Journal of Pediatric Psychology, Vol. 28, No. 3, 2003, pp. 191-196
© 2003 Society of Pediatric Psychology
Brief Report: Cross-Cultural Replication of an Anomalous Psychometric Pattern in Children With Type 1 Diabetes
1 Children's National Medical Center, 2 The George Washington University, 3 Georgetown University, 4 University of Puerto Rico, 5 Virginia Commonwealth University
All correspondence should be sent to Randi Streisand, Children's National Medical Center, Dept. of Psychology, 111 Michigan Ave., NW, Washington, District of Columbia 20010. E-mail: rstreis{at}cnmc.org. Anne Kazak, PhD, ABPP, former Editor, served as accepting editor on this article.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodResultsDiscussionReferences |
|---|
Objective To replicate an anomalous psychometric profile previously documented in children with Type 1 diabetes living in the mainland United States with a cross-cultural sample selected from Puerto Rico. Methods Ninety-three Spanish-speaking children (M age = 12.8 years) with Type 1 diabetes living in Puerto Rico were administered the Puerto Rican version of the Wechsler Intelligence Scale for Children-Revised (WISC-R). The factor structure of the Puerto Rican sample's WISC-R was then compared to that of a United States sample (n = 95) in which an anomalous factor structure in children with diabetes was first documented. Results As in the United States sample, a four-factor IQ structure was obtained. Instead of the traditional three-factor structure of the WISC-R, the Perceptual Organization factor split into a Spatial Conceptual factor and an anomalous Visual Discrimination factor. Conclusions Results support previous findings and suggest anomalies in the psychometric profiles of children with Type 1 diabetes. Cross-cultural replication of the anomalous IQ factor structure, and atypical visual discrimination, suggests that differences are illness-related, and consideration may therefore be warranted when administering some subtests of the Wechsler scales to children with Type 1 diabetes.
Key words: psychometric profile; Type 1 diabetes; IQ.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Type 1 or insulin-dependent diabetes mellitus (IDDM) is the most common endocrine/metabolic disorder of childhood (Sperling, 1997
) and requires
adherence to a complex daily regimen to maintain near normal metabolic
control. Even with careful adherence, blood glucose fluctuations decrease the
supply of oxygen and glucose to the brain and result ultimately in brain
insult. Differences in the cognitive functioning of children with diabetes
have been well documented elsewhere
(Holmes, Cant, Fox, Lampert, & Greer,
1999; Ryan, 1990).
For example, research has demonstrated clinical impairments in children with
diabetes on measures of reasoning, visuomotor performance, verbal abilities,
and memory (Ryan, Vega, Longstreet, &
Drash, 1984; Rovet &
Ehrlich, 1999). Though the majority of these studies have examined
differences in test scores between children with diabetes and healthy
controls, the impact of diabetes has also been suggested through atypical
psychometric profiles.
Specifically, a psychometrically anomalous factor structure of the Wechsler
Intelligence Scale for Children-Revised (WISC-R;
Wechsler, 1974) has emerged
for children with diabetes (Holmes,
Cornwell, Dunlap, Chen, & Lee, 1992). Factor analytic studies
of the WISC-R, based on the original standardization sample of children who
were disease-free, revealed a three-factor structure (Verbal Comprehension,
Perceptual Organization, and Freedom From Distractibility) that has been
consistently replicated across diverse samples, including groups of Caucasian,
African American, and Hispanic children in the United States
(Juliano, Haddad, & Carroll,
1988; Kaufman,
1975; Taylor & Ziegler,
1987), as well as over time
(Wechsler, 1974). However,
Holmes et al. (1992) found that
within a group of children with diabetes, the Perceptual Organization factor
split into two separate factors. The first factor was a Spatial Conceptual
factor consisting of Block Design and Object Assembly. The Spatial Conceptual
factor acted as a measure of nonverbal IQ and intercorrelated with the other
IQ subtests. The other Perceptual Organization factor that emerged was an
anomalous Visual Discrimination factor, consisting of Picture Completion and
Picture Arrangement, that did not correlate with the other IQ subtests. This
anomalous psychometric pattern was first described in a Midwestern sample of
children with diabetes and was extended cross-regionally to a Southern sample
of children with diabetes (Holmes et al.,
1993). Such replication in disparate geographic samples and
medical centers provides further support for the hypothesis that the cognitive
characteristics of diabetes may include an atypical psychometric Visual
Discrimination factor.
While the impact of Type 1 diabetes on the cognitive functioning in children continues to be investigated in children living in the United States, fewer data are available on children reared in other cultures. If cognitive difficulties as well as anomalous psychometric patterns are found across cultures, languages, and other group differences, the illness itself, diabetes, is implicated as the etiologic underpinning.
