To the Editor.
—Dr Peterson and colleagues1 report that significant differences in the utilization of invasive cardiac procedures between white and black patients treated in the Department of Veterans Affairs (VA) hospitals persist while controlling for confounding variables. In response to this study, Jesse Brown, secretary of the VA, stated his concern about possible "racial bias in the delivery of care at VA hospitals."2 However, since several important potential confounders could not be included in their regression models because of limitations of the database, the implication that racial biases influence the decision-making process in the VA is unwarranted.Their list of patient characteristics does not include a variable for smoking status, any indicator of socioeconomic status (SES) such as income or education, or any specific indicators of comorbidities such as pulmonary disease. Differences in the pattern of any of these characteristics between white and black patients within the