5 Arguments against the Use of Racialized Categories as Genetic Variables in Biomedical Research:
What Are They, and Why Are They Being Ignored?
This chapter provides a detailed analysis of published criticisms and concerns surrounding the use of racialized categories as (if they were) markers of genetic variation in genetics and related biomedical research. Using a systematic citation-based literature search, we identified 335 articles containing such criticisms and concerns and subjected these to in-depth thematic analysis. Two broad groups of arguments were observed: those relating to the use of racialized categories as (if they were) genetic variables, and those exploring their use as such within biomedical research. Five specific groups of arguments against the use of racialized categories as (if they were) markers of genetic variation emerged from these articles. These drew on concerns that most genetic variation is found within all racialized groups; only a modest amount of genetic variation loosely clusters around racialized groups (but does not result in distinct packages of genetic traits that are unique to each racialized group); genetically “pure” racialized populations do not exist (and have never existed); variation in phenotypic traits among racialized groups cannot be assumed to reflect variation in genotypic traits; and using racialized categories as (if they were) markers of genetic variation tends to reify these categories as essential qualities of the individuals and groups concerned (rather than as context-specific and fluid forms of sociocultural identity). Six related themes emerged concerning the use of racialized categories as (if they were) genetic variables in biomedical research, drawing on concerns about the lack of consensus regarding the definition of racialized categories; the limited reliability and external validity of racialized categories; the limited internal validity of racialized categories; the way in which using racialized categories as (as if they were) markers of genetic variation tends to encourage their widespread use to (erroneously) infer genetic causality for disparities in health among racialized groups; the way in which research emphasizing differences in health risk and health care need among racialized groups can lead to the development of inappropriately targeted services for different racialized groups; and how the use of racialized categories to explore disparities in health (and particularly the possible genetic basis for these) can lead to stigmatization and stereotyping. We discuss how these arguments are not simply overlooked, ignored, rejected, or circumvented by contemporary geneticists and biomedical scientists within the context of contemporary genetics and biomedical science, but are selectively engaged with to generate a largely self-referential case for the continued use of racialized categories as (if they were) markers for genetic variation between populations by the researchers involved. As such, while the criticisms and concerns reviewed in this chapter provide powerful arguments for removing the use of racialized categories as (if they were) genetic variables from most genetic (and related biomedical) research, selective engagement with these arguments, together with the self-referential interpretation of findings from analyses using racialized categories as (if they were) genetic variables, continues to make their use as such appear useful to geneticists and biomedical researchers alike.
Introduction
While the creation of racialized categories dates from late-eighteenth-century European science, today, these same categories (albeit with some modification in taxonomy and nomenclature) continue to be associated with the contemporary science of genetics and with biomedical studies of social disparities in health. The principal argument against using these categories within genetics and biomedical research is not simply that by using the term race (or related racialized categories such as ethnicity, nationality, or, more recently, biogeographical ancestry), scientists are likely to repeat the abuses of the past. Instead, it seems clear that the political climate has shifted, as has the scientific rationale for using these categories. With the benefit of new scientific knowledge concerning the nature and extent of variation across the human genome, arguments over the use of racialized categories in genetics and biomedical research increasingly focus on their utility as markers for genetic variation (and their related applicability to whatever innovations in clinical practice may emerge from new developments in genetic technology). Indeed, mainstream genetic research does not appear to be especially preoccupied with looking for alleles capable of differentiating between racialized groups (Cavalli-Sforza et al. 1991), but instead focuses on exploring whether a better understanding of the extent and nature of genetic variation within and between racialized groups might benefit these groups and/or the population as a whole. Much of this work seeks to examine whether the persistence of disparities in health among racialized groups can be accounted for by genetic variation between them. Given the exponential growth in gene-disease association studies, alongside the continued categorization of populations according to racialized categories (not least their use in contemporary national censuses and health information systems), it seems unlikely that these categories will become redundant in the near future (Ellison and Jones 2002). The question has therefore shifted from whether using racialized categories is morally a good or bad thing to whether, on balance, the use of such categories can be justified by the genetic data or by the health benefits that accrue to racialized groups (and/or the population as a whole; Ellison 2005).
This chapter examines the various criticisms leveled at the continuing use of racialized categories within genetics and related biomedical research as (if they were) markers of genetic variation and aims to identify common themes emerging from a review of the literature on this topic. These themes are then analyzed with reference to previous qualitative research we (and our colleagues) have conducted with a range of geneticists and biomedical scientists (Outram and Ellison 2006a, 2006b; Ellison et al. 2007, 2008a). In this way, we hoped to establish how and why criticisms of this practice have been overlooked, ignored, rejected, or circumvented by contemporary geneticists and biomedical scientists. Our analyses suggest that they have adopted an approach characterized by selective engagement and the self-referential interpretation of criticisms against the use of racialized categories as (if they were) markers of genetic variation. The analyses also illuminate the interminable debate concerning the extent of genetic variation within and between racialized groups, which dates back to Richard Lewontin’s work on protein polymorphisms in 1972 and has been reinvigorated by recent developments in genetic technology and genomewide analyses.
When comparing the themes emerging from articles criticizing the use of racialized categories as (if they were) markers of genetic variation against the explanations offered by geneticists and biomedical scientists for the continued use of these categories in this way, we were at pains to accept that geneticists are now able to correlate clinical patterns of genetic variation with a range of racialized categories. We also accepted that these correlations have been successfully used to generate credible predictions of the likely affiliation of individuals to racialized groups (albeit with varying degrees of accuracy, depending on the populations studied, the number and type of ancestry informative markers used, and related methodological constraints; Liu et al. 2006; Shriver et al. 1997; Risch et al. 2002; Rosenberg et al. 2002; Tang et al. 2005). For some commentators, such as Edwards (2003), these correlations and predictions seriously undermine Lewontin’s (1972) conclusion (that racialized categories were of little use as genetic variables), and this interpretation is likely to play an important role in the continuing use of racialized categories within genetic research.1 However, we have adopted a more skeptical interpretation of these correlations and predictions and conclude that the persistent use of such categories results, instead, from the choices made by geneticists and biomedical scientists about what to emphasize and what to ignore. This involves a narrow focus on what one might call the biogenetic correlates of racialized categories, as opposed to their sociopolitical origins and associated characteristics. In this we agree with those who argue that a new framework is required to fully understand what otherwise appear to be heritable and ostensibly innate disparities in health between racialized groups. This framework would recognize how the integral biological and social attributes of racialized categories contribute to disparities in health. It suggests that a more appropriate use of racialized categories as (if they were) markers of genetic variation would be one that recognizes these as, first and foremost, social categories that may only be usefully incorporated into biomedical research and practice in those instances (and these seem likely to be rare) when they are strongly associated with genetic differences responsible for disparities in health.
As such, an important limitation, or rather, an important qualification of the analyses that follow is that these focus on the ways in which racialized categories are used in genetics and related biomedical science; the arguments against their use therein; and why many geneticists and biomedical scientists have failed to engage with these.
Finally, it is important to note that this chapter does not attempt to enter into any discussions situated outside the study of racialized groups within genetics and biomedicine. As such, we have focused only on arguments against the use of racialized categories within genetics and biomedical research. And while the issues raised herein may resonate with, and may have potential implications for, identity politics and policy making, our chapter is an attempt to focus on the roots of the ongoing debate about the biogenetic nature of racialized categories—a debate that is frustrating (due to the failure of recent advances in genetic knowledge to deliver resolution or consensus; Weiss and Fullerton 2005), but also highly productive (in the way it illuminates the interpretative frameworks and methods used within genetics and the biomedical sciences). Thus our chapter does not seek to provide a comprehensive analysis of the various concerns raised about the use of racialized categories within other natural or social science disciplines, nor do we explore the social, political, and moral debates concerning the development or use of official racialized categories (such as those common in contemporary censuses) or unofficial racialized categories (such as those recognized more generally within contemporary society). Our chapter also makes no attempt to distinguish between the many different categories and labels used to classify racialized groups within science or society, not least because there is so little consensus on what these mean and because it seems clear to us that all have the potential to reify discredited notions of naturally occurring human populations with innate and heritable biogenetic (and/or sociocultural) characteristics. These issues have been discussed by Baum (2006), Jenkins (2001), and Woodward 2004), among others, and elsewhere in this volume.
Methods
To access articles debating the use and utility of racialized categories in genetics and related biomedical research, a citation-based (or snowball) literature search (Hart 1998, 2001) was conducted based on any articles that had cited McKenzie and Crowcroft’s (1996) editorial in the British Medical Journal—“Describing Race, Ethnicity, and Culture in Medical Research”—which accompanied the journal’s “Guidelines for Research, Audit, and Publication” on ethnicity, race, and culture (the first such guidelines to be developed by a major biomedical journal; British Medical Journal 1996).
The ISI Web of Knowledge (http://isiwebofknowledge.com/) was used to identify any articles that had cited the editorial up until the end of 2007. These were selected for inclusion in our review on the basis of whether they had discussed in detail the use of racialized categories in genetics and/or biomedical research. This was determined by reading each article’s abstract and, when this provided insufficient information on which to base a decision, by reading the main text of the article itself. Any articles selected for inclusion in the review were then forward- and back-referenced, to identify any relevant articles referenced within these and any relevant articles that had cited these, respectively.
All the articles selected for inclusion in the review were analyzed using thematic analysis, as described by Boyatsiz (1998) and Braun and Clarke (2006). More general guidance on the qualitative analysis of scientific literature was drawn from Silverman (2000, 2001). On the basis of these techniques, the first stage of the analyses involved identifying distinct criticisms and concerns raised within each of the selected articles and extracting verbatim quotes that illustrated the various forms in which these criticisms and concerns were presented. The second stage of the analysis involved assembling these quotes into common themes, which were continually reviewed and refined as additional data from subsequent articles were identified and extracted. This process aimed to develop a comprehensive yet coherent thematic overview of the criticisms and concerns raised, organized into thematic groupings to reflect the ontological and epistemological nature of the issues they contained. In this way, the number of themes fluctuated as subsequent articles were read, with new themes emerging and existing themes coalescing or separating out as the analysis proceeded.
