Improving the Collection and Use of Racial and Ethnic Data in HHS. 4. Measures of Discrimination


The President's measures of discrimination initiative will further test the availability of racial and ethnic data in HHS data sets and may raise further issues concerning gaps in such data. This initiative will require drawing methodological distinctions between measuring disparities in access, services, health status and quality of care, and measuring how intentional and/or subconscious biases may restrict access or result in differential quality of care based on race or ethnicity. Data may reveal disparities, but it is considerably more difficult to isolate specific causes of differential health outcomes, including measuring the extent to which racial and ethnic discrimination play a role.

Many studies have identified differences, according to race and sex, in the treatment of patients with cardiovascular disease in the U.S. A recent New England Journal of Medicine (NEJM) article on the effects of race and sex on physicians' recommendations for cardiac catheterization reports on a study that addressed directly the question of physician responsibility relative to treatment recommendations for patients with various types of chest pain (Schulman, et al., 1999). The study used a computerized survey instrument and videotaped actors portraying patients with particular characteristics. The methodology used by the Shulman, et al. and their results were criticized in a NEJM Sounding Board article and further called into question in subsequent letters to the journal. Nevertheless, there are certain lessons that can be learned. In a sense, this study utilized a type of paired testing, a methodology used most prominently by HUD as a measure of the extent of racial and ethnic discrimination in housing markets. In effect, paired testing utilizes two subjects who have like symptoms, but who belong to different racial groups or have disparate language skills. Shulman, et al. suggest possible means by which HHS could measure racial and ethnic discrimination in other health care settings. The Department's activities in support of the new Presidential Initiative on measures of racial and ethnic discrimination will place further demands on ensuring the adequacy and accuracy of racial and ethnic data in the Departmental data sets. Currently, HHS is supporting a comprehensive review of the state of the art in discrimination measurement in health care. Consequently the focus of future HHS activity, data collection, methodology design and analysis will depend on the outcome of this effort. Whatever is done, however, will place further demands on systems for detailed racial and ethnic data.

A complicating factor in assessing discrimination is that the diversity of the U.S. population will be changing rapidly over future decades. As a result, the key racial and ethnic groups who need to be covered could change as a result of shifting patterns of immigration. Immigration is an area that needs to be carefully studied.

As part of the measures of discrimination initiative, in FY 2000, HHS, as a first step, is expected to assess what current data sets or surveys could be used or supplemented to measure racial and ethnic discrimination and determine whether new methodologies need to be developed and tested. The goal of the overall measures of discrimination initiative is to expand existing knowledge on appropriate and credible ways to measure the presence of racial and ethnic discrimination and to support empirical studies that measure the scope of racial and ethnic discrimination using new and existing techniques and data.

However, given the early nature of the measures of discrimination initiative, few specific data gaps relating directly and only to measures of racial or ethnic discrimination have been identified. As we endeavor to develop methods and identify effective measures of racial or ethnic discrimination, it is probable that racial and ethnic data gaps beyond those noted elsewhere in this report may arise as a significant impediment to successful use of current data sets.