The association between trajectories of endocrine therapy adherence and mortality among women with breast cancer.

Abstract

Studies examining adherence to endocrine therapy (ET) and breast cancer-related outcomes have traditionally used the proportion of days covered (PDC) by medication to define adherence which may mask true treatment-outcome associations for patients with different medication use behaviors. We use group-based trajectory models to examine the association between ET adherence patterns and mortality compared to a standard PDC adherence measure. Using Surveillance, Epidemiology and End Results-Medicare data we included 9492 women with breast cancer who initiated ET between 2007 and 2010. We excluded women who died/recurred in the 12 months after ET initiation. We used monthly group-based trajectory models to characterize longitudinal adherence patterns and adjusted Cox proportional hazard models to estimate the association between ET adherence and mortality, comparing trajectory-based adherence to traditional PDC-based measures. Trajectory models identified five adherence groups: (i) high (56.2%); (ii) quick decline (9.5%); (iii) moderate decline (7.9%); (iv) quick decline, then increase (16.0%); and (v) slow decline (10.5%). Mortality was significantly associated with group assignment; compared to the high adherers, there was a significantly higher risk of death among quick declines (HR = 1.41, 95%CI = 1.09-1.72) and moderate declines (HR = 1.25, 95%CI = 1.00-1.55). Using the standard PDC adherence measure women with adherence <80% over the year had a higher risk of death than those with adherence ≥80% (HR = 1.21, 95%CI = 1.06-1.38). Defining ET adherence using trajectory models improved adherence measurement. These models could inform clinical practice by helping to identify common adherence patterns, potential areas for intervention and better isolate adherence-related outcomes in comparative effectiveness studies. Copyright © 2016 John Wiley & Sons, Ltd.