Accuracy of Current Body Mass Index Obesity Classification for White, Black, and Hispanic Reproductive-Age Women : Obstetrics & Gynecology

Secondary Logo

Journal Logo

Original Research

Accuracy of Current Body Mass Index Obesity Classification for White, Black, and Hispanic Reproductive-Age Women

Rahman, Mahbubur MD, PhD; Berenson, Abbey B. MD, MMS

Author Information
Obstetrics & Gynecology 115(5):p 982-988, May 2010. | DOI: 10.1097/AOG.0b013e3181da9423
  • Free


To compare the National Institutes of Health's (NIH) body mass index (BMI)-based classification to identify obesity in comparison with the World Health Organization (WHO), which uses percent body fat, among white, black, and Hispanic reproductive-aged women.


Body weight, height, BMI, and percent body fat (dual-energy X-ray absorptiometry generated) were determined for 555 healthy adult women aged 20–33 years (mean±standard deviation 26.5±4.0 years). Diagnostic accuracy of the NIH-based obesity definition (BMI of 30 kg/m2 or higher) was determined using the WHO criterion standard (percent body fat greater than 35%).


Obesity as defined by the NIH (BMI 30 kg/m2 or higher) and by WHO (percent body fat greater than 35%) classified 205 (36.9%) and 350 (63.1%) of the women as obese, respectively. The NIH-defined obesity cutoff values had 47.8%, 75.0%, and 53.9% sensitivity in white, black and Hispanic, women, respectively. White and Hispanic women had 2.9% greater percent body fat than black women for a given BMI. Receiver operating characteristics curves analyses showed that the respective sensitivities improved to 85.6%, 81.3%, and 83.2%, and that 311 women (56.0%) were classified as obese as a whole when race or ethnic-specific BMI cutoff values driven by our data (BMI at or above 25.5, 28.7, and 26.2 kg/m2 for white, black, and Hispanic women, respectively) were used to detect percent body fat–defined obesity.


Current BMI cutoff values recommended by the NIH failed to identify nearly half of reproductive-aged women who met the criteria for obesity by percent body fat. Using race or ethnic-specific BMI cutoff values would more accurately identify obesity in this population than the existing classification system.



Obesity is fast approaching tobacco as the top preventable cause of death in the United States.1,2 It exerts adverse effects on health through multiple organ systems of the body and reduces life expectancy.3,4 In 2000, the World Health Organization (WHO) labeled obesity as the most blatantly visible, but most neglected, public-health problem worldwide.4

Body mass index (BMI) has been used to identify who is overweight or obese for decades. Further, this method has been used consistently in most clinical and behavioral studies and is the key measure to assess weight loss programs.5–7 However, arguments against using a single universal cutoff value worldwide to define obesity are widespread.8–19 For example, WHO recommended BMI values of 23 kg/m2 and 25 kg/m2 as the cutoff points for overweight and obesity for Asians, respectively.20 Several Asian countries have also recommended using lower cutoff values to identify overweight and obesity.12–19

The accuracy of the National Institutes of Health's (NIH) definition of obesity21 as those with a BMI 30 kg/m2 or greater compared with WHO's criterion standard for obesity (percent body fat greater than 25% in men and 35% in women)6 has been recently challenged as well.9–11 For example, Romero-Corral et al9 observed that this obesity cutoff value was too high for U.S. adult men and women and missed more than half of the obese individuals. Evans et al10 and Blew et al11 made similar observations in their studies of postmenopausal women. Moreover, white and Hispanic women demonstrated significantly higher percent body fat for a given BMI than black women.22 Thus, data on accurate BMI cutoff values based on percent body fat to define obesity in women of different races/ethnicities are needed to better assess an individual's risk of obesity related morbidity and mortality. This is especially important to determine in reproductive-aged women as they are more likely to be obese than similarly aged men,23 which may make them vulnerable to cardiovascular disease risk factors and other obesity related diseases. To obtain this critical information, we investigated the accuracy of the currently used BMI cutoff value in comparison with WHO's percent body fat-based criterion standard to identify obesity among reproductive-aged white, black, and Hispanic women.


