Archives of Physical Medicine and Rehabilitation

journal homepage: www.archives-pmr.org

Archives of Physical Medicine and Rehabilitation 2018;-:-------

ORIGINAL RESEARCH

Measuring the Reliability and Construct Validity of

the Pediatric Evaluation of Disability
InventoryeComputer Adaptive Test (PEDI-CAT)
in Children With Cerebral Palsy
Benjamin J. Shore, MD, MPH, FRCSC,a,b Benjamin G. Allar, MD,a Patricia E. Miller, MS,a
Travis H. Matheney, MD, MLA,a,b Brian D. Snyder, MD, PhD,a,b
Maria Fragala-Pinkham, PT, DPTc
From the aDepartment of Orthopedic Surgery, Boston Children’s Hospital, Boston, MA; bHarvard Medical School, Boston, MA; and cFranciscan
Children’s Hospital, Boston, MA.

Abstract
Objective: The purpose of this study was to (1) investigate the construct validity and (2) test-retest reliability of the Pediatric Evaluation of
Disability Inventory-Computer Adaptive Test (PEDI-CAT) in children with cerebral palsy (CP).
Design: A prospective convenience cross-sectional sample.
Setting: Multidisciplinary CP clinic in a tertiary level pediatric children’s hospital.
Participants: English- and Spanish-speaking school-aged children (NZ101) with a diagnosis of CP, stratified by Gross Motor Function
Classification System level, who presented to our multidisciplinary clinic. Participants were excluded if they underwent recent surgery (<6mo) or
botulinum neurotoxin A injection (<3mo). A subset of 17 families participated in retest reliability.
Main Outcome Measures: Convergent and divergent validity were evaluated using Spearman correlation coefficient analysis; test-retest reliability
was assessed using intraclass correlation coefficients (ICCs).
Results: Mean age was 12
3.7 years. Convergent validity was established between Mobility (PEDI-CAT) and Functional Mobility Scale (FMS)
(5 m, rZ0.85; 50 m, rZ0.84; 500 m, rZ0.76; P<.001). In ambulant children, convergent validity was established between Daily Activities
(PEDI-CAT vs Pediatric Quality of Life CP [PedsQL-CP] [rZ0.85, P<.001]) and between Social/Cognitive (PEDI-CAT) and Speech and
Communication (PedsQL-CP) (rZ0.42, P<.001). In nonambulant children, convergent validity was established between Daily Activities (PEDI-
CAT) and Personal Care (Caregiver Priorities and Child Health Index of Life with Disabilities [CPCHILD]) (rZ0.44, P<.001) and between
social/cognitive (PEDI-CAT) and Communication (CPCHILD) (rZ0.64, P<.001). A lack of correlation between Daily Activities, Social/
Cognitive, and Responsibility (PEDI-CAT) and FMS and between the Mobility (PEDI-CAT) and Communication (PedsQL) domains confirmed
divergent validity. Test-retest reliability was excellent for all domains of the PEDI-CAT (ICCZ0.96-0.99).
Conclusions: The PEDI-CAT is an outcome measure that demonstrates strong construct validity and reliability in children with CP.
Archives of Physical Medicine and Rehabilitation 2018;-:-------
ª 2018 by the American Congress of Rehabilitation Medicine

Over the last 20 years several outcome measures have been

Presented to the Pediatric Orthopedic Society of North America, April 29, 2015, Atlanta, GA;
the American Academy of Cerebral Palsy and Developmental Medicine, October 22, 2015, San
Diego, CA.
Supported by the Pediatric Orthopaedic Society of North AmericadYoung Investigator Grant
2014.
Disclosures: Maria Fragala-Pinkham is one of the senior authors of the PEDI-CAT and receives
less than $100 per year in royalties. The other authors have nothing to disclose.
developed for children with cerebral palsy (CP); however, each
measure has demonstrated variable performance according to the
International Classification of Functioning, Disability and Health
and no universal gold standard has been identified. The Pediatric
Evaluation of Disability Inventory (PEDI) is a common parent-
report outcome measures used to report functional skills in

