Assessing Population Risk for Postmenopausal Osteoporosis: A New Strategy Using Data From the Behavioral Risk Factor Surveillance System (BRFSS)

Transcription

1 JOURNAL OF BONE AND MINERAL RESEARCH Volume 23, Number 1, 2008 Published online on October 1, 2007; doi: /JBMR American Society for Bone and Mineral Research Assessing Population Risk for Postmenopausal Osteoporosis: A New Strategy Using Data From the Behavioral Risk Factor Surveillance System (BRFSS) Christine R Schneyer, 1 Helio Lopez, 2 Mary Concannon, 3 and Marc C Hochberg 4 ABSTRACT: Osteoporosis public health measures are hindered by the inability to easily identify subclinical disease. We have now estimated state-specific osteoporosis prevalences using a simple formula (OST Index) to analyze age and weight of 62,882 older women; the prevalences determined are similar to those based on BMD. This new method has potential use for guiding implementation of osteoporosis prevention/treatment programs. Introduction: Although osteoporosis-related fractures are a major U.S. public health issue, population-based prevention programs have not yet been developed. One contributing factor has been lack of a suitable screening test to detect asymptomatic high-risk individuals. Materials and Methods: We estimated state-specific prevalences of postmenopausal osteoporosis using the Osteoporosis Self-Assessment Tool Index (OST Index; [self-reported weight in kg age] 0.2) to analyze data from 62,882 women 50 yr of age who participated in the 2002 Behavioral Risk Factor Surveillance System (BRFSS). The OST Index, designed to assess an individual s risk of disease, has previously been shown to have modest positive and high negative predictive value for osteoporosis defined by BMD criteria. Based on this index, women from each state were distributed among high-, moderate-, and low-risk OST categories. Calculated percentages for each category were weighted to U.S. Census Bureau population projections for By adjusting results to reflect previously validated percentages of women with osteoporosis in each risk category, we estimated the prevalence of postmenopausal osteoporosis in each state. Results: Our calculated weighted prevalence estimates agreed closely with those of the National Osteoporosis Foundation derived from actual femoral neck BMD measurements obtained in the third National Health and Nutrition Examination Survey ( ) and projected to U.S. census state population predictions for Comparison of unweighted BRFSS-OST results and NHANES BMD data revealed similar percentages of osteoporosis among all women 50 yr of age (BRFSS, 18.5%; NHANES, 18.0%; p 0.47) and also among white women (BRFSS, 19.0%; NHANES, 20.0%; p 0.28). However, the percentages of osteoporosis among blacks and Hispanics did not correspond, at least partly because of the lack of race-specific reference standards for BMD measurements and OST index ranges. Conclusions: Analysis of readily available BRFSS data with the OST index formula is a simple, no-cost technique that provides state prevalence estimates of postmenopausal osteoporosis that could be used to guide allocation of resources to statewide osteoporosis prevention programs. J Bone Miner Res 2008;23: Published online on October 1, 2007; doi: /JBMR Key words: osteoporosis, epidemiology, population studies, health services and economics, menopause The authors state that they have no conflicts of interest. INTRODUCTION WITH AN ESTIMATED prevalence of 10 million cases in 2002 (1) and projected doubling of the population 65 yr of age by 2030, (2) osteoporosis is a significant public health challenge. In addition to excess morbidity and mortality caused by osteoporosis-related fractures, (3) the economic burden placed on our health care system is severe; direct expenditures in the United States for fracture care in 2005 were >19 billion dollars. (4) Despite the magnitude of this problem and broad availability of effective antifracture therapy, (5) population-based interventions have not yet been developed. (6) One important reason for this deficiency has been the inability to perform population-based screen- 1 Division of Endocrinology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA; 2 Maryland Department of Health and Mental Hygiene,, Baltimore, Maryland, USA; 3 University of Maryland Cooperative Extension, Timonium, Maryland, USA; 4 Departments of Medicine and Epidemiology and Preventive Medicine, University of Maryland School of Medicine, Baltimore, Maryland, USA. 151

