Persisting problems related to race and ethnicity in public health and epidemiology research

A recent and comprehensive review of the use of race and ethnicity in research that address health disparities in epidemiology and public health is provided. First it is described the theoretical basis upon which race and ethnicity differ drawing from previous work in anthropology, social science and public health. Second, it is presented a review of 280 articles published in high impacts factor journals in regards to public health and epidemiology from 2009-2011. An analytical grid enabled the examination of conceptual, theoretical and methodological questions related to the use of both concepts. The majority of articles reviewed were grounded in a theoretical framework and provided interpretations from various models. However, key problems identifi ed include a) a failure from researchers to differentiate between the concepts of race and ethnicity; b) an inappropriate use of racial categories to ascribe ethnicity; c) a lack of transparency in the methods used to assess both concepts; and d) failure to address limits associated with the construction of racial or ethnic taxonomies and their use. In conclusion, future studies examining health disparities should clearly establish the distinction between race and ethnicity, develop theoretically driven research and address specifi c questions about the relationships between race, ethnicity and health. One argue that one way to think about ethnicity, race and health is to dichotomize research into two sets of questions about the relationship between human diversity and health. DESCRIPTORS: Ethnicity and Health. Race or Ethnic Group Distribution. Health Inequalities. Social Inequity. Review. 105 Rev Saúde Pública 2013;47(1):104-15 A universal epidemiological fact is that health and disease are unevenly distributed across groups of populations, regardless of the way these are characterized either on the basis of race, ethnicity, nationality, religious affi liation, socioeconomic level, gender, and so on. Racial and ethnic disparities are reported in the United States for virtually all physical and mental health conditions, health-related risk factors, as well as in the use and quality of health care services.19 The real public health challenge is to explain how and why these disparities exist in order to properly inform policies and the development of programs aimed at reducing them.31,50 The methods of classifying groups of populations have a direct impact on the ways health disparities are examined and interpreted. The theory and methods used to study racial and ethnic disparities have been the subject of important criticisms over the last decades in regards to public health and epidemiology.5,6,10,13,14,16,22,27,28,30,35,38 Four comprehensive reviews have examined the use of race and ethnicity concepts in the fi elds of epidemiology and public health research.1,12,25,56 These reviews revealed several key problems including: a) a failure from researchers to differentiate between the concepts of race and ethnicity; b) an inconsistency and lack of transparency in the methods used to assess both RESUMO Realizou-se revisão recente e abrangente da utilização de raça e etnia em pesquisas dedicadas às disparidades de saúde em epidemiologia e saúde pública. Foi descrita a base teórica sobre qual raça e etnia diferem nos métodos de trabalhos em ciência, antropologia social e de saúde pública. A revisão foi feita com base na seleção de artigos publicados em periódicos de alto fator de impacto no que diz respeito à saúde pública e epidemiologia, no período de 2009-2011. O total de artigos selecionados foi de 280. A revisão foi baseada sobre um conjunto de questões conceituais, teóricas e metodológicas relacionadas ao uso de ambos os conceitos. A maioria dos artigos revisados foi fundamentada em um referencial teórico e desde interpretações de vários modelos. No entanto, os principais problemas identifi cados incluem: a) falha de pesquisadores para diferenciar conceitos de raça e etnia; b) utilização indevida de categorias raciais para atribuir etnia; c) falta de transparência nos métodos utilizados para avaliar ambos os conceitos; e d) falta de limites de endereços associada à construção de taxonomias raciais ou étnicas e a sua utilização. Concluiu-se que os futuros estudos que objetivem examinar as disparidades de saúde devem estabelecer claramente a distinção entre raça e etnia, desenvolver pesquisas com orientação teórica que trata de questões específi cas sobre as relações entre raça, etnia e saúde. Argumenta-se que uma maneira de pensar sobre raça, etnia e saúde é dicotomizar a pesquisa em dois conjuntos de questões sobre a relação entre a diversidade humana e da saúde. DESCRITORES: Etnia e Saúde. Distribuição por Raça ou Etnia. Desigualdades em Saúde. Iniquidade Social. Revisão.