0003-9993/18/$36 - see front matter ª 2018 by the American Congress of Rehabilitation Medicine
https://doi.org/10.1016/j.apmr.2018.07.427
2 B.J. Shore et al
children and youth with CP.1-4 However, limitations with this
instrument exist: the length of application (>200 items) makes it
challenging to administer in busy clinical settings; in older chil-
dren with higher levels of functional ability, ceiling affects have
been recorded and normative scores exist only for children 6
months to 7 years of age.5 The PEDI-Computer Adaptive Test
(CAT) represents a novel clinical assessment for children and
youth from birth to 20 years of age which uses CAT and was
developed on the basis of years of experience, feedback, and
formal research with the original PEDI.5-8 CATs are built from a
set of coordinated items (item banks) that define a common
dimension.9 The CAT uses a simple form of artificial intelligence,
which selects questions that are directly tailored to an individual,
thus shortening the test to achieve desired precision. CAT testing
is currently being used in pediatric and adult health services,10,11
demonstrating the potential to be the assessment platform of
choice for future heath care appraisal programs.
The PEDI-CAT was developed at the Boston University Health
and Disability Research Institute and released in October 2012.
Preliminary psychometrics studies on the PEDI-CAT have
demonstrated good concurrent validity with the Alberta Infant
Motor Scale for young children admitted to postacute care12 and
with the PEDI Functional Mobility Scale (FMS) for children with
neurodevelopmental disabilities.13 Excellent test-retest reliability
(intraclass correlation coefficient [ICC]Z0.96-0.99) has been
reported for all 4 domains of the PEDI-CAT for a small sample of
children with and without disabilities (nZ23).14 With the devel-
opment of any new assessment tool, validity must be determined.
Recently, the discriminant validity of the PEDI-CAT was tested in
the same cohort of children with CP, demonstrating excellent
discriminant validity between ambulant and nonambulant partic-
ipants across Gross Motor Functional Classification System
(GMFCS) and Manual Ability Classification System levels.15
However, the construct validity of the PEDI-CAT in children
with CP has yet to be studied.
The Pediatric Quality of Life CP (PedsQL-CP)16 and the
Caregiver Priorities and Child Health Index of Life with Dis-
abilities (CPCHILD) questionnaire17 were both developed to
measure health-related quality of life or caregiver burden specif-
ically for children with CP. Both these measures are completed by
parent report, and they have similar constructs related to self-care,
mobility, communication, and social function as the PEDI-CAT.
The PedsQL-CP has been shown to discriminate across most
GMFCS levels for Daily Activities and Movement and Balance;
however, it is less discriminative across GMFCS levels III-V. Face
and content validity has been reported for the CP-CHILD which
was designed specifically to measure health and caregiver burden
for children with severe CP (GMFCS IV and V).17 The FMS
List of abbreviations:
CP cerebral palsy
CPCHILD Caregiver Priorities and Child Health Index
of Life with Disabilities
FMS Functional Mobility Scale
GMFCS Gross Motor Functional Classification System
ICC intraclass correlation coefficient
ICF International Classification of Functioning,
Disability and Health
PEDI-CAT Pediatric Evaluation of Disability Inventory-
Computer Adaptive Test
PedsQL-CP Pediatric Quality of Life cerebral palsy
describes functional mobility over 3 distances representing
mobility in home, school, and community settings. Construct
content and concurrent validity of the FMS have been demon-
strated in children with CP.18 Because these 3 measures were
developed specifically for children with CP, and they are parent
report measures, they were used to assess validity of the
PEDI-CAT.
The purpose of this study was to report the construct validity
(including convergent and divergent validity) of the PEDI-CAT
against the FMS, PedsQL-CP, and CPCHILD and test-retest reli-
ability in children and youth with CP.
Methods
This prospectively gathered cross-sectional cohort included a group
of 101 children and their parents or caregivers who presented to a
multidisciplinary CP clinic at a tertiary level pediatric children’s
hospital between June 2013 and May 2014. English- and Spanish-
speaking parents and children (between 6 and 20y) with a confirmed
diagnosis of CP (GMFCS I-V) and with a consistent primary
caregiver present during the clinic visit were recruited. Children
were excluded if their neuromuscular diagnosis was unclear or if
recent surgery (<6mo) or botulinum neurotoxin A administration
(<3mo) was performed. Informed consent was obtained prior to
study participation, and data collection and Institutional Review
Board approval occurred prior to commencement of this study.
During the study period, a total of 525 patients were identified as
potential study candidates; however, on the day of the clinic, pa-
tients were often being seen concurrently, and thus 111 of these
eligible patients were approached for participation, of which 101
consented to the study (fig 1).
Caregivers of children with CP completed the PEDI-CAT
(version 1.3.9) using iPad technology with the aid of a trained
research assistant, while our inclusion criteria included Spanish-
speaking families, in every consented case, the caregiver who
completed the assessments spoke and read English. In addition,
for ambulant children (GMFCS I-III), the caregiver completed the
PedsQL-CP and for nonambulant children (GMFCS IV and V), the
caregiver completed the CPCHILD questionnaire. Demographic
information was recorded for each participant including race,
language, sex, ethnicity, level of education (child and caregiver),
socioeconomic status, and type of caregiver completing the form.
The treating physician assigned a GMFCS, Manual Ability
Classification System, and FMS score for each child with input
from the caregiver about the child’s typical functioning. Test-retest
reliability was measured in a convenience sample of 17 partici-
pants between 10 and 44 days from their initial clinical visit.
The PEDI-CAT uses a computer adaptive platform with an
item bank of 276 items in the domains of (1) Daily Activities; (2)
Mobility; (3) Social/Cognitive; and (4) Responsibility.7,8 The 4
PEDI-CAT domains are self-contained and can be used separately
or in combination with the other domains. The PEDI-CAT is
completed by parent or caregiver report or professional judgment
of clinicians who are familiar with the child’s typical
performance.
There are 2 versions of the PEDI-CAT: Speedy and Content-
balanced. The Speedy PEDI-CAT was used for this study. It is the
more efficient version because it provides precise score estimates
while administering only 5-15 items per domain. The score report
for the Speedy CAT includes a percentile score and a T score that
can be used to describe a child’s performance compared to other
www.archives-pmr.org
Construct validity of PEDI-CAT in CP 3