This study sought to determine the generalizability of the anomalous
diabetes Wechsler factor structure across cultures and languages to a sample
of Spanish-speaking youths with Type 1 diabetes living in Puerto Rico. A
cross-cultural methodology has the potential to verify illness-related
cognitive patterns while minimizing the possible contribution of language and
cultural differences. To date, there have been no previous attempts at
cross-cultural replication of the growing body of diabetes-related cognitive
sequelae to non-English-speaking children outside the mainland United States,
nor have there been attempts to verify the anomalous factor structure of the
WISC-R despite its possible ramification for intellectual assessment
(Holmes et al., 1992).
|
Method |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Participants included 93 Spanish-speaking children with Type 1 diabetes living in Puerto Rico. The original English-speaking mainland U.S. sample (Holmes et al., 1992
) of 95
children with diabetes served as the comparison group. The Puerto Rican (P.R.)
sample was matched in age restrictions to that of the U.S. diabetes sample;
children were between the ages of 8 and 16 years (M P.R. and U.S.
ages = 12.8 and 12.5 years, respectively). Within an age level (e.g., all
8-year-olds), an equal gender distribution was sought such that at least five
boys and five girls were recruited for participation. Trained psychology research assistants recruited children for the study during their participation in a summer diabetes camp offered to children living across Puerto Rico, free of charge. Informed consent and assent were obtained on the first day of camp registration from parents and children. Children who attended camp received routine medical care from physicians who provided services to both public and private patients. Thus, participants were sampled from a pool of children representing a range of socioeconomic strata selected for study participation on the basis of gender, age, and camp scheduling availability. Children who participated were compensated with $20 for their assistance. The Puerto Rican adaptation of the Spanish version of the WISC-R (the Escala de Inteligencia Wechsler para Ninos-Revisada de Puerto Rico [EIWN-R PR]; Herrans & Rodriguez, 1992), which has the same three-factor structure as the WISC-R, was individually administered to all children. All test protocols were scored twice to ensure accuracy.
|
Results |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Examination of demographic differences between the P.R. and U.S. samples indicated no age differences between groups of children, t(179) = 1.36, p > .05. There also were no differences between the groups in average disease duration (M duration = 4.7 vs. 5.4 years, respectively), t(160) = 1.32, p > .05, nor age of onset (M onset age = 7.9 vs. 7.4 years, respectively), t(160) = -.85, p > .05. IQ scores for both the P.R. and U.S. groups were in the average range and were normally distributed.
P.R. and U.S. Factor Structures With Confirmatory Factor
Analysis
Age-adjusted scaled scores from each of the Wechsler intelligence scales,
the WISC-R and the EIWN-R PR, were used for factor analysis and factor
structure comparisons. The first hypothesis tested whether the IQ factor
structure for P.R. children with IDDM fit the four-factor model indicated in
Holmes et al. (1992).
Confirmatory factor analysis using the maximum likelihood method with the
structural equation modeling program
(Bentler, 1989) was applied to
test this hypothesis.
Prior to conducting the confirmatory factor analysis, PRELIS
(Joreskog & Sorbom, 1996)
software was used to evaluate the assumption of multivariate normality
required by the maximum likelihood method. Results indicated that all 11
EIWN-R PR subtests demonstrated a univariate skewness index with an absolute
value less than 1. For the multivariate normality assumption, the combined
skewness and kurtosis had a 
2 = 1.383 with p = 0.51,
indicating multivariate normality for the 11 IQ subscales. The covariance
matrices for the two samples with the 11 IQ subscales are presented in
Table I.
|
The results from the confirmatory factor analysis showed that the
four-factor solution fit the P.R. sample well, with the likelihood ratio
2 (38) = 42.35, p = 0.29, a Bentler-Bonett Nonnormed
Fit Index (BBNFI) = .98, and the Comparative Fit Index (CFI) = .99. A model
with a BBNFI or CFI fit index of greater than .9 is considered a good fit. The
BBNFI and the CFI indices are relatively insensitive to sample sizes. With
both BBNFI and CFI values near 1, the fitted four-factor pattern is regarded
as stable. For comparison purposes, a similar confirmatory factor analysis of
the four-factor model was also conducted for the U.S. sample with 95 children.
This analysis resulted in 2 (38) = 47.94, p = 0.13,
the BBNFI = .95, and CFI = .97, indicating that the four-factor model also fit
the data for the U.S. sample.
