Introduction

Both attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) have strong heritable components. Although their underlying etiologies are yet to be fully elucidated, evidence from twin and family studies indicate shared heritability (1,2). Individuals with ADHD may manifest different forms of ASD, from a fully developed syndromic form to a milder form of ASD symptomatology that involves the exhibition of autistic traits (ATs). However, in contrast with the well-developed literature addressing ASD, much less is known about the morbidity and dysfunction associated with ATs for individuals with and without ADHD.

Recent studies have shown that symptoms of autism or ATs appear in 20% to 30% of children with ADHD (3-5). Children with ADHD and ATs appear to be more impaired and dysfunctional than children with ADHD and no ATs. With these literature findings in mind, we assessed the prevalence and correlates of ATs in youth with and without ADHD, with a diagnosis of autism being exclusionary (6). ATs were operationalized with the use of a profile from the Child Behavior Checklist (CBCL) by using extreme values from the sums of the Withdrawn, Social, and Thought Problems T-scores (7).

In the prior study, a positive AT profile was significantly more prevalent among children with ADHD as compared with control children (18% vs. 0.87%; p < .01). Children with ADHD with positive AT profiles were significantly more impaired than other children with ADHD and control children with regard to psychopathology and interpersonal, school, family, and cognitive functioning. However, because our results and the results of previous reports examined ATs in referred samples of children with ADHD (3-5), it remains unknown whether these findings can be generalized to non-referred samples of children with and without ADHD. Studies of such samples would allow clinicians to identify children at high risk for adverse outcomes in multiple domains of functioning. This information could further lead to the development of appropriate intervention strategies aimed at mitigating the adverse outcomes associated with ATs.

The aim of the current study was to examine both the prevalence and correlates of ATs in non-referred youth with and without ADHD. To this end, we used data from an existing large-scale sample of non-referred siblings of probands with and without ADHD. On the basis of the findings in probands, we hypothesized that ATs would be identifiable in non-referred siblings and that the presence of ATs would be associated with higher levels of morbidity and dysfunction.

Methods

Assessment Procedures

We used an empirically derived profile from the CBCL to define ATs (CBCL-AT) by using a cutoff of 195 or more from the combined T-scores of the Withdrawn, Social Problems, and Thought Problems subscales. We previously reported that this profile correctly classified 78% of all subjects with ASD in a psychiatrically referred sample of youth with and without ASD (7). CBCL emotional dysregulation profiles were created from the combined T-scores of the Anxiety/Depression, Aggression, and Attention subscales. Deficient emotional self-regulation was defined as a combined T-score of 180 or more but less than 210; severe emotional dysregulation was defined as a combined T-score of 210 or more.

Psychiatric assessments relied on the Kiddie Schedule for Affective Disorders and Schizophrenia for School-Age Children, Epidemiological Version (K-SADS-E) (10,11). Indirect interviews were conducted with each subject’s parent or guardian, usually the mother. Direct interviews were conducted with subjects who were at least 12 years old. After combining data from direct and indirect interviews, a diagnosis was considered positive if it was endorsed in either interview.

Interviews were administered by highly trained and closely supervised raters with bachelor’s or master’s degrees in psychology or related fields. Raters were blinded to the referral source and the proband’s diagnostic status (i.e., ADHD vs. control). On the basis of 500 assessments obtained via interviews of children and adults, the median κ coefficient of agreement between a rater and an experienced clinician was 0.98.

The parent or guardian also provided information about pregnancy and delivery history and the child’s infancy. Socioeconomic status (SES) was established using categories delineated by Hollingshead (12). The Global Assessment of Functioning (GAF) scale was used to measure overall adaptive functioning (13,14).

Interpersonal and psychosocial functioning was assessed with the use of the Social Adjustment Inventory for Children and Adolescents (SAICA) (15). To better gauge social difficulties, we computed an index of social dysfunction that we previously called Social Disability. This index was computed with the use of methodology recommended by Reynolds (16); it was previously implemented by our team to assess learning disabilities on the basis of the discrepancy between the expected SAICA scaled score (derived from the estimated full-scale intelligence quotient) and the actual SAICA scaled score (17).

To evaluate school functioning, three indices of school difficulties were used: placement in special classes, extra tutoring, and repeated grades as reported by the parent or guardian.

Intellectual functioning was assessed through the Vocabulary, Block Design, Digit Span, Digit Symbol, Digit Coding, and Arithmetic subtests of the Wechsler Intelligence Scale for Children–Revised (WISC-R) (18). By using procedures suggested by Sattler (19), we estimated the full-scale intelligence quotient from the Block Design and Vocabulary subtests of the WISC-R using age-corrected scaled scores. We computed the Freedom from Distractibility intelligence quotient using the Digit Span, Digit Coding, and oral Arithmetic subscales of the WISC-R. Reading and arithmetic achievement were assessed using subtests of the Wide Range Achievement Test–Revised (WRAT-R) (20).

FIGURE 1.