Correlation Threshold

<0.30
<0.30

<0.30
<0.30

<0.30
<0.30

<0.30
<0.30
Divergent Validity Comparison Construct

CPCHILD Communication
PedsQL Communication

FMS 5, 50, 500m

FMS 5, 50, 500m

FMS 5, 50, 500m

FMS 5, 50, 500m

FMS 5, 50, 500m

FMS 5, 50, 500m
Fig 1 Strobe diagram of study population.

children of the same age. The score report also contains a scaled
score which is not age related. Scaled scores provide information
on a child’s status along a continuum of easy to difficult levels of Correlation Threshold
function. The PEDI-CAT scaled scores are currently on a 20-80
metric, which allows for further development of the measure.
A higher score translates to higher function functional ability.
To prevent floor and ceiling effects, further items can be
0.60
0.50
0.50

0.60
0.40

0.40
0.40

0.40
0.40
validated and added with future research to aid in the respon-
siveness of the measure. The total duration of this test is on
average 12 minutes.14
The PEDI-CAT Mobility domain items range from basic
motor skills of turning head while in supine and sitting without
CPCHILD Activities of Daily Living/Personal Care

CPCHILD Activities of Daily Living/Personal Care

PedsQL School Activities and/or Communication
support to more difficult motor skills of running, walking 3 CPCHILD Communication and Social Interaction
Convergent and divergent hypotheses for each PEDI-CAT domain

CPCHILD Positioning, Transferring and Mobility

PedsQL Daily, School, and/or Eating Activities

miles, or climbing a step ladder to put a heavy box on a shelf.19
Convergent Validity Comparison Construct

The use of walking devices (crutches, canes, or walker) is also

included in this domain. The PEDI-CAT Daily Activities domain
includes self-care activities related to keeping clean, mealtime
activities, and dressing; specific test items include fastening a
PedsQL Movement and Balance

watch band, using a knife to butter bread, and wiping self with
toilet paper after a bowel movement. The PEDI-CAT Social/
PedsQL Daily Activities

Cognitive domain includes skills needed for effective social

exchange and to function safely. Social/Cognitive domain in-
FMS 5, 50, 500 m

cludes 60 items related to 4 content areas of interaction (eg,

follows the gaze of another person to look at the same place or
object), communication (eg, uses words, gestures or signs to ask
for something), everyday cognition (eg, recognizes his/her
printed name), and self-management (eg, when upset responds
without hitting, punching, or biting). The Responsibility domain
provides information at the participation level according to the
International Classification of Functioning, Disability and Health
(ICF) levels by examining a child’s engagement in typical
All participants
PEDI-CAT Domain

Social/cognitive

functional activities or life tasks. Responsibility items comprise a

Nonambulant

Nonambulant

Nonambulant

Nonambulant
Daily Activities

Responsibility

combination of functional skills needed to carry out life skills.