2 152 ing of asymptomatic at-risk persons to identify the percent of the population at high risk for fracture. In this paper, we propose a new and simple solution to this problem. The major consequence of osteoporosis is fractures, and one of the best predictors of fracture risk is reduced BMD. (7) However, DXA, the gold standard for BMD measurement, (8) is too expensive and inefficient for application to large populations. (9,10) As a result, several risk assessment tools have been proposed to distinguish high-risk individuals who are likely to have low BMD and who would therefore be the best candidates for BMD testing. (11) Each of these risk indices consists of a scorecard of two or more clinical risk factors (e.g., history of fracture after age 50, female sex, advanced age, low body weight, maternal history of hip fracture). (12 17) The simplest screening tool to identify candidates for BMD testing, and therefore the most appealing for use in large populations, is the Osteoporosis Self-Assessment Tool (OST). The OST index, (14) an integral value based only on age and weight, is calculated using this equation: OST index (self-reported weight in kg age in years) 0.2 In performing this calculation, the value of the OST index is truncated to an integral value by dropping numbers to the right of the decimal point. A recent review on OST has been published elsewhere. (18) The World Health Organization (WHO) has defined osteoporosis as a BMD measured by DXA at the hip of 2.5 SDs below the mean for non-hispanic white women, yr of age, (19) and this is the reference standard to which OST indices have been compared in studies on American and European women. (20 23) Using the WHO definition, the OST index has been divided into three risk categories. Those at high risk have OST Indices of 4 (the cut-point), those at moderate risk have indices ranging from 3 to 1, and those at low risk have the highest scores of 2. (20 23) The prevalence of osteoporosis is 60% for those in the high-risk group, 21% in moderate-risk women, and 4% or less for those in the low-risk category, and these were the prevalences applied to the risk categories generated in our study. As described in the Materials and Methods section, Asian women have lower cut-points. (14) In this study, by applying the OST index to age and weight data from the Behavioral Risk Factor Surveillance System (BRFSS; see following paragraph), we estimated the state-specific prevalence of osteoporosis in women 50 yr of age. We compared these estimates to those published by the National Osteoporosis Foundation that were based on results of BMD measurements from the third National Health and Nutrition Examination Survey. (1) MATERIALS AND METHODS Program description We estimated state and national prevalence of postmenopausal osteoporosis by calculating OST indices for women, 50 yr of age, who self-reported age and weight (as well as race) in the 2002 BRFSS. (24) The BRFSS is a random-digitdialed telephone survey of noninstitutionalized adults, 18 yr of age. It is conducted annually by health departments in U.S. states, the District of Columbia, and territories, in collaboration with the Centers for Disease Control and Prevention (CDC). The results are in the public domain, accessible at In 2002, the median BRFSS survey response rate among U.S. states and territories was 44.5% (ranging from 25.2% in New Jersey to 79.3% in Puerto Rico). (25) We restricted our study to women, 50 yr of age, because most of the studies correlating OST indices with BMD have been performed in this high-risk group. (18) We compared our weighted prevalence estimates to similar predictions reported by the National Osteoporosis Foundation (NOF). (1) The NOF estimates were derived from femoral neck BMD measurements performed on 3311 older women who had participated in the Third National Health and Nutrition Examination Survey (NHANES III, ). (26) NHANES III is a national probability sample of the civilian noninstitutionalized population of the United States. (27) Of NHANES III eligible selected participants, 63% responded positively to having BMD measurements. (28) BMD respondents were more likely than nonrespondents to be non-hispanic whites, to be physically active, and to have used hormone replacement therapy. However, these differences were reportedly minor and did not produce any major response biases affecting the BMD data. (28) Because the NOF prevalence estimates were derived from NHANES III BMD studies, we compared (1) distributions of age and race and (2) frequency of postmenopausal osteoporosis across age and weight categories in the BRFSS and NHANES III BMD samples. Proportions of osteoporotic women in the BRFSS sample were calculated from OST indices; in the NHANES sample, they were determined from BMD measurements. NHANES BMD data were analyzed using the same WHO diagnostic criteria for femoral neck osteoporosis as have been applied to OST-BMD correlation studies in white women (i.e., BMD measurement 2.5 SD below the mean for young non-hispanic white women). Non ageadjusted femoral neck measurements were used to determine the percentage of women by race (which was selfreported) with osteoporosis. Data analysis SCHNEYER ET AL. OST indices were calculated for all women of 50 yr of age who had self-reported age and weight data in the 2002 BRFSS. For this purpose, a categorical variable was created for the OST index formula with a coding scheme for low-, moderate-, and high-risk OST categories, as defined above. Weighted frequencies for the three OST risk groups were calculated by adjusting the sum of the unweighted BRFSS frequencies to reflect the state-specific populations projected for 2002 by the U.S. Census Bureau. State-specific weighted prevalence figures for postmenopausal osteoporosis were derived by adding together the number of women predicted to have osteoporosis within the weighted frequencies of each OST risk category (i.e., 60%, 21%, and

