Approximately half of people living with HIV (PWH) experience HIV-associated neurocognitive disorders (HAND), yet HAND often goes undiagnosed. There is an ongoing need to find efficient, cost-effective ways to screen for HAND and monitor its progression in order to intervene earlier in its course and more effectively treat it. Prior studies that analyzed brief HAND screening tools have demonstrated that certain cognitive test pairs are sensitive to HAND cross-sectionally and outperform other screening tools such as the HIV Dementia Scale (HDS). However, few studies have examined optimal tests for longitudinal screening. This study aims to identify the best cognitive test pairs for detecting cognitive decline longitudinally.

Participants were HIV+ adults (N=132; ages 25-68; 59% men; 92% Black) from the Temple/Drexel Comprehensive NeuroHIV Center cohort. Participants were currently well treated (98% on cART, 92% with undetectable viral load, and mean current CD4 count=686). They completed comprehensive neurocognitive assessments longitudinally (328 total visits, average follow-up time=4.9 years). Eighteen participants (14% of the cohort) demonstrated significant cognitive decline, defined as a decline in global cognitive z-score of 0.5 (SD) or more. In receiver operating characteristic (ROC) analyses, tests with an area under the curve (AUC) of greater than .7 were included in subsequent test pair analyses. Further ROC analyses examined the sensitivity and specificity of each test pair in detecting significant cognitive decline. Results were compared with the predictive ability of the Modified HIV Dementia Scale (MHDS).

The following test pairs demonstrated the best balance between sensitivity and specificity in detecting global cognitive decline: Grooved Pegboard dominant hand (GPD) and category fluency (sensitivity=.89, specificity=.60, AUC=.75, p<.001), GPD and Coding (sensitivity=.76, specificity=.70, AUC=.73, p<.001), letter fluency and Trail Making Test (TMT) B (sensitivity=.82, specificity=.63, AUC=.73, p<.001), and GPD and TMT B (sensitivity=.81, specificity=.64, AUC=.73, p<.001). Change in MHDS predicted significant decline no better than chance (sensitivity=.61, specificity=.47, AUC=.53, p=.65).

Several cognitive test pairs, particularly those that include GPD, are sensitive to HIV-associated cognitive change, and far more sensitive and specific than the MHDS. Cognitive test pairs can serve as valid, rapid, cost-effective screening tools for detecting cognitive change in PWH, thereby better enabling early detection and intervention. Future research should validate the present findings in other cohorts and examine the implementation of test pair screenings in HIV care settings. Most of the optimal tests identified are consistent with the well-established impact of HAND on frontal-subcortical motor and executive networks. The utility of category fluency is somewhat unexpected as it places more demands on temporal semantic networks; future research should explore the factors driving this finding, such as the potential interaction of HIV with aging and neurodegenerative disease.