Ambulant

Ambulant

Ambulant

Ambulant

The FMS gives a rating of assistance required by children

Mobility
Table 1

with CP for mobility in 3 environmental settings using distances

in meters to differentiate settings of home, 5 m; school, 50 m;
and community, 500 m. We hypothesize that the PEDI-CAT

www.archives-pmr.org
4 B.J. Shore et al

Mobility domain will exhibit convergence with the FMS 5, 50,

Table 2 Cohort description (NZ101)
and 500m.
The Pediatric Quality of Life-CP version (PedsQL-CP) is a Freq. (%)
valid and reliable modular approach to measuring health-related Patient characteristics
quality of life in children with CP. The PedsQL-CP is made up of Sex (% men) 54 (53)
7 domains including Movement and Balance, Daily Activities, Mean age
SD (y) 11.9
3.70
School Activities, Communication, Pain and Hurt, Fatigue, and GMFCS
Eating Activities. We hypothesize that for ambulant CP partici- I 23 (23)
pants, the PEDI-CAT Mobility domain will exhibit convergence II 19 (19)
with the PedsQL-CP Movement and Balance domain; the PEDI- III 17 (17)
CAT Daily Activities domain will exhibit convergence with the IV 21 (21)
PedsQL-CP Daily Activities domain; the PEDI-CAT Social/ V 21 (21)
Cognitive domain will exhibit convergence with at least one of the Ethnicity
PedsQL-CP School Activities and Communication domains; and Hispanic 16 (16)
the PEDI-CAT Responsibility domain will exhibit convergence Not Hispanic 85 (84)
with at least one of the PedsQL-CP Daily Activities, School Ac- Race
tivities, and Eating Activities domains. White 77 (76)
The Caregiver Priorities and Child Health Index of Life with Black 15 (15)
Disabilities (CPCHILD) questionnaire measures caregivers’ per- Asian 4 (4)
spectives on the health status, comfort, well-being, and ease of American Indian or Alaskan Native 1 (1)
care giving of children with severe developmental disabilities, Other 4 (4)
including nonambulant children with CP.17 The CPCHILD con- Caregiver characteristics
sists of 37 items distributed across the following 6 domains: Ac- Caregiver type
tivities of Daily Living/Personal Care (9 items); Positioning, Biological parent 87 (87)
Transferring and Mobility; Comfort and Emotion; Communication Nonbiological parent 9 (8)
and Social Interaction; Health; and Overall Quality of Life, and a Guardian 4 (4)
total score over all domains. We hypothesize that for nonambulant Caregiver highest education
CP participants, the PEDI-CAT Daily Activities and Re- Some high school or less 7 (7)
sponsibility domains will exhibit convergence with the CPCHILD High school diploma or GED 17 (17)
Activities of Daily Living/Personal Care domain; the PEDI-CAT Some college, vocational school, or AD 20 (20)
Mobility domain will exhibit convergence with the CPCHILD College or university degree 32 (32)
Positioning, Transferring and Mobility domain; and the PEDI- Postgraduate degree 24 (24)
CAT Social/Cognitive domain will exhibit convergence with the Household income
CPCHILD Communication and Social Interaction domains. <$10,000 4 (4)
$10,000-$39,000 17 (18)
Statistical analysis $40,000-$69,000 21 (22)
$70,000-$99,000 17 (18)
Convergent validity was defined as the extent to which 2 measures >$100,000 38 (39)
relate to the same construct, whereas divergent validity was Abbreviations: AD, associate’s degree; GED, general equivalency
defined as the extent to which 2 measures that should not be diploma.
related were not related. Both convergent and divergent validity
were assessed using Spearman correlation coefficient analysis.
Varying degrees of correlation were expected based on the type of
validity tested and the particular domains being compared Results
(table 1).20 Test-retest reliability for each of the 4 domains of the A total of 101 children were included in our analysis (54 boys, 47
PEDI-CAT was assessed on participants returning to clinic (10- girls) with an average age of 11.9
3.7 years and a near-equal
44d after their first completion) using ICCs along with 95% distribution according to GMFCS level (table 2). Mean scores and
confidence intervals as well as a paired t test for the mean dif- standard deviations for all of the outcomes are presented in
ference in response. table 3. In our cohort, all caregivers who completed the assess-
Power analysis determined that we would require a minimum ment spoke English and we did not administer the Spanish ver-
of 13 participants to power a 1-sided test for a test-retest reliability sions of the form.
coefficient (ICC) of at least 0.8 compared with an ICC of 0.95 to
achieve 80% power with alpha set to 0.05. In addition, we would
require a minimum of 41 participants for a 1-sided test for a Convergent validity
correlation coefficient of at least of 0.40 to achieve 80% power
with significance set to 5%. Similar analyses for correlation hy- PEDI-CAT Mobility
potheses larger than 0.4 would require fewer participants. There- The Mobility (PEDI-CAT) domain demonstrated convergent val-
fore, the number of patients included in the entire cohort (NZ101) idity with high correlations for the FMS 5 m, FMS 50 m, and FMS
and for the ambulant (nZ59) and nonambulant (nZ42) subgroups 500 m (rZ0.85, 0.84, and 0.76, respectively; P<.001) across
provided sufficient power for our primary and secondary hy- all participants (table 4). In addition, convergence was
pothesis testing. demonstrated with the Movement and Balance (PedsQL) domain

