Racial differences in bone density and fracture risk in the United States


  • Marcella D. WALKER,

    1. Division of Endocrinology, Metabolic Bone Diseases Unit, College of Physicians & Surgeons, Columbia University, New York, New York 10032
    Search for more papers by this author

    1. Division of Endocrinology, Metabolic Bone Diseases Unit, College of Physicians & Surgeons, Columbia University, New York, New York 10032
    Search for more papers by this author

: Marcella D. Walker, MD, Assistant Professor of Clinical Medicine, 630 West 168th Street, PH8 West – 864, New York, NY 10032. Email: mad2037@columbia.edu


Many studies indicate racial differences in areal bone mineral density (aBMD). African-Americans tend to have the greatest aBMD, followed by Caucasian, Hispanic and Native Americans. Asian Americans, on average, have the lowest aBMD. Differences in aBMD are attenuated when adjustments are made for bone size, weight and lifestyle factors. While many studies have focused on racial differences in bone density, it is important to consider fracture risk since it is the clinically meaningful outcome and is determined by many factors other than bone density. African- and Asian-Americans have a relatively low risk of fracture compared to other groups, despite drastically different bone density, as measured by dual X-ray absorptiometry (DXA). Factors that cannot be completely evaluated by DXA, such as bone size, volumetric density, bone microarchitecture and turnover may be important in explaining disparate fracture rates. Newer technologies such as central quantitative computed tomography (CT) and high-resolution peripheral CT promise to contribute important knowledge in this regard. Despite recent advances, more fracture risk data is needed for non-Caucasian groups to assist in the development of relevant screening and treatment guidelines.

Differences in fracture risk among races have largely been attributed to differences in bone mineral density (BMD). The data are most abundant between African American and Caucasian individuals, with limited information in other racial and ethnic groups. Newer data suggest that other aspects of bone quality, apart from areal BMD, are important in explaining racial differences in fracture risk.


Available data indicate that African American men and women have higher bone density as measured by dual energy X-ray absorptiometry (DXA) than other racial groups, including Caucasian, Asian, Hispanic and Native Americans.1 Since DXA is a two-dimensional measurement (g/cm2), we will refer to such measurements as areal BMD (aBMD). The greater BMD among African Americans is attenuated, but still persists, when the data are adjusted for weight, bone size (by calculating bone mineral apparent density) and other covariates, such as physical activity, calcium intake, smoking and alcohol use.2,3 The higher bone mass observed in African American individuals may be due to more favourable calcium absorption and retention,4 along with differences in other genetic factors that govern acquisition of peak bone mass. While African Americans tend to have lower 25-hydroxyvitamin D levels5 and similar6 or higher5 parathyroid hormone (PTH) levels, sensitivity to the resorptive effects of PTH may be reduced.7 Histomorphometric studies also indicate that premenopausal African American women have a lower rate of mineralized matrix apposition and longer formation period than Caucasians. These mechanisms may lead to greater deposition of bone mineral,6 and perhaps, bone of higher quality.

Most studies indicate that Hispanic Americans have similar3,8,9 or slightly higher10 bone density as compared to Caucasians. One recent study noted slightly lower bone density in a group of Hispanic American women of different ethnic backgrounds compared to Caucasians. Unlike some prior studies, bone density values were adjusted for risk factor covariates, which may explain the disparity.1 The majority of investigations in the Hispanic community have focused on Mexican Americans, with much less data available in other Hispanic groups. We have studied a female Hispanic population from the Dominican Republic residing in New York City. This group has higher aBMD, unadjusted for weight, as compared to Caucasian reference values at the femoral neck and total hip. Areal BMD in Dominican Americans may also be higher compared to Mexican American reference values at the total hip across some age ranges.11

Studies in Asian populations have demonstrated lower aBMD in comparison to Caucasians12–14 and other racial groups.1,2 Weight or bone size explains the majority of these differences.2,15 Moreover, some data suggest that when adjustments are made for differences in bone size and other covariates,2,16 BMD is actually greater among individuals of Chinese descent.

Bone density data in Native Americans is incomplete but two recent studies indicate similar bone density in Native Americans compared to Caucasian women.1,17,18 The available information on unadjusted areal bone density can be recapitulated by a hierarchy in the following manner: African Americans > Caucasian, Hispanic and Native Americans > Asian Americans.


