

169. Chriqui JF, Taber DR, Slater SJ, Turner L, Lowrey K, Chaloupka FJ. The impact of state

safe routes to school-related laws on active travel to school policies and practices in US

elementary schools. Health Place. 2012;18:8.

170. Buliung R, Faulkner G, Beesley T, Kennedy J. School travel planning: mobilizing school

and community resources to encourage active school transportation. J Sch Health.

2011;81:704-712.

171. Active and Safe Routes to School. A Program of Green Communities of Canada, 2013.

Available at: http://www.saferoutestoschool.ca/. Accessed September, 2013.

172. Active Healthy Kids Canada. Mandate, 2013. Available at:

http://www.activehealthykids.ca/AboutUs.aspx. Accessed September, 2013.

173. Metrolinx. The Big Move. Toronto, Canada, 2008. Available at:

http://www.metrolinx.com/thebigmove/Docs/big_move/TheBigMove_020109.pdf.

Accessed September, 2013.

174. Metrolinx. Stepping It Up-Final Report. Toronto, 2012. Available at:

http://www.metrolinx.com/en/projectsandprograms/schooltravel/SteppingItUpReportEN

G.pdf. Accessed September, 2013.

Page 180: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

164

175. McKee R, Mutrie N, Crawford F, Green B. Promoting walking to school: results of a

quasi-experimental trial. J Epidemiol Community Health. 2007;61:818-823.

176. Rowland D, DiGuiseppi C, Gross M, Afolabi E, Roberts I. Randomised controlled trial of

site specific advice on school travel patterns. Arch Dis Child. 2003;88:8-11.

177. City of Toronto. Traffic Calming Policy, 2010. Available at:

http://www.toronto.ca/transportation/traffic/pdf/traffic_calming_policy_summary.pdf.

Accessed August, 2013.

178. City of Toronto. Missing Sidewalk Program, 2011. Available at:

http://www.toronto.ca/transportation/walking/newsletter/issue1.htm. Accessed

September, 2013.

179. U.S. Department of Transportation FHA. Highway Safety Improvement Program (HSIP)

MAP-21 Interim Eligibility Guidance, 2012. Available at:

http://www.fhwa.dot.gov/map21/guidance/guidehsip.cfm. Accessed September, 2013.

180. Merom D, Tudor-Locke C, Bauman A, Rissel C. Active commuting to school among

NSW primary school children: implications for public health. Health Place. 2006;12:678-

687.

181. Watson M, Dannenberg AL. Investment in safe routes to school projects: public health

benefits for the larger community. Prev Chronic Dis. 2008;5:A90.

182. Kingham S, Ussher S. An assessment of the benefits of the walking school bus in

Christchurch, New Zealand. Transp Res Part A: Policy Pract. 2007;41:502-510.

183. Southworth M. Designing the walkable city. J Urban Plan. 2005;131:246.

184. Stewart O. Findings from Research on Active Transportation to School and Implications

for Safe Routes to School Programs. J Plan Lit. 2011;26:127-150.

185. Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic

reviews and meta-analyses of studies that evaluate health care interventions: explanation

and elaboration. PLoS Medicine. 2009;6:e1000100.

186. Jacobs G, Aeron-Thomas A, Astrop A. Estimating global road fatalities. Berkshire, UK,

2000. Available at: http://www.transport-

links.org/transport_links/filearea/publications/1_329_TRL445.pdf. Accessed September,

2013.

187. Pucher J, Dijkstra L. Promoting safe walking and cycling to improve public health:

lessons from the Netherlands and Germany. Am J Public Health. 2003;93:1509.

188. Bassett Jr DR, Pucher J, Buehler R, Thompson DL, Crouter SE. Walking, cycling, and

obesity rates in Europe, North America, and Australia. J Phys Act Health. 2008;5:795-

814.

Page 181: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

165

189. New Zealand Ministry of Transport. Raising the profile of walking and cycling in New

Zealand: A guide for decision makers. Wellington, NZ, 2008. Available at:

http://www.transport.govt.nz/ourwork/Documents/RaisingtheProfileWalkingCyclinginN

Z.pdf. Accessed September, 2013.

190. Genaidy AM, Lemasters GK, Lockey J, et al. An epidemiological appraisal instrument: a

tool for evaluation of epidemiological studies. Ergonomics. 2007;50:920-960.

191. Downs SH, Black N. The feasibility of creating a checklist for the assessment of the

methodological quality both of randomised and non-randomised studies of health care

interventions. Journal of Epidemiology and Community Health. 1998;52:377-384.

192. Sanderson S, Tatt ID, Higgins JP. Tools for assessing quality and susceptibility to bias in

observational studies in epidemiology: a systematic review and annotated bibliography.

International Journal of Epidemiology. 2007;36:666-676.

193. Tester JM, Rutherford GW, Wald Z, Rutherford MW. A matched case-control study

evaluating the effectiveness of speed humps in reducing child pedestrian injuries. Am J

Public Health. Apr 2004;94:646-650.

194. Jones SJ, Lyons RA, John A, Palmer SR. Traffic calming policy can reduce inequalities

in child pedestrian injuries: database study. Inj Prev. 2005;11:152-156.

195. Mueller BA, Rivara FP, Lii SM, Weiss NS. Environmental factors and the risk for

childhood pedestrian-motor vehicle collision occurrence. Am J Epidemiol. Sep

1990;132:550-560.

196. Carver A, Timperio A, Hesketh K, Crawford D. Are safety-related features of the road

environment associated with smaller declines in physical activity among youth? J Urban

Health. 2009;87:29-43.

197. Stevenson MR, Jamrozik KD, Spittle J. A case-control study of traffic risk factors and

child pedestrian injury. Int J Epidemiol. 1995;24:957-964.

198. Stevenson M, Jamrozik K, Burton P. A case-control study of childhood pedestrian

injuries in Perth, Western Australia. J Epidemiol Community Health. 1996;50:280-287.

