The Relationship Between Age and IQ in Adults With Williams Syndrome

VOLUME 109, NUMBER 3: 231–236
MAY 2004
AMERICAN JOURNAL ON MENTAL RETARDATION
The Relationship Between Age and IQ in Adults
With Williams Syndrome
Yvonne M. Searcy, Alan J. Lincoln, Fredric E. Rose, Edward S. Klima, and Nasim Bavar
The Salk Institute for Biological Studies, Laboratory for Cognitive Neuroscience
(La Jolla, CA)
Julie R. Korenberg
Cedars-Sinai Medical Center, Laboratory for Human Molecular Genetics (Los Angeles)
Abstract
The relationship between age and IQ was evaluated in a cross-sectional sample of 80
individuals with Williams syndrome (17 to 52 years). The relationship between age and
WAIS-R subtest scores was such that increases and decreases in raw scores occurred at a
rate sufficient to maintain stability of age-corrected scaled scores, indicating a develop-
mental trajectory similar to that of the WAIS-R normative sample. Despite stability of age-
corrected scaled scores with age, increased age was related to higher Performance IQ. This
disparity, which occurs during the conversion of sums of scaled scores to IQs, may be
unique to the WAIS-R. Although Performance IQ increased with age, results imply that
the overall IQ of an adult with Williams syndrome will likely remain stable.
Identifying and understanding the develop-
velopmental syndromes. In adults with Down
mental trajectory of intellectual ability in persons
syndrome, for example, longitudinal studies show
with neurodevelopmental syndromes is important
declines in IQs with age in about one third of
to the characterization of the neurocognitive phe-
those under 45, in over 70% of those 45 to 49,
notype of the syndrome. Researchers often em-
and diminishing IQ into old age (Carr, 1994; Fen-
ploy standardized intelligence tests for this pur-
ner, Hewitt, & Torpy, 1987). Males with fragile X
pose. Intelligence tests are also administered to
syndrome show intellectual growth and stability
adults with developmental disabilities in order to
of IQs until about 10 to 15 years of age, at which
determine competence, monitor functioning, as-
point IQs begin to decline (Dykens et al., 1989;
sess employability, and assess eligibility for social
Hagerman et al., 1989). Conversely, persons with
services. Thus, information regarding potential de-
autism and no mental retardation show evidence
velopmental changes in intellectual ability in in-
of verbal intellectual skills that improve through
dividuals with a specific neurodevelopmental syn-
adolescence and into adulthood (Kuck, Lincoln,
drome can provide caregivers with valuable infor-
& Heaton, in press). The developmental trajectory
mation useful in planning for continuing educa-
of IQ in Williams syndrome has not been fully
tion and long-term support.
explored.
Because IQs are derived with respect to the
Williams syndrome is a genetic disorder as-
performance of same-age peers, IQs of typically
sociated with the deletion of one copy of the gene
developing individuals remain relatively stable
for elastin and several surrounding genes on chro-
with age (Schaie, 1983). Age-related changes in
mosome 7q11.23. The phenotype of Williams
IQs that vary from the normal pattern of devel-
syndrome includes heart defects, such as supra-
opment have been reported for several neurode-
valvular aortic stenosis, as well as dysmorphic
American Association on Mental Retardation
231

