1722

JOURNAL-BASED CME ARTICLE (ARTICLE 1 OF 2)

Spinal Cord Injury-Functional Index: Item Banks to MeasurePhysical Functioning in Individuals With Spinal Cord InjuryDavid S. Tulsky, PhD, Alan M. Jette, PT, PhD, Pamela A. Kisala, MA, Claire Kalpakjian, PhD,Marcel P. Dijkers, PhD, Gale Whiteneck, PhD, Pengsheng Ni, MD, MPH, Steven Kirshblum, MD,Susan Charlifue, PhD, Allen W. Heinemann, PhD, Martin Forchheimer, MPP, Mary D. Slavin, PT, PhD,Bethlyn Houlihan, MSW, MPH, Denise G. Tate, PhD, Trevor Dyson-Hudson, MD, Denise G. Fyffe, PhD,

Steve Williams, MD, Jeanne Zanca, MPT, PhD

Both articles below must be read to complete the one 2-hour CME activity.

Article 1: Spinal Cord Injury-Functional Index: Item Banks to Measure PhysicalFunctioning of Individuals With Spinal Cord Injury

David S. Tulsky, PhD; Alan M. Jette, PT, PhD; Pamela A. Kisala, MA; Claire Kalpakjian, PhD;

Marcel P. Dijkers, PhD; Gale Whiteneck, PhD; Pengsheng Ni, MD, MPH; Steven Kirshblum, MD;

Susan Charlifue, PhD; Allen W. Heinemann, PhD; Martin Forchheimer, MPP; Mary D. Slavin, PT,

PhD; Bethlyn Houlihan, MSW, MPH; Denise G. Tate, PhD; Trevor Dyson-Hudson, MD; DeniseFyffe, PhD; Steve Williams, MD; Jeanne Zanca, MPT, PhD

Article 2: Development and Initial Evaluation of the Spinal Cord Injury-Functional Index

Alan M. Jette, PT, PhD; David S. Tulsky, PhD; Pengsheng Ni, MD, MPH; Pamela A. Kisala, MA;Mary D. Slavin, PT, PhD; Marcel P. Dijkers, PhD; Allen W. Heinemann, PhD; Denise G. Tate, PhD;Gale Whiteneck, PhD; Susan Charlifue, PhD; Bethlyn Houlihan, MSW, MPH; Steve Williams, MD;Steven Kirshblum, MD; Trevor Dyson-Hudson, MD; Jeanne Zanca, MPT, PhD; Denise Fyffe, PhD

Statement of Need

A major treatment goal in the rehabilitation of persons with spinal cord injury (SCI) is to maximizethe restoration of physical functioning. Documenting the extent of recovery is imperative for: 1)assessing treatment efficacy; 2) evaluating the cost-effectiveness of treatment interventions; 3) examiningthe impact of policy changes on patient outcomes; 4) evaluating the quality of care being provided; and5) providing appropriate, long-term prognostic information to patients and their families, as well as toinsurance carriers. In order to document recovery of rehabilitation interventions, reliable and valid toolsare necessary to assess physical functioning outcomes in the SCI population.

Several outcomes measures are currently used to assess physical functioning in SCI. The mostcommonly used scales (e.g., Functional Independence Measure) have 2 important shortcomings withrespect to their use in this population: comprehensiveness of the measure’s content to assess the fullrange of SCI severity and the breadth of content to ensure all important aspects of physicalfunctioning are covered, including the perspective of individuals with SCI in assessing outcomes. Itis difficult for any single instrument to include the large number of items necessary to cover therange of severity levels seen among persons with SCI.

These 2 articles will describe the development and evaluation of the Spinal Cord Injury-Functional Index (SCI-FI) a new comprehensive outcomes measurement tool for persons with SCI.

Accreditation Statement

This journal-based activity has been planned and developed in accordance with the Essential Areasand policies of the Accreditation Council for Continuing Medical Education (ACCME) through thesponsorship of Professional Education Services Group (PESG).

PESG is accredited by the ACCME to provide continuing medical education (CME) for physicians.

Credit Designation Statement

PESG designates this Journal-based CME activity for a maximum of 2.0 AMA PRA Category1 Credit(s)™. Physicians should claim only the credit commensurate with the extent of theirparticipation in the activity.

All other health care professionals completing continuing education credit for this activity willbe issued a certificate of participation.

Educational Objectives

To support the attainment of knowledge, competence, and performance, the learner should beable to achieve the following objectives:

1. Describe the current outcomes measurement tools being used with persons with SCI.2. Discuss the limitations of the current outcomes measurement tools being used with persons with SCI.3. Describe the development of a new outcome measurement tool with the item content and

structure being designed specifically for persons with SCI.4. Explain how the Spinal Cord Injury-Functional Index (SCI-FI) outcomes measurement tool

improves upon existing outcome measurement tools used with persons with SCI.

Planning Committee

Susan Charlifue, PhD; Marcel P. Dijkers, PhD; Trevor Dyson-Hudson, MD; Martin Forch-heimer, MPP; Denise Fyffe, PhD; Allen W. Heinemann, PhD; Bethlyn Houlihan, MSW, MPH; AlanM. Jette, PT, PhD; Claire Kalpakjian, PhD; Steven Kirshblum, MD; Pamela A. Kisala, MA;Pengsheng Ni, MD, MPH; Mary D. Slavin, PT, PhD; Denise G. Tate, PhD; David S. Tulsky, PhD;Gale Whiteneck, PhD; Steve Williams, MD, Jeanne Zanca, MPT, PhD; PESG staff.

Faculty Profiles & Disclosure Information

As a sponsor accredited by the ACCME, it is the policy of PESG to require the disclosure ofanyone who is in a position to control the content of an educational activity. All relevant financial

Arch Phys Med Rehabil Vol 93, October 2012

relationships with any commercial interests and/or manufacturers must be disclosed to participantsat the beginning of each activity. The faculty of this educational activity disclose the following:

Susan Charlifue, PhD

Craig Hospital, Englewood, CONo financial conflicts to disclose.

Marcel P. Dijkers, PhD

Mt. Sinai School of Medicine, New York, NYNo financial conflicts to disclose.

Trevor Dyson-Hudson, MD

University of Medicine and Dentistry of New Jersey-New Jersey Medical School, Department ofPhysical Medicine and Rehabilitation, Newark, NJNo financial conflicts to disclose.

Martin Forchheimer, MPP

University of Michigan Medical School, Department of Physical Medicine and Rehabilitation,Ann Arbor, MINo financial conflicts to disclose.

Denise Fyffe, PhD

University of Medicine and Dentistry of New Jersey-New Jersey Medical School, Department of PhysicalMedicine and Rehabilitation, Newark, NJNo financial conflicts to disclose.

Allen W. Heinemann, PhD

Northwestern University and Rehabilitation Institute of Chicago, Department of Physical Medicineand Rehabilitation, Chicago, ILNo financial conflicts to disclose.

Bethlyn Houlihan, MSW, MPH

Boston University School of Public Health, Health and Disability Research Institute, Boston, MANo financial conflicts to disclose.

Alan M. Jette, PT, PhD

Boston University School of Public Health, Health and Disability Research Institute, Boston, MANo financial conflicts to disclose.

Claire Kalpakjian, PhD

University of Michigan Medical School, Department of Physical Medicine and Rehabilitation,Ann Arbor, MINo financial conflicts to disclose.

