American Journal of Epidemiology
Copyright © 2004 by the Johns Hopkins Bloomberg School of Public Health
All rights reserved
Vol. 159, No. 6
Printed in U.S.A.
DOI: 10.1093/aje/kwh084
SPECIAL ARTICLE
Four-Year Review of the Use of Race and Ethnicity in Epidemiologic and Public
Health Research
R. Dawn Comstock1,2, Edward M. Castillo3, and Suzanne P. Lindsay4
1
Epidemic Intelligence Service, Epidemiology Program Office, Centers for Disease Control and Prevention, Atlanta, GA.
Injury Prevention Service, Oklahoma State Department of Health, Oklahoma City, OK.
3 Division of Emergency Medical Services, Health and Human Services Agency, County of San Diego, San Diego, CA.
4 Institute for Public Health, Graduate School of Public Health, San Diego State University, San Diego, CA.
2
Received for publication April 24, 2003; accepted for publication October 24, 2003.
To determine how current researchers address the use of race and ethnicity as variables in epidemiologic and
public health studies, the authors conducted a comprehensive review of 1,198 articles published in the American
Journal of Epidemiology and the American Journal of Public Health from 1996 to 1999. Seventy-seven percent
(n = 919) of the articles referred to race or ethnicity. The number of variable categories ranged from 0 to 24, with
an average of 3.14 per article. An enormous diversity of terms was used to describe the concepts of race and
ethnicity as variables as well as to describe the categories used to assess these variables. Researchers
frequently failed to differentiate between the concepts of race and ethnicity, to state the context in which these
variables were used, to state the study methods used to assess these variables, and to discuss significant study
results based on race or ethnicity. Continued professional commitment is needed to ensure the scientific integrity
of race and ethnicity as variables. At a minimum, researchers should clearly state the context in which these
valuable epidemiologic and public health study variables are being used, describe the method used to assess
and categorize these variables, and discuss all significant findings.
epidemiologic methods; ethnic groups; population groups
It is common to compare the prevalence of and risk
factors for various diseases and health outcomes between
members of different racial and ethnic groups in epidemiologic and public health studies. When these comparisons
are made, health disparities between racial and ethnic
groups have been well documented. In fact, one of the two
primary goals of the US Department of Health and Human
Services’ “Healthy People 2010” is to “eliminate health
disparities among segments of the population, including
differences that occur by gender, race or ethnicity, education or income, disability, geographic location or sexual
orientation” (1, p. 11).
Despite such a clearly identified goal, use of the terms
race and ethnicity in epidemiologic and public health
studies has generated a great deal of careful thought,
discussion, and controversy. The challenge facing
researchers is how to accurately measure such concepts and
how to interpret the results of studies that examine health
outcomes using these concepts. Race and ethnicity have no
widely accepted standardized definitions, and, despite the
fact that several authors have discussed the distinctions
between race and ethnicity (2–11), researchers often fail to
differentiate between them. Conversely, these terms are
commonly used interchangeably or as a single variable
labeled “race or ethnicity,” “race and ethnicity,” “raceethnicity,” or “race/ethnicity” (12).
Two major concerns exist when the results of studies using
these variables are interpreted. First is the potential for inaccurate classification or misrepresentation of a person’s race
or ethnicity. Many methods adopted to assess race and
ethnicity do not use an objective measurable characteristic
(12), no good evidence exists that current racial classifications accurately capture biologic or genetic similarities (12),
and accurate categorization is difficult even when persons
Correspondence to Dr. Edward M. Castillo, Emergency Medical Services, 6255 Mission Gorge Road, San Diego, CA 92120 (e-mail:
edward.castillo@sdcounty.ca.gov).