We conducted secondary analyses of data gathered to examine the effects of hormonal contraception on bone mineral density (BMD). As a part of larger study, 805 healthy, reproductive-aged non-Hispanic white, non-Hispanic black, and Hispanic women aged between 16 and 33 years were recruited between October 9, 2001, and September 14, 2004. Determination of race was based on self-identification. The methods for the larger study are reported in detail elsewhere.24 Briefly, recruitment was planned to achieve a sample that was balanced by race/ethnicity, age group (16–24 years and 25–33 years), and contraceptive method. We excluded women from participation in the larger study if they weighed more than 300 pounds (due to safety limitations of the dual-energy X-ray absorptiometry machine), were not eligible to receive hormonal contraceptive containing estrogen, wished to become pregnant in 3 years or less, had received oral contraceptive pills or depot medroxyprogesterone acetate in the last 3 or 6 months, respectively, had used medications or had a medical condition known to affect BMD, or had a dietary intake known or suspected to be high in isoflavones. In addition, to avoid including women with a possible medical condition that could affect their BMD, those with abnormal serum levels of vitamin D, thyroid stimulating hormone, or liver function tests were excluded. We obtained child assent and parental permission for participants under 18 years of age and written informed consent from all others. Of the 805 women who consented to participate, 92 failed additional screening tests and five were removed from the study after the baseline bone scan due to results indicative of osteoporosis (T-score -2.5 or less). Those excluded (n=97) did not differ from women included in the analyses (n=708) by age, but they were more likely to be black (22% compared with 10% Hispanic and 2% white, P<.001) and to have a higher BMI (28.4 kg/m2 compared with 24.4 kg/m,2P<.001). Data reported in this paper were collected at the baseline visit for the longitudinal study. The study received approval from the Institutional Review Board at the University of Texas Medical Branch at Galveston.

Our analysis was limited to adults aged 20 years or older who had percent body fat data available at baseline. We excluded 146 women who were younger than 20 years because the obesity definition of this age group differs from that of the adult population. In addition, seven adult women did not have percent body fat data available at baseline. Thus, our final analyses were based on 555 healthy adult women aged 20–33 years. In the present analyses, we included data on age, height, weight, BMI, and body composition measures collected at baseline in the clinic for the larger study. We obtained body composition measures using dual-energy X-ray absorptiometry (Hologic QDR 4500W densitometer, Marlborough, MA). The coefficient of variation of percent body fat among adults for repeated measurements ranged from 2.8% to 3.3%11,25; the interclass correlation coefficient was 0.994.25 We calculated percent body fat by dual-energy X-ray absorptiometry using the following formula: [fat mass (g)/{fat mass (g)+lean mass (g)+total bone mineral content (g)}]×100. We measured standing height and weight with women wearing light indoor clothing and no shoes. Standing height was measured to the nearest 0.1 centimeters using a stadiometer. Body weight was measured using a digital scale accurate to the nearest 0.1 kg. Body mass index was calculated as weight (kg) divided by the square of the height (m).

We performed univariable comparisons among three race/ethnic groups using one-way analysis of variance with Bonferroni corrections. We used multiple regression analyses to examine the relationship between percent body fat and BMI and the effect of age and race/ethnicity on this relationship. We also used nonlinear terms to estimate whether the relationship between percent body fat and BMI is linear or curvilinear.

The sensitivity and specificity of the currently used BMI cutoff value (30 kg/m2 or greater) to define obesity were compared with the criterion standard definition of obesity in women using percent fat (greater than 35%) proposed by the WHO.6 Corresponding 95% confidence intervals (CIs) were computed using exact methods.26 We also examined racial differences with regard to BMI accuracy. In addition, we evaluated appropriateness of various BMI values for classifying obesity using receiver operating characteristics curves. Analysis using receiver operating characteristics curves generated sensitivity and specificity rates corresponding to various BMI cutoff values. We identified race/ethnicity specific optimum BMI cutoff levels by maximizing the classification accuracy after calculating overall performance (sensitivity plus specificity). In addition, we determined the area under the curve statistic, which is a global measure of the overall diagnostic accuracy of BMI, to determine clinical status (obesity).27,28 We also examined the accuracy of our data-driven race/ethnicity-specific BMI cutoff values using the same criterion standard of obesity. Improvement of sensitivity with race/ethnic specific BMI cutoff values were compared using McNemar's χ2 tests. All analyses were performed using STATA 10 (Stata Corporation, College Station, TX).


