A Multi-Level Strategy for De-implementing Mammography Overuse Among Older Women

De-implementation, defined as "reducing (frequency and/or intensity) or stopping the use or delivery of health services or practices that are ineffective, unproven, harmful, overused, inappropriate, and/or low-value by practitioners and delivery systems to patients", is recognized as a critical but understudied area within implementation science. Despite growing research on determinants of and strategies for implementation of evidence-based practices, there has been little empirical work on factors that influence de-implementation, or effective strategies for facilitating de-implementation. Further, there are gaps in knowledge in determining optimal methods and approaches for identifying and selecting de-implementation strategies to match barriers to de-implementation. Addressing screening mammography overuse among older women offers excellent opportunities for both improving healthcare delivery and outcomes among older women and advancing the science of de-implementation.

Routine screening mammography is widely implemented among older women despite the following: 1) screening mammography does not significantly reduce breast cancer mortality among older women who have shorter life expectancies, greater competing health risks, and lower risk for clinically significant or rapidly progressive breast tumors, and 2) screening mammography among older women poses substantial harms including anxiety and complications from follow-up diagnostic tests, false positive results, and potential over-diagnosis and over-treatment of tumors that would not have progressed. The current American College of Physicians guidelines recommend discontinuation (e.g., removal) of mammography in women aged ≥75 years at average risk for breast cancer; other national guidelines (e.g., United States Preventive Services Task Force, American Cancer Society) do not support routine screening mammography in older women and recommend consideration of morbidities, life expectancy and patients' informed preferences.

Informed by (1) multi-level qualitative interviews among patients and providers on the drivers of routine screening mammography in older women (n=36); (2) the Knowledge-to-Action Model and the Dual Process Model; (3) an innovation tournament - a crowdsourcing method, successfully applied as an emerging participatory approach for generating implementation solutions (n=47); and (4) a discrete choice experiment among patients (n=673), the research team identified, designed, and tailored a multilevel de-implementation strategy involving the patient, provider, and organizational levels. The research team will conduct a cluster randomized controlled trial (at the provider level) to test the impact of the provider- and patient-level components of the multilevel strategy on screening mammography use, and secondarily, on provider referrals/orders for screening mammography, across a large New York City healthcare system serving a racially and ethnically diverse population. The organizational level components of the multilevel strategy will be implemented among all participants, not via random assignment.

Patient level criteria

Inclusion criteria

• English- or Spanish-speaking women based on preferred language in electronic health record
• Aged 75 and older at the time of their scheduled primary care visit
• Women who have a primary care visit scheduled within the next 2-4 weeks
• Women who have not had a screening mammogram 6 months prior to their scheduled primary care visit but have had a screening mammogram 7-18 months prior to their scheduled primary care visit

Exclusion criteria

• Women with a history of atypical ductal hyperplasia (ADH) or non-invasive or invasive breast cancer (assessed via ICD 10 code)
• Women with dementia (assessed via ICD 10 code)
• ICD codes to use across exclusion criteria: F01-F03, Z85.3, G30-G31, N60, C50, or D05

Provider level criteria

Inclusion criteria

• Primary care clinicians attributed to the 2 intervention clinics that serve adult patients, including those ≥75 years

Exclusion criteria

• Specialists
• Primary care clinicians for patients < 18 years old
• Non-clinicians
• Physician assistants