International Journal of Obesity (2000) 24, 1188±1194 ß 2000 Macmillan Publishers Ltd All rights reserved 0307±0565/00 $15.00 www.nature.com/ijo Body mass index in a US national sample of Asian Americans: effects of nativity, years since immigration and socioeconomic status DS Lauderdale1* and PJ Rathouz1 Department of Health Studies, University of Chicago, Chicago, IL 60637, USA OBJECTIVE: To examine body mass index (BMI) and the proportion overweight and obese among adults age 18 ± 59 in the six largest Asian American ethnic groups (Chinese, Filipino, Asian Indian, Japanese, Korean, Vietnamese), and investigate whether BMI varies by nativity (foreign- vs native-born), years in US, or socioeconomic status DESIGN: Cross-sectional interview data were pooled from the 1992 ± 1995 National Health Interview Survey (NHIS) SUBJECTS: 254,153 persons aged 18 ± 59 included in the 1992 ± 1995 NHIS Sample sizes range from 816 to 1940 for each of six Asian American ethnic groups MEASUREMENTS: Self-reported height and weight used to calculate BMI and classify individuals as overweight (BMI 25 kgam2) or obese (BMI 30 kgam2), age, sex, years in the US, household income and household size RESULTS: For men, the percentage overweight ranges from 17% of Vietnamese to 42% of Japanese, while the total male population is 57% overweight For women, the percentage overweight ranges from 9% of Vietnamese and Chinese to 25% of Asian Indians, while the total female population is 38% overweight The percentage of Asian Americans classi®ed as obese is very low Adjusted for age and ethnicity, the odds ratio for obese is 3.5 for women and 4.0 for men for US - vs foreign-born Among the foreign-born, more years in the US is associated with higher risk of being overweight or obese The association between household income for women is similar for US-born Asian Americans and Whites and Blacks, but is much weaker for foreign-born Asian Americans CONCLUSIONS: While these data ®nd low proportions of Asian Americans overweight at present, they also imply the proportion will increase with more US-born Asian Americans and longer duration in the US International Journal of Obesity (2000) 24, 1188±1194 Keywords: Asian Americans; body mass index; obesityaethnology; acculturation; socioeconomic factors Introduction The public health signi®cance of obesity derives largely from its well-documented associations with chronic conditions such as diabetes mellitus, osteoarthritis and hypertension, conditions whose prevalence varies by race and ethnicity The percentage of obese adults has increased markedly in recent decades In the Third National Health and Nutrition Examination Survey (NHANES III), conducted from 1988 to 1994, the proportion of the adult population in the United States with body mass index (BMI, kgam2) of 25 or higher was approximately 54%.1 That proportion varied substantially by the racial and ethnic categories available in NHANES III (non-Hispanic White, nonHispanic Black and Mexican American), with 67% of Mexican Americans, 63% of Blacks and 53% of Whites overweight (BMI 25) Obesity (BMI 30) *Correspondence: DS Lauderdale, Department of Health Studies, University of Chicago, 5841 S Maryland Avenue, MC 2007, Chicago, IL 60637, USA E-mail: lauderdale@health.bsd.uchicago.edu Received 27 September 1999; revised April 2000; accepted 18 April 2000 similarly varied by ethnicity and race, with 21% of White, 30% of Black, and 28% of Mexican American adults found to be obese There is a paucity of national data on the health status of Asian Americans, a numerous, rapidly increasing, and ethnically diverse minority group NHANES III, for example, over sampled Mexican Americans and Blacks, but not Asian Americans Nor can one distinguish Asian Americans in the publiclyreleased ®les, since they are classi®ed as `Other' Further, there are no other national data collections with physical examinations and suf®cient numbers to characterize BMI and obesity of Asian Americans Our knowledge of this and other cardiovascular risk factors among Asian Americans derives disproportionately from the study of one ethnic group in an environment atypical of the US as a whole, Japanese Americans in Hawaii Mortality data from the National Center for Health Statistics suggest that Asian Americans in general are a uniquely healthy group with the highest life expectancy in the country.2 However, since the majority of Asian American adults are foreign-born, the question arises of whether Asian American good health is owing to a `healthy immigrant' effect The two most frequently presented Asian American body mass index DS Lauderdale and PJ Rathouz explanations for a `healthy immigrant' effect are (1) the self-selection of people able to and choosing to immigrate, and (2) the maintenance of healthy behaviors associated with a traditional lifestyle, including diet and physical activity In this