Results
Articles Identified for Inclusion in the Review
Up until the end of 2007, McKenzie and Crowcroft’s (1996) editorial had been cited by a total of seventy-one articles, thirty-two of which (45%) were judged to have discussed the use of racialized categories in genetics and/or biomedical research. These thirty-two articles contained a total of 1,327 references and had themselves been cited a total of 695 times. Of these references and citations, 303 were judged to have discussed the use and/or utility of racialized categories in genetics and/or biomedical research so that, together with the original 32 articles, 335 unique articles were included in the analyses that follow. This process of selection led to the following types of articles being excluded:
• Articles describing empirical studies of disparities in health among racialized groups that did not discuss the use of racialized categories as research variables, for example, an epidemiological study concerned with exploring variation in hypertension among racialized groups that did not question the use of the categories themselves.
• Articles exploring the formation of social identities that did not relate these processes to the use of racialized categories in genetic or biomedical research, for example, a study of identity among second- and third-generation migrants.
• Articles examining the historical and legal construction of racialized categories that did not relate these processes to genetics and biomedical research, for example, a study concerned with how the notion of a racial hierarchy became intertwined with the trans-Atlantic slave trade.
• Articles discussing appropriate methods for conducting studies in cross-cultural environments, for example, a study assessing whether the social identity of the interviewer might influence the conduct of interviews.
• Articles containing clinical advice concerning specific diseases, for example, studies that identified differences in disease risk or therapeutic efficacy among racialized groups and provided related clinical advice concerning the care of patients from different groups.
Issues Addressed by Articles Included in the Review
Seven types of articles were selected for inclusion in the review, although most of the articles took more than one of these forms:
• Articles describing studies of genotypic variation using racialized categories that discussed how such data might be applied elsewhere, for example, Liu et al. (2006) and Mountain and Risch (2004).
• Articles exploring how disparities in health among racialized groups are often interpreted as evidence of innate (i.e., genetic) differences by biomedical researchers, for example, Cooper (1984) and Osbourne and Feit (1992).
• Articles examining the methods and analytical assumptions used in epidemiological research to explore disparities in health between racialized groups, for example, Kaufman and Cooper (1999) and LaVeist (1994).
• Articles assessing available methods for classifying and categorizing racialized groups in biomedical research, for example, Hahn et al. (2002) and Sugarman (1996).
• Articles discussing the conceptual basis underlying the racialized categories used in biomedical research, for example, Aspinall (1997) and Cartmill (1999).
• Articles reviewing the use of racialized categories in studies published by biomedical journals, for example, Comstock, Castillo, and Lindsay (2004) and Ellison and de Wet (1997).
• Articles tackling the terminology and nomenclature used to identify racialized groups in biomedical research, for example, Bhopal (2004) and Gimenez (1989).
Thematic Analysis of Articles Selected for Inclusion in the Review
Thematic analysis of the 335 articles selected for inclusion in the review identified two broad groups of themes relating to criticisms and concerns about the use of racialized categories in genetics and biomedical research: the use of racialized categories as (if they were) markers for genetic variation per se and the use of racialized categories as (if they were) markers of genetic variation in biomedical research. These two groups of themes will be analyzed separately in the following sections.
The Use of Racialized Categories as (If They Were) Markers of Genetic Variation
There were five key themes relating to debates about the use of racialized categories as (if they were) genetic variables or markers of genetic variation. These themes included the following criticisms and concerns: (1) that most genetic variation is found within all racialized groups, (2) that a modest amount of genetic variation loosely clusters around racialized groups but does not reflect distinct packages of genetic traits that are unique to each racialized group, (3) that there are no genetically pure populations (and never have been), (4) that variation in phenotypic traits cannot be assumed to reflect variation in genotypic traits, and (5) that using racialized categories as (if they were) genetic variables or markers of genetic variation tends to reify these categories as innate and immutable entities, rather than recognizing these to be context-specific and fluid forms of sociocultural identity.
• Most genetic variation is found within all racialized groups. Perhaps the most powerful criticism leveled at the use of racialized categories as (if they were) markers of genetic variation stems from the observation that most genetic variation among human beings occurs within, rather than between, racialized groups. In Goodman’s (1997, 24) exploration of racialized categories in forensic anthropology—Bred in the Bone?—he wrote, “Some thirty years ago the population geneticist Richard C. Lewontin of Harvard University conducted a statistical study of [genetically determined] blood groups with two of the more common forms. On average, he found about 94 percent of the variation in blood forms occurred within perceived races; fewer than 6 percent could be explained by variations among races.” Such is the importance of Lewontin’s analysis that the Race, Ethnicity, and Genetics Working Group (2005, 521), along with many others, cite Lewontin’s (1972, 397) analysis, which found that only “5%–15% of genetic variation occurs between large groups living on different continents [i.e., groups commonly classified as different races], with the remaining majority of the variation occurring within such groups.” Lewontin (1972, 397) concluded that given that genetic variation within groups was greater than that between them, “racial classification is now seen to be of virtually no genetic or taxonomic significance either [and, therefore] no justification can be offered for its continuance.” Indeed, many of the concerns identified in the paragraphs that follow derive from Lewontin’s (1972, 397) conclusion that race is of “virtually no genetic or taxonomic significance.”
• (Only) a modest amount of genetic variation loosely clusters around racialized groups. Despite the observation that most genetic variation among humans occurs within racialized groups, a modest proportion is associated with racialized categories (i.e., varies between these). Nonetheless, because these associations are loose, there are no distinct or definitive genetic differences between racialized groups. To this end, Cooper (2005, 72) noted that “race, a quantitative distinction within a species, has no equivalent defining criterion—that is, genetic variability is not restricted to discrete packages,” whereas Cartmill (1999, 654) has pointed out that “since there are thousands of separate, independently assorting variable loci in the human genome, it is highly unlikely a priori that variation at any particular locus will covary with any other.” Turning specifically to one of the most commonly used racialized categories, Braun et al. (2007, 1423) argued that “assuming that ‘African’ origin can capture the complexity of migrations, artificial boundaries, and gene drift is scientifically unsupportable.” Likewise, relating these clustering arguments to specific populations, the Race, Ethnicity, and Genetics Working Group (2005, 521) observed that “samples taken from India and Pakistan affiliate with Europeans or eastern Asians rather than separating into a distinct cluster… [whereas] samples from the Kalash, a small population living in northwestern Pakistan, form their own cluster on a level comparable with those of the major continental regions.” Cooper (1984, 716) summed up the inadequacy of racialized categories under these circumstances: “historical efforts to use the race concept scientifically have ended in failure” because racialized categories do not “delineate important and consistent genetic differences [in the form of]… a ‘package’ of different genes… between groups.” Thus, while a modest amount of genetic variation might cluster within geographical areas and sociocultural groups, such clustering does not fall within discrete racialized groups in such a way that these can be categorized as genetically distinct.
• There are no genetically pure populations (and never have been). Leading from arguments about the distribution of genetic variation between populations are concerns about the belief that racialized groups are genetically homogeneous—or at least, that they were before these groups started to move and mix. Two distinct criticisms of this belief were evident among the articles included in this review. The first, as summarized by Freeman (1998, 224), is simply that “races, in the sense of genetically homogeneous populations, do not exist in the human species today, nor is there any evidence that they have ever existed in the past.” The second, as argued by Cartmill (1999, 653), is that although “it is true that human populations in some parts of the world were more uniform and distinctive a thousand years ago than they are at present… populations like those of modern North America, with high levels of phenotypic variability maintained partly by migration and gene flow from elsewhere, are not a new phenomenon.” In essence, Cartmill’s argument is that the idealized notion of genetically pure racialized groups is limited in both place (“some parts of the world”) and time (“a thousand years ago”). Although these arguments seem to differ as to whether genetic homogeneity ever existed within (some, isolated and/or pure) racialized groups, both still draw the same conclusion regarding contemporary populations: that the notion of racialized groups as genetically homogeneous is limited and somewhat questionable, given that few of the world’s populations have ever lived in sufficient isolation from one another to generate or maintain discrete genetic characteristics. As such, the assumption that racialized populations are genetically homogeneous (and, as we saw earlier, genetically distinct) is felt to be flawed or very limited in its applicability.
• Variation in phenotypic traits cannot be assumed to reflect variation in genotypic traits. While a range of phenotypic, sociocultural, and geographical markers are commonly used to classify racialized groups—that is, what you look like and where you or your family come from—it is widely accepted that the most salient historical and contemporary marker for classifying racialized groups is skin color. Indeed, the importance of skin color has led researchers to develop and test hypotheses based on skin color per se, and this is a practice that has been specifically addressed in several articles that question the utility of such phenotypic characteristics as markers of wholesale genotypic differences. For example, Garte (2002, 421) argued that “present definitions of race based on superficial characteristics, or on other phenotypes strongly influenced by natural selection, such as skin color, simply do not correlate with data from the whole genome.” Likewise, Williams (1997, 323) pointed out that “skin and hair color, facial features, and other superficial external characteristics do not correlate well with biochemical or other genetic characteristics.” As such, these writers conclude that focusing on observable phenotypic markers of racialized categories (especially, but not exclusively, skin color) to estimate genetic relatedness is a gross oversimplification and (mis)leads researchers to assume genetic homogeneity (erroneously) on the basis of shared phenotypic traits.
