Escolar Documentos
Profissional Documentos
Cultura Documentos
2
and RMSEA are absolute ft indices, whereas the NFI, TLI,
and CFI are incremental ft measures. Te Akaike information
criterion (AIC) is a comparative measure of ft. RMSEA values
<0.05, 0.060.08, 0.080.10, and >0.1 indicate good, reason-
able, mediocre, and poor ft, respectively. NFI and TLI > 0.90
also indicate a good ft. Te CFI is independent of sample
size and quantifes the amount of variation and covariation
accounted for by the hypothesized model by comparing its
ft to that of an independent model of uncorrelated variables.
Te upper limit of the CFI is 1.00, and a value exceeding
0.90 indicates a good ft. Lower AIC values indicate a better
ft. Goodness-of-ft indices were used to determine the ade-
quacy of the models [23]. Te Analysis of Moment Structure
(AMOS) 20.0 statistical program was employed to perform
the CFA for obtaining maximum-likelihood estimates of
model parameters and provide goodness-of-ft indices.
EFA is uses to identify the number of factors and the pat-
tern of factor loadings. Principle component analysis was
applied for factor extraction method. Rotation was employed
to improve the meaningfulness and interpretation of the
generated factors. Tere are two rotation methods, orthogo-
nal and oblique. Because the factors were expected to be
correlated, an oblique rotation was used [24]. Factors were
selected if they had eigenvalues of 1 with loadings on that
factor exceeding 0.30 [25]. Te correlations between factors
and internal consistency reliability for each factor were also
calculated. EFA was performed using SPSS 20.0.
First, CFA was performed for calculating model ft indices
for each 1- and 2-factor model. And then EFA was performed
for identifying the number of factors and factor loadings.
Lastly, CFA was used again for examining the construct of
WeeFIM based EFA.
Results
Internal consistency of the WeeFIM
As shown in Table II, the overall scale reliability of the
WeeFIM was good. Cronbachs coefcient was 0.98 with a
95% confdence interval (range, 0.970.98). Table II presents
Cronbachs coefcients according to the three factors of the
WeeFIM.
CFA of the WeeFIM based on original 1- and 2-factor
models
Table III presents the ft indices for each 1- and 2-factor
model. A 1-factor model posits the entire 18-item load on
1-factor. In the 2-factor model, items are hypothesized to load
on 2 factors: motor and cognitive factors. Te ft indices did
not support for any of the models for the WeeFIM among
children with spastic CP.
EFA of the WeeFIM
Table IV shows the factor structures based on eigenvalues
of 1. Bartletts test for a signifcant correlation matrix in this
sample produced a value of p < 0.01. Te KaiserMyerOlkin
test measuring the sample adequacy was 0.93, which is excel-
lent. Te analysis revealed three factors (self-care, motor, and
cognitive), explaining 87.12% of the variance.
Table II. Reliability according to the three factors of the WeeFIM.
Factor Cronbachs 95% CI
Self-care 0.97 0.960.97
Motor 0.96 0.950.97
Cognitive 0.97 0.970.98
ADLs 0.98 0.970.98
ADL, activities of daily living; CI, confdence interval.
Table III. Comparison of the two models of the WeeFIM.
Factor df
2
NFI TLI CFI RMSEA AIC
1-factor 126 936.46 0.85 0.84 0.87 0.18 1062.46
2-factor 125 770.99 0.88 0.87 0.89 0.16 898.98
AIC, Akaike Information Criterion; CFI, Comparative Fit Index; NFI, Normed Fit
Index; RMSEA, Root mean square error of approximation; TLI, Tucker-Lewis Index.
Construct of WeeFIM in children with spastic CP1 1469
2013 Informa UK, Ltd.
In the 3-factor model of the EFA, items loaded on the
frst factor were nested with the original cognitive factor of
the item. Only the eating item of the original motor factor
was loaded on the cognitive factor. Te items comprising the
original motor factor were subdivided into two factors. Toilet,
bed/chair/wheelchair, tub/shower, walk/wheelchair, stair,
bowel, and bladder were loaded on the second factor. Dressing
upper, dressing lower, bathing, toileting, and grooming were
loaded on the third factor. In the EFA, factors are determined
on the basis of factor loading and interpretability. Te eating
items have high factor loading on the frst and third factors.
First and third factor loading of eating item were 0.80 and
0.79, respectively. Considering factor loading and interpret-
ability, eating was interpreted as an item of the third factor.