3 OSTEOPOROSIS PREVALENCE ESTIMATES USING BRFSS DATA 153 TABLE 1. AGE AND RACE DISTRIBUTION OF WOMEN 50 YR OF AGE IN THE BRFSS AND NHANES III SAMPLES COMPARED WITH U.S. CENSUS ESTIMATES FOR 2002 BRFSS (%) (n = 62,882) NHANES III BMD (%) (N = 3311) U.S. Census 2002 (%) Age of non-hispanic whites (yr) Race White Black Hispanic Asian 1.5 NA 3.2 Other NHANES III subjects included only those women whose BMD measurements were included in the study. NA, not available. 4% of women in high-, moderate-, and low-risk OST categories, respectively; see Introduction). These percentages apply to white but not to Asian women because of the latter s generally smaller body habitus. In that population, OSTA (OST for Asians) risk categories have necessarily been defined by lower cut-points (i.e., < 4 [high risk], 4 to 1 [moderate risk], and > 1 [low risk]), (14,29,30) and BMD measurements have been compared with a young-adult reference population of Asian women (rather than non- Hispanic white women). To enable comparisons between the BRFSS and NHANES survey areas, weighted state-specific BRFSS data were modified to exclude Guam, Puerto Rico, and the Virgin Islands. The proportion of women with postmenopausal osteoporosis (by age and race) in the BRFSS and NHANES BMD samples were compared using two-sided t-tests. p < 0.05 was considered significant. All analyses of BRFSS data were performed using SAS Version (SAS Institute, Cary, NC, USA). (31) RESULTS Characteristics of BRFSS 2002 and NHANES III ( ) cohorts Table 1 shows the distribution of age and race of older women in the BRFSS (n 62,882) and the NHANES III samples (n 3311), together with the corresponding U.S. Census Bureau population estimates for (32) Table 1 presents unweighted data because, to our knowledge, the weighted data used by the NOF for their 2002 state-specific osteoporosis prevalence estimates (see below) are not publicly available. In general, the BRFSS race and age distributions were more consistent with those of the U.S. census. Table 2 shows the proportion of women with osteoporosis, across age and race categories, calculated from OST categories in the BRFSS sample and from BMD measurements in the NHANES sample. There were no statistically significant differences between the calculated proportions either within the total sample (BRFSS, 18.5%; NHANES, 18.0%; p 0.47) or for the sample of white women (BRFSS, 19%; NHANES, 20%; p 0.28). However, there were significant differences between the proportions calculated for both blacks and Hispanics. There was no separate category for Asian women in the NHANES BMD data. However, we calculated the percentage of Asian women with osteoporosis in the BRFSS sample using two different sets of OST index ranges. With OST cut-points validated in whites, the proportion of Asian women with osteoporosis was 26.9% (95% CI, 23.4, 30.4). However, when using cutpoints validated in Asians (OSTA), (14,29,30) the proportion fell to 15.8% (95% CI, 12.9, 18.7; p < ). Percentile ranking of states in OST risk categories The 2002 BRFSS survey collected self-reported age and weight data from a sample of women in every state. Sample sizes varied widely among states, ranging from 476 for Alaska to 3625 for Pennsylvania. From these data, we calculated state-specific weighted percentages of subjects constituting the three OST risk categories (Table 3). Hawaii, the New England states, Arizona, and Florida ranked among the top quarter of states for both high-risk (9.9 14%) and moderate-risk ( %) OST index scores, suggesting that these states have the highest population prevalences of postmenopausal osteoporosis. Conversely, Texas, Louisiana, and Alaska shared the lowest percentile rankings for high- ( %) and moderate-risk ( %) OST index scores, indicating that older women in these states have relatively low risk for this disease. Because we generated the values in Table 3 using OST Index cut-points for non-asian women, and because Hawaii has a large Asian population, we thought it likely that the results for Hawaii were falsely elevated. Of 926 Asian women surveyed nationwide, 658 were in Hawaii (nearly one half of the 1542 Hawaiian women in the sample). Therefore, we performed a sensitivity analysis to determine the effect of subtraction of Asians from the national sample. Removal of Asian women caused the weighted percentage of Hawaiian women to drop from 14.0% to 7.7% in the high-risk category (from a ranking of 1st to 43rd) and from 57.3% to 47.9% in the moderate-risk