© 2001 Elsevier Science Inc. All rights reserved. 1047-2797/01/$–see front matter 655 Avenue of the Americas, New York, NY 10010 PII S1047-2797(01)00211-3 INTRODUCTION Department of Veterans Affairs (VA) national databases contain extensive information about health care services received by veterans in VA facilities (1). A distinctive fea- ture of VA databases is that they contain information on veterans’ current vital status. This important information can be used scientifically to study a variety of questions of interest to epidemiologists and health services researchers, such as mortality among veterans with lung disease (2) or colorectal cancer (3). It is crucial that researchers understand the validity and limitations of this vital status information when studying mortality. An early study of VA records determined that 95% of 779 deaths among veterans known to have died in 1962 were recorded in the file (4). Since then, others have studied the VA death files, usually comparing them to the National Death Index or the Social Security death files (5– Assessment of Vital Status in Department of Veterans Affairs National Databases: Comparison with State Death Certificates JASON A. DOMINITZ, MD, MHS, CHARLES MAYNARD, PhD, and EDWARD J. BOYKO, MD, MPH PURPOSE: To determine the extent to which Department of Veterans Affairs (VA) database vital status information agrees with Washington state death certificates. METHODS: Using each data source, vital status was determined for 19,481 Washington state resident veterans hospitalized in Washington VA hospitals from 1994 to 1997, and for 33,602 Washington state resident veterans who were seen as outpatients during 1997. RESULTS: The agreement between VA and Washington state records was excellent for hospitalized vet- erans (kappa = 0.91, p  0.0001). Three thousand one hundred-eight individuals (86.2% of all deaths) appeared in both files. Of those deaths missing in the VA files, 71% had no service-connected disability, VA pension, or other compensation. Among outpatients, agreement between the death files was very good (kappa = 0.82, p  0.001). Three hundred seventy-two individuals (69.8% of all deaths) appeared in both files. Of those deaths missing in the VA files, 63% had no service-connected disability or VA pen- sion or other compensation. CONCLUSIONS: The VA death files are a valid source of vital status information for veterans hospital- ized in recent years. For veterans having exclusively outpatient visits, however, the VA files miss a sub- stantial proportion of deaths. For these patients, alternative means of vital status ascertainment are warranted. Ann Epidemiol 2001;11:286–291. © 2001 Elsevier Science Inc. All rights reserved. KEY WORDS: Databases, Death Certificate, Mortality, Veterans, Vital Statistics. 10). A study of veterans who received cardiac care in both Veterans Affairs and Medicare hospitals reported that there was 98% agreement with respect to the reporting of deaths in the VA and Medicare (11). The purpose of this study is to determine the extent to which VA vital status information agrees with Washington state death certificates. MATERIALS AND METHODS Study Population This study was approved by the University of Washington Human Subjects Committee and the VA Puget Sound Health Care System Research and Development Commit- tee. Two study populations were compiled. First, from the Patient Treatment File (PTF), there were 24,929 individu- als who either died during hospitalization or were dis- charged from one of three Washington state VA medical centers between January 1, 1994, and December 31, 1997. We excluded 5448 individuals who, according to the PTF, were not Washington state residents. This resulted in 19,481 Washington state resident veterans who were hospi- talized in VA facilities. Second, using the VA outpatient clinic file (OPC), we identified 33,602 unique Washington state resident veterans who were seen as outpatients in Washington state VA clinics during fiscal year 1997. This From the Department of Medicine, University of Washington School of Medicine, Seattle, WA (J.A.D., E.J.B.); the Veterans’ Administration Ep- idemiologic Research and Information Center, Seattle, WA (J.A.D., C.M., E.J.B.); and the Department of Health Services, University of Washington, Seattle, WA (C.M.). Address reprint requests to: Jason A. Dominitz, MD, MHS, VAPSHCS, 111-S-Gastro, 1660 S. Columbian Way, Seattle, WA 98108-1597. Received June 23, 2000; revised November 9, 2000; accepted November 13, 2000.