Harnessing national data systems to understand circumstances surrounding veteran suicide: linking Department of Veterans Affairs and National Violent Death Reporting System Data.

Abstract

Background: Veterans are at elevated risk for suicide compared to non-Veteran U.S. adults. Data sources and analyses to inform prevention efforts, especially for those who do not use Department of Veterans Affairs (VA) healthcare services, are needed. This study aimed to link VA and CDC's National Violent Death Reporting System (NVDRS) data to create a novel data source to characterize the circumstances precipitating and preceding suicide among Veterans, including among those who did not use VA healthcare.

Methods: Multi-variable, multi-stage, deterministic linkage of VA-Department of Defense (DoD) Mortality Data Repository (MDR) and NVDRS-Restricted Access Database suicide and undetermined intent mortality records within 189 state-year strata (42 states, 2012-2018). Three linkage stages: (1) exact (matched on: age, sex, death date, underlying cause of death, day of month of birth, first initial of last name); (2) probable (all but one variable matched); (3) possible (all but 2 variables matched). Linkage success and accuracy of NVDRS-documented military history were assessed.

Results: Across all state-years, 22,019 matches (89.20% of 24,685 MDR Veteran records) were identified (65.47% exact). When high missingness (2 + matching variables in > 10% of records; n = 23) or incomplete reporting (n = 12) state-years were excluded, match rate increased to 94.29% (77.15% exact). NVDRS-documented military history (ever served) was accurate for 87.79% of matched records, with an overall sensitivity of 84.62%. Sensitivity was lower for female (61.01%) and younger (17-39 years; 77.51%) Veterans.

Conclusions: Accurate linkage of VA-DoD and NVDRS data is feasible and offers potential to improve understanding of circumstances surrounding suicide among Veterans.