Abstract Background Completion of the recommended 5-10 year course of adjuvant endocrine therapy (AET) course is low, and few risk factors for discontinuation have been identified. We examined factors associated with early AET discontinuation in a large cohort with patient-reported outcomes (PROs) and other measures routinely used in clinical care at four sites within an academic health system's multidisciplinary cancer center. Methods We examined postmenopausal women with stage I-III hormone-receptor positive breast cancer identified using the registry at an academic multidisciplinary cancer center 2014-19 who received a prescription for AET, and followed them for up to 5 years. Breast cancer extent of disease and treatments, Elixhauser comorbidity index, medication number and sociodemographic characteristics (age, race/ethnicity, insurance, area deprivation index of zip code of residence) were abstracted from the electronic health record (EHR). PRO measures included the Distress Thermometer (DT, scored 0-10) and its associated problem list (39 yes/no questions), and the Patient Health Questionnaire-9 (depression), all administered routinely as part of clinical care between diagnosis date and start of AET. If >1 PRO performed during that time, the measure closest to AET start was used. DT was dichotomized using the clinically accepted score of 4 as the threshold for severe distress, as was Patient Health Questionnaire-9 (PHQ-9) depression ( >5). AET prescription information included both prescription date from the EHR, and prescription fills assessed using an all-payer pharmacy data source. AET was defined as discontinued (d/c) when no prescription had been filled at any pharmacy for at least 90 days. We used nonparametric hypotheses test for two-samples (d/c vs. no d/c) as well as Kaplan-Meier product-limit time-to-event analyses with censoring for recurrence, death, or change of provider, and advanced machine learning to examine the association of these variables with discontinuation prior to five years. 54 curated covariates were examined in all, corresponding to a Bonferroni statistical significance p-value threshold of 0.001 to control type I error. Sensitivity analyses were performed to 1)use alternative date of DT (closest to diagnosis date) and 2)to examine d/c defined as 180 days. Results Among the cohort of 985 women (74% stage I, 25% Elixhauser comorbidity score >2), 36.9% had a DT >4, and 6.7% a PHQ-9 >5. The cohort had a median followup time of 954 days, and during followup 291 women ( 29.5) % discontinued before completion of the 5-year course or censoring for death, change to a provider outside the system or study end date in 2021. Over 40% of women who stopped filling prescriptions never had their prescriptions discontinued by their physician in the EHR. None of the examined variables including those for Distress and PL, PHQ-9 and measures of extent of disease, cancer treatments, medication number and sociodemographic characteristics were associated with early discontinuation. Results were similar in analyses using the DT date closest to diagnosis or a 180-day prescription fill gap. We also used advanced machine learning with respect to time to event discontinuation had similar results. Conclusions. Clinically available EHR data including two PRO measures were not reliable predictors of early discontinuation. However, many women who had discontinued still had orders for AET in the EHR, suggesting that their physicians were unaware that they had stopped filling prescriptions. Future work should examine whether routine monitoring of prescription fills might be enlisted to identify these women and assist with education, supportive care or other assistance to restart. Citation Format: Joan Neuner, Rodney Sparapani, Jacob Tiegs, Vaia Makris, Melinda Stolley, Sailaja Kamaraju, Bradley Crotty, Kathryn Flynn. Pharmacy fills but not baseline distress or other clinical factors can identify those at high risk for early adjuvant endocrine therapy discontinuation [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-17-05.
Read full abstract