Group-based trajectory modeling to identify longitudinal patterns and predictors of adherence among older adults on concomitant triple therapy (oral antidiabetic, renin-angiotensin-system antagonists, statins).

Abstract

Background: Diabetes, hypertension, and hyperlipidemia frequently co-occur in older adults, significantly increasing their risk for cardiovascular disease, a leading cause of mortality in the United States. Managing these conditions often requires concomitant triple therapy, which includes antihypertensives, oral antidiabetics, and statins. Although medication adherence is critical for reducing cardiovascular risk, adherence to complex regimens is often suboptimal in older populations, further complicating disease management. Medicare's STAR metrics assess adherence to these medications as a measure of care quality. Traditional methods, like the proportion of days covered (PDC), provide single adherence estimates, but fail to capture the dynamic nature of adherence over time. Group-based trajectory modeling (GBTM) offers a more comprehensive approach, graphically depicting patterns of adherence behavior. This study seeks to understand longitudinal patterns and predictors of adherence of concurrent triple therapy among elderly patients under Managed Care using GBTM.

Objective: To evaluate adherence patterns to concurrent triple therapy (antidiabetic, antihypertensive, and lipid-lowering medications) among older patients using GBTM and identify predictors associated with each adherence trajectory.

Methods: Patients on concurrent triple therapy were identified using a Texas Medicare Advantage dataset from July 2016 to December 2016. Patients included had an overlap of 30 days of triple therapy, a second prescription of each component of triple therapy within the identification period, and a 12-month follow-up after the triple therapy. Monthly adherence was measured using PDC during follow-up. Patients were defined as adherent if they had at least 80% (24 out of the 30 days) covered for all 3 medications. The monthly PDC was incorporated into a logistic GBTM to provide distinct patterns of adherence. Two to five adherence groups were estimated using the second-order polynomial function of time. Predictors of adherence were identified using multinomial logistic regression, guided by the Anderson Behavioral Model.

Results: Of the 7,847 patients included, the following 4 distinct adherence trajectories were identified: adherent (42.5%), gaps in adherence (28.9%), gradual decline (13.4%), and rapid decline (15.3%). Female patients had higher odds of being in the gaps in adherence or rapid decline groups compared with males. Low-income subsidy recipients were less likely to experience rapid decline. Prior hospitalizations increased the likelihood of rapid decline in adherence.

Conclusions: This study identified heterogeneous adherence patterns among older adults on triple therapy for cardiovascular disease risk factors. Targeted interventions tailored to specific adherence trajectories are needed to improve medication adherence and health outcomes in this high-risk population.