A universal epidemiological fact is that health and disease are unevenly distributed across groups of populations, regardless of the way these are characterized either on the basis of race, ethnicity, nationality, religious affi liation, socioeconomic level, gender, and so on.Racial and ethnic disparities are reported in the United States for virtually all physical and mental health conditions, health-related risk factors, as well as in the use and quality of health care services. 19The real public health challenge is to explain how and why these disparities exist in order to properly inform policies and the development of programs aimed at reducing them. 31,50e methods of classifying groups of populations have a direct impact on the ways health disparities are examined and interpreted.The theory and methods used to study racial and ethnic disparities have been the subject of important criticisms over the last decades in regards to public health and epidemiology. 5,6,10,13,14,16,22,27,28,30,35,38ur comprehensive reviews have examined the use of race and ethnicity concepts in the fi elds of epidemiology and public health research. 1,12,25,56These reviews revealed several key problems including: a) a failure from researchers to differentiate between the concepts of race and ethnicity; b) an inconsistency and lack of transparency in the methods used to assess both RESUMO Realizou-se revisão recente e abrangente da utilização de raça e etnia em pesquisas dedicadas às disparidades de saúde em epidemiologia e saúde pública.Foi descrita a base teórica sobre qual raça e etnia diferem nos métodos de trabalhos em ciência, antropologia social e de saúde pública.A revisão foi feita com base na seleção de artigos publicados em periódicos de alto fator de impacto no que diz respeito à saúde pública e epidemiologia, no período de 2009-2011.O total de artigos selecionados foi de 280.A revisão foi baseada sobre um conjunto de questões conceituais, teóricas e metodológicas relacionadas ao uso de ambos os conceitos.A maioria dos artigos revisados foi fundamentada em um referencial teórico e desde interpretações de vários modelos.No entanto, os principais problemas identifi cados incluem: a) falha de pesquisadores para diferenciar conceitos de raça e etnia; b) utilização indevida de categorias raciais para atribuir etnia; c) falta de transparência nos métodos utilizados para avaliar ambos os conceitos; e d) falta de limites de endereços associada à construção de taxonomias raciais ou étnicas e a sua utilização.Concluiu-se que os futuros estudos que objetivem examinar as disparidades de saúde devem estabelecer claramente a distinção entre raça e etnia, desenvolver pesquisas com orientação teórica que trata de questões específi cas sobre as relações entre raça, etnia e saúde.Argumenta-se que uma maneira de pensar sobre raça, etnia e saúde é dicotomizar a pesquisa em dois conjuntos de questões sobre a relação entre a diversidade humana e da saúde.

INTRODUCTION
concepts and c) a dearth in the interpretation of study results based upon race or ethnicity.Reviews conducted in biomedicine 33 and nursing research 15 identified similar key problems which must be addressed.Recent papers addressed methodological issues surrounding the use of statistical analyses and causal inference in the study of racial and ethnic differences 17 as well as problems in the defi nition and psychometrics of acculturation scales, 8,36,51 and racial discrimination scales. 4comprehensive and updated review of the use of race and ethnicity in public health and epidemiology is clearly needed.The most recent review covered manuscripts published over a decade ago. 12This paper fi lls this gap and provides an evaluation as to how and if researchers differentiate between race and ethnicity.These questions are essential since we noted that a failure to distinguish both concepts is found in recent papers that discussed theoretical, conceptual and methodological issues regarding the use of race and ethnicity. 37,38,45 this paper we fi rst draw from anthropology, social science and public health and describe the basis upon which race and ethnicity differ as concepts related to the categorization of human diversity and as social constructions used to understand the nature of social interaction between human populations.Second, a review of the ways race and ethnicity are used to address health disparities in public health and epidemiology research is presented.Using a sample of 280 articles published between 2009-2011 in high impacts factor journals, we identifi ed key problems and concluded with recommendations to guide future studies.

