Nvivo version 11 pro

The quantitative data were analyzed with Stata version 16 (Statacorp, College Station, TX, USA), while the qualitative data were analyzed by using Nvivo version 11 Pro (QSR International Pty Ltd., Victoria, Australia). Frequencies and percentages were used to summarize the categorical variables for the close-ended questions. Chi-square and Fisher’s exact tests were used to compare the differences in knowledge by group (HCWs vs. HMOs) and by counties. Theme codes were used to summarize the open-ended (qualitative) questions, especially the respondents’ attitudes and preferences. The lead authors (BON, VAO), assisted by RAs, familiarized themselves with all the transcripts before entering the data into Nvivo. Abstracted sections were assigned codes as the authors compared key points by consensus.

All video-recorded focus groups were transcribed into verbatims and imported into the qualitative data analysis software NVivo version 11 pro (QSR International). Field notes, including facial and non-verbal cues, were added to the verbatims. A list of nodes (i.e. a code, theme, or idea) was generated a priori using the themes covered in the discussion guide. Additional nodes were created during coding as new themes emerged from the dataset and were validated with the research team to ensure their relevance. Each node was defined using Aristotelian definitions (a good practice in writing definitions) to ensure accuracy of coding [17 ]. The two first focus groups were coded and simultaneously discussed together by the two coders (RD and NS) and entirely reviewed by a third research member (AB). The six other focus groups were then coded independently by the two coders (RD and NS). All coding discrepancies were resolved through discussion until a consensus was reached. When consensus was not reached, a third reviewer was involved (AB). Using an iterative process, the research team met to discuss the coding of the transcripts periodically. We established that data saturation was obtained when no new themes or codes emerged from the dataset.