group (from a ranking of 2nd to 42nd). California had the second highest number of Asian women (34 of 992 surveyed). However, omission of Asians from the California analysis had minimal effect on its ranking (data not shown). The distribution of weighted frequencies of high-risk OST scores by state is shown in Fig. 1. Quintile 1 has the highest percentage of these scores, whereas quintile 5 has the lowest (Fig. 1). With the exceptions of Florida, Arizona, and New Mexico, there is a concentration of quintiles 1, 2, and 3 in the northern states. State-specific and national osteoporosis prevalence: comparison of BRFSS- with NHANES-dependent methods Table 3 also compares the state-specific osteoporosis prevalence estimates derived in the current analysis with

4 154 SCHNEYER ET AL. TABLE 2. FREQUENCIES OF POSTMENOPAUSAL OSTEOPOROSIS ACCORDING TO AGE AND ETHNIC GROUP: A COMPARISON OF BRFSS AND NHANES III SAMPLES BRFSS NHANES n % (95% CI) n % (95% CI) p* Non-Hispanic whites by age (yr) , (9.4,10.5) (3.7,9.3) < , (14.5,15.9) (11.7,19.9) , (23.6,25.4) (24.2,33.4) < , (40.1,42.9) (41.6,53.2) <0.05 Race/ethnicity Total, all races 62, (18.1,18.9) 3, (16.4,19.6) 0.47 White 53, (18.6,19.4) 1, (17.8,22.2) 0.28 Black 3, (11.0,13.6) (3.0,7.0) < Hispanic 2, (14.9,18.7) (7.1,12.9) < For BRFSS, n corresponds to the total number of women 50 yr of age whose self-reported age and weight were included in the 2002 survey; % is the proportion of women in the sample with osteoporosis as calculated from OST risk categories. For NHANES, n is the number of women 50 yr of age with BMD measurements included in the study; % is the proportion of these women with BMD evidence of osteoporosis (femoral neck T-score 2.5). * Two-sided t-test. The total also includes an Other category, not shown. corresponding estimates published by the NOF. (1) The NOF estimates of state-specific osteoporosis prevalence were determined by applying the NHANES III BMD results by age and race (26) (these results are the NHANES percentages reproduced in Table 2) to the U.S. Census Bureau age-, sex-, and race-specific state population predictions for (1) Our figures were similar to those derived by the NOF from the NHANES BMD measurements. The national prevalence for postmenopausal osteoporosis was estimated to be 7.3 million women when using estimates derived from BRFSS data and 7.7 million based on NHANES III BMD data. The similarity between these weighted prevalences is consistent with the fact that there was no statistically significant difference between the sample proportions shown in Table 2. DISCUSSION The major significance of this report is that the OST index, a simple risk assessment tool developed to estimate the osteoporosis risk of individuals, can also be used to assess the risk burden for postmenopausal osteoporosis in large populations. Table 3 shows close agreement between our estimates of weighted state-specific osteoporosis prevalence and those of the NOF (the only other nationwide state-specific prevalence estimates currently available). As described above, the NOF estimates presented in Table 3 were derived from the NHANES III BMD data shown in Tables 1 and 2. Although the similarity between the two sets of estimates is reassuring, the NOF estimates cannot be considered a gold standard. One reason for this is that the age and race distributions of the NHANES BMD data, on which the prevalence estimates are based, differ from the population distributions estimated by the U.S. Census Bureau (Table 1). Another reason is that BMD studies in minorities, in contrast to whites, are of questionable validity because of a lack of race-specific normative databases (see below). Support for the reliability of our approach is provided in Table 2, which presents the percentage of women with osteoporosis in different age and race categories determined from BRFSS and NHANES data. There were no statistically significant differences in the percentages for the total sample of older women or for the subgroup of white women. In contrast, the differences in percentages for blacks and Hispanics were statistically significant, but no definitive conclusion can be drawn concerning the differences in minorities. This is because, as with the NHANES BMD data involving racial subgroups, the BRFSS-OST findings are questionable because OST index ranges validated in whites had to be used to analyze the data of minority women (no race-specific OST cut-points were available for blacks or Hispanics). A recent systematic review determined that the OST index ranges developed in studies focused on white women detected femoral neck osteoporosis, as determined by DXA, with a median sensitivity of 92% (range, 88 96%), meaning that only 8% of cases are missed. The specificity was 39% (range, 30 71%). (18) The review concluded that, in white women, the performance of OST was moderate in ruling out femoral neck osteoporosis (negative likelihood ratio 0.19; 95% CI, ), and there were similar findings (but with greater heterogeneity) for Asian women using OSTA cut-points. (18) This same review concluded that the use of the OST index is more reliable in women 65 yr of age and less so in early postmenopausal women. (18) This may have been a factor in the differences between BRFSS and NHANES results regarding age in Table 2. A potential problem with BRFSS data is that the weight values are self-reported; these are often inaccurate (usually in the downward direction). (33) Our results suggest geographical differences in osteoporosis prevalence among older U.S. women. As shown in the map in Fig. 1, states with the highest osteoporosis prevalence estimates (quintiles 1 and 2) are located primarily in the northern parts of the country. In contrast, with the exceptions of Florida, Arizona, and New Mexico, there is a

5 OSTEOPOROSIS PREVALENCE ESTIMATES USING BRFSS DATA 155 TABLE 3. STATE-SPECIFIC PREVALENCE OF OSTEOPOROSIS IN WOMEN 50 YR OF AGE, USING BRFSS, 2002 BRFSS OST risk categories (weighted to state population) High Moderate Low PMP osteoporosis prevalence estimates STATE % (95% CI) n % (95% CI) n % (95% CI) n BRFSS (n) NHANES (n)* Alabama 8.0 ( ) 55, ( ) 336, ( ) 303, , ,900 Alaska 2.7 ( ) 1, ( ) 26, ( ) 33,117 7,973 12,000 Arizona 10.6 (8.2 13) 86, ( ) 445, ( ) 279, , ,600 Arkansas 8.5 ( ) 36, ( ) 220, ( ) 176,467 75,369 86,700 California 7.7 ( ) 356, ( ) 2,641, ( ) 1,622, , ,200 Colorado 8.2 ( ) 47, ( ) 314, ( ) 217, , ,200 Connecticut 11.5 ( ) 58, ( ) 287, ( ) 161, , ,000 Delaware 8.6 ( ) 10, ( ) 61, ( ) 46,679 20,757 21,300 DC 7.7 ( ) 6, ( ) 40, ( ) 39,408 14,194 8,200 Florida 10.2 ( ) 295, ( ) 1,625, ( ) 980, , ,100 Georgia 6.4 ( ) 68, ( ) 526, ( ) 476, , ,000 Hawaii 14 ( ) 26, ( ) 106, ( ) 53,308 40,064 37,500 Idaho 9.4 ( ) 16, ( ) 84, ( ) 69,381 30,124 40,000 Illinois 8.7 ( ) 156, ( ) 855, ( ) 779, , ,400 Indiana 10.4 ( ) 93, ( ) 410, ( ) 392, , ,200 Iowa 8.8 ( ) 40, ( ) 235, ( ) 190,552 81, ,100 Kansas 9.4 ( ) 35, ( ) 198, ( ) 147,987 69,091 80,400 Kentucky 5.8 ( ) 35, ( ) 315, ( ) 265,003 98, ,000 Louisiana 5.4 ( ) 34, ( ) 296, ( ) 298,732 94, ,700 Maine 9.9 ( ) 20, ( ) 108, ( ) 80,659 38,486 42,700 Maryland 7.5 ( ) 57, ( ) 385, ( ) 326, , ,500 Mass 11.0 ( ) 105, ( ) 513, ( ) 340, , ,000 Michigan 9.0 ( ) 134, ( ) 689, ( ) 668, , ,200 Minnesota 9.2 ( ) 65, ( ) 347, ( ) 295, , ,100 Mississippi 8.1 ( ) 33, ( ) 191, ( ) 188,315 67,909 71,300 Missouri 7.8 ( ) 69, ( ) 449, ( ) 381, , ,600 Montana 9.6 ( ) 12, ( ) 73, ( ) 45,878 24,960 31,900 Nebraska 9.6 ( ) 23, ( ) 127, ( ) 95,146 44,675 53,800 Nevada 8.3 ( ) 23, ( ) 154, ( ) 110,653 51,211 55,900 New Hampshire 8.2 ( ) 14, ( ) 101, ( ) 64,877 32,941 36,200 New Jersey 10.2 ( ) 132, ( ) 670, ( ) 494, , ,200 New Mexico 10.1 ( ) 26, ( ) 132, ( ) 102,258 47,611 44,700 New York 8.8 ( ) 253, ( ) 1,478, ( ) 1,153, , ,100 North Carolina 7.9 ( ) 95, ( ) 605, ( ) 502, , ,000 North Dakota 9.5 ( ) 8, ( ) 49, ( ) 35,940 17,265 21,600 Ohio 8.2 ( ) 140, ( ) 874, ( ) 700, , ,300 Oklahoma 10 ( ) 52, ( ) 269, ( ) 199,086 96, ,700 Oregon 10.8 ( ) 57, ( ) 259, ( ) 220,398 98, ,300 Pennsylvania 9.3 ( ) 187, ( ) 1,060, ( ) 760, , ,500 Rhode Island 11.2 ( ) 18, ( ) 97, ( ) 50,277 33,741 32,400 South Carolina 7.6 ( ) 45, ( ) 314, ( ) 244, , ,500 South Dakota 9.3 ( ) 10, ( ) 60, ( ) 40,491 20,676 24,300 Tennessee 8.6 ( ) 76, ( ) 437, ( ) 375, , ,800 Texas 7.1 ( ) 185, ( ) 1,235, ( ) 1,199, , ,300 Utah 8.5 ( ) 19, ( ) 112, ( ) 98,895 39,452 51,600 Vermont 8.9 ( ) 8, ( ) 50, ( ) 34,569 16,931 19,300 Virginia 9.2 ( ) 93, ( ) 474, ( ) 453, , ,000 Washington 9.5 ( ) 77, ( ) 405, ( ) 330, , ,700 West VA 10.8 ( ) 34, ( ) 146, ( ) 137,397 57,044 69,100 Wisconsin 8.6 ( ) 68, ( ) 407, ( ) 319, , ,900 Wyoming 9.8 ( ) 6, ( ) 36, ( ) 26,559 12,834 15,700 Range 11.3 ( ) 15.2 ( ) 25 ( ) Mean 8.7 ( ) 51.3 ( ) 40.0 ( ) Median IQR Total 3,623,044 21,351,897 16,613,995 7,322,284 7,748,700 The BRFSS OST categories show the weighted percentage, CI, and weighted n for each state, calculated by adjusting the unweighted state-specific BRFSS frequencies to the 2002 U.S. population. The two right columns compare prevalence estimates for postmenopausal osteoporosis derived from BRFSS data with those based on BMD measurements of NHANES III subjects. * n, state-specific prevalence estimates for postmenopausal osteoporosis were calculated from BRFSS data as follows: (high-risk OST n)(0.60) + (moderate-risk OST n)(0.21) + (low-risk OST n)(0.04). Multiplication factors for each OST risk category represent the fraction of women in that category with osteoporosis. OST scores for Hawaiian women were falsely elevated because of a large proportion (43%) of Asian women in the state sample. Recalculation of OST scores after removal of Asian women from the national sample substantially altered results for Hawaii, but not for any other state. OST scores for Hawaii with Asian women excluded were as follows: high risk, 7.7% ( %), n 7341; moderate risk, 47.9% ( %), n 45,424; low risk, 44.4% ( %), n 42,069. PMP, postmenopausal.

