Health Services Research & Development

Veterans Crisis Line Badge
Go to the ORD website
Go to the QUERI website

2006 HSR&D National Meeting Abstract

1008 — Death, Where is Thy Record? Death Ascertainment for Veterans

Author List:
Arnold NL (VIReC)
Sohn M (VIReC)
Hynes DM (VIReC)
Maynard C (Seattle ERIC)

Mortality is one of the most frequently studied outcomes in health services research and clinical trials. In this study we (1) examined the completeness and reliability of death dates found in VA data sources, (2) determined the costs and benefits associated with use of alternative sources for death ascertainment, and (3) developed a strategy for building a central database or registry of death data that would be used by VA researchers for death ascertainment.

A finder file of over eight million veterans was linked to the BIRLS Death File, the SSA Death Master File, the VHA Medical SAS Inpatient Datasets, and the Medicare Vital Status File to obtain death dates from each source. These death dates were compared for consistency. Stratified random samples were sent to the National Death Index (NDI) to resolve differences in death dates among the sources and assess accuracy of the different sources. We evaluated the value added of using the NDI. In addition, a presumed living status was created, and random samples were sent to the NDI and the SSA Epidemiological Vital Status Search to validate its accuracy.

By combining the four sources available to VA researchers, we achieved a sensitivity of 98.3% when compared to the NDI. The sensitivity of each source alone compared to the NDI was: 77.4% for the BIRLS Death File; 83.2% for the Medicare Vital Status File; and 92.1% for the SSA Death Master File. There were no deaths identified in the random samples of veterans categorized as presumed living sent to the NDI and the SSA.

Use of the NDI has limited value for death ascertainment over combining currently available death sources and using a presumed living status. VIReC is currently working with National Data Systems to build a Death Registry of the combined data for use by researchers and VA operations.

Results of this study provide research with a standard methodology for death ascertainment that allows for easier and more complete death ascertainment.