Article Title

Working Towards De-Implementation: A Mixed-Methods Study in Breast Cancer Surveillance Care

Publication Date



implementation science, mixed-methods research


Background/Aims: De-implementing commonly used but ineffective clinical practices is an important component of quality. Oncology offers several opportunities to reduce use of ineffective practices based on guidelines from the American Society of Clinical Oncology (ASCO). We studied the use of one such practice, biomarker blood tests for breast cancer surveillance, within an integrated health care system. We documented utilization patterns and explored provider perceptions and attitudes to inform de-implementation efforts.

Methods: Using a sequential explanatory mixed-methods design, we identified a cohort of early-stage breast cancer survivors and calculated the number and frequency of biomarker tests during an 18-month posttreatment period. We identified high- and low-utilizing medical centers and conducted semi-structured qualitative interviews with oncologists in both types of centers, guided by the Theoretical Domains Framework. Interviews were transcribed, coded and analyzed.

Results: Among 7,363 patients diagnosed during 2009–2012, 40,114 biomarker tests were ordered for 41% of patients. We found significant variation by medical center: 5–78% of eligible patients received a test. We interviewed 18 oncologists in high- and low-utilizing centers. Several themes emerged, including: 1) Awareness of nonadherence: oncologists reported awareness of current ASCO guidelines and agreed that biomarkers are not clinically useful; high utilizers acknowledge intentional nonadherence (“We all know we shouldn’t do this but do it anyways.”); 2) Anxiety: despite agreeing that biomarkers aren’t useful, some oncologists are anxious about missing a recurrence and want to do “all possible” to prevent this; high utilizers perceive that patients are highly anxious and desire a quantitative test for reassurance (“They need a number.”); and 3) Perceived patient expectations: oncologists perceive competition from other systems and are concerned about perception of withholding care (“If [competitor] does it … patients expect it.”) and implications for patient satisfaction.

Conclusion: Barriers to de-implementation are numerous and complex. Traditional strategies of practice change based on increasing awareness and knowledge (provider education, electronic alerts) are unlikely to be effective. Multifaceted, multilevel strategies deployed to address consumer-, clinician- and system-related barriers are required. Research-based development and evaluation of multilevel de-implementation strategies is critical and will help develop valuable insights and theories regarding the determinants of clinical practices and opportunities to influence them.




June 27th, 2016


August 12th, 2016