Research Article: Racial Categories in Medicine: A Failure of Evidence-Based Practice?

Date Published: September 25, 2007

Publisher: Public Library of Science

Author(s): George T. H Ellison, Andrew Smart, Richard Tutton, Simon M Outram, Richard Ashcroft, Paul Martin

Abstract: Race and ethnicity are imprecise markers of the genotypic and sociocultural determinants of health, argue the authors.

Partial Text: In this issue of PLoS Medicine, Lundy Braun and colleagues from the Race, Medicine, and Science Workshop discuss “the trouble with race” [1]. They argue that the heterogeneity of racial and ethnic categories (and a lack of consensus on how these categories should be defined and measured) make them poor markers for underlying differences in the genotypic, cultural, and structural characteristics responsible for inequalities in health (where “structural characteristics” include a diverse range of historical and contemporary socio-political factors responsible for differential access to health resources and differential exposure to health risks). Such racial and ethnic categories, they argue, are also poor markers for the differential efficacy of diagnostic and therapeutic interventions.

In particular, we have found that there is a lack of consensus about what race and ethnicity mean [3] and how these should be operationalised [4]. As a result, researchers and practitioners may conflate the utility of racial and ethnic categories for sampling diverse study populations with their ability to identify and address aetiological variation therein [5]. At the same time, there is widespread concern about the socio-political sensitivity of such categories and a desire to avoid stigmatising populations and groups found to be at greater risk of disease [6]. This concern has led many of the researchers we interviewed to adopt the more socially acceptable term “ethnicity” in preference to “race” [6]. It has also led some researchers to use crude socio-political classifications (such as the categories developed by the United States Office of Management and Budget [OMB] and by the United Kingdom Office for National Statistics [ONS]) not only for sampling diverse study populations and describing population differences in health and health care, but also for attributing these differences to innate genotypic and/or socio-cultural causes [4].

The use of crude socio-political categories of race and ethnicity to describe variation in health risks and health needs, and to attribute these differences to innate genotypic and socio-cultural factors, has a long and discredited history, as Braun and colleagues describe [1]. However, more recently this practice seems to have been resurrected and validated, however unwittingly, by the adoption of such categories to promote equitable participation of “minorities” in research and to ensure that researchers are able to identify and describe any differences between racial and ethnic groups that warrant further investigation and intervention [7].

The attributive use of crude socio-political categories of race and ethnicity in biomedical research and practice is not, however, an inevitable consequence of policies to promote equitable inclusion in research or to describe disparities in health and health care that warrant further investigation and intervention. Researchers are not required to limit the variables they use to explore disparities in health, and clinicians are not obliged to use crude racial and ethnic categories to select the care they provide. But researchers and clinicians do need to be encouraged to use more specific attributive markers of genotype, culture, and structural disadvantage wherever appropriate.

So while we would support Braun and colleagues’ call for international consensus to improve the use of racial and ethnic categories in biomedical research and practice [1], such categories cannot and should not be standardised for use in all scientific, social, and clinical contexts, even if only for use as descriptive variables. Instead, we need to recognise that different racial and ethnic categories are needed to describe inequalities in health and health care in contexts where these have different salience and meaning. At the same time, alternative attributive markers of genotypic, cultural, and structural determinants of variation in health and access to health care need to be developed in order to: (1) improve the aetiological precision of biomedical research; and (2) facilitate the translation of research on the causes of variation in health across racial and ethnic groups into appropriate care for individual patients. This more precise approach would help to address a long-standing problem with evidence-based practice, which often struggles to apply the results of epidemiological research on populations to the clinical care of individuals [15].

An alternative approach would be to target the gatekeepers of research dissemination, particularly the editors of biomedical journals, who have displayed increasing willingness to improve the use of racial and ethnic categories in published research [18]. Certainly, our research has identified many journals that have generated a range of dedicated guidelines for authors [4,17], even if many biomedical journals simply encourage authors to “justify” the categories they use. These include, for example, the 648 journals signed up to the International Committee of Medical Journal Editors’ Uniform Requirements for Manuscripts Submitted to Biomedical Journals: Writing and Editing for Biomedical Publication, which recommend that “When authors use variables such as race or ethnicity, they should define how they measured the variables and justify their relevance” [19].