Although differences in bone density are notable, data indicating differences in fracture risk are perhaps of more significance since fracture is the clinically relevant end point. Moreover, fracture risk is determined by elements of bone quality that are not captured completely in the bone density measurement. Many studies indicate reduced rates of hip19–21 and other fractures22 in African Americans as compared to Caucasians. While the greater bone density among African Americans is undoubtedly relevant, examination of other groups suggest more to this observation. For example, despite low bone density, Asian23 and Asian-American21,24 individuals have relatively low rates of hip fracture, while vertebral fracture rates in most studies are similar to that of Caucasian populations.25,26 Recent data from the National Osteoporosis Risk Factor Assessment (NORA) study also indicate that Asian and African American women, two groups with distinctly different bone mineral densities, share a relatively low risk of forearm osteoporotic fractures compared to Caucasian, Hispanic and Native American women.1

As already alluded to, other factors besides areal bone density, may influence fracture rates. A number of hypotheses to account for the lower risk of osteoporotic fracture in Asian and African Americans have been offered. Faulkner et al. demonstrated that hip axis length predicts hip fracture risk independently of height, weight, age, and femoral neck BMD in Caucasian women.27 Some later studies supported hip axis length as a risk factor for only certain types of hip fractures and others have found no correlation whatsoever between hip axis length and fracture risk.28–30 Subsequently, Cummings did show that on average Asians and African Americans have shorter hip axis length than Caucasians.31 However, a recent study found no racial differences in hip axis length adjusted for hip girth when non-Hispanic white, African-American, and Mexican-American women were compared.32 Differences in hip axis length may account, in part, for the variation in hip fracture incidence between Asians and Caucasians.

Larger bone size may also afford protection from fracture in African Americans.33 A larger cross-sectional bone diameter increases strength (i.e. resistance to bending) by placing more mass further from the axis of the long bone.34,35 Asian individuals are thought generally to have smaller bone size, which would increase the risk of fracture, for a given bone density. However, small bone size also leads to an artificially lower areal bone density when measured by DXA. Thus when DXA is used to measure BMD,15,29 small bone size could mask a higher ‘true’ bone density, a value that can only be determined by a measurement of true bone density, namely a volumetric quantity (g/cm3). A recent multiracial study of men aged ≥ 65 indicated thicker femoral neck and shaft cortices and higher femoral neck trabecular volumentric BMD in Asian and Black men compared with Whites.36 Femoral neck cross-sectional area was greater in Black compared to White men, but similar between Whites and Asians. Shaft cross-sectional area was smaller in Asian compared to White men. None of the femoral neck/shaft dimensions or BMD measures differed significantly between the Hispanic and White men. Such differences in Black and Asian men may be advantageous with regard to risk of hip fracture.

Additionally, protection from fracture may be afforded by racial differences in other ‘bone qualities’, besides bone size and volumetric bone density, such as the bone microarchitecture, mineralization density, and collagen cross-linking, although the limited available data in African Americans thus far does not necessarily support these possibilities.6,37 Other potential explanations for differences in fracture risk include fewer falls38 and better muscle strength. Slower rates of age-related bone loss39,40 or remodeling41 might also be protective. In a longitudinal investigation, Cauley et al. have demonstrated a slower rate of bone loss at the total hip and femoral neck in African American compared to Caucasian postmenopausal women.39 Comparisons in older Black and White men are also suggestive of slower rates of bone loss among Black men.40 Cross-sectional findings in Asian and Hispanic American individuals suggest that differences in rates of bone loss may also exist,11–14,42 but need to be confirmed with longitudinal studies.


Referent database

Because data regarding the relationship between BMD and fracture risk in non-Caucasians are limited, it remains unclear whether a Caucasian or ethnic database should be used for fracture risk assessment in non-Caucasian individuals within the United States. Some advocate a single Caucasian referent standard to which all racial groups and even both sexes should be related.43,44 Others feel that because of inherent differences in skeletal qualities among races and the sexes, separate referent databases should be established with which race- and sex-specific risk can be determined. With data from the third National Health and Nutrition Examination Survey (NHANES III), ethnic-specific hip BMD reference databases were created for Caucasian, Hispanic (Mexican Americans), and African-American male and female populations in the United States. Unfortunately, Pacific Islanders, Asian and Native American individuals were not included in the NHANES III bone density assessment; BMD databases are not therefore commercially available for these other populations in the United States. While other countries have created and utilize BMD databases for their populations, evaluation of fracture risk with these databases may be difficult because environmental factors and acculturation are also likely to cause differences in bone density and perhaps fracture risk.45,46

DXA manufacturers have adopted disparate methods to the reference databases utilized for non-Caucasians. Some use race-specific databases for the calculation of both T- and Z-scores and others use race-specific databases for only Z-scores or women.44 Osteoporosis prevalence rates will obviously differ if an ethnic group's T- or Z-scores are calculated from a Caucasian or an ethnic-specific database. Likewise, the T-score, and thus the diagnosis of osteoporosis for a given non-Caucasian patient, is dependent upon which specific young reference database is used. For example, our data indicate that the osteoporosis detection rate in a cohort of Chinese Americans is reduced by approximately half if a Chinese American referent database is used for calculation of T-scores rather than a Caucasian database.14