Steven Kirshblum, MD

Kessler Institute for Rehabilitation, West Orange, NJNo financial conflicts to disclose.

Pamela A. Kisala, MA

University of Michigan Medical School, Department of Physical Medicine and Rehabilitation,Ann Arbor, MINo financial conflicts to disclose.

Pengsheng Ni, MD, MPH

Boston University School of Public Health, Health and Disability Research Institute, Boston, MANo financial conflicts to disclose.

Mary D. Slavin, PT, PhD

Boston University School of Public Health, Health and Disability Research Institute, Boston, MANo financial conflicts to disclose.

Denise G. Tate, PhD

University of Michigan Medical School, Department of Physical Medicine and Rehabilitation,Ann Arbor, MINo financial conflicts to disclose.

David S. Tulsky, PhD

University of Michigan Medical School, Department of Physical Medicine and Rehabilitation,

Ann Arbor, MINo financial conflicts to disclose.

1723SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

Gale Whiteneck, PhD

Craig Hospital, Englewood, CONo financial conflicts to disclose.

Steve Williams, MD

Boston Medical Center, New England Regional Spinal Cord Injury Center, Boston, MANo financial conflicts to disclose.

Jeanne Zanca, MPT, PhD

Mt. Sinai School of Medicine, New York, NYNo financial conflicts to disclose.

PESG Staff

No financial conflicts to disclose.

Resolution of Conflict of Interest

PESG has implemented a process to resolve conflict of interest for each CME activity. Inorder to help ensure content objectivity, independence, and fair balance, and to ensure that the

content is aligned with the interest of the public, PESG has resolved the conflict by external

content review.

Unapproved/Off-Label Use Disclosure

PESG requires CME faculty to disclose to the participants:

1. When products or procedures being discussed are off-label, unlabeled, experimental, and/or

investigational (not US Food and Drug Administration [FDA] approved); and

2. Any limitations on the information presented, such as data that are preliminary or that

represent ongoing research, interim analyses, and/or unsupported opinion. Faculty may

discuss information about pharmaceutical agents that is outside of FDA-approved labeling.

This information is intended solely for CME and is not intended to promote off-label use ofthese medications. If you have questions, contact the medical affairs department of themanufacturer for the most recent prescribing information.

Intended Audience

This program is intended for physicians and healthcare professionals responsible for the

comprehensive care for individuals with chronic illness and disabilities.

Method of Participation

In order to claim credit, participants must complete the following:1. Pre-activity self-assessment questions2. Read the 2 articles included in this activity.3. Complete the CME Test and Evaluation. Participants must achieve

a score of 70% on the CME Test.Participants can complete the pre-activity self-assessment and CME

Test and Evaluation online by logging on to http://acrm.cds.pesgce.com.Upon successful completion of the online tests and evaluation form, youcan instantly download and print your certificate of credit.

To better define and meet the CME needs of health care profession-

comes-measurement survey following the conclusion of the program.This follow-up survey is designed to measure changes to participants’practice behaviors as a result of their participation in this CME activity.You will be contacted by email 60 days following the conclusion of thisactivity with an outcomes measurement survey. We would greatlyappreciate your participation.

CME Inquiries

For all CME certificate inquiries, please contact us at support@pesgce.com.This continuing education activity is active starting October 1,

2012 and will expire September 30, 2013.

als and enhance future CME activities, PESG will conduct an out- Estimated Time to Complete This Activity: . . . . . . . . 2.0 hours

Arch Phys Med Rehabil Vol 93, October 2012

itt

o(a

d

Ol

ort

p

r

1724 SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

A

ABSTRACT. Tulsky DS, Jette AM, Kisala PA, KalpakjianC, Dijkers MP, Whiteneck G, Ni P, Kirshblum S, Charlifue S,Heinemann AW, Forchheimer M, Slavin MD, Houlihan B,Tate DG, Dyson-Hudson T, Fyffe D, Williams S, Zanca J. Spi-nal Cord Injury-Functional Index: item banks to measure phys-ical functioning in individuals with spinal cord injury. ArchPhys Med Rehabil 2012;93:1722-32.

Objectives: To develop a comprehensive set of patient-reportedtems to assess multiple aspects of physical functioning relevant tohe lives of people with spinal cord injury (SCI), and to evaluatehe underlying structure of physical functioning.

Design: Cross-sectional.Setting: Inpatient and community.Participants: Item pools of physical functioning were devel-

ped, refined, and field tested in a large sample of individualsN�855) with traumatic SCI stratified by diagnosis, severity,nd time since injury.

Interventions: None.Main Outcome Measure: Spinal Cord Injury-Functional In-

ex (SCI-FI) measurement system.Results: Confirmatory factor analysis (CFA) indicated that a

5-factor model, including basic mobility, ambulation, wheel-chair mobility, self-care, and fine motor function, had the bestmodel fit and was most closely aligned conceptually withfeedback received from individuals with SCI and SCI clini-cians. When just the items making up basic mobility weretested in CFA, the fit statistics indicated strong support for aunidimensional model. Similar results were demonstrated foreach of the other 4 factors, indicating unidimensional models.

Conclusions: Though unidimensional or 2-factor (mobilityand upper extremity) models of physical functioning make upoutcomes measures in the general population, the underlyingstructure of physical function in SCI is more complex. A5-factor solution allows for comprehensive assessment of keydomain areas of physical functioning. These results informedthe structure and development of the SCI-FI measurementsystem of physical functioning.

Key Words: Activities of daily living; Mobility limitation;utcome assessment (health care); Psychometrics; Quality of

ife; Rehabilitation; Self care; Spinal cord injuries; Walking.

From the Department of Physical Medicine and Rehabilitation, University ofMichigan Medical School, Ann Arbor, MI (Tulsky, Kisala, Kalpakjian, Forchheimer,Tate); Health and Disability Research Institute, Boston University School of PublicHealth, Boston, MA (Jette, Ni, Slavin, Houlihan); Mount Sinai School of Medicine,New York, NY (Dijkers, Zanca); Craig Hospital, Englewood, CO (Whiteneck, Char-lifue); Kessler Institute for Rehabilitation, West Orange, NJ (Kirshblum); Departmentof Physical Medicine and Rehabilitation, University of Medicine and Dentistry ofNew Jersey-New Jersey Medical School, Newark, NJ (Kirshblum, Dyson-Hudson,Fyffe); Department of Physical Medicine and Rehabilitation, Northwestern Univer-sity, Chicago, IL (Heinemann); Rehabilitation Institute of Chicago, Chicago, IL(Heinemann); Kessler Foundation, West Orange, NJ (Kirshblum, Dyson-Hudson,Fyffe); and New England Regional Spinal Cord Injury Center, Boston MedicalCenter, Boston, MA (Williams).

Supported by the U.S. Department of Education, National Institute of Disability andRehabilitation Research (grant nos. H133N060022, H133N060024, H133N060014,H133N060005, H133N060027, and H133N060032) and by the National Institutes ofHealth, National Institute of Child Health & Human Development, National Center forMedical Rehabilitation Research, and National Institute of Neurological Disorders andStroke (grant no. 5R01HD054659).

No commercial party having a direct financial interest in the results of the researchsupporting this article has or will confer a benefit on the authors or on any organi-zation with which the authors are associated.