At baseline, the total sample included 189 white, 159 black, and 207 Hispanic women with a mean age of 26.5 (±4.0) years. Whites were older than blacks but had lower percent body fat than Hispanics (Table 1). Black women were more likely to have higher values for body weight and BMI relative to white and Hispanic women, while height was similar among blacks and whites.

Table 1:
Descriptive Characteristics of the Women in the Study Population by Race/Ethnicity

Regression analysis between percent body fat and BMI along with significant BMI2 terms showed that the relationship is curvilinear (Fig. 1). The model building strategy showed that the model with BMI, BMI-squared, and race×BMI interaction had the highest explained variance (R2 76.2%) although the model with BMI, BMI2, and race showed almost the same variance (R2 75.8%). No other predictor added additional variance. Age and age×BMI interaction were not predictive of percent body fat and did not add any extra variance to the model. The following equation represents the final model without the interaction term. Percent body fat can be estimated in black women by putting white=0 and Hispanic=0 in the equation below.

Fig. 1.:
Relationship between body mass index and percent body fat measured by dual-energy X-ray absorptiometry in reproductive-aged women (aged 20–33 years; n=555).Rahman. Accuracy of BMI Cutoff Value To Define Obesity. Obstet Gynecol 2010.

This equation shows that for a given BMI, white and Hispanic women will have 2.9% higher percent body fat than black women, which implies that the BMI cutoff value equivalent to 35% body fat will differ in black, white, and Hispanic women. Although the race×BMI interaction was significant for Hispanic women (regression coefficient=−0.181, standard error 0.059, P=.002), the additional contribution was minimal as explained variance improved very little. The interpretation of the significant race×BMI interaction for Hispanic women is as follows: for a given BMI, the difference in percent body fat between black and Hispanic women narrowed with increased BMI.

Of the 555 reproductive-aged women we examined, the BMI cutoff value of 30 kg/m2 suggested by the NIH identified 205 women as obese (percent body fat greater than 35%). Although the NIH-recommended BMI cutoff point for obesity had high specificity (96.8–100%) in different races or ethnicities, the sensitivity was relatively low (47.8–75.0%) (Table 2). The overall sensitivity and specificity of this cutoff value was 57.7% (95% CI 52.5–62.8%) and 98.5% (95% CI 95.8–99.5%), respectively. The sensitivity was significantly higher in black women (75.0%, 95% CI 65.5–82.6%) compared with white (47.8%, 95% CI 38.7–57.0%) and Hispanic women (53.9%, 95% CI 45.7–61.8%, P<.001 both for black compared with white and black compared with Hispanic). It did not differ between white and Hispanic women (P=.335). Specificities (96.7–100%) were statistically similar in different races/ethnicities.

Table 2:
Accuracy of Body Mass Index Cutoff Value to Define Obesity in White, Black and Hispanic Reproductive-Aged Women Based on National Institutes of Health Guideline and the Current Study

Receiver operating characteristics curve analysis based on our study data showed that the greatest accuracy to identify obesity using 35% body fat corresponded to BMI values of 25.5, 28.7 and 26.2 kg/m2 in white, black, and Hispanic women, respectively. The higher BMI cutoff value for black than white and Hispanic women support the finding based on our regression analysis that for a given BMI, black women have a lower percent body fat than the other two groups of women. The respective area under the curve was 0.967 (95% CI 0.946–0.988), 0.946 (95% CI 0.916–0.977), and 0.927 (95% CI 0.894–0.960). The sensitivity and overall performance of the race/ethnic specific BMI values generated by the current study were improved over those recommended by the NIH, particularly in white and Hispanic women (Table 2). Of the 555 women, race/ethnic specific BMI cutoff values identified 292 obese women out of 350 actually obese women (90 more than that identified with the NIH-based cutoff value). The sensitivity increased to 85.6% from 47.8% (P<.001) in white, 81.3% from 75.0% in black (P=.031), and 83.2% from 57.7% (P<.001) in Hispanic women. The overall sensitivity was significantly improved from 57.7% to 83.4% (P<.001). A greater improvement in sensitivity was observed in white (37.8%) and Hispanic women (29.3%) than among black women (5.3%).

