Methods
Participants
One hundred and five adolescents aged 11–16 years old and resident in the State of Sao Paulo, Brazil, were recruited. The sample comprised four groups: (1) 22 adolescents with WBS, (2) 22 adolescents with DS, (3) 37 adolescents with ASD, and (4) 24 healthy controls (HC).
Adolescents with WBS were recruited from the Brazilian Association of Williams–Beuren Syndrome and all individuals presented diagnostic confirmation by cytogenetic analysis by Fluorescence in situ Hybridization (FISH). Of the 28 individuals with WBS registered with the Brazilian Association of Williams–Beuren Syndrome (aged 11–16 years), 24 individuals agreed to participate in this study. One participant with WBS was excluded due to unfinished questionnaires, and socioeconomic class data were missing for another individual.
Individuals with DS were recruited from the Association of Parents and Friends of Exceptional Children, a non-profit organization that provides social services to people with intellectual disability. The diagnosis of DS was confirmed by examining the karyotype in all individuals. One case was excluded due to comorbidity with ASD and another due to missing data.
Adolescents with ASD were recruited from the specialized clinic in ASD of the Child and Adolescent Psychiatric Unit at the Department of Psychiatry of the Federal University of Sao Paulo (UNIFESP) and from two psychosocial care centers for children and adolescents. Psychosocial care centers for children and adolescents are the main centers for the diagnosis and treatment of children and adolescents with ASD in Brazil. Specialized and experienced professionals in the field carried out the diagnosis of ASD in accordance with the DSM-IV diagnostic criteria, as these assessments took place between 2011 and 2013. One participant with ASD was excluded because of missing data.
The HC group comprised brothers, cousins and friends of the participants with WBS and this group were recruited from events organized by the Brazilian Association of Williams–Beuren Syndrome. The parents or legal guardians of all participants signed informed consent forms, as did the adolescents. The study was approved by UNIFESP’s Research Ethics Committee.
Instruments
We used the Vineland Adaptive Behavior Scales (VABS) to measure adaptive behavior. VABS evaluates the ability of individuals to cope with environmental changes, learn everyday skills, and demonstrate independence [
2]. The scale is based on a structured interview, in which adaptive behavior information is obtained from a significant caregiver. Completion time was approximately 25–90 min. VABS is organized in a structure with three main domains: communication, daily living, and socialization. The raw scores obtained from the domains were weighted to adjust for chronological age, according to the VABS manual, and standardized to obtain a common metric [
2]. VABS results can determine strengths and weaknesses in specific adaptive behavior areas, and the scale has extensive representative normative data as well as strong psychometric properties [
2].
To estimate IQ we used the Wechsler Intelligence Scale for Children–Third Edition (WISC-III). This instrument evaluates children between 6 and 16 years of age [
40]. We estimated IQ by using the weighted sum of the subtests Cubes and Vocabulary [
41].
The Critério de Classificação Econômica Brasil (
Brazil Economic Classification Criteria) developed by the Associação Brasileira de Empresas de Pesquisa [ABEP] (2011) [
42] was used to measure socioeconomic status. The classification estimates the income of Brazilian families living in urban areas by evaluating their consumption of durable goods and also assesses their educational level. The socioeconomic classes range from A (highest income) to E (lowest income) [
42].
Statistical analysis
To describe the sample’s characteristics, the frequency of participants’ responses for each categorical variable was calculated. Continuous variables were described based on measures of central tendency (mean) and dispersion (standard deviation).
Mann–Whitney Wilcoxon post hoc test was used to adjust p values when bivariate comparisons of continuous variables were performed, as these variables had a non-normal distribution.
When a comparison between two diagnostic groups resulted in a statistically significant difference in any of the clinical scales, we ran a multivariate linear logistic regression model to control for the effect of demographic characteristics and IQ as potential confounders.
These comparisons resulted in three multivariate linear logistic regression models, in which the following pairs of diagnostic groups were compared: (1) WBS vs. DS, (2) WBS vs. ASD, and (3) DS vs. ADS. We repeated these three models for each of the following outcomes: (1) VABS total score, (2) VABS communication domain, (3) VABS socialization domain, and (4) VABS daily living domain.
Finally, we tested for interactions between diagnoses and socioeconomic classes through linear logistic regression models, controlling for IQ. The Vineland dimension (communication, socialization, and daily living) was defined as a dependent variable, whereas diagnosis, in interaction with economic classes, was entered as an independent variable.
In the linear logistic regression models, HC from the A/B classes were regarded as the reference category. For each diagnosis, when the difference between the A/B and C/D classes (e.g., WBS A/B classes vs. WBS C/B classes) was higher than 20%, we considered that this indicated an interaction between diagnosis and economic class, meaning that differences in performance between A/B and C/D classes were clinically significant. We arbitrarily established 20% as a parameter to define the interaction between diagnosis and economic classes, as this can be considered a clinically significant difference in adaptive behavior performance. Among many statisticians and epidemiologists, it is acceptable to set an “arbitrary” parameter such as this when the literature does not provide established parameters [
43,
44].
The level of significance was set at
p < .05. Effect sizes (Cohen’s
d) were calculated as the difference between means divided by a pooled standard deviation (difference between HC and WBS, DS, and ASD groups). An effect size of ≤.2 indicates a small change, between .2 and .5 a moderate change, and an effect size ≥.8 a large clinical change [
45].
Discussion
The findings of this study suggest that there are distinct adaptive behavior profiles in adolescents with WBS, DS, and ASD. Individuals with WBS had better adaptive behavior skills than individuals with DS, especially in communication, as well as better socialization skills than individuals with ASD. Furthermore, the DS group performed better than the ASD group in the socialization domain. These results remained significant after adjusting for IQ, since the groups were not matched for this variable when selected.