Reprint requests to David S. Tulsky, PhD, Dept of Physical Medicine and Reha-bilitation, North Campus Research Complex, 2800 Plymouth Rd, Building NCRCB520, Office 3210, Ann Arbor, MI 48109-2800, e-mail: dtulsky@med.umich.edu.

In-press corrected proof published online on Jul 30, 2012, at www.archives-pmr.org.

0003-9993/12/9310-00191$36.00/0http://dx.doi.org/10.1016/j.apmr.2012.05.007

rch Phys Med Rehabil Vol 93, October 2012

© 2012 by the American Congress of RehabilitationMedicine

AMAJOR TREATMENT goal in the rehabilitation of per-sons with spinal cord injury (SCI) is to maximize the

restoration of physical functioning. Documenting the extent ofrecovery is imperative for (1) assessing treatment efficacy; (2)evaluating the cost-effectiveness of treatment interventions; (3)examining the impact of policy changes on patient outcomes;(4) evaluating the quality of care being provided; and (5)providing appropriate, long-term prognostic information to pa-tients and their families, as well as to insurance carriers.1-3 Inrder to document recovery of rehabilitation interventions,eliable and valid tools are necessary to assess physical func-ioning outcomes in the SCI population.

Several outcomes measures are currently used to assesshysical functioning in SCI.4-13 The most commonly used

scales (eg, the FIM) have 2 important shortcomings with re-spect to their use in this population: comprehensiveness of themeasure’s content to assess the full range of SCI severity andthe breadth of content to ensure that all important aspects ofphysical functioning are covered, including the perspective ofindividuals with SCI, in assessing outcomes. It is difficult forany single instrument to include the large number of itemsnecessary to cover the range of severity levels seen amongpersons with SCI. For example, an instrument designed for usein individuals with high-level tetraplegia is not likely to havesufficient range to be meaningful if used with individuals withparaplegia. Yet, comprehensive range is an essential measure-ment property for an outcome instrument designed to assesschange in physical functioning after any level and complete-ness of SCI. The current measures of physical functioning varyin terms of both their range and the breadth of physical func-tioning items covered; for instance, the Walking Index forSpinal Cord Injury9 focuses only on ambulation, and the Quad-iplegia Index of Function (QIF)13 on physical functioning

among individuals with tetraplegia. The Spinal Cord Indepen-dence Measure III14 and the FIM4 focus on a greater breadth ofphysical functioning, but this breadth does not necessarilycover all aspects of physical functioning. For example, none ofthe current measures assesses fine motor function apart fromthat which may be implied by self-care. Measures developedfor the general population often assess ambulation wholly interms of mobility, while for persons with SCI, the use of a

List of Abbreviations

AM-PAC Activity Measure for Post-Acute CareCAT computer adaptive testCFA confirmatory factor analysisCFI comparative fit indexIRT item response theoryNeuro-QOL Neurology Quality of Life Measurement

SystemPROMIS Patient-Reported Outcomes Measurement

Information SystemPRO patient-reported outcomeQIF Quadriplegia Index of FunctionRMSEA root mean square error of approximationSCI spinal cord injurySCI-FI Spinal Cord Injury-Functional IndexSCIMS Spinal Cord Injury Model System

TLI Tucker-Lewis Index

tisampsabnairctacm

stMtnemg

ifi

itcbhan

eiSsaaQwvNcag

1725SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

wheelchair (which some consider an extension of their bod-ies15) constitutes mobility.

An important shortcoming of existing measures is the failureo incorporate the perspective of individuals with SCI in thetem development process, and the content of the current mea-ures (eg, FIM, QIF) being selected by rehabilitation cliniciansnd researchers. These measures rely exclusively on perfor-ance, observation, and the use of clinician raters. While

rofessional observer–rated methods provide an importantource of information, there are many activities that cannot bessessed directly, and observational methods are subject to biasy raters. Patient-reported outcomes (PROs) provide an alter-ative method of assessment and are increasingly recognizeds an essential component of outcome measurement in clin-cal research and clinical practice. A PRO is defined as anyeport on a patient’s health condition or health status thatomes directly from the patient, without the interpretation ofhe response by a clinician or anyone else.16 Many havergued that PROs are essential to the measurement of out-omes and should be collected alongside performance-basedeasures.17-19

In response to growing recognition of the value of PROs, theNational Institutes of Health initiated the Patient-ReportedOutcomes Measurement Information System (PROMIS) in2004 to develop new measures for use in clinical research andhealth care delivery.20 In a parallel project, to address thepecific health-related quality of life issues relevant to popula-ions with neurologic disorders, the Neurology Quality of Life

easurement (Neuro-QOL) System was developed.21 Despitehe broad scope of the work, the PROMIS and Neuro-QOL didot address issues specific to the SCI population. Like otherxisting instruments, the PROMIS and Neuro-QOL instru-ents respectively conceptualize physical functioning as 1

eneral domain (PROMIS—overall physical functioning)22 or2 general domains (Neuro-QOL—mobility and upper extrem-ity functioning).23 With SCI, however, given the nature of thenjury, impairments impact many areas of functioning in dif-erent ways, leading one to conclude that physical functionings a multifactorial construct15 and to question whether a 1- or

2-factor measurement scale is adequate, especially for thoseclinical and research situations where a detailed assessment ofthe person’s abilities and disabilities in multiple aspects ofphysical functioning is needed.

The aims of this study were to develop a comprehensive setof items that assess multiple aspects of physical functioningrelevant to the lives of people with SCI, and to evaluate theunderlying conceptual structure of physical functioning inthese individuals. Specifically, we sought to determine whethera 1- or 2-factor model of physical functioning or a morecomplex model better summarizes the empirical Spinal CordInjury-Functional Index (SCI-FI) data and is therefore moreappropriate for use in an SCI population. If distinct dimensionsof functioning are shown to exist, item banks to measure eachsubdomain separately are necessary. The current work sets thestage for item response theory (IRT) analyses and the devel-opment of a computer adaptive test (CAT) for each item bank,modern methods to create and administer measures that arereported in a related article in this issue.24 The instrument wedeveloped is called the SCI-FI.

METHODS

Development of the SCI-FI Item PoolsSCI-FI development (fig 1) began with a thorough literature

review, input from focus groups conducted with individuals

with SCI and SCI clinicians, and feedback from experts in SCI

rehabilitation. As reported by Slavin et al,15 12 focus groupswith individuals who had SCI and 6 focus groups with SCIclinicians were held at 6 Spinal Cord Injury Model System(SCIMS) centers, and a rigorous qualitative analysis was con-ducted to extract potential activities, skills, and tasks to beincluded in the SCI-FI.25 Focus group feedback was used todentify important domains and subdomains of physical func-ion, including several categories of mobility (manual wheel-hair, power wheelchair, ambulation, and transfers/changingody position), self-care and upper extremity functioning (fineand use/manipulating objects, lifting objects, reaching/handnd arm use, toileting), sexual functioning, and use of commu-ication devices.15

