A novel methodology to identify and survey physicians participating in medical aid-in-dying.

Physicians who participate in medical-aid-in-dying (MAID) cannot be easily identified and studied due to cost and anonymity barriers. We developed and empirically tested a novel methodology to identify and survey physicians highly likely to participate in MAID activities. We used a state-level comprehensive administrative claims database to identify a cohort of patients with diagnoses and hospice enrollment similar to those known to have filled a prescription for MAID from 2017-2018. We then identified physicians who provided routine outpatient care to these patients using National Provider Identifier numbers. We surveyed these physicians in 3 waves (n = 583 total surveys), ranking physicians in order of their likelihood of being asked about MAID for each wave based on characteristics including specialty and the number of unique patients they had provided care to. We re-ranked physicians in waves 2 and 3 based on responses from prior waves. Physicians were surveyed only once and there was no follow-up to preserve anonymity. Surveys assessed the proportion of respondents who participated in MAID activities (discussions, referrals, and/or prescriptions). We identified 6369 physicians that provided care to 2960 patients. In survey waves one, two, and three respectively, response rates (55%, 52%, and 55%; p = 0.98) and the proportion of respondents that participated in MAID activities (58%, 56%, and 42%; p = 0.05) were similar. Small adjustments made to physician ranking criteria in waves two and three did not increase the proportion of physicians that participated in MAID activities. We used a novel methodology using administrative data to identify and survey physicians at high likelihood of participating in MAID activities. We achieved good overall response rates (52%), and a high proportion of respondents that participated in MAID activities (52%), demonstrating that it is possible to overcome cost and anonymity barriers to conducting quantitative research on MAID. This methodology could be used in larger scale studies of MAID or other bioethical issues with "hidden" physician populations.