Development and testing of study tools and methods to examine ethnic bias and clinical decision-making among medical students in New Zealand: The Bias and Decision-Making in Medicine (BDMM) study

Abstract

Background: Health provider racial/ethnic bias and its relationship to clinical decision-making is an emerging area of research focus in understanding and addressing ethnic health inequities. Examining potential racial/ethnic bias among medical students may provide important information to inform medical education and training. This paper describes the development, pretesting and piloting of study content, tools and processes for an online study of racial/ethnic bias (comparing Māori and New Zealand European) and clinical decision-making among final year medical students in New Zealand (NZ). Methods: The study was developed, pretested and piloted using a staged process (eight stages within five phases). Phase 1 included three stages: 1) scoping and conceptual framework development; 2) literature review and identification of potential measures and items; and, 3) development and adaptation of study content. Three main components were identified to assess different aspects of racial/ethnic bias: (1) implicit racial/ethnic bias using NZ-specific Implicit Association Tests (IATs); (2) explicit racial/ethnic bias using direct questions; and, (3) clinical decision-making, using chronic disease vignettes. Phase 2 (stage 4) comprised expert review and refinement. Formal pretesting (Phase 3) included construct testing using sorting and rating tasks (stage 5) and cognitive interviewing (stage 6). Phase 4 (stage 7) involved content revision and building of the web-based study, followed by pilot testing in Phase 5 (stage 8). Results: Materials identified for potential inclusion performed well in construct testing among six participants. This assisted in the prioritisation and selection of measures that worked best in the New Zealand context and aligned with constructs of interest. Findings from the cognitive interviewing (nine participants) on the clarity, meaning, and acceptability of measures led to changes in the final wording of items and ordering of questions. Piloting (18 participants) confirmed the overall functionality of the web-based questionnaire, with a few minor revisions made to the final study. Conclusions: Robust processes are required in the development of study content to assess racial/ethnic bias in order to optimise the validity of specific measures, ensure acceptability and minimise potential problems. This paper has utility for other researchers in this area by informing potential development approaches and identifying possible measurement tools.