199. Stevenson M. Childhood pedestrian injuries: what can changes to the road environment

achieve? Aus N Z J Public Health. 1997;21:33-37.

200. Wazana A, Rynard VL, Raina P, Krueger P, Chambers LW. Are child pedestrians at

increased risk of injury on one-way compared to two-way streets? Can J Public Health.

2000;91:201-206.

201. Christie N. The High Risk Child Pedestrian: Socio-Economic and Environmental Factors

in Their Accidents. Vol 117. Berkshire, UK: Transport Research Library; 1995.

202. Agran PF, Winn DG, Anderson CL, Tran C, Del Valle CP. The role of the physical and

traffic environment in child pedestrian injuries. Pediatrics. 1996;98:1096-1103.

Page 182: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

166

203. Roberts I, Norton R, Jackson R, Dunn R, Hassall I. Effect of environmental factors on

risk of injury of child pedestrians by motor vehicles: a case-control study. BMJ.

1995;310:91-94.

204. Kupferberg-Bendavid NK, Rice RG. The Role of the physical environment in child

pedestrian accidents. J Adv Transport. 1994;28:171-187.

205. Buliung R, Larson K, Faulkner G, Stone M. The "path" not taken: Exploring structural

differences in mapped-versus shortest-network-path school travel routes. Am J Public

Health. 2013;103:1589-1596.

206. Panter JR, Jones AP, van Sluijs EMF, Griffin SJ. Attitudes, social support and

environmental perceptions as predictors of active commuting behaviour in school

children. J Epidemiol Community Health. 2010;64:41-48.

207. Napier MA, Brown BB, Werner CM, Gallimore J. Walking to school: Community design

and child and parent barriers. J Environ Psychol. 2010;31:45-51.

208. Gallimore JM, Brown BB, Werner CM. Walking routes to school in new urban and

suburban neighborhoods: An environmental walkability analysis of blocks and routes. J

Environ Psychol. 2011;31:184-191.

209. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in

epidemiology. JAMA. 2000;283:2008-2012.

210. Roberts I, Marshall R, Leejoe T. The urban traffic environment and the risk of child

pedestrian injury: a case-crossover approach. Epidemiology. 1995;6:169-171.

211. TDSB. Transportation of Students. Toronto, Canada, 2005. Available at:

http://www.tdsb.on.ca/_site/viewitem.asp?siteid=106&menuid=332&pageid=259.

Accessed September, 2013.

212. Falb MD, Kanny D, Powell KE, Giarrusso AJ. Estimating the proportion of children who

can walk to school. Am J Prev Med. 2007;33:269-275.

213. Larsen K, Gilliland J, Hess P, Tucker P, Irwin J, He M. The influence of the physical

environment and sociodemographic characteristics on children's mode of travel to and

from school. Am J Public Health. 2009;99:520-526.

214. Marler N, Montgomery R. Urban road traffic surveys: Overseas road note 11.

Berkshire, United Kingdom, 1993. Available at: http://www.transport-

links.org/transport_links/filearea/publications/1_708_ORN%2011-%20.pdf. Accessed

September, 2013.

215. Donroe J, Tincopa M, Gilman RH, Brugge D, Moore DAJ. Pedestrian road traffic injuries

in urban Peruvian children and adolescents: case control analyses of personal and

environmental risk factors. PLoS ONE. 2008;3:e3166.

Page 183: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

167

216. TDSB. The 2011 Learning Opportunities Index: Questions and Answers. Toronto,

Canada, 2011. Available at:

http://www.tdsb.on.ca/wwwdocuments/about_us/external_research_application/docs/LOI

2011.pdf. Accessed September, 2013.

217. Statistics Canada ISD. Low income cut-offs for 2008 and low income measures for 2007.

Ottawa, Canada, 2009. Available at:

http://www.statcan.gc.ca/pub/75f0002m/75f0002m2009002-eng.pdf. Accessed

September, 2013.

218. Hosmer DW, Lemeshow S. Applied logistic regression. Vol 354. New York, NY: Wiley-

Interscience; 2004.

219. Chatterjee S, Hadi A. Regression analysis by example. New Jersey: John Wiley & Sons;

2006.

220. Agresti A. An introduction to categorical data analysis. 2nd ed. Hoboken, New Jersey:

John Wiley & Sons, Inc.; 2007.

221. Martin SL, Lee SM, Lowry R. National prevalence and correlates of walking and

bicycling to school. Am J Prev Med. 2007;33:98-105.

222. Delaware Department of Transportation. School site audit. Delaware Safe Routes to

School Program Sourcebook, Available at:

http://www.deldot.gov/information/community_programs_and_services/srts/pdf/srts.pdf

Accessed September, 2013.

223. MPAC. Facts about Municipal Planning Assessment Corporation (MPAC). Pickering,

Ontario, 2013. Available at:

http://www.mpac.on.ca/about/corporate_overview/facts_about_mpac.asp. Accessed

September, 2013.

224. Teranet. About Teranet. Toronto, Canada, 2013. Available at: http://www.teranet.ca/our-

company/about-teranet. Accessed September, 2013.

225. ESRI. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research

Institute; 2011.

226. Giuliano G. New directions for understanding transportation and land use. Berkeley,

CA, 1989. Available at: http://www.uctc.net/papers/002.pdf. Accessed September,

2013.

227. Atkins DC, Gallop RJ. Rethinking how family researchers model infrequent outcomes: a

tutorial on count regression and zero-inflated models. J Family Psychol. 2007;21:726.

228. D'agostino R. A second look at analysis of variance on dichotomous data. J Educ Meas.

1971;8:327-333.

Page 184: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

168

229. Kieschnick R, McCullough BD. Regression analysis of variates observed on (0, 1):

percentages, proportions and fractions. Stat Modelling. 2003;3:193-213.

230. Harten N, Olds T. Physical Activity: Patterns of active transport in 11–12 year old

Australian children. Aus N Z J Public Health. 2004;28:167-172.

