Learning from the "tail end" of de-implementation: the case of chemical castration for localized prostate cancer

Background: Men with prostate cancer are often treated with the suppression of testosterone through long-acting injectable drugs termed chemical castration or androgen deprivation therapy (ADT). In most cases, ADT is not an appropriate treatment for localized prostate cancer, indicating low-value care. Guided by the Theoretical Domains Framework (TDF) and the Behavior Change Wheel's Capability, Opportunity, Motivation Model (COM-B), we conducted a qualitative study to identify behavioral determinants of low-value ADT use to manage localized prostate cancer, and theory-based opportunities for de-implementation strategy development.

Methods: We used national cancer registry and administrative data from 2016 to 2017 to examine the variation in low-value ADT use across Veterans Health Administration facilities. Using purposive sampling, we selected high- and low-performing sites to conduct 20 urology provider interviews regarding low-value ADT. We coded transcripts into TDF domains and mapped content to the COM-B model to generate a conceptual framework for addressing low-value ADT practices.

Results: Our interview findings reflected provider perspectives on prescribing ADT as low-value localized prostate cancer treatment, including barriers and facilitators to de-implementing low-value ADT. We characterized providers as belonging in 1 of 3 categories with respect to low-value ADT use: 1) never prescribe 2); willing, under some circumstances, to prescribe: and 3) prescribe as an acceptable treatment option. Provider capability to prescribe low-value ADT depended on their knowledge of localized prostate cancer treatment options (knowledge) coupled with interpersonal skills to engage patients in educational discussion (skills). Provider opportunity to prescribe low-value ADT centered on the environmental resources to inform ADT decisions (e.g., multi-disciplinary review), perceived guideline availability, and social roles and influences regarding ADT practices, such as prior training. Provider motivation involved goals of ADT use, including patient preferences, beliefs in capabilities/professional confidence, and beliefs about the consequences of prescribing or not prescribing ADT.

Conclusions: Use of the TDF domains and the COM-B model enabled us to conceptualize provider behavior with respect to low-value ADT use and clarify possible areas for intervention to effect de-implementation of low-value ADT prescribing in localized prostate cancer.

Trial registration: ClinicalTrials.gov , NCT03579680.