Date Published: September 25, 2007
Publisher: Public Library of Science
Author(s): Maggie Brown
Partial Text: A study is published reporting that a new drug works better in black people than in whites. Is this an informative study or is it based on archaic, incorrect, even harmful notions of human difference?
In this issue of PLoS Medicine, Braun et al.  discuss some of the difficulties and dangers that go along with racial and ethnic designations in the clinic and in biomedical research. They propose that a clear consensus needs to be reached about exactly how to use race and ethnicity as variables and as designations and when it is valid to use them. They state further that while race may be used legitimately as a “descriptive” quality, it should not be used as a “biological attribution.” In other words, race as a social construct may result in differences in treatment that affect health outcomes, but such descriptive use does not imply that race can be used as a proxy for biological difference. In a commentary on the Fausto-Sterling essay, Ellison et al.  agree that standards of definition need to be improved, but caution that potentially important data may be at stake: citing the example of variance in histocompatibility antigen markers as important in transplantation research, they argue that variation in the frequency of genotypic markers among racial and ethnic groups should not be unilaterally discarded simply because of the risk that such information may be misused. They add that while international consensus on improved categorization is important, any guidelines deriving from such consensus will need to be made flexible: “such categories cannot and should not be standardised for use in all scientific, social, and clinical contexts.” This proposal echoes a Nature Genetics editorial that argued for inclusion of multiple types of information (e.g., “ancestral and environmental”) so that the data can be “grouped flexibly to serve the needs of medical geneticists, epidemiologists, and biological anthropologists” . To complicate matters, a constraint facing many researchers is the requirement that US National Institutes of Health grantees in clinical research collect information on race that follows the current US census designations [5,6].
What can journals do? Should designations of race and ethnicity be left to an author’s discretion, since, after all, authors know the most about their own studies? Some editors have adopted such an approach, arguing that the author and referees are best equipped to offer expert guidance on the matter [1,9]. Other journals, instead of creating specific instructions for authors, have published position papers that seek to persuade, rather than force, authors to apply more consistent, rigorous standards of terminology and science (e.g., [10,11]). Kaplan and Bennett , for example, suggest that, whenever possible, studies of racial differences should include analyses to control or adjust for other variables, such as socioeconomic status, nutrition, environmental exposures, etc. (see also ). In 2004 Nature Genetics devoted an entire special issue (“Genetics for the Human Race”) to a discussion of how to conceptualize, define, and study human racial and genetic differences (content freely available at http://www.nature.com/ng/journal/v36/n11s/index.html). As noted above, a few journals have added instructions or guidance to their author guidelines (e.g., Journal of the American Medical Association, http://jama.ama-assn.org/misc/ifora.dtl#ReportingRaceEthnicity; Canadian Medical Association Journal, http://www.cmaj.ca/authors/policies.shtml; Genetics in Medicine, http://edmgr.ovid.com/gim/accounts/ifauth.htm).
Although race and ethnicity as contentious variables in research and clinical medicine are the most discussed in the literature, they are not the only possible sources of incorrect generalizations and possibly harmful bias. Others are sex/gender, age, sexual orientation, disease/disability, religion, socioeconomic status, and many more. For example, the AMA Manual of Style (10th edition, section 11.10 ) and the CSE manual (7th edition, section 7.5 ) offer advice on inclusive language in the areas of race/ethnicity, age, disease/disabilities, religion, and sexual orientation, emphasizing in part that terminology should be nonstigmatizing and reflect the preferred designations of groups or individuals. In all of these areas humans have been subject to stereotyping and discrimination; thus a critical examination of all the names we call ourselves and others is warranted, and at least general guidelines should be developed for these areas, although consensus may take time.
Whose responsibility should it be to ensure that race, ethnicity, and other human variables are described appropriately? Authors and referees, who know their research and fields better than most journal editors do? Or editors, whose mandate it is to uphold the scientific and ethical quality of their journals? The answer is both, working collaboratively.