231. Timperio A, Crawford D, Telford A, Salmon J. Perceptions about the local neighborhood

and walking and cycling among children. Prev Med. 2004;38:39-47.

232. Rothman L, To T, Buliung R, Macarthur C, Howard A. Influence of social and built

environment features on children’s walking to school: an observational study. Prev Med.

Prev Med. 60; 2014:10-15.

233. Laflamme L, Diderichsen F. Social differences in traffic injury risks in childhood and

youth: a literature review and a research agenda. Inj Prev. 2000;6:293-298.

234. Braddock M, Lapidus G, Gregorio D, Kapp M, Banco L. Population, income, and

ecological correlates of child pedestrian injury. Pediatrics. 1991;88:1242-1247.

235. City of Toronto. Open Data, 2013. Available at:

http://www1.toronto.ca/wps/portal/contentonly?vgnextoid=9e56e03bb8d1e310VgnVCM

10000071d60f89RCRD. Accessed Septembear, 2013.

236. Base SAS 9.3, Procedures Guide, Statistical Procedures [computer program]. Cary, NC,

USA: SAS Institute Inc.; 2012.

237. Kleinbaum D, Kupper L, KE M. Applied Regression Analysis and Other Multivariable

Methods. Boston, Mass: PWS-Kent Publishing; 1998.

238. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public

Health. 1989;79:340-349.

239. Roberts I, Carlin J, Bennett C, et al. An international study of the exposure of children to

traffic. Inj Prev. 1997;3:89-93.

240. Robinson DL. Safety in numbers in Australia: more walkers and bicyclists, safer walking

and bicycling. Health Prom J Aust. 2005;16:47-51.

241. Rothman L, Buliung R, Macarthur C, To T, Howard A. Walking and child pedestrian

injury: a systematic review of built environment correlates of safe walking. Inj Prev.

Published Online First: 24, May, 2013. doi:10.1136/injuryprev-2012-040701.

242. Bunn F, Collier T, Frost C, Ker K, Roberts I, Wentz R. Area-wide traffic calming for

preventing traffic related injuries. Cochrane Database of Syst Rev. 2003;1.

243. Rivara FP, Barber M. Demographic analysis of childhood pedestrian injuries. Pediatrics.

1985;76:375-381.

Page 185: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

169

244. Dougherty G, Pless IB, Wilkins R. Social class and the occurrence of traffic injuries and

deaths in urban children. Can J Public Health. 1990;81:204.

245. Bagley C. The urban setting of juvenile pedestrian injuries: a study of behavioural

ecology and social disadvantage. Accid Anal Prev. 1992;24:673-678.

246. Roberts I, Norton R, Taua B. Child pedestrian injury rates: the importance of" exposure

to risk" relating to socioeconomic and ethnic differences, in Auckland, New Zealand. J

Epidemiol Community Health. 1996;50:162-165.

247. von Kries R, Kohne C, Bohm O, von Voss H. Road injuries in school age children:

relation to environmental factors amenable to interventions. Inj Prev. 1998;4:103-105.

248. DiMaggio C, Durkin M. Child pedestrian injury in an urban setting: descriptive

epidemiology. Acad Emerg Med. 2002;9:54-62.

249. Koepsell T, McCloskey L, Wolf M, et al. Crosswalk markings and the risk of pedestrian–

motor vehicle collisions in older pedestrians. JAMA. 2002;288:2136-2143.

250. Stevenson M, McClure R. Use of ecological study designs for injury prevention. Inj Prev.

2005;11:2-4.

251. McDonald NC, Brown AL, Marchetti LM, Pedroso MS. US school travel, 2009: an

assessment of trends. American Journal of Preventive Medicine. 2011;41:146-151.

252. Richmond S, Willan A, Rothman L, et al. The impact of pedestrian countdown signals on

pedestrian – motor vehicle collisions: A re-analysis of data from a quasi-experimental

study. Inj Prev. 2012;Published Online First: September, 23, 2013 doi:10.1136/

injuryprev-2012-040717.

253. Clifton K, J., Kreamer-Fults K. An examination of the environmental attributes

associated with pedestrian-vehicular crashes near public schools. Accid Anal Prev.

2007;39:708-715.

254. Calhoun AD, McGwin JG, King WD, Rousculp MD. Pediatric pedestrian injuries: a

community assessment using a hospital surveillance system. Acad Emerg Med.

1998;5:685-690.

255. Joly MF, Foggin P, Pless I. [Socioecological determinants of the risk of accidents in

young pedestrians]. Rev Epidemiol Sante Publique. 1991;39:345-351.

256. Frank L, Kerr J, Chapman J, Sallis J. Urban form relationships with walk trip frequency

and distance among youth. Am J Health Promot. 2007;21:305-311.

257. Department for Transport. Tomorrow's roads: safety for everyone - summary document,

2002. Available at:

http://www.ukroads.org/ukroadsafety/articlespapers/tomorrowsroadssaferforeveryone.pdf

. Accessed September, 2013.

Page 186: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

170

258. Schmidt M, Murray C, Nguyen H. Cohesive Communities, Improved Outcomes: A Case

for Small Schools. Education Canada. 2007;47:59-62.

259. Peel Public Health. Peel Health's Position Statement on the Impacts of the Built

Environment on Public Health 2006. Available at:

http://www.peelregion.ca/health/urban/pdf/PEEL-HEALTH-POSITION-

STATEMENT_Legal_2.0.pdf. Accessed September 2013.

260. Dunn J, Creatore M, Peterson E, Weyman J, Glazier. Final Report: Peel Healthy

Development Index. Toronto, Canada, 2009. Available at:

http://www.peelregion.ca/health/urban/pdf/HDI-report.pdf. Accessed September, 2013.

261. Parry D, Salsberg J, Macaulay A. A guide to researcher and knowledge-user

collaboration in health research, 2009. Available at: http://www.cihr-

irsc.gc.ca/e/documents/Guide_to_Researcher_and_KU_Collaboration.pdf. Accessed

September, 2013.