A central goal in developing the item pool was to constructdirect links with some existing PRO scales, which would beaccomplished by placing items from these different existingPRO measurement tools on a common metric and creating astatistical link between the scores of the 2 instruments.26 Bymbedding items from the Neuro-QOL and PROMIS verbatimnto the SCI-FI, a statistical link can be created to convertCI-FI scores to scores yielded by those measures, throughcore conversion tables. Such linking is important when scoresre being compared with those obtained in other populationsnd groups. For the SCI-FI study, the PROMIS and Neuro-OL item banks were reviewed for content that overlappedith the suggestions made in the focus groups with SCI indi-iduals and clinicians, and where appropriate, PROMIS andeuro-QOL items were embedded verbatim into the pool of

andidate items of the SCI-FI measure. The PROMIS containslarge item bank of 124 calibrated items measuring a single,

eneral factor of physical functioning,22 whereas the Neuro-QOL contains 2 item banks measuring upper extremity (20items) and mobility (19 items), respectively.23 After a detailedreview of the PROMIS and Neuro-QOL items, a total of 33PROMIS items and 37 Neuro-QOL items (19 of which werecommon to both measures) were selected based on content rele-vance. The Neuro-QOL item banks incorporate a series of itemsfrom the Activity Measure for Post-Acute Care (AM-PAC)26,27;therefore, these AM-PAC items were also included in the SCI-FI.By including the AM-PAC and the Neuro-QOL items within theSCI-FI and cocalibrating all items using IRT, it is possible tocreate linking lookup tables, and therefore investigators can obtaina total score equivalent for either the AM-PAC or Neuro-QOL,even though only a subsample of items for either measure iscontained within the SCI-FI CAT. Furthermore, items were in-corporated from a pediatric SCI-targeted measure of physicalfunction,28,29 which will allow linking across the lifespan. Thislinking is important when children with SCI are enrolled in studiesand will be retested longitudinally into adulthood. The final link-age step was to ensure that there were common items with aPROMIS physical functioning supplemental scale for individualswho use adaptive technology, which was being developed con-currently with the SCI-FI.17

The current investigators added a total of 619 new items tothese legacy items based on the new physical functioningcontent suggested by the focus groups. After investigatorsmade an initial pass at deleting items for redundancy, severalsteps were taken (see fig 1) to ensure the content validity of theremaining new items (the legacy items had previously under-gone similar scrutiny). First, cognitive debriefing interviewswere conducted with individuals with SCI who in a structuredinterview read each item, responded based on their level offunctional ability, and then reviewed the meaning of the itemand their thought process behind their response. A minimum of5 individuals with SCIs reviewed each newly written item and

discussed the language, comprehensibility, ambiguity, and,

Arch Phys Med Rehabil Vol 93, October 2012

tfUJMCvtu(pas

cess.

1726 SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

A

most importantly, the relevance of the item to individuals withSCI. The cognitive interview process we used has been de-scribed by DeWalt et al.30

As a second step to ensure that the phrasing of the new itemswas robust and would be amenable to future Spanish transla-tion, a team of native Spanish-speaking translation scienceexperts reviewed each item for translatability to Spanish andprovided feedback on specific words and phrases that wouldlose meaning during translation. This team identified multipleinstances of problematic wording and grammar, and in eachcase suggested an alternative way to phrase the item so that thefinal scale would be translation-ready.

A third step was to perform a reading level analysis, reviewingall of the new items for readability, ensuring that no item waswritten above a third-grade reading level. We used the LexileFrameworka to evaluate each item. Items were rewritten if thereview identified that they were above a third-grade reading level.

At this stage, the 124 legacy items were merged back in with the204 final new items, for a complete item pool consisting of 328 items.Finally, the coinvestigators reviewed the item banks and suggestedwording changes to new items and binned the items into subdomainsof physical functioning (eg, changing and maintaining body position,transfers, walking and running, wheelchair mobility, bathing, eating,grooming, toileting, sexual functioning, fine hand use/manipulatingobjects, and use of communication devices) (see fig 1).15 The quali-

Be

Changing &

Maintaining Body

Posi on

Walking &

Running Bathin

Wheelchair

Use gnirrefsnarT

Exis ng Items

AMPAC NeuroQOL PROMIS PROMIS AT Shriners

124

25 37 33 6

57

CFA: Confirm

Models Basic Mobility Ambula on

Wheelchair Mobility

Basic Mobility Ambula on Wheelchair

Mobility

65 93 45

CFA: Confirm

Unidimension-

ality

Fig 1. SCI-FI item bank development pro

tative review process, from focus groups and item development to

rch Phys Med Rehabil Vol 93, October 2012

item binning, suggested that the structure of physical functioning inindividuals with SCI is more complex than a simple 1- or 2-factormodel, and the next phase of the study (described below) evaluatedalternative, more complicated structures of physical functioning.

Calibration StudyParticipants. The item calibration sample used for field

esting included 855 participants with traumatic SCI recruitedrom 6 SCIMS centers: New England Regional SCI Center,niversity of Michigan Model SCI System, Northern New

ersey SCI System, Rocky Mountain Regional SCI System,ount Sinai SCI Model System, and Midwest Regional SCIare System. The institutional review board at each site re-iewed and approved this study. Inclusion criteria specifiedhat participants were 18 years or older and could speak andnderstand English fluently. The sample was stratified by levelparaplegia vs tetraplegia) and completeness of injury (com-lete vs incomplete), as well as time since injury (�1y, 1–3y,nd �3y) to ensure that the sites recruited a heterogeneousample of individuals with SCI.31

Data collection procedures. All items were presented toparticipants by trained data collectors in an interview format,either in person or over the phone. Because some items (eg,ambulation, wheelchair use, bowel and bladder management)apply to only certain subgroups of individuals, participants first

Literature Focus Groups

Cogni on Debriefing

Item Wri ng (k=619)

Translatability Evalua on

Reading Level Analysis

Field Tes ng (k=328)

g Items

Grooming Toile ng Sexual

Func oning

Fine Hand Use/ Manipula ng

Objects

Communica-

on Device

Use ng

Subset 1 k=119; Subset 2 k=115

New SCI-FI Items, k=204

r Func�onotoMeniF

r Func�onotoMeniF

era-C fleS

era-C fleS

63 09Final Items k=275

Abbreviation: AT, assistive technology.

ginnin

g aE

answered screening questions to determine the need to admin-

i

d

mbmm

cwl.

wswawfrRbst

1727SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

ister these supplemental items beyond the basic set adminis-tered to every respondent. Participants were asked to respondto each item based on their capacity to perform the activitywithout special equipment or help from another person, exceptwhen such help was explicitly stated in the question. Partici-pants were shown a response card to help guide them throughthe interview, because the response options differed somewhatfrom one block of items to the next. All responses were enteredinto a web-based data collection system, which allowed data tobe automatically uploaded and stored on a secure server im-mediately. The study team reviewed the data for quality andadherence to the sampling stratification quotas.

Data analysis. Based on the focus group feedback regard-ng the important subdomains of physical functioning in SCI,15

several competing models of physical functioning were devel-oped and tested. Confirmatory factor analysis (CFA) was usedfor this analysis because it allows for direct comparison ofalternative models. Because of the ordinal nature of the items,the CFAs were conducted based on polychoric correlationmatrices with weighted least squares with mean and varianceadjustment parameter estimation.

