The Relative Importance of Unmeasured Covariates in Racial/Ethnic Disparities Research

To the Editor:

Dr. Barnato and colleagues (1) present results from a population-based analysis of race-specific incidence and case fatality in severe sepsis. Using hospital discharge data from six states, the authors determined that the rate of severe sepsis for blacks was almost double that of whites. We commend the authors for documenting this racial/ethnic disparity. Identifying disparities in health outcomes and treatment effects is an important first step in eventually curtailing their occurrence. An important subsequent step involves appropriately identifying the fundamental causes of such disparities. It is here that we take issue with the article.

Confronted with a residual association between race/ethnicity and sepsis incidence after adjusting for measured confounders, the authors acknowledge that unmeasured confounders may be the cause. This is the most obvious conclusion given the crude indicators of social status used in their analysis. However, they are more enthusiastic about a second possibility: unexplained racial disparities in sepsis incidence could be due to “different genetic susceptibility to sepsis between blacks and whites.” The common argument that, after accounting for social/environmental factors, genetic differences are responsible for residual racial disparities in complex diseases is flawed for several reasons (2–4). From a purely probabilistic standpoint, even in situations where social factors can be measured without error, and are appropriately accounted for in the analysis, social/environmental explanations for the higher rates of sepsis among black patients are far more likely than genetic ones (2, 5). This is because the variability in social/environmental factors that influence health across racial/ethnic groups (e.g., education, health insurance, poverty) is orders of magnitude greater than any genetic variability between groups (4). In other words, race is likely to be much more highly correlated with social/environmental factors than with genetic factors.

The authors' enthusiasm for future research efforts that would determine the genetic basis of racial differences in the pathophysiology of sepsis ignores the fact that race serves as a risk factor for both genetic and environmental effects. Issues of confounding, coupled with the weak correlation between human genetic diversity and racial/ethnic categories, will ultimately limit the promise of such research efforts (6).

While we appreciate the assertion that “hypotheses regarding potential race-associated differences in innate susceptibility do not belie the [importance of] differences in healthy lifestyles,” we feel that the purported genetic and biologic differences between blacks and whites with sepsis are mainly conjecture and detract from the known social and environmental causes that likely explain the vast majority of the difference.