262. Bagley C. The urban environment and child pedestrian and bicycle injuries: Interaction of

ecological and personality characteristics. J Community Appl Soc Psychol. 1992;2:281-

289.

263. Alton D, Adab P, Roberts L, Barrett T. Relationship between walking levels and

perceptions of the local neighbourhood environment. Arch Dis Child. 2007;92:29-33.

264. Black C, Collins A, Snell M. Encouraging walking: the case of journey-to-school trips in

compact urban areas. Urban Stud. 2001;38:1121.

265. Buck C, Pohlabeln H, Huybrechts I, et al. Development and application of a moveability

index to quantify possibilities for physical activity in the built environment of children.

Health Place. 2011;17:1191-1201.

266. Carson V, Kuhle S, Spence JC, Veugelers PJ. Parents' perception of neighbourhood

environment as a determinant of screen time, physical activity and active transport. Can J

Public Health. 2010;101:124-127.

267. Johansson M. Environment and parental factors as determinants of mode for children's

leisure travel. J Environ Psychol. 2006;26:156-169.

268. McDonald NC. Critical factors for active transportation to school among low-income and

minority students. Evidence from the 2001 National Household Travel Survey. Am J Prev

Med. 2008;34:341-344.

269. Pabayo R, Gauvin L, Barnett TA. Longitudinal changes in active transportation to school

in canadian youth aged 6 through 16 years. Pediatrics. 2011;128:e404-e413.

270. Rodriguez A, Vogt CA. Demographic, environmental, access, and attitude factors that

influence walking to school by elementary school-aged children. J Sch Health. Jun

2009;79:255-261.

Page 187: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

171

271. Ziviani J, Scott J, Wadley D. Walking to school: incidental physical activity in the daily

occupations of Australian children. Occup Ther Int. 2004;11:1-11.

272. Green J, Muir H, Maher M. Child pedestrian casualties and deprivation. Accid Anal Prev.

2011;43:714-723.

273. McGuigan D. Deprivation and child pedestrian road casualties: Final Report. Belfast,

UK, 2010.

274. Pitt R, Guyer B, Hsieh CC, Malek M. The severity of pedestrian injuries in children: an

analysis of the Pedestrian Injury Causation Study. Accid Anal Prev. Dec 1990;22:549-

559.

275. Roberts I, Marshall R, Norton R. Child pedestrian mortality and traffic volume in New

Zealand. BMJ. Aug 1 1992;305:283.

276. Roberts I, Crombie I. Child pedestrian deaths: Sensitivity to traffic volume - Evidence

from the USA. Journal of Epidemiology & Community Health. 1995;49:186-188.

277. Rothman L, Howard A. Pedestrian crossing location influences injury severity in urban

areas. Inj Prev. 2012;18:365-370.

278. Stevenson MR, Laing BA, Lo SK. Factors contributing to the severity of childhood

pedestrian injury in Perth, Western Australia. Asia Pac J Public Health. 1992;6:25-31.

279. Wedagama P, D. M., Bird RN, Metcalfe AV. The influence of urban land-use on non-

motorised transport casualties. Accid Anal Prev. Nov 2006;38:1049-1057.

280. Yiannakoulias N, Scott DM, Rowe BH, Voaklander DC. Child pedestrian injuries and

urban change. Inj Prev. Feb 2011;17:9-14.

Page 188: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

172

Appendices

Appendix A: Search Strategies

Database Walkability Search Terms Pedestrian-Motor Vehicle

Collision Search terms

Medline walking or walkab* or pedestrian*

AND

social planning or city planning or

environment design or Geographic

Information Systems or Residence

Characteristics or Environment or global or

geographic* system* or gis or neighbo*or

*urban health or suburban health

accidents or accident prevention

or accidents, traffic

AND

social planning or city planning or

environment design or Geographic

Information Systems or Residence

Characteristics or Environment or

Walking or global or geographic*

system* or gis or walkab* or

neighbo*

PsycINFO As Medline As Medline

Scopus As Medline As Medline

Embase Pedestrian or walking or walkable

AND

city planning or environmental planning or

policy or geographic information system or

demography or Residence Characteristic* or

global or geographic* position* system* or

gis or urban health suburban health

traffic accident or accident

prevention

AND

city planning or environmental

planning or policy or geographic

information system or

demography or Residence

Characteristic* or walking or

global or geographic* position*

system* or gis or walkab* or

neighbo*

Transport Pedestrian or walkability or walkability

indicators or walkable communities or

walking

AND

Traffic concentration or traffic condition or

traffic conditions or traffic congestion or

traffic control or traffic control or traffic

control sign location or traffic control

systems or traffic density or traffic density

maps or traffic distribution or traffic

environment or traffic environments or

traffic fatalities or traffic filtering or traffic

flow or traffic incidents or traffic

infrastructure or traffic light or traffic

maintenance or traffic management or traffic

management or traffic schools or "traffic

sign and signals" or traffic studies or traffic

study or traffic surveillance or built

environment or built environments or

environment or environmental design or

traffic accident or traffic accidents

AND

(as in previous cell)

Page 189: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

173

environmental friendliness or social

planning or social policy or city or city

driving or city planning or city size or city

traffic or residence or residential areas or

residential area or residential areas or

residential density or residential

development or residential streets or

geographic information or geographic

information services or geographic

information systems or geographic

information systems or geographic

information systems or geographical

characteristics or geographical differences or

geographical distribution or geographical

information system or geographical

information systems or geographic

information system or urban or urban

automobiles or urban cars or urban design or

urban driving or urban health or suburb or

neighborhood walkability scale or

neighborhood or neighbourhood

Dissertations and

Theses

walking or walkability or pedestrian

AND

“social planning” OR environment* OR

“city planning” OR policy OR “geographic

information systems” OR demography OR

“residence characteristics” OR gis OR

neighbo*OR urban OR suburban

Traffic accident or accident

prevention

AND “social planning” OR

environment* OR “city planning”