To explore the factor structure, different models of in-creasing complexity were created, fit to the data, and com-pared with each other using a competing models approach.Several indices of model fit for each of the more compli-cated models were compared against the fit indices of ageneral 1-factor model of physical functioning and a 2-fac-tor model (specifying a general mobility factor and generalself-care factor). The 1-factor model parallels the PROMISstructure, whereas the 2-factor model parallels the Neuro-QOL structure. Several additional, more complex modelswere also outlined that reflect constructs outlined in theInternational Classification of Functioning, Disability andHealth model32 of impairments, activity limitations and par-ticipation restrictions, and distinctions of function that weresuggested by our qualitative review of focus groupfeedback.

Each successive measurement model was evaluated in termsof indices of goodness-of-fit. These indices were selected be-cause they are moderately sensitive to effects of sample sizeand degrees of freedom.33-36 The chi-square index divided byegrees of freedom (�2/df) was used to calculate the Tucker-

Lewis Index (TLI), which is a nonnormed comparative fitindex (CFI) that makes adjustment for the number of degrees offreedom in the model.37 Bentler’s CFI,38 which also compares

odel specifications to baseline models, was calculated. Foroth the TLI and CFI, values greater than .90 indicate goododel fit, with those above .95 indicating extremely goododel fit.35 In addition, the root mean square error of

approximation (RMSEA),39 which compensates for modelomplexity, was calculated. Whereas exact fit to a modelould be indicated by an RMSEA value of .00, values of

ess than .08 indicate reasonable model fit, while values of06 or lower indicate a very close fit.

To prepare for the CFAs, a final content analysis of the itemsas performed to evaluate the authors’ binning of items and to

elect the most representative items for the initial analyses,hich is described below. Model specification proceeded par-

llel to this effort, and items from the binning were includedithin the specified models. Within the self-care and fine motor

unction item banks, there were 3 content-specific subdomains,elated to communication, toileting, and sexual functioning.ather than develop these areas of functioning as separate itemanks, the research team binned these items along with otherelf-care and fine motor function items because the item con-

ent focused on the use of upper extremity or, more specifically,

fine motor movement that is involved in each of these activities(eg, using the keypad on a touch-tone phone). Because of theinclusion of sex-specific items (eg, shaving, tampon use), theself-care items completed by men and women varied slightly,and therefore this item bank was assessed separately by sex.Model specifications can be found in table 1.

A limitation of the CFA procedure was that given the largepool of 328 items, a sample size of 855 participants is notsufficient for model identification; there are too many items tobe analyzed simultaneously. However, the items were believedto demonstrate sufficient redundancy, and it was assumed that

Table 1: Model Specifications for CFA

Model and Factors Domain(s)

Model 1 (1 factor) Physical functionModel 2 (2 factors)

Factor 1 Upper extremityFactor 2 Mobility

Model 3 (3 factors)Factor 1 Upper extremityFactor 2 Mobility (basic, wheelchair,

ambulation)Factor 3 Communication

Model 4 (3 factors)Factor 1 Upper extremity–self-careFactor 2 Upper extremity–fine motor

functionFactor 3 Mobility (basic, wheelchair,

ambulation)Model 5 (4 factors)

Factor 1 Basic mobilityFactor 2 Wheelchair mobilityFactor 3 AmbulationFactor 4 Upper extremity

Model 5a (same as model 5 butrun without ambulation)

Model 5b (same as model 5 butrun without wheelchair)

Model 6 (4 factors)Factor 1 Upper extremity–fine motor

functionFactor 2 Upper extremity–self-careFactor 3 Basic mobilityFactor 4 Ambulation and wheelchair

mobilityModel 7 (4 factors)

Factor 1 Upper extremity–fine motorfunction

Factor 2 CommunicationFactor 3 Upper extremity–self-careFactor 4 Mobility (basic, wheelchair,

ambulation)Model 8 (5 factors)

Factor 1 Upper extremity–self-careFactor 2 Upper extremity–fine motor

functionFactor 3 Basic mobilityFactor 4 Wheelchair mobilityFactor 5 Ambulation

Model 8a (same as model 8 butrun without ambulation)

Model 8b (same as model 8 but

run without wheelchair)

Arch Phys Med Rehabil Vol 93, October 2012

embnolo

1728 SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

A

model evaluation could proceed with a subset (approximatelyone-third) of the items. Therefore, items were reviewed and asubset of 119 items selected for the initial analyses, withrepresentative items (ie, those items that were expected to havethe highest loading on each respective factor) being selectedfrom among all of the bins (or domains) that are describedabove. The initial evaluation included 8 alternative factor mod-els. To determine the robustness of the factor structure andensure that the initial results were not an artifact of the itemselection process, the models specified in the CFAs were thenretested using a second subset of 115 items, selected fromacross all of the bins or domains that were not included in theinitial analyses, and the model fit statistics for the 8 alternativefactor models were compared between the 2 item subsets. Thisreplication analysis allowed us to determine the robustness ofthe factor structure and ensured that the initial results were notan artifact of the item selection process.

In preparing the data for analysis, we imputed all responsesto the walking items as unable to do/cannot do for subjects withcomplete injuries who responded that they were never able towalk in the relevant screening question. This allowed us toinclude them in the overall analyses. Moreover, a large per-centage of the sample used a wheelchair or ambulated, but didnot do both. As the alternative factor models were developed,there was concern that the strong inverse relationship betweenwheelchair and ambulation items would impact the factorstructure. Specifically, the lack of covariance between theambulation and wheelchair mobility factors in SCI populationscould cause model identification problems when the factorstructure was tested through structural equation modeling orCFA methods.

Consequently, we decided that some models were to betested including the ambulation items but without the wheel-chair items or, alternatively, tested with the wheelchair itemswhile excluding the ambulation items.

After these initial CFAs, each bank was analyzed separatelyto assess its unidimensionality. Rather than pit successive mod-els against one another, for this analysis single unidimensionalmodels were run for each of the factors identified as significantin the previous step (eg, basic mobility). These subdomainanalyses were conducted and evaluated using the fit statisticsdescribed above. All CFAs were conducted using MPlus ver-sion 6.0.b To ensure that the item banks met the assumptionsnecessary for subsequent IRT analysis,24 it was necessary toassess the local independence of items in each bank, whichmeans that for individuals with the same level of the constructbeing measured, responses to each item should be independentof one another.40 If this is not the case, something other thanthe construct being measured is influencing the variation ina cluster of 2 or more items, which violates the assumptionof unidimensionality. To test for local item dependence,residual correlations between items were examined, anditems exhibiting absolute residual correlations greater than

Table 2: SCI-FI Item

SCI-FI Version New Neuro-QOLShriners’ Pediatri

SCI-CAT

Original item pool 204 37 57Final item banks 165 34 49

*The number in parenthesis is the number of AM-PAC items that is†The number in parenthesis is the number of PROMIS items that is‡The number in parenthesis is the number of PROMIS adaptive tech

0.2 were removed. N

rch Phys Med Rehabil Vol 93, October 2012

RESULTSAs shown in table 2, the SCI-FI item pool for field testing

comprised 328 items, including 204 new, SCI-specific physicalfunctioning items, 37 Neuro-QOL physical health items, 25AM-PAC items, 33 PROMIS physical functioning items, 57

ribution by Source

AM-PAC*PROMIS

Version 1.0†PROMIS Adaptive

Technology Supplement‡ Total

25 (9) 33 (19) 6 (6) 32823 (9) 29 (16) 5 (5) 275

edded in the Neuro-QOL.edded in the Neuro-QOL.gy items that is embedded in the PROMIS.