611
Am J Epidemiol 2004;159:611–619
612 Comstock et al.
FIGURE 1. Comparison of the proportion of journal articles that assessed race or ethnicity. Health Services Research data are for midyear of
the time periods shown in table 3 of Williams (10), American Journal of Epidemiology data for 1921–1990 are estimated from information in figure
2 of Jones et al. (6), and American Journal of Public Health data for 1980–1989 are estimated from information in figure 1 of Ahdieh and Hahn (2).
self-report their race or ethnicity because a person’s perception of race or ethnicity can change over time (3). The US
Office of Management and Budget classifies persons into
five racial groups (White, Black or African American,
American Indian or Alaskan Native, Asian, and Native
Hawaiian or other/Pacific Islander) and one ethnic category
(Hispanic/non-Hispanic origin) (13). However, these categories are broad, and the heterogeneity within each category is
great. Thus, even though these categories are widely used in
health-related research, they may have only limited value in
helping us to truly understand health disparities.
The second major concern is the possibility that
researchers will erroneously attribute health disparities to
racial or ethnic group differences and fail to adequately
study and understand the actual underlying causes of the
disparities. Factors such as socioeconomic status, social
status, educational opportunities, cultural views of disease
and health, acculturation, positions of power or powerlessness, inability to appropriately access health care, and racism
are most likely the actual driving forces behind these disparities (14). Focusing on race or ethnicity, and failing to
adequately recognize and measure underlying factors, limits
our ability to truly address these disparities.
Methodological recommendations for using race and
ethnicity as research variables presented by previous authors
(2, 4–6, 8–10) are fairly consistent. In general, these recom-
mendations have stated that research in which race or
ethnicity is used should
1. Differentiate between the concepts of race and ethnicity.
2. Provide justification for the use of race or ethnicity.
3. Describe in detail the method used to assess race or
ethnicity.
4. Provide the rationale for the categorization or grouping
of race or ethnicity.
5. Attempt to collect information that may be potentially
associated with race or ethnicity, such as socioeconomic
status or country of origin.
6. Interpret and discuss findings based on race or ethnicity.
7. Be cautious when comparing or generalizing across
studies, particularly when potential population differences
may be present.
However, we are aware of only three previous studies of
researchers’ practices in using race and/or ethnicity as variables in public health research that assessed the use of these
variables in journal articles published prior to 1991 to determine whether such methodological recommendations have
been accepted (2, 6, 10). Much has been written and recommended since that time, including a Centers for Disease
Control and Prevention/Agency for Toxic Substances and
Disease Registry Workshop Summary on the “Use of Race
and Ethnicity in Public Health Surveillance” (15). The
purpose of this study was to update our knowledge of how
researchers are currently using the variables race and
Am J Epidemiol 2004;159:611–619
Use of Race/Ethnicity in Health Research 613
TABLE 1. Diversity of variable names* used to refer to the concepts of race and ethnicity in articles
published in the American Journal of Epidemiology and the American Journal of Public Health, 1996–1999
Ancestral group
Ethnicity
Race/ethnicity
Birthplace/race
Ethnicity/national origin
Race/ethnicity code
Culture
Ethnicity/race
Race-ethnicity
Ethnic/ethnicity
Ethnic-specific
Racial groups
Ethnic background
Heritages
Racial or ethnic group
Ethnic category
Maternal race
Racial stock
Ethnic distribution
National origin
Racial/ethnic background
Ethnic groups
Race
Racial/ethnic group
Ethnic origin
Race or ethnicity
Racial/ethnic subgroup
* Alphabetical list.
ethnicity by conducting a comprehensive 4-year historical
review (1996–1999) of the recent epidemiologic and public
health scientific literature.
MATERIALS AND METHODS
The American Journal of Epidemiology and the American
Journal of Public Health are well-established, peerreviewed journals that publish scientific research of public
health importance. We reviewed each of the 1,918 articles
(1,016 from the American Journal of Epidemiology and 902
from the American Journal of Public Health) published in
the 1996, 1997, 1998, and 1999 volumes to determine how
researchers publishing in these journals used race and
ethnicity as scientific variables. These journals were chosen
because of their status as respected representatives of the
literature in the fields of epidemiology and public health and
because of a precedent for this type of literature review study
design in both journals (6, 16–18).
