Background: Effective cognitive screening instruments are needed to differentially assess and manage milder forms of HIV-associated neurocognitive disorders (HAND). We assessed concurrent validity of four screening tests against the gold standard for HAND diagnosis in people living with HIV.
Methods: 220 adults (mean age: 51 years; mean education: 14 years; 86% men) attending one HIV outpatient clinic in Toronto were included in our analyses. Four neurocognitive screening tests: Cogstate Brief Battery (CBB), HIV Dementia Scale (HDS), Computer Assessment of Memory and Cognitive Impairment (CAMCI), and Montreal Cognitive Assessment (MoCA) were administered. Impairment was defined as raw score of >10 (HDS), ≥30 percentile (CAMCI), a score of >26 (MoCA), and impairment in two or more domains (CBB). Participants completed comprehensive neuropsychological battery assessing processing speed, attention/working memory, learning/memory, and executive functions. Clinical HAND diagnosis was made according to Antinori (2007) criteria independent of screening test scores. Validity of tests was assessed using sensitivity, specificity, and area under the curve (AUC) estimates against the gold standard.
Results: 129 participants (59%) had a clinical diagnosis of HAND (ANI=20; MND=94; HAD=15). Sensitivity estimates were: 72% (MoCA), 61% (CBB), 42% (HDS), and 31% (CAMCI). Specificity estimates were: 98% (CAMCI), 95% (HDS), 82% (CBB), and 73% (MoCA). AUC estimates were 0.723 (MocA), 0.714 (CBB), 0.682 (HDS), and 0.644 (CAMCI). Combining of any two screening tests (test positive by either one or both tests) resulted in modest classification accuracy improvements (AUC ranges: 0.736-0.758). All four screening tests were better at detecting symptomatic HAND (10%-32% higher AUC) compared to non-symptomatic HAND.
Conclusions: Our results suggest that the MoCA and CBB screening tests have only modest global classification accuracy for assessing mild HAND in people with HIV. Work is underway in our laboratory to determine the clinical utility and generalizability of newer instruments which may have better diagnostic accuracy.

Download the e-Poster