There is strong evidence that African Americans are at elevated cancer risk, compared to European Americans. In addition, it is generally accepted that African Americans face more aggressive disease, with poorer treatment outcome. Whether these differences reflect biologic factors inherent in lifestyle or genomic characteristics that affect risk and treatment outcome is not well understood. Racial differences in risk could reflect differences in exposure to environmental toxins, to diet, to physical activity or possibly obesity; they could reflect genetically determined physiologic or metabolic processes. Racial differences in treatment outcome could result from compliance with treatment directives, with stage at diagnosis, or with the quality of medical care or with the aggressiveness with which treatment is pursued; they could also be a product of genetically structured physiologic, metabolic or detoxification processes. Racial differences in quality of care could result from differences in the institutions from which African Americans and European Americans receive cancer care.
To the degree that the quality of medical care is largely a function of the institution from which care is delivered, holding the institution constant should eliminate or greatly lessen racial differences. A comparison of the treatment outcomes of African Americans and European Americans at Roswell Park should lessen the likelihood that treatment differences would affect survival differences. Treatment differences would in all likelihood be minimized within a single institution; they would be conditioned by the disease site and stage of disease, rather than by institution-specific preferences. In addition, within a single institution, disease site, along with stage at diagnosis, age and gender could be held constant.
We have used a number of patient data sets collected at Roswell Park over the past 50 years to compare treatment outcomes of African Americans and European Americans. We consider racial differences in mortality treated first as a dichotomy at 5 and at 10 years, and with the relative hazard of mortality examined by use of Cox proportional hazards analysis. We adjust for stage of disease at diagnosis, age and gender for comparison. We also consider effect modification by disease stage, age and gender.
Citation Information: Cancer Epidemiol Biomarkers Prev 2010;19(10 Suppl):PL08-03.
- Copyright © September 30, 2010, American Association for Cancer Research