X. Zou, J. Shi
Background: Marital status at diagnosis may capture aspects of social support, but it is not a direct measure of support quality. This study examined its association with suicide mortality after breast cancer diagnosis while accounting for competing mortality and exploring variation by county-level income and rural-urban residence. Methods: Adults of both sexes diagnosed with first primary breast cancer in SEER between 2000 and 2022 were studied using a baseline complete-case cohort. Marital status was classified as married/partnered or unmarried/non-partnered. Stabilized inverse probability of treatment weighting (sIPTW) was combined with multivariable Fine-Gray regression, treating non-suicide death as a competing event. Cox, landmark, exploratory subgroup, interaction, and unweighted treatment-adjusted sensitivity analyses were performed. Results: The analytical cohort included 825,047 patients, of whom 40.7% were unmarried/non-partnered. All baseline standardized mean differences were below 0.01 after weighting. During 6,816,981.42 person-years, 506 suicide deaths occurred (7.42 per 100,000 person-years). Unmarried/non-partnered status was associated with higher suicide mortality in the primary Fine-Gray model (sHR 1.34, 95% CI 1.12-1.60) and Cox model (HR 1.45, 95% CI 1.21-1.73). Landmark sHRs at 1, 3, and 5 years were 1.39, 1.46, and 1.34, respectively. The treatment-adjusted sensitivity estimate was 1.26 (95% CI 0.99-1.60; P = 0.064). Income interaction analysis suggested heterogeneity (joint P = 0.042), but the Q4 conditional sHR was imprecise (1.06, 95% CI 0.80-1.39). There was no statistically significant rural-urban interaction (P = 0.488). Conclusions: Unmarried/non-partnered status was associated with higher suicide mortality after breast cancer diagnosis, although the absolute rate was low. The observational design, sensitivity-analysis uncertainty, and exploratory income interaction warrant cautious interpretation and independent validation.