This research was approved by the University of Chicago institutional review board and patients provided written informed consent for survey participation and retrospective data analysis.
Study sample
Women who had ER + breast cancer were identified by the cancer registry at the University of Chicago. Patients were eligible for this study if they were English speaking, under 80 years of age at initial cancer diagnosis, were no more than 10 years out from their cancer diagnosis, and had been offered hormone therapy [either tamoxifen or an aromatase inhibitor (letrozole, anastrozole, or exemestane)] as part of their post-surgery treatment. Women with stage IV breast cancer, history of cancer prior to their breast cancer diagnosis, or a known BRCA1/2 germ-line mutation were excluded from the study.
Survey
We designed and implemented a paper survey to assess factors associated with self-reported adherence to AHT. Patients were asked to self-report demographic variables such as race, income, and education. In addition, patients were asked questions about pre-existing chronic health conditions such as diabetes and hypertension, the number of prescription medications taken, whether or not they had insurance coverage for their medications, their total out-of-pocket cost of medications per month, and perceived risk for breast cancer recurrence. The Survey Research Lab at the University of Chicago conducted pretesting of the survey questions to assess both validity and reliability. All eligible patients (N = 381) were contacted by telephone and then mailed the 15 page self- administered survey. Regardless of survey completion, all participants were given a $10 dollar gift card.
To assess adherence, women were asked how often they missed their adjuvant hormone therapy on six categories: 1) never missed a dose; 2) missed no more than 1 dose per month; 3) missed no more than 2 doses per month; 4) missed 1–2 doses per week; 5) missed ≥3 doses per week; and 6) never took the therapy. Regardless of whether or not they chose to start AHT, participants were asked how important they thought AHT was in decreasing their risk for breast cancer recurrence (not important, a little important, moderately important, or very important). Participants were also asked how worried they were about their risk for breast cancer recurrence (very, somewhat, or not worried) and how this compared to their fears at the time of their cancer diagnosis. Other questions included how worried they were about the long-term side effects of taking AHT and concerns about their overall medication costs and the potential for AHT to interfere with their other medicines. In addition, participants were asked how heavily they weighed their doctor’s opinions in their decision to take or not take adjuvant hormone therapy (almost entirely, partly, or not at all).
Measurements
Prior studies using both self- report and pharmacy prescription refill records have used an 80% cut-off to define adherence since <80% adherence has been associated with increased mortality (Kahn et al. 2007; Molfenter et al. 2012; Stavropoulou 2011). Thus we defined adherence in our cohort by self report of taking ≥ 80% of prescribed pills on average per month and patients were considered non-adherent if they missed greater than or equal to 3 doses per week on average (which corresponded to taking therapy less than 80% of the time per month) or if they had not initiated therapy at the time of the survey.
Persistence was defined as completion of 5 years of therapy and was verified by chart review. Chart review documentation in clinic notes from the breast surgeon, radiation oncologist, or medical oncologist were used to assess completion of 5 years of AHT. At the time of chart review, 25 women (12.6%) were in their fourth year of therapy. These women were included as having completed 5 years of AHT. We felt this was appropriate since prior studies have shown that women who have taken most of their therapy are very likely to complete therapy (Molfenter et al. 2012). We also conducted sensitivity analyses by including and then excluding these 25 women and found no difference in either persistence rates or predictors of persistence. Patients with missing data were excluded from the overall analysis.
Compliance was defined as a composite of being both adherent and persistent. Of the women who were found to be persistent, only one patient (0.7%) reported non-adherence. Therefore, our definition of compliance relies heavily on persistence.
Statistical analysis
All statistical analyses were conducted with Stata 12.1 (StataCorp, College Station, TX, USA). Demographic characteristics and variables previously mentioned were compared for adherence and compliance rates using a Spearman’s rank correlation or Kruskal-Wallis test. Originally, adherence to hormone therapy was measured as a 6-scale ordinal outcome based on self-report. However, the categories for non-adherence were very small, and thus were combined into one category. We used logistic regression models to evaluate how perceived importance of therapy, concern about breast cancer recurrence, concern about side effects of therapy, and how heavily they valued their doctor’s opinion factored into their decision to adhere and complete therapy. Self- reported race and value placed on their doctor’s opinions were adjusted in the multivariable model to remove confounding effects (Model B) and is considered the full model. Concern about side effects was tested as an additional confounder (Model C) but was not included in the full model. To test if the associations differed by race, we generated an interaction term between race and the variable of interest, and used the Likelihood-ratio test to determine if the interaction was significant.