Developing and Validating a Computerized Adaptive Test to Measure Broad and Specific Factors of Internalizing in a Community Sample

Highly efficient assessments that better account for comorbidity between mood and anxiety disorders (internalizing) are required to identify individuals who are most at risk of psychopathology in the community. The current study examined the efficiency and validity associated with a multidimensional computerized adaptive test (CAT) to measure broad and specific levels of internalizing psychopathology. The sample comprised 3,175 respondents to an online survey. Items from five banks (generalized anxiety, depression, obsessive-compulsive disorder, panic disorder, social anxiety disorder) were jointly calibrated using a bifactor item response theory model. Simulations indicated that an adaptive algorithm could accurately (rs ≥ 0.90) estimate general internalizing and specific disorder scores using on average 44 items in comparison with the full 133-item bank (67% reduction in items). Scores on the CAT demonstrate convergent and divergent validity with previously validated short severity scales and could significantly differentiate cases of DSM-5 disorder. As such, the CAT validly measures both broad and specific constructs of internalizing disorders in a manner similar to the full item bank and a static brief form but with greater gains in efficiency and, therefore, a reduced degree of respondent burden.

Clinical reappraisal of the Composite International Diagnostic Interview Screening Scales (CIDI-SC) in the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS)

A clinical reappraisal study was carried out in conjunction with the Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS) All-Army Study (AAS) to evaluate concordance of the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) diagnoses based on the Composite International Diagnostic Interview Screening Scales (CIDI-SC) and post-traumatic stress disorder (PTSD) checklist (PCL) with diagnoses based on independent clinical reappraisal interviews (Structured Clinical Interview for DSM-IV [SCID]). Diagnoses included: lifetime mania/hypomania, panic disorder, and intermittent explosive disorder; six-month adult attention-deficit/hyperactivity disorder; and 30-day major depressive episode, generalized anxiety disorder, PTSD, and substance (alcohol or drug) use disorder (abuse or dependence). The sample (n = 460) was weighted for over-sampling CIDI-SC/PCL screened positives. Diagnostic thresholds were set to equalize false positives and false negatives. Good individual-level concordance was found between CIDI-SC/PCL and SCID diagnoses at these thresholds (area under curve [AUC] = 0.69-0.79). AUC was considerably higher for continuous than dichotomous screening scale scores (AUC = 0.80-0.90), arguing for substantive analyses using not only dichotomous case designations but also continuous measures of predicted probabilities of clinical diagnoses.