A comparison of direct medical costs across racial and ethnic groups among children with cancer.

OBJECTIVE: Previous studies reported that some minority childhood cancer patients are likely to develop worse outcomes than white children. This study examines whether there are racial and ethnic disparities in health expenditures among children with cancer. RESEARCH DESIGN AND METHODS: A retrospective study was conducted among children (younger than 20) with cancer diagnoses in the Medical Expenditure Panel Survey (MEPS; 1996 to 2004). Total health expenditures and the following subcategories were examined across racial and ethnic groups: (1) office-based visits; (2) outpatient visits; (3) inpatient and emergency room visits; (4) home health care; (5) prescription drugs; and (6) dental, vision, and other health care expenditures. Consumer price indexes were used to convert all expenditures to 2004 dollars. A classical linear model was analyzed using the natural logarithm of health expenditures as the dependent variable, with the purpose of determining whether there were racial and ethnic differences in health expenditures after adjusting for confounding factors.
RESULTS: Study sample included 394 non-Hispanic whites (weighted to 4 958 685), 53 non-Hispanic blacks (weighted to 352 534), and 94 Hispanic whites (weighted to 424 319). Hispanic blacks and other minority populations were excluded from the analysis due to insufficient sample size. The annual total health expenditure for treating each child with cancer was $3467.40, $2156.15, and $5545.34, respectively, among non-Hispanic whites, non-Hispanic blacks, and Hispanic whites. The differences in the various subcategories of health expenditures across racial and ethnic groups were generally not significant according to both descriptive and analytical analyses with very few exceptions.
CONCLUSIONS: This study did not identify significant racial and ethnic disparities in health care costs. However, one important study limitation is the small sample size of the minority populations in the study sample.