Distinguishing Race and Ethnicity
In recent anthropological 14,22 and public health papers, 16 race and ethnicity are clearly differentiated as concepts that address two distinct facets of human diversity.
From a biological perspective, the concept of human races postulates the existence of discrete and nonoverlapping biological divisions of the human species which can be identifi ed using physical (e.g., skin color), morphological, geographic (e.g., continental location) or genetic markers.Six racial categories are currently defi ned by the US Offi ce of Management and Budget in the United States: 41 White or Caucasian, Black or African American, Asian, American Indian or Alaska Native, Native Hawaiian or Pacifi c Islander, and some other Race.However, on the basis of current knowledge in science and genetics, the Institute of Medicine 24 (2009) recognizes that such racial groups are not mutually exclusive on the basis of biology or genetics.
Gravlee 22 (2009) reviewed the lines of evidence upon which the existence of human races have been rebutted.Recent genetics studies show that while both clines and clusters are part of human genetic variation, clusters explain relatively little of total variation.In fact, the three conventional racial groupings which were originally based on continental origins (Europe, Africa and Asia) differ from one another in about 10 to 15 % of their genes; as such, there is more variation within than between conventional races.Also, most human genetic variation is non-concordant, whereby phenotypes used to distinguish races have no value in predicting other aspects of biology.Moreover, physical variations in skin and hair color amongst the human species occur gradually rather than abruptly over geographic areas. 7Today, some geneticists now suggest that direct assessment of disease-related genetic variation at the individual level may be more accurate and benefi cial in understanding health disparities than the use of racial categories. 26ile its biological signifi cance has been dismissed, public health scholars have suggested the use of race as a social construct to understand the nature of social interaction and how people perceive and relate to race. 16,30As Gravlee 22 (2009) explained; the sociocultural reality of race has biological consequences for racially defi ned groups through two reinforcing mechanisms: social inequalities that may shape the biology of racialized groups and embodied inequalities which perpetuate a racialized view of human biology.
Various models and theoretical considerations have been devoted to explain how psychosocial stress stemming from the experience of discrimination, social exclusion and stereotype may produce health disparities in health care services, access to resources and other health-related factors. 23,27As pointed by Dressler et al 14 (2005), since discrimination is not only based on racial but also ethnic characteristics, the term 'ethnoracial categories' may be used to refer to the social construction of ethnic and racial differences and the way these shape social interactions.
From a sociocultural perspective, the concept of ethnicity refers to the existence of cultural diversity amongst the human species and therefore can be seen as synonymous to humanity. 27Ethnic identity refers to the expression of ethnicity that is particular to each human group; it is constructed through a process of communalization born out of social relations and based on the belief in a common origin and history. 55ecifi cally, it is possible to distinguish three main components of ethnic identity: the cultural, ancestral and the referential. 16The cultural includes shared models of the mundane (language, diet, dress, rituals), and the symbolic (beliefs, world vision etc.); the ancestral refers to a common history, a shared territory and/or kinship, while the referential implies the defi nition of an established frontier between a group and the rest of the world.Various models have been developed to explain how cultural factors specifi c to an ethnic group may affect health, including behaviors, norms, attitude, and beliefs. 16thropologist Fredrik Barth insisted that ethnic studies should focus on the fundamental existence of cultural boundaries that exist between groups of populations 3 (this is similar to the referential dimension described by Dressler et al 14 (2005).In a related approach, Juteau 27 (1999) proposed that the process of ethnic communalization is based on the delineation of an internal frontier, which defi nes the group based on the belief in a common history and shared culture, and an outer boundary that characterizes the balance of power between ethnic groups.One essential point made by Juteau 27 (1999) is that even if ethnic identity is socially constructed, it remains that ethnicity is a reality inseparable from humanity and necessary to the understanding of social dynamics.Like the effect of race, discrimination based on ethnic identity does shape the health and life of individuals and groups, in processes of migration, cultural contact and acculturation experiences. 13ere are various ways and debates regarding specifi c markers that should be used to identify ethnic categories, including; language, religion, nativity, etc. 11 For example Hispanic ethnicity is defi ned on the basis of spoken language. 41The Institute of Medicine 24 defi nes granular ethnicity on the basis of national and sub-national origin (e.g., Korean, Puerto Rican, and German from Russia).Regardless of the markers used to defi ne ethnic group, adopting a cultural perspective of ethnicity implies two important methodological considerations.
First, biophysical traits should not be part of the defi nition of ethnicity since there is no essential correlation between biology and culture. 47To be clear, categories defi ned upon skin color such as "Blacks" and "Whites" are not appropriate to represent the cultural and ancestral dimensions of ethnicity. 39Second, current ethnic categorizations defi ned on the basis of continental markers or languages are heterogeneous overall in terms of culture.Such groupings may have in certain cases, a useful value in identifying general health disparities, 11 however they do not hold a great potential for understanding the root determinants of disease variation attributed to cultural factors since broad categories such as Asian, Hispanic or Western, lump together various groups that substantially differ both socially and culturally. 5,16,39

