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
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
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.
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.
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.
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).
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
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
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
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
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
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
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
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
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
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
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
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.
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).
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.
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
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
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
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
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
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
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.
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
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
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
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
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).
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
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).
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,
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
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
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
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.
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.
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
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
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
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.
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).
.
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
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
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.
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).
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
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
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
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
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.
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.
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.
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
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
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.
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)
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
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.
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
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%).
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.
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)
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.
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.
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
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
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
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.
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.
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.
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
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.
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.
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.
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
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
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.
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
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.
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)
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
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
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)
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
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
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.
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.
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
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
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
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.
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
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
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
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)
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.
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)
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.
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.
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
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
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
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)
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.
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
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.
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
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)
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
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.
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.
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
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.
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
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).
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.
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)
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)
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)
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)
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%).
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
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
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
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.
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
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
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
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
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.
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.
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).
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.
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.
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
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
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
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
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.
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
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.
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
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.
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
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
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
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.
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.
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
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
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.
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
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.
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
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
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.
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
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.
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.
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
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
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.
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.
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.
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.
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:
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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)
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
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
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
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
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
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
179
Appendix D: Elementary school boundaries (TDSB) and pre-amalgamated City of Toronto
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 __________________
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? � � � �
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:
______________________________________________________________________________
______________________________________________________________________________
______________________________________________________________________________
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.