Data Management and Analysis

We audio recorded and took written notes during interviews and focus groups and transcribed them verbatim. We used NVivo qualitative software for data management (QSR International). We recorded demographic data in Excel (Microsoft). Using a constant comparative approach, we reviewed transcripts (which were not returned to participants) with clinical investigators (T.D.G. and J.M.K.) to inductively develop a codebook for thematic content analysis. Once the codebook was developed (see data supplement), two researchers (K.J.R. and E.A.C.) cocoded transcripts, resolving by consensus any coding that fell below 90% agreement, until percent coding agreement reached 98% (kappa = 0.81). Following acceptable intercoder agreement, coding proceeded independently. The codebook consisted of themes, subthemes, and definitions (e.g., theme of risk-management strategies with subthemes of extubate, not extubate, and rationale). The results are presented as themes and subthemes with supporting quotes. As outlined in the data supplement, we used the Consolidated Criteria for Reporting Qualitative Studies (23).