• Using racialized categories as genetic variables reifies these categories as innate. Finally, a number of arguments link critiques concerning the use of racialized categories as genetic variables to their (inappropriate) use in biomedical research. These arguments stress that assuming genetics to be an essential element of any disparities in health among racialized groups makes racialized categories appear inherently innate characteristics of the groups and individuals concerned (rather than fluid social/cultural/political forms of identity). For example, Williams, Lavizzo-Mourey, and Warren (1994, 27) argued that “research on racial variations in health has been dominated by a genetic model that views race as primarily reflecting biological homogeneity and black-white differences in health as largely genetically determined.” This essentialization of racialized categories as genetic variables detracts from studies on nongenetic alternatives, as Muntaner, Nieto, and O’Campo (1996, 532) pointed out: “without any evidence from genetic studies, the observation that blacks tend to have lower leukocyte counts than whites for example, led scientists to the conclusion that neutropenia is probably a normal genetically determined characteristic in people of African descent.” Likewise, Liu (1998, 1765) observed that “simple logic notes that races are different physically and that these differences are determined by genes: some that are responsible for the pigmentation of the skin and others dictate hair color or the contour of the nose. Because there are differences in cancer rates and mortality between the races, specifically between blacks and whites, the same logic has led many to believe that these distinctions must also be determined genetically.” Liu goes on to say that “this argument, however, is fundamentally flawed both on genetic and on social grounds.” Bradby (1995, 406) extended this critique of deterministic logic by pointing out that “the existence of one disease-causing gene at a higher frequency in a population defined by a particular physical appearance, say dark skin and tightly curled black hair, does not mean that those so described have poor health in other respects.” In summary, these arguments focus on the way in which the essentialization of racialized categories as genetic entities has led biomedical researchers to assume that differences in health among racialized groups are predominantly genetically determined, and thereby led them away from examining the structural, socioeconomic, political, and cultural issues that many believe make a significant (if not larger) contribution to disparities in health.
The Use of Racialized Categories as Genetic Variables in Biomedical Research
In addition to the five themes described in the preceding section, related to the use of racialized categories (as if they were) markers of genetic variation, six additional themes emerged from articles included in this review that referred more generally to the use of racialized categories as (if they were) markers of genetic variation within biomedical research. These included (1) a lack of consensus regarding the definition of racialized categories, (2) the limited reliability and external validity of racialized categories, (3) the limited internal validity of racialized categories, (4) how using racialized categories as markers of genetic variation can encourage their widespread use to infer genetic causality for disparities in health among racialized groups, (5) how research emphasizing differences in health risk and health care need between racialized groups can lead to the development of inappropriately targeted interventions for different racialized groups, and (6) how using racialized categories to explore disparities in health (and particularly the genetic basis thereof) can lead to stigmatization and stereotyping.
• Lack of consensus regarding the definition of racialized categories. A starting point for many of the critiques leveled at the use of racialized categories in biomedical research was the lack of consensus regarding how such groups and categories are defined—something that is evident from any brief scan of biomedical dictionaries (Ellison 1999). Indeed, within the literature included in this review, LaVeist (1994) provided a detailed discussion of medical, biological, psychological, and epidemiological definitions of race, which identified two significant features thereof: the first being a tendency to emphasize race as a genetically meaningful taxonomic category, and the second being a tendency toward uncertainty, ambiguity, and inconsistency. Sometimes both these are evident, as in the example LaVeist (1994, 5) quoted from Becker and Landav’s International Dictionary of Medicine and Biology, which defined “race as a biological concept that defies discrete categorization: A subspecies or other division or subdivision of a species.” These two features are also evident in the changing definitions of race offered in subsequent editions of some influential dictionaries, such as the Dictionary of Epidemiology (Last 2001, 150), the most recent edition of which emphasizes the use of race as a biological category but notes that the “biologic classification of human races is difficult because of significant genetic overlaps among population groups.” In contrast, the previous version of this dictionary (Last 1995, 139) led with the sentence that races comprised “persons who are relatively homogeneous with respect to biological inheritance,” but followed this with the sentence stating that “in a time of political correctness, classifying by race is done cautiously,” with a footnote referenced to Cooper and David (1986) and Osbourne and Feit (1992). Such reviews of dictionary definitions suggest that race is an (increasingly) ambiguous, contentious, and politically sensitive term when defined as a biological and/or genetic entity.
The ongoing debate over the meaning (and utility) of race has led a number of writers to examine whether ethnicity might be a preferable term. As such, Wiencke (2004, 79) distinguished between the two by proposing that “race, as it is used in common discourse, is a subdivision of a species formed by a group of individuals that share common biological characteristics that distinguish them from other groups… [while the] concept of ethnicity emphasizes cultural, socioeconomic, religious and political qualities of human groups.” This view, that race and ethnicity are conceptually distinct, is also evident in Huth’s (1995, 620) review, in which he argued that “in contrast [to race], ethnicity represents a concept that makes no claims to biological precision and that reflects a broader view of factors that may influence the susceptibility of individuals to etiologic agents and the ways in which they may respond to those agents.” Significantly, Huth’s definition added another dimension to ethnicity—that of continuous contextual and temporal variation. This emphasis on the inherent flexibility of ethnicity is also evident in Nazroo’s (1998, 723) argument that “ethnic identity cannot be considered as fixed, because culture is not an autonomous and static feature in an individual’s life.” As such, ethnicity has been distinguished from race by its fluid and predominantly social attributes, rather than the fixed biological and/or genetic attributes traditionally associated with the notion of race.
However, although racial and ethnic categories can be conceptualized very differently and are sometimes articulated as such, the two terms often overlap, not least when they refer to the same social groups or are classified using the same phenotypic, sociocultural, and geographical characteristics, and similar nomenclature. Moreover, as Karlsen (2004, 108) pointed out, there is “an assumption dominant in epidemiological research that the ethnic differentials found among various social and economic characteristics are a consequence of innate characteristics related to ‘ethnic’ or ‘racial’ difference: that ethnic differences are to some extent natural.” Karlsen’s argument is that within epidemiological research, there is a tendency to downplay the fluidity of ethnic (as well as racial) categories and to interpret these as markers for a set of fixed, and to some extent innate and naturalized, attributes. Indeed, as Pfeffer (1998, 1382) noted, “essentialism can be social as well as biological. Essentialist versions of ethnicity (defined as a belief in a shared destiny) see history as an unchanging truth.… Essentialist accounts of culture are found in ‘factfiles,’ information resources produced for health professionals. They present cultures as fixed products rather than dynamic processes.” As such, there may be a tendency in health research to see ethnicity as a set of fixed social and cultural entities, rather than as fluid sociocultural concepts, just as race is routinely viewed as a fixed and natural biogenetic category, rather than a socially constructed identity in its own right. In the process, race and ethnicity tend to become conflated, however distinct they might be in the view of the researchers concerned.
• The limited reliability and external validity of racialized categories. Contextual variation in the meaning of race and ethnicity (and related racialized categories) is also at the root of concerns over the reliability and external validity of categories when operationalized within genetics and biomedical research. In particular, the contemporary practice of using self-identification to operationalize racialized categories (not least in the collection of data by contemporary censuses and health services information systems, which go on to be used by biomedical researchers) has been questioned in terms of both reliability (i.e., their repeatability) and external validity (i.e., their generalizability across contexts; see chapter 6). Changes to self-identification can be rapid and substantial. For example, Williams (1996, 487) reported that one “study of a large national population found that one-third of the U.S. population reported a different racial or ethnic status one year after their initial interview.” Moreover, Jones (2001, 300) described how “I would have been counted as a slave [before the U.S. Civil War], in 1850 as either Black or ‘mulatto,’ in 1890 as one of Black, mulatto, ‘quadroon,’ or ‘octoroon’ ancestry, in 1950 as ‘Negro,’ and in 2000 as ‘Black, African American, or Negro’ plus ‘White’ and ‘American Indian or Alaska Native’ if I so chose.” Jones’s account provides a powerful example of how changing social structures and norms challenge the concept of racialized categories as fixed over time (and place). Elsewhere, LaVeist (1994, 3) described how Brazil, Japan, and the United States have each produced “five different policies for assigning racial status” in under ten years. As such, the populations and individuals included under each racialized category differ over time and place, making it difficult to assume that these categories are equivalent from one data set to the next.
An additional consequence of these variable and unstable categories is that researchers might overlook associated problems with external validity. Hammerschmidt (1999, 10) highlighted how little attention has been paid to this issue: “it is as though authors have considered racial assignment to be as unambiguous and straightforward as gender assignment, as though terms such as ‘black’ and ‘white’ will mean the same thing in one study (or study population) as they mean in another. Such is demonstrably not the case.” Hahn, Truman, and Baker’s (1995) discussion drew the issues of reliability and external validity together by pointing out the very different measurement systems then in use to allocate racialized categories within the United States, which included self-identification and classification by proxy, by interviewer, or by funeral director. These authors concluded that the classification of a person’s racialized identity “varies over time and by method of ascertainment, relationship between person classifying and person classified, vital status of subject, and specific category of ancestry” (Hahn, Truman, and Baker 1995, 79). Unsurprisingly, different classification techniques tend to generate different numbers of people assigned to each category, making it difficult to identify the extent or possible causes of health disparities among racialized groups. This led McKenney and Bennett (1994, 19) to suggest that “different data collection methods, different content and format of the questions, and different definitions and classifications for race and ethnicity” were all “possible explanations for results to differ by race and ethnicity when data from the Bureau of the Census and the public health surveillance systems are used as the denominator and numerator.” Taken together, the points these writers made all serve to highlight the problems that occur as a result of different data collection methods (over time and place) and the lack of standardized data collection techniques (even in the same place and at the same time) as well as the paucity of information provided by researchers regarding how they operationalized the racialized categories they used (as discussed previously). All these issues combine to severely curtail the reliability and comparability of studies exploring disparities in health among racialized groups.
