Sampling and protocol
In line with this method, we conducted in-depth individual interviews with purposively sampled experienced generalist physicians who practice in diverse clinic settings and were willing to tell stories about memorable personal errors. We defined “experienced” as being in post-training practice for at least five years. We created this criterion to ensure that our sample of physicians had a depth of clinical experiences by which to contextualize their errors; although, we did not specifically limit the timeframe from which participants would select their stories. Furthermore, all physicians were based in Canada or the United States. We involved a wide range of physicians in order to speak with those who were willing to share rich and detailed stories. Physicians from Canada and the United States were included due to an assumed similarity in culture around medical errors that may arise due to the influence of organizations or projects like the Associated Medical Services Phoenix Project, which has members from across North America57.
Given these stories may be seen by some as reflecting poorly on their professional competence or ability, we identified confidentiality to be of utmost importance to eliciting rich data. We restricted our data collection to generalist physicians in the hopes that the commonalities among their practices and treating and diagnosing patients with a wide variety of complaints would facilitate comparison of stories across the whole dataset. We chose not to focus on a single location or specialty in an effort to look for transferable ideas that are consistent across a variety of clinical contexts. See Table 1 for detailed inclusion and exclusion criteria.
(INSERT TABLE 1 AROUND HERE)
After identifying and recruiting participants, we conducted in-depth interviews either in-person or over the telephone, based on participant preference. Telephone interviews can drive the gathering of sensitive information when anonymity is preferred59 and facilitate the inclusion of participants who were both socially and geographically distant from our own networks. All interviews were conducted by non-clinicians with experience conducting qualitative interviews (MV, EC, SK).
The interviewees were asked to share two memorable stories of personal errors—one which they chose to share with others and another which they kept more private. The interviewer elicited the entire story of the error and asked follow-up questions about how the physician dealt with the error with a focus on how it affected their practice over time. All interviews were audio-recorded and transcribed verbatim. See Appendix 1.0 for the detailed interview guide.
We used the theory of information power to establish data sufficiency. We judged the information from 26 participants to be sufficient due to the narrow scope of our project, strong dialogue from participants that produced rich data, and the use of established theory. The research aim, specificity, dialogue, and analysis drive us toward higher information power and smaller sample size58.