METHODS OF REVIEW
This study aimed to review the ways in which race and ethnicity are used to address health disparities in public health and epidemiology research.We confi ned our review to articles published in the years 2009, 2010 and 2011 in high impacts factor authoritative peer-reviewed journals in the fi elds of epidemiology and public health.These journals included the American Journal of Public Health (AJPH), the American Journal of Epidemiology (AJE), Social Science and Medicine (SSM), the American Journal of Preventive Medicine (AJPM), the Journal of Epidemiology and Community Health (JECH), the International Journal of Epidemiology (IJE) and the European Journal of Epidemiology (EJE).This selection offers the most updated and comprehensive reviews since previous ones were limited to one or two peer-reviewed journals.
MEDLINE was searched to identify relevant articles using the following keywords: Rac*, Ethni*, White, Black, European, Caucasian, African, Asian, Latin*, Hispanic, Pacifi c, and Indian.These words cover the concepts and category terms used in the standard racial and ethnic classifi cation of the Offi ce of Management and Budget (1997) and in most classifi cations in the United Kingdom, Canada, New Zealand, and Europeans countries.
Out of this search, a total of 508 articles were identifi ed.Abstracts were then systematically reviewed to identify the relevant research articles which focused on health disparities and were available at the university library.Only fi ve out of 508 were not available (less than 1%).Program evaluation research papers were excluded from the main review because they would not allow for the use of the same analytical grid.These papers accounted for less than 3% of all articles (n = 14/508).
Commentaries, theoretical and methodological papers, and reviews were scrutinized and their content was included in the theoretical development and discussion of results.
A total of 280 research articles were retained and examined in their full length to answer the following questions: • What are the study designs used to address racial and ethnic health disparities?
• What are the health outcomes measured and justifications given by authors to examine these outcomes in terms of race and ethnicity?
• What are the concepts used by authors to address group of populations; race, ethnicity, both or neither?Do researchers differentiate between the concepts of race and ethnicity?
• What are the methods used to access race or ethnicity?
• Do researchers discuss and recognize the limits inherent to the methods of racial and ethnic classifi cations?
• What is the nature of the interpretations given in the studies to explain racial and ethnic health differences?
An analytical grid comprised of 22 variables was developed and used to structure the analysis of each article and to address the aforementioned questions.Data was coded by the author to produce a series of descriptive statistics using SPSS 18.0.

RESULTS
Table 1 presents the distribution of articles reviewed by journals, years of publications and research settings.Notably, research was predominantly conducted in the United States (81.4%), with other research undertaken mostly in western countries.In the following sections, results are reported for all articles considered as a single sample.Important differences between journals and country of setting are also reported when pertinent.Authors provided different types of justifi cations to support their study of racial or ethnic differences in health.First and foremost, almost every author (95.0% of all articles) provided statistics on disease prevalence, service use, or mortality rates and identifi ed at-risk groups which they then examined in their study.Additionally, some authors justifi ed the use of granular ethnicity to examine subpopulations (e.g., Korean Chinese) by explaining the limits of racial or ethnic aggregation into a larger group (e.g., Asian American).However, only 63.6% of authors went beyond the reporting of mere statistics and the identifi cation of at-risk groups to actually provide a theoretical basis for the study of racial and ethnic health disparities.Seven articles presented no justifi cations at all in relation to why racial or ethnic differences in health were being investigated.

Concepts
We next examined which concepts (race, ethnicity, both or neither) were used by authors to defi ne groups of populations, and whether or not they differentiated race and ethnicity.As shown in Table 3, most authors used both race and ethnicity in their study (170 or 60.7% of articles).Ethnicity was preferred in 61 or 21.8% of articles along with others terms such as ancestry, culture and pan-ethnic.Race was the sole concept used in 28 or 10.0% of manuscripts, sometimes referring to 'people of color'.A remaining 21 or 7.5% of authors avoided referring to either race or ethnicity, and instead opted to refer directly to group labels (e.g., African Americans).address the heterogeneity of large groups, or to analyze the ways multiple dimensions of ethnic and racial affi liations interact.Some concepts related to acculturation or immigration measures and included nativity, immigration status, years of residence in the host country, and language spoken or language profi ciency.Other concepts such as ethnic density, racial segregation and ethnic enclave were used to defi ne a measure of race or ethnicity at the neighborhood level.Finally, racism or discrimination scales were used in several studies.Interestingly, the use of such fi ne grain concepts was mainly seen in SSM (58.1% of the journal articles), whereas in other journals the use of such concepts were found in about 25-30% of articles.Two authors developed unique concepts using a relational approach to address social relations in function of race; these are the concept of colourism, expressed racial identity and refl ected racial identity 53 and racial centrality. 9erall, 64.3% of authors failed to differentiate race and ethnicity.Such a failure was evidenced by the presence of at least one of the following criteria: a) the systematic use of the expression "race/ethnicity" or "racial/ethnic" when referring to either racial or ethnic differences or both; b) the use of the terms race and ethnicity interchangeably; c) the use of skin color based categories to ascribe ethnicity without justifi cations (e.g., White and Blacks taken as ethnic groups) or; d) the comparison of ethnic categories with racial ones without justifi cations (e.g., Mexicans compared to Blacks in a study examining ethnicity).Examples of sentences illustrating the confusion between race and ethnicity include: "The classifi cation of ethnicity was performed by the healthcare professional on the basis of race and country of birth" (p.697), 43 "Hispanics have become the largest racial/ethnic minority group in the United States" (p.145), 49 and most evidently "Ethnicity was measured by asking respondents, What race do you consider yourself to be?" (p.563) 57