Table 3: Background Characteristics of the Sample

Variable Mean � SD or n (%)

Age (y) 43.1�15.3Age at injury (y) 36.3�15.7Time since injury (y) 6.8�9.3Sex

Men 657 (77)Women 198 (23)

EthnicityHispanic 97 (11)Non-Hispanic 751 (88)Unknown/refused 7 (1)

RaceWhite 602 (70)Black 148 (17)Asian 17 (2)American Indian/Alaskan Native 5 (1)More than 1 race 72 (8)Declined 11 (1)

DiagnosisParaplegia 390 (46)Tetraplegia 465 (54)

Type of injuryComplete 393 (46)Incomplete 462 (54)

Central cord syndrome 30 (4)Mechanism of injury

Motor vehicle accident 300 (35)Fall 205 (24)Gunshot wound/violence 99 (12)Diving 73 (9)Other sports 75 (9)Medical/surgical complication 42 (5)Other 58 (7)

Current living situationHome 665 (78)Initial rehabilitation 166 (19)Skilled nursing or long-term care 24 (3)

Use a bowel and bladder program 679 (79)Walk some or all of the time 228 (27)Use a manual wheelchair some or all

of the time 438 (51)Use a power wheelchair some or all of

the time 358 (42)

Dist

c

emb

OTE. N�855.

hRhtsa3s

mwan

*

*

1729SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

Shriners Pediatric SCI-CAT items, and 6 PROMIS assistivetechnology items.

Demographic data for the calibration study participants areshown in table 3. Approximately 46% of the individuals werediagnosed with paraplegia and 54% with tetraplegia; 46% ofthe sample had complete injuries and 54% had incompleteinjuries. Finally, 31% of the sample was injured less than 1year ago, 28% were injured 1 to 3 years ago, and 41% hadlong-term injuries (�3y). The sample closely matches theproposed stratification by injury type, injury severity, and timesince injury.

The model specifications are outlined in table 1, while theresults of the model assessments for the initial subset of items(ie, k�119) are shown in table 4. All of the assessed modelsad very good fit to the model, as assessed by the CFI, TLI, andMSEA. Model 1, the 1-factor model of physical functioning,ad fit statistics demonstrating acceptable model fit. Model 2,he 2-factor upper extremity and mobility structure, showed alight improvement over the 1-factor model, indicating that itlso had exceptional fit to the data. Models 3 and 4, alternate-factor models, also had good model fit, but they were notuperior to the 2-factor solution.

Models 5 and 8 had a more complicated structure, whereobility was divided into basic mobility, ambulation, andheelchair mobility. These models had inadmissible statistics

bove 1, indicative of model specification errors. This result isot surprising given the nature of SCI, where a substantial

Table 4: Initial CFA for S

Model �2 df

1 26,139.64 69022 23,427.16 67843 23,430.95 67824 23,403.68 67825 14,633.93 68925a* 12,664.35 42715b† 11,038.87 48446 18,776.84 66637 23,412.52 67798 14,613.07 68928a* 12,615.55 42718b† 11,032.13 4844

Model run without ambulation.†Model run without wheelchair.

Table 5: Replication of CFA

Model �2 df

1 24,884.820 49492 22,406.285 48493 22,267.892 48474 22,163.492 48475 17,461.436 49395a* 12,613.592 35635b† 11,612.653 30746 21,874.597 47467 22,123.961 48448 17,316.829 49398a* 12,323.367 35638b† 11,429.536 3074

Model run without ambulation.†Model run without wheelchair.

proportion of the participants either does not ambulate or doesnot use a wheelchair. As a result, the variance in these itembanks is zero, making it difficult to analyze wheelchair mobil-ity and ambulation within the same model. Thus, we developedalternative models that removed either wheelchair or ambula-tion items or the corresponding subdomain factors from themodel (models 5a and 8a [excluding ambulation items] andmodels 5b and 8b [excluding wheelchair mobility items] dem-onstrated improved model fit). Models 8a and 8b had amongthe best fit statistics of all of the models. However, it is unclearif this improvement in fit alone would justify adoption ofmodels 8a/8b specifically, because all models were supportedby the fit statistics. However, models 8a/8b were more repre-sentative of the focus group data and expert input. All of themultifactor models demonstrated slight improvement in modelfit over the 1-factor model.

We replicated the model tests with the second subset ofitems (k�115) (table 5). As before, all models demonstratedvery good fit. Models with both ambulation and wheelchairitems on separate factors did not converge, and the covariancematrices were not positive definite. Models 8a and 8b demon-strated slightly improved model fit. These models did, how-ever, conform to the structure suggested in the focus groupswith distinct factors of self-care, fine motor function, basicmobility, ambulation, and wheelchair mobility.

The last set of CFAs determined whether the basic mobility,wheelchair mobility, ambulation, self-care, and fine motor do-

I—Sample of 119 Items

f CFI TLI RMSEA

0.986 0.986 0.0570.988 0.988 0.0540.988 0.988 0.0540.988 0.988 0.054

Not positive definite0.992 0.992 0.0480.995 0.992 0.0390.991 0.991 0.0460.988 0.988 0.054

Not positive definite0.992 0.992 0.0480.995 0.995 0.039

CI-FI—Sample of 115 Items

f CFI TLI RMSEA

3 0.968 0.968 0.0692 0.972 0.971 0.0659 0.972 0.972 0.0657 0.972 0.972 0.0654 Not positive definite4 0.984 0.984 0.0558 0.985 0.985 0.0571 0.973 0.972 0.0657 0.972 0.972 0.0651 Not positive definite6 0.985 0.984 0.0542 0.986 0.985 0.056

CI-F

�2/d

3.793.453.453.452.122.972.282.823.452.122.952.28

for S

�2/d

5.04.64.54.53.53.53.74.64.53.53.43.7

Arch Phys Med Rehabil Vol 93, October 2012

n

bipu.CRf

idccsqfnmmm

ililbAeltasita

bss

* s are

1730 SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

A

mains were sufficiently unidimensional to be developed ascalibrated item banks. For example, the 54 items that had beenbinned into the basic mobility subdomain were analyzed in aunidimensional analysis. Table 6 summarizes the results ofeach model. Unlike the previous analyses where fit statisticswere compared with one another, the fit statistics in table 6 areot meant for comparative review.After the unidimensional CFAs, we removed several items

ecause of local item dependence (10 items), missing data (7tems), and content concerns (5 items). Each of these itemools demonstrates exceptionally good fit, supporting theirnidimensionality. The CFI and TLI for all banks exceeded929, and for the ambulation and fine motor function banks, theFI and TLI were .999 and .998, respectively. Similarly, theMSEA was less than .08 for all banks except basic mobility,

or which it was .081.

