Background:
Polypharmacy is associated with worse cognitive health among people with HIV (PWH), however explanatory models have not been established. This study leveraged data from a large and well characterized cohort of virally suppressed individuals age 40 and older to identify risk factors that explain the association between polypharmacy and cognitive impairment among PWH.
Methods:
Data obtained at enrollment into ACTG A5322 (‘HAILO’) were included. Cognitive performance was measured using four tests of verbal learning, attention/psychomotor speed, and fine motor speed and dexterity. Hierarchical density-based spatial clustering, an unsupervised machine learning based algorithm, was used to identify sub-groups based on cognitive performance. Polypharmacy (≥5 non-ART medications), and hyperpolypharmacy (≥10 medications) as well as demographic, clinical, and psychosocial variables were compared across clusters using multinomial regression, adjusted for multiple comparisons.
Results:
Participants were 870 PWH (18.4% female, 52.2% non-White), with a median age of 51. Analyses identified 8 cognitive clusters. There were no differences in average age across clusters. Cluster 1 (33% of the sample) included participants with the best test scores whereas clusters 6, 7, and 8 (collectively 48%) had the worst test scores. Polypharmacy was more common in clusters 6 and 8 compared to cluster 1 (reference group) and hyperpolypharmacy was more common in clusters 7 and 8 compared to cluster 1 (ps<.05; Table). Participants in clusters 6, 7, and 8 were also more likely to be women, Black or Hispanic, less educated, and have higher rates of cardiovascular disease, diabetes, hepatitis C, peripheral neuropathy, and substance use compared to participants in cluster 1 (ps<.05). Participants in clusters 6, 7, and 8 also reported a shorter duration of ART, lower CD4+ T-cell count and nadir, and lower CD4/CD8 ratio compared to cluster 1. Use of psychoactive medications did not differ between the clusters. Black or Hispanic women were more likely to have polypharmacy or hyperpolypharmacy (OR=1.4; 95th CI [.97-2.0]) and cognitive impairment (OR=2.8; 95th CI [1.7-4.6]) compared to any other demographic group.
Conclusions:
Psychosocial determinants of health, particularly those that disproportionately impact Black and Hispanic females with HIV, associate with an increased risk for polypharmacy/hyperpolypharmacy and cognitive impairment. Intervention/prevention efforts aimed at these high risk groups are warranted.