---

VOLUME 109, NUMBER 3: 231–236
MAY 2004
AMERICAN JOURNAL ON MENTAL RETARDATION
Age and IQ in Williams syndrome
Y. M. Searcy et al.
body and facial features, brain abnormalities in
man, & Rosenfeld, 1985; Sattler, 1988; Spitz,
both gross anatomy and cytoarchitecture, and vi-
1988). Thus, there is an increased likelihood that
suospatial and visuomotor integrative impair-
spurious improvement in IQs would be observed
ments (Bellugi & St. George, 2000; Pober & Dy-
in intellectually impaired individuals tested with
kens, 1996). Most individuals with Williams syn-
the WISC-R as a child and then the WAIS-R as
drome exhibit some degree of intellectual impair-
an adult. Due to the limits on direct comparability
ment, with the majority of adults scoring in the
between the WISC-R and WAIS-R, researchers
mild range of mental retardation (55 to 69 points)
who employ the WISC-R and/or WAIS-R to ex-
on standardized intelligence tests (Howlin, Da-
amine the relationship between age and IQs are
vies, & Udwin, 1998).
restricted by the respective age-range limitations of
Much of what is known about the relationship
the tests.
between IQ and aging in adults with Williams syn-
Although longitudinal analyses, particularly
drome comes from studies by Udwin, Davies, and
those involving cross-sequential cohort designs,
Howlin (1996), whose findings suggest no decline
may be the optimal method of examining age-
in IQs from adolescence into early adulthood. For
related changes in IQs (see Schaie, 1983), Kauf-
example, they conducted a longitudinal study of
man (1990) showed that evaluating age-related
23 individuals with Williams syndrome over an
changes in the sums of the scaled scores relative
8.5-year period. Participants ranged from 10 to
to changes in standard scores earned on the
15.75 years at initial testing with the Wechsler In-
WAIS-R may be useful in ascertaining whether
telligence Scale for Children-Revised WISC-R
there are true developmental changes in intellec-
(Wechsler, 1974), and from 19 to 24.83 at re-testing
tual functioning with age in a cross-sectional sam-
with the Wechsler Adult Intelligence Scale-
ple. Using this approach in the present study, we
Revised WAIS-R (Wechsler, 1981). Unlike the IQs
employed the WAIS-R to evaluate age-related
of individuals with fragile X or Down syndrome,
changes in IQ in a cross-sectional sample of 80
IQs of individuals with Williams syndrome fol-
individuals with Williams syndrome between the
lowed the same trend as the general population in
ages of 17 and 52 years. We evaluated the rela-
showing a slight increase from WISC-R testing to
tionship between chronological age (CA) and raw
WAIS-R re-testing, indicating no apparent decline
scores as well as between CA and age-corrected
in intellectual functioning over time. Howlin et al.
scaled scores. With this approach we assessed
(1998) compared the mean WAIS-R scores of 62
whether potential age-related changes in IQs (Ver-
individuals with Williams syndrome (19 to 39 years
bal, Performance, and Full-Scale) were due to ‘‘los-
of age) to the mean WISC-R scores reported for
ing ground’’ compared to the typically developing
44 children in an earlier study that included some
normative sample or related to true developmen-
of the same individuals (Udwin & Howlin, 1987,
tal change in intellectual ability (increases or de-
as cited in Howlin et al., 1998). They reported that
creases in raw scores) with age.
55% of the children for whom the WISC-R was
administered had an IQ below 50, with 22% of the
Method
sample scoring below the basal level of 40. In con-
trast, all of the adults who completed the WAIS-R
Participants
scored above the basal level, and only 4.8% had
Eighty individuals with Williams syndrome
an IQ below 50, suggesting ‘‘some changes in score
(35 males, 45 females; M age
29.8 years, range
levels over time’’ (Howlin et al., 1998, p. 187).
17 to 52) participated. They all had a clinical
Although the two studies above provide lim-
diagnosis of Williams syndrome and obtained a
ited evidence suggesting that IQ in people with
score of at least 3 points on the Williams syn-
Williams syndrome does not decline with age from
drome Diagnostic Score Sheet, indicating the
adolescence to age 39, such conclusions must be
presence of a minimum threshold for common
tempered due to the methodological confounds
medical and physical characteristics associated
imposed by the psychometric relationship between
with Williams syndrome in clinical studies (Amer-
the WISC-R and WAIS-R. It is well-established
ican Academy of Pediatrics, 2001). In addition, 79
that the WAIS-R yields higher IQs than does the
participants tested positive on a FISH test (fluo-
WISC-R and that this disparity becomes more ev-
rescence in situ hybridization) for the absence of
ident toward the lower end of the intelligence
one copy of the gene for elastin on chromosome
curve (Avery, Slate, & Chovan, 1989; Rubin, Gold-
7 (Korenberg et al., 2000). The remaining partic-
232
American Association on Mental Retardation