A standardized article assessment form developed by the
researchers was used to collect information about the use of
race and ethnicity in each article. We reviewed and tested the
assessment tool to assure standardization and ease of
completion. One researcher reviewed all articles from the
American Journal of Epidemiology; another studied all articles from the American Journal of Public Health.
Throughout review, researchers were in close contact to
resolve problems or answer questions as they arose. The
1,918 articles were initially screened for study eligibility,
with inclusion restricted to full or brief papers presenting
research involving human subjects from the United States.
On the basis of this criterion, 720 articles presenting research
of non-US study populations, nonhuman subjects, and methodological issues (those whose main purpose was to present
technique) were excluded. The remaining 1,198 articles (570
from the American Journal of Epidemiology and 628 from
the American Journal of Public Health) were included in the
study.
Questions on the article assessment form addressed
1. Whether the concepts of race or ethnicity were used in
the article
2. The stated purpose of including the variables race or
ethnicity, and in what context (as a social, cultural, biologic,
Am J Epidemiol 2004;159:611–619
or strictly demographic variable, etc., or as a marker for
acculturation, socioeconomic status, racism, etc.)
3. How these variables were assessed and reported
(including the categories of race/ethnicity used and the
number of categories reported)
4. How race and ethnicity were specifically referred to in
the analysis (as a risk factor, confounder, stratifier, covariate,
effect modifier, etc.)
5. The reported results related to race/ethnicity
6. Whether the implications of the race/ethnicity results
were addressed in the article (explanatory reasons for the
results, suggestions for further research, policy recommendations, etc.).
Both article reviewers (R. C. and E. C) were doctoral-level
graduate students in epidemiology. The reviewers did not
make any assumptions about the use of race or ethnicity in
the articles reviewed. In all cases, the authors’ specific
language was recorded. Using practice articles, both
reviewers independently pilot tested the article assessment
form for interrater reliability prior to beginning the complete
review. In addition, the reviewers met weekly to discuss the
review process and to address any challenges encountered to
ensure that the review procedures and results were consistent
throughout the entire period of review. Descriptive analysis
of the data was performed by using SPSS software, version
10.0 (SPSS, Inc., Chicago, Illinois).
RESULTS
Of the 1,198 articles included in the study, 919 (76.7
percent) contained references to race or ethnicity; 420 (45.7
percent) were published in the American Journal of Epidemiology and 499 (54.3 percent) in the American Journal of
Public Health. There was no apparent trend over time in the
proportion of articles assessing race or ethnicity published in
either journal (figure 1).
Table 1 shows the diversity of variable names that authors
used to describe the concept of race or ethnicity. Twentyseven different variable names were identified, nine of
which (33.3 percent) combined race and ethnicity into one
variable (i.e., race/ethnicity, race or ethnicity, racial or ethnic
group, etc.). Each variable listed in table 1 also had its associated categories.
