Use of a Discrete Choice Experiment to Inform De-implementation of Mammography Overscreening: A US-Based National...

Mammography overscreening, defined as any routine screening in women ≥ 75 years, particularly with limited life expectancy, persists.

Identify preferences for de-implementing mammography overscreening among older women.

A national survey using the NORC AmeriSpeak panel, a probability-based panel representative of US households. Informed by qualitative methods, we constructed a discrete choice experiment (DCE) based on a hypothetical patient activation de-implementation strategy (the Rethink Resource) for prompting patient/provider discussions about whether to stop getting mammograms.

Women ≥ 70 years old selected using sampling strata based on age, race/ethnicity, and education and without a breast cancer history.

Attributes (levels) included modality (electronic, paper, in-person); context (reviewed with provider, group, on their own); content (mammography pros/cons, patient story/testimonial); frequency (once, yearly); and decision-making principles (age/health calculator, personal preferences/responsibilities checklist). We estimated a random utility model to quantify patient preferences and calculate importance scores.

There were 673 eligible participants; the weighted mean age was 77.5 (standard deviation: 5.3); 72.0% were Non-Hispanic White, 10.5% Non-Hispanic Black, and 9.8% Hispanic; 69.3% had less than a college degree; 49.6% agreed with the idea of stopping mammography based on age and health. In order of importance, participants preferred (mean [standard error]) the Rethink Resource be reviewed: with their healthcare provider (1.52 [0.08]) or on their own (1.22 [0.07]), include pros/cons (0.79 [0.05]), and be delivered on paper (0.81 [0.07]) or electronically (0.60 [0.07]) on a yearly basis (0.34 [0.05]). There were no significant preferences for decision-making principles (-0.01 [0.05]).

In the first DCE for de-implementation strategies, we found that women express clear preferences for how and with whom information is relayed, but do not have strong preferences for calculators/checklists.