• The limited internal validity of racialized categories. The use of racialized categories as if these reflected discrete and homogeneous groups drew criticism from many of the articles included in this review, including those that emphasized the limited validity of such categories as markers of wholesale genetic difference and those that questioned their ability to capture socioculturally distinct groups. Some, like Hammerschmidt (1999, 11), argued that using self-identification to apply “one of five broad [racialized] categories may be a useful tool in monitoring for evidence of discrimination, but it falls far short of scientific assignment to a group expected to have greater-than-chance homogeneity in some genetic or physiologic variable of interest.” Others, like Caldwell and Popenoe (1995, 614), acknowledged that “single-word racial labels such as ‘black’ or ‘white’ are of occasional help to the clinician” but nonetheless recognized that, as a result of “their broad scope and lack of scientific clarity, these terms often poorly represent information—for example, about genetic risks and perceptions of disease—that they are supposed to convey.” In his critique, Nazroo (1998, 713) argued that “many studies use ‘ethnic’ groupings with quite inappropriate boundaries, such as Black or South Asian. The data are then interpreted as though the individuals within them are ethnically (i.e. genetically and culturally) homogeneous, even though such categories are heterogeneous, containing ethnic groups with different cultures, religions, migration histories, and geographical and socio-economic locations.” For these reasons, Caldwell and Popenoe (1995, 614) pointed out that “in many instances, [racialized categories]… are superficial and potentially misleading terms that fail to serve the patient’s medical needs.” As with the concerns over external validity described previously, the arguments presented here highlight another of the often hidden problems facing the use of racialized categories in biomedical research, particularly when this involves (or invokes) assumptions regarding the extent of genetic and/or sociocultural homogeneity within, and heterogeneity between, racialized groups.
• The use of racialized categories to infer genetic causality for disparities in health. Relating concerns about internal validity to the contentious issue of causal inference, LaVeist (1994, 8) pointed out that “a statistically significant coefficient for the race binary variable without further analysis often leads to such illogical, yet commonly published conclusions as, ‘race is a significant determinant of prenatal care utilization.’ Such a conclusion eventually filters into medical and public health practice… [yet] Clearly, a person’s skin color does not determine prenatal care utilization.” LaVeist’s argument, that racialized categories themselves should not be considered the specific or direct cause of disparities in health among racialized groups, was supported by many of the other articles included in this review. For example, Smaje (1996, 141) suggested that “on the one hand… [race] has been regarded as an explanatory principle sui generis, which can itself explain various dimensions of human action. On the other, it can be constituted as an object of analysis, something to be explained with reference to other modes of human action.” The causal interpretation, whereby racialized categories are, of themselves, seen as sufficient to explain disparities in health—rather than as phenomena that need to be explained—has also been criticized by Buescher, Gizlice, and Jones-Vessey (2005, 397), who argued that “racial group is at best a crude marker for particular health problems, and certainly not a risk factor or cause.”
Elsewhere, several authors of articles included in this review have criticized the tendency among biomedical scientists to use well-known and highly penetrative single-gene disorders, associated with racialized groups (particularly hemoglobinopathies), as models for understanding broader inequalities in health on three grounds: first, that the distribution of the alleles responsible for these single-gene disorders are not necessarily concordant with the distribution of other health-related alleles; second, that the impact of these highly penetrative single-gene disorders provide a poor model for understanding the impact of genetic variation on health-related phenotypes dependent on an interaction between more than one gene and more than one aspect of the environment; and third, that only a tiny proportion of disparities in health among racialized groups are the result of highly penetrative single-gene disorders. For example, Krieger et al. (1993, 85) argued that “the accumulated evidence indicates that for virtually every racial/ethnic group, a handful of genetic diseases seem specifically associated with geographic and biological heritage, yet these diseases nonetheless account for only a minute percentage of each group’s overall morbidity and even less of their mortality.” Similarly, Kaufman, Cooper, and McGee (1997, 621) pointed out that “single gene disorders such as hemoglobinopathies collectively make up less than 0.3% of the differential mortality burden [between blacks and whites in the United States].” Indeed, Krieger and Bassett (1986, 77) have suggested that “such uncommon genetic maladies have become important strictly because of their metaphorical value: they are used to support genetic explanations of racial differences in the ‘big diseases’ of the twentieth century—heart disease, stroke, and cancer. Yet no current evidence exists to justify such an extrapolation.”
Meanwhile, a further concern regarding the analytical use of racialized categories in biomedicine is the incompleteness of statistical adjustment for differences in socioeconomic status among different racialized groups—an issue that Kaufman, Cooper, and McGee (1997, 621) felt was a major “threat to the validity” of such analyses. If residual confounding from incomplete adjustment is present, the assumption that any residual differences observed following adjustment are free from socioeconomic influences cannot be justified. The possibility of residual confounding was also highlighted by O’Loughlin (1999, 153), who pointed out that confounding will continue to “be a problem if measurement of socioeconomic status is incomplete,” and by Karter (2003, 2193), who concluded that “one can safely assume that many [socioeconomic and social] factors are missing (not collected or not specified) from a statistical model that attempts to explain racial/ethnic differences.” However, an important qualification of this concern, and one that questions the rationale for socioeconomic adjustment in the first place, was raised by Smaje (1996, 159), who pointed out that adjusting for socioeconomic status when exploring disparities in health among racialized groups “can have the unfortunate effect of directing attention away from a key analytical question, namely the nature of the relationship between ethnicity, socio-economic status and health.” In essence, Smaje was concerned that socioeconomic status might lie on the causal pathway between racialized categories and disparities in health because socioeconomic status is likely to be the result of historical and contemporary discrimination. This is an issue that had previously been addressed by Cooper and David (1986, 111), who argued that “explaining racial differentials by education, income, etc. could in a causal sense be considered ‘over-control’; race is not confounded by other variables, it is antecedent to them.” Moreover, Kaufman and Cooper (2001) have since pointed out that it might not be possible to use racialized categories in multivariate analyses of this sort if, as they and others suggest, socioeconomic status is partly determined by the differential treatment of racialized groups, because these differences are largely inseparable from the racialized categories concerned. This important technical point hinges on the fact that racialized categories cannot be used in the sorts of multivariate analyses capable of generating plausible etiological insights because the categorical variables involved need to be amenable to counterfactual comparisons—that is, analyses capable of assessing what outcomes might have occurred had the study participants had different characteristics. Since it is often difficult to envisage how racialized categories might be reassigned—either hypothetically or empirically (Erasmus and Ellison 2005; 2008)—it is usually impossible to separate out those aspects of disparities in health that are associated with racialized groups from those associated with the socioeconomic differences such groups experience.
Taken together, these arguments present a strong critique of analyses that assume that disparities in health among racialized groups offer a definitive insight into their genetic, structural, and/or sociocultural determinants. Indeed, they suggest that alongside assumptions about the genetic (and sociocultural) homogeneity of racialized groups, which make them appear a self-evident explanation for disparities in health, the use of racialized categories in biomedical research is likely to detract from the more detailed, in-depth investigations required to explore the multifactoral issues responsible.
• Research on racialized groups can lead to the development of inappropriately targeted interventions. For the most part, the articles included in this review focused on conceptual and methodological issues related to the operationalization of racialized categories as markers of genetic (and, to some extent, sociocultural) differences in biomedical research (as outlined previously). However, some of the articles also voiced criticisms and concerns regarding the application of research using racialized categories to clinical practice. These manifested as concerns that the routine use of racialized categories in biomedical research might imply that specific health care services should be developed for different racialized groups and lead to the stigmatization and stereotyping of the least healthy groups. The first of these issues is discussed under this theme, while the second will be discussed in the theme that follows.
Commenting on how common it can be for racialized categories to be invoked in clinical contexts, Caldwell and Popenoe (1995, 614) highlighted how “in many institutions, medical students are routinely taught to begin their case presentations with a statement describing the age, sex, and ‘race’ of the patient.” While such practices appear benign, they may inadvertently lead medical practice along diagnostic and therapeutic pathways that mistakenly assume that racialized groups are accurate markers of etiological risk. Witzig (1996, 676) referred to this as the medicalization of racialized categories—the process of transforming them into medical (rather than social) variables—and concluded that “unfortunately, the continued appearance of race taxons in the medical literature has legitimized them as acceptable descriptive labels for patients and has thus made them seem integral to the proper diagnosis and treatment of diseases.” This reification of racialized categories as integral to (if not crucial for) clinical diagnosis and treatment, based as it is on common underlying assumptions that such categories are helpful markers of innate biological (i.e., genetic) causes, has been criticized for unduly narrowing the range of diagnoses and potential treatments applied to individual patients. In particular, Williams (1997, 324) described how “physicians can use assumptions about a patient’s race to prematurely eliminate possible diseases or to inappropriately narrow the focus to one disease in the diagnosis of patients.”
Given their common, if not routine, use in the clinical encounter, it is perhaps unsurprising that racialized categories have also provided a framework for pharmacogenomic research, in which assumptions about innate differences in the efficacy and safety of drugs among different racialized groups have made racialized categories key variables in the development, prescription, and monitoring of drugs. The most prominent example of this framework has been the heart failure drug BiDil (Lillquist and Sullivan 2004; Ellison 2006; Ellison et al. 2008b). Clearly the routine collection of data on racialized groups, together with common assumptions about genetic homogeneity within such groups, has resulted in practices whereby diagnostic and treatment regimes (including drugs) have been tailored to specific groups. While some such tailoring may be beneficial (i.e., when the assumptions involved turn out to be correct), given the concerns about the limited reliability and validity of racialized categories as useful markers of difference (be they genetic or sociocultural), it seems likely that translating such research into clinical practice would exclude some populations from gaining access to appropriate medical treatment and pharmaceutical products.