www.archives-pmr.org
Construct validity of PEDI-CAT in CP 5

(rZ0.45, P<.001) in ambulant children and the Positioning,

Table 3 Outcome measurements by domain for all participants
Transfer and Mobility (CPCHILD) domain (rZ0.61, P<.001) in
nonambulant children. Outcome Mean
SD Range
PEDI-CAT (NZ101)*
PEDI-CAT Daily Activities Daily Activities 50.2
9.78 29.7-70.2
The Daily Activities (PEDI-CAT) domain exhibited convergence Mobility 55.4
11.46 26.8-76.3
with the Daily Activities (PedsQL-CP) domain (rZ0.85, P<.001) Social/Cognitive 63.4
9.73 28-76.7
in ambulant children and the Activities of Daily Living/Personal Responsibility 44.3
11.47 24.5-65.2
Care (CPCHILD) domain (rZ0.44, P<.001) in non- FMS (median [IQR]) (nZ95)
ambulant children. 5m 3 (1-6) 1-6
50m 2 (1-5) 1-6
PEDI-CAT Social/Cognitive 500m 1 (1-5) 1-6
The Social/Cognitive (PEDI-CAT) domain exhibited convergence PedsQL (ambulant only; nZ58)
with the School Activities (PedsQL-CP) and Communication Movement and Balance 65.3
23.06 12.5-100
(PedsQL-CP) domains (rZ0.51 and 0.42, respectively; P<.001) in Daily Activities 65.8
28.51 0-100
ambulant children and the Communication and Social Interaction School Activities 67.7
27.80 0-100
(CPCHILD) domain (rZ0.64, P<.001) in nonambulant children Speech and Communication 81.3
27.00 0-100
(see table 4). Pain and Hurt 73.0
20.89 12.5-100
Fatigue 63.7
24.26 18.8-100
PEDI-CAT Responsibility Eating Activities 77.3
21.89 5-100
The Responsibility (PEDI-CAT) domain exhibited convergence CPCHILD (nonambulant; nZ41)
with the Daily Activities (PedsQL-CP), School Activities Total score 57.1
15.94 23.2-93.7
(PedsQL-CP), and Eating Activities (PedsQL-CP) domains Positioning 40.7
18.54 1.4-83.3
(rZ0.66, 0.47, and 0.49, respectively; P<.001) in ambulant chil- Personal care 42.0
17.89 2.5-95.1
dren; however, correlation with the Activities of Daily Living/ Communication 59.9
25.92 0-100
Personal Care (CPCHILD) domain was negligible (PZ.07) in Comfort 83.0
14.36 36.7-100
nonambulant children (see table 4). Health 61.7
20.04 20-93.3
Quality of life 70.7
17.94 40-100
Divergent validity Abbreviation: IQR, interquartile range.
* The number in parentheses represents the number of participants
Divergent validity was illustrated by the lack of correlation be- with data available for that measure.
tween the Daily Activities, Social/Cognitive, or the Responsibility
(PEDI-CAT) domains and FMS scores (see table 4). In addition,
the Mobility (PEDI-CAT) domain exhibited no correlation with serially monitor the effect of interventions over a range of
PedsQL constructs not directly related to mobility such as cognitive and physical function disability levels.21 We report the
Communication (PedsQL) (see table 4). There was, however, a construct validity and the test-retest reliability of the PEDI-CAT in
larger than expected correlation between the Mobility (PEDI- a population of children with CP to support this premise.
CAT) domain and the Communication (CPHILD) domain The PEDI has been used worldwide as a comprehensive
(rZ0.38) (see table 4). functional assessment instrument designed to quantify function
and measure change after treatment interventions in children with
physical disability. McCarthy et al1 found that in children with CP,
Test-retest reliability the PEDI demonstrated better data quality, reliability, and validity
About 17 families returned to repeat the PEDI-CAT at an average compared to other commonly used measures (Pediatric Outcomes
of 25 days (range, 10-44d) after the first questionnaire. Test-retest Data Collection Instrument and Child Health Questionnaire). In
reliability was high for all 4 domains with intraclass correlations addition, Wren et al3 found that the PEDI may be the preferred
ranging between 0.96 (Daily Activities) to 0.99 (Social/Cognitive) outcome tool if specific functional activities are desired, when
(table 5). In addition, all paired mean differences between first and compared to the Pediatric Outcomes Data Collection Instrument
second tests were negligible (all P>.10). and Child Health Questionnaire. Despite these strengths of the
PEDI, with more than 200 items to administer, it is time
consuming and impractical in a busy clinical setting.
Discussion Children with CP demonstrate a broad spectrum of function
and disability. In this study, we used the GMFCS to dichotomize
Using valid and reliable outcome measures to track functional our patient population into ambulant (GMFCS I-III) and non-
abilities at the ICF levels of activity and participation for children ambulant (GMFCS IV and V) groups to help investigate the
and youth with CP is critical for measuring the effects of medical, convergent validity of the PEDI-CAT. Ambulant GMFCS I and II
surgical, and rehabilitation interventions. The ideal goal of children undergo orthopedic surgery for gait correction, whereas
outcomes-based research in CP is to develop a core set of out- nonambulant GMFCS IV and V children undergo major ortho-
comes for each of the ICF levels. The PEDI-CAT demonstrates pedic and medical interventions to improve their comfort or
potential for being included in this battery at the Activity and quality of life. The accurate measurement of function and
Participation level because it is a comprehensive performance- disability using 1 outcome instrument across a population of both
based functional outcome measure that is quick to administer, ambulant and nonambulant children with CP is challenging due to
precise, and user friendly with the potential to enable clinicians to significant floor and ceiling effects.17

