Body mass index was calculated as weight in kilograms divided by height
in meters squared. Estimates were adjusted for age and sex and weighted to
reflect national population estimates. Data for underweight respondents are
not reported. P<.001 for trend in the age- and
sex-adjusted prevalence of obesity with longer duration of US residence among
foreign-born individuals. Error bars represent 95% confidence intervals.
Goel MS, McCarthy EP, Phillips RS, Wee CC. Obesity Among US Immigrant Subgroups by Duration of Residence. JAMA. 2004;292(23):2860-2867. doi:10.1001/jama.292.23.2860
Author Affiliations: Division of General Internal
Medicine, Department of Medicine, Feinberg School of Medicine, Northwestern
University, Chicago, Ill (Dr Goel); Division of General Medicine and Primary
Care, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard
Medical School, Boston, Mass (Drs McCarthy, Phillips, and Wee).
Context The prevalence of obesity has increased substantially since the 1980s.
While immigrants are the fastest growing segment of the US population, little
is known about obesity or clinician counseling about diet and exercise in
Objectives To estimate the prevalence of obesity among immigrant subgroups and
quantify the magnitude of the association with duration of US residence, and
to describe reported diet and exercise counseling by birthplace, race, and
Design, Setting, and Participants Cross-sectional study using data from the 2000 National Health Interview
Main Outcome Measures Body mass index (BMI, measured as weight in kilograms divided by the
square of height in meters) based on self-reported height and weight measurements,
and self-reported rates of diet and exercise counseling.
Results Of 32 374 respondents, 14% were immigrants. The prevalence of obesity
was 16% among immigrants and 22% among US-born individuals. The age- and sex-adjusted
prevalence of obesity was 8% among immigrants living in the United States
for less than 1 year, but 19% among those living in the United States for
at least 15 years. After adjusting for age, sociodemographic, and lifestyle
factors, living in the United States for 10 to 15 and at least 15 years was
associated with BMI increases of 0.88 and 1.39, respectively. The association
for 15 years or more was significant for all immigrant subgroups except foreign-born
blacks. Additionally, immigrants were less likely than US-born individuals
to report discussing diet and exercise with clinicians (18% vs 24%, P<.001; 19% vs 23%, P<.001,
respectively). These differences were not accounted for by sociodemographic
characteristics, illness burden, BMI, or access to care among some subgroups
Conclusions Among different immigrant subgroups, number of years of residence in
the United States is associated with higher BMI beginning after 10 years.
The prevalence of obesity among immigrants living in the United States for
at least 15 years approached that of US-born adults. Early intervention with
diet and physical activity may represent an opportunity to prevent weight
gain, obesity, and obesity-related chronic illnesses.
Since the 1980s, the US population has become more obese, with similar
trends documented worldwide.1- 3 The
World Health Organization recently described “globesity” as a
global epidemic of obesity affecting at least 300 million people, with a 3-fold
or greater increase since 1980 in parts of Eastern Europe, the Middle East,
the Pacific Islands, and China.4 Nevertheless,
the prevalence of obesity in most parts of the world is lower than in the
The high prevalence of obesity has produced a major burden of obesity-related
illnesses. In the year 2000, overweight and obesity accounted for nearly 17%
of all deaths in the United States, a mortality rate surpassed only by that
of tobacco use.5According to the Centers for
Disease Control and Prevention, poor diet and physical inactivity, both modifiable
behaviors, cause a large part of this mortality from overweight.5 Thus,
many guidelines recommend obesity prevention through the promotion of exercise
and healthy diet.6- 8
Evidence suggests significant racial and ethnic differences in the prevalence
of obesity and the susceptibility to obesity-related illnesses, particularly
among individuals who are black, Latino, and Asian, relative to whites.9,10 However, few data are available about
the epidemiology of obesity among immigrants, the fastest growing segment
of the US population, currently comprising more than 11% of the total US population
and an even larger proportion of many minority groups.11 Immigrants
generally originate from countries where the prevalence of obesity is lower
than that of the United States, but acculturation to US norms over time may
lead to an increasing prevalence of obesity among this population.3,12 The magnitude of the change in body
mass index (BMI) and the consistency of the effect of acculturation on BMI
among various immigrant subgroups, however, remain unclear.