• Using racialized categories to explore disparities in health can lead to stereotyping. The last of the criticisms and concerns leveled at the use of racialized categories in biomedical research focused on their potential impact on the stigmatization and stereotyping of less healthy racialized groups. According to Williams, Lavizzo-Mourey, and Warren (1994, 34), assuming that racialized categories are in some way causal determinants of disparities in health, and thereby failing “to identify the specific factors that contribute to group differences, can reinforce racial prejudices and perpetuate racist stereotypes, diverting both public opinion and research dollars from the larger social factors that ultimately account for the patterns of disease variation.” In Europe and North America, where much of this research takes place, a principal reason for such stereotyping is simply that the majority “white” population tends to be treated either as nonracialized or the norm, against which other populations are thereby racialized and appear deviant. This is evident in Drevdahl, Phillips, and Taylor’s (2006, 56) assertion that “by classifying all forms of racial or ethnic diversity into ‘non-white’ or ‘ethnic minority,’ ‘white’ was the implied but often unstated norm against which all ‘other’ groups were compared.” Because racialized deviance is commonly assumed to be genetic (or at least the product of heritable, innate, and immutable biosocial characteristics), it runs the risk of deflecting attention away from structural and related socioeconomic causes of disparities in health. Indeed, recent advances in genetic technology have elicited a shift in emphasis onto genetic explanations for a whole range of essentially social problems, as McCann-Mortimer, Augoustinos, and LeCouteur (2004, 412) explained: “entrenched social problems such as poverty, educational underachievement, mental illness, delinquency, alcoholism, violence, and criminal behaviour… [are] being increasingly attributed to ‘deficient’ or ‘problematic’ genes by experts, rather than to the social conditions in which people lived.” Thus, while discrimination against racialized groups may not be an inevitable product of contemporary biomedical research using racialized categories, some of the articles included in this review have argued, as do Osbourne and Feit (1992, 275), that “it is naïve to believe that research methods can always compensate for racial bias.” As a result, even health researchers who have explicitly set out to reduce health inequalities between racialized groups may inadvertently contribute to a circular argument, in which disparities in health among racialized groups are accepted as reflecting innate differences that are not amenable to intervention and thereby entrench and reproduce the disparities observed—disparities that may, directly and indirectly, perpetuate these inequalities by justifying inaction or discrimination. Whatever the actual impact of biomedical research using racialized categories as (if they were) accurate markers of innate difference, it seems clear that researchers and clinicians should recognize the potential impact of such research on future research and clinical practice as well as on lay perceptions of, and justifications for, persistent disparities in health among racialized groups.
Discussion
Limitations
Given the wealth of literature engaging critically with the use of racialized categories in genetics and biomedical research, this review was necessarily limited in scope. First, it focused exclusively on articles that had problematized the use of racialized categories as (if they were) markers of genetic variation in biomedical research and on the criticisms and concerns these articles contained. As such, the potential benefits of using racialized categories were not recorded or analyzed. This is an important qualification given that such benefits might, under some circumstances, outweigh any disadvantages of using racialized categories in genetics and biomedical research (Ellison 2005). Meanwhile, it is also important to point out that the literature review presented here will have been limited by the citation search strategy used (based on the McKenzie and Crowcroft editorial of 1996), which excluded any unconnected articles (i.e., any outside this citation network). It is therefore possible that this approach might have generated a largely self-referential collection of articles that excluded alternative, minority, and/or controversial views. Certainly some of the most influential articles on this topic—such as Lewontin’s (1972) assessment of genetic variation within and between racialized groups—were not included in the articles in this citation network and were therefore not included in this review. However, while it is possible that the citation search strategy reduced the diversity of opinions found, it is clear from the review’s findings that the articles included expressed a wide range of different, and occasionally contradictory, views. It therefore seems unlikely that the search strategy used generated an assessment of the criticisms and concerns that was unduly biased or limited, except inasmuch as this drew exclusively on articles within a citation network emanating from an editorial in a medical journal (i.e., McKenzie and Crowcroft 1996).
Key Findings
Notwithstanding these limitations, the key unifying element in many of the arguments reviewed here is that their foundation lies in the unresolved interpretation of human genetic variation. More specifically, as Lewontin (1972, 397) put it, given that “less than 15% of all human genetic diversity is accounted for by differences between human groups,” it can be argued that “racial classification is now seen to be of virtually no genetic or taxonomic significance either, [and, therefore] no justification can be offered for its continuance.” While this may seem a very straightforward proposition, when the same genetic data are analyzed in terms of their correlation structure (i.e., when patterns among numerous genetic traits, rather than the presence or absence of individual genetic traits in different populations, are analyzed within populations), they can be used to argue that “the ‘taxonomic significance’ of genetic data in fact often arises from correlations amongst the different loci, for it is these that may contain the information which enables a stable classification to be uncovered” (Edwards 2003, 798). Indeed, from Edwards’s perspective, Lewontin’s often quoted figure of “less than 15%” is essentially an illusion (if not a fallacy) because Lewontin’s “argument ignores the fact that most of the information that distinguishes populations is hidden in the correlation structure of the data and not simply in the variation of the individual factors” (Edwards 2003, 798). This very different interpretation of very similar data is partly what underlies contemporary disagreements about the genetic utility of racialized categories (Smart et al. 2006), and failure to resolve these disagreements has resulted in stalemate between those scientists who want to use racialized categories within genetics and biomedicine and those who criticize this practice. What is clear, however, is that technological improvements in our ability to measure and characterize genetic variation since Lewontin (1972) have not fundamentally changed the genetic arguments against the use of racialized categories as (if they were) markers of genetic variation. As Weiss and Fullerton (2005, 168) argue, “We already know the facts. In that sense, the endless cycling could stop, because we’re already there. But people stubbornly continue to see what they want to see in the facts—either that or what they want to see determines which facts count.” While the arguments identified in this review might justifiably be used to limit claims made with respect to the genetic basis of disparities in health among racialized groups, they do not mean that these claims will always be either incorrect (according to the genetic data) or necessarily counterproductive (in terms of subsequent biomedical analyses or applications). Instead, the use of racialized categories leaves genetics and biomedical science in something of a quandary—how should they deal with a fluid sociopolitical category that sometimes appears to be associated with what might turn out to be a useful amount of genetic variation?
Why have the criticisms leveled at racialized categories failed to curtail their use in genetics and biomedical research? Given the substantial volume and breadth of the criticisms and concerns included in this review, and the challenges these pose more broadly for science and clinical practice elsewhere, one might assume that geneticists and biomedical researchers would be only too pleased to abandon racialized categories. However, in the discussion that follows, we will argue that although arguments against the use of racialized categories appear well founded, they leave enough room for interpretation to make the use of such categories an attractive option for geneticists and biomedical researchers. During the course of the discussion, this process of interpretation is referred to as selective engagement—that is, the way in which the arguments themselves leave room for scientists to acknowledge the limitations of racialized categories and engage with these arguments, while, at the same time, continuing to emphasize the utility of racialized categories in their work.
This process of selective engagement can be broken down into three parts. The first is the process of selective engagement with definitional arguments and the methods employed to operationalize racialized categories. The second is selective engagement with the analysis and interpretation of observable health inequalities—particularly how selective engagement provides a framework through which epidemiological studies of disparities in health among racialized groups are interpreted as largely the product of genetic variation. Finally, it is argued that this genetic framework perspective has come to dominate clinical practice. This, arguably, produces a largely self-fulfilling frame of reference when setting research objectives and interpreting research findings—one that appears highly attractive to researchers wishing to find genetic answers to a variety of health questions, including the presumed genetic basis of disparities in health among racialized groups (Fine, Ibrahim, and Thomas 2005). However, we will argue that this process of selective engagement is only partially successful, leaving geneticists and biomedical researchers with the difficult task of justifying their work in the public sphere, while at the same time attempting to maintain the boundaries between science and society they have sought to create/reinforce.
While the lack of conceptual and definitional consensus surrounding racialized categories implies instability, which is, in turn, contrary to the requirements of scientific research (Rivara and Finberg 2001; Winker 2004), it is also indicative of the wide range of choices available to researchers with respect to the meanings applied to, and operationalization of, such categories. The failure to clarify these issues within biomedical dictionaries, and elsewhere, leaves researchers with an opportunity to view racialized categories as “relatively homogeneous with respect to biological inheritance” (Last 1995, 198). Although more recent definitions of some dictionaries appear to emphasize caution, this does not entirely discount the notion that such categories can be used as markers for genetic variation. Subsequently, researchers are relatively free to conceptualize and operationalize these categories in a variety of ways. For example, this is likely to manifest itself in moving from a conceptualization of racialized categories that assumes/implies that these are biogenetic entities but uses inherently nonbiological (and, to some extent, nonscientific) methods for collecting data thereon (such as those based on self-identification and official census categories; Smart et al. 2008). The ambiguity and variability in data collection methods, together with the absence of a specific conceptual rationale, is a common theme in articles critiquing the use of these categories in genetics and biomedical research (not least with respect to their internal and external validity), and these issues have also been highlighted in more recently published articles reviewing the use of such categories in human genetics and biomedical research (e.g., Hunt and Megyesi 2008; Sankar, Cho, and Mountain 2007; Shanawani et al. 2006). In this context, racialized categories become unexamined, commonsense entities, sustained without critical examination by their frequent appearance in the genetic and biomedical literature. As Epstein (2004, 1995) has argued, although “racial categories are culturally variable and change over time… [this] does little to disturb the common-sense understanding of racial difference.” To summarize, within the scientific community, the popular conceptual basis underlying the social and scientific meaning of racialized categories escapes close scrutiny by a selective, partial engagement with both the data on genetic and sociocultural variation and the arguments against these (Outram and Ellison 2006a). And by routinely avoiding the explicit recognition of either the conceptual rationale or the methodological basis for using racialized categories, these scientists avoid close scrutiny of any glaring contradictions, not least those inherent in the collection of genetic markers using self-identification, and uncritical assumptions about the reliability and validity of such data.