OR policy OR “geographic

information system” OR

demography OR “residence

characteristics” OR gis OR

neighbo*OR urban OR suburban

OR walk*

SafetyLit walk or walkability or pedestrian

AND

environment or geographic or

neighbourhood or neighbourhood

Traffic or collision or crash

AND

social or environment or city or

geographic or gis or demography

or neighbourhood or neighborhood

or urban or walk

Web of Science walk* or walkability or pedestrian

AND

social or environment or city or geographic

or gis or demography or neighb* or urban or

city

Traffic or collision or crash

AND

social or environment or city or

geographic or gis or demography

or neighb* or urban or city

CINAHL walking or walkab* or pedestrian*

AND

residence characteristics or urban areas or

environment or geographic information

systems

Accidents, traffic

AND

residence characteristics or

walking or urban areas or

environment or geographic

information systems or pedestrians

Page 190: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

174

Appendix B: Summary of walking publications

Author, year

Design Participation

Rate

Outcome(s) of

interest

Location Walking

Data

Years

Population

/Sample Size

Age/Grade Walking

Data

Source

Built

environment

data source

Covariates

Alton, 2007 263 CS 82% students, 33%

parents

Walking frequency

last 7 days

Birmingham, UK - 6 schools,

473 participants

9-11 a,b a SES

Other demo

Black, 2001264 CS 36% Car usually to school UK* 1996 51 schools,

4214 participants

<11 b b SES

Other demo

Boarnet, 2005128 CS 39% Increased

walking/biking with

construction project

along route to school

California, US 2002-

2003

10 schools,

862 participants

grades 3-5 b g None

Braza, 2004123 CS 23% schools, 100%

students

Walked/biked to

school today

California,

US

1999 2993, 34 schools 9-11, grade 5 h g SES,

Other demo

Bringolf-Isler,

2008124

CS 65% Non-active commuting

usually to school

CH** 2004-

2005

1031 6-7 (grade 1),

9-10 (grade 4),

13-14 (grade 8)

b b,g SES,

Other demo

Buck, 2011265 CS - Walked to/from school Delmenhorst, DE 2007-

2008

596 6-10 b g SES,

Other demo

Buliung, 200937 CS - Walked to/from school

day prior to survey

Toronto, CA 2006 2009 11-13 d g Age

Carson, 2010266 CS 80% schools, 61%

students

Walk/bikes usually to

/from school

Alberta, CA 2008 148 schools,

3421 participants

grade 5 b b SES,

Other demo

Carver, 2009196 L 60% Change in walking/

biking trips /wk

Melbourne, AU 2004,

2006

19 schools,

170 participants

8-9 b g Other demo

de Vries, 2010130 CS - Walking/biking trips

per week for

transportation,

recreation/school

ND*** 2004-

2005

20 schools,

448 participants

6-11 g h SES,

Other demo

DiGuiseppi,

1998109

CS 84% Car to/from school London, UK - 2086 6-7, 9-10 b n SES,

Other demo

Evenson,

2006133

CS - Biking/walking

/skating to school > 1

day/wk

US**** 2002 480 Girls, 10-12,

(grades 6, 8)

a a Other demo

Ewing, 2004107 CS - Walking trips to

school

Alachua County

FLA,US

2000-

2001

709 school trips k-grade12 g g SES

Frank, 2007256 CS 30% At least 1 walking trip

in 2 days or Walked

>.5 mile/day

Atlanta, US 2001-

2002

3161 5-20 g g SES,

Other demo

Fyhri, 2009110 CS 62% Independent active

transport usually to

school/other activities

National, NO 2005 840 6-12 b c SES,

Other demo

Page 191: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

175

Gallimore,

2011208

CS - Ever walks to school

in typical wk

Utah, US 2007-8 2 schools grade 5 c h SES

Giles-Corti,

2011134

CS 69% Regular walking to

school > 6 trips/wk

Perth, AU 2007 238 schools,

1480 students,

1314 parents

10-12 b g SES,

Other demo

Greene, 2009132 CS 44% Walks/bikes usually to

school, Days in last

week walking >10

min/day

Oregon, US 2006 801 5-11 d g SES,

Other demo

Harten, 2004230 CS 73% schools, 80%

students

Active transportation

trips, previous Sunday,

2 weekdays

Adelaide, AU - 8 schools,

136 participants

11-12 e c SES,

Other demo

Johanson,

2006267

CS 67% Independent

walking/cycling to

leisure activities

Lund, Malmo, SE - 357 8 -11 b h SES,

Other demo

Kerr, 2006120 CS 28% time 1, 93%

time 2

Walks/bikes to school

> 1/week

Seattle, US - 259 4 -18 b b,g SES,

Other demo

Kerr, 2007126 CS 30% Walked > once over 2

days

Atlanta, US 2001-2 3161 5-18 g g SES,

Other demo

Kweon, 2006125 CS 56% Walks to school

usually, typical wk

College Station, US - 2 schools,

150 participants

grades 5-6 b b,g None

Leslie, 2010135 CS 51% schools, 92%

students

Walks/bikes usually

to/from school

3 states, AU 2006 231 schools,

2961 participants

10-14

(grades 6-8)

