Diabetes Mellitus as a Risk Factor for Hip Fracture in Mexican American Older Adults

Transcription

1 Journal of Gerontology: MEDICAL SCIENCES 2002, Vol. 57A, No. 10, M648 M653 Copyright 2002 by The Gerontological Society of America Diabetes Mellitus as a Risk Factor for Hip Fracture in Mexican American Older Adults Kenneth J. Ottenbacher, 1,2,4 Glenn V. Ostir, 2,3 M. Kristen Peek, 2,4 James S. Goodwin, 2,3,4 and Kyriakos S. Markides 2,4 1 Division of Rehabilitation Sciences, School of Allied Health Sciences, University of Texas Medical Branch at Galveston. 2 Sealy Center on Aging, University of Texas Medical Branch at Galveston. 3 Division of Geriatrics, Department of Medicine, University of Texas Medical Branch at Galveston. 4 Department of Preventive