P4-12-12: Patient-, Illness-, and/or Treatment-Related Baseline Predictors of Nonadherence to Oral Hormonal Therapy.

Abstract Background: Nonadherence to oral hormonal therapy is problematic for women with breast cancer. Patient-, illness-, and treatment-related factors have been associated with nonadherence, but with inconsistent findings. Therefore, our aim was to explore predictors of nonadherence to hormonal therapy for women with early stage breast cancer from the baseline assessment (pre-hormonal therapy) to 6 months post-treatment. Methods: A secondary analysis was performed to explore potential patient-, illness-, and treatment-related predictors of nonadherence for 198 women enrolled in either: 1) The Anastrozole Use in Menopausal Women Study (AIM Study, n=162), or, 2) Predictors of Adherence to Hormonal Therapy in Breast Cancer (ONS Study, n=36). Nonadherence was defined as the percentage of prescribed administrations of hormonal therapy that were not taken during the first 6 months of therapy as measured using an electronic Drug Exposure Monitor (eDEM) (AARDEX, Ltd.). Chi-square tests of independence and Mann-Whitney U tests were performed to determine whether data from the two studies could be pooled. Descriptive statistics were performed to characterize the sample. Multiple linear regression analyses were performed to identify the best model in two stages: 1) univariate relationships between each candidate predictor variable and the outcome variable (6-month nonadherence) were assessed using a cut-off of p=.20; and, 2) candidate predictors meeting the criteria were retained for further exploration in model-building multiple linear regression analyses (stepwise and backward) to determine the predictors of 6-month nonadherence summary scores. Candidate predictors were retained in the model if they remained associated at p<.10 in the multiple linear regression analysis. Results: Women were 98.3% Caucasian with a mean age 59.1 years (SD 7.5) and mean number of years of education of 15.0 (SD 2.9). Overall mean nonadherence was 11.6% (13.2% AIM Study, 4.6% ONS Study). Chi-square and Mann-Whitney U tests demonstrated that the two samples could be pooled, given the data were similar on key variables (number of years of education, depression, anxiety, fatigue, symptoms, and nonadherence). Both stepwise and backward elimination modeling algorithms demonstrated evidence of 3 significantly significant variables associated with nonadherence; however, the backward elimination model best represented the sample (R2=.106, adjusted R2=.086, s=0.28490). Women who worked (p=.082), whose primary occupation was clerical or administrative (p=.029), had DCIS tumor type (p=.017), or who had higher gastrointestinal (GI) symptoms scores (p=.013) were associated with nonadherence. The potential for interactions between primary occupation, DCIS tumor type, and higher GI symptoms was explored and was not significant. In addition, while all tumor types were examined as candidate predictors, participants with DCIS also had another tumor to be eligible for the parent studies. Conclusions: Our study offers insight into potential predictors of nonadherence for women participating in one of two cohort studies. The findings suggest additional examinations of nonadherence concerning work and symptom burden and their relationship to nonadherence are indicated. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P4-12-12.