a a SES,

Other demo

Martin, 2007221 CS 32% cohort 1, 44%

cohort 2

Active travel > 1

day/wk

National, US 2004 2649 9 -15 d g SES,

Other demo

McDonald,

200735

TS 1990-87%, 1995-

34.3%, 2001-

38.9%- other years

Walked/biked to

school on survey day

National, US 1969, 77

83, 90,

95, 2001

- 5-18 b b SES,

Other demo

McDonald,

2008111

CS - Walks/bikes to school National, US 2001-2 6508 5-13 b b SES,

Other demo

McDonald,

2008268

CS 34% Walks/bikes to school National, US 2001-2 14,553 5-18 b b SES,

Other demo

McMillan,

2007117

CS - Walks/bikes to school California, US 16 schools,

1128 participants

grades 3-5 b b,h SES,

Other demo

Merom, 2006180 CS 67% Walks/bikes usually to

school on Mondays

NSW, AU 2002 812 5-12 d d SES,

Other demo

Mitra, 2010122 CS - Walked to/from school

day before interview

Toronto, CA 2001 8009 school trips 11-13 d g SES,

Other demo

Mitra, 2010137 CS - Walked to/from school

day before interview

Toronto, CA 2001 1088 traffic

analysis zones

11-13 d g SES,

Other demo

Mitra, 2012105 CS - Walked /biked to/from

school day before

Toronto, CA 2006 2190 11-12 d g SES,

Other demo

Napier, 2010207 CS 53% parents, 58%

students

Walk frequency/wk to

school

Daybreak, US 2010 193 students,

177 parents

grade 5 a g SES,

Other demo

Page 192: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

176

Pabayo, 2011269 TS 34% Walks/bikes usually to

school

National, CA 1996/97,

98/09,

2000/01

7690 k-10 f f SES,

Other demo

Panter, 2010206 CS - Walks usually to

school

Norfolk, UK 2007 2012 9-10 c cg SES,

Other demo

Panter, 2010121 CS - Walks usually to

school

Norfolk, UK 2007 2012 9-10 c g,h SES,

Other demo

Rodriguez,

2009270

CS 84% Walked to school

today

Michigan, US 2004 1897 7-13,

grades 3-5

a a SES,

Other demo

Rosenberg,

2009127

CS - Walks to

park/shops/to/from

school > 1/week

Boston, Cincinnati,

San Diego, US

2005 116 5-11 b b SES,

Other demo

Rossen,201168 CS - Usually walks to/from

school

Baltimore, US 2007 365 8-13,

(grades 3-6)

a h SES,

Other demo

Salmon, 2007112 CS 27% Walks/bikes usually >

once/week

8 Capitol Cities, AU 2004 720 4-13 d d SES,

Other demo

Timperio, 2006 108

CS 27% ages 5-6, 44%

ages 10-12

Walks/bikes usually >

once/wk

Melbourne, AU 2001 19 schools,

912 participants

10 -12,

(grades 5-6)

c b,g SES,

Other demo

Trapp, 2011129 CS 57% Walking > 1/2 of all

to/from school trips

Perth, AU 2007 25 schools,

1314 participants

9 to 13 a,g c,g SES,

Other demo

Wen, 2008113 RCT 87% students

71% parents

Active transport

usually to/from school

Sydney, AU 2005 24 schools,

1966 participants

10-12 a,b b Age

Yarlagadda,

2008136

CS 82% Walks to/from school

alone/with mother

San Francisco, US 2000 4352 <18 g b,g SES,

Other demo

Yelavich,

2008114

CS 85% schools, , 68%

parents

Walked to school

today

Dunedin, NZ 2004 1157 5-11,

(grades 1-6)

h b SES,

Other demo

Yeung, 2008115 CS 64% Active transport

to/from school

Brisbane,AU - 318 4-12 b b Other demo

Zhu, 2008116 CS 27% Walks usually to/from

school

Austin, US 2007 8 schools,

1281 participants

Elementary

schools

b b SES,

Other demo

Zhu, 2011131 CS 25% Walks usually to/from

school

Austin, US 2007 4 schools,

680 participants

Elementary

schools

b b SES,

Other demo

Ziviani, 2004271 CS 46% Walks to school >1 wk Brisbane, AU - 164 grades 1-7 b b Other demo

- Not specified

Design: CS = Cross sectional, L = Longitudinal, TS= Time Series

Walking data source: a = child questionnaire, b= parent/caregiver questionnaire, c= both child/parent questionnaire, d= telephone survey, e= child interview, f= parent

interview, g= travel diary, h= hand count

Built environment data source: a = child questionnaire, b= parent/caregiver questionnaire, c= both child/parent questionnaire, d= telephone survey, e= child interview,

f= parent interview, g= data files/GIS, h = field surveys

* Hampshire, Stoke- on-Trent, Crewe, ** Berne, Biel/Bienne, Payerne, *** Harlem, Rotterdam, Amersfoort, Schiedam, Vlaardingen, Hengelo, **** Baltimore,

Columbia, New Orleans, Minneapolis, Tucson, San Diego

Page 193: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

177

Appendix C: Summary of child pedestrian-motor vehicle collision publications

First author, year

Design

Outcome of

Interest Location

Injury Data

Years

Population/

Participants

Age of

Interest

Collision

Data source

Built

Environment

Data Source

Covariates

Agran, 1996202 CC Injury/fatality Orange County, US 1991-1993 39 cases, 62

controls

0-14 a,b

a Other demo

Bagley, 1992262 CSR, CC,

EC

Pedestrian/cyclist

injury/fatality

Brighton, UK - 400 <14 a

d SES

Braddock, 1991234 CSR, EC Injury/fatality Hartford, US 1986-1987 198 <15 e

d SES,

Other demo

Calhoun,254 CSR, EC Injury Jefferson County, US 1989-1991 91 <15 a

d SES,

Other demo

Christie, 1995201 CC Injury UK* - 152 cases,

484 controls

5-16 a,c

a,c SES,

Other demo

Clifton, 2007253 CSR, EC Injury severity

Collision

(injury/fatality)

Baltimore, US 2000-2002 163 public

schools,

1513

collisions

<15

e

d SES,

Other demo

Cloutier, 2008151 CSR, EC Injury/fatality Montreal, CA 1999-2003 968 5-14 e,f

d SES,

Other demo

DiMaggio, 2002248 CSR Injury/fatality New York City, US 1991-1997 32,578 0-19 e d Age

Dissanayake,

2009150

CSR, EC Injury/fatality Newcastle upon Tyne,

UK

2000-2005 522 <15 e

d None

Green, 2011272 CSR, EC Injury/fatality Leeds, Bradford, UK 2000-2005 2670 <17 e d SES