6 156 SCHNEYER ET AL. FIG. 1. Weighted state-specific BRFSS high-risk OST frequencies in decreasing quintiles: (1) %; (2) %; (3) 8.6 to <9.3%; (4) 7.9 to <8.6%; (5) %. OST score for Hawaii was calculated after subtraction of Asian women from the national sample. OST score for District of Columbia fell within the fifth quintile (data not shown). predominance of states in southern regions with lower prevalence estimates (quintiles 4 and 5). At first glance, this geographical pattern would seem related to generally better bone health in the southern United States because of more extensive sun exposure and superior vitamin D status. (34,35) In addition, differences in the racial makeup of state populations (e.g., relatively large proportions of blacks who typically have relatively high BMD (36) ) might also contribute to lower estimates of osteoporosis prevalence in the southeastern states. (37) However, the apparent effects of geography may be more complex. On the one hand, there is evidence that bone health in the South may be poorer than expected despite increased sun exposure. BMD data from NHANES III revealed that the mean value for total femur BMD was slightly lower in the South (although sufficient data for this analysis were available only for non-hispanic whites). (28) Also, the incidence of hip fractures is higher among women in southern states (38) (although there have been suggestions that hip fracture rates depend on the region of residence early in life (39) ). On the other hand, there is a higher prevalence of obesity in southern states, (40) and it has been widely reported that greater weight is associated with higher bone mass (recall that weight is one of two variables in the OST index formula). This would explain the generally higher OST indices across the South. Why then would hip fractures be more common in the South? One reason is that factors other than low BMD increase fracture risk, and lower socioeconomic status is a major contributor. (41) Another reason is that the relationship between body weight and BMD is less straightforward than previously thought because of emerging evidence that obesity (specifically, increasing fat mass), contrary to dogma, may actually be a risk factor for osteoporosis. (42 45) If these reports are confirmed, it could require reevaluating the correlations between OST indices and BMD for obese individuals as has been done already for ethnic groups such as Asians (see above). Age, the critically important second variable in the OST index equation, (46) has a multifactorial association with the pathogenesis of osteoporosis. For example, increasing age is associated with estrogen deficiency, reduced cutaneous synthesis of vitamin D, decreased osteoblastic activity, and greater comorbidity, all of which contribute to the development of osteoporosis. (47) Therefore, one obvious explanation for the high frequency of postmenopausal osteoporosis in Florida is that the proportion of its population 65 yr of age is the largest in the country (nearly 18% in 2000 compared with a national average of 12.4%). (48) Alaska, at the opposite end of the age spectrum, had only 6% of residents in the 65-yr age group in 2000, (48) a demographic that presumably contributed to its distinction of having the lowest osteoporosis prevalence estimate in the United States (Table 3). What are the effects of race and ethnicity? The agreement between BRFSS osteoporosis prevalence estimates and those derived from NHANES BMD data (Table 3) was likely favored by the white majority in the two study samples (Table 1). The reason, as mentioned above, is that both the NHANES III BMD measurements and the OST index ranges used in our calculations were based on white reference data. In contrast, the dramatic effect of using Asian-specific OST indices (OSTA) to estimate osteoporosis in an Asian population highlights the need for ethnicspecific OST index ranges. The current practice for measuring BMD by DXA in all women, regardless of race or ethnicity, is to use a young female white non race-adjusted normative database as the reference standard. (49) The use of white BMD reference standards for Hispanics is probably less problematic than for blacks because Hispanic and white women have more similar BMD and fracture profiles. (36,50) However, it has been suggested that race-specific normative databases would be more appropriate. (51) To date, OST has been validated not only in white and Asian women, but also in a group of South American women. (52) One preliminary study has been conducted in black women. (17) In future studies, the impact of geographical differences, race, and age, as well as multiple other factors (e.g., physical inactivity, smoking, alcohol consumption, arthritis, depression, health care access), on state-specific osteoporosis prevalence may be easily studied within the BRFSS dataset. Cross-tabulation of OST index findings with additional data obtained from the BRFSS core questionnaire and optional

