e19113 Background: The risk of suicide among cancer survivors more than double that of the general population, highlighting the need to mitigating risk factors for suicide. While several studies have described marital status, a surrogate for social support, as associated with cancer mortality, it is inconclusive whether marital status impacts suicide as a competing cause of cancer mortality. We tested this hypothesis by describing the association of marital status and suicide among survivors of four cancer sites with the highest suicide mortality rates in the United States. Methods: Adult cancer patients were identified from the Surveillance, Epidemiology and End Results database from 2004 to 2016 for four index cancer sites previously identified with highest suicide mortality rates: pancreas, head and neck, lung/bronchus and stomach ( n = 800,798). Cumulative incidence curves stratified by marital status (divorced/separated, widowed, never unmarried, and married/partnered) estimated unadjusted probability of suicide (outcome of interest). A multivariable competing risk proportional hazards model yielded sub-distribution hazard ratios (sdHRs) and 95% confidence intervals (CI) to estimate the association of marital status with suicide for each cancer site, while controlling for clinical and nonclinical factors. Results: Half (50.7%) of the cohort were married/partnered, males (56.8%), and non-Hispanic whites (71.0%). Mean age at diagnosis was 67.3 years. Most patients (60.9%) had cancer in the lung/bronchus, 17.9% head and neck, 13.8% pancreas, and 8.3% stomach. Unadjusted probability of suicide was highest among head and neck cancer survivors (0.3%). In the fully adjusted model, mortality by suicide was more likely among divorced/separated patients vs. married/partnered patients across cancer sites (sdHRhead and neck = 1.81; 95% CI 1.38, 2.37; sdHRlung/bronchus = 1.68; 95% CI 1.28, 2.19; sdHRpancreas = 2.19; 95% CI 1.27, 3.78; and sdHRstomach = 2.38; 95% CI 1.17, 4.58). Additionally, for lung/bronchus cancer, patients who were never married patients were more likely to die by suicide than those married/partnered (sdHRlung/bronchus = 1.47; 95% CI 1.09, 1.98). Conclusions: Marital status is associated with suicide mortality among cancer survivors, and divorced/separated survivors may have greater suicide mortality risks, independent of cancer site. As overall probability of suicide remains low, these findings might help identify cancer survivors who may be candidates for ongoing surveillance and psychosocial support to mitigate suicide mortality risks.
Age-Period-Cohort Analysis of Suicide Mortality Rates in Spain, 1959–1991