614 Comstock et al.
TABLE 2. Examples of terms used to refer to specific categories of race or ethnicity in articles published in the American Journal of
Epidemiology and the American Journal of Public Health: 1996–1999*,†
Category
Asian
(n = 37)
Black
(n = 16)
Hispanic
(n = 46)
White
(n = 32)
Other
(n = 38)
Unknown
or missing
(n = 9)
Mixed race
or ethnicity
(n = 7)
Additional terms
(n = 34)
Asian
African
Caribbean
Hispanic
Anglo American
All others
Missing
Biethnic
American
Asian/Oriental
African American
Central/South
American
British Whites
Missing or other
Not indicated
Biracial
Ashkenazi Jews
Asian/Pacific
Islander
Black Americans
Cuban
Caucasian
Multiple/other/
unknown
Not ascertained
Black/White
Canadian
Asian American
Black and other
Cuban American
Caucasian/other
Neither Black nor
White
Not classified
Mixed ethnicity
East Indian
Asian or Filipino
Black Hispanic
Dominican
European
Non Black (all
others)
Unknown/refused
Multiethnic
Egyptian
Asian or other
races
Black, nonHispanic
Foreign born
Latina/Latino
Non-Hispanic
White
Non-American
Indian
Multiracial
Eskimo/Aleut/
Alaskan
Asian or Pacific
Islander
Black, other
Hispanic
Scandinavian
Non-Anglo
cultural groups
Partly Native
American
Foreign born
Asian other
Black/other races
Hispanic/Puerto
Rican
Southern
European
Non-Latino
Hawaiian
Asian-Indians
Blacks and other
minorities
Hispanic = White,
Spanish
surnamed
White/Anglo
Nonminority
(other)
Jewish
Cambodian
United States/
Blacks
Hispanic all races
White/other
Non US born
Middle
Easterners
Chinese
Hispanic
American
White American
Non-White/
Hispanics
Native American
Chinese
(Americans)
Hispanic Black
White and Asian
Non-White ethnic
minority
Pakistanis
Filipino
Hispanic ethnicity
White and other
Non-White or
Hispanic
United States
born
Japanese
Hispanic origin
White ethnicity
Other/mixed
Japanese
American
Hispanic
surname
White Hispanic
Other minorities
Korean
Hispanic White
White nonHispanic
Other races/
ethnicities
Korean
(Americans)
Latin American
White Spanish
surnamed
Other nonHispanic
Laotian
Latino/Hispanic
Non-Hispanic
Asian
Latino not Black
Orientals
Mexican
Pacific Islander
Mexican
American
Southeast Asian
Mexican
immigrants
Thai
Non-White
Hispanic
Vietnamese
Other Hispanic
Other non-White
Puerto Rican
South American
Spanish
US-born Hispanic
White-Hispanic of
Mexican
ancestry
* Terms shown are only a subset of those identified during article review.
† n, no. of terms used to describe each category.
Am J Epidemiol 2004;159:611–619
Use of Race/Ethnicity in Health Research 615
TABLE 3. Number of categories of race or ethnicity and the most common categories reported in articles
published in the American Journal of Epidemiology and the American Journal of Public Health, 1996–1999
American Journal of
Epidemiology
American Journal of
Public Health
Total
Year
Mean
Minimum,
maximum
Mean
Minimum,
maximum
Mean
Minimum,
maximum
No. of categories of race or ethnicity reported by year
1996
3.26
0, 16
3.29
1, 16
3.28
0, 16
1997
3.47
0, 24
2.96
1, 8
3.17
0, 24
1998
2.92
1, 13
2.98
1, 7
2.95
1, 13
1999
3.26
0, 10
3.12
1, 8
3.19
0, 10
All years
3.22
0, 24
3.07
1, 16
3.14
0, 24
Category
American Journal of
Epidemiology
(n = 339)
American Journal of
Public Health
(n = 448)
No.
No.
%
%
Total
(n = 787)
No.
%
Most common categories of race or ethnicity in articles reporting more than one racial or ethnic category
White
300
88.5
383
85.5
683
86.8
Black
255
75.2
347
77.5
602
76.5
Hispanic
44.1
122
36.0
225
50.2
347
Asian
82
24.2
94
21.0
176
22.4
Other
123
36.3
176
39.3
299
38.0
Table 2 shows examples of the exact terms used by article
authors to describe racial or ethnic categories and demonstrates the extraordinary diversity of categories in the 919
articles that used a racial or ethnic variable. For display, we
have grouped the diverse terms by the following, more
familiar or traditional general categories: 1) Asian, 2) Black,
3) Hispanic, 4) White, 5) other, 6) unknown or missing,
7) mixed race or ethnicity, and 8) additional terms. These
groupings are for the purposes of this study only, and we
recognize that others perhaps would group terms differently.
The heterogeneity, which can exist within broad racial or
ethnic categories, is displayed in the 46 terms used to
describe different peoples of Hispanic ethnicity (e.g., Cuban,
Dominican, Mexican, Puerto Rican, Spanish, US-born
Hispanic). Additionally, these examples of terms indicate
that racial or ethnic categories used in different studies may
be contradictory (e.g., Black Hispanic and Black nonHispanic, White/other and Black and other, and neither
Black nor White, nonminority (other), and other non-White).