Additionally, immigrants face more barriers to quality health care13 and are less likely to receive preventive health
care than persons born in the United States.14 Whether
immigrants are less likely to discuss diet and exercise with clinicians is
We therefore examined the relationship between prevalence of obesity
and years of US residence among immigrants nationally and explored whether
counseling about diet and exercise may differ between immigrants and US-born
We used data from the Sample Adult Module of the 2000 National Health
Interview Survey (NHIS), an in-person health survey of the civilian, noninstitutionalized
population, administered by the US Bureau of the Census for the National Center
for Health Statistics.15 Latino and black populations
are oversampled to allow for more precise estimation of these minority groups.
A total of 100 618 respondents from 38 633 households provided information
about basic measures of health status, utilization of health services, and
sociodemographics including country of birth. In addition, 1 randomly selected
adult per household, aged 18 years or older, was asked to complete the Sample
Adult Module (n=32 374), which elicited detailed information on health
care services, behavior, and health status including height and weight. The
combined response rate to both components of the survey, based on the American
Association for Public Opinion Research standards for Response Rate 5,16 was 72%. Sample weights provided by National Center
for Health Statistics account for the complex sampling design of NHIS and
also account for nonresponse. These weights are modified for poststratification
adjustments for census sex, age, and race/ethnicity population controls. This
weighting of data allows estimates that generalize to the civilian, non-institutionalized
population. The survey is administered in only Spanish or English languages
and does not allow proxy respondents for Sample Adult questions. Family members
may translate for a non–English- or non–Spanish-speaking respondent
who is present in the home.
We received institutional review board exemption from Beth Israel Deaconess
Medical Center for this study.
We defined obesity as having a BMI of more than 30 (measured as weight
in kilograms divided by the square of height in meters) among adults.17 The National Health Interview Survey calculates BMI
from self-reported information on height and weight, measures previously established
as largely valid for BMI when used in combination with adjustments for age.18 We used BMI as a continuous outcome for all linear
regression analyses examining the relationship between years of residence
in the United States and BMI. For all other analyses, we classified BMI as
a categorical outcome based on the National Institutes of Health classification
scheme17: underweight (BMI <18.5), normal
weight (18.5-<25), overweight (25-<30), obese class I (30-<35), obese
class II (35-<40), and obese class III (≥40).
We were also interested in exploring differences in receipt of dietary
and exercise counseling, which were ascertained only in respondents who reported
seeing or talking with a health care professional in the past year (n=26 402).
We considered those responding to the following questions in the affirmative
as having received that counseling. Dietary counseling was assessed with the
question “During the past 12 months has a doctor or other health professional
talked with you about your diet and eating habits?” Exercise counseling
was assessed with the question “During the past 12 months did a doctor
or other health professional recommend that you begin or continue to do any
type of exercise or physical activity?”
Race/ethnicity was ascertained with 2 questions. Respondents were asked,
“Do you consider yourself to be Hispanic, or Latino?” and “What
race do you consider yourself to be?” Respondents were asked to select
1 or more of 16 options for the latter question. We considered foreign birth
as a proxy for immigrant status and defined foreign birth as birthplace either
in a US territory or outside of the United States, based on responses to the
question “Where were you born?” Choices for responses included
specific states within the United States, the United States, other countries,
or territories. The National Health Interview Survey
categorized respondents born in US territories as foreign born; as these respondents
were likely to be culturally more similar to other foreign-born respondents
than to US-born respondents, we used this grouping also.
For our analyses, we considered race/ethnicity and birthplace together.
We classified respondents into the following 8 nonoverlapping categories:
US-born white, black, Latino, and Asian (including Asian American and Pacific
Islander); and foreign-born white, black, Latino, and Asian (including Asian
American and Pacific Islander).
Finally, we categorized respondents according to their years of residence
in the United States (<1 year, 1-<5 years, 5-<10 years, 10-<15
years, or ≥15 years) based on their response to the question “About
how long have you been in the United States?”