Multigroup Method Comparing U.S. and P.R. Factor Structures
The second hypothesis predicted no significant differences between the P.R.
and U.S. factor structures for the children with IDDM in these samples. This
null hypothesis was tested using the multigroup method available in structural
equation modeling, a method flexible in the way that it can be programmed to
test the similarity of factor structures across multiple samples. Factor
structures can be tested at different levels ranging from completely identical
to the lowest level of similarity, which is the same number of factors.
Because the primary interest of this study was whether the IQ subtests
clustered into similar factors for the U.S. and P.R. samples, we tested the
factor structure similarity at this level. In other words, no assumptions were
made about the similarity of the factor item loadings across samples nor were
statistical tests conducted to compare the similarity of the factor item
loadings. Specifically, a multigroup comparison program with the same
four-factor model, with no parameter constraints across the two samples, was
tested using structural equation modeling. This test resulted in a likelihood
ratio of 2 (76) = 90.29, p = .13, BBNFI = .97 and CFT
= .98, indicating no significant difference in terms of the number of factors
nor the clustering of the Wechsler subtests on the relative factors between
the U.S. and P.R. samples of children with IDDM (n = 182).
To demonstrate the degree of similarity for the factor patterns of the two
samples, the total congruence coefficient
(Chan, Ho, Leung, Chan, & Yung,
1999) also was calculated. Using this formula, the total
congruence coefficient for the P.R. and U.S. samples based on the four-factor
model was .91, indicating acceptable similarity between factor structures
(Mulaik, 1972).
Factor Loadings for the U.S. and P.R. Samples From Exploratory Factor
Analysis
In addition to the confirmatory factor analyses, exploratory factor
analyses were conducted in each sample because they avoid a priori assumptions
of similarity between the factors of each independent sample, as must occur
with confirmatory factor analysis. Maximum likelihood factor analysis,
followed by a varimax rotation, was utilized to be consistent with the method
used in Holmes et al. (1992)
and the method in the standardization data of the WISC-R
(Wechsler, 1974) and EWIN-R PR
(Herrans & Rodriquez,
1992). The factor loadings from the exploratory factor analysis
for the U.S. and the P.R. samples are presented in
Table II. As can be seen, both
of the exploratory factor analyses detected a four-factor structure, with
Picture Completion (P.R. only) and Picture Completion and Picture Arrangement
(U.S.) forming a separate fourth factor, the Visual Discrimination factor.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Results revealed that across two different cultures, youths with Type 1 diabetes displayed an atypical, yet similar, psychometric pattern: a four-factor structure on the frequently utilized Wechsler intelligence scales for children. These results suggest that cognitive sequelae of diabetes, as measured in this study by the Wechsler scales, may transcend culture, language, and geographical region. Two culturally different samples of children with Type 1 diabetes, an illness whose first chronic complication typically involves visual anomalies (Diabetes Control and Complications Trial Research Group, 1994
), displayed a four-factor structure, in which
the traditional Perceptual Organization factor split into two separate
factors. Block Design and Object Assembly, subtests that demand spatial
conceptual skills, loaded together and created a Spatial Conceptual factor.
Previous work with the U.S. sample suggests that this factor acts as a
traditional measure of nonverbal IQ that correlates well with academic
achievement and other IQ subtest scores
(Holmes et al., 1992). In
contrast, Picture Completion and Picture Arrangement, two subtests that
require fine visual discrimination skills, loaded together for the U.S. sample
to comprise a Visual Discrimination factor, while Picture Completion alone
comprised the Visual Discrimination factor for the P.R. children. In the
standardization data of both the WISC-R and EIWN-R PR, the Picture Arrangement
subtest loads almost equally on the Perceptual Organization factor, where it
has the highest loadings, and the Verbal Comprehension factor, where the
loadings are slightly lower (i.e., .45 vs. .43; Carroll, Herrans, &
Rodriguez, 1995; Kaufman,
1975), perhaps explaining the loading of Picture Arrangement on
the Verbal Comprehension factor in the P.R. diabetes sample. Previous work in
our sample of children with diabetes in the United States
(Holmes et al., 1992) indicates
that this Visual Discrimination factor is anomalous and does not act
as a traditional measure of nonverbal IQ. Specifically, it does not correlate
with academic achievement or with other IQ subtests, suggesting that the
visual discrimination or perceptual demands of the task may eclipse other
nonverbal cognitive task demands for children with diabetes. Intact
performance on the Spatial Conceptual factor (i.e., its intercorrelations with
other IQ and achievement tasks) suggests that general nonverbal intelligence
is unaffected in these children with diabetes, although our hypothesis of an
isolated perceptual difficulty underlying the Visual Discrimination factor,
which is unlinked to general nonverbal intelligence, remains speculative and
requires further documentation.