Only a small number of categories representing mixed race
or ethnicity were reported.
Despite the diversity of categories displayed in table 2, the
number of racial or ethnic categories described in each
article ranged from 0 to 24, with an average of 3.14 categories per variable. In addition, the overall mean number of
categories that each journal used to identify race or ethnicity
was similar. An article that mentioned race or ethnicity but
did not provide categories was reported to have 0. The
average number of categories of race or ethnicity reported in
each of the 919 American Journal of Epidemiology and
American Journal of Public Health articles was less than
four, and the average number of categories varied only
Am J Epidemiol 2004;159:611–619
slightly between 1996 and 1999. In both journals, the
number of categories of race or ethnicity reported during
1998 decreased slightly, but the mean number of categories
reported in both journals in 1999 was similar to the means
reported in 1996. Of the 919 articles that contained references to race/ethnicity, 787 (85.6 percent) described more
than one category for this variable. Table 3 illustrates the
proportion of articles using the five most common categories
of race or ethnicity among the journal articles that used more
than one category (n = 787). The most common racial or
ethnic category reported was “White” (86.8 percent),
followed by “Black” (76.5 percent), “Hispanic” (44.1
percent), and “Asian” (22.4 percent). The “other” category
was reported in 299 (38.0 percent) articles.
Table 4 describes where in the article (the position) the
variables race or ethnicity were mentioned, the stated
purpose of using these variables, and the stated method of
use in the analysis (in the article authors’ own words). In
both journals, race or ethnicity was mentioned in results
tables most often compared with other sections (85.6
percent). The highest single specified purpose of use was as
a demographic variable (17.5 percent); however, in 57.4
percent of the articles, the purpose for using race or ethnicity
as variables was not described. Similarly, although the
highest single specified method of use was to adjust for (16.8
percent), almost half of the articles reviewed (48.7 percent)
failed to state the method of using race or ethnicity in the
analysis phase of the reported study. Figures 2 and 3 illustrate slight differences between the American Journal of
Epidemiology and the American Journal of Public Health in
trends over time in terms of purpose and method of using
race and ethnicity.
616 Comstock et al.
TABLE 4. Review of the use of race or ethnicity as scientific variables in articles identifying more than one
racial/ethnic group published in the American Journal of Epidemiology and the American Journal of Public
Health, 1996–1999
Use of race or ethnicity
American Journal of
Epidemiology
(n = 339)
American Journal of
Public Health
(n = 448)
Total
(n = 787)
No.
%
No.
%
No.
%
Introduction
103
30.4
126
28.1
229
29.1
Methods
281
82.9
331
73.9
612
77.8
Results text
268
79.1
348
77.7
616
78.3
Results tables
306
90.3
368
82.1
674
85.6
Discussion
189
55.8
184
41.1
373
47.4
Social variable
3
0.9
29
6.5
32
4.1
Cultural variable
8
2.4
12
2.7
20
2.5
Biologic variable
1
0.3
3
0.7
4
0.5
74
21.8
64
14.3
138
17.5
Position in article
Stated purpose of use
Demographic variable
Sociodemographic
28
8.3
29
6.5
57
7.2
Other purpose
121
35.7
30
6.7
151
19.2
Not stated
149
44.0
303
67.6
452
57.4
Risk factor
53
15.6
12
2.7
65
8.3
Confounder
38
11.2
25
5.6
63
8.0
Stratifier
31
9.1
29
6.5
60
7.6
8
2.4
28
6.3
36
4.6
Covariate
31
9.1
20
4.5
51
6.5
Control for
13
3.8
40
8.9
53
6.7
Adjust for
101
29.8
31
6.9
132
16.8
84
24.8
59
13.2
143
18.2
131
38.6
252
56.3
383
48.7
Subject self-report
35
10.3
82
18.3
117
14.9
Preexisting records
39
11.5
106
23.7
145
18.4
Defined by others
3
0.9
0
0.0
3
0.4
Visually interpreted
2
0.6
2
0.4
4