We also considered other sociodemographic characteristics, illness burden,
measures of access to health care, and health behaviors. Sociodemographic
characteristics included age (in decades), sex, marital status (married, not
married), region of residence (Northeast, Midwest, South, West), level of
education (<high school graduate, high school graduate, some post–high
school education, ≥college graduate), and annual household income (<$20 000,
≥$20 000). We defined illness burden using several variables: self-reported
health status (excellent/very good, good, fair/poor), which has previously
been shown to be associated with mortality in a multiethnic cohort19; smoking status (never, current, past); alcohol use
(rare [<1 drink/wk], moderate [between 1 drink/wk and no more than 2 drinks/d],
heavy [>2 drinks/d]); presence of concurrent illnesses (diabetes, hypertension,
coronary artery disease, stroke, ulcer, arthritis, cancer other than nonmelanoma
skin cancer, or other); and hospitalizations in the past year (0, 1, ≥2).
We measured access to care using the following proxies: type of health insurance
(none, Medicare, Medicaid, private, other), and usual source of care (general
practitioner, specialist including obstetrician/gynecologist, no usual clinician
but usual source of care [such as a health center, physician’s office,
hospital clinic, or some other place not including an emergency department],
no usual source of care).
Because the level of current physical activity might influence rates
of obesity and receipt of dietary and exercise counseling, we included level
of leisure physical activity as a potential confounder. We categorized physical
activity level as high (vigorous activity ≥2 times/wk or moderate activity
≥4 times/wk), moderate (vigorous activity 1 time/wk or moderate activity
1-3 times/wk), or sedentary (no vigorous or moderate activity/week) based
on validated methods described previously.20
Additionally, because respondent occupation may influence level of physical
activity, in secondary subgroup analyses we also examined the association
of occupation on change in BMI with duration of residence, and on reported
receipt of dietary and exercise counseling in a subgroup of respondents with
information on occupation. We qualitatively categorized NHIS-defined occupations
into sedentary (eg, administrators, engineers, health care providers) or active
(eg, police/firefighters, construction laborers, farm/agricultural workers)
by adapting methods used by Wee et al.21We
were able to classify occupational activity levels for 54% of those who reported
having an occupation (n=19 707), and excluded those with occupations
that were difficult to classify (eg, personal service, farm operators/managers).
We conducted bivariable analyses comparing baseline characteristics
across our 8 groups of interest defined according to race/ethnicity and birthplace.
We used χ2 statistics for all categorical variables.
To determine differences in BMI among foreign-born adults according
to duration of residence in the United States, we first described the age-
and sex-adjusted prevalence of obesity among the foreign-born by years of
US residence, and used the χ2test for trend in this relationship. We then fit a linear regression model with BMI as the
continuous outcome and the categorical variable of years in the United States
as the primary association of interest, collapsing the lowest 2 categories
after ensuring no significant differences (0-<5 years, 5-<10 years,
10-<15 years, ≥15 years), and adjusting for age, sex, race/ethnicity,
education, and income. To examine whether differences in BMI could be attributable
to differences in lifestyle, we additionally adjusted for health behaviors
such as alcohol use, smoking, and leisure physical activity. We then further
adjusted for occupation (sedentary, active) among a subgroup with information
on level of occupational activity. To examine the association between duration
of US residence and BMI among various subgroups, we stratified our primary
analysis by sex, then by race/ethnicity.
To explore differences in reported dietary or exercise counseling by
race/ethnicity and birthplace, we fit logistic regression models for each
respective outcome, adjusting for potential confounders: sociodemographic
characteristics, illness burden, and access to care, as previously defined.
To examine the influence of physical activity, we further adjusted for level
of leisure physical activity (high, moderate, sedentary) and performed subgroup
analyses among those with classifiable occupational activity (active, sedentary).
For all multivariable analyses, we assessed confounding using a 10%
change in the estimated β coefficients from the multivariable models
as our criterion. For all analyses, data were weighted to reflect national
population estimates and analyzed using SAS-callable SUDAAN software, version
7.5 (Research Triangle Institute, Research Triangle Park, NC) to adjust for
the complex sampling design. We considered a 2-tailed P value of ≤.05 statistically significant for all analyses.
Because this is a secondary analysis of a survey database, we were limited
to post hoc power analyses. Based on these, we estimated an 80% power to detect
the following differences in BMI between immigrants living in the United States
less than 5 years and those living in the United States for 10 to 15 years:
0.8 for all immigrants, 1.6 for whites, 2.7 for blacks, 1.0 for Latinos, and
1.6 for Asians. Comparing immigrants living in the United States less than
5 years and those living in the United States for 15 or more years, we could
detect the following differences in BMI: 0.7 for all immigrants, 1.2 for whites,
2.5 for blacks, 1.0 for Latinos, and 1.3 for Asians.