www.archives-pmr.org
6 B.J. Shore et al

Table 4 Correlation coefficients between PEDI-CAT domains and other outcome instrument domains
Outcome Measure Mobility Daily Activities Social/Cognitive Responsibility
FMS (nZ95)
5m 0.85* 0.18 0.03 0.05
50 m 0.84* 0.20 0.02 0.02
500 m 0.76* 0.20 0.13 0.17
PedsQL (ambulant only, nZ58)
Movement and Balance 0.45* 0.33y 0.23 0.25
Daily Activities 0.61* 0.85* 0.70* 0.66*
School Activities 0.38* 0.62* 0.51* 0.47*
Communication 0.24 0.49* 0.42* 0.24
Pain and Hurt 0.14 0.01 0.15 0.08
Fatigue 0.43* 0.28y 0.19 0.19
Eating Activities 0.41* 0.76* 0.57* 0.49*
CPCHILD (nonambulant only, nZ41)
Total score 0.53* 0.61* 0.56* 0.40y
Positioning 0.61* 0.58* 0.46* 0.36y
Personal care 0.43* 0.44* 0.38y 0.29
Communication 0.38y 0.54* 0.64* 0.37y
Comfort 0.34y 0.40y 0.31 0.16
Health 0.40* 0.60* 0.60* 0.53*
Quality of life 0.13 0.12 0.10 0.07
* Significant correlation at the 1% level.
y
Significant correlation at the 5% level.