Taxonomies
As presented in Table 4, seven different types of taxonomies were found in the reviewed articles to classify group of populations on the basis of race and ethnicity.
In the following paragraphs we described how these taxonomies were used by researchers in function of the concepts they chose (race, ethnicity, both or neither).
First, the majority of authors who used both race and ethnicity (n = 119/170) used the Offi ce of Management and Budget (1997) classifi cation system (Taxonomy 1).Categories crossing race and ethnicity were often created by these researchers (e.g., Non-Hispanic Blacks).A second group of authors (n = 13/170) defined racial and ethnic groups on the basis of country of residence or country of birth (Taxonomy 2).A third group of researchers (n = 34/170) created and compared groups by combining the Office of Management and Budget (1997) classifi cation with country or region of birth (Taxonomy 3).For example, authors sub-divided the racial Black group by region of birth place to examine pre-migration exposure to racism and discrimination. 32A small group of authors (n = 4/170) defi ned specifi c ethnic groups on the basis of various cultural markers including language and religion, etc. (Taxonomy 4).Overall, we found that 93.5% (n = 159/170) of all authors that used both race and ethnicity failed to differentiate both terms.
Studies in which ethnicity was solely used (n = 61) were mostly done outside the US where the term "race" is a A failure to differentiate race and ethnicity was evidenced by the presence of at least one of the following criteria: a) the systematic use of the expression race/ ethnicity' or 'racial/ethnic' when referring to either racial or ethnic differences, or both b) the use of the terms 'race' and 'ethnicity' interchangeably, c) the use of skin color based categories to ascribe ethnicity without justifi cations (e.g., Black taken as an ethnic group), or d) the comparison of ethnic categories with racial ones without justifi cations (ex: Mexicans compared to Blacks in a study examining ethnicity) b The defi nition of race or ethnicity as a social construct was evidenced by the presence of any comments regarding the social and/or cultural dimension of race or ethnicity in the entire revised paper avoided.However, a great variety of taxonomic systems is found.The fi rst group of authors (n = 9/61) used the Offi ce of Management and Budget (1997) classifi cation system to defi ne ethnic groups (Taxonomy 1).Most researchers however (n = 18/61) defi ned ethnic groups on the basis of countries or regions of provenance (Taxonomy 2).This is exemplified by Netherland studies where ethnicity is based on country of birth of the father and mother.A third group of authors (n = 7/61) combined the Offi ce of Management and Budget system with countries (Taxonomy 3) while a fourth group (n = 6/61) defi ned ethnic groups on the basis of specifi c cultural markers (Taxonomy 4).The fi fth group of authors include all studies conducted in the UK (n = 16/61), where British scholars avoid the use of "race" and preferred ethnicity but nonetheless used a system that combined skin color with geographical/country location (Taxonomy 5).Lastly, in studies conducted in New Zealand (n = 5/61), ethnic classifi cation was based on a prioritization system based on the Maori people (see Taxonomy 6).Overall, amongst all authors who used only ethnicity, a third (34.4% or n = 21/61) failed to distinguish it with race.For example, these authors used categories such as White, White British and Blacks as ethnic groups.
Authors using only the concept of race in their studies all used the Offi ce of Management and Budget (1997) classifi cation system (Taxonomy 1) (n = 27/28), except one who used Taxonomy 3. None of the authors who used solely race were found to have mixed race and ethnicity; however, only 1 out of 28 authors acknowledge the use of race as a social construct.
Finally, studies who used neither race nor ethnicity used the Offi ce of Management and Budget system (Taxonomy 1) (n = 12/21) or combined it with countries (Taxonomy 3) (n = 1/21).Other authors (n = 8/21) relied on countries or regions of provenance (Taxonomy 2).No clear evidence was found that authors who avoided the use of race and ethnicity failed to differentiate both concepts.