PRISMA diagram. Siblings were excluded if they were not 6 – 18 years old or if they were missing CBCL data. Our final sample was restricted to the siblings of ADHD probands because only 3 siblings of control probands had ATs

Statistical Analyses

All analyses were performed using regression models with robust standard errors to account for the non-independence of siblings. Differences in sociodemographic characteristics were assessed with the use of linear regression for continuous outcomes and logistic regression for binary outcomes. We controlled for any demographic confounder that reached significance at the 0.05 alpha level. Rates of ATs in siblings of probands with ADHD and control probands were analyzed using logistic regression. Analyses for psychiatric disorders, emotional dysregulation, school functioning, social disability, and perinatal complications were performed using exact logistic regression with permutation testing to deal with the issues of small expected numbers and non-independence. Analyses of the CBCL, GAF, SAICA, WISC-R, and WRAT-R items were performed with the use of linear regression. All tests were two-tailed, with the alpha set at 0.05. We calculated all statistics using Stata software version 14.0 (StataCorp, College Station, TX).

Results

Although there were 712 siblings in total, 221 siblings were excluded because they did not meet the age criteria of 6 to 18 years or because they were missing CBCL information (Figure 1). Thus, our final sample included 257 siblings of probands with ADHD and 234 siblings of control probands. Siblings had mean ± standard deviation ages of 10.7 ± 3.2 years and SES scores of 1.8 ± 0.9. Of the 230 subjects who reported their race, 94% were Caucasian.

FIGURE 2.

Rate of ATs in siblings of Control probands (n = 234), siblings of ADHD probands without ADHD (n = 201), and siblings of ADHD probands with ADHD (n = 56). ^a Compared to siblings of Control probands. ^b Compared to siblings of ADHD probands without ADHD. *p < 0.05, **p < 0.005, ***p < 0.001

As shown in Figure 2, rates of ATs were lowest among the siblings of control probands, intermediate among siblings of probands with ADHD without ADHD, and highest among siblings of probands with ADHD with ADHD. Siblings of probands with ADHD had significantly higher rates of ATs as compared with siblings of control probands (6% vs. 1%; χ² = 5.92; p = .02). Because only three siblings of control probands had positive AT profiles, analyses were limited to siblings of probands with ADHD with ATs (n = 15) and without ATs (n = 242).

Sociodemographic Characteristics

The siblings with ATs were of similar age and sex as the siblings without ATs but were of a more disadvantaged SES status. Siblings with ATs scored an average of 2.3 on the Hollingshead measure of SES as compared with a score of 1.9 for siblings without ATs (Table 1). Therefore, all subsequent analyses corrected for SES.

Patterns of Psychiatric Comorbidity and Psychopathology

The average number of psychiatric disorders was significantly higher for siblings with ATs as compared with siblings without ATs (t₍₁₇₄₎ = 3.68; p < .001; Figure 3, A). As compared with siblings without ATs, siblings with ATs had a significantly higher prevalence of ADHD, other disruptive behavior disorders, mood disorders, multiple (two or more) anxiety disorders, language disorders, and elimination disorders (p < .005 for all; see Figure 3, B). Likewise, individual and composite scores on all CBCL clinical scales as well as the two CBCL emotional dysregulation profiles were significantly more impaired among siblings with ATs as compared with those without ATs (p ≤ .001 for all; see Figure 3, C and D).

Psychosocial Functioning

As compared with siblings without ATs, siblings with ATs had significantly more impaired GAF scores (t₍₁₇₄₎ = -4.75; p < .001; Figure 4, A), significantly more impaired scores on the CBCL social and school competence scales (p < .001 for both; see Figure 4, B), significantly more impaired SAICA scaled scores on 7 of the 10 subscales (p < .05 for all; see Figure 4, C), and a significantly higher prevalence of social disability (Exact; p = .001; see Figure 4, D).

School Functioning

Siblings with ATs were significantly more likely to receive tutoring (χ² = 10.69; p = .001) and to be placed in special classes (Exact; p < .001) as compared with siblings without ATs (Figure 5, A).

Neuropsychological Functioning

Siblings with ATs had lower composite cognitive and achievement WISC-R and WRAT-R scores than siblings without ATs (see Figure 5, B) as well as lower scores on the five WISC-R subscales (see Figure 5, C). However, these differences failed to reach our a priori threshold for statistical significance.

FIGURE 3.

Rates of psychopathology in sibling and without ATs. A. Average number of psychiatric disorder. B. Comorbid disorder. C. Clinical CBCL D. CBCL emotional dysregulation profiles. ^a Compare to siblings without ATs. *p < 0.05, **p < 0.005, ***p < 0.001

FIGURE 4.

Social functioning in siblings with and without ATs. A. GAF score. B. CBCL social function ing scale. C. SAICA individual item scores. D. Social disability. ^a Compared to siblings without ATs. *p<0.05, **p<0.005, ***p<0.001

FIGURE 5.

School functioning in siblings with and without ATs. ^a Compared to siblings without ATs. *p<0.05, **p<0.005, ***p<0.001

FIGURE 6.

