Document Type


Publication Date


Publication Title

BMC Health Services Research


Background: Veterans who are hospitalized in both VA and non-VA hospitals within a short timespan may be at risk for fragmented or conflicting care. To determine the characteristics of these “dual users,” we analyzed administrative hospital discharge data for VA-enrolled veterans of any age in seven states, including any VA or non-VA hospitalizations they had in 2004 – 2007. Method: For VA enrollees in Arizona, Iowa, Louisiana, Florida, South Carolina, Pennsylvania, or New York in 2007, we merged 2004 – 2007 discharge data for all VA hospitalizations and all non-VA hospitalizations listed in state health department or hospital association databases. For patients hospitalized in 2007, we compared those younger or older than 65 years who had one or multiple hospitalizations during the year, split into users of VA hospitals, non-VA hospitals, or both (“ dual users ”), on demographics, priority for VA care, travel times, principal diagnoses, co-morbidities, lengths of stay, and prior (2004 – 2006) hospitalizations, using chi-square analysis or ANOVA. Multiply hospitalized patients were compared with multinomial logistic regressions to predict non-VA and dual use. Payers for non-VA hospitalizations also were compared across groups. Results: Of unique inpatients in 2007, 38% of those 65 or older were hospitalized more than once during the year, as were 32% of younger patients; 3 and 8%, respectively, were dual users. Dual users averaged the most index-year (3.7) and prior (1.5) hospitalizations, split evenly between VA and non-VA. They also had higher rates of admission for circulatory diseases, symptoms/signs/ill-defined conditions, and injury and poisoning, and more admissions for multiple diagnostic categories; among younger patients they had the highest rate of mental disorders admissions. Higher income, non-rural residence, greater time to VA care, lower VA priority, prior non-VA hospitalization, no prior VA hospitalization, and several medical categories predicted greater non-VA use. Among younger patients, however, mental disorders predicted more dual use but less exclusively non-VA use. Dual users’ non-VA admissions were more likely than others’ to be covered by payers other than Medicare or commercial insurance. Conclusions: Younger dual users require more medical and psychiatric treatment, and rely more on government funding sources. Effective care coordination for these inpatients might improve outcomes while reducing taxpayer burden.