However, it is not clear if a T-score derived using an ethnic database would be more predictive of fracture risk, as prospective data do not currently exist. The International Society of Clinical Densitometry (ISCD) recommends that a Caucasian DXA reference be used for the calculation of T-scores and diagnosis of osteoporosis in non-Caucasians in the United States for several reasons. During the ISCD Position Development Conference, the panel recognized practical considerations such as the difficulty in defining ethnicity and creating relevant databases for every ethnicity. Further, use of a Caucasian database as a referent in the NORA study indicated a ‘similar pattern of increasing fracture risk with decreasing peripheral BMD level, regardless of race’.1 Lastly, the majority of data relating fracture risk to bone density exist for Caucasian women.47 The ISCD does recommend Z-score calculation using ethnic-specific reference data if available since the purpose is to ‘compare a patient's BMD with a closely matched normal population’.47,48

Screening and treatment recommendations in non-Caucasians in the US

The NORA study indicated that peripheral BMD ‘predicted fractures equally well within ethnic groups’1, but the absolute risk of fracture was lower in African and Asian Americans.1 The authors suggested that ethnic-specific screening and treatment recommendations might be helpful. The recently published World Health Organization (WHO) algorithm allows calculation of an estimated 10-year probability of hip (or major osteoporotic) fracture that is race-specific. This algorithm may help with treatment decisions for non-Caucasian individuals by providing a more accurate reflection of fracture risk from the T-score. While the algorithm provides improved fracture risk information for non-Caucasian patients, more ethnic-specific data would clearly be beneficial.49


Little data are available regarding risk factors for fracture and low bone density among Chinese American women. Chinese American women represent a growing population at risk for osteoporosis. By the year 2050, one-tenth of the US population will be of Asian descent, many of them Chinese American, according to the United States Census Bureau.50 In a cross-sectional study, we investigated predictors of BMD in 359 ambulatory Chinese American women, ages 20–90, using stepwise multiple regression analysis.46 Variables in the model included age, weight, height, menarche age, years since menopause, immigration age, years in the US, percentage of life in the US, number of pregnancies, oral contraceptive use, family history of osteoporosis, daily calcium intake, exercise, time outdoors, alcohol consumption and tobacco use. Among premenopausal women, weight was the most robust predictor of BMD, accounting for 10.5% of the variance at the lumbar spine (LS), 15.2% at the total hip (TH) and 16.6% at the femoral neck (FN). Time outdoors was also a positive predictor of BMD (1.4% at LS, 2.8% at TH, 1.6% at FN), while family history of osteoporosis (1.4% at TH) and age (3.7% at FN) were negative predictors. Among postmenopausal women, greater BMD at the LS and TH was associated with greater weight and earlier immigration age. Weight explained 16.4% of the variance at the LS and 19.8% at the TH; immigration age accounted for 3.1% of the variance at the LS and 4.1% at the TH. At the FN, years since menopause and weight were predictors of BMD, accounting for 14.4% and 8.7% of the variance, respectively.

As in other racial groups weight is the major predictor of BMD in Chinese American women. Limited nutritional information was collected, but calcium intake was not an independent determinant of BMD in this study. However, calcium intake was low [612 ± 17 mg (mean ± SD) daily]. Among premenopausal women, time outdoors was a significant predictor of BMD. This variable may reflect vitamin D status, physical activity or general health. Among postmenopausal women, older age of immigration to the US, independent of weight or age, had a negative effect on BMD. We hypothesize that later age of immigration may have a negative effect on BMD because51 older individuals are less likely to take on a Western lifestyle when they do immigrate or1 individuals who emigrate past a certain age may not be able to significantly alter BMD with lifestyle changes. More data are needed to differentiate between these possibilities. Bone mineral density in Chinese American women is influenced by a number of biological and lifestyle factors, including immigration. The results of this study provide new insights into risk factors for low bone density as they relate to environmental determinants in the growing population of Chinese American women.

In summary, racial differences in BMD are influenced predominantly by weight and bone size. Differences in African Americans persist even with adjustment for these factors. Racial differences in fracture rates are not completely explained by differences in areal BMD. The relationship between BMD and fracture risk may be altered by racial differences in bone size, hip axis length (hip fracture), other bone qualities or non-skeletal factors. More data are needed to provide screening and treatment recommendations that are relevant for non-Caucasian individuals.


Portions of this article have been adapted from Walker et al.51 Some of the data presented in this paper was supported in part by NIH K23AR053507. This work was also supported in part by a grant from the National Osteoporosis Foundation, the Mary and David Hoar Fellowship Program of the New York Community Trust and The New York Academy of Medicine as well as UL1 RR0241576 from the National Center for Resources (NCRR), a component of the National Institutes of Health (NIH) and NIH Roadmap for Medical Research.