Abstract
Background: Observational studies report strong associations between polypharmacy and adverse outcomes and suggest that people with HIV (PWH) may be at greater risk than uninfected comparators. Yet, factors driving excess risk for PWH have not been characterized. We explore the role of demographics, physiologic frailty and known pairwise drug interactions (KPDI) in risk of hospitalization and mortality associated with polypharmacy by HIV status. Methods: KPDI were identified for each patient by linking data from the Veterans Aging Cohort Study (VACS) to DrugBank Version 5.0, mapping active medications in 2009 on 9,186 PWH and 37,930 uninfected comparators. Outcomes (hospitalization and mortality) were observed from 2010-19. Using simulation across the range of common medication counts (2-12), we estimated the KPDI expected by chance and compared with observed. We then calculated average observed association with mortality for each KPDI and summed these creating a KPDI Index. In nested models, the association between medication count and hospitalization and mortality was explored sequentially adjusting for demographics, physiologic frailty (VACS Index) and KPDI Index. Findings: When simulation was compared with observation, each additional medication was associated with 5-6 times the KPDI expected by chance and PWH had more KPDI than comparators (2.9 for PWH vs. 2.7 for comparators). After adjusting for demographics and physiologic frailty, adjustment for KPDI Index reduced associations of medication count with both outcomes. But, after full adjustment, PWH still had greater residual risk of hospitalization (HR 1.07, 95% CI 1.06, 1.08) than comparators (HR 1.04, 3 95% CI 1.03, 1.05, p for interaction=0.02) and of mortality (HR 1.03, 95% CI 1.01, 1.04) than comparators (HR 0.98, 95% CI 0.97, 0.99; p for interaction=0.03). Interpretation: Controlling for medication count, PWH have more KPDI than comparators likely due to increased interactivity of ART medications. After adjustment for KPDI, PWH also had more residual risk. Higher level drug interactions, drugsubstance use interactions, mitochondrial toxicity and/or a greater susceptibility to toxicity may explain excess harm for PWH.
