Increased Psychiatric Complexity of Autism Spectrum Disorder: Explaining Diagnostic Inconsistencies

Background: Recent studies have indicated that the Autism Diagnostic Observation Schedule (ADOS) has reduced specificity in samples with a broad range of developmental and behavioral disorders (Molloy et al. 2011), and a pilot study found that “false-positive” ADOS results may be attributable to symptoms of co-morbid psychiatric conditions (Stadnick et al. 2015). However, literature has yet to appropriately quantify phenotypic characteristics of individuals with diagnostic inconsistencies in ADOS and community diagnosis.

Objectives: To identify whether clinical characteristics differ in individuals with discordance between community diagnosis and ADOS as compared to individuals with concordance in these same measures.

Methods:  The sample was comprised of 762 individuals (ages 3 – 18; M_age = 13.27 years, SD = 9.45, 78.3% male) selected from a state-wide ASD registry. Participants entered the registry either with an existing diagnosis or due to a concern of an ASD. All participants were administered the ADOS-2 upon enrollment. Participants were grouped into four diagnostic categories depending on status of community diagnosis and ADOS-2 result: (1) Community ASD diagnosis and a positive ADOS-2 (N=533); (2) Community diagnosis and a negative ADOS-2 (N=54); (3) No community diagnosis and a positive ADOS-2 (N=109); and (4) No community diagnosis and a negative ADOS-2 (N=66). The presence of psychiatric diagnoses and the reports of current medications were obtained through caregiver-completed questionnaires. The number of different psychotropic medications were summed to create four dichotomous variables representing antidepressants, Attention-deficit/hyperactivity disorder (ADHD) medications, antipsychotics and mood stabilizers. Autism severity was measured by standardized scores available through a caregiver-completed Social Responsiveness Scale, Second Edition (SRS-2).

Results: A series of logistic regressions of co-morbid diagnoses with age and gender as covariates revealed that the discordant subgroups were more likely to have depression (χ² = 61.47, p < .001), anxiety (χ² = 51.93, p < .001), ADHD (χ² = 18.70, p = .002) and oppositional defiant disorder (χ² = 13.97, p = .016). Chi-square analyses revealed a significant difference between diagnostic groups and the likelihood of taking antipsychotic (p = .031) and ADHD (p = .020) medication. However, these effects did not survive follow-up logistic regression analyses controlling for age and gender. Further, there was a significant difference between diagnostic groups on SRS-2 severity (F(3,659) = 5.34; p = .001). Participants with a positive ADOS-2 result and community diagnosis scored higher on the SRS-2 compared to those with no ASD diagnosis (p= .003). Results support the sensitivity of the SRS-2 to measure ASD symptom severity in a subgroup more likely to have a true positive diagnosis.

Conclusions: Children presenting to an ASD registry with inconsistencies between community diagnosis and ADOS-2 findings had increased rates of caregiver-reported psychiatric diagnoses and psychotropic medications as compared to those whose ADOS-2 result was concordant with community diagnosis of ASD. The results concur with prior reports of reduced specificity of the ADOS-2 in the presence of co-occurring psychiatric symptoms. These findings are both empirically and clinically relevant, confirming the need for cautious interpretation of positive ADOS-2 results in the presence of psychiatric complexity.