DISCUSSIONThe SCI-FI marks a significant advance in measuring phys-

cal functioning in individuals with SCI. Its content is based onirect feedback from individuals with SCI as well as SCIlinicians, and psychometric analyses24 provide evidence of theontent validity of the SCI-FI. We followed the PROMIStandards for scale development throughout the qualitative anduantitative phases of this study. No other measure of physicalunctioning is as comprehensive and uniquely tailored to theeeds of individuals with SCI as the computer adaptive SCI-FIeasure, with its 5 distinct, unidimensional item banks: basicobility, wheelchair mobility, ambulation, self-care, and fineotor function.We sought to develop patient-reported physical functioning

tem banks that are both targeted to individuals with SCI andinked with new generic physical functioning measures, allow-ng cross-condition comparisons. By using items from theegacy item pools verbatim, scores derived from the SCI-FI cane linked to scores generated from the PROMIS, Neuro-QOL,M-PAC, and Shriner’s Pediatric SCI-CAT measures. An

xample of this methodology for and potential of this type ofinking can be found in Haley et al’s26 linkage of the AM-PACo the Neuro-QOL. However, such an extensive series of link-ges has never been accomplished before, and allowing re-earchers and clinicians the ability to translate scores from onenstrument to another will greatly facilitate comparisons be-ween different medical populations as well as between pedi-tric and adult SCI samples.

The 5-factor model (model 8) tested in this study includedoth ambulation and wheelchair mobility and yielded modelpecification errors (ie, the model was not positive definite),ignaling the presence of inadmissible parameter estimates.41

Inadmissible estimates can result from issues such as samplingvariation, outliers, small samples with only 2 indicator vari-ables per factor, linear dependency of one variable on another,

Table 6: Goodness-of-Fit Indices for Unidim

Subscale No. of Items �2

Ambulation 39 1619.886Basic mobility 54 9090.642Fine motor function 36 1802.012Self-care (women) 85 10,739.826Self-care (men) 84 11,087.981Wheelchair 56 5412.631

The self-care bank includes a total of 90 items; for women, 85 item

or model specification error.33 In this case, model specification

rch Phys Med Rehabil Vol 93, October 2012

error was likely to be the problem, because the correlationbetween ambulation and wheelchair mobility took on an im-possible value (ie, .012). This finding is likely because of themutual exclusivity of responses to walking and wheelchairitems—an artifact of the variety of consequences of SCI, wheremany people who use wheelchairs for mobility are unable towalk at all, while others who are able to walk do not use awheelchair. The results from models 8a and 8b support thisinterpretation. Specifically, when the items related to wheel-chair mobility only are tested (model 8a) or items related toambulation only are tested (model 8b), the fit statisticssupport the inclusion of ambulation and wheelchair mobil-ity, respectively, as distinct from basic mobility. Conceptu-ally, ambulation and wheelchair mobility are critical func-tional abilities for individuals with SCI, and both cliniciansand researchers will have need for separate assessments. Asa result, the 5-factor model is preferable for use in an SCIpopulation.

The CFA examination of the unidimensionality of the 5 finalitem banks (see table 6) demonstrated that all of the itemswithin each bank form a unidimensional hierarchy. This fur-ther supports the 5-factor model and provides support for thevalidity of calibrating the items using IRT and developing asa CAT (see Jette et al24), where any given respondent willonly be asked a fraction of the items included in a given itembank.

The CFA fit statistics support all of the models tested withinthis study, including the 1-factor (model 1) and the 2-factor(model 2) models that more closely resemble the PROMIS andNeuro-QOL structures of physical functioning, respectively.CFA fit statistics also support the other models, and model fitimproves slightly with successive increases in the complexityof the model structure. When mobility is divided into multiplesubdomains (ie, basic mobility, wheelchair mobility, and am-bulation) and self-care is separated into a general activities ofdaily living factor that is distinct from the fine motor abilityfactor (models 8a and 8b), the fit statistics improve over othermodels. This more complicated structure (in models 8a and 8b)more closely reflects the feedback that was received from boththe individuals with SCI and the SCI clinicians who partici-pated in the focus group phase of the study. Together, theseresults suggest that while there may be a place for 1- or 2-factormodels when studying physical function across multiple im-pairment groups, when the clinical or research focus is exclu-sively on people with SCI, more specific, refined aspects ofphysical functioning are important in assessing functional abil-ity and ability limitations.

Study LimitationsThe 5-factor structure was adopted by the research team in

the belief that 5 individual item banks would have more clinicalutility in the SCI population. Such clinical utility data do not

onal CFAs for the SCI-FI Final Item Banks*

df �2/df CFI TLI RMSEA

702 2.308 0.999 0.999 0.0391377 6.602 0.969 0.968 0.081594 3.034 0.998 0.998 0.049

3485 3.082 0.993 0.993 0.0493402 3.259 0.992 0.992 0.0521430 3.785 0.932 0.929 0.063

included, for men, 84 items.

ensi

yet exist, and future research will need to test the responsive-

ppiiscibattd

1731SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

ness of these 5 item banks to meaningful changes in physicalfunctioning, and assess the utility of separate measures inclinical management. Future research should also compare theresults of clinician ratings of physical functioning performancewith PROs of physical functioning to determine the incremen-tal validity of the SCI-FI.

As discussed earlier, the large number of SCI-FI items(k�328) makes it impossible to evaluate the factor structurewith structural equation modeling of all items in a singleanalysis. The procedures that were followed represent an ap-proach to evaluate the factor structure in light of this limitation.Selection of a subset of items with varied content across all ofthe domains and subdomains facilitated examination of thefactor structure. These results were replicated with a differentsubset of items, helping to both ensure the robustness of theCFA results as well as providing assurance that the initialselection of items did not bias the results. The final set ofanalyses, which examined all of the items within their sub-domain to ensure that each item bank was unidimensional,demonstrates that the 5 item banks meet the assumptions forthe IRT analyses and can be calibrated and developed as 5 CATbanks.24

CONCLUSIONSThe SCI-FI represents an ambitious attempt to develop a

atient-reported measure of physical functioning in the SCIopulation. A rigorous item development process occurred thatnvolved literature reviews, integration with previously exist-ng items and scales, expert opinion, and feedback from con-umers. After these procedures, a large multisite field test wasonducted in which 855 individuals with traumatic SCI partic-pated. The results of this study support a 5-factor solutionreaking down physical functioning into key subdomains thatre relevant to individuals with SCI and the clinicians who treathem. These analyses and results have informed the structure ofhe final SCI-FI scale, which has been calibrated using IRT andeveloped as 5 distinct item banks.24

References1. Lammertse DP, Jackson AB, Sipski ML. Research from the model

spinal cord injury systems: findings from the current 5-year grantcycle. Arch Phys Med Rehab 2004;85:1737-9.

2. Ditunno JF. Outcome measures: evolution in clinical trials ofneurological/functional recovery in spinal cord injury. SpinalCord 2010;48:674-84.

3. Tator CH. Review of treatment trials in human spinal cord injury:issues, difficulties, and recommendations. Neurosurgery 2006;59:957-82.

4. Uniform Data System for Medical Rehabilitation (UDS). ClinicalGuide for the Uniform Data Set for Medical Rehabilitation (in-cluding the FIMTM instrument). Buffalo: State University of NewYork at Buffalo; 1997.

5. Keith RA, Granger CV, Hamilton BB, Sherwin FS. The Func-tional Independence Measure: a new tool for rehabilitation. AdvClin Rehabil 1987;1:6-18.

6. Catz A, Itzkovich M, Steinberg F, et al. The Catz-Itzkovich SCIM:a revised version of the Spinal Cord Independence Measure.Disabil Rehabil 2001;23:263-8.

7. Alexander MS, Anderson KD, Biering-Sorensen F, et al. Outcomemeasures in spinal cord injury: recent assessments and recommen-dations for future directions. Spinal Cord 2009;47:582-91.

