Sensitivity and Specificity of the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) in Acute Care Settings

Background: ASD affects an estimated 1 in 68 children in the US (CDC, 2015) increasing substantially since the 1990s. Children with ASD often develop co-morbid psychiatric conditions (Simonoff et al., 2008), and often require acute psychiatric services (Mandell et al., 2008). Despite the need for specialized psychiatric services, this population is often hospitalized in general pediatric psychiatry units (Gabriels et al., 2012). These units may include assessment services, especially to guide disposition planning and service access (Mandell et al., 2005). Consequently, assessment measures designed for use in outpatient settings are used in acute care settings, at times with limited guidance from neurodevelopmental specialists. These tools may capture a unique snapshot of extreme behaviors related to a psychiatric crisis, calling into question the validity of the clinical findings, particularly for long term/developmental diagnoses.

Objectives: This study investigates the sensitivity and specificity of the ADOS-2 when administered in an acute psychiatric care setting.

Methods: Participants were selected from the Rhode Island Consortium for Autism Research and Treatment (RI-CART), a state-wide registry of individuals diagnosed with ASD or related neuro-developmental disorders. Selected RI-CART participants (n = 80; 70% male, Mean Age = 11.5, sd = 2.9) were enrolled in the registry while admitted to either a pediatric psychiatric inpatient unit (n = 11) or partial hospital program (n = 69); assessment was acquired during admission. Assessments included an ADOS-2 administered by a research-reliable examiner, and other cognitive (K-BIT 2), emotional/behavioral (SRS-2), and adaptive measures (VABS). Parents completed a demographics form. A confirmed diagnosis of ASD was indicated if participants medical charts contained an active diagnosis of ASD and ADOS-2 results were positive for ASD (n = 34). Additional data derived from the Autism and Developmental Disorders Inpatient Collaborative (ADDIRC; n = 736 with confirmed ASD diagnosis and positive ADOS-2; n = 178 with discordant clinical diagnosis and ADOS-2 scores) is being prepared and subsequent analysis will include: 1) more in-depth examination of demographics characteristics, 2) second sensitivity and specificity analysis with the full sample, 2) an examination of individual ADOS-2 items that reliably distinguish between ASD positive and ASD negative participants and, 3) a receiver operating characteristic (ROC) curve to examine how alternative algorithm cut-offs may increase the sensitivity and specificity of the ADOS-2 when used in acute care settings.

Results: Participants presented with numerous psychiatric comorbidities (mean = 2.7, range 1-5 psychiatric dx). Preliminary data include 51 Module 3 and 29 Module 4 ADOS-3 results. In this sample, the Module 3 algorithm produced a sensitivity of 36% and a specificity of 79%; Module 4 algorithm produced a sensitivity of 83% and a specificity of 71%. No differences in sensitivity and specificity were attributed to participants’ gender (p = 0.17).

Conclusions: Results from these analyses suggest that sensitivity and specificity of the ADOS-2 may be lower in acute care compared to previous findings in outpatient samples. This, combined with upcoming analysis examining alternative algorithm cut-offs, aim to provide clinical diagnostic guidelines to improve the accuracy of the ADOS-2 when used in acute care.