Neuropsychological functioning in siblings. A. Composite cognitive and achievement scores for WISC-R and WRAT. B. WISC-R subscales

Perinatal Complications

Rates of all but two pregnancy and infancy characteristics were significantly higher among siblings with ATs as compared with siblings without ATs (p < .005 for all), and the rate of accidents during pregnancy was significantly lower among siblings with ATs (Exact; p = .04; Table 2).

Discussion

Our findings reveal that ATs can be identified in a sizeable number of non-referred children, particularly in those with ADHD. The results also show that the presence of ATs heralds a significantly more compromised clinical presentation that is characterized by higher rates of psychopathological, neuropsychological, and interpersonal deficits as compared with children without these traits. These results in non-referred youth are highly consistent with the findings of previous reports (3-5) and with our findings from probands with ADHD (21). Thus, these results provide further support for the clinical relevance of ATs, irrespective of referral status.

Non-referred youth with ATs exhibited significantly higher rates of comorbid psychopathology as compared with those without ATs as expressed by the significantly higher mean number of comorbid psychiatric disorders; the significantly higher rates of disruptive behavior, mood, anxiety, language, and elimination disorders; the marked impairments noted in all individual and composite CBCL clinical scales; and higher rates of emotional dysregulation. These findings are highly consistent with those previously reported for referred probands with ADHD with ATs and therefore stress the heavy burden of psychopathology associated with ATs in both referred and non-referred youth. These findings are also in line with the prevailing literature, which highlights higher-than-expected rates of major psychiatric disorders among youth with ASD (22-24).

Also consistent with previously reported findings in probands with ADHD (6) is that non-referred siblings with ATs experienced greater social and interpersonal deficits as manifested by significantly worse GAF scores, significantly more impaired CBCL social and school competence scores, significantly more impaired SAICA scores, and higher rates of social disability as compared with siblings without ATs. Taken together, these findings provide further evidence that ATs severely affect the social functioning of afflicted children, irrespective of referral status.

Although the results did not reach our a priori threshold for statistical significance, we found that non-referred siblings with ATs had similarly lower full-scale intelligence quotients, Freedom from Distractibility intelligence quotients, and Block Design and Digit Symbol WISC-R scores as well as lower WRAT-R reading and arithmetic scores. These findings, which mirror what we reported for probands with ADHD, suggest that neurocognitive functioning is compromised in the presence of ATs.

Also in line with the findings reported for probands with ADHD is that non-referred children with ATs had higher rates of most pregnancy and infancy complications as compared with those of other children without ATs. These findings further support the hypothesis that perinatal complications alone—or in combination with genetic risk factors—could account for the development of ATs in some children.

Extending findings related to ATs to non-referred children has important clinical and scientific implications. Documenting the broad and severe spectrum of morbidity and dysfunction associated with ATs in non-referred youth supports the clinical relevance of ATs independent of referral status. Considering how easy it is to screen for ATs with the use of the CBCL, clinicians in the community have access to a rapid, inexpensive, and informative tool to screen children in clinical practice. Significant morbidity and dysfunction are associated with ATs, so the ability to identify children with ATs may facilitate the development of appropriate interventions that target afflicted children. This knowledge will help to inform future research aimed at identifying distinct genetic, neural, and therapeutic targets related to ATs.

The present study contains several strengths. Most notably, it relies on a large number of non-referred youth and an operationalized definition of ATs that is based on a unique profile of the empirically derived CBCL for the identification of ATs, which can be easily used in the clinical setting.

However, our findings need to be viewed in light of certain limitations. Despite the strength of the findings, our sample of non-referred siblings with ATs was relatively small. Although ATs were more strongly associated with ADHD, the sample size was too small to allow for a separate analysis limited to siblings with ADHD. Because we lacked an adequate sample of siblings of control probands with ATs, the analysis was limited to siblings of probands with ADHD. Although autism was excluded in the probands, it was not systematically ruled out in the siblings, thereby allowing for the possibility that some of the children may have had undiagnosed ASD. Although this analysis was limited to the siblings of probands with ADHD with and without ATs, we have previously documented the high discriminant validity of the CBCL-AT profile for the identification of ASD in a psychiatrically referred population of youth. (5) It is also notable that the presence of ATs in the siblings of children with ADHD was associated with lower SES, even though the analysis was controlled for social class. The present findings are in line with our previous report of poorer SES being more highly associated with children with ADHD with ATs than without ATs (6). Our findings are consistent with those of the prevailing literature, which suggest a higher association of ASD with low SES (25). Furthermore, we found a lower rate of accidents during pregnancy among siblings with ATs. This was an unexpected finding, so additional studies are needed to verify this result. Finally, our sample was largely Caucasian, but our findings may not generalize to other ethnic groups.

Despite these considerations, our work shows that ATs can be identified in a sizeable minority of non-referred children and that these children are at high risk for significant morbidity and disability. More work is needed to replicate these findings and to further examine their prognostic usefulness.

Clinical Significance

The current findings suggest that elevated scores on the CBCL-AT subscale may indicate a need to clinically assess a child for ASD and ADHD; mood, anxiety, and disruptive behavior disorders; emotional dysregulation; and impaired social and school functioning.