Abstract
BACKGROUND: Poor adherence to oral endocrine therapy (OET) is a significant problem among patients with hormone receptor-positive breast cancer as it results in higher risk of recurrence and mortality. Non-adherence to OET is prevalent among underserved patients, often attributable to socioeconomic factors and limited healthcare access. We evaluated OET adherence patterns over time using group-based trajectory modeling (GBTM) and identified predictors of suboptimal adherence trajectory among patients seen at Harris Health System, serving underserved patients in Houston, Texas. METHODS: A single-center, retrospective study was conducted from October 2019 through December 2020. OET adherence was measured using proportion of days covered (PDC). A logistic GBTM was conducted using 2-5 adherence groups considering the Bayesian information criteria, clinical relevance, and a 5% minimum membership requirement. Multinomial logistic regression was used to assess the predictors of non-adherence trajectories. RESULTS: Among 496 patients, majority were Hispanic (62.50%) or African American (15.12%) and <65 years of age (82.66%). Four distinct adherence trajectories were identified: consistent high adherence (41.4%); constant PDC at ~0.6 (32.6%); rapid decline (14.6%); low adherence with gradual decline (11.5%). African Americans had higher likelihood of having low adherence with gradual decline [odds ratio (OR): 2.462 (confidence interval (CI): 1.1149-5.276), p=0.0205]. Patients with diabetes were more likely to have constant PDC at ~0.6 [OR: 1.714 (CI: 1.042-2.820), p=0.0338]. Longer time (4 or greater years) on therapy predicted low adherence with gradual decline [OR: 2.463 (CI: 1.266-4.793), p=0.008) and constant PDC at ~0.6 (OR: 1.966 (CI: 1.141-3.388), p=0.0149] trajectories. CONCLUSION: The identified predictors, including comorbidities like diabetes, African American descent, and longer OET treatment are crucial considerations when developing patient-centered interventions to enhance OET adherence among underserved populations. These insights can guide the implementation of initiatives such as mobile health applications, community-based educational programs, and financial aid efforts.