study, as one indicator of whether lifestyle risk factors for chronic diseases are associated with nativity (ie US-born vs foreign-born) for Asian Americans and whether risk factors change for the foreignborn following immigration, we use national interview data to examine variation in BMI, the proportion overweight and the proportion obese for the six major Asian American ethnic groups (Chinese, Filipino, Japanese, Asian Indian, Korean and Vietnamese) We ask whether BMI and the odds of being overweight or obese vary by nativity Among the foreignborn, we further investigate whether BMI and the odds of being overweight or obese vary with years since immigration Finally we examine whether the association between socioeconomic status and BMI is the same for Asian Americans as it is for Whites and Blacks Methods Data The National Health Interview Survey (NHIS) is a nationally representative annual cross-sectional interview survey conducted by the National Center for Health Statistics (NCHS) The target population is civilian, non-institutionalized residents of the US The sampling plan follows a multistage area probability design that permits the representative sampling of households Data are collected through a personal household interview conducted by interviewers employed and trained by the US Bureau of the Census according to procedures speci®ed by NCHS Translators are used to collect information from persons with limited English pro®ciency NHIS data are collected annually in 36,000 ± 47,000 households from 92,000 ± 125,000 persons.3 The annual response rate is greater than 90% Beginning in 1992, the NHIS added Asian ethnic group detail to the `race' item for Japanese, Chinese, Filipino, Asian Indian, Korean and Vietnamese These six groups together comprise about 90% of the Asian American population Then in 1996, the categories for the three smaller groups, Asian Indian, Korean and Vietnamese, were combined into the `other Asian and Paci®c Islander' group for data release Because the number of Asian Americans surveyed in each ethnic group in one single year is as few as 100, we combine data from the 1992 ± 1995 NHIS to achieve adequate sample sizes of subjects with maximum ethnic detail.4 ± This study includes persons aged 18 ± 59 y of age Persons 60 and older are not included both because health effects of BMI may differ for the elderly and because BMI may increasingly re¯ect the consequences of ill-health as age increases Among the core questions asked of each member of sampled households are raceaethnicity, age, sex, height, weight, years in US for the foreign-born ( ` y, ± 5, ± 10, 10 ± 15, b 15 y), household income and household size We re-categorize years in the US as less than y, ± 15 y and 15 y or more We categorize income by $10,000 increments up to $50,000 or greater, the highest category collected by NHIS Income is adjusted for household size in regression models We use household income rather than education as an indicator of socioeconomic status in this study because equating the number of years of school for persons educated in the US and the dissimilar educational systems in each Asian country is clearly problematic We calculate the BMI by converting height and weight to the metric system and calculating kgam2 for each subject Persons with missing data for height, weight, age, race or sex are not included in this study (approximately 2% of sample) Approximately 15% of records lack household income data, and they are omitted only from the analyses using that variable 1189 Analysis Data collected in the NHIS are obtained through a multistage complex sample design involving both strati®cation and clustering Moreover, the sampling frame for the NHIS is redesigned every 10 y, and there were two major changes to the sampling design beginning in 1995: an increased number of primary sampling locations to permit estimation at the state level, and oversampling of the Hispanic population in addition to the prior oversampling of the Black population Since this study pools data from the 1992 ± 1995 NHIS, the change in sampling frame introduces an additional level of complexity to the estimation procedures Extrapolating from NCHS guidelines for combining 1994 and 1995 data, all four years were concatenated into a single data set, with each year read independently and treated as a stratum Stratum and primary sampling unit variables were created for 1995 to be consistent with previous years Exploratory data analysis was carried out with STATA (Stata corporation, College Station, TX) Then, prevalence estimates and logistic and linear regression models were ®tted with the SUDAAN software package (Research Triangle Institute, Triangle Park, North Carolina), as recommended by the NCHS Using the method of generalized estimating equations, SUDAAN ®ts models and obtains standard errors that correctly account for multistage strati®ed