Making structure visible: contextual analysis of stigma in semi-structured interviews

Background: The negative social and health impacts of stigma against people with substance use disorders and other mental health conditions are well-documented. Yet qualitative studies rarely specify how stigma is operationalized in the coding and analysis of interview transcripts, which may lead to underestimating the prevalence of structural stigma and miscoding clinically relevant data.Methods: We built on a thematic analysis of 27 interviews with 29 administrators of skilled nursing facilities (SNFs) concerning barriers and facilitators to SNF admission for people with opioid use disorder. Where coding of stigma was disputed among team members, we triangulated passages using within-interview contextualization, reference interviews selected by five observable criteria, and external evidence.Results: We identify three modes in which stigma presented in semi-structured qualitative interviews: directly observable stigma, contextualized clinical stigma, and hidden structural stigma. We explore how dissensus among team members with different positionalities, including team members with lived experience, can inform data analysis.Conclusions: This paper provides methodological guidance on using triangulation to identify structural dimensions of stigma in semi-structured interviews analyzed using thematic analysis and related approaches, contributing to a more rigorous operationalization of stigma as a multi-level construct in qualitative research. Our findings demonstrate the possibility of identifying reference interviews which provide contextualized, patient-centered accounts, and triangulating them within and across interviews and with external evidence to make structural stigma legible. The approach we present is particularly suited to interdisciplinary teams that include members with lived experience, clinical knowledge, and advocacy perspectives, whose divergent readings of the same data can be treated as an analytic resource.