---

VOLUME 109, NUMBER 3: 231–236
MAY 2004
AMERICAN JOURNAL ON MENTAL RETARDATION
Age and IQ in Williams syndrome
Y. M. Searcy et al.
ipant was not tested for the elastin deletion but
conducting profile analysis and comparing an in-
was diagnosed with Williams syndrome based
dividual’s performance directly with that of same-
upon having sufficient characteristics of this syn-
age peers as well as for making subtest interpre-
drome according to the Diagnostic Score Sheet.
tations and comparisons (Sattler & Ryan, 1988).
All WAIS-R data were obtained as part of an on-
The IQs and age-corrected scaled scores were de-
going research project conducted at the Salk In-
rived according to the test manual instructions.
stitute’s Laboratory for Cognitive Neuroscience.
Genetic evaluation of the elastin deletion by FISH
was performed at the Cedars-Sinai Medical Center
Results
Laboratory for Human Molecular Genetics.
Approximately 46% of the 80 individuals test-
ed achieved a FSIQ in the borderline range (be-
Materials and Procedures
tween 85 and 70 points), 45% scored in the mild
The WAIS-R was administered to all partici-
range of mental retardation (69 to 55 points), and
pants. This instrument is composed of 11 subtests,
9% fell in the moderate range (54 to 40 points)
including 6 Verbal Scale and 5 Performance Scale
(American Psychiatric Association, 1994). Mean
subtests. Raw scores are obtained for each subtest,
IQs, subtest age-scaled scores, and raw scores are
from which scaled scores (M
10, SD
3) are
reported in Table 1. There were no differences, p
derived. Verbal (VIQ), Performance (PIQ), and
.05, between males and females on VIQ, PIQ,
Full-Scale (FSIQ) deviation IQs (M
100, SD
or FSIQs nor were there gender differences on any
15) are derived from the appropriate sums of
of the Verbal subtests or the Performance subtests.
scaled scores. The WAIS-R scaled scores are based
The participants’ WAIS-R VIQs were signifi-
on a reference group (age 20 to 34 years). How-
cantly higher than PIQs by an average of 5.5
ever, the WAIS-R manual also provides age-cor-
points, t(79)
7.72, p
.001. The significance
rected scaled score equivalents for each of the
of the VIQ–PIQ difference for each individual in
nine age groups in the standardization sample.
our sample was tested using the age appropriate
These age-corrected scaled scores are useful in
critical value of the VIQ–PIQ difference at the .05
Table 1. Mean Raw Scores and Age-Corrected Scaled Scores and Their Correlation With Age
Sum of scaled scores
IQ
WAIS-R
Mean
SD
r
Mean
SD
r
FSIQ
47.8
15.5
.09
67.4
8.3
.07
VIQ
28.1
9.2
.01
71.5
8.2
.01
PIQ
19.8
7.3
.18
66.0
8.0
.26*
Raw scores
Age-corrected scaled scores
Verbal subtests
Information
7.24
4.2
.06
4.55
2.2
.06
Digit Span
8.01
2.8
.10
4.94
1.9
.04
Vocabulary
19.20
11.4
.15
5.16
2.2
.03
Arithmetic
3.59
1.8
.03
3.73
1.6
.06
Comprehension
8.17
4.8
.08
4.64
2.1
.01
Similarities
11.45
4.8
.18
6.40
1.8
.13
Performance subtests
Picture Completion
7.51
3.4
.17
4.99
1.7
.08
Picture Arrangement
4.67
3.4
.22
5.17
2.1
.02
Block Design
6.29
5.9
.07
3.96
1.7
.15
Object Assembly
11.51
7.3
.03
3.55
2.3
.02
Digit Symbol
23.50
9.2
.33**
3.89
1.4
.01
*p
.05. **p
.01.
American Association on Mental Retardation
233

