Data Analysis
Audio recordings of the interviews were transcribed in their language of
origin using NVivo Transcription and cleaned to remove participant
identifiers. Each transcript was inductively coded by two to three study
team members, including bilingual research assistants and the PI, using
a conventional content analysis approach.37 All
interviews were coded in their language of origin to increase
trustworthiness and prevent loss of meaning during
translation;38,39 the interviewer was included in each
coding team. The coders held standing meetings with the PI to establish
the codebook, compare coding structures and categories, and achieve
intercoder agreement. As guided by qualitative descriptive
methodology,26,40 efforts were made to remain close to
the data. Discordance in coding was resolved through team discussions;
the team employed a reflexive approach during their coding, individually
and as a team.41 After initial coding was completed
and saturation was reached across cross-cutting themes, the coding team
reviewed the coded transcripts to explore similarities or differences by
participant role and language.42 Saturation tables
were created and reviewed to ensure agreement of codes and
categories.43 Interpretations of the data were
enhanced through consultation of the notes taken during member checking
discussions to confirm the findings fit the experiences of the
participants.44