Identification of subgroups of early breast cancer patients at high risk of nonadherence to adjuvant hormone therapy: results of an italian survey.

The aim of this study was the identification of subgroups of patients at higher risk of nonadherence to adjuvant hormone therapy for breast cancer. Using recursive partitioning and amalgamation (RECPAM) analysis, the highest risk was observed in the group of unmarried, employed women, or housewives. This result might be functional in designing tailored intervention studies aimed at improvement of adherence. Background: Adherence to adjuvant endocrine therapy (HT) is suboptimal among breast cancer patients. A high rate of nonadherence might explain differences in survival between clinical trial and clinical practice. Tailored interventions aimed at improving adherence can only be implemented if subgroups of patients at higher risk of poor adherence are identified. Because no data are available for Italy, we undertook a large survey on adherence among women taking adjuvant HT for breast cancer. Patients and Methods: Patients were recruited from 10 cancer clinics in central Italy. All patients taking HT for at least 1 year were invited, during one of their follow-up visit, to fill a confidential questionnaire. The association of sociodemographic and clinical characteristics of participants with adherence was assessed using logistic regression. The RECPAM method was used to evaluate interactions among variables and to identify subgroups of patients at different risk of nonadherence. Results: A total of 939 patients joined the study and 18.6% of them were classified as nonadherers. Among possible predictors, only age, working status, and switching from tamoxifen to an aromatase inhibitor were predictive of nonadherence in multivariate analysis. RECPAM analysis led to the identification of 4 classes of patients with a different likelihood of nonadherence to therapy, the lowest being observed in retired women with a low level of education, the highest in the group of unmarried, employed women, or housewives. Conclusion: The identification of these subgroups of “real life” patients with a high prevalence of nonadherers might be functional in designing intervention studies aimed at improving adherence