7 OSTEOPOROSIS PREVALENCE ESTIMATES USING BRFSS DATA 157 modules could provide information helpful for the development of targeted osteoporosis programs. Inclusion of self-reported age and weight in the BRFSS core questions is a guarantee that annual monitoring of changes in osteoporosis prevalence will be possible using OST index analyses. The major value of combining OST with BRFSS is that it allows for the first time an easy and cost-free method for population surveillance of osteoporosis, a wide-spread chronic disease for which no practical surveillance method has been available. Surveillance data are crucial to guide the planning and implementation of public health programs for prevention and treatment of osteoporosis. Unfortunately, no federal funding is provided currently for collection of osteoporosis data in the BRFSS core questionnaire. Consequently, osteoporosis remains the one chronic disease of public health dimension that the BRFSS survey does not specifically address. Despite this shortcoming, we showed that BRFSS core questions can still serve as a source of osteoporosis prevalence data by using the OST index to analyze self-reported age and weight data. The methodology involved is not only remarkably simple, but it is also free. Use of this technique to document state-specific osteoporosis prevalence would provide federal agencies and national organizations with the information necessary for allocation of funds to states and territories with the greatest need for bone health programs. (53) REFERENCES 1. America s Bone Health 2002 The State of Osteoporosis and Low Bone Mass in Our Nation. National Osteoporosis Foundation, Washington, DC, USA. 2. U.S. Census Bureau, Population Division, Population Projections Branch 2004 U.S. interim projections by age, sex, race, and Hispanic origin Available at census.gov/ipc/www/usinterimproj/. Accessed March 2, Cummings SR, Melton LJ 2002 Epidemiology and outcomes of osteoporotic fractures. Lancet 359: Burge R, Dawson-Hughes B, Solomon DH, Wong JB, King A, Tosteson A 2007 Incidence and economic burden of osteoporosis-related fractures in the United States, J Bone Miner Res 22: Gass M, Dawson-Hughes B 2006 Preventing osteoporosisrelated fractures: An overview. Am J Med 119(Suppl 1):S3 S Bone Health and Osteoporosis 2004 A Report of the Surgeon General. U.S. Department of Health and Human Services, Washington, DC, USA. 7. Johnell O, Kanis JA, Oden A, Johansson H, Eisman JA, Fujiwara S, Kroger H, Honkanen R, Melton LJ III, O Neill T, Reeve J, Silman A, Tenenhouse A 2005 Predictive value of BMD for hip and other fractures. J Bone Miner Res 20: Lewiecki EM, Watts NB, McClung MR, Petak SM, Bachrach LK, Shepherd JA, Downs RW Jr, International Society for Clinical Densitometry 2004 Official positions of the international society for clinical densitometry. J Clin Endocrinol Metab 89: Melton LJ III, Johnell O, Lau E, Mautalen CA, Seeman E 2004 Osteoporosis and the global competition for health care resources. J Bone Miner Res 19: Cummings SR, Bates D, Black DM 2002 Clinical use of bone densitometry: Scientific review. JAMA 288: Wehren LE, Siris ES 2004 Beyond bone mineral density: Can existing clinical risk assessment instruments identify women at increased risk of osteoporosis? J Intern Med 256: Cadarette SM, Jaglal SB, Kreiger N, McIsaac WJ, Darlington GA, Tu JV 2000 Development and validation of the Osteoporosis Risk Assessment Instrument to facilitate selection of women for bone densitometry. CMAJ 162: Michaelsson K, Bergstrom R, Mallmin H, Holmberg L, Wolk A, Ljunghall S 1996 Screening for osteopenia and osteoporosis: Selection by body composition. Osteoporos Int 6: Koh LK, Sedrine WB, Torralba TP, Kung A, Fujiwara S, Chan SP, Huang QR, Rajatanavin R, Tsai KS, Park HM, Reginster JY, Osteoporosis Self-Assessment Tool for Asians (OSTA) Research Group 2001 A simple tool to identify Asian women at increased risk of osteoporosis. Osteoporos Int 12: Sedrine WB, Chevallier T, Zegels B, Kvasz A, Micheletti MC, Gelas B, Reginster JY 2002 Development and assessment of the Osteoporosis Index of Risk (OSIRIS) to facilitate selection of women for bone densitometry. Gynecol Endocrinol 16: Lydick E, Cook K, Turpin J, Melton M, Stine R, Byrnes C 1998 Development and validation of a simple questionnaire to facilitate identification of women likely to have low bone density. Am J Manag Care 4: Wallace LS, Ballard JE, Holiday D, Turner LW, Keenum AJ, Pearman CM 2004 Evaluation of decision rules for identifying low bone density in postmenopausal African-American women. J Natl Med Assoc 96: Rud B, Hilden J, Hyldstrup L, Hrobjartsson A 2007 Performance of the Osteoporosis Self-Assessment Tool in ruling out low bone mineral density in postmenopausal women: A systematic review. Osteoporos Int 18: Kanis JA, Delmas P, Burckhardt P, Cooper C, Torgerson D 1997 Guidelines for diagnosis and management of osteoporosis. The European Foundation for Osteoporosis and Bone Disease. Osteoporos Int 7: Geusens P, Hochberg MC, van der Voort DJ, Pols H, van der Klift M, Siris E, Melton ME, Turpin J, Byrnes C, Ross P 2002 Performance of risk indices for identifying low bone density in postmenopausal women. Mayo Clin Proc 77: Cadarette SM, McIsaac WJ, Hawker GA, Jaakkimainen L, Culbert A, Zarifa G, Ola E, Jaglal SB 2004 The validity of decision rules for selecting women with primary osteoporosis for bone mineral density testing. Osteoporos Int 15: Richy F, Gourlay M, Ross PD, Sen SS, Radican L, De Ceulaer F, Ben Sedrine W, Ethgen O, Bruyere O, Reginster JY 2004 Validation and comparative evaluation of the osteoporosis self-assessment tool (OST) in a Caucasian population from Belgium. QJM 97: Rud B, Jensen JE, Mosekilde L, Nielsen SP, Hilden J, Abrahamsen