Transparency in methods and limits of classifi cation
Overall, only 57.9% of studies provided details regarding the manner in which study subjects were assigned to the taxonomy categories.Such methods when given included self-reporting race or ethnicity, parent reported race or ethnicity, mother's race and origin listed on the infant's birth certifi cate, and identifi cation in medical records.Outstanding positive cases  were found.For example; one study contained an entire paragraph devoted to the description and methods of ethnicity assignment. 18At the other end, one study provided no details on the defi nition or methods used to identify a specifi c and rarely studied ethnic group; the Roma people of Europe. 54 found that only 19.6% of authors provided any discussion regarding the limits of using an ethnic or racial categorization.Of those mentioned, these limits were related to issues of group aggregation (e.g., Asian), generalization outside a given territory or state, generalization related to generational status, racial or ethnic misclassifi cation by individuals or health-care providers, arbitrary measures of novel concepts (e.g.: ethnic enclave), and missing data on race or ethnicity.The homogeneity of racial or ethnic groups was simply assumed in 86.9% of articles reviewed.
Finally, very few researchers (6.4%) addressed the issues surrounding the biological signifi cance of race or the social construction of the concept of race and ethnicity.More importantly, we found that only 10.3% of researchers who specifi cally aimed to describe and analyze psychosocial health related outcomes related to race or ethnicity addressed the issues surrounding the social meaning of these concepts.

Interpretation of ethnic or racial disparities
Lastly, we examined the nature of the interpretations given by authors to explain health disparities according to race or ethnicity (Table 5).First, we found that 69.3% of authors provided interpretations which were based on the actual analyzed data.Amongst these (n = 194), a variety of interpretations were used with the most popular being variations of socioeconomic (16.5%) and psychosocial (16.5%) models, followed by acculturation or immigration models (10.8%).Other models are shown in Table 5.It is noteworthy to mention that 26.8% of authors provided more than one interpretation model based on several factors and covariates being measured in their study.
We found however that 40 or 14.3% of studies provided interpretations that were not based on actual analyzed data either because the interpretations provided belonged to a domain (e.g., socioeconomic) which was not included in the study design or because interpretation model (e.g, racism) was addressed by a proxy (e.g., race) and not by a more specifi c measure (e.g., perceived racism).Lastly, in 46 or 16.4% of all studies, authors provided no interpretations at all of ethnic or racial health disparities observed because the study was merely descriptive.

DISCUSSION
This paper presented the most recent and comprehensive review of the use of race and ethnicity to address health disparities in public health and epidemiology research.The strengths of this review include the cover of multiple high impact factors journals several of which had never been previously reviewed.Our review also included a wide range of questions and topics which had not been addressed in past reviews.
a If researchers interpreted health disparities using a specifi c theoretical model b The interpretation provided belonged to a model (e.g.socioeconomic) which was not included in the study design or the interpretation model (e.g, racism) was addressed by a proxy (e.g., race) and not by a more specifi c measure (e.g., perceived racism); 3 Interpretation models were classifi ed by types using open and focused coding procedure.These types are: socioeconomic (e.g., based on poverty, education), psychosocial (e.g., based on the effects of racism or discrimination), acculturation/ immigration (e.g., health disparities are attributed to changes in the process of migration and/or cultural change), health care services (e.g., disparities are due to language barriers in the use of health care services), norms, attitudes and beliefs (e.g., health disparities are caused by differences in health beliefs, norms or attitudes), behavioral (e.g., based on behavioral differences such as drug or tobacco use) biological/genetic (e.g., health disparities are attributed to different biological or genetic constitution specifi c to racial or ethnic groups), ethnic density (e.g., neighborhood effects), physical environment (e.g., health differences are attributed to differential exposure of groups to the physical environment), political/ social organization (health differences are attributed to different social or political organization particular to ethnic or racial groups) Several fi ndings reported may help orient future research that addresses racial and ethnic health disparities.We identifi ed key problems including: a) a failure from researchers to differentiate between the concepts of race and ethnicity where both concepts are often used interchangeably or merged into a single entity termed "race/ ethnicity"; b) an inappropriate use of racial categories to ascribe ethnicity; c) a lack of transparency in the methods used to assess both concepts; d) failure to address limits associated with the classifi cation and use of racial or ethnic taxonomies and; e) failure to recognize the social meaning of race in discrimination and racism studies.
The confusion between race and ethnicity, as well as the inconsistency and lack of transparency in the methods used to assess race or ethnicity, are clearly not a new problem in the fi eld. 12Plainly, this problem continues to persist even ten years later and must be addressed once and for all.More importantly, this confusion was noted in theoretical and methodological papers.For example; in an interesting discussion of how the theory of Durable Inequality may apply to ethnic health studies, authors avoided the term race and preferred ethnicity, yet they considered Whites and Blacks as ethnic groups. 37dditionally, in a paper addressing the importance of multilevel modeling, both race and ethnicity were used rather interchangeably and inconsistently. 45 is important to acknowledge the positive fi ndings we observed as well.For instance, several authors employed variables in addition to race and ethnicity which are more specifi c to the research questions and which are clearly related to underlying factors of health disparities.These include acculturation measures, immigration status, as well as concepts grounded in a relational approach such as racial centrality, colourism, and racial expressed identity.
Another positive fi nding was that the majority of articles reviewed were grounded in a theoretical framework and provided interpretations from various models and were based on analyzed data.This fi nding is important to consider since problems in the interpretation of results were reported by Comstock et al 12 (2004), where they found that only "30.4 percent of authors discussed their fi ndings in terms or race and ethnicity" (p.617).Clearly, the situation has improved over the years however the need to address it remains.