Te frst, second, and third factors were cognitive, motor, and
self-care, respectively.
CFA of the WeeFIM based on EFA
In this study, the absolute ft indices included the
2
and
RMSEA. Incremental ft measures included the NFI and
CFI. Te AIC was used for a comparative measure of ft. As
shown in Table V, the 3-factor model showed good ft indices.
RMSEA scores <0.05 indicate a good model ft. Te scores
for the NFI and CFI were >0.9, which also indicates a good
model ft. Te AIC was lower than the original 1- and 2-factor
models. Correlation matrix of factors and regression weights
are presented in Tables VI and VII, respectively.
Discussion
Although the WeeFIM is used to identify changes in func-
tional outcomes afer intervention, there is lack of studies on
the structure of the WeeFIM. Terapists must not only deter-
mine the changes in the total score of the WeeFIM, but also
examine the changes according to the domains to confrm
treatment efects. Tus, it is important to identify the struc-
ture of the WeeFIM.
Te results of the present study suggest that the WeeFIM
has good internal consistency for measuring functional
outcomes in children with spastic CP. In this study, the
Cronbachs was 0.98; according to the rule of thumb for
describing internal consistency, values above 0.90 are consid-
ered excellent [26].
CFA was selected to verify the factor structure, and EFA
was employed to determine the factor structure. Te 1- and
2-factor models suggested in the previous study and a theo-
retical model were tested by CFA. Because these models did
not exhibit adequate ft indices, EFA was executed to deter-
mine the ft model in children with spastic CP. Te 3-factor
models suggested from the results of EFA were tested by CFA
across the sample. Te results support the 3-factor structure
of self-care, motor, and cognition. Te internal consistency
and construct validity of the 3-factor model were verifed.
Terefore, the 3-factor WeeFIM is a useful measurement tool
in children with CP.
Table IV. Factor structure based on eigenvalue of n = 207.
Items Cognitive Motor Self-care Communalities
Memory 0.95 0.55 0.64 0.92
Social interaction 0.94 0.54 0.66 0.90
Problem solving 0.93 0.55 0.66 0.89
Expression 0.93 0.57 0.70 0.89
Comprehension 0.90 0.50 0.62 0.83
Eating 0.80 0.66 0.79 0.79
Toilet 0.55 0.95 0.69 0.92
Bed, chair, wheelchair 0.57 0.94 0.70 0.91
Tub, shower 0.55 0.93 0.74 0.90
Walk/wheelchair 0.60 0.92 0.59 0.84
Stair 0.41 0.86 0.59 0.78
Bowel 0.79 0.81 0.44 0.87
Bladder 0.79 0.81 0.42 0.88
Dressing upper 0.68 0.68 0.93 0.90
Dressing lower 0.64 0.66 0.93 0.88
Bathing 0.65 0.64 0.91 0.85
Toileting 0.66 0.72 0.91 0.87
Grooming 0.74 0.61 0.89 0.85
Explained variance (%) 70.37 10.44 6.31 87.12
ADLs (M SD) 73 33.15
ADL, activities of daily living.
Table V. Fit indices of the 3-factor model of the WeeFIM.
df
2
NFI TLI CFI RMSEA AIC
123 514.45 0.92 0.92 0.94 0.12 646.49
AIC, Akaike Information Criterion; CFI, Comparative Fit Index; NFI, Normed Fit
Index; RMSEA, Root mean square error of approximation; TLI, Tucker-Lewis Index.
Table VI. Correlation matrix of factors.
Factors Cognitive Motor Self-care
Cognitive
Motor 0.79**
Self-care 0.78** 0.67**
**p < 0.01
Table VII. Regression weights.
B
a
b
SE
c
CR
d
MemoryCognition 1.00 0.93
Problem solvingCognition 0.96 0.93 0.03 31.86**
Social interactionCognition 0.97 0.93 0.03 31.36**
ExpressionCognition 1.04 0.95 0.04 23.57**
ComprehensionCognition 0.92 0.89 0.05 19.87**
EatingSelf-care 1.00 0.86
GroomingSelf-care 1.00 0.92 0.05 19.22**
BathingSelf-care 0.93 0.90 0.05 18.49**
Dressing UpperSelf-care 1.00 0.92 0.05 19.58**
Dressing lowerSelf-care 1.00 0.90 0.05 18.60**
ToiletingSelf-care 1.07 0.92 0.06 19.33**
BladderMotor 1.00 0.74
BowelMotor 1.02 0.74 0.03 35.51**
Bed, chair, wheelchairMotor 1.27 0.94 0.09 14.06**
ToiletMotor 1.32 0.95 0.09 14.05**
Tub, showerMotor 1.25 0.92 0.09 13.74**
Walk/wheelchairMotor 1.13 0.91 0.08 13.66**
StairMotor 1.08 0.85 0.09 12.61**
a
Unstandard regression weight.
b
Standard regression weight.
c
Standard error.
d
Critical
ratio.