---

VOLUME 109, NUMBER 3: 231–236
MAY 2004
AMERICAN JOURNAL ON MENTAL RETARDATION
Age and IQ in Williams syndrome
Y. M. Searcy et al.
level of confidence as defined by Wechsler (1981).
The VIQ–PIQ difference was significant for 19
participants (23.8%), and for one individual, the
PIQ was significantly higher than the VIQ.
The relationship between age and WAIS-R
scores was assessed using Pearson correlations (see
Table 1). Increased age was related to lower raw
scores on the Digit Symbol subtest, r
.33, p
.01, with raw scores showing a slow and steady
decline after age 24, as occurs in the normative
sample. Although age was not related to VIQ or
FSIQ, increased age was related to higher PIQs, r
.26, p
.02. Longitudinal data from 4 of the
WAIS-R participants support the latter findings.
The WAIS-R had previously been administered to
these 4 individuals at a mean age of 32.9 years.
Upon re-testing an average of 9.2 years later (M
age
42.1), VIQs remained stable (Ms
75.8
and 75.2, respectively), but small gains occurred
in FSIQ (Ms
69.5 and 73.0, respectively), t(3)
12.12, p
.001, due primarily to increases in
PIQs (Ms
65.8 and 72.0, respectively), t(3)
5.64, p
.01, with all 4 participants gaining PIQ
points (4, 5, 7, and 9 points).
Figure 1 illustrates the developmental trajectory
of the sums of scaled scores for our Williams syn-
drome sample compared to that of the WAIS-R
standardization sample (Wechsler, 1981, p. 26). In
the standardization sample, the sum of scaled scores
of the Verbal subtests increased slightly and gradu-
Figure 1. The developmental trajectory of the
ally from age 16 through 34 and began to decline
mean sums of scaled scores earned by each age
after age 44, whereas the sum of scaled scores of the
group of our Williams syndrome sample com-
Performance subtests increased gradually through
pared to the mean sums of scaled scores earned
age 24 before beginning to decline after age 34.
by each relevant age group of the WAIS-R nor-
Note that although the developmental trajectory of
mative standardization sample (Wechsler, 1981, p.
the Verbal, Performance, and Full-Scale sum of
26).
scaled scores remains relatively stable with age in
Williams syndrome, PIQ in this Williams syndrome
sample is, nevertheless, positively correlated with
liams syndrome tend to be lower than average at
age. This phenomenon will be addressed below.
any age, changes in the abilities measured by the
various subtests of the WAIS-R occurred at a rate
similar to typically developing same-age peers.
Discussion
The stability of age-corrected scaled scores with
Although decline in IQs with aging has been
age further indicates that the pattern of strengths
reported for other neurodevelopmental disorders,
and weaknesses that are part of the phenotypic
our results indicate no such decline in adults with
variability in Williams syndrome is maintained
Williams syndrome from age 17 to 52 years. In-
through adulthood. Similarly, Howlin et al. (1998)
creases and decreases with age in WAIS-R subtest
reported stability of the characteristic Williams
raw scores of individuals with Williams syndrome
syndrome profile of cognitive strengths and weak-
occurred at rates comparable to the WAIS-R nor-
nesses from childhood (age 6 to 14) into early
mative sample, resulting in stability of age-cor-
adulthood (age 18 to 39).
rected scaled scores in Williams syndrome with
Although the stability of sums of scaled scores
age. Thus, although IQs of individuals with Wil-
with age in the present study resulted in stability
234
American Association on Mental Retardation

