Primary Submission Category: Design-Based Causal Inference
Causal Effects of Health-Related Social Needs on Mammography Screening: Propensity-Score Weighting in Complex Survey Data
Authors: Fode Tounkara,
Presenting Author: Fode Tounkara*
Screening mammography lowers breast cancer mortality, yet many women in the United States remain overdue for recommended screening. Uptake is especially low among women facing social and economic hardship. Health-related social needs (HRSNs), such as housing instability, food insecurity, and transportation barriers, can interfere with preventive care by limiting access, increasing stress, and competing with daily survival needs. Although these factors are widely recognized, their cumulative impact on mammography use has rarely been evaluated using causal methods in nationally representative data.
We analyzed data from the 2023 National Health Interview Survey to examine whether unmet HRSNs are associated with being up to date on screening mammography among women aged 40 to 74 years. Both individual social needs and a cumulative burden index (0, 1, or ≥2 unmet needs) were assessed. To address confounding while accounting for the complex survey design, we applied inverse probability of treatment weighting (IPW) integrated with survey weights, stratification, and clustering. Covariate balance was evaluated using standardized mean differences. Weighted logistic regression models estimated marginal associations overall and within age groups (40–49, 50–64, and 65–74 years).
After weighting, covariate balance improved substantially across exposure groups. Unmet social needs were associated with lower odds of guideline-concordant mammography, with stronger effects observed as the num