sampling designs Regression parameter estimates are consistent and, if the covariance model is correctly speci®ed, ef®cient Even if the covariance model is misspeci®ed, however, the robust standard errors are correct, and hence con®dence intervals will have the correct coverage probabilities Although the International Journal of Obesity Asian American body mass index DS Lauderdale and PJ Rathouz 1190 population of main interest in this study is Asian Americans aged 18 ± 59, no NHIS observations were deleted since analysis of subsetted data may result in incorrectly computed standard errors.8 Instead, the subpopulation option (SUBPOPN) in SUDAAN procedure was used to target a subdomain from the full design database Results There were 7263 Asian Americans aged 18 ± 59 in the six major ethnic groups interviewed by the NHIS from 1992 ± 1995 (Table 1) By ethnic group, the numbers ranged from 816 Korean to 1940 Chinese As expected from the immigration history,9 a large majority of each group except the Japanese is foreign-born For each of the Korean, Vietnamese and Asian Indian groups, less than 10% and fewer than 100 persons were US-born Among the foreign-born, the subjects were distributed in roughly equal numbers into the three categories of years in the US (Table 2) Overall, 26% had been in the US fewer than y, 42% from to 15 y, and 33% for more than 15 y, and these percentages were generally similar across ethnic groups Figures and display box-and-whisker plots of BMI for men and women by race category All of the Asian American groups have lower median BMI than the White or Black groups There is, however, signi®cant variation among Asian American groups For men, the Japanese and Filipino have higher median BMI than the other groups, while for women it is the Filipino and Asian Indian groups with higher median BMI For both men and women, the Vietnamese have the lowest median BMI Table Distribution of six Asian American ethnic groups by birthplace for persons aged 18 ± 59, from the National Health Interview Survey, 1992 ± 1995, based on the unweighted sample Ethnicity Foreign-born Chinese Filipino Korean Vietnamese Japanese Asian Indian 1636 1399 747 888 342 1041 All Asian 6053 (83.3%) (84.3%) (83.1%) (91.5%) (98.1%) (41.1%) (95.9%) Native-born 304 284 69 17 491 45 Total (15.7%) (16.9%) (8.46%) (1.9%) (58.9%) (4.1%) 1940 1683 816 905 833 1086 1210 (16.7%) 7263 Table Distribution of six Asian American ethnic groups by years in the United States for foreign-born persons aged 18 ± 59, from the National Health Interview Survey, 1992 ± 1995, based on the unweighted sample Ethnicity Chinese Filipino Korean Vietnamese Japanese Asian Indian All Asian International Journal of Obesity `5y 437 256 169 258 142 305 (26%) (19%) (23%) (28%) (39%) (29%) 1567 (26%) ± 15 y 770 567 301 422 74 461 (44%) (40%) (39%) (45%) (20%) (43%) 2595 (42%) b 15 y 469 591 297 244 141 292 (29%) (42%) (38%) (27%) (41%) (28%) 2034 (33%) BMI, proportion overweight and proportion obese are all higher for US-born Asian Americans than for the foreign-born, and the effect is similar for men and women For Asian American men, adjusting for age and ethnicity, mean BMI is 1.31 kgam2 (95% CI 0.90 ± 1.72) lower for the foreign-born than the native-born For women, the difference in mean BMI is 1.14 kgam2 (95% CI 0.77 ± 1.51) Adjusting for age and ethnicity, the odds ratio of being overweight for the US-born compared to the foreign-born is 1.85 for men (95% CI 1.52 ± 2.26) and 1.94 for women (95% CI 1.46 ± 2.58), and the odds ratio for obese is 4.03 (95% CI 2.40 ± 6.78) for men and 3.51 (95% CI 1.74 ± 7.10) for women Among the foreign-born, the odds of being overweight or obese increases for Asian Americans with longer duration in the US (Table 3) Finally we examine whether the association between socioeconomic status, measured as family income (adjusted for family size), and BMI is the same for Asian Americans as it is for Blacks and Whites Using the same y of NHIS, we ®nd that family income is strongly inversely related to BMI for women The magnitude and direction of the association is very similar for White and Black women (Table 4) For White women, each $10,000 in income is associated with a BMI 0.55 kgam2 lower, or a total difference of 2.75 kgam2 between the highest and lowest income categories For Black women, BMI is 0.59 kgam2 lower per $10,000, a difference of 2.95 kgam2 between the highest and lowest income categories For White men, there is a very small inverse association between family income and BMI, 0.03 lower per $10,000 For Black men, there is a modest positive association, with BMI 0.07 higher for each $10,000 family income For US-born Asian American women, there is also a strong inverse association between BMI and income, although the magnitude of the association, 0.38 kgam2 for each $10,000 income, is somewhat smaller than for White and Black women