Table 3 shows the obesity rates and 95% CIs based on percent body fat, the NIH definition, and our data using race or ethnic-specific BMI cutoff values. Of the 555 women we examined, 205 (36.9%) were classified as obese based on NIH guidelines (BMI 30 kg/m2 or greater). The obesity rate in black (46.5%) and Hispanic women (37.7%) was significantly higher than that observed in white women (28.0%). However, WHO criterion (percent body fat greater than 35%) classified 350 women as obese (63.1% of total). When percent body fat was used, the obesity rate was highest in Hispanic women (69.1%) and the rates were similar in white (58.7%) and black women (60.4%). When race/ethnic specific BMI cutoff values were used, 311 women were labeled as obese (56.0%) with 52.9% of whites, 52.8% of blacks and 61.4% of Hispanics classified as obese.

Table 3:
Obesity Rate in Our Study Population Based on World Health Organization, National Institutes of Health, and Race/Ethnicity-Specific Body Mass Index Cutoff Values


Our study shows that the currently used BMI cutoff value for obesity recommended by the NIH (BMI 30 kg/m2 or greater) may be too high and does not reflect actual body fatness by race or ethnicity among reproductive-aged women. Use of this definition resulted in the misclassification of many obese women when compared with use of WHO criteria despite having very good specificity. Similar to our findings, Romero-Corral et al9 also observed that the NIH-based BMI cutoff value to define obesity had low sensitivity (49%) in U.S. women aged 20–80 years. Evans et al10 found similar results in white (47.1%) and black (52.6%) postmenopausal women. Several smaller studies have shown similar results.29,30 Blew et al11 observed even lower sensitivity (25.6%) when this definition was used in mostly white postmenopausal women. Together, these studies provide evidence that the NIH-based BMI cutoff value is not accurate enough to identify obesity among a large number of adult women residing in the United States.

Our data-driven race or ethnic-specific BMI cutoff values to define obesity agree with those of several other U.S. studies that included diverse populations.9–11 Evans et al10 identified obesity as those with BMI values 26.9 kg/m2 or greater among white women and 28.4 kg/m2 or greater among black women while our study showed BMI values 25.5 or greater, 28.7, and 26.2 kg/m2 for white, black, and Hispanic women, respectively. However, Blew et al11 observed even lower BMI cutoff values (24.9 kg/m2) in mostly white postmenopausal women. Romero-Corral et al9 found that the cut off value should be 25.5 kg/m2 among multiethnic women. Moreover, sensitivities of the revised BMI cutoff values generated in our study are also similar to previously published studies.9–11 This suggests that the BMI cutoff value should not only be lower than the value currently used, but it also should differ by race or ethnicity.

The difference between actual and observed obesity rates in whites (59% compared with 28%) and Hispanic (69% compared with 38%) women could be a threat to the success of obesity awareness and programs in the United States. The NIH-based obesity rate calculations, which show that black women have the highest obesity rate, were not supported by percent body fat data in this study. In contrast, Hispanic women had the highest obesity rates based on percent body fat classified obesity. Thus, there is a need to organize the obesity prevention programs targeting all three race or ethnic groups equally with a special emphasis on Hispanic women. It is a serious public health concern that more than two-thirds of Hispanic reproductive-aged women are obese.

Moreover, obesity rates based on NIH guidelines in white and Hispanic women are severely underestimated, which needs to be corrected. The current BMI cutoff value results in about half of women with actual obesity (greater than 35% body fat) being labeled as normal or overweight. Thus, the opportunity to reduce body weight by appropriate intervention in this group of people is missed. It is possible that the improvement in sensitivity in white and Hispanic women using race/ethnic specific BMI cutoff values will result in labeling a few women as obese who are not, causing them additional stress. However, considering that fewer women will be misclassified by the revised cutoff values and the myriad public health implications of obesity, any potential harm would be outweighed by the benefit of identifying an increased number of actually obese women.

Our finding that the NIH classified obesity rate was 36.9% among reproductive-aged women is consistent with population based reports of its prevalence in 20–39-year-old women (29.1%). According to the National Health and Nutrition Examination Survey 1999–2002 data,23 24.9% of non-Hispanic whites, 46.6% of non-Hispanic blacks, and 31.2% of Hispanic women between the age of 20 and 39 years were obese (BMI 30 kg/m2 or greater) compared with 28.0%, 46.5% and 37.7%, respectively, in the current study. The similarity of obesity rates between the current study and the National Health and Nutrition Examination Survey–based study increases the external validity of our study results.