**p < 0.001.
1470 E.-Y. Park et al.
Disability & Rehabilitation
Originally, the WeeFIM had a 2-factor structure: motor
and cognitive. In this study, the results for the 3-factor model
showed that the model ftted reasonably with the data. Most
researchers do not rely strictly on cutof values. Hayduk et al.
[27] argue that cutofs for a ft index can be misleading and
subject to misuse. Te
2
-test is widely used to analyze model
ft. Nevertheless, it was not used as a ft statistic because it
is sensitive to sample size. Moreover, RMSEA is commonly
used to measure ft. However, researchers can never know the
appropriate population size for RMSEA. It is not optimal to
strive for single-test accept/reject decisions because the nature
of such tests is very diferent from conventional hypothesis
tests such as the t-test. Hence, it is important to use other
goodness-of-ft measures to determine global model ft and
attend to diagnostics for the sources of model misft [28]. In
addition, if the sample size is <250, RMSEA is less preferable
[29]. We accepted the 3-factor model of the WeeFIM even
though the RMSEA ft index was >0.10. Te ft indices of the
3-factor model of the WeeFIM such as the NFI, TLI, and CFI
were at acceptable levels.
Te PEDI and WeeFIM are frequently used measures of
functional outcomes. Te items of the PEDI were developed
to ft the Rasch model, which estimates item difcultly and
personal ability and indicates a 1-dimensional construct [30].
Moreover, CFA has been used to estimate factor scores [31]. In
contrast, the WeeFIM was adapted from the adult FIM
TM
. Te
items of the WeeFIM were developed to use traditional testing
methods. Te construct validity of the WeeFIM was examined
while comparing other developmental assessments such as the
Vineland Adaptive Behavior Scales, Battelle Developmental
Inventory, and Clinical Linguistic Auditory Milestone Scale
[30]. For this reason, Rasch analysis has been performed
on the items of the WeeFIM to examine its construct valid-
ity in many studies [1,30,32]. Te conventional method of
determining the construct validity of measurement is factor
analysis. However, no study evaluating the construct validity
of the WeeFIM through factor analysis has been performed.
In this study, EFA and CFA were employed to determine the
construct validity of the WeeFIM in children with spastic CP,
and the results of this study recommend the 3-factor model.
Previous researchers have suggested that the motor domain
is not unidimensional. Chen et al. [30] performed Rasch anal-
ysis and reported that bladder and bowel management of the
motor domain do not ft well. Tey also suggested that blad-
der and bowel management measure a distinct aspect of func-
tion. Tur et al. [1] report that the motor factor rather than the
one-dimensional model is divided into several dimensions in
children with CP. Moreover, it is reported that other items do
not load on motor or cognitive scales and do not ft well. A
validity study for the internal construct of the WeeFIM was
conducted to determine factors by using principal compo-
nent analysis. In that study, stair-climbing, eating, grooming,
walking, and bathing items did not load into any factor; the
remaining motor items were loaded into the frst factor; and
all cognitive items were loaded into the second factor [33].
Tsuju et al. [32] administered the WeeFIM to 225 children
aged 6 months to 7 years without developmental delays. Te
scores were converted to an interval scale by Rasch analysis,
which assumes unidimensionality of the measurement items,
for determining the ft index of the assumption and decide
item difculty. Te results show that when the WeeFIM items
were divided into motor and cognition to minimize misft,
the degree of misft was acceptable except for eating, bladder
management, tub/shower transfer, and comprehension. Tis
suggests that items of the WeeFIM should not only be divided
into motor and cognitive domains.
In factor analytic theory, it is of primary interest to deter-
mine the underlying constructs that give rise to collected data.
Traditionally, EFA is used to explore the possible underlying
factor structure of a set of observed variables without impos-
ing a preconceived structure on the outcome. By performing
EFA, the underlying factor structure is identifed. CFA is a
statistical technique used to verify the factor structure of a
set of observed variables. In this study, the three underlying
constructs of the WeeFIM in children with spastic CP were
determined using EFA and supported though CFA.