There is no signi®cant association for US-born Asian American men, while the point estimate, 0.06, suggests a modest positive association and is similar to the point estimate for Black men The inverse association between income and BMI, however, is very weak and of marginal statistical signi®cance for foreign-born Asian American women, just 0.06 kgam2 per $10,000 income For foreign-born Asian American men, the positive association is somewhat greater than for other men, 0.11 per $10,000 income, and the trend is highly signi®cant When years in the US is also entered into the model, the positive association is attenuated Discussion We have found that BMI and the proportions overweight and obese are lower among each of the Asian American ethnic groups than the US population in Asian American body mass index DS Lauderdale and PJ Rathouz 1191 Figure Body mass index for men by race categories from the National Health Interview Survey, 1992 ± 1995 Each box corresponds to the interquartile range of the data, the 25th to 75th percentiles, and the line in the middle is the median The line above the box extends to the largest data point less than or equal to the 75th percentile plus 1.5 times the interquartile range The lower line is formed analogously Observations beyond this are individually plotted However, the ®gure does not display values greater than a BMI of 40 Figure Body mass index for women by race categories from the National Health Interview Survey, 1992 ± 1995 Each box corresponds to the interquartile range of the data, the 25th to 75th percentiles, and the line in the middle is the median The line above the box extends to the largest data point less than or equal to the 75th percentile plus 1.5 times the interquartile range The lower line is formed analogously Observations beyond this are individually plotted However, the ®gure does not display values greater than a BMI of 40 International Journal of Obesity Asian American body mass index DS Lauderdale and PJ Rathouz 1192 Table The effect of duration in US on odds of overweight and obese for foreign-born Asian American men and women, adjusted for age and ethnicity Overweight Years in US `5y ± 15 y b 15 y Obese Men Women Men Women OR (95% CI OR (95% CI) OR (95% CI) OR (95% CI) 0.62 (0.48 ± 0.80) 0.81 (0.67 ± 0.98) 1.00 (ref) 0.77 (0.55 ± 1.08) 0.98 (0.76 ± 1.26) 1.00 (ref) 0.63 (0.31 ± 1.28) 0.69 (0.40 ± 1.19) 1.00 (ref) 0.28 (0.11 ± 0.70) 0.50 (0.27 ± 0.92) 1.00 (ref) OR odds ratio CI con®dence interval Table The effect of household income on body mass index Men a White Blacka US born Asian Americanb Foreign born Asian Americanb a Women b coef®cient per $10,000 P-trend b coef®cient per $10,000 P-trend 0.3 0.07 0.06 0.11 0.008 0.016 0.664 0.004 0.55 0.59 0.38 0.06 ` 0.001 ` 0.001 0.002 0.046 Adjusted for age and family size Adjusted for age, family size and ethnicity b general However, the proportions vary by ethnicity and nativity US-born Asian Americans are signi®cantly more likely to be obese or overweight than the foreign-born Among the foreign-born, the number of years spent in the US is directly related to the risk of being overweight or obese For White and Black women in the US, there is a strong inverse association between BMI and economic status This association is also seen for US-born Asian American women, although the effect is weaker However, there is only weak evidence of an association between economic status and BMI for foreign-born Asian American women There is a positive association between BMI and economic status for foreign-born Asian American men This study has several methodologic limitations, the most signi®cant of which is that height and weight are self-reported Previous studies have assessed the validity of self-reported height and weight, and generally the level of agreement between self-report and measured height and weight has been found to be very high.10 However, there is evidence of modest systematic bias towards under-reporting weight and overreporting height.11,12 Such systematic mis-reporting would underestimate prevalence of overweight and obesity, which are calculated relative to ®xed BMI values Nonetheless, the prevalence of overweight found in the examination-based NHANES III data (collected 1988 ± 1994) for persons age 20 ± 741 is in close agreement with the prevalence calculated for 1992 ± 1995 NHIS for the same age range Speci®cally, Flegel et al reported from NHANES III a crude prevalence overweight of 60% for men and 51% for women From NHIS, the crude prevalence is 61% for International Journal of Obesity men and 51% for women Because the emphasis in this study is on comparisons and relative measures, systematic mis-reporting would be less of a limitation than mis-reporting which was related to variables under investigation, such as nativity or ethnicity While there is some evidence from the UK that quality of self-report may