Joly, 1991255 CSR, EC Injury Montreal, CA 1980-1982 1006 0-14 a d None

Jones, 2005194 TS, EC Change in

collisions,

Injury/fatality

Two cities in UK 1992-2000 1560 4 to 16

e

a SES

Kupferberg-

Bendavid,1994 204

CSR Injury/fatality Montreal, CA 1980-1982 786 1-14 e

d

Other demo

LaScala, 2004152 CSR, EC Injury/fatality 4 Californian

communities, US

1992-1996 717 <16 e

d SES,

Other demo

Macpherson,

199842

CSR Injury/fatality Montreal, CA 1990-1994 2501 5 to 12 e

b None

McGuigan, 2010273 CSR, EC Injury/fatality Northern IE 1999-2008 3, 235 0-15 e

d SES,

Other demo

Mueller, 1990195 CC Injury/ fatality King County, US 1985-1986 98 cases, 196

controls

<15 c

a SES,

Other demo

Petch, 2000154 CSR, EC Injury/ fatality Salford, UK 1995-1998 556 <15 a,e d SES

Pitt, 1990274 CSR Injury/ fatality National, Urban US 1977-1980 1035 <20 e d Other demo

Page 194: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

178

Roberts, 1992275 TS, EC Fatality National, NZ 1967-1987 - <15 f d None

Roberts, 1995276 TS, EC Fatality National, US 1970-1988 - 0 to 14 f d None

Roberts, 1995210 CCROSS Injury/ fatality Auckland, NZ 1992-1994 46 5 to 15 a,b a Other demo

Roberts, 1995203 CC Injury/ fatality Auckland, NZ 1992-1994 190 cases,

380 controls

<15 a,b

a SES,

Other demo

Rothman, 2012277 CSR Injury severity Toronto, CA 2000-2009 1394 <18 e d Other demo

Stevenson, 1992278 CSR Injury severity Perth, AU 1980-1989 1282 0 to 14 d d Other demo

Stevenson, 1995197 CC Injury only Perth, AU 1991-1993 100 case, 200

control sites

1 to 14 a,d

a,c SES,

Other demo

Stevenson, 1996198 CC Injury only Perth, AU 1991-1993 97 cases, 360

controls

4 to 14 a,d

a,c,d SES,

Other demo

Stevenson, 1997199 CC Injury only Perth, AU 1991-1993 100 cases,

400 controls

1 to 14 a,d

a,c SES,

Other demo

Tester, 2004193 CC Injury/fatality Oakland, US 1995-2000 100 cases,

200 controls

<15 c,e

d SES,

Other demo

von Kries, 1998247 CC Injury/fatality Dusseldorf, DE 1993-1995 174 cases,

174 controls

6-14 e

a Age

Warsh, 200939 CSR, EC Injury/fatality Toronto, CA 2000-2005 2717 <18 e d None

Wazana, 2000200 CSR, EC Injury/fatality Hamilton, CA 1978-1994 2091 0-14 e d Other demo

Wedagama,

2006279

CSR, EC Injury/fatality Newcastle upon Tyne,

UK

1998-2001 - <17 e

d None

Yiannakoulias,

200240

CSR, EC Injury Edmonton, CA 1995-1999 258 0-15 a

d None

Yiannakoulias,

2011280

TS, EC Change in injury Edmonton, CA 1996-2007:

all injuries,

2000-2007:

severe

985 injuries.

162 severe

injuries

<19 c d SES

-Not specified

*Bradford, Bristol, London, Merthyr Tydfil, Reading

Design: CSR= Cross-sectional, retrospective, CC = Case control, CCROSS = Case crossover, TS = Time series, EC = Ecological

Collision data source: a = Hospital surveillance, b- Coroner surveillance, c = Trauma database, d = Police surveillance, e = Police reported database, f =Other databases

(e.g. insurance)

Built environment data source: a= Field survey, b=Parent report, c= Parent and child report, d= GIS

Page 195: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

179

Appendix D: Elementary school boundaries (TDSB) and pre-amalgamated City of Toronto

Page 196: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

180

Appendix E: Observational counts data collection form

School

Name:

School

ID:

Primary Location □ Secondary Location □

Observer’s Name:

_____________________

Date: ____________________

mm/dd/yy

Location of Observer (please circle side of school) North South East West

Weather (please circle as many as apply): Sun Cloud Rain Cold Hot

Car occupant tally: Younger (JK-grade 3) Older (grades 4-6)

Total number (jk-grade3): Total Number (grade 4-6):

Pedestrian Tally: Younger (JK-grade 3)

Older (grades 4-6)

Total number (jk-grade3): Total Number (grade 4-6):

Cyclist Tally: Younger (JK-grade 3)

Older (grades 4-6)

Total number (jk-grade3): Total Number (grade 4-6):

Scooters/roller blades: Younger (JK-grade 3) Older (grades 4-6)

Total number (jk-grade3): Total Number (grade 4-6):

% Confidence in Counts __________________

Page 197: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

181

Appendix F: Site survey

School name:

School ID:

Observer: Weather Conditions (check all that

apply):

� Sun � Cloud � Rain � Cold

� Hot

Date: ___/___/___

dd/mm/yy

Start Time:

___:___ am

Items to complete during school

drop-off period

1. Roadways surrounding school Yes No

a. Do cars appear to be driving too fast creating a dangerous pedestrian

environment on any roadways near the school?

� �

b. Please fill in posted speed limits : Front of school ______km/hr Border road 2 ______km/hr

Border road 3 ______km/hr Border road 4 ______km/hr

2. Student Car Drop-off

a. Please indicate whether a car drop-off area was visible to you during

observation period. Only answer b-f if you answered “yes”

� �

b. Is there designated car drop-off area (s)? How many? _______ � �

c. Where are children being dropped-off? Please check all that apply

On a roadway � In a driveway � In a parking lot � Other _____________________________

d. Did you see any drivers double park when dropping children off? � �

e. Do drivers drop off children on opposite side of the road from school and

children cross midblock with no crossing controls?

