Race, Ethnicity, and Language of Patients: Hospital Practices Regarding Collection of Information to Address Disparities in Health Care

2003 Institute of Medicine (IOM) report, Unequal Treatment: Confronting Racial and Ethnic Disparities in Health Care, identified strong evidence of disparities in the health care of racial and ethnic minorities not explained by differences in health insurance coverage and income. Among ways to eliminate these inequalities the IOM report recommended enhanced collection of patient race and ethnicity data. National Public Health and Hospital Institute researchers surveyed 500 non-federal acute care hospitals on their collection of patient race, ethnicity and preferred language information to understand data collection practices in the U.S. hospital industry. The researchers also surveyed 64 safety net hospitals — which typically have very diverse patient populations and presumed data collection experience — on their collection and use of patient race, ethnicity and language preference data. Key Findings: Most hospitals collect data about the race, ethnicity and language preference of their patients. Over three-quarters (78.4 percent) collect race information and one-half collect data on patient ethnicity (50.4 percent) and language preference (50.2 percent). Fewer than one in five hospitals use the data to assess and compare care quality, health services utilization, health outcomes or patient satisfaction. The most common barrier to data collection for hospitals that do not collect these data is the sense that the data are not important, with more than half of non-collecting hospitals identifying this as a barrier to collection. All of the surveyed safety net hospitals routinely collect race and ethnicity data, although only 20 percent have formal data collection policies. Eighty-four percent of these hospitals have a required field for race in their automated registration system and 28 percent have a field for ethnicity that is generally optional. While 80 percent have a field for language it is rarely required and its recording varies widely