OBJECTIVES: To model the baseline and time-varying determinants of longitudinal statin adherence, and to identify and characterize the heterogeneity in statin adherence patterns over time. METHODS: : We used Marketscan commercial claims data from 2008-2013. Adults newly initiating a statin, with at least 3 years of continuous eligibility after statin initiation were included. The 3 year follow-up period was divided into 12 periods of 90 days; a period was considered an adherent one if the patient was exposed to statins for 80% of the days. Baseline covariates included age, sex, and history of congestive heart failure (CHF), hypertension (HTN), stroke, asthma, chronic obstructive pulmonary disease (COPD), depression, and diabetes (DM). Time varying covariates included new onset of clinical conditions (same conditions as baseline plus liver dysfunction and myopathy) and average non-statin daily pill burden in a period (<1, 1-2, or >2 pills/day). Generalized estimating equations (GEE) were used as primary analysis. We estimated Group Based Trajectory Models (GBTM) while adjusting for all baseline and time-varying covariates as secondary analysis. RESULTS: :