Investigating Gendered Implicit Bias Among Healthcare Workers in South Africa: Development and Piloting of an Implicit Association Test in a High Tuberculosis and HIV Setting

Background In sub-Saharan Africa, men with HIV and/or tuberculosis (TB) report negative experiences within health services and demonstrate poorer engagement and treatment outcomes compared to women. We developed and piloted an Implicit Association Test (IAT) to assess healthcare workers’ (HCWs’) gender-based associations characterizing patients with the concepts of “good” versus “bad.” Methods Standardized photographs of Black male and female faces from the Chicago Face Database were paired with locally derived descriptors of “good” and “bad” patients. The IAT was piloted among nurses providing TB or TB/HIV services in government clinics in Eastern Cape Province, South Africa. Following IAT administration, focus groups explored user experience and elicited discussions of “good” versus “bad” patients and perceptions of male and female patients. IAT data were analyzed using D -scores; focus group fieldnotes and artifacts were analyzed using a descriptive qualitative approach. Results Eleven nurses participated and all completed the IAT, with 98% of trials retained after standard processing, indicating strong task engagement. Participants described the tool as understandable and usable, although some reported initial anxiety about responding “correctly.” The mean D -score was 0.234 (SD = 0.492), indicating a small relative association between male patients and negative attributes, though this association did not differ significantly from zero (95% CI: − 0.096, 0.565; p  = 0.146). Qualitative findings revealed explicit characterizations of men as more likely to exhibit “bad” patient behaviors, aligning directionally with quantitative patterns. Conclusions This pilot study demonstrates that a locally developed gender-based IAT can be feasibly administered among public-sector health workers in South Africa. Although the quantitative signal was small and not statistically significant, qualitative findings revealed explicit gendered evaluative perceptions, supporting the instrument’s internal face validity. This tool provides a contextually grounded approach to examining how gendered bias may shape health service delivery and establishes a foundation for larger-scale validation and implementation research.