B 2005 Performance of four clinical screening tools to select peri- and early postmenopausal women for dual X-ray absorptiometry. Osteoporos Int 16: Centers for Disease Control and Prevention (CDC) Behavioral Risk Factor Surveillance System Data. Available at cdc.gov/brfss/index.htm. Accessed February 7, Centers for Disease Control and Prevention (CDC) 2002 Behavioral Risk Factor Surveillance System: Summary Data Quality Report. Available at technical_infodata/quality.htm. Accessed January 28, Looker AC, Orwoll ES, Johnston CC Jr, Lindsay RL, Wahner HW, Dunn WL, Calvo MS, Harris TB, Heyse SP 1997 Prevalence of low femoral bone density in older U.S. adults from NHANES III. J Bone Miner Res 12: National Center for Health Statistics (NCHS) Plan and operation of the Third National Health and Nutrition Examination Survey, Vital Health Statistics. Available at Accessed January 14, Looker AC, Wahner HW, Dunn WL, Calvo MS, Harris TB, Heyse SP, Johnston CC Jr, Lindsay R 1998 Updated data on proximal femur bone mineral levels of US adults. Osteoporos Int 8: Park HM, Sedrine WB, Reginster JY, Ross PD 2003 Korean

8 158 experience with the OSTA risk index for osteoporosis: A validation study. J Clin Densitom 6: Chan SP, Teo CC, Ng SA, Goh N, Tan C, Deurenberg-Yap M 2006 Validation of various osteoporosis risk indices in elderly Chinese females in Singapore. Osteoporos Int 17: SAS Institute 2004 SAS OnlineDoc [computer program]. Version SAS Institute, Cary, NC, USA. 32. U.S. Census Bureau U.S. Census Bureau Population Estimates. Available at Accessed September 5, McAdams MA, Van Dam RM, Hu FB 2007 Comparison of self-reported and measured BMI as correlates of disease markers in US adults. Obesity (Silver Spring) 15: Holick MF 2006 High prevalence of vitamin D inadequacy and implications for health. Mayo Clin Proc 81: Calvo MS, Whiting SJ 2003 Prevalence of vitamin D insufficiency in Canada and the United States: Importance to health status and efficacy of current food fortification and dietary supplement use. Nutr Rev 61: Barrett-Connor E, Siris ES, Wehren LE, Miller PD, Abbott TA, Berger ML, Santora AC, Sherwood LM 2005 Osteoporosis and fracture risk in women of different ethnic groups. J Bone Miner Res 20: McKinnon J 2003 The black population in the United States: March Available at pubs/p pdf. Accessed February 5, Karagas MR, Lu-Yao GL, Barrett JA, Beach ML, Baron JA 1996 Heterogeneity of hip fracture: Age, race, sex, and geographic patterns of femoral neck and trochanteric fractures among the US elderly. Am J Epidemiol 143: Lauderdale DS, Thisted RA, Goldberg J 1998 Is geographic variation in hip fracture rates related to current or former region of residence? Epidemiology 9: Blanck HM, Dietz WH, Galuska DA, Gillespie C, Hamre R, Kettel Khan L, Serdula MK, Ford ES, Garvin WS, Mokdad AH 2006 State-specific prevalence of obesity among adults United States, Morbid Mortal Wkly Rev 55: Zingmond DS, Soohoo NF, Silverman SL 2006 The role of socioeconomic status on hip fracture. Osteoporos Int 17: Zhao LJ, Liu YJ, Liu PY, Hamilton J, Recker RR, Deng HW 2007 Relationship of obesity with osteoporosis. J Clin Endocrinol Metab 92: Hsu YH, Venners SA, Terwedow HA, Feng Y, Niu T, Li Z, Laird N, Brain JD, Cummings SR, Bouxsein ML, Rosen CJ, Xu X 2006 Relation of body composition, fat mass, and serum lipids to osteoporotic fractures and bone mineral density in Chinese men and women. Am J Clin Nutr 83: Núñez NP, Carpenter CL, Perkins SN, Berrigan D, Jaque SV, SCHNEYER ET AL. Ingles SA, Bernstein L, Forman MR, Barrett JC, Hursting SD 2007 Extreme obesity reduces bone mineral density: Complementary evidence from mice and women. Obesity (Silver Spring) 15: Looker AC, Flegal KM, Melton LJ III 2007 Impact of increased overweight on the projected prevalence of osteoporosis in older women. Osteoporos Int 18: Steiger P, Cummings SR, Black DM, Spencer NE, Genant HK 1992 Age-related decrements in bone mineral density in women over 65. J Bone Miner Res 7: Raisz LG 2005 Pathogenesis of osteoporosis: Concepts, conflicts, and prospects. J Clin Invest 115: Dept. of Health & Human Services, Administration on Aging 2004 A profile of older Americans, 2002: Geographic distribution. Available at 6.asp. Accessed March 5, The International Society For Clinical Densitometry 2005 Official Positions. Available at positions/officialpositionstext.cfm. Accessed September 1, Lauderdale DS, Jacobsen SJ, Furner SE, Levy PS, Brody JA, Goldberg J 1998 Hip fracture incidence among elderly Hispanics. Am J Public Health 88: Cauley JA, Lui L-Y, Ensrud KE, Zmuda JM, Stone KL, Hochberg MC, Cummings SR 2005 Bone mineral density and the risk of incident nonspinal fractures in black and white women. JAMA 293: Sen SS, Rives VP, Messina OD, Morales-Torres J, Riera G, Angulo-Solimano JM, Neto JF, Frisoli A Jr, Sáenz RC, Geling O, Ross PD 2005 A risk assessment tool (OsteoRisk) for identifying Latin American women with osteoporosis. J Gen Intern Med 20: Frieden TR 2004 Asleep at the switch: Local public health and chronic disease. Am J Public Health 94: Address reprint requests to: Christine R Schneyer, MD, MHS Division of Endocrinology Johns Hopkins University School of Medicine 1830 East Monument Street, Suite 333 Baltimore, MD 21287, USA cschneyer@jhmi.edu Received in original form March 17, 2007; revised form September 11, 2007; accepted September 26, 2007.