For rates of reported dietary counseling, we had 80% power to detect
the following absolute differences relative to our reference group of US-born
whites: 2.3% for US-born blacks, 3.1% for US-born Latinos, 10.1% for US-born
Asians, 4.3% for foreign-born whites, 6.5% for foreign-born blacks, 2.8% for
foreign-born Latinos, and 5.3% for foreign-born Asians. This was similar for
reported exercise counseling.
The 32 374 eligible respondents represent an estimated 201 million
adults in the United States. Of these respondents, 14% were foreign born.
Compared with US-born respondents, foreign-born respondents were generally
older, had lower annual household incomes and education, had lower illness
burden, and had poorer access to care (Table 1 and Table 2). The foreign-born
respondents were less often obese than those born in the United States (16%
vs 22%, P<.001), but they were also more often
sedentary than US-born respondents.
In this cross-sectional analysis, adjusted for age and sex, the foreign-born
were generally less likely than the US-born to be overweight and obese (Figure). However, the proportion of overweight
and obese foreign-born individuals increased with longer duration of residence
in the United States. The prevalence of obesity among foreign-born respondents
living in the United States for less than 1 year was 8%. In contrast, the
BMI distribution of foreign-born respondents living in the United States for
at least 15 years approached that of US-born respondents, with 41% at normal
weight, 38% overweight, and 19% obese, compared with 41%, 35%, and 22% of
the US-born, respectively.
Examining the relationship between the magnitude of change in BMI and
duration of US residence showed that living in the United States for 10 years
or more was associated with a significantly higher BMI (Table 3). Results were similar after additional adjustment for health
behaviors and in a subgroup with classifiable occupational activity. Years
of US residence was associated with a significant increase in BMI for men,
women, and all racial/ethnic groups, except for foreign-born blacks. We found
no significant interaction between race/ethnicity and number of years in the
Overall, 24% of respondents reported discussing their diet and eating
habits with a clinician in the past year; foreign-born respondents were less
likely to report counseling than were US-born respondents (18% vs 24%, P<.001). After adjustment, foreign-born blacks and Latinos
were less likely to report dietary counseling than US-born whites (Table 4). Results were similar after additional
adjustment for level of leisure physical activity and in a subgroup analysis
incorporating occupational activity.
Overall, 23% of respondents reported that a clinician recommended beginning
or continuing exercise within the last 12 months: 19% of foreign-born compared
with 23% of US-born respondents (P<.001). After
adjustment, foreign-born black, but not foreign-born Latino or Asian, respondents
were less likely to report exercise counseling than were US-born whites; US-born
black respondents were also significantly less likely to report exercise counseling
(Table 4). Relationships were similar
after adjustment for level of leisure physical activity and in a subgroup
analysis adjusting for occupational activity.
Our study shows that among a number of immigrant subgroups, longer duration
of residence in the United States is associated with higher BMI, such that
the distribution of obesity among immigrants residing in the United States
for 15 or more years approaches that of the US born. To illustrate the impact,
for a typical 5′4″ (1.63-m) immigrant woman and a typical 5′9″
(1.75-m) immigrant man this amounts to an excess 9 lb (4.05 kg) and 11 lb
(4.95 kg), respectively, in addition to any weight gained due to aging or
other factors. Using a nationally representative sample of foreign-born individuals,
we found that this trend is present among immigrant whites, Latinos, and Asians.
Our study also suggests that some immigrant minority groups may be less likely
to discuss diet or exercise with their clinicians.
Using data from the 1993-1994 NHIS, Singh and Siahpush22 reported
that immigrants with longer duration of residence in the United States appear
to be at higher risk of obesity than recent US immigrants. As part of a larger
study examining causes of morbidity and mortality among immigrants, they described
a cross-sectional relationship of increasing prevalence of obesity among immigrants
and longer duration of residence in the United States, after adjusting for
sociodemographic factors. Kaplan et al23 confirmed
this among Hispanics using 1998 NHIS data. Our study, using more recent data
from 2000, in addition to being consistent with these previous results, estimated
the magnitude of the increase in BMI associated with longer duration of residence
in the United States and described these relationships across racial/ethnic
subgroups. We found that BMI did not increase substantially until after living
in the United States for at least 10 years, suggesting a threshold effect.