Published studies show that the influence of race/ethnicity on the relationship between BMI and percent body fat may not be consistent.10,22,31,32 For example, Fernandez et al30 did not observe any difference in percent body fat between white and black postmenopausal women for a given BMI while Evans et al10 observed that white women had 1% higher percent body fat than black postmenopausal women. In contrast, our study showed a difference of almost 3%. Aloia et al32 also found that black women at the same percent body fat had significantly higher BMI than white perimenopausal women. Differences in age distribution could be the reason for these discrepancies. However, further studies on age-related changes in percent body fat based on 10-year increments by race/ethnicity are needed to shed more light on this issue.

This study has several limitations. First, we examined diagnostic performance of BMI in only 20–33-year-old women, so we don't know whether similar findings would be observed in other age groups. However, similar findings in studies of postmenopausal white and black women10 suggest that similar race/ethnic specific cutoff values might work for other age groups of women residing in the United States. Second, our study is not based on a random sample, and thus, our sample may not be representative of all white, black, and Hispanic women. However, similar obesity rates in the current study and National Health and Nutrition Examination Survey–based study23 increase the external validity of the study. Third, the Hispanic women in our study were predominantly of Mexican descent, so extension of these data to Hispanic women of other origins should be done with caution. Finally, use of a single site could limit the generalizability of our findings.

In conclusion, our findings show that the currently accepted BMI cutoff value to identify obesity is too high for many reproductive-aged women residing in the United States. This suggests that women whose BMI is between 25 and 29.9 kg/m2 (in addition to 30 kg/m2 or greater), may require additional counseling on how to reduce their body weight to avoid obesity-related morbidity. Furthermore our data suggest that race-specific BMI classifications need to be established to more accurately identify reproductive-aged women who are obese so they can be counseled appropriately. Substantial increases in sensitivity in white (38% increase) and Hispanic women (30% increase) make the BMI cutoff values generated in this study reasonable to consider for reproductive-aged women. As a validation measure, however, these proposed criteria and their relationship to cardiovascular risk factors need to be further examined using an independent nationally representative sample.


