health disparities


We found differences—but not those we expected—in the adjusted risk of death by race and ethnicity. We found that the adjusted risk of death for African-American discharges was not significantly different from that of white discharges, but that Asian discharges had a significantly lower risk of death compared to whites.

Many studies using hospital administrative data have found racial and ethnic disparities in morbidity and mortality in other disease groups. As noted by the recent Institute of Medicine review of racial and ethnic health disparities, the majority of the evidence for racial and ethnic disparities in health outcomes is derived from studies utilizing administrative data for which risk adjustment for comorbid disease and other patient characteristics was often inadequate. We assessed the independent effect of race and ethnicity after comprehensive adjustment for comorbid illness and other baseline patient characteristics by including all diagnoses reported as present on admission, grouped into categories of comorbid disease and conditions related to aspiration pneumonia.

Our multivariable logistic regression model contained 223 predictor variables for comorbid disease categories and an additional 36 predictor variables for categories of conditions closely related to aspiration pneumonia. This comprehensive adjustment for baseline characteristics improved predictive accuracy over existing risk adjustment models; however, this improved statistical performance was not without tradeoffs. Specifically, the CCS groups were dichotomized into either comorbid categories or categories of closely related conditions—an “either/or” decision that did not accommodate “gray” areas. For example, the CCS category “septicemia” included ICD-9 codes that some panel members scored as “equivocal,” meaning they may be related to the principal diagnosis of aspiration pneumonia. However, the overall CCS category, like the few others that included ICD-9 codes considered “equivocal,” was modeled as most likely a comorbid condition.

We found differences in the frequencies of comorbidities across racial and ethnic categories. Except for thyroid disorders, coronary atherosclerosis, and cardiac dysrhythmias, all the comorbid illnesses we studied were more prevalent among African Americans and Asians than among whites. Among Hispanics, all the comorbidities except thyroid disorders were more prevalent.

Given these differences, we considered the possibility that the effects of race and ethnicity on mortality may be expressed indirectly through differences in the prevalence and effects of comorbid diseases. Seven of the 10 most commonly occurring comorbid diseases in the study population with statistically significant differences in prevalence by race or ethnicity were more prevalent among African Americans and Asians than among whites and more prevalent among Hispanics than among non-Hispanics. However, when we tested the statistical significance of the interactions between these common comorbid diseases and racial and ethnic categories, we failed to find any that were statistically significant. We also examined these relationships in subgroups of the population. Though some of the differences found in our subgroup analyses were statistically significant, these may have occurred by chance because of the multiple comparisons considered in the sequences of the multivariable logistic regression model.
buy antibiotics canada


Several limitations in the study attend our findings. Differences in the prevalence and incidence rates of aspiration pneumonia in the general population may be different by race and by ethnicity. This possibility is not addressed in this study of hospitalized patients because the baseline population rates are unknown.

This study included only California hospitalization and may not be generalizable to other populations. Hospital administrative data may be incomplete regarding the list of comorbid conditions, and using such lists may still overlook baseline differences in health status. Our measures of mortality risk did not include adjustments for potentially important socioeconomic factors, such as occupation, income, or social.

Our use of in-hospital death as an outcome measure is also limited by potential systematic differences by race and ethnicity in hospital length of stay (LOS) that could bias results. While there were no statistically significant differences in mean length of stay by race, we did find statistically significant differences by ethnicity. However, we found the mean LOS for Hispanics was significantly less than that for non-Hispanics, which strengthens our finding that in-hospital mortality for Hispanics was lower than that for non-Hispanics.
The study populations in our supplemental analysis of the effects of comorbid disease in race and ethnic subgroups include small numbers of cases, and these subgroups may be too small to generate statistically significant results. Finally, other factors, such as choice of antibiotics, hospital characteristics, variances in physician practices, and other unmeasured or unknown influences, may have affected the risk of inpatient death in this population.
generic viagra online


Racial and ethnic disparities in hospital mortality have been reported in many disease groups. While other research has demonstrated that patients hospitalized with aspiration pneumonia are at high risk of death, the effects of race on this risk had not been addressed. Our research indicates that compared to whites, Asians have a lower risk of death, and the risk of death for African Americans is not significantly different than that of whites. Our research also demonstrates that while the prevalence of comorbid diseases varies by race and ethnicity, these differences do not appear to confound the effect of race on mortality risk. Race- and ethnic-specific effects of comorbid diseases on the risk of death were not statistically significant.

Reviews of the evidence for racial disparities in studies of other disease groups using administrative data suggest that this evidence is limited by the adequacy of adjustments for comorbid disease and other patient characteristics. The adequacy of mortality risk adjustment in studies of racial and ethnic disparities is a key methodological issue that deserves greater attention. We plan to assess the effects of race and ethnicity on the risk of in-hospital mortality in discharges hospitalized with other conditions for which the risk of inpatient death is high, including acute myocardial infarction and lung cancer, where prior research has shown racial and ethnic disparities in mortality outcomes. Optimal adjustments for comorbid disease will improve the quality of evidence about racial and ethnic disparities in in-hospital mortality. viagra soft