In spite of the limitations in the socialization and communication domains in participants with WBS when compared to the HC group, they still performed better in socialization compared to individuals with ASD, and in communication when compared to individuals with DS. This outcome is consistent with the literature with respect to adolescents and adults with WBS who showed satisfactory performance in social interaction [
17,
18] and communication skills and poor performance in the remaining adaptive behavior domains [
21]. One study showed that the performance of adolescents with WBS was better than that of adolescents with DS in all VABS domains [
34], which is in agreement with the results presented in this study. It is reasonable to suggest that although individuals with WBS within the age range of this study, 7.67–46.41 years of age have an interest in interpersonal relationships (socialization), they still have difficulty performing this skill to the same level as HC [
46].
Thus, we suggest that the adaptive behavior profile for WBS, in terms of strongest to weakest performance in the three domains, would be in the order: socialization, communication, daily living skills. Our results agree with the personality description of individuals with WBS as “hypersocial” [
47].
Our results regarding the adaptive behavior profile of the DS group corroborate other studies that showed discrete differences between domains but a better performance in the socialization domain [
34,
48]. Moreover, the DS group showed better performance in the socialization domain than the ASD group, but a worse performance in the communication domain when compared to individuals with WBS. Our results therefore agree with the pattern suggested by other studies that found relatively stronger social skills but weaker communication in individuals with DS between ages 1–11.5 years [
9,
24]. DS acquire their adaptive skills ceiling scores at about the age of 12 years, furthermore, presented lower performance than HC children in 65% approximately [
49]. A review study identified that the main difficulties in the DS cognitive profile are in communication, such as: auditory-verbal processing and short-term memory, expressive language, grammar, and pronunciation [
7]. Therefore, the adaptive behavior profile for DS in this study would be, in terms of strongest to weakest domains: socialization, daily living skills, communication.
Previous adaptive behavior findings in children, adolescents and young adults with ASD showed significant deficits in the socialization domain, compared to a relative ability in the daily living skills and communication domains [
30,
50,
51]. Thus, some results described in the literature corroborate with the clinical definition of ASD, which is characterized by impairment in social development. Accordingly, in our study the WBS and DS groups showed better performance in the socialization domain than the ASD group, after controlling for IQ. The adaptive behavior profile for ASD in this study would be, in terms of strongest to weakest domains: Daily Living Skills > Communication > Socialization.
Subjects with WBS, DS, and ASD presented lower adaptive behavior performance when compared to the HC group. These difficulties in functional performance in all assessed diagnostic groups in this study are likely associated with limitations derived from the disorder itself. For some adaptive behavior difficulties, IQ might play an important role in adolescents with WBS, DS, and ASD, as previous studies have observed a positive correlation between adaptive behavior and IQ in ASD [
31,
52‐
54], DS [
25], and WBS [
19]. Therefore, the main differences in terms of performance in adaptive behavior between the groups in this study can only be observed when taking into account IQ. For this reason, we believe it is essential describe the influence of intellectual disability in adaptive behavior performance.
Finally, the result of the analysis identified possible associations between adaptive behavior profiles and socioeconomic class, suggesting that being in the C/D classes might be related to increased adaptive behavior impairment. This result corroborates the theory that parental behavior is adapted by socioeconomic and demographic conditions with a subsequent effect on their children [
39], with previous studies demonstrating a strong relationship between socioeconomic status and expressive communication during preschool years through third grade [
37,
38]. A study evaluating the risk and protective factors of parental well-being compared parents of children with intellectual disability with control children and found that differences in economic hardship and self-rated health were the strongest predictors for well-being [
55]. Thus, Olsson and Hwang [
55] confirmed the importance of the relationship between socioeconomic status and the well-being of parents of children with intellectual disability.
As there are no parameters in the literature that define a significant difference in the Vineland scores, a difference of 20% was chosen as being clinically significant based on our clinical experience. The proposed 20% can be a good starting point to demonstrate the relationship between socioeconomic class and the condition, and the effect on adaptive behavior. Nonetheless, it would be worthwhile conducting more specific studies to test this parameter, which, if confirmed, possibly could indicate how interventions that improve families’ socioeconomic conditions (i.e., the income and education level) most likely could help to alleviate the effects of genetic syndromes on adaptive behavior.
When interpreting the results of our study, some limitations should be considered. First, the sample size was defined based on the available number of adolescents with WBS, which is a rare syndrome. Despite this limitation, our statistical models were able to demonstrate differences between the groups that were consistent with our hypotheses and with the literature. Second, only two subscales of the WISC-III were used to estimate IQ. Even though it has been shown that this approach provides scores that are similar to those obtained through the full WISC-III, it does not include other dimensions of intellectual function that may be related with adaptive behavior. The third limitation was not having collected information such as type of school (mainstream or specialist school), other co-morbid or medical conditions and maternal education. The final limitation was the gender differences between the ASD group and the other groups. However, our multivariate analyses showed that gender did not play a significant role as a potential predictor of adaptive behavior in our sample, although females with ASD tend to be somewhat more functional than males in relation to adaptive behavior [
56]. Future studies should explore in more depth whether adaptive behavior profiles also differ between male and female individuals with ASD.
Authors’ contributions
DCCG performed the data collection and wrote the manuscript; CSC contributed to the design of the research project, data collection and manuscript revision; RW performed the statistical analyzes in the study and revised the manuscript; KAM assisted in the analysis of the results and in their description; JAP oversaw the research project and reviewed the manuscript. All authors read and approved the final manuscript.