---

VOLUME 109, NUMBER 3: 231–236
MAY 2004
AMERICAN JOURNAL ON MENTAL RETARDATION
Age and IQ in Williams syndrome
Y. M. Searcy et al.
of VIQ and FSIQ with age, PIQ increased signif-
dinal WAIS-R scores. Upon re-testing with the
icantly with age. The positive relationship be-
WAIS-R after an average delay of 9.2 years, these
tween age and PIQ without concurrent improve-
4 individuals showed small gains in FSIQ, due
ment in raw scores on any of the Performance
primarily to gains of 4 to 9 points in PIQs, with
subtests, or increases in the scaled scores used to
no consistent increase in Performance subtest raw
calculate IQs, indicates that the increase in PIQ
scores.
in this sample is likely an artifact of how the PIQ
In sum, our results indicate that although IQs
is derived by the WAIS-R. For example, the mean
in adults with Williams syndrome were generally
PIQ of our Williams syndrome sample was 66. In
below average, subtest raw scores, and, thus, the
the WAIS-R standardization sample, from age 16
sums of scaled scores used to calculate IQs, nei-
to 54, an individual’s Performance Scale sum of
ther increased nor decreased differently than nor-
scaled scores can decline by up to 8 points while
mal with age. This stability of sums of scaled
maintaining a PIQ of 66. Because the Perfor-
scores resulted in VIQs and FSIQs that remained
mance Scale sum of scaled scores for our Williams
stable with age. Because PIQ increased with age
syndrome sample remained stable with age, the
despite a slight normal decline in Performance
mean PIQ of our sample actually increased with
subtest raw scores, we suggest that this increase in
age. In contrast, in order to maintain a VIQ of 72
PIQ with age may be a phenomenon unique to
over the same age range, one would need to
the WAIS-R. Therefore, one might expect that the
achieve an increase of up to 4 points in the sum
IQs obtained by an individual with Williams syn-
of the Verbal scaled scores.
drome as a young adult will generally remain rel-
Evaluation of IQs of individuals in a special
atively stable through their early 50s, as they do
population for the purpose of assessing the sta-
in the typically developing population. The find-
bility of intellectual functioning with aging and
ing of stability of intellectual functioning with age
comparison of their IQs to those of a normative
in Williams syndrome provides impetus for ad-
sample is informative only in determining wheth-
vocating continued enrichment and/or education-
er the rate of improvement or decline in the spe-
al opportunities throughout the life-span of indi-
cial population is comparable to that of the nor-
viduals with Williams syndrome.
mative sample. Our findings highlight the need to
consider changes in raw scores when evaluating
References
meaningful change in intellectual functioning
with age. By considering both raw and age-scaled
American Psychiatric Association. (1994). Diag-
scores, we differentiated age-related changes in
nostic and statistical manual of mental disorders
IQs that actually reflected the ability of the indi-
(4th ed.). Washington, DC: American Psy-
viduals taking the test from changes in IQs re-
chiatric Association.
sulting from the psychometric properties of the
American Academy of Pediatrics. (2001). Health
test.
care supervision for children with Williams
Because our sample was cross-sectional and
syndrome. Pediatrics, 107, 1192–1204.
included individuals from families who volun-
Avery, R. O., Slate J. R., & Chovan, W. (1989). A
teered to participate in an extended research pro-
longitudinal study of WISC-R and WAIS-R
tocol, care should be taken in generalizing the
scores with students who are educable mental-
findings. Nonetheless, our results are consistent
ly handicapped. Education and Training in Men-
with those of Mervis, Morris, Bertrand, and Rob-
tal Retardation, 24, 28–31.
inson (1999), who also found that Block Design
Bellugi, U., & St. George, M. (Eds.). (2000). Link-
and Digit Span ability scores (raw scores) on the
ing cognitive neuroscience and molecular ge-
Differential Ability Scales (Elliot, 1990) remained
netics: New perspectives from Williams syn-
stable with age in a cross-sectional sample of
drome [Special issue]. Journal of Cognitive
adults with Williams syndrome (18 to 46 years).
Neuroscience, 12(1).
Further, our results are consistent with those from
Carr, J. (1994). Long-term outcome for people
longitudinal studies indicating no evidence of de-
with Down’s syndrome. Journal of Child Psy-
cline in IQs in persons with Williams syndrome
chology and Psychiatry, 35, 425–439.
from childhood into early adulthood (e.g., How-
Dykens, E. M., Hodapp, R. M., Ort, S., Finucane,
lin et al., 1998), and with our concurrent finding
B., Shapiro, L. R., & Leckman, J. F. (1989).
for 4 individuals for whom we obtained longitu-
The trajectory of cognitive development in
American Association on Mental Retardation
235