Previous factor-analytic work has traditionally evaluated pediatric
populations or different cultural groups separately. For example, a study of
young children with epilepsy found a two-factor WISC-R structure (i.e., Verbal
IQ/Performance IQ; Richards, Fowler,
Berent, & Boll, 1980). Studies with English-speaking Hispanic
children on the WISC-R have either found a two-factor solution in students
referred for psychological services (Gutkin
& Reynolds, 1980) or the Kaufman three-factor solution in a
nonreferred sample of Chicano children
(Reschly, 1978). Regardless of
whether a two- or three-factor structure was found, the Picture Completion and
Picture Arrangement subtests loaded with Block Design and Object Assembly to
form either a Performance IQ score or a Perceptual Organization factor. This
study is the first to concurrently compare a cross-cultural pediatric sample,
and the creation of an anomalous Visual Discrimination factor with Picture
Completion or Picture Arrangement loading separately is unique to samples of
children with diabetes in this study and previously (Holmes et al.,
1992,
1993). These findings also
provide the first important cross-cultural information that the anomalous
diabetes Visual Discrimination factor occurs in both Caucasian and native
Hispanic children with diabetes.
The U.S. and P.R. samples were not significantly different in age, duration
of illness, nor age of disease onset, providing support for the hypothesis
that these findings are likely a result of biologic or functional sequelae
common to diabetes, rather than demographic factors. Given that these striking
psychometric profiles occurred similarly across two different cultures, yet in
neither culture's test standardization sample, the validity of attributing the
anomalous finding to some aspect of the disease itself is strengthened. The
present cross-cultural methodology can help rule out potential cultural
contribution effects that may relate to psychometric performance differences,
such as cultural testing expectations, different values about speeded
responding, and so forth. Diabetes-related performance similarities that
occurred cross-culturally help to confirm that some aspect of the illness is
invariant across cultures, possibly early visual anomalies, that is related to
this psychometric pattern. Because of a possible Visual Discrimination
anomaly, caution is warranted in the interpretation of Perceptual Organization
test results for children with diabetes, from both the United States and from
different cultures. Some of these subtests, particularly Picture Completion,
may not act as standard measures of nonverbal intelligence, as they do for
children without diabetes (Holmes et al.,
1992).
The WISC-R has largely been replaced by the newer third edition (WISC-III;
Wechsler, 1991), although the
two versions are composed of the same Verbal Comprehension and Perceptual
Organization factors (Kaufman,
1994), and IQ scores on each version have been significantly
correlated with one another (Lyon,
1995). The similarity between the WISC-R and WISC-III Perceptual
Organization factors suggests that the diabetes-related findings presented may
generalize to the WISC-III. However, use of the P.R. version of the WISC-R is
a limitation of this study, and only future studies with the WISC-III, and a
P.R. adaptation of the WISC-III, can confirm such generalizations. At present,
though, a P.R. adaptation of the WISC-III does not exist. Examining data on
children's socioeconomic status would also strengthen future studies.
At a minimum, results from this study indicate that the psychometric profiles of children with diabetes, and other pediatric conditions, may differ from the profiles of children for whom tests were initially developed. More important, professionals conducting intellectual assessments with the Wechsler scales in children who have diabetes should consider that these findings may implicate disease-specific visual discrimination difficulties, particularly on the Picture Completion subtest. Continued research on the neurocognitive functioning of children with diabetes will help to elucidate these findings. In an effort to document the underlying processes that account for these factor analytic differences, future studies of children with diabetes should examine perceptual organization and visual discrimination functioning in children longitudinally.
|
Acknowledgments |
|---|
Portions of this research were supported by grants DK37545 and DK to Clarissa S. Holmes. We thank Ana Marie Quartro, Silvia Cochran, Pedro Martinez, and Pilar Marie for their assistance with data collection and Mara Richards and Rachel Rodriguez for their assistance with data management. Most important, we are grateful to the children and parents who volunteered their time to participate in this project.
Received September 27, 2001; revision received March 21, 2002; accepted May 28, 2002
|
References |
|---|
|
TopAbstractIntroductionMethodResultsDiscussionReferences |
|---|
Bentler, P. M. (1989). EQS: Structural Equations Program manual. Los Angeles: BMDP Statistical Software.
Carroll, J. F., Herrans, L., & Rodriquez, J. M. (1995). Analisis factorial de la EIWN-R de Puerto Rico, con ninos de 11 niveles de edad, entre los 6 y los 16 anos. Revista-Latinoamericana-de-Psicologia, 27, 187-206.