8. Catz A, Itzkovich M, Tesio L, et al. A multicenter internationalstudy on the Spinal Cord Independence Measure, version III:Rasch psychometric validation. Spinal Cord 2007;45:275-91.

9. Ditunno PL, Dittuno JF. Walking Index for Spinal Cord injury

(WISCI II): scale revision. Spinal Cord 2001;39:654-6.

10. Burns AS, Delparte JJ, Patrick M, Marino RJ, Ditunno JF. Thereproducibility and convergent validity of the Walking Index forSpinal Cord Injury (WISCI) in chronic spinal cord injury. Neu-rorehabil Neural Repair 2011;25:149-57.

11. Mahoney F, Barthel D. Functional evaluation: the Barthel Index.Md Med State J 1965;14:61-5.

12. Granger CV, Albrecht GL, Hamilton BB. Outcome of compre-hensive medical rehabilitation: measurement by PULSES pro-file and the Barthel Index. Arch Phys Med Rehabil 1979;60:145-54.

13. Gresham GE, Labi ML, Dittmar SS, Hicks JT, Joyce SZ, StehlikMA. The Quadriplegia Index of Function (QIF)–sensivitiy andreliability demonstrated in a study of 30 quadriplegic patients.Paraplegia 1986;24:38-44.

14. Itzkovich M, Gelernter I, Biering-Sorensen F, et al. The SpinalCord Independence Measure (SCIM) version III: reliability andvalidity in a multi-center international study. Disabil Rehabil2007;29:1926-33.

15. Slavin MD, Kisala PA, Jette AM, Tulsky DS. Developing acontemporary functional outcome measure for spinal cord injuryresearch. Spinal Cord 2010;48:262-7.

16. U.S. Department of Health and Human Services FDA. Guidancefor industry patient-reported outcome measures: use in medicalproduct development to support labeling claims. Silver Spring:U.S. Department of Health and Human Services, Food and DrugAdministration; 2009.

17. Amtmann D, Cook KF, Johnson KL, Cella D. The PROMISInitiative: involvement of rehabilitation stakeholders in develop-ment and examples of applications in rehabilitation research. ArchPhys Med Rehabil 2011;92(10 Suppl):S12-9.

18. Chang C-H. Patient-reported outcomes measurement and manage-ment with innovative methodologies and technologies. Qual LifeRes 2007;16:157-66.

19. Acquadro C, Berzon R, Dubois D, et al. Incorporating the pa-tient’s perspective into drug development and communication: anad hoc task force report of the patient-reported outcomes (PRO)harmonization group meeting at the Food and Drug Administra-tion, February 16, 2001. Value Health 2003;6:522-31.

20. Cella D, Riley W, Stone A, et al. The Patient-Reported OutcomesMeasurement Information System (PROMIS) developed andtested its first wave of adult self-reported health outcome itembanks: 2005-2008. J Clin Epidemiol 2010;63:1179-94.

21. Cella D, Nowinski C, Peterman A, et al. The Neurology Quality-of-Life Measurement Initiative. Arch Phys Med Rehabil 2011;92(10 Suppl):S28-36.

22. Rose M, Bjorner JB, Becker J, Fries JF, Ware JE. Evaluation ofa preliminary physical function item bank supported the ex-pected advantages of the Patient-Reported Outcomes Measure-ment Information System (PROMIS). J Clin Epidemiol 2008;61:17-33.

23. Gershon RC, Lai JS, Bode R, et al. Neuro-QOL: quality of lifeitem banks for adults with neurological disorders: item develop-ment and calibrations based upon clinical and general populationtesting. Qual Life Res 2012;21:475-86.

24. Jette AM, Tulsky DS, Ni P, et al. Development and initial eval-uation of the Spinal Cord Injury-Functional Index. Arch Phys MedRehabil 2012;93:1733-50.

25. Kisala PA, Tulsky DS. Opportunities for CAT applications inmedical rehabilitation: development of targeted item banks.J Appl Meas 2010;11:315-30.

26. Haley SM, Ni P, Lai JS, et al. Linking the activity measure forpost acute care and the quality of life outcomes in neurologicaldisorders. Arch Phys Med Rehabil 2011;92(10 Suppl):S37-43.

27. Haley SM, Coster WJ, Andres PL, et al. Activity outcomemeasurement for postacute care. Med Care 2004;42(1 Suppl):

I49-61.

Arch Phys Med Rehabil Vol 93, October 2012

1732 SCI-FUNCTIONAL INDEX FACTOR ANALYSIS, Tulsky

A

28. Calhoun CL, Haley SM, Riley A, Vogel LC, McDonald CM,Mulcahey MJ. Development of items designed to evaluate activityperformance and participation in children and adolescents withspinal cord injury. Int J Pediatr 2009;2009:854904.

29. Mulcahey MJ, DiGiovanni N, Calhoun C, Homko E, Riley A,Haley SM. Children’s and parents’ perspectives about activityperformance and participation after spinal cord injury: initialdevelopment of a patient-reported outcome measure. Am J OccupTher 2010;64:605-13.

30. DeWalt DA, Rothrock N, Yount S, Stone AA. Evaluation of itemcandidates: the PROMIS qualitative item review. Med Care 2007;45(5 Suppl 1):S12-21.

31. National Spinal Cord Injury Statistical Center. Spinal cord injury: factsand figures at a glance. Birmingham: University of Alabama; 2011.

32. World Health Organization. International Classification of Functioning,Disability and Health. Geneva: World Health Organization; 2008.

33. Bollen KA. Structural equations with latent variables. New York:Wiley; 1989.

34. Bollen KA, Long JS. Introduction. In: Bollen KA, Long JS, editors.Testing structural equation models. Newbury Park: Sage; 1993. p 1-9.

35. Hu L, Bentler PM. Cutoff criteria for fit indexes in covariancestructure analysis: conventional criteria versus new alternatives.

Struct Equation Model 1999;6:1-55.

rch Phys Med Rehabil Vol 93, October 2012

36. Marsh HW, Balla JR, McDonald RP. Goodness-of-fit indexes inconfirmatory factor analysis: the effect of sample size. PsycholBull 1988;103:391-401.

37. Tucker L, Lewis C. A reliability coefficient for maximum likeli-hood factor analysis. Psychometrika 1973;38:1-10.

38. Bentler PM. Comparative fit indexes in structural models. PsycholBull 1990;107:238-46.

39. Browne MW, Cudeck R. Alternative ways of assessing model fit.In: Bollen KA, Long JS, editors. Testing structural equationmodels. Newbury Park: Sage; 1993. p 136-62.

40. Steinberg L, Thissen D. Use of item response theory and the testletconcept in the measurement of psychopathology. Psychol Meth-ods 1996;1:81-97.

41. Millis S, Malina AC, Bowers DA, Ricker J. Confirmatory factoranalysis of the Wechsler Memory Scale-III. J Clin Exp Neuro-psychol 1999;21:87-93.

Suppliersa. LEXILE framework for reading [computer program]. Durham:

MetaMetrics, Inc, 1000 Park Forty Plz #120, Durham, NC 27713.b. Mplus [computer program]. Version 6. Los Angeles: Muthen &

Muthen, 3463 Stoner Ave, Los Angeles, CA 90066.