Trends in obesity among immigrants may reflect acculturation and adoption
of the US lifestyle, such as increased sedentary behavior and poor dietary
patterns. They may also be a response to the physical environment of the United
States, with increased availability of calorically dense foods and higher
reliance on labor-saving technologies. After 10 years, BMI appears to increase
more substantially, a pattern that was consistent across sex and race/ethnicity,
except among immigrant blacks for whom there was no change in BMI associated
with years of US residence. Because of the cross-sectional nature of our data,
we cannot establish a causal relationship between duration of residence and
BMI. Our findings could reflect a cohort effect in which immigrants residing
in the United States for a longer duration were more obese at the time of
immigration than more recent immigrants; however, this is unlikely given the
global trend toward obesity and the consistency between our findings and earlier
Increasing obesity with longer duration of US residence is concerning
given the rapid growth of the immigrant population11 and
the adverse health care consequences associated with obesity.5 Unfortunately,
our findings also suggest that clinicians may be paying less attention to
diet and exercise among some immigrant groups. We found that foreign-born
minorities were generally less likely to report discussing diet and exercise
with their clinicians, and US-born blacks were also less likely to report
Reasons for these differences are unclear. Variation in sociodemographics,
illness burden, and access to care may have contributed. However, the measured
differences may be an artifact of how questions about counseling in the NHIS
were interpreted by different cultural groups.24 Research
is needed to replicate our findings using more culturally sensitive methods
and should also explore potential mechanisms for any observed differences
such as language barriers, cultural discordance between patient and clinician,
and clinician perceptions about lifestyle behaviors and obesity risk among
Our study has some important limitations. Our results were based on
self-reported information, which may lead to recall bias. Whether foreign-born
and US-born adults have different biases in reporting weight, height, receipt
of counseling, illness burden, or health behaviors is unclear, given that
these concepts may be understood differently in different cultural groups.24 The NHIS attempts to account for the nonresponse
rate of 28% in the weighting procedure; however, the possibility of residual
bias remains, which could limit the generalizability of our findings. It is
unclear what proportion of non–English- and non–Spanish-speaking
adults were excluded and what proportion used translators. In addition, we
had no direct measures of acculturation, except for years of US residence,
thereby limiting our ability to examine mechanisms of weight gain among immigrants.
Finally, our study may have missed some important associations due to inadequate
power to detect potentially clinically important differences in reported counseling
rates, up to 5.3% for foreign-born Asians.
In summary, immigrants appear to assume a similarly high prevalence
of obesity as US-born adults with longer duration of residence. With the growing
immigrant population in the United States, early clinician intervention on
diet and physical activity may represent an important opportunity to prevent
weight gain, obesity, and obesity-related chronic illnesses.
Corresponding Author: Mita Sanghavi Goel,
MD, MPH, Northwestern University, Division of General Internal Medicine, 676
North St. Clair Ave, Suite 200, Chicago, IL 60611 (firstname.lastname@example.org).
Author Contributions: Dr Goel had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Goel, McCarthy, Phillips,
Analysis and interpretation of data: Goel,
McCarthy, Phillips, Wee.
Drafting of the manuscript: Goel, McCarthy,
Critical revision of the manuscript for important
intellectual content: Goel, McCarthy, Phillips, Wee.
Statistical analysis: Goel, McCarthy, Phillips,
Administrative, technical, or material support:
Study supervision: Goel, McCarthy, Phillips,
Funding/Support: Dr Goel was supported by an
institutional National Research Service Award (5T32PE11001) and by the Ryoichi
Sasakawa Fund when this research was conducted. Dr Wee is supported by a career
development award from the National Institute of Diabetes and Digestive and
Kidney Diseases (K23DK02962). Dr McCarthy is the recipient of a First Independent
Research and Transition Award from the National Cancer Institute (R29 CA79052).
Dr Phillips is supported by a Mid-Career Investigator Award from the National
Institutes of Health (K24 AT00589-01A1).
Role of the Sponsors: None of the organizations
funding the investigators had any role in the design and conduct of the study;
collection, management,analysis, and interpretation of the data;
and preparation, review,or approval of the manuscript.