� �

f. Do cars wait blocking the vision of other motorists and pedestrians? � �

g. Is there congestion and backup around the school during drop-off time? � �

h. Please describe any particularly dangerous situation(s) re: car drop-offs, and potential solutions:

3. Bus Loading Zones Yes No

a. Please indicate whether bus loading zone was visible to you during

observation period. Only answer b and c if you answer “Yes”.

� �

b. Are children at risk of being hit by other vehicles when dropped off by bus? � �

c. Do other child pedestrians appear to be at risk of being hit by a school bus? � �

d. Please describe any particularly dangerous situation(s) in bus loading zones, & potential solutions:

4. Traffic/speed control measures around the school Please check all that apply:

� Different pavement surfaces � Non-white paint

� Speed bumps � Other Please describe:____________________

5. Adjacent Intersections to school Intersection 1

Streets________ +

_____

Intersection 2

Streets________ +

_______

a. Was this intersection visible to you during

observation period? Only answer b + c if “yes”

Yes

No

Yes No

� �

b. Was this a dangerous intersection? � � � �

Page 198: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

182

c. Please circle traffic control for each intersection

Stoplight

Stop sign(s)

Crossing guard

Other________

Stoplight

Stop sign(s)

Crossing guard

Other________

Please circle why you think intersection was dangerous -Circle intersection # for all that apply

ONLY ANSWER IF 5B = ‘YES”.

1 2 Children not listening to crossing guard

1 2 Children not following other traffic controls (e.g. lights)

1 2 High speed traffic

1 2 Cars parked blocking crossing controls

1 2 Cars not adequately following traffic controls

1 2 Large volume of cars

1 2 TTC vehicles blocking view

1 2 Other (please describe):

6. Are there any other areas where children cross that aren’t intersections? Yes � No �

If Yes, please answer:

Other Crossing Locations Crossing 1

Street

_________

Crossing 2

Street___________

a. Was this crossing location visible to you during the

observation period? Only answer b + c if “yes”

Yes

No

Yes No

� �

b. Was this a dangerous crossing location? � � � �

c. Are there any traffic controls (e.g. crossing guard)?

Please specify:

_____________

_______

___________________

Please circle why the crossing location was dangerous -Circle crossing location # for all that

apply. ONLY ANSWER IF 6B = ‘YES”.

1 2 Children not listening to crossing guard

1 2 Children not following other traffic controls (e.g. lights)

1 2 High speed traffic

1 2 Cars parked blocking crossing controls

1 2 Cars not adequately following traffic controls

1 2 Large volume of cars

1 2 TTC vehicles blocking view

1 2 Parents parking on opposite side of street to drop off

1 2 Parents backing up vehicles

1 2 Other (please describe):

Additional comments:

______________________________________________________________________________

______________________________________________________________________________

______________________________________________________________________________

Page 199: Child Pedestrian-Motor Vehicle Collisions and Walking to ... · light density, school crossing guards and lower school socioeconomic status were related to higher collision rates.

183

Appendix G: Vehicle speed data collection form

Please mark off 82 ft and record the number of seconds it takes for the car to travel the 82 ft.

School name:

School ID:

Data Collector Name: Date:___/___/___

Mm/dd/yr

Road Name:

Between which streets:

From (street name):

To (street name):

Time Measurements Done (total time

should be 20 minutes):

From:

To:

Weather (check all that apply):

� Sun � Cloud � Rain � Cold � Hot

Sampling Frame: Every ___________ car timed

Guidelines – small roads: every car, major roads: every 3-4 cars.

Vehicle # Duration Vehicle # Duration

1 ___ . ___ ___ secs 31 ___. ___ ___ secs

2 ___. ___ ___ secs 32 ___. ___ ___ secs

3 ___. ___ ___ secs 33 ___. ___ ___ secs

4 ___. ___ ___ secs 34 ___. ___ ___ secs

5 ___. ___ ___ secs 35 ___. ___ ___ secs

6 ___. ___ ___ secs 36 ___. ___ ___ secs

7 ___. ___ ___ secs 37 ___. ___ ___ secs

8 ___. ___ ___ secs 38 ___. ___ ___ secs

9 ___. ___ ___ secs 39 ___. ___ ___ secs

10 ___. ___ ___ secs 40 ___. ___ ___ secs

11 ___. ___ ___ secs 41 ___. ___ ___ secs

12 ___. ___ ___ secs 42 ___. ___ ___ secs

13 ___. ___ ___ secs 43 ___. ___ ___ secs

14 ___. ___ ___ secs 44 ___. ___ ___ secs

15 ___. ___ ___ secs 45 ___. ___ ___ secs

16 ___. ___ ___ secs 46 ___. ___ ___ secs

17 ___. ___ ___ secs 47 ___. ___ ___ secs

18 ___. ___ ___ secs 48 ___. ___ ___ secs

19 ___. ___ ___ secs 49 ___. ___ ___ secs

20 ___. ___ ___ secs 50 ___. ___ ___ secs

21 ___. ___ ___ secs 51 ___. ___ ___ secs

22 ___. ___ ___ secs 52 ___. ___ ___ secs

23 ___. ___ ___ secs 53 ___. ___ ___ secs

24 ___. ___ ___ secs 54 ___. ___ ___ secs

25 ___. ___ ___ secs 55 ___. ___ ___ secs

26 ___. ___ ___ secs 56 ___. ___ ___ secs

27 ___. ___ ___ secs 57 ___. ___ ___ secs

28 ___. ___ ___ secs 58 ___. ___ ___ secs

29 ___. ___ ___ secs 59 ___. ___ ___ secs

30 ___. ___ ___ secs 60 ___. ___ ___ secs

Please use back of sheet if there are more than 60 cars in the 20-minute interval.


Recommended