With respect to the analytical and interpretative arguments against the use of racialized categories as (if they were appropriate) markers of genetic variation within genetics and biomedical research, a similar pattern of selective engagement occurs. Thus genetic scientists are prone to argue, when questioned about their use of racial categories, that they are able to use race as a scientific category within their research as a loose marker of genetic affiliation (and, by implication, as a useful biomedical marker in clinical practice), while at the same time distancing themselves from the sociopolitical attributes of these categories (Outram and Ellison 2006b). Critics of this position argue that such selective distancing leaves genetic science open to potential misinterpretation (or deliberate manipulation) to subscribe a genetic cause to inequalities in health between populations and deflect attention away from the impact of discriminatory practices within society (and within the medical establishment) that continue to impact on health. While many of the articles included in this review argued that the separation of the biological from the social is untenable (due to residual confounding and the absence of an appropriate counterfactual thesis; Kaufman and Cooper 2001), geneticists and biomedical researchers act as if they have been able to achieve this separation by framing the interpretation of racialized analyses in ways that appear to make them immune from such critiques. As such, while these scientists might accept some of the arguments against using racialized categories as interesting insights into alternative (i.e., nongenetic) explanations for disparities in health among racialized groups, they may still assume that the disparities in health they have studied are primarily the product of genetic variation. As Krieger (2005, 2156) has argued, this produces a largely self-referential body of analytical interpretation: “unobserved ‘innate’ biological differences lead to observed biological differences—which in turn prove that unobserved ‘innate’ biological differences exist.” This a priori focus on genetic etiology has taken hold of a considerable body of epidemiological analysis and is based on what Kaufman (2008, 1668) calls an “apparent eagerness to embrace the message of racial essentialism,” which “seems to represent a very strong prior belief on the part of many researchers.” Kaufman (2008, 1668) concludes that “until this strong predilection for racial essentialism in biological thinking abates, there would seem to be little hope that a more sensible and honest approach to statistical inference in observational data will take hold more widely”—that is, one that questions the independence of genetic, structural, and sociocultural factors and engages more fully with the challenges facing the analysis of the multifactoral causes for disparities in health among racialized groups. In this context, the lack of a thoroughly examined theoretical or conceptual basis for using racialized categories in genetics and biomedical research appears to produce a largely unexamined and self-perpetuating analytical/interpretative framework (Weed 2000). Once racialized categories are used in this self-referential manner, it becomes difficult to interpret disparities in health among racialized groups (or, for that matter, differences in any other characteristics) as anything other than genetically derived. Moreover, as Ahmad and Bradby (2007, 798) have argued, this conceptual, analytical, and interpretive framework may not only distort the perception of why health disparities exist (i.e., their etiology), but may also extend to victim blaming: “using culture and ethnicity as an explanation of inequality between groups [which] distorts perceptions of how ethnic relations, and the related inequities of power are produced. Defined by those in power, the disadvantage of minority ethnic groups too often continues to be seen as ‘caused’ by their diseased genetic and dysfunctional cultural inheritance.” In attempting to rectify the tension between the potential utility and the potential dangers of using racialized categories, we have sought to steer a course whereby arguments against the use of racial groups within genetics and biomedical research have been identified and discussed; set against the context of what we know about genetics, health, and health care; and explored in relation to the sociopolitical and individual impact of racialization. Overall, we have argued that the analytical and interpretative framework that assumes racial identity to be strongly affiliated to genetic variation is neither scientifically defensible nor politically sustainable in the context of its potential social harms.
Finally, with respect to concerns about the application of racialized categories as genetic variables, it can be argued that both selective engagement and disengagement occur. First, with respect to the narrowing of diagnoses and treatment options, it can be seen that arguments against the use of racialized categories as (if they were appropriate) markers of genetic variation follow directly from the narrowing of analytical interpretations described previously. A preoccupation with genetic causes among researchers examining the etiology of health disparities among racialized groups can, and arguably has (Witzig 1996), led to a routine emphasis on racialized categories within medicine such that these categories increasingly influence the diagnostic and treatment pathways selected and are increasingly seen as an essential element of day-to-day clinical practice. Together with what Frank (2007, 1981) has referred to as the “head-long rush toward a genetic explanation for any race/ethnic difference in disease prevalence or etiology,” it is easy to see how the triangular association between race, genes, and health has become such a robust and self-referential framework.
While most biomedical research is carried out with the explicit or implicit intention of improving clinical practice, a focus on racialized groups in such research is likely to have a number of (unanticipated and negative) social consequences. For example, efforts to target health services at specific racialized groups—such as the screening programs for sickle-cell anemia and Tay-Sachs disease that have been developed at various times in the past—are ostensibly attempts to use the findings of research on racialized groups to better direct limited health resources to groups at greater risk of disease (Aspinall, Dyson, and Anionwu 2003; Dyson et al. 2006). Yet they also provide mechanisms for stigmatizing and stereotyping the groups concerned as inherently unhealthy (Ahmad and Bradby 2007; Brandt-Rauf et al. 2006). Geneticists and biomedical researchers may find these bioethical implications difficult to engage with within the context of scientific work that is intended to be, and is often interpreted and presented as, objective and value-free. Indeed, it is not possible to deal with such concerns without either abandoning the distinctions between different scientific disciplines (and between science and society) that are so central to the culture of much scientific research (Gieryn 1995; Goodman and Leatherman 1998) or crossing over academic and contextual boundaries between the natural and social sciences and between pure and applied research. By focusing inward—on scientific methods and results that support the triangular framework of race, genes, and health—while largely ignoring the complexity and conceptual ambiguity of the tools used, it would seem that this form of scientific practice can present itself as being untainted by social values, pragmatic concerns, or unintended consequences. However, by embracing the racialized categories and labels that have become recognized as socioculturally meaningful (not least through their adoption by censuses as sociopolitical realities), genetics and biomedical science run the risk of entering a contentious social, political, and ethical arena in which scientific practice is subject to unprecedented scrutiny and control. In the process, scientific practice can become aligned to broader sociopolitical enterprises, as was the case with policies designed to tackle the underrepresentation of “minority” populations in biomedical research (which began in the United States and United Kingdom during the 1980s)—policies that have, somewhat paradoxically, resulted in an even greater emphasis on differences between racialized groups than had previously occurred (Epstein 2004).
Conclusion
Looking back over the criticisms and concerns identified during the course of this review, it is difficult to see how the routine use of racialized categories as (if they were) markers of genetic variation—one that is seldom accompanied by an explicit or adequate conceptual, methodological, or analytical rationale—might do anything other than undermine biomedical research and practice. Indeed, there appears to be a strong prima facie case for assuming that racialized categories are inappropriate approximations of medically important aspects of genetic variation. Likewise, there seems to be a strong case against assuming that disparities in health between racialized groups are usually or necessarily the result of genetic variation. Of course, in a few very specific contexts—that is, with very specific populations and for very specific genetic conditions—racialized categories might offer sufficiently precise markers of genetic variation to be etiologically or therapeutically useful (Ellison 2005). But such contexts appear to be far less common than is generally assumed, and even within these (very specific) contexts, it may still be more appropriate to appraise all potential determinants of disparities in health (i.e., structural, sociocultural, and genetic), rather than focusing on just one of these—not least whenever there is any etiological, diagnostic, or therapeutic uncertainty. This more holistic approach would have the benefit of establishing the precise relationship between race, genes, and health, including the role that structural and/or sociocultural factors might play in generating, exacerbating, or sustaining the impact of genetic variation on health.
Meanwhile, an ancillary conclusion that might be drawn from the articles examined in this review is that the use of racialized categories within genetics and biomedical research cannot be dissociated from their sociopolitical contexts or consequences—not least because both race and health are biosocial phenomena (i.e., with biological and sociocultural characteristics). In negotiating between the potential benefits of using racialized categories within scientific contexts and the potential structural and sociocultural ills this might cause, we have argued that conceptually distancing science from society when referring to racialized groups is likely to be counterproductive. Viewing the sociopolitical attributes of racialized categories as beyond the sphere of scientific enquiry may leave geneticists and biomedical scientists unable to understand the full meaning(s) of such categories and the ultimate consequences of using these as (if they were appropriate) markers of genetic variation. This self-imposed restriction on engaging with alternative meanings and interpretations of racialized categories denies geneticists and biomedical scientists the benefit of insight from the social sciences. In turn, it also limits the ability of the former to enter the clinical and political arenas whenever they see their work being inappropriately applied to health care policies and practices, or used to stigmatize, stereotype, and discriminate.
Thus, somewhat paradoxically, while a propensity among geneticists and biomedical scientists to selectively engage with arguments against the use of racialized categories as (if they were appropriate) markers of genetic variation seems to reify these categories as scientifically meaningful, the ultimate appeal of racialized categories remains their ability to capture the meld of genetic and social characteristics capable of generating persistent (and socially heritable) disparities in health between racialized groups. More specifically, these categories are likely to be crucial for any analyses of disparities in health within racialized societies simply because they are essential for identifying, exploring, and thereby addressing the consequences of injustice based on notions of racialized groups as genetically distinct. However, it is increasingly recognized that conceptualizing racialized categories as either sociopolitical or biogenetic constructs undermines the ability of genetics and biomedical research to identify their true biosocial character and how this character generates the circular arguments that sustain inequity (Dingwall, Nerlich, and Hillyard 2006; Duster 2003). By recognizing the biosocial (or what Goodman and Leatherman [1998] have called the biocultural) character of racialized categories, it may yet be possible to use these sensitively within the natural and social sciences as well as within scientific and sociopolitical arenas to address disparities in health (and related structural and sociocultural disparities), without generating the evidence needed to sustain the stigmatization, stereotyping, and unfair treatment of racialized groups.
Notes
1. We have previously conducted two quantitative reviews to examine the contemporary relevance of these issues. In the first of these, we surveyed the 102 journals categorized under “Medicine, General and Internal” within the ISI Web of Knowledge’s Science Citation Index (SCI) Journal Citation Reports. This survey found that between 1995 and 2005, 90 of these 102 (88%) journals contained articles that referred to “race,” “racial,” “ethnic,” and/or “ethnicity” in their titles and/or abstracts. A comparable survey of the 120 journals listed under the SCI category “Genetics and Heredity” found that 70 (58%) contained articles that referred to “race,” “racial,” “ethnic,” and/or “ethnicity” in their titles and/or abstracts (Outram and Ellison 2006c). These figures appear to correspond reasonably well with other surveys of genetics journals (Sankar et al. 2007) and biomedical and public health journals (Ahdieh and Hahn 1996; Comstock, Castillo, and Lindsay 2004; Drevdahl, Taylor, and Phillips 2001; Ellison and de Wet 1997; Jones, LaVeist, and Lillie-Blanton 1991; Walsh and Ross 2003; Williams 1994), which have found that anything between 40% and 75% of articles referred to racialized categories in some form or another As such, the issue of racialized categories (i.e., their use and the criticisms associated with this use) cannot simply be seen as a side issue to much of the research published in these fields. Instead, concerns regarding the use of racialized categories in genetics and biomedical research raise fundamental questions about such research and, by implication, about the clinical services and public health programs based thereon, not least those seeking to address persistent disparities in health between racialized groups in a range of different countries; recent examples of such work include Jamieson, Armfield, and Roberts-Thomson (2007), Matijasevich et al. (2008), Goh et al. (2007), and Mann et al. (2008)—studies from Australia, Brazil, Malaysia, the United Kingdom, and the United States.