Questions About the Relationship Between Human Diversity and Health
To address the key problems noted in this review we must start with the recognition that race and ethnicity are conceptually different.Only then can the proper theory and methods be appropriately chosen to study either, or the interactions of both in the production of health problems.We believe the best ways to think about ethnicity, race and health is to dichotomize research into two sets of questions about the relationship between human diversity and health. 40However, before describing these sets of questions, we must be clear as to what kind of diversity are we speaking about.Two points need to be addressed.
First, if researchers aim to study health disparities upon the rationale of biological diversity, that is where health differences are attributed to biological or genetic constitutions, they should clearly state it and address the question of biological plausibility.Relevant markers must be used to identify groups of populations that differ on a biological basis, including genetic markers.Today, rather than using phenotype or continental location, human biological variation can be assessed using genetic markers, 44 along with other forms of genealogical and historical knowledge. 46cond, social diversity, in terms of education, class, employment and income, is a separate rationale upon which health disparities can be studied.Some scholars recommend that researchers should separate the effects of socioeconomics factors from those of race and ethnicity. 35his is the case in studies we excluded from this review because they did not address nor frame their research questions in terms of ethnicity, but rather in socioeconomic and political terms (e.g., studies comparing different political regimes across countries). 20,34Indeed cross-country comparisons address social, economic and political disparities, not ethnic disparities.
The reminder of this section focuses on cultural diversity and introduces two sets of questions.
The fi rst set of questions is how do cultural factors particular to an ethnic group such as religious beliefs, dietary traditions, behaviors, beliefs and attitudes affect health?These questions correspond to what Juteau 27 (1999) named the inner frontier of ethnic identity, i.e., the characteristics that are share by an ethnic group.It also corresponds to what Ford & Harawa 16 (2010) called "the attributional" dimension of ethnicity.One example of a model embedded in this fi rst set of questions is the health behavior model where health differences are attributed to behavioral factors particular to ethnic groups. 16An understudied research area that fi ts into this fi rst set of question are assets-based approaches that identify strengths particular to ethnic groups and that confer them specifi c health advantages. 16e second set of questions is concerned with the ways human groups interact together in the social arena, on the basis of both race and ethnicity.These questions are grounded in a relational approach, and correspond to the outer frontiers of ethnic groups, or as Ford & Harawa 16 (2010) explain; the relational dimension of ethnicity.
One model embedded such a relational approach is the cultural adaptation model, in which health disparities are the product of changes and challenges experienced through acculturation and migration processes. 23Another model is the psychosocial stress model in which health disparities are associated with stress related problems stemming from the experience of discrimination. 14eighborhood effects of ethnic segregation, racial and ethnic discrimination, language or beliefs barriers in the use of health services, all fall into this second relational dimension of human diversity.
A number of methodological challenges must be addressed when using these two sets of questions about the relationship between human diversity and health.Ethnicity and race are concepts that are complex, subjective, and vary according to the demographic and migration patterns, and sociopolitical history of each country. 14Clearly, researchers are faced with the problem of using groups that are not too heterogeneous and suffi ciently large in sample size.However, the use of heterogeneous ethnic or racial categories (such as Hispanics or Blacks) will only generate imprecise data.The only universal recommendation would be to choose the methods to identify ethnic or racial groups in direct relation to the research question and setting.
However, several points are clear: a) ethnicity must be defi ned by cultural markers and not biological ones; b) continental categories such as Africans or Asians are too culturally heterogeneous to be used by themselves; c) countries are socio-political entities and not necessarily culturally ones; d) while language is a cultural marker, categories such as Hispanic are culturally heterogeneous; e) Whites and Blacks are socially constructed racial categories, not ethnic groups, and f) discrimination is established on the basis of both ethnicity and race.
Above all, authors must be transparent in their methods to ascribe race or ethnicity and acknowledge the limits of the classifi cation they choose to use.In cases where large aggregated groups are used, researchers should always justify their aggregation in terms of the research questions and the variables of interest. 48The extremely low number of researchers that described their methods of group assignment suggests that these authors take race and ethnicity categories for granted and are not aware of or concerned by the limits and issues related to the use of these categories.
There are also promising developments such as the use of novels methods to defi ne race and ethnicity employed by researchers using a constructivist and/or relational approach. 9,16,21,53Also, instead of addressing ethnic identity directly, some researchers are dissecting its components (e.g., religion, language, norms, beliefs, etc.) and using fi ne grain methods to analyze their relation to health and disease.

