EDITORIAL COMMENT

What We’re Talking AboutWhen We Talk About Race*Edward P. Havranek, MD, FACC,Frederick A. Masoudi, MD, MSPH, FACCDenver, Colorado

It has been pointed out repeatedly (1,2) that reportedassociations between race and illness should be subject tocareful scrutiny. Establishing race as an independent riskfactor for a disease is even more complicated than theusually difficult task of establishing epidemiologic associa-tion. Among the usual difficulties are selection of a repre-sentative sample, collection of accurate data, identificationof appropriate controls, and adjustment for confounders.Unlike other variables in epidemiologic studies, however,race has complex social and biologic implications.

See page 429

Viewing race as a social construct is clearly important inepidemiology. Race is linked to a wide range of socioeco-nomic and cultural factors that impart important modifyingeffects on health, disease, and the delivery of health care.These factors include income, education, diet, health be-liefs, access to health insurance, and exposure to discrimi-nation (3). African American adults are approximately 2.5times as likely as white Americans to have incomes belowthe poverty level, and African American children are ap-proximately four times as likely to live in families withincomes below the poverty level. According to the 2000U.S. Census, 78.5% of African Americans and 88.4% ofwhite Americans over the age of 25 are high schoolgraduates. Differences in college graduation rates are evengreater: 16.5% for African Americans and 28.1% for whiteAmericans. The 2000 Center for Disease Control NationalHealth Interview Survey found that although 12.5% ofwhite Americans lacked health insurance, the comparablefigure for African Americans was 20%. Apparent racialdifferences may be markedly attenuated after accounting forconfounders such as these. For instance, nationwide studiesof patterns of care for heart failure (HF) have found lowerrates of angiotensin-converting enzyme inhibitor prescrip-tion for African American patients (4,5). In contrast, arecent study of Medicare patients (6), who presumably havelower barriers to gaining access to care, found no racialdifferences in rates of guideline-based therapy. Race should

not be used as a proxy for socioeconomic or cultural factors.When differences in risk factors for a disease are the resultof measurable socioeconomic and cultural factors, thiscausative link should be made explicit.

Race has been viewed in medicine as a biologic constructas well. Race classifies humans on the basis of facial features,hair type, and skin color. The boundaries of these classifi-cations are never explicitly stated and vary across cultures.Thus, there is no standardized method in the medicalliterature of assigning race. Race may be assigned by theinvestigator based on the subject’s appearance, or it may beself-assigned by the subject. Self-assignment may be fromfixed sets of categories that vary among studies or from anopen-ended question. Although racial designations areloosely associated with the continent of origin of an indi-vidual’s ancestors, much of the medical literature implicitlyattaches a greater genetic significance to race than theavailable data warrant. Estimates of the genetic variabilityattributable to individual variation within populations rangefrom 85% to 95% (7,8). Of the remaining genetic variability,only about half is attributable to between-continent differ-ences. Thus, two individuals from different racial groupsmay very well share more genes than two individuals fromthe same racial group (9). It has been estimated thatapproximately 30% of the variable genetic make-up ofAfrican Americans is the result of admixture with people ofnon-African origin (10). Seen in this light, race may haverelatively little value as a biologic construct. Exceptions mayexist, however. Adrenergic receptor (11) and G protein (12)polymorphisms that may predispose one to HF are moreprevalent in African Americans compared with whites.Although further work is needed to confirm the validity ofthese findings, they may help explain why, for instance,African Americans with hypertension appear to be morelikely to develop left ventricular hypertrophy (13). The factremains, however, that race is an imprecise categorizationscheme and is a poor proxy for genomic differences.

In this issue of the Journal, Ruo et al. (14) presentevidence that HF patients identified in the medical record asAfrican American are approximately half as likely as thoseidentified as white to have atrial fibrillation (AF). Among1,373 randomly chosen patients hospitalized with HFduring a one-year period, 506 had AF. The prevalence ofAF was 19.7% in the African American patients and 38.3%in the white patients. Although relative risks are notreported, the odds ratio of 0.51 for racial prevalence in theirfinal multivariable model suggests that adjustment for age,gender, known risk factors for AF, comorbidity, medica-tions, and HF severity does not substantially alter thisrelationship. The magnitude of the computed relative risk isstriking. How well has the current study avoided the pitfallsof inadequate adjustment for confounding by socioeconomicand cultural factors and of unwarranted assumption ofgenetic explanations?

With respect to recognizing confounding by socioeco-

*Editorials published in the Journal of the American College of Cardiology reflect theviews of the authors and do not necessarily represent the views of JACC or theAmerican College of Cardiology.

From the Denver Health Medical Center and the University of Colorado HealthSciences Center, Denver, Colorado. Dr. Masoudi is supported by NIH ResearchCareer Award K08 AG-01011.

Journal of the American College of Cardiology Vol. 43, No. 3, 2004© 2004 by the American College of Cardiology Foundation ISSN 0735-1097/04/$30.00Published by Elsevier Inc. doi:10.1016/j.jacc.2003.11.011

nomic and cultural factors, they have studied a sampledrawn from a Kaiser Permanente managed care populationthat has been demonstrated to be representative of the statepopulation. This makes it less likely that black–whitedifferences are the result of socioeconomic factors, but it isunclear whether the African American patients in the studyare representative of African American patients in general.Second, although the population studied is relatively ho-mogenous from a socioeconomic standpoint, unmeasuredsocioeconomic confounders are likely to have been present.Ascertainment of detailed socioeconomic data from themedical record is difficult, and more explicit adjustmentswould not have been possible with the information availableto the investigators.