Chan, W., Ho, R., Leung, K., Chan, D. K., & Yung, F. (1999). An alternative method for evaluating congruence coefficients with procrustes rotation. Psychological Methods, 4, 378-402.[CrossRef]
Diabetes Control and Complications Trial Research Group. (1994). Effect of intensive diabetes treatment on the development and progression of longterm complications in adolescents with insulin-dependent diabetes mellitus: Diabetes Control and Complications Trial. Journal of Pediatrics, 125, 228-229.[CrossRef][ISI][Medline]
Gutkin, T. B., & Reynolds, C. R. (1980). Factorial similarity of the WISC-R for Anglos and Chicanos referred for psychological services. Journal of School Psychology, 18, 34-39.[CrossRef]
Herrans, L., & Rodriquez, J. M. (1992). Manual for the Escala de Inteligencia Wechsler para Ninos—Revisada de Puerto Rico (EIWN-R PR). San Antonio, TX: Psychological Corporation.
Holmes, C. S., Cant, M. C., Fox, M. A., Lampert, N. L., & Greer, T. (1999). Disease and demographic risk factors for disrupted cognitive functioning in children with insulin-dependent diabetes mellitus (IDDM). School Psychology Review, 28, 215-227.
Holmes, C. S., Cornwell, J., Dunlap, W. P., Chen, R. S., & Lee, C. Y. (1992). Anomalous factor structure of the WISC-R for diabetic children. Neuropsychology, 6, 341-350.[CrossRef]
Holmes, C. S., Dunlap, W. P., Chen, R. S., Cornwell, J., Weissman, L., Obach, M., et al. (1993). Postpubertal disease status in diabetes and factor structure anomaly on the WISC-R. Journal of Clinical and Experimental Neuropsychology, 15, 843-848.[ISI][Medline]
Joreskog, K. G., & Sorbom, D. (1996). PRELIS 2 user's reference guide. Chicago: Scientific Software International.
Juliano, J. M., Haddad, F. A., & Carroll, J. L. (1988). Three-year stability of WISC-R factor scores for Black and White, female and male children classified as learning-disabled. Journal of School Psychology, 26, 317-325.
Kaufman, A. S. (1975). Factor analysis of the WISC-R at eleven age levels between 6.5 and 16.5 years. Journal of Consulting and Clinical Psychology, 43, 135-147.[CrossRef]
Kaufman, A. S. (1994). Intelligent testing with the WISC-III. New York: John Wiley.
Lyon, M. A. (1995). A comparison between WISC-III and WISC-R scores for learning disabilities reevaluations. Journal of Learning Disabilities, 28, 253-258.
Mulaik, S. A. (1972). The foundations of factor analysis. New York: McGraw-Hill.
Reschly, D. J. (1978). WISC-R factor structures among Anglos, Blacks, Chicanos, and Native-American Papagos. Journal of Consulting and Clinical Psychology, 46, 417-422.[CrossRef]
Richards, H. C., Fowler, P. C., Berent, S., & Boll, T. J. (1980). Comparison of the WISC-R factor patterns for younger and older epileptic children. Journal of Clinical Neuropsychology, 2, 333-341.
Rovet, J. F., & Ehrlich, R. M. (1999). The effect of hypoglycemic seizures on cognitive function in children with diabetes: A 7 year prospective study. Journal of Pediatrics, 134, 503-506.[CrossRef][ISI][Medline]
Ryan, C. M. (1990). Neuropsychological consequences and correlates of diabetes in childhood. In C. S. Holmes (Ed.), Neuropsychological and behavioral aspects of diabetes (pp. 58-84). New York: Springer-Verlag.
Ryan, C., Vega, A., Longstreet, C., & Drash, A. (1984). Neuropsychological changes in adolescents with insulin-dependent diabetes. Journal of Consulting and Clinical Psychology, 52, 335-342.[CrossRef][ISI][Medline]
Sperling, M. A. (1997). Aspects of the etiology, prediction, and prevention of insulin-dependent diabetes mellitus in childhood. Pediatric Clinics of Northern America, 44, 269-284.
Taylor, R. L., & Ziegler, E. W. (1987). Comparison of the first principal factor on the WISC across ethnic groups. Educational and Psychological Measurement, 47, 691-694.[Abstract]
Wechsler, D. (1974). Manual for the Wechsler Intelligence Scale for Children-Revised. New York: Psychological Corporation.
Wechsler, D. (1991). Manual for the Wechsler Intelligence Scale for Children-Third Edition. New York: Psychological Corporation.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. D. Nelson, B. S. Aylward, and R. G. Steele Structural Equation Modeling in Pediatric Psychology: Overview and Review of Applications J. Pediatr. Psychol., August 1, 2008; 33(7): 679 - 687. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||