A comparison of the sensitivity, stability, and reliability of three diagnostic schemes for HIV-associated neurocognitive disorders

HIV-associated neurocognitive disorders (HAND) occur in approximately 50% of HIV-infected individuals, yet available diagnostic criteria yield varying prevalence rates. This study examined the frequency, reliability, and sensitivity to everyday functioning problems of three HAND diagnostic criteria (DSM-5, Frascati, Gisslén). Participants included 361 adults with HIV disease and 199 seronegative adults. Neurocognitive status as defined by each of the three diagnostic systems was determined via a comprehensive neuropsychological battery. Everyday functioning was evaluated through self-report and clinician ratings. Results of logistic regressions revealed an association of HIV serostatus with Frascati-defined neurocognitive impairment ( p = .027, OR = 1.7[1.1, 2.7]), but not DSM-5 or Gisslén-defined criteria ( p s > .05). Frascati and DSM-5 criteria demonstrated agreement on 71% of observations, Frascati and Gisslén showed agreement on 80%, and DSM-5 and Gisslén criteria showed agreement on 46%, though reliability across the three criteria was poor. Only Frascati-defined neurocognitive impairment significantly predicted everyday functioning problems ( p = .002, OR = 2.3[1.4, 3.8]). However, when both neurocognitive and complaint criteria were considered, the DSM-5 guidelines demonstrated significant relationships to everyday functioning, serostatus, and also increased reliability overtime compared to neurocognitive criteria alone (all p s < .05). A subset ( n = 118) of the HIV+ group was assessed again after 14.0 (2.2) months. DSM-5 criteria evidenced significantly higher rates of incident neurocognitive disorder compared to both Frascati ( p = .003) and Gisslén ( p = .021) guidelines, while there were fewer remitting neurocognitive disorder diagnoses when Gisslén criteria were applied to the study sample compared to Frascati ( p = .04). Future studies should aim to identify gold standard biological markers (e.g., neuropathology) and clinical outcomes associated with specific diagnostic criteria.