1. Mokdad AH, Marks JS, Stroup DF, Gerberding JL. Correction: actual causes of death in the United States, 2000. JAMA 2005;293:293–4.
2. McGinnis JM, Foege WH. Actual causes of death in the United States. JAMA 1993;270:2207–12.
3. Haslam DW, James WP. Obesity. Lancet 2005;366:1197–209.
4. World Health Organization. Obesity: preventing and managing the global epidemic. WHO Technical Report Series 2000;894:1–252.
5. National Heart, Lung, and Blood Institute. Clinical Guidelines on the Identification, and Treatment of Overweight and Obesity in Adults: The Evidence Report. NIH Publication no. 98–4083. Washington, DC: National Institutes of Health; September 1998.
6. World Health Organization. Physical status: the use and interpretation of anthropometry. Report of a WHO expert committee. World Health Organ Tech Rep Ser 1995;8541–52.
7. Seidell JC, Kahn HS, Williamson DF, Lissner L, Valdez R. Report from a Centers for Disease Control and Prevention Workshop on use of adult anthropometry for public health and primary health care. Am J Clin Nutr 2001;73:123–6.
8. Razak F, Anand SS, Shannon H, Vuksan V, Davis B, Jacobs R, Teo KK, et al. Defining obesity cut points in a multiethnic population. Circulation 2007;115:2111–8.
9. Romero-Corral A, Somers VK, Sierra-Johnson J, Thomas RJ, Collazo-Clavell ML, Korinek J, et al. Accuracy of body mass index in diagnosing obesity in the adult general population. Int J Obes 2008;32:959–66.
10. Evans EM, Rowe DA, Racette SB, Ross KM, McAuley E. Is the current BMI obesity classification appropriate for black and white postmenopausal women? Int J Obes 2006;30:837–43.
11. Blew RM, Sardinha LB, Milliken LA, Teixeira PJ, Going SB, Ferreira DL, et al. Assessing the validity of body mass index standards in early postmenopausal women. Obes Res 2002;10:799–808.
12. Zhou BF, Cooperative Meta-Analysis Group of the Working Group on Obesity in China. Predictive values of body mass index and waist circumference for risk factors of certain related diseases in Chinese adults–study on optimal cut-off points of body mass index and waist circumference in Chinese adults. Biomed Environ Sci 2002;15:83–96.
13. Deurenberg-Yap M, Chew SK, Deurenberg P. Elevated body fat percentage and cardiovascular risks at low body mass index levels among Singaporean Chinese, Malays and Indians. Obes Rev 2002;3:209–15.
14. Stevens J, Nowicki EM. Body mass index and mortality in Asian populations: implications for obesity cut-points. Nutr Rev 2003;61:104–7.
15. Ko GT, Tang J, Chan JC, Sung R, Wu MM, Wai HP, et al. Lower BMI cut-off value to define obesity in Hong Kong Chinese: an analysis based on body fat assessment by bioelectrical impedance. Br J Nutr 2001;85:239–42.
16. He M, Tan KC, Li ET, Kung AW. Body fat determination by dual energy X-ray absorptiometry and its relation to body mass index and waist circumference in Hong Kong Chinese. Int J Obes Relat Metab Disord 2001;25:748–52.
17. Chang CJ, Wu CH, Chang CS, Yao WJ, Yang YC, Wu JS, et al. Low body mass index but high percent body fat in Taiwanese subjects: implications of obesity cutoffs. Int J Obes Relat Metab Disord 2003;27:253–9.
18. Kanazawa M, Yoshiike N, Osaka T, Numba Y, Zimmet P, Inoue S. “Criteria and classification of obesity in Japan and Asia-Oceania.” Asia Pac J Clin Nutr 2002;11(suppl 8):S732–7.
19. Chen YM, Ho SC, Lam SS, Chan SS. Validity of body mass index and waist circumference in the classification of obesity as compared to percent body fat in Chinese middle-aged women. Int J Obes 2006;30:918–25.
20. WHO/IASO/IOTF. The Asia-Pacific Perspective: Redefining Obesity and Its Treatment. Melbourne (Australia): Health Communications Australia 2000.
21. National Heart Lung and Blood Institute. Clinical guidelines on the identification, evaluation and treatment of overweight and obesity in adults: the evidence report. National Institutes of Health. Obes Res 1998;6:51S–209S.
22. Rahman M, Temple JR, Breitkopf CR, Berenson AB. Racial differences in body fat distribution among reproductive-aged women. Metabolism 2009;58:1329–37.
23. Hedley AA, Ogden CL, Johnson CL, Carroll MD, Curtin LR, Flegal KM. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA 2004;291:2847–50.
24. Berenson AB, Rahman M, Breitkopf CR, Bi LX. Effects of depot medroxyprogesterone acetate 20-microgram oral contraceptives on bone mineral density. Obstet Gynecol 2008;112:788–99.
25. Russell-Aulet M, Wang J, Thornton J, Pierson RN Jr. Comparison of dual-photon absorptiometry systems for total-body bone and soft tissue measurements: dual-energy X-rays versus gadolinium 153. J Bone Miner Res 1991;6:411–5.
26. Collett D. Modelling binary data. London (UK): Chapman & Hall; 1991.
27. Hanley JA, McNeil BJ. The meaning and use of area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29–36.
28. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental tool in clinical medicine. Clini Chem 1993;39:561–77.
29. Romero-Corral A, Somers VK, Sierra-Johnson J, Jensen MD, Thomas RJ, Squires RW, et al. Diagnostic performance of body mass index to detect obesity in patients with coronary artery disease. Eur Heart J 2007;28:2087–93.
30. Smalley KJ, Knerr AN, Kendrick ZV, Colliver JA, Owen OE. Reassessment of body mass indices. Am J Clin Nutr 1990;52:405–8.
31. Fernández JR, Heo M, Heymsfield SB, Pierson RN Jr, Pi-Sunyer FX, Wang ZM, et al. Is percentage body fat differentially related to body mass index in Hispanic Americans, African Americans, and European Americans? Am J Clin Nutr 2003;77:71–5.
32. Aloia JF, Vaswani A, Mikhail M, Flaster ER. Body composition by dual-energy X-ray absorptiometry in black compared with white women. Osteoporos Int 1999;10:114–9.
© 2010 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.