Limits
Our sample included research predominantly done in the United States.Different methods and perspectives to investigate the question of human diversity may apply to research done elsewhere in the world.Our review suggests however that our results are also true for research done in all western countries.Second, the review has been conducted by a single rater.A list of all reviewed articles is provided in a supplemental fi le and is open to criticisms on aspects the author might have been mistaken on.Finally, this review did not cover journals in specialized fi elds that are of public health and epidemiological health relevance (e.g., in psychiatry).We are confi dent however that by choosing high impact journals we are presenting the current trends in the use of race and ethnicity in the most state-of-the art research.Reviews conducted in biomedical research 33 and nursing science 15 suggests that our results refl ect the wider phenomenon of health research.

CONCLUSIONS
Ethnicity has overtaken race in medical science over the course of the second half of the 20 th century. 2 However this shift is useless unless it is accompanied by a theoretical understanding of what race and ethnicity are as concepts related to human diversity.Similarly, experienced researchers from a Latin/North American workshop have called for more theoretically driven and specifi c oriented research to address the main priority of eliminating health disparities. 31To undertake such an important quest, researchers must cease to systematically mix race with ethnicity, and understand the theoretical basis upon which each of the concepts affect the health of individuals and populations.Ultimately, the construction and operationalization of race and ethnicity not only determine the quality of research but also affect the way heath disparities are portrayed by the media, perceived by the public and tackled by politics and prevention practices. 29,42,52 c (non-Maori), Asians (non-Maori and non-Pacifi c), and Europeans (non-Maori, non-Pacifi c and non-Asian (nMnPnA) 1.8

Table 2
presents data on the study design, outcomes measured, and types of justifications provided by authors to examine these outcomes in terms of race and ethnicity.Most articles were cross-sectional and used quantitative data (67.9%).A limited number of studies analyzed qualitative data or utilized a mixed methods design.The majority of the examined studies (77.9%) were comparative where health disparities were measured between two or more racial or ethnic groups, whereas the remaining articles were limited to a single group.

Table 3
also shows the use of fi ne grain concepts in addition to race and ethnicity noted in 102 or 36.4% of articles.These concepts were used by researchers to

Table 1 .
Characteristics of articles addressing racial and ethnic health disparities in public health and epidemiology, 2009 to 2011.(n = 280)

Table 2 .
Study design, outcomes and justifi cations provided in the study of racial and ethnic disparities in public health and epidemiology research, 2009 to 2011.(n = 280) a Outcome types are: biological (e.g., morbidity, mortality), behavioral (e.g., drug use, physical activity), healthservices (e.g., health care use), psychosocial (e.g., racism), socioeconomic (e.g., education, income), physical environment (e.g., pollution exposure) b Types of justifi cations provided by authors to examine health outcomes in terms of race and ethnicity

Table 3 .
Concepts used to study racial and ethnic health disparities in public health and epidemiology research, 2009 to 2011.(n = 280)

Table 4 .
Six taxonomies used in the study of ethnic and racial health disparities in public health and epidemiology, 2009 to 2011.

Table 5 .
Interpretations models used to explain health disparities in public health and epidemiology research, 2009 to 2011.(n = 280)