Do medication adherence trajectories predict clinical endpoints in a cohort with HIV and chronic comorbidities?

With little known about medication adherence for non-HIV chronic conditions in people living with HIV (PLHIV), this research evaluated antiretroviral (ART) adherence and non-ART composite medication adherence patterns among PLHIV and non-HIV chronic conditions (i.e., type 2 diabetes, hypertension, and hyperlipidemia) and their association with corresponding treatment goals.

Group-based multi-trajectory modeling of ART and non-ART adherence was used to identify medication adherence trajectories (MATs) over a 37-month observation period (9/2018-9/2021) in 525 continuously-enrolled PLHIV with multiple chronic conditions. ART and non-ART adherence were estimated using monthly proportion of days covered for diabetes, renin-angiotensin system antagonist, and statin medications. Relationships between MAT group membership and viral load, and a composite endpoint reflecting hemoglobin A1c, blood pressure, and/or total cholesterol control were evaluated using multivariable logistic regression.

A six MAT model was identified: MAT-1 had inadequate, decreasing ART and non-ART adherence (6.3%); MAT-2 had inadequate, increasing ART and non-ART adherence (6.0%); MAT-3 had inadequate, divergent ART and non-ART adherence (11.5%); MAT-4 had inadequate, stable ART and non-ART adherence (17.6%); MAT-5 had low adequate ART and non-ART adherence (32.6%); and MAT-6 had high adequate ART and non-ART adherence (25.9%). Compared to MAT-6, MAT-1 (aOR = 0.24, p = 0.004) and MAT-2 (aOR = 0.24, p = 0.004) had decreased odds of achieving a viral load <20 copies/mL. Additionally, MAT-1 (aOR = 0.31, p = 0.038) and MAT-4 (aOR = 0.49, p = 0.033) had decreased odds of achieving the composite endpoint compared to MAT-6.

Varying patterns of ART and non-ART adherence are differentially associated with achieving intermediate clinical endpoints suggesting tailored intervention to address unique medication adherence needs.