Social
Causal diagrams for sexual and gender minority health disparities Travis Salway* Vandad Hazrati Noor Bainwait Ashleigh Rich Ace Chan Amit Gupta Christoffer Dharma Ayden Scheim Travis Salway
Social epidemiologists have long deliberated the application of directed acyclic graphs (DAGs) in understanding disparities linked to non-modifiable characteristics, such as gender/sex and race. However, these methodological challenges are less frequently explored within the context of sexual and gender minority (SGM) health research. While SGM identity is intrinsic and self-determined, related constructs like status disclosure and minority stress are dynamic and integral to causal pathways influencing health outcomes.
We conducted a scoping review of SGM epidemiological studies indexed in Medline in 2024, resulting in 19 articles that inform our analysis. Drawing on these examples, this talk explores how SGM epidemiologists are adapting DAGs to enhance causal inference in health disparities research. Three key strategies emerge:
- Measurement Bias: SGM status can act as an imperfect proxy for daily disclosure behaviors, influenced by factors like age, race, socioeconomic status, and modality of data collection.
- Stigma as Exposure: External stigma (e.g., harassment or discrimination) can be reframed as modifiable exposure variables. Policies and structural interventions aimed at reducing stigma provide opportunities for addressing disparities, positioning SGM status as a fundamental but not sole causal factor, akin to how racism drives race-related health inequities.
- Effect Modification: Emerging evidence supports the idea that the health effects of generalized stigma vary across SGM subgroups, with certain identities (e.g., bisexual, gay/lesbian) experiencing amplified negative impacts. These variations necessitate subgroup-specific analyses to refine causal interpretations.
Through these examples, we propose actionable solutions for researchers employing DAGs to study SGM health disparities, advocating for greater integration of nuanced data sources and subgroup-specific analyses.
