County Sociodemographic Conditions and Hospital Capacity are Associated with Suicide Mortality in the United States

Objective: To examine associations of county socioeconomic composition and hospital capacity with suicide mortality. Background. Community-level analyses frequently report that hospital capacity is associated with lower suicide mortality. Because hospitals are concentrated in populous counties, and suicide rates differ by settlement scale, such associations may reflect the characteristics of places rather than the effects of capacity. Methods. CDC WONDER county-year deaths assigned underlying-cause ICD-10 codes X60–X84 were linked to American Community Survey measures of county poverty, percentage Black, and income inequality, and to hospital capacity constructed from Medicare cost reports. Because cost reports were available through 2023 and mortality through 2024, the analytic period was restricted to 2011–2023, yielding 12 494 county-years across 1490 counties. Five capacity measures were examined: general and psychiatric hospital presence, general and psychiatric beds per 100,000, and the critical access share. Pooled models included year effects with county-clustered standard errors; fixed-effects models included both county and year effects, with the within transformation applied to all regressors. Sensitivity analyses excluded rates flagged unreliable and restricted the sample to counties of at least 100,000 residents. Results. General hospital presence (−4.35), psychiatric hospital presence (2.36), psychiatric beds per 100,000 (−0.58), and the critical access share (3.40) were each associated with the suicide rate. In pooled models, county poverty was associated with a higher suicide rate (1.35 per 100,000 per standard deviation, 95% CI 0.45 to 2.25), county percentage Black with a lower rate (−1.50, −1.86 to −1.14), and county population with the largest coefficient (−4.91, −5.42 to −4.41). In two-way fixed-effects models, no capacity measure retained an association: general hospital presence 0.52 (95% CI −0.70 to 1.75) and psychiatric beds per 100,000 0.01 (−0.21 to 0.23). Conclusions. Associations between hospital capacity and county suicide mortality observed in pooled models were not reproduced when counties were compared with themselves over time. Capacity indicators based on facility counts appear to track settlement scale. Because within-county variation in capacity was limited, these null results indicate an absence of evidence rather than evidence of no effect.