Influence of contextual factors on death by suicide in rural and urban settings

To examine the association between contextual factors, represented by geographic and community health variables, and suicide among rural and urban Department of Veterans Affairs health care users (VA users).

We performed a retrospective cohort study of 12,700,847 VA users between 2003 and 2017. We assigned contextual factors based on individuals' home address at the ZIP Code (area deprivation), county (sunlight exposure, altitude, and community health), and state level (firearm ownership), using publicly available data sources. We grouped contextual factors by quintiles or prespecified thresholds, depending on the nature of each variable. We obtained mortality data from the National Death Index. We measured the effect of living in a place with the highest versus lowest level of each contextual factor on odds of suicide using logistic regression, adjusting for individual compositional factors abstracted from VA electronic medical records data. We used random forest modeling to build prediction models for suicide based on contextual factors among rural and urban veterans.

Almost all contextual factors we examined were significantly associated with suicide among rural and urban VA users, even after adjusting for individual compositional factors. However, no contextual variables were strong protective or risk factors (0.52.0), and prediction models leveraging these contextual factors had poor accuracy among both rural (0.51, 95% CI: 0.48-0.54) and urban (0.53, 95% CI: 0.51-0.55) VA users.

A wide variety of contextual factors is significantly associated with suicide among rural and urban VA users. However, the factors we measured contributed very little to individual-level suicide risk.