0.5
Surname surmised
7
2.1
4
0.9
11
1.4
17
5.0
20
4.5
37
4.7
255
75.2
244
54.5
499
63.4
Reported significant findings
197
58.1
238
53.1
435
55.3
Discussed findings
117
34.5
122
27.2
239
30.4
86
25.4
61
13.6
147
18.7
128
37.8
98
21.9
226
28.7
20
5.9
34
7.6
54
6.9
Stated method of use
Interaction term
Other method
Not stated
Stated method for assessing race or
ethnicity
Other method
Not stated
Reported results of the use of race or
ethnicity
Called for further research
Sited references for findings
Recommended policy
Table 4 also describes the method used to assess race or
ethnicity in the articles that were reviewed. The most
commonly stated methods of assessing race were preexisting
records such as medical records (18.4 percent) and subject
self-report (14.9 percent). The least commonly stated
methods in both journals were definition by others (0.4
Am J Epidemiol 2004;159:611–619
Use of Race/Ethnicity in Health Research 617
FIGURE 2. Comparison of the proportion of journal articles that used race or ethnicity to describe population demographics. Health Services
Research data are for midyear of the time periods shown in table 5 of Williams (10), and American Journal of Epidemiology data for 1921–1990
are estimated from information in figure 3 of Jones et al. (6). American Journal of Epidemiology and American Journal of Public Health data apply
to articles that used race or ethnicity to describe populations demographically, sociodemographically, socially, or culturally.
percent) and visual interpretation (0.5 percent). Race or
ethnicity was surname-surmised in 1.4 percent of the articles
reviewed. However, the majority of articles (63.4 percent)
failed to state how race or ethnicity was assessed.
Finally, table 4 shows that, overall, 55.3 percent of the articles described statistically significant findings related to race
or ethnicity, while only 30.4 percent of authors discussed
their findings and only 18.7 percent called for further
research related to their findings. Only rarely were policy
recommendations made on the basis of findings associated
with race or ethnicity.
DISCUSSION
Three previous reviews of the use of race and ethnicity as
research variables in the scientific literature reported methodological problems associated with the use of these variables. In 1991, Jones et al. (6) concluded from their study of
articles published in the 1921, 1930, 1940, 1950, and 1960
volumes of the American Journal of Hygiene and the 1965,
1970, 1975, 1980, 1985, and 1990 volumes of the American
Journal of Epidemiology that researchers should justify the
use of race as a scientific variable, explain how race was
measured, include measures of other factors such as socioeconomic status as appropriate, and interpret any observed
race-associated differences. In 1994, Williams (10)
concluded from his study of articles published in Health
Am J Epidemiol 2004;159:611–619
Services Research from 1966 to 1990 that there was a need
for more careful attention to the conceptualization and
measurement of race, that researchers should as much as
possible avoid conglomerate terms for racial groups, that
there was a need for more accurate definitions of racial and
ethnic status, and that researchers should always report how
race was assessed. In 1996, Ahdieh and Hahn (2) concluded
from their study of articles published in the American
Journal of Public Health from 1980 to 1989 that concepts
and terminology for race, ethnicity, and national origin
should be clearly and explicitly defined and that the scientific community needed to reach a consensus about the
meaning of such concepts.
This comprehensive historical review of articles published
in the American Journal of Epidemiology and the American
Journal of Public Health from 1996 to 1999 provides a more
current assessment of the use of race and ethnicity as scientific variables in epidemiologic and public health research.
