**Background:** The COVID-19 pandemic disproportionately affected the LGBTQ community, with sexual minority women (SMW) facing elevated stress, economic hardship, and health disparities. According to the minority stress model, excess stress from marginalization may drive health risk behaviors as coping mechanisms. Social support may buffer or exacerbate these effects, but limited evidence existed on how perceived stress and different types of social support (emotional, material, virtual, in-person) were associated with changes in health behaviors among SMW during the pandemic.
**Methods:** In January 2021, an online convenience sample of 501 SMW (mean age 23.6 years, 76.4% bisexual, 60.1% non-Hispanic White, 57.9% with income <$20,000/year) was recruited via Prolific. Participants completed measures of demographics, the 4-item Perceived Stress Scale (PSS-4; Cronbach's α=0.77), and four questions assessing frequency of receiving emotional, material, virtual, and in-person social support. The primary outcomes were self-reported changes (increased, decreased, or stayed the same) in fruit/vegetable intake, physical activity, sleep, combustible tobacco use, e-cigarette use, alcohol use, and other substance use compared with pre-pandemic levels. Multinomial logistic regression models estimated associations of perceived stress and social support with behavior changes, controlling for sexual orientation, age, race/ethnicity, and income. Interactions between perceived stress and social support were also tested.
**Key Results:** A 1-unit increase in PSS was associated with higher odds of both decreased (OR=1.20, 95% CI: 1.06–1.36, p=0.01) and increased (OR=1.12, 95% CI: 1.00–1.26, p=0.04) fruit/vegetable intake, as well as decreased (OR=1.24, 95% CI: 1.08–1.41, p<0.01) and increased (OR=1.20, 95% CI: 1.06–1.35, p<0.01) sleep, and increased substance use (OR=1.19, 95% CI: 1.01–1.39, p=0.04). In-person support more than weekly was associated with higher odds of both decreasing (OR=10.10, 95% CI: 2.79–36.58, p<0.001) and increasing (OR=7.35, 95% CI: 1.88–28.76, p<0.01) combustible tobacco use, and increasing e-cigarette use (OR=5.96, 95% CI: 1.12–31.77, p=0.04). Emotional support more than weekly was associated with lower odds of increasing alcohol use (OR=0.41, 95% CI: 0.18–0.94, p=0.04). Material support more than weekly was associated with lower odds of increasing alcohol use (OR=0.32, 95% CI: 0.13–0.75, p=0.01) and substance use (OR=0.30, 95% CI: 0.10–0.95, p=0.04). In-person support more than weekly was associated with higher odds of increasing alcohol use (OR=2.63, 95% CI: 1.23–5.60, p=0.01). Virtual support at least monthly or more than weekly was associated with lower odds of both decreasing and increasing fruit/vegetable intake. A significant interaction showed that among SMW who never received material support, increased PSS was associated with higher odds of increasing alcohol use (OR=1.25, 95% CI: 1.07–1.47, p=0.005).
**Clinical Implications:** This study provides the first evidence linking perceived stress and multiple types of social support to health behavior changes among SMW during the COVID-19 pandemic. The mixed findings—where social support was associated with both beneficial and harmful behavior changes—suggest that not all support is equally helpful. In-person support was linked to increased tobacco, e-cigarette, and alcohol use, possibly reflecting social norms around substance use within some networks. Conversely, emotional and material support appeared protective against increased alcohol and substance use. The finding that lack of material support amplified the stress–alcohol use relationship highlights the importance of addressing financial strain. Clinicians and public health interventions should consider the type and source of social support when designing programs for SMW, and aim to bolster supportive networks that promote healthy behaviors while reducing risk behaviors. Limitations include convenience sampling, self-report bias, and small cell sizes for some analyses, which may affect precision and generalizability.