With respect to possible genetic explanations, the currentstudy is intriguing. First, the direction of the associationbetween race and AF runs counter to that which one mightexpect if the relationship was based on socioeconomic andcultural factors. The poorer socioeconomic status of AfricanAmericans typically is associated with more severe impair-ment of health status and poorer outcomes, yet the lowerincidence of AF among African Americans is a marker ofless severe HF. Second, the authors have studied a condi-tion, AF, for which genetic predisposition exists. Familialforms of AF have been described (15), and genetic loci forthe arrhythmia have been mapped to chromosome 10 (16),chromosome 6 (17), and perhaps to others (18). Althoughthe existence of a heritable form of AF may suggest thepossibility that the findings of the present study reflectdistinct genetic differences, the reality is much less clear.The genetic forms identified thus far account for a smallminority of cases of AF, and none of the existing AFchromosomal links have been specifically associated withrace. Although current understanding of the contribution ofgenotype to the development of AF is incomplete, it wouldbe premature to conclude that racial differences in AFincidence reflect genetic differences rather than a findingconfounded by socioeconomic or cultural factors.

If confirmed, the findings of Ruo et al. (14) may be moreimportant for generating future mechanistic studies than forelucidating epidemiology. Such future studies would collectdata that more explicitly account for the socioeconomicstatus of the patients involved. If predisposition to AF orprotection from AF can be mapped to a specific gene orgenes, genotyping performed as part of these studies shouldaid in the identification of relevant polymorphisms. Thecurrent study, although limited in its ability to adjust forsocioeconomic and cultural confounding and to assess forgenetic polymorphisms directly, credibly neither neglectsthe social nature of race nor uncritically assumes a biologic

basis for race. As our society struggles to eliminate race-associated disparities in care and eagerly awaits the advent ofgenomic medicine, it is absolutely critical that we all knowwhat we’re talking about when we talk about race.

Reprint requests and correspondence: Dr. Edward P. Havranek,Denver Health Medical Center #0960, 777 Bannock Street, Denver,Colorado 80204-4507. E-mail: ehavrane@dhha.org.

REFERENCES

1. Cooper R, Kaufman J, Ward R. Race and genomics. N Engl J Med2003;348:1166–70.

2. Kaplan J, Bennett T. Use of race and ethnicity in biomedical literature.JAMA 2003;289:2709–16.

3. Institute of Medicine. Unequal Treatment: Confronting Racial andEthnic Disparities in Health Care. Washington, DC: Institute ofMedicine, 2002.

4. Ayanian JZ, Weissman JS, Chasan-Tauber, Epstein AM. Quality ofcare by race and gender for congestive heart failure and pneumonia.Med Care 1999;37:1260–9.

5. Stafford RS, Sagalm D, Blumenthal D. National patterns ofangiotensin-converting enzyme inhibitor use in congestive heart fail-ure. Arch Intern Med 1997;157:2460–4.

6. Rathore SS, Foody JM, Wang Y, et al. Race, quality of care, andoutcomes of elderly patients hospitalized with heart failure. JAMA2003;289:2517–24.

7. Rosenberg N, Pritchard J, Weber J, et al. Genetic structure of humanpopulations. Science 2002;298:2381–5.

8. Barbujani G, Magagni A, Minch E, Cavalli-Sforza LL. An appor-tionment of human DNA diversity. Proc Natl Acad Sci USA 1997;94:4516–9.

9. American Anthropological Association. American AnthropologicalAssociation statement on “race.” Available at: http://www.aaanet.org/stmts/racepp.htm. Accessed August 27, 2003.

10. Graves JL. The Emperor’s New Clothes: Biological Theories of Raceat the Millennium. New Brunswick, NJ: Rutgers University Press,2001:201–3.

11. Small KM, Wagoner LE, Levin AM, Kardia SLR, Liggett SB.Synergistic polymorphisms of �1- and �2C-adrenergic receptors andthe risk of congestive heart failure. N Engl J Med 2002;347:1135–42.

12. Dong Y, Zhu H, Sagnella GA, et al. Association between the C825Tpolymorphism of the G protein beta-3 subunit gene and hypertensionin blacks. Hypertension 1999;34:1193–6.

13. Gardin JM, Wagenknecht LE, Anton-Culver H, et al. Relationship ofcardiovascular risk factors to echocardiographic left ventricular mass inhealth young black and white adult men and women. The CARDIAStudy. Circulation 1995;92:380–7.

14. Ruo B, Capra AM, Jensvold NG, Go AS. Racial variation in theprevalence of atrial fibrillation among patients with heart failure: theEpidemiology, Practice, Outcomes and Costs of Heart failure (EP-OCH) study. J Am Coll Cardiol 2004;43:429–35.

15. Mestroni L. Genomic medicine and atrial fibrillation. J Am CollCardiol 2003;41:2193–6.

16. Brugada R, Tapscott T, Czernuszewicz GZ, et al. Identification of agenetic locus for familial atrial fibrillation. N Engl J Med 1997;336:905–11.

17. Ellinor P, Shin J, Moore R, Yoerger D, MacRae C. Locus for atrialfibrillation maps to chromosome 6q14-16. Circulation 2003;107:2880–3.

18. Darbar D, Herron K, Ballew J, et al. Familial atrial fibrillation is agenetically heterogeneous disorder. J Am Coll Cardiol 2003;41:2185–92.

437JACC Vol. 43, No. 3, 2004 Havranek and MasoudiFebruary 4, 2004:436–7 Editorial Comment