In the current study, we found that the PEDI-CAT performed communication; however, after review of the literature, we found
equally well in comparison to legacy measures: PedsQL (ambu- evidence to confirm the relation between mobility and commu-
lant children) and CPCHILD (nonambulant children). For ambu- nication in children with CP.22
lant children with CP, we found the PEDI-CAT Mobility, Daily The PEDI-CAT Responsibility domain did not meet the
Activities, and Social/Cognitive domains demonstrated conver- threshold for convergence with CPCHILD personal care (r<0.40)
gence across the 3 FMS settings and PedsQL-CP Daily Activities as we originally hypothesized. We suspect these results are a
and Speech/Communication domains. For nonambulant children function of variable measurement: with the PEDI-CAT measuring
with CP, we found that the PEDI-CAT Mobility, Daily Activities, child functional performance and the CPCHILD measuring care-
and Social/Cognitive domains were convergent with the giver burden and ease of care provision.
CPCHILD Positioning, Transferring and Mobility; Personal Care;
and Communication and Social Interaction domains. Initially we
did not anticipate a significant correlation between mobility and Study limitations
Test-retest reliability was high for all 4 domains in our 17 patient
test-retest reliability convenience sample. Although this sample
Table 5 Test-retest reliability of PEDI-CAT domains may appear small, we had appropriate power to demonstrate retest
reliability and we believe that these results are accurate. Previous
Domain ICC 95% CI work on the PEDI-CAT has demonstrated high test-retest reli-
Mobility 0.98 (0.96-0.99) ability when used to measure function in children with and
Daily Activities 0.96 (0.90-0.99) without developmental disabilities14 and in patients from other
Social/Cognitive 0.99 (0.97-1.00) countries (Brazil).23 Future research should be directed toward
Responsibility 0.98 (0.95-0.99) evaluating a larger sample of children to demonstrate test-retest
Test-retest Cohort Summary (nZ17) reliability of the PEDI-CAT at each of the GMFCS and MAC
levels to determine if it is consistent across all levels.
Characteristic Freq. (%) Only school-aged children were studied so we cannot gener-
Mean age
SD (y) 13.2
4.72 alize these findings to children with CP younger than 6 years.
Sex (% men) 10 (59) There may have been unintentional selection bias with caregivers
GMFCS level who chose to participate in this study. Most caregivers in this
I 4 (24) study were white, middle-income parents so these findings may
II 1 (6) not be representative of caregivers from other cultural and/or
III 2 (12) socioeconomic backgrounds. For the purpose of this study, we
IV 4 (24) excluded those children who had undergone recent surgical pro-
V 6 (35) cedures or botulinum neurotoxin A injections, thus keeping the
Abbreviation: 95% CI, 95% confidence interval.
study population relatively homogeneous. One of the main
features of any outcome tool is to demonstrate responsiveness to a