---

VOLUME 109, NUMBER 3: 231–236
MAY 2004
AMERICAN JOURNAL ON MENTAL RETARDATION
Age and IQ in Williams syndrome
Y. M. Searcy et al.
males with fragile X syndrome. Journal of the
Sattler, Assessment of children (3rd ed.). San Di-
American Academy of Child and Adolescent Psy-
ego: Sattler Publisher.
chiatry, 28, 422–426.
Schaie, K. W. (1983). The Seattle Longitudinal
Elliot, C. D. (1990). Differential Ability Scales. San
Study: A twenty-one year exploration of psy-
Diego: Harcourt, Brace, Jovanovich.
chometric intelligence in adulthood. In K. W.
Fenner, M. E., Hewitt, K. E., & Torpy, D. M.
Schaie (Ed.), Longitudinal studies of adult psy-
(1987). Down’s syndrome: Intellectual and
chological development (pp. 64–135). New York:
behavioural functioning during adulthood.
Guilford Press.
Journal of Mental Deficiency Research, 31, 241–
Spitz, H. H. (1988). Inverse relationship between
246.
the WISC-R/WAIS-R score disparity and IQ
Hagerman, R. J., Schreiner, R. A., Kemper, M. B.,
level in the lower range of intelligence. Amer-
Wittenberger, M. D., Zahn, B., & Habicht, K.
ican Journal on Mental Retardation, 94, 376–
(1989). Longitudinal IQ changes in fragile X
378.
males. American Journal of Medical Genetics, 33,
Udwin, O., Davies, M., & Howlin, P. (1996). A
513–518.
longitudinal study of cognitive abilities and
Howlin, P., Davies, M., & Udwin, O. (1998). Cog-
educational attainment in Williams syn-
nitive functioning in adults with Williams
drome. Developmental Medicine and Child Neu-
syndrome. Journal of Child Psychology and Psy-
rology, 38, 1020–1029.
chiatry and Allied Disciplines, 39, 183–189.
Wechsler, D. (1974). Wechsler Intelligence Scale for
Kaufman, A. (1990). Assessing adolescent and adult
Children-Revised. New York: Psychological
intelligence. Boston: Allyn & Bacon.
Corp., Harcourt Brace Jovanovich.
Korenberg, J. K., Chen, X., Hirota, H., Lai, Z.,
Wechsler, D. (1981). Wechsler Adult Intelligence
Bellugi, U., Burian, R., & Matsuoka, R.
Scale-Revised. New York: Psychological Corp.,
(2000). Genome structure and cognitive map
Harcourt Brace Jovanovich.
of Williams syndrome. In U. Bellugi & M. St.
Received 5/8/02, accepted 9/26/03.
George (Guest Eds.), Linking cognitive neu-
Editor-in-charge: Elisabeth M. Dykens
roscience and molecular genetics: New per-
spectives from Williams syndrome [Special Is-
The second author is also affiliated with Alliant
sue]. Journal of Cognitive Neuroscience, 12(1),
International University, San Diego. Williams Syn-
89–107.
drome: Bridging Cognition and Gene is an ongoing
Kuck, J., Lincoln, A. J., & Heaton, R. (in press).
research project conducted at The Salk Institute
Age related changes in intellectual abilities
for Biological Studies Laboratory for Cognitive
among individuals with autistic disorder. Jour-
Neuroscience in collaboration with Cedars-Sinai
nal of Autism and Developmental Disorders.
Medical Center Laboratory for Human Molecular
Mervis, C. B., Morris, C. A., Bertrand, J., & Rob-
Genetics. The project is supported by grants
inson, B. F. (1999). Williams syndrome: Find-
awarded to Ursula Bellugi from the National In-
ings from an integrated program of research.
stitutes of Health (PO1 HD33113) as well as the
In H. Tager-Flusberg (Ed.), Neurodevelopmental
Oak Tree Foundation and the James S. Mc-
disorders (pp. 65–110). Cambridge: MIT Press.
Donnell Foundation (P50 NS22343). We thank
Pober, B. R., & Dykens, E. M. (1996). Williams
Teresa Doyle for helpful comments made on ear-
syndrome: An overview of medical, cognitive,
lier drafts of this paper. Special thanks go to the
and behavioral features. Child and Adolescent
Williams Syndrome Association and to the indi-
Psychiatric Clinics of North America, 5, 929–
viduals with Williams syndrome and their families
943.
who participated in this study. Some of these data
Rubin, H. H., Goldman J. J., & Rosenfeld, J. G.
were previously presented in Technical Report
(1985). A comparison of WISC-R and WAIS-R
2110, Institute for Neural Computation, Univer-
IQs in a mentally retarded residential popu-
sity of California, San Diego. Requests for reprints
lation. Psychology in the Schools, 22, 392–397.
should be sent to Yvonne M. Searcy, The Salk
Sattler, J. M. (1988). Assessment of children (3rd ed.).
Institute for Biological Studies, Laboratory for
San Diego: Sattler Publisher.
Cognitive Neuroscience, 10010 N. Torrey Pines
Sattler, J. M., & Ryan, J. J. (1988). Wechsler Adult
Rd., La Jolla, CA 92037. E-mail: searcy@lcn.
Intelligence Scale-Revised (WAIS-R). In J. M.
salk.edu.
236
American Association on Mental Retardation

---