Material and method
In February 2018, 71 first year junior doctors at Akershus University Hospital and Nordland Hospital were invited to take part in a pilot study on uncertainty in clinical practice. The study was in two parts and consisted of a questionnaire with 30 questions about tolerance for uncertainty, and a qualitative interview. Of the 65 who responded to the questionnaire, 48 agreed to be contacted about a qualitative interview.
Based on the responses from the questionnaire survey at Akershus University Hospital (n = 42), a hierarchical cluster analysis of response patterns was performed. Six groups of doctors with similar response patterns were identified. One doctor in each group was invited to a qualitative interview. Informed by this method and with the aim of ensuring a similar stratification of interviewees at Nordland Hospital, four doctors who had consented to an interview were invited. These four were sent an email or text message, consented to participation and were included in the study.
Six informants were women, and eight informants were educated in Norway (all four Norwegian universities were represented). Before starting their specialty training, the first year junior doctors had an average of nine months' experience of working with a medical licence (0–23 months). We therefore felt confident that we had a representative sample in terms of gender, hospital size, place of education and work experience, and the personal diversity was sufficient for us to expect a rich data material.
Qualitative interviews were conducted according to a semi-structured interview guide consisting of ten questions (see the appendix). The study deliberately did not use a predetermined definition of medical uncertainty, and the interview guide started with the following two questions: 1) When I say 'uncertainty in clinical practice', what first springs to mind? 2) When you think about uncertainty, what do you think it entails?
At Akershus University Hospital, the interviews were conducted 8–9 months into the year by PG, whom the doctors had not met since the induction week. At Nordland Hospital, the interviews were conducted 9–10 months into the year by a research assistant, who was a qualified doctor and was trained in conducting qualitative interviews. The reason for this was that the local study member (EHO) was the head consultant in the emergency department, and we wanted the interviewees to be able to speak freely to an interviewer with whom they had no working relationship. The interviews lasted 15–60 minutes, and most were 30–45 minutes long. The interviews were transcribed verbatim and translated to English so that PKH could participate in the analysis.
An initial core team of BN, PG and EHO used a detailed American codebook from a similar study for all statements in the interviews (4), which sharpened our focus on nuances in interpretations and classifications related to English and Norwegian. Due to changes in the working situation, we subsequently had to assemble a new core team consisting of EHO, KA and PG. We decided to take a fresh look at the material and chose the systematic text condensation method (10). The core team's analyses were presented to and discussed with BN, TG and PKH in several rounds.
The study was assessed and approved by the Norwegian Centre for Research Data (project number 59292) in 2018 as well as the data protection officers at Akershus University Hospital and Nordland Hospital.