Race, Quality of Care and Patient Outcomes: What Can We Learn from the Department of Veterans Affairs?

Article excerpt


Research over the last 20 years has shown persistent racial differences in health care access, utilization, and outcomes. Numerous studies have documented that minorities, particularly African Americans, experience higher mortality and increased hospitalization rates for acute and chronic conditions (Gaskin and Hoffman 2000; Davis et al. 2003; Institute of Medicine 1999, 2002). Facility segregation is one of the hypotheses that may explain these facts. Several recent studies (Ordin and Krumholz 2003; Baicker et al. 2004; Barnato et al. 2005; Skinner et al. 2005) investigate the link between the level of segregation in the health care facility and minority health outcomes, but reach conflicting conclusions. One of the main difficulties in interpreting empirical analyses of this phenomenon is that the degree of health care facility segregation and the socio-economic and health status of the patient population may be jointly determined by an unobserved factor.

This paper offers several contributions. It is the first study to investigate the performance of the Department of Veterans' Affairs (DVA) outpatient community care clinics. Community based clinics were opened in the late 1990s and early 2000s in the largest expansion of outpatient care for veterans since the inception of the DVA. These clinics offer free health care for all veterans. In addition, I employ an instrumental variables approach to examine the causal effect of racial segregation at the health facility on patient outcomes. Time variation in clinic openings during the reform of the DVA provides an instrument for changes in the racial profiles of patients served by existing outpatient clinics. Using an instrumental variable estimation overcomes the omitted variable bias that may have affected previous attempts to isolate the pure effect of an increase in the proportion of minority patients on the quality performance of health facilities. Finally, I link outpatient clinical data to demographic data, and am able to follow the same patients and facilities for up to 6 years. This allows me to work with within-facility variation in the racial profile of patients, controlling for characteristics of the facility that are fixed over time such as differences in the quality of care due to geographic location.

Several additional advantages come from the nature of the health care system serving the population studied. Universal access and reduced financial incentives for physicians at the VA allow me to avoid common pitfalls with private health data such as underutilization because of scarce resources, or under provision of care by physicians based on lower socio-economic status (SES). The VA operates inpatient, specialty, and outpatient facilities. The highly integrated nature of the health organization ensures that physicians' incentives to "outsource" some services to private practices are greatly reduced. Finally, the demographic data provide a larger set of controls than other national patient datasets.

I show that within this highly integrated, free access health care system, a higher proportion of African-American patients within a clinic improves individual outcome for all patients. This result is robust to controlling for cross-sectional differences between clinics and is confirmed by results from the instrumental variables estimation. Several possible explanations for this finding are offered in the text.

The analysis focuses on a national sample of all patients diagnosed with Chronic Heart Failure (CHF) who accessed a VA outpatient clinic in the period 1998-2004. Patients suffering from CHF are a particularly appropriate study group for several reasons. First, heart disease is the leading cause of death in the Medicare population. It has become the largest Medicare expense in recent years. Cardiovascular disease is a major contributor to the mortality difference between white Americans and African Americans. …