Our analyses show that the WeeFIM has distinct self-care,
motor, and cognitive domains. Te self-care domain includes 6
items: eating, grooming, dressing upper, dressing lower, bath-
ing, and toileting. Te motor domain includes 7 items: blad-
der and bowel management, changing positions from chairs,
getting on and of the toilet, getting in and out of showers and
bathtubs, self-mobility indoors and outdoors, and ascending
and descending stairs. Te 5 items of the cognitive domain
are the same as those of the original cognitive domain. For
the 3-factor model, the results showed that the model ftted
the data reasonably well. However, the 1-factor and original
2-factor models were inadequate for children with spastic CP.
Recently, the WeeFIM Clinical Guide reported that the
bladder and bowel items are included in the self-care domain
[19]. However, the present results indicate that these items
were not loaded in the self-care domain. Tis result might
be because of the characteristics of these items. Te bladder
and bowel items can be rated both physiologically (i.e. con-
tinence) and according to performance [30]. Bladder and
bowel management is usually related to central neuromotor
control dysfunction and to the cognitive development status
of the child. Te results of EFA show that these items have
high factor loading on both the cognitive and motor domains.
As a result, it would not be meaningful to compare children
with spastic CP with nondisabled children.
Usefulness and responsiveness of the WeeFIM instrument
may be afected by subtotal scores. Te results of this study
suggest that the 3-factor WeeFIM is more acceptable than the
original 2-factor WeeFIM for children with spastic CP. Tus,
the self-care, motor, and cognitive domains should be treated
as separate scales in children with spastic CP. Tese results sup-
port the belief that the psychometric properties of a population
should be examined prior to the use of a given instrument.
Although the reliability and validity of the 3-factor
WeeFIM were examined in this study, further studies are
needed to determine additional psychometric properties. In
addition, comparative data should be accumulated using a
variety of methodologies. Because only children with spastic
CP participated in the present study, the present results may
not be representative of other types of CPs.
Construct of WeeFIM in children with spastic CP1 1471
2013 Informa UK, Ltd.
Declaration of Interest: Te authors report no conficts of
interest.
References
1. Tur BS, Kkdeveci AA, Kutlay S, Yavuzer G, Elhan AH, Tennant A.
Psychometric properties of the WeeFIM in children with cerebral palsy
in Turkey. Dev Med Child Neurol 2009;51:732738.
2. Aneja S. Evaluation of a child with cerebral palsy. Indian J Pediatr
2004;71:627634.
3. Gunel MK, Mutlu A, Tarsuslu T, Livanelioglu A. Relationship among
the Manual Ability Classifcation System (MACS), the Gross Motor
Function Classifcation System (GMFCS), and the functional sta-
tus (WeeFIM) in children with spastic cerebral palsy. Eur J Pediatr
2009;168:477485.
4. Kwolek A, Majka M, Pabis M. Te rehabilitation of children with cere-
bral palsy: problems and current trends. Ortop Traumatol Rehabil
2001;3:499507.
5. Msall ME, Rogerss BT, Ripstein H, Lyon N, Wllezenski F. Measurements
of functional outcomes in children with cerebral palsy. Ment Retard Dev
Disabil Res Rev 1997;8:194203.
6. 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 children with spas-
tic cerebral palsy. Dev Med Child Neurol 2002;44:468476.
7. Mayston MJ. People with cerebral palsy: efects of and perspectives for
therapy. Neural Plast 2001;8:5169.
8. Russell DJ, Rosenbaum PL, Cadman DT, Gowland C, Hardy S,
Jarvis S. Te gross motor function measure: a means to evaluate the
efects of physical therapy. Dev Med Child Neurol 1989;31:341352.
9. Haley SM, Coster WJ, Ludlow LH, Haltiwanger JT, Andrellos PJ (eds).
Pediatric evaluation of disability inventory (PEDI). Version 1.0. devel-
opment, standardization and administration manual. Boston: New
England Center Hospital, 1992.
10. Graham HK, Harvey A, Rodda J, Nattrass GR, Pirpiris M. Te Functional
Mobility Scale (FMS). J Pediatr Orthop 2004;24:514520.
11. Ottenbacher KJ, Msall ME, Lyon NR, Dufy LC, Granger CV,
Braun S. Interrater agreement and stability of the Functional
Independence Measure for Children (WeeFIM): use in children with
developmental disabilities. Arch Phys Med Rehabil 1997;78:13091315.