vary by demographic factors,13 we are unaware of studies which have investigated the validity of self-reported height and weight for the foreign-born or for Asian Americans Another limitation to this study is that conclusions regarding the effect of years since immigration are based on crosssectional rather than longitudinal data, possibly confounding acculturation with cohort effects related to year and age at immigration We cannot examine the effect of age at immigration in models which already include age and years in the US Finally, sample sizes for US-born Korean, Vietnamese and Asian Indian Americans are too small, even combining y of NHIS, to permit ethnicity-speci®c evaluations of nativity Our study is limited by the use of current household income (adjusted for family size) This measure may not re¯ect lifetime socioeconomic status (SES) as well for the foreign-born as the USborn The highest income category in the NHIS is $50,000 and higher, which fails to distinguish gradations among those with high incomes Despite these limitations, this study addresses a remarkable lack of information concerning the health of Asian Americans The immigration history, both the proportion US-born and the years of peak immigration, is different for each Asian American group Since both ethnicity and immigration status (nativity and duration in the US) may be related to Asian American body mass index DS Lauderdale and PJ Rathouz health, the ability to stratify Asian Americans by both ethnicity and immigration status is key to avoiding unmeasured confounding of one by the other The NHIS provides the opportunity to study a few health indicators for nationally-representative samples of the six largest Asian American ethnic groups With no oversampling of Asian Americans in the NHIS, individual years not provide adequate sample size for most analyses, nor the supplements, such as the 1992 Cancer Control Supplement, which are administered to a subsample of respondents We overcome this limitation by combining several years of data To our knowledge, only one previous project has investigated BMI and obesity in a multi-ethnic Asian American population with ethnicity-level detail Klatsky and Armstrong14 assessed cardiovascular risk factors among Asian Americans who volunteered for an examination at a northern California prepaid heath plan between 1978 and 1985 Their data, which include both mean BMI and proportion with BMI greater than 24.4 (overweight), are presented for Chinese, Japanese, Filipino and other Asian The proportions overweight are very similar to the proportions found in the NHIS data For Chinese, 27% of men and 13% of women were overweight; for Filipinos, 42% of men and 26% of women were overweight; and for Japanese, 38% of men and 18% of women were overweight They also found evidence of increased odds of being overweight for those born in the US, although the effect was only statistically signi®cant for men Further evidence of a nativity effect for BMI comes from the National Longitudinal Study of Adolescent Health Popkin and Udry found that US-born Asian American adolescents (in aggregate) were more than twice as likely to be overweight as the foreign-born adolescents.15 The association between SES and obesity has been studied in diverse cultures, with generally consistent ®ndings.16 In developed countries, there is a strong inverse association between SES and obesity for women and no consistent association for men In developing countries, by contrast, there is a positive association for both men and women; the higher prevalence of obesity among those with greater wealth likely re¯ects both greater access to food and a related cultural preference for physical evidence of such access.17 The SES effects we have found for USborn Asian Americans conform to expectations for persons in a developed country The lack of association for foreign-born women and the moderate positive association for foreign-born men suggest an effect intermediate in direction between those of developed and developing countries The health signi®cance of BMI in part derives from its correlation with adiposity Several previous studies have investigated whether the ability of BMI to predict percentage body fat varied by ethnicity Gallagher and others found that, while the association between BMI and adiposity did vary by age and sex, it did not differ between Black and White adults in a study conducted in New York City.18 Comparing BMI and percentage body fat for Asians and Whites, Wang and others did ®nd some differences.19 Asians, although mean BMI was lower, had higher percentage body fat and more upper-body subcutaneous fat In a meta-analysis which included data for three Asian groups, Deurenberg and others found that the percentage body fat was higher than predicted at low BMI levels for Chinese Body fat was underestimated across all BMI levels for