We found enormous diversity in the terms used by
researchers to refer to race and ethnicity. When our findings
are compared with those of prior researchers (6, 10), it is
evident that the diversity of terms used to refer to the
concepts of race and ethnicity as well as the diversity of
terms used to categorize these concepts have expanded over
time. The enormous diversity could be a result of one of two
things. It could indicate that serious efforts are being made to
include much more detailed information about the racial
618 Comstock et al.
FIGURE 3. Comparison of the proportion of journal articles that used race or ethnicity to stratify or adjust for in analysis. Health Services
Research data are for midyear of the time periods shown in table 5 of Williams (10), and American Journal of Epidemiology data for 1921–1990
are estimated from information in figure 3 of Jones et al. (6). American Journal of Epidemiology and American Journal of Public Health data apply
to articles that used race or ethnicity to stratify, control for, or adjust for in analysis.
background, cultural heritage, ancestry, and ethnicity of
study subjects; or the diversity could be a result of an
increasing trend to be more politically correct regarding
these terms. Because of low rates of reporting of why these
variables are used in research, it will be difficult to truly
understand authors’ intentions. However, regardless of the
reason for the diversity of terms, when they were collapsed
into aggregated categories, it was clear that “White” and
“Black” continue to predominate the published research
despite the changing demographics of our country.
We found a high prevalence of the use of race and
ethnicity in the articles reviewed (76.7 percent referred to
concepts of race or ethnicity), but we also found that the
majority of the articles failed to describe the purpose for
including race or ethnicity in the study; the method of
assessing race or ethnicity, including the rationale behind the
categorization of these variables; or the implications of
significant findings pertaining to these variables. When our
results are compared with prior research (2, 6, 10), it appears
that the prevalence of the use of race or ethnicity has
remained fairly constant since the mid-1980s in the American Journal of Epidemiology, while it has increased in the
American Journal of Public Health (figure 1). A similar
comparison with prior work (6, 10) reveals that the stated
purpose and method of using race and ethnicity in the epide-
miologic and public health literature has varied over time
and by journal (figures 2 and 3).
In 2000, Lin and Kelsey discussed the use of race and
ethnicity in epidemiologic research, and, similar to prior
authors (2, 6, 10), concluded that epidemiologists “should
continue to study racial and ethnic patterns of health and
disease, but they should seek to improve upon the ways in
which these concepts are captured” (12, p. 195). Some of
their suggestions follow: 1) researchers should work to
improve methods of assessing ethnicity, including
addressing a subject’s identification with multiple ethnic
groups; 2) researchers should try to separate the effects of
socioeconomics, social class, education, and other factors
from those of race and ethnicity; and 3) researchers should
explain how race and ethnicity classifications were identified. Our review indicates that there is currently no
consensus concerning the definition and categorization of
race and ethnicity as scientific variables, the methods used to
assess race and ethnicity, and the appropriate interpretation
of the data obtained from using these variables. Given all of
the possible ways in which racial and ethnic characteristics
can be described and all of the potential causes that might
explain racial disparities, perhaps it is not possible for the
scientific community to come to a consensus on guidelines
or standards for using these variables. Perhaps it is not even
a desirable or achievable goal. However, it is possible for
Am J Epidemiol 2004;159:611–619
Use of Race/Ethnicity in Health Research 619
researchers to carefully record and discuss why race or
ethnicity is being used, how it is being assessed, and what the
potential findings based on its use may imply.
Research should also begin to more frequently consider
disparities within each racial or ethnic group rather than
simply focusing on differences between them (19). There are
distinct differences within racial groups in different
geographic regions of the United States, such as the differences between the Hispanic populations of the Southwest
and those of the Southeast. This factor will continue to
increase in importance as racial groups in the United States
become more intertwined. As the many diverse ingredients
in the country’s “melting pot” become increasingly hard to
define, differences based on racial classification will become
more difficult to interpret. The etiologic clues embedded in
observed racial differences will be better understood only
after the complexities of the concepts of race and ethnicity
are taken into account.
In conclusion, the use of race or ethnicity in epidemiologic
and public health research affects the quantification and
explanation of health outcomes, including health disparities.
Although previous authors who have questioned the value of
using race and ethnicity as scientific variables have proposed
methodological guidelines aimed at increasing the integrity
of these variables, it is clear from our study that researchers
have not yet come to a consensus concerning their use.