12. Ottenbacher KJ, Msall ME, Lyon N, Dufy LC, Ziviani J, Granger CV,
Braun S, Feidler RC. Te WeeFIM instrument: its utility in detecting
change in children with developmental disabilities. Arch Phys Med
Rehabil 2000;81:13171326.
13. Ziviani J, Ottenbacher KJ, Shephard K, Foreman S, Astbury W,
Ireland P. Concurrent validity of the Functional Independence Measure
for Children (WeeFIM) and the Pediatric Evaluation of Disabilities
Inventory in children with developmental disabilities and acquired
brain injuries. Phys Occup Ter Pediatr 2001;21:91101.
14. Oefnger DJ, Tylkowski CM, Rayens MK, Davis RF, Gorton GE 3rd,
DAstous J, Nicholson DE, et al. Gross Motor Function Classifcation
System and outcome tools for assessing ambulatory cerebral palsy: a
multicenter study. Dev Med Child Neurol 2004;46:311319.
15. Braun SL, Granger CV. A practical approach to functional assessment in
pediatrics. Occup Ter Pract 1991;2:4651.
16. Wong V, Wong S, Chan K, Wong W. Functional Independence Measure
(WeeFIM) for Chinese children: Hong Kong Cohort. Pediatrics
2002;109:E36.
17. Guide for the Uniform Data Set for Medical Rehabilitation (Adult FIM),
version 4.0. Bufalo, NY: State University of New York at Bufalo, 1993.
18. Finch E, Brooks D, Stratford PW, Mayo NE (eds). Physical Rehabilitation
Outcome Measures: a guide to enhanced clinical decision making.
Hamilton: Lippincott Williams and Wilkins, 2002.
19. Uniform Data System for Medical Rehabilitation. Te WeeFIM Clinical
System Guide, Version 5.01. New York: State University of New York,
2000.
20. MacCallum RC, Widaman KF, Zhang S, Hong S. Sample size in factor
analysis. Psychol Methods 1999;4:84.
21. Kunce JT, Cook WD, Miller DE. Random variables and correlational
overkill. Educat Psychol Measurement 1975;35: 529524.
22. Finn JD, West SG. Te investigation of personality structure: Statistical
models. J Res Personality 1997;31:439485.
23. Tabachnick B, Fidell L. Using multivariable statistics. New York: Allyn &
Bacon, 2007.
24. Pett MA, Lackey NR, Sullivan JJ. Making sense of factor analysis.
California: Sage, 2003.
25. Costello AM, Osborn JW. Best practices in exploratory factor analysis:
four recommendations for getting the most from your analysis. Pract
Assess Res Eval 2005;10:19.
26. Nunnally JC, Bernstein IH. Pshychometric theory. New York: McGraw-
Hill, 1994.
27. Hayduk L, Cummings GG, Boadu K, Pazderka-Robinson H, Boulianne
S. Testing! Testing! One, two three Testing the theory in structural
equation models!. Pers Individ Dif 2007;42: 841850.
28. Bentler P. On tests and indices for evaluating strucutural models. Pers
Individ Dif 2007;42: 825829.
29. Chen F, Curran PJ, Bollen KA, Kirby J, Paxton P. An Empirical Evaluation
of the Use of Fixed Cutof Points in RMSEA Test Statistic in Structural
Equation Models. Sociol Methods Res 2008;36:462494.
30. Chen CC, Bode RK, Granger CV, Heinemann AW. Psychometric
properties and developmental diferences in childrens ADL item hier-
archy: a study of the WeeFIM instrument. Am J Phys Med Rehabil
2005;84:671679.
31. Chen KL, Hsieh CL, Sheu CF, Hu FC, Tseng MH. Reliability and validity
of a Chinese version of the Pediatric Evaluation of Disability Inventory
in children with cerebral palsy. J Rehabil Med 2009;41:273278.
32. Tsuji T, Liu M, Toikawa H, Hanayama K, Sonoda S, Chino N. ADL
structure for nondisabled Japanese children based on the Functional
Independence Measure for Children (WeeFIM). Am J Phys Med Rehabil
1999;78:208212.
33. Canan A, Gulten E, Atilla E, Hulya S, Sumru O. ADL assessment of non-
disabled Turkish children with the WeeFIM instrument. Am J Phys Med
Rehabil 2007;86:176182.