Application of statin medication adherence trajectory models in an integrated financing and care delivery system [abstract] Abstract uri icon

abstract

  • INTRODUCTION: Group-based trajectory models have been applied to classify patients based on medication adherence patterns over time, rather than a single summary measure (e.g., proportion of days covered). Previous studies of trajectory group membership used claims data to predict trajectory group inclusion. RESEARCH QUESTION OR HYPOTHESIS: The study objective was to classify 12-month statin medication adherence patterns and develop prediction formulas for forecasting individual patients’ adherence trajectories using baseline claims and EMRbased clinical data. STUDY DESIGN: Retrospective descriptive analysis METHODS: Members of a Midwestern health plan who initiated new statin therapy in 2012 were included. Members were partitioned into six group-based trajectory models of utilization using latent class analysis. Logistic regressions were conducted with membership in each adherence group as the outcome. Demographic, clinical, and healthcare utilization variables were explored as possible predictors of inclusion into adherence groups. Akaike Information Criterion (AIC) was used to determine the best multivariate predictive model for each group. RESULTS: Six adherence trajectory groups were defined with distinct adherence and persistence patterns over 12 months; including ongoing persistence, early discontinuation, and fluctuation in utilization. The trajectory most easily predicted was discontinuation at approximately two months (C-statistic = 0.74). Predictors for this group included: black or Hispanic race/ethnicity; Charlson score; diagnosis of a chronic lung condition, chronic kidney disease, obesity, or depression; the use of multiple prescribers or multiple pharmacies; and total number of chronic prescription medications. CONCLUSION: Baseline clinical and claims-based characteristics may be used to identify new statin users likely to follow one of six specific adherence trajectories. This information may support early targeted interventions to improve adherence in high-risk groups.

publication date

  • 2016