Continued professional commitment is needed to ensure the
scientific integrity of race and ethnicity as variables. At a
minimum, researchers should clearly state the context in
which these valuable epidemiologic and public health
research variables are being used, describe the method used
to assess these variables, and discuss all significant findings.
Doing so will ensure continued constructive scientific dialog
about the interpretation of findings regarding race or
ethnicity and will promote the successful development of
intervention strategies aimed at eliminating health disparities
linked to race and ethnicity.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
ACKNOWLEDGMENTS
The research for this manuscript was completed while Dr.
Comstock and Dr. Castillo were students in the Joint
Doctoral Program in Public Health, San Diego State University/University of California, San Diego.
REFERENCES
1. US Department of Health and Human Services. Healthy
people 2010: understanding and improving health. 2nd ed.
Am J Epidemiol 2004;159:611–619
16.
17.
18.
19.
Washington, DC: US GPO, 2000. (Also available at http://
www.healthypeople.gov/Document/pdf/uih/uih.pdf).
(Accessed January 19, 2004).
Ahdieh L, Hahn RA. Use of the terms ‘race’, ‘ethnicity’, and
‘national origins’: a review of articles in the American Journal
of Public Health, 1980–1989. Ethn Health 1996;1:95–8.
Hahn RA, Truman BI, Barker ND. Identifying ancestry: the
reliability of ancestral identification in the United States by self,
proxy, interviewer, and funeral director. Epidemiology 1996;7:
75–80.
Hahn RA, Stroup DA. Race and ethnicity in public health surveillance: criteria for the scientific use of social categories.
Public Health Rep 1994;109:7–15.
Jones CP. Invited commentary: “Race,” racism, and the practice of epidemiology. Am J Epidemiol 2001;154:299–304.
Jones CP, LaVeist TA, Lillie-Blanton M. “Race” in the epidemiologic literature: an examination of the American Journal of
Epidemiology, 1921–1990. Am J Epidemiol 1991;134:1079–
84.
Kaufman JS, Cooper RS. Commentary: considerations for use
of racial/ethnic classification in etiologic research. Am J Epidemiol 2000;154:291–8.
Senior PA. Ethnicity as a variable in epidemiological research.
BMJ 1994;309:327–30.
Singh SP. Ethnicity in psychiatric epidemiology: need for precision. Br J Psychiatry 1997;171:305–8.
Williams DR. The concept of race in Health Services Research:
1966 to 1990. Health Serv Res 1994;29:261–74.
Last JM, ed. A dictionary of epidemiology. 4th ed. New York,
NY: Oxford University Press, 2001.
Lin SS, Kelsey JL. Use of race and ethnicity in epidemiologic
research: concepts, methodological issues, and suggestions for
research. Epidemiol Rev 2000;22:187–202.
Office of Management and Budget. Standards for maintaining,
collecting, and presenting federal data on race and ethnicity.
Federal Register 62 FR 58781–58790, 1997. (Also available at
http://www.whitehouse.gov/omb/fedreg/ombdir15.html).
(Accessed January 19, 2004).
Jones CP. Levels of racism: a theoretic framework and a gardener’s tale. Am J Public Health 2000;90:1212–15.
Use of Race and Ethnicity in Public Health Surveillance. Summary of the CDC/ATSDR workshop. Atlanta, Georgia, March
1–2, 1993. MMWR Recomm Rep 1993 Jun 25;42:1–16.
Jackson LW, Lee NL, Samet JM. Frequency of policy recommendations in epidemiologic publications. Am J Public Health
1999;89:1206–10.
Ottenbacher KJ. Quantitative evaluation of multiplicity in epidemiology and public health research. Am J Epidemiol 1998;
147:615–19.
Savitz DA, Tolo K, Poole C. Statistical significance testing in
the American Journal of Epidemiology, 1970–1990. Am J Epidemiol 1994;139:1047–52.
Khoshnood B, Lee K, Wall S, et al. Short interpregnancy intervals and the risk of adverse birth outcomes among five racial/
ethnic groups in the United States. Am J Epidemiol 1998;148:
798–805.