METHODOLOGICAL CONSIDERATIONS RELATED TO EQUITY, DIVERSITY, AND INCLUSION IN CLINICAL EPIDEMIOLOGY

Objectives: Understanding how social categories like gender, migration background, lesbian/gay/bisexual/transgender (LGBT) status, education, and their intersections affect health outcomes is crucial. Challenges include avoiding stereotypes and fairly assessing health outcomes. This paper aims to demonstrate how to analyze these aspects. Study Design and Setting: The study used data from N 5 19,994 respondents from the German Socio-Economic Panel 2021 data collection. Variations between and within intersectional social categories regarding depressive symptoms and self-reported depression diagnosis were analyzed. We employed intersectional Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy to assess the impact of gender, lesbian/gay/bisexual/transgender status, migration, education, and their interconnectedness. A Configuration-Frequency Analysis assessed typicality of intersections. Differential Item Functioning analysis was conducted to check for biases in questionnaire items. Results: Intersectional multilevel analysis of individual heterogeneity and discriminatory accuracy analysis revealed significant interactions between these categories for depressive symptoms and depression diagnosis. The Configuration-Frequency Analysis showed that certain combinations of social categories occurred less frequently compared to their expected distribution. The Differential Item Functioning analysis showed no significant bias in a depression short scale across social categories. Conclusion: Results reveal interconnectedness between the social categories, affecting depressive symptoms and depression probabilities. More privileged groups had significant protective effects, while those with less societal privileges showed significant hazardous effects. Statistical significance was found in some interactions between categories. The variance within categories outweighs that between them, cautioning against individual-level conclusions. (c) 2024 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.