www.archives-pmr.org
Construct validity of PEDI-CAT in CP
treatment intervention. With our small sample size, we felt that we
would not be able to measure the convergent validity and
responsiveness concurrently and therefore excluded the patients
above. Our next step in testing or validating the PEDI-CAT in
children with CP will be to prospectively examine the respon-
siveness of the PEDI-CAT after single event orthopedic
surgerydto identify the minimal detectable change and minimal
clinically important difference. Preliminary evidence suggests that
the PEDI-CAT should demonstrate sensitivity to change that is at
least as good as the original PEDI.24
Conclusions
This study demonstrates the construct validity and test-retest
reliability of the PEDI-CAT across a sample of children with CP.
This study demonstrates that the PEDI-CAT is a valid outcome
instrument across a population of children with CP with diverse
functional capabilities.
Keywords
Cerebral palsy; Rehabilitation; Validation studies
Corresponding author
Benjamin J. Shore, MD, MPH, FRCSC, Department of
Orthopedic Surgery, Boston Children’s Hospital, 300 Longwood
Avenue, Boston, MA 02115. E-mail address: Benjamin.shore@
childrens.harvard.edu.
References
1. McCarthy ML, Silberstein CE, Atkins EA, Harryman SE,
Sponseller PD, Hadley-Miller NA. Comparing reliability and validity
of pediatric instruments for measuring health and well-being of chil-
dren with spastic cerebral palsy. Dev Med Child Neurol 2002;44:468-
76.
2. Harvey AR, Morris ME, Graham HK, Wolfe R, Baker R. Reliability of
the functional mobility scale for children with cerebral palsy. Phys
Occup Ther Pediatr 2010;30:139-49.
3. Wren TA, Sheng M, Hara R, et al. Agreement among three in-
struments for measuring functional health status and quality of life in
pediatric orthopaedics. J Pediatr Orthop 2007;27:233-40.
4. Oeffinger D, Gorton G, Bagley A, et al. Outcome assessments in
children with cerebral palsy, part I: descriptive characteristics of
GMFCS Levels I to III. Dev Med Child Neurol 2007;49:172-80.
5. Haley SM, Coster WI, Kao YC, et al. Lessons from use of the Pedi-
atric Evaluation of Disability Inventory: where do we go from here?
Pediatr Phys Ther 2010;22:69-75.
6. World Health Organization. International Classification of Func-
tioning, Disability and Health(ICF). Geneva, Switzerland: World
Health Organization; 2001.
www.archives-pmr.org
7
7. Dumas HM, Fragala-Pinkham MA, Feng T, Haley SM. A preliminary
evaluation of the PEDI-CAT Mobility item bank for children using
walking aids and wheelchairs. J Pediatr Rehabil Med 2012;5:29-35.
8. Dumas H, Fragala-Pinkham M, Haley S, et al. Item bank development
for a revised pediatric evaluation of disability inventory (PEDI). Phys
Occup Ther Pediatr 2010;30:168-84.
9. Haley SM, Chafetz RS, Tian F, et al. Validity and reliability of
physical functioning computer-adaptive tests for children with cere-
bral palsy. J Pediatr Orthop 2010;30:71-5.
10. Jette AM, Haley SM. Contemporary measurement techniques for
rehabilitation outcomes assessment. J Rehabil Med 2005;37:339-45.
11. Cella D, Gershon R, Lai JS, Choi S. The future of outcomes mea-
surement: item banking, tailored short-forms, and computerized
adaptive assessment. Qual Life Res 2007;16(Suppl 1):133-41.
12. Dumas HM, Fragala-Pinkham MA, Rosen EL, Lombard KA,
Farrell C. Pediatric Evaluation of Disability Inventory Computer
Adaptive Test (PEDI-CAT) and Alberta Infant Motor Scale (AIMS):
validity and responsiveness. Phys Ther 2015;95:1559-68.
13. Dumas HM, Fragala-Pinkham MA. Concurrent validity and reliability
of the Pediatric Evaluation of Disability Inventory-Computer Adaptive
Test Mobility domain. Pediatr Phys Ther 2012;24:171-6.
14. Dumas HM, Fragala-Pinkham MA, Haley SM, et al. Computer
adaptive test performance in children with and without disabilities:
prospective field study of the PEDI-CAT. Disabil Rehabil 2012;34:
393-401.
15. Shore BJ, Allar BG, Miller P, et al. Evaluating the discriminant val-
idity of the Pediatric Evaluation of Disability Inventory: computer
adaptive test (PEDI-CAT) in children with cerebral palsy. Phys Ther
2017;96:669-76.
16. Sullivan E, Barnes D, Linton JL, et al. Relationships among functional
outcome measures used for assessing children with ambulatory CP.
Dev Med Child Neurol 2007;49:338-44.
17. Narayanan UG, Fehlings D, Weir S, Knights S, Kiran S, Campbell K.
Initial development and validation of the Caregiver Priorities and
Child Health Index of Life with Disabilities (CPCHILD). Dev Med
Child Neurol 2006;48:804-12.
18. Graham HK, Harvey A, Rodda J, Nattrass GR, Pirpiris M. The
Functional Mobility Scale (FMS). J Pediatr Orthop 2004;24:514-20.
19. Kirshner B, Guyatt G. A methodological framework for assessing
health indices. J Chronic Dis 1985;38:27-36.
20. Abma IL, Rovers M, van der Wees PJ. Appraising convergent validity
of patient-reported outcome measures in systematic reviews: con-
structing hypotheses and interpreting outcomes. BMC Res Notes
2016;9:226.
21. Msall ME. Tools for measuring daily activities in children: promoting
independence and developing a language for child disability. Pediat-
rics 2002;109:317-9.
22. Hidecker M, Ho N, Dodge N, et al. Inter-relationships of functional
status in cerebral palsy: analyzing gross motor function, manual
ability, and communication function classification systems in children.
Dev Med Child Neurol 2012;54:737-42.
23. Mancini MC, Coster WJ, Amaral MF, Avelar BS, Freitas R, Sampaio RF.
New version of the Pediatric Evaluation of Disability Inventory (PEDI-
CAT): translation, cultural adaptation to Brazil and analyses of psy-
chometric properties. Braz J Phys Ther 2016;20:561-70.
24. Iyer LV, Haley SM, Watkins MP, Dumas HM. Establishing minimal
clinically important differences for scores on the pediatric evaluation
of disability inventory for inpatient rehabilitation. Phys Ther 2003;83:
888-98.