Volume 42 | Number 4 | August 2007

Abstract List

Kevin G. Volpp, Roslyn Stone, Judith R. Lave Ph.D., Ashish K. Jha M.D., M.P.H., Mark Pauly, Heather Klusaritz, Huanyu Chen, Liyi Cen, Nancy Brucker, Daniel Polsky


To examine the source of observed lower risk‐adjusted mortality for blacks than whites within the Veterans Affairs (VA) system by accounting for hospital site where treated, potential under‐reporting of black deaths, discretion on hospital admission, quality improvement efforts, and interactions by age group.

Data Sources

Data are from the VA Patient Treatment File on 406,550 hospitalizations of veterans admitted with a principal diagnosis of acute myocardial infarction, stroke, hip fracture, gastrointestinal bleeding, congestive heart failure, or pneumonia between 1996 and 2002. Information on deaths was obtained from the VA Beneficiary Identification Record Locator System and the National Death Index.

Study Design

This was a retrospective observational study of hospitalizations throughout the VA system nationally. The primary outcome studied was all‐location mortality within 30 days of hospital admission. The key study variable was whether a patient was black or white.

Principal Findings

For each of the six study conditions, unadjusted 30‐day mortality rates were significantly lower for blacks than for whites (<.01). These results did not vary after adjusting for hospital site where treated, more complete ascertainment of deaths, and in comparing results for conditions for which hospital admission is discretionary versus nondiscretionary. There were also no significant changes in the degree of difference by race in mortality by race following quality improvement efforts within VA. Risk‐adjusted mortality was consistently lower for blacks than for whites only within the population of veterans over age 65.


Black veterans have significantly lower 30‐day mortality than white veterans for six common, high severity conditions, but this is generally limited to veterans over age 65. This differential by age suggests that it is unlikely that lower 30‐day mortality rates among blacks within VA are driven by treatment differences by race.