References
Ahdieh, L., and R. A. Hahn. 1996. Use of the terms “race,” “ethnicity,” and “national origins”: A review of articles in the American Journal of Public Health, 1980–1989. Ethnicity and Health 1:95–98.
Ahmad, W. I. U., and H. Bradby. 2007. Locating ethnicity and health: Exploring concepts and contexts. Sociology of Health and Illness 29:795–810.
Aspinall, P. J. 1997. The conceptual basis of ethnic group terminology and classifications. Social Science and Medicine 45:689–698.
Aspinall, P. J., S. M. Dyson, and E. N. Anionwu. 2003. The feasibility of using ethnicity as a primary tool for antenatal selective screening for sickle cell disorders: Pointers from the research evidence. Social Science and Medicine 56:285–297.
Baum, B. 2006. The rise and fall of the Caucasian race: A political history of identity. New York: New York University Press.
Bhopal, R. 2004. Glossary of terms relating to ethnicity and race: For reflection and debate. Journal of Community Health 58:441–445.
Boyatsiz, R. E. 1998. Transforming qualitative information: Thematic analysis and code development. Thousand Oaks, CA: Sage.
Bradby, H. 1995. Ethnicity: Not a black and white issue. A research note. Sociology of Health and Illness 17:405–417.
Brandt-Rauf, S. I., V. H. Raveis, N. F. Drummond, J. A. Conte, and S. M. Rothman. 2006. Ashkenazi Jews and breast cancer: The consequences of linking ethnic identity to genetic disease. American Journal of Public Health 96:1979–1988.
Braun, L., A. Fausto-Sterling, D. Fullwiley, E. M. Hammonds, A. Nelson, W. Quivers, S. M. Reverby, and A. E. Shields. 2007. Racial categories in medical practice: How useful are they? PLoS Medicine 4:e271.
Braun, V., and V. Clarke. 2006. Using thematic analysis in psychology. Qualitative Research in Psychology 3:77–101.
British Medical Journal. 1996. Style matters: Ethnicity, race, and culture: Guidelines for research, audit, and publication. British Medical Journal 312:1094–1095.
Buescher, P. A., Z. Gizlice, and K. A. Jones-Vessey. 2005. Discrepancies between published data on racial classification and self-reported race: Evidence from the 2002 North Carolina live birth records. Public Health Reports 120:393–398.
Caldwell, S. H., and R. Popenoe. 1995. Perceptions and misperceptions of skin color. Annals of Internal Medicine 122:614–617.
Cartmill, M. 1999. The status of the race concept in physical anthropology. American Anthropologist 100:651–660.
Cavalli-Sforza, L. L., A. C. Wilson, C. R. Cantor, R. M. Cook-Deegan, and M. C. King. 1991. Call for a worldwide survey of human genetic diversity: A vanishing opportunity for the Human Genome Project. Genomics 11:490–491.
Comstock, R. D., E. M. Castillo, and S. P. Lindsay. 2004. Four-year review of the use of race and ethnicity in epidemiologic and public health research. American Journal of Epidemiology 159:611–619.
Cooper, R. S. 1984. A note on the biological concept of race and its application to epidemiologic research. American Heart Journal 8:715–723.
Cooper, R. S. 2005. Race and IQ—Molecular genetics as deus ex machina. American Psychologist 60:71–76.
Cooper, R. S., and R. David. 1986. The biological concept of race and its application to public health. Journal of Public Health Politics, Policy, and Law 11:97–116.
Dingwall, R., B. Nerlich, and S. Hillyard. 2006. Biological determinism and its critics: Some lessons from history. In The nature of difference: Science, society, and human biology, ed. G. T. H. Ellison and A. Goodman. Boca Raton, FL: Taylor and Francis, 17–34.
Drevdahl, D. J., D. A. Philips, and J. Y. Taylor. 2006. Uncontested categories: The use of race and ethnicity variables in nursing research. Nursing Inquiry 13:52–63.
Drevdahl, D., J. Y. Taylor, and D. A. Phillips. 2001. Race and ethnicity as variables in nursing research, 1952–2000. Nursing Research 50:305–313.
Duster, T. 2003. Buried alive: The concept of race in science. In Genetic nature/culture: Anthropology and science beyond the two-culture divide, ed. A. H. Goodman, D. Heath, and M. S. Lindee. Berkeley: University of California Press, 258–277.
Edwards, A. W. 2003. Human genetic diversity: Lewontin’s fallacy. BioEssays 25:798–801.
Ellison, G. T. H. 1999. Contemporary definitions of “race” and “ethnicity” in medical dictionaries. Annals of Human Biology 27:104–105.
Ellison, G. T. H. 2005. “Population profiling” and public health risk: When and how should we use race/ethnicity? Critical Public Health 15:65–74.
Ellison, G. T. H. 2006. Medicine in black and white: BiDil®, race and the limits of evidence-based medicine. Significance 3:118–121.
Ellison, G. T. H., and T. de Wet. 1997. The use of “racial” categories in contemporary South African health research. South African Medical Journal 87:1671–1679.
Ellison, G. T. H., and I. R. Jones. 2002. Social identities and the “new genetics”: Scientific and social consequences. Critical Public Health 12:265–282.
Ellison, G. T. H., J. Kaufman, R. F. Head, P. Martin, and J. Kahn. 2008b. Flaws in the US Food and Drug Administration’s rationale for supporting the development and approval of BiDil® as a treatment for heart failure in black patients. Journal of Law, Medicine, and Ethics 36:449–457.
Ellison, G. T. H., A. Smart, R. Tutton, S. M. Outram, R. Ashcroft, and P. A. Martin. 2007. Racial categories in medicine: A failure of evidence-based practice? PLoS Medicine 4:e287.
Ellison, G. T. H., R. Tutton, S. M. Outram, P. A. Martin, R. Ashcroft, and A. Smart. 2008a. An interdisciplinary perspective on the impact of genomics on the meaning of “race” and the future role of racialized categories in biomedical research. NTM Journal of the History of Science, Technology, and Medicine 16:378–386.
Epstein, S. 2004. Bodily differences and collective identities: The politics of gender and race in biomedical research in the United States. Body and Society 10:183–203.
Erasmus, Y., and G. T. H. Ellison. 2005. Race reclassification and socio-political hierarchy in apartheid South Africa. Annals of Human Biology 32:792.
Erasmus, Y., and G. T. H. Ellison. 2008 What can we learn about the meaning of race from the classification of population groups during apartheid? South African Journal of Science 104: 450–452.
Fine, M. J., S. A. Ibrahim, and S. B. Thomas. 2005. The role of race and genetics in health disparities research. American Journal of Public Health 95:2125–2128.
Frank, R. 2007. What to make of it? The (re)emergence of a biological conceptualization of race in health disparities research. Social Science and Medicine 64:1977–1983.
Freeman, H. P. 1998. The meaning of race in science—considerations for cancer research. Cancer 82:219–225.
Garte, S. 2002. The racial genetics paradox in biomedical research and public health. Public Health Reports 117:421–425.
Gieryn, T. F. 1995. Boundaries of science. In Handbook of science and technology studies, ed. S. Jasinoff, G. E. Markle, J. C. Petersen, and T. Pinch. Thousand Oaks, CA: Sage, 145–167.
Gimenez, M. E. 1989. Latino/“Hispanic”—who needs a name? The case against a standardized terminology. International Journal of Health Services 19:557–571.
Goh, K. L., P. L. Cheah, M. D. Noorfaridah, K. F. Quek, and N. Parasakthi. 2007. Ethnicity and H. pylori as risk factors for gastric cancer in Malaysia: A prospective case control study. American Journal of Gastroenterology 102:40–45.
Goodman, A. H. 1997. Bred in the bone? Sciences 37:20–25.
Goodman, A. H., and A. H. Leatherman. 1998. Traversing the chasm between biology and culture. In Building a new biocultural synthesis: Political-economic perspectives on human biology, ed. A. H. Goodman and T. L. Leatherman. Ann Arbor: University of Michigan Press, 3–41.
Hahn, R. A., B. I. Truman, and N. D. Baker. 1995. Identifying ancestry: The reliability of ancestral identification in the United States by self, proxy, interviewer, and funeral director. Epidemiology 7:75–80.
Hahn, R. A., S. F. Wetterhall, C. A. Gay, D. S. Harshbarger, C. A. Burnett, R. G. Parrish, and R. J. Orend. 2002. The recording of demographic information on death certificates: A national survey of funeral directors. Public Health Reports 117:37–43.
Hammerschmidt, D. E. 1999. It’s as simple as black and white! Race and ethnicity as categorical variables. Journal of Laboratory and Clinical Medicine 133:10–12.
Hart, C. 1998. Doing a literature review: Releasing the social science research imagination. London: Sage.
Hart, C. 2001. Doing a literature search: A comprehensive guide for the social sciences. London: Sage.
Hunt, L. M., and M. S. Megyesi. 2008. The ambiguous meanings of the racial/ethnic categories routinely used in human genetics research. Social Science and Medicine 66:349–361.
Huth, E. J. 1995. Identifying ethnicity in medical papers. Annals of Internal Medicine 122:619–621.
Jamieson, L. M., J. M. Armfield, and K. F. Roberts-Thomson. 2007. Indigenous and non- indigenous child oral health in three Australian states and territories. Ethnicity and Health 12:89–107.
Jenkins, R. P. 2001. Rethinking ethnicity: Arguments and explorations. London: Sage.
