2026 Theses Doctoral

Social Determinants of Physical Activity Across Diverse Populations

Lopez Veneros, David

Physical activity (PA) is widely recognized as a key modifiable factor for promoting mental and physical health, quality of life, and overall well-being. Despite well-established evidence of the benefits of PA, substantial disparities in PA persist across population subgroups, including racial and ethnic minority adults and sexual and gender minority (SGM; lesbian, gay, bisexual, transgender) individuals. Growing evidence suggests that gender minority (GM; transgender, nonbinary) adults engage in lower levels of PA than their cisgender (i.e., non-GM) counterparts. Despite this, the multilevel social and psychological determinants of PA among SGM populations remain poorly understood. The aim of this dissertation was to advance understanding of PA disparities among diverse populations – with particular attention to GM individuals – through three complementary studies guided by the socioecological model and the minority stress model.

In the first study (Chapter 2), a systematic review of 24 quantitative, cross-sectional articles representing 13 unique datasets examined differences in PA prevalence between GM and cisgender adults and identified factors associated with PA among GM adults. The majority of studies found lower levels of aerobic and muscle-strengthening PA among GM adults compared to cisgender adults. Transgender women showed the most consistent evidence of lower PA engagement, while nonbinary adults reported higher PA than both transgender women and men, highlighting meaningful heterogeneity within the GM population. Only three articles highlighted factors associated with lower PA among GM adults, including lower social support, negative self-perceptions, higher depression and stress, and lower resilience. The overall quality of evidence was rated as low based on the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) criteria, due to the observational, cross-sectional designs of all of the included studies. Additional methodological limitations across the literature included an overreliance on non-validated self-report PA measures and limited racial and ethnic diversity across study samples.

In the second study (Chapter 3), k-means clustering, an unsupervised machine learning technique, was applied to data from the All of Us Research Program (N = 8,406) to examine how two dimensions of social connectedness – social support at the interpersonal level and neighborhood social cohesion at the community level – jointly influenced objectively measured PA. Three distinct social connectedness clusters were identified: (1) a cluster characterized by high social support and high neighborhood social cohesion, (2) a cluster characterized by high social support and low neighborhood social cohesion, and (3) a cluster characterized by low social support and low neighborhood social cohesion. In fully adjusted logistic regression models, adults in the two clusters characterized by low neighborhood social cohesion – one with high and one with low social support – had significantly lower odds of meeting recommended guidelines for moderate-vigorous PA (MVPA) and had lower odds of walking ≥7,500 steps/day compared to those with high social connectedness across both levels. Approximately half of participants reported high social support alongside low neighborhood social cohesion, and participants in this cluster still had lower odds of meeting MVPA guidelines. Findings suggest that interpersonal social support alone is insufficient to promote optimal PA engagement. After conducting moderations analyses, we found that the associations of social connectedness with PA did not differ significantly by gender identity, sexual orientation, or race and ethnicity.

In the third study (Chapter 4), we examined the associations between three GM-specific stressors – GM-related discrimination, stigma consciousness, and internalized transphobia – measured at Wave 2 and self-reported MVPA measured approximately 1 year later at Wave 3 among GM individuals. These analyses used data from Waves 2 and 3 of Project AFFIRM, a multisite, racially and ethnically diverse cohort of GM individuals living in New York, San Francisco, and Atlanta (N = 265). In fully adjusted models, internalized transphobia at Wave 2 was significantly associated with lower minutes of MVPA and lower odds of meeting MVPA guidelines at Wave 3. GM-related discrimination and stigma consciousness were not significantly associated with PA. Neither social support nor transgender congruence moderated these associations.

Overall, the findings of this dissertation highlight important social determinants of PA among diverse populations, including SGM individuals. GM adults consistently engaged in lower levels of PA than their cisgender counterparts. Further, both neighborhood social cohesion and internalized transphobia are potentially modifiable factors associated with PA engagement among diverse populations. These results support the need for multilevel interventions for PA that incorporate gender-affirming mental health care and transgender community peer support to address internalized transphobia, community-level investment in increasing neighborhood social cohesion, and policies that protect SGM individuals’ access to safe and inclusive PA environments and gender-affirming care.

The findings also point toward an opportunity to extend the minority stress model by more explicitly theorizing how various dimensions of minority stress, such as experiences of discrimination and internalized stigma, may differentially influence health behaviors such as PA. Specifically, our findings suggest that internalized stigma – rather than experienced discrimination or stigma consciousness – may be the dimension of minority stress most directly linked to PA engagement and potentially other health behaviors among GM individuals.