Thais and Indonesians.20 In a study of women in Hawaii, Novotny and others found that Asian women had a greater percentage of body fat than did White women with the same BMI.21 One implication of such data is that the accepted cutoff values for overweight and obese may be less appropriate for Asian American populations in terms of their association with heart disease and proximal risk factors such as hypertension, glucose intolerance or lipid pro®les While the proportions overweight and obese are much lower for Asian Americans than other racial categories at present, the strong associations with birthplace and years since immigration suggest these proportions may increase signi®cantly as the demography of the population shifts, with increased duration of residence in the US and a higher proportion US-born The limited data concerning the correlation between BMI and adiposity suggest that health effects of BMI may differ for Asian Americans Higher levels of BMI could have a signi®cant impact on morbidity and mortality 1193 Acknowledgements We thank Ye Luo, PhD, for computer programming and Kate Pickett, PhD, for comments on the manuscript Data were presented in part at the annual meeting of the Society for Epidemiological Research (Baltimore MD, June 1999) References Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL Overweight and obesity in the United States: prevalence and trends, 1960 ± 1994 Int J Obes Relat Metab Disord 1998; 22: 39 ± 47 Peters KD, Kochanek KD, Murphy SL Deaths: ®nal data for 1996 National vital statistics reports; 47(9) National Center for Health Statistics: Hyattsville, MD, 1998 National Center for Health Statistics web site, 20 September 1999 http://www.cdc.gov/nchswww/products/catalogs/subject/ nhis/nhis.htm#description1 US Department of Health and Human Services National Center for Health Statistics, Hyattsville, MD National Health Interview Survey, 1992 Inter-university Consortium for Political and Social Research: Ann Arbor, MI, 1994 US Department of Health and Human Services, National Center for Health Statistics, Hyattsville, MD National Health Interview Survey, 1993 Inter-university Consortium for Political and Social Research: Ann Arbor, MI, 1995 US Department of Health and Human Services, National Center for Health Statistics, Hyattsville, MD National Health Interview Survey, 1994 Inter-university Consortium for Political and Social Research: Ann Arbor, MI, 1996 International Journal of Obesity Asian American body mass index DS Lauderdale and PJ Rathouz 1194 US Department of Health and Human Services, National Center for Health Statistics, Hyattsville, MD National Health Interview Survey, 1995 Inter-university Consortium for Political and Social Research: Ann Arbor, MI 1997 Graubard BI, Korn EL Survey inference for subpopulations Am J Epidemiol 1996; 144: 102 ± 106 Barringer HR, Gardner RW, Levin MJ Asians and Paci®c Islanders in the United States (The Population of the United States in the 1980s) NYL Russell Sage Foundation: New York, 1993 10 Willett W Nutritional Epidemiology, 2nd edn New York: Oxford University Press, 1998, p 247 11 Palta M, Prineas RJ, Berman R, Hannan P Comparison of self-reported and measured height and weight Am J Epidemiol 1982; 115: 223 ± 230 12 Stewart AW, Jackson RT, Ford MA, Beaglehole R Underestimation of relative weight by use of self-reported height and weight Am J Epidemiol 1987; 125: 122 ± 126 13 Hill A, Roberts J Body mass index: a comparison between self-reported and measured height and weight J Public Health Med 1998; 20: 206 ± 210 14 Klatsky AL, Armstrong MA Cardiovascular risk factors among Asian Americans living in northern California Am J Public Health 1991; 81: 1423 ± 1428 International Journal of Obesity 15 Popkin BM, Udry JR Adolescent obesity increases signi®cantly in second and third generation U.S Immigrants: the National Longitudinal Study of Adolescent Health J Nutr 1998; 128: 701 ± 706 16 Sobal J, Stunkard AJ Socioeconomic status and obesity: a review of the literature Psychol Bull 1989; 105: 260 ± 275 17 Stunkard AJ Socioeconomic status and obesity Ciba Found Symp 1996; 201: 174 ± 187 18 Gallagher D, Visser M, SepuÂlveda D, Pierson RN, Harris T, Heyms®eld SB How useful is body mass index for comparison of body fatness across age, sex and ethnic groups? Am J Epidemiol 1996; 143: 228 ± 239 19 Wang J, Thornton JC, Russell M, Burastero S, Heyms®eld S, Pierson RN Jr Asians have lower body mass index (BMI) but higher percent body fat than Whites: comparisons of anthropometric measurements Am J Clin Nutr 1994; 60: 23 ± 28 20 Deurenberg P, Yap M, van Staveren WA van Body mass index and percent body fat: a meta analysis among different ethnic groups Int J Obes Relat Metab Disord 1998; 22: 1164 ± 1171 21 Novotny R, Davis J, Ross P, Wasnich R Adiposity and blood pressure in a multiethnic population of women in Hawaii Ethnic Health 1998; 3: 167 ± 173