Jones, C. P. 2001. Invited commentary: “Race,” racism, and the practice of epidemiology. American Journal of Epidemiology 154:299–304.
Jones, C. P., T. A. LaVeist, and M. Lillie-Blanton. 1991. “Race” in the epidemiologic literature: An examination of the American Journal of Epidemiology, 1921–1990. American Journal of Epidemiology 134:1079–1084.
Karlsen, S. 2004. “Black like Beckham”? Moving beyond definitions of ethnicity based on skin colour and ancestry. Ethnicity and Health 9:107–137.
Karter, A. J. 2003. Commentary: Race, genetics, and disease—in search of a middle ground. International Journal of Epidemiology 32:26–28.
Kaufman, J. S. 2008. Epidemiologic analysis of racial/ethnic disparities. Social Science and Medicine 66:1659–1669.
Kaufman, J. S., and R. S. Cooper. 1999. Seeking causal explanations in social epidemiology. American Journal of Epidemiology 150:113–120.
Kaufman, J. S., and R. S. Cooper. 2001. Commentary: Considerations for use of racial/ethnic classification in etiologic research. American Journal of Epidemiology 154:291–298.
Kaufman, J. S., R. S. Cooper, and D. L. McGee. 1997. Socioeconomic status and health in blacks and whites: The problem of residual confounding and the resiliency of race. Epidemiology 8:621–628.
Krieger, N. 2005. Stormy weather: Race, gene expression, and the science of health disparities. American Journal of Public Health 95:2155–2160.
Krieger, N., and M. Bassett. 1986. The health of black folk: Disease class, and ideology in science. Monthly Review 38:74–85.
Krieger, N., D. L. Rowley, A. A. Herman, B. Avery, and M. T. Phillips. 1993. Racism, sexism, and social class: Implications for studies of health, disease, and well-being. American Journal of Preventive Medicine 9:82–122.
Last, J. M., ed. 1995. A dictionary of epidemiology. 3rd ed. Oxford: Oxford University Press.
Last, J. M., ed. 2001. A dictionary of epidemiology. 4th ed. Oxford: Oxford University Press.
LaVeist, T. A. 1994. Beyond dummy variables and sample selection: What health services researchers ought to know about race as a variable. Health Services Research 29:1–15.
Lewontin, R. C. 1972. An apportionment of human diversity. In Evolutionary biology 6, ed. T. Dobzhansky, M. K. Hecht, and W. C. Steere. New York: Appleton-Century-Crofts, 381–398.
Lillquist, E., and C. A. Sullivan. 2004. The law and genetics of racial profiling in medicine. Harvard Civil Rights–Civil Liberties Law Review 39:391–483.
Liu, E. T. 1998. The uncoupling of race and cancer genetics. Cancer 83:1765–1769.
Liu, X. Q., A. D. Paterson, E. M. John, and J. A. Knight. 2006. The role of self-defined race/ethnicity in population structure control. Annals of Human Genetics 70:496–505.
Mann, A. G., C. L. Trotter, M. A. Balogun, and M. E. Ramsay. 2008. Hepatitis C in ethnic minority populations in England. Journal of Viral Hepatitis 15:421–426.
Matijasevich, A., C. G. Victora, A. J. D. Barros, I. S. Santos, P. L. Marco, E. P. Alernax, and F. C. Barros. 2008. Widening ethnic disparities in infant mortality in southern Brazil: Comparison of 3 birth cohorts. American Journal of Public Health 98:692–698.
McCann-Mortimer, P., M. Augoustinos, and A. LeCouteur. 2004. “Race” and the Human Genome Project: Constructions of scientific legitimacy. Discourse and Society 15:409–432.
McKenney, N. R., and C. E. Bennett. 1994. Issues regarding data on race and ethnicity: The Census Bureau experience. Public Health Reports 109:16–25.
McKenzie, K., and N. S. Crowcroft. 1996. Describing race, ethnicity, and culture in medical research—describing the groups studied is better than trying to find a catch all name. British Medical Journal 312:1054.
Mountain, J. L., and N. Risch. 2004. Assessing genetic contributions to phenotypic differences among “racial” and “ethnic” groups. Nature Genetics 36(11 Suppl. l):48–53.
Muntaner, C., F. J. Nieto, and P. O’Campo. 1996. The Bell Curve: On race, social class, and epidemiologic research. American Journal of Epidemiology 144:531–536.
Nazroo, J. Y. 1998. Genetic, culture or socio-economic vulnerability? Explaining ethnic inequalities in health. Sociology of Health and Illness 20:710–730.
O’Loughlin, J. 1999. Understanding the role of ethnicity in chronic disease: A challenge for the new millennium. Canadian Medical Association Journal 161:161–162.
Osbourne, N. G., and M. D. Feit. 1992. The use of race in medical research. Journal of the American Medical Association 267:275–279.
Outram, S. M., and G. T. H. Ellison. 2006a. The truth will out: Scientific pragmatism and the geneticisation of difference. In The nature of difference: Science, society, and human biology, ed. G. T. H. Ellison and A. H. Goodman. Boca Raton, FL: Taylor and Francis, 157–179.
Outram, S. M., and G. T. H. Ellison. 2006b. Anthropological insights into the use of race/ethnicity to explore genetic determinants of disparities in health. Journal of Biosocial Science 38:83–102.
Outram, S. M., and G. T. H. Ellison. 2006c. Improving the use of race and ethnicity in genetic research: A survey of instructions to authors in genetics journals. Science Editor 29:78–81.
Pfeffer, N. 1998. Theories in health care and research—theories of race, ethnicity and culture. British Medical Journal 14:1381–1384.
Race, Ethnicity, and Genetics Working Group. 2005. The use of racial, ethnic, and ancestral categories in human genetics research. American Journal of Human Genetics 77:519–532.
Risch, N., E. Burchard, E. Ziv, and H. Tang. 2002. Categorization of humans in biomedical research: Genes, race, and disease. Genome Biology 3:1–12.
Rivara, F. P., and L. Finberg. 2001. Use of the terms race and ethnicity. Archives of Pediatrics and Adolescent Medicine 155:119.
Rosenberg, N. A., J. K. Pritchard, J. L. Weber, H. M. Cann, K. K. Kidd, L. A. Zhivotovsky, and M. W. Feldman. 2002. Genetic structure of human populations. Science 298:2381–2385.
Sankar, P., M. K. Cho, and J. Mountain. 2007. Race and ethnicity in genetic research. American Journal of Medical Genetics, Part A 143:961–970.
Shanawani, H., L. Dame, D. A. Schwartz, and R. Cook-Deegan. 2006. Non-reporting and inconsistent reporting of race and ethnicity in articles that claim associations among genotype, outcome, and race or ethnicity. Journal of Medical Ethics 32:724–728.
Shriver, M. D., M. W. Smith, L. Jin, A. Marcini, J. M. Akey, R. Deka, and R. E. Ferrell. 1997. Ethnic-affiliation estimation by use of population-specific DNA markers. American Journal of Human Genetics 60:957–964.
Silverman, D. 2000. Doing qualitative research: A practical handbook. London: Sage.
Silverman, D. 2001. Interpreting qualitative data. London: Sage.
Smaje, C. 1996. The ethnic patterning of health: New directions for theory and research. Sociology of Health and Illness 18:139–171.
Smart, A., R. Tutton, R. Ashcroft, P. Martin, and G. T. H. Ellison. 2006. Can science alone improve the measurement and communication of race and ethnicity in genetic research? Exploring the strategies proposed by Nature Genetics. Biosocieties 1:307–318.
Smart, A., R. Tutton, P. Martin, G. T. H. Ellison, and R. Ashcroft. 2008. The standardization of race and ethnicity in biomedical science editorials and UK biobanks. Social Studies of Science 38:407–423.
Sugarman, J. R. 1996. Improving American Indian cancer data in the Washington State cancer registry using linkages with the Indian health service and tribal records. Cancer 78:1564–1568.
Tang, H., T. Quertermous, B. Rodriguez, S. L. R. Kardia, X. Zhu, A. Brown, J. S. Pankow, et al. 2005. Genetic structure, self-identified race/ethnicity, and confounding in case-control association studies. American Journal of Human Genetics 76:268–275.
Tutton, R., A. Smart, R. Ashcroft, P. Martin, and G. T. H. Ellison. 2010. From self-identity to genotype: The past, present and future of ethnic categories in post-genomic science. In What’s the use of race? Modern governance and the Biology of Difference, ed. Ian Whitmarsh and David S. Jones. Cambridge, MA: MIT Press.
Walsh, C., and L. F. Ross. 2003. Whether and why pediatric researchers report race and ethnicity. Archives of Pediatrics & and Adolescent Medicine 157:671–675.
Weiss, K. M., and S. M. Fullerton. 2005. Racing around, getting nowhere. Evolutionary Anthropology 14:165–169.
Weed, D. L. 2001 Methods in epidemiology and public health: Does practice match theory? Journal of Epidemiology and Community Health 55: 104–110.
Wiencke, J. K. 2004. Impact of race/ethnicity on molecular pathways in human cancer. Nature Reviews Cancer 4:79–84.
Williams, D. R. 1994. The concept of race in health services research: 1966–1990. Health Services Research 29:1392–1393.
Williams, D. R. 1996. Race/ethnicity and socioeconomic status. International Journal of Health Services 26:483–505.
Williams, D. R. 1997. Race and health: Basic questions, emerging directions. Annals of Epidemiology 7:322–333.
Williams, D. R., R. Lavizzo-Mourey, and R. C. Warren. 1994. The concept of race and health status in America. Public Health Reports 109:26–41.
Winkler, M. A. 2004. Measuring race and ethnicity: Why and how? Journal of the American Medical Association 292 (13): 1612–1614.
Witzig, R. 1996. The medicalization of race: Scientific legitimization of a flawed social construct. Annals of Internal Medicine 125:675–679.
Woodward, K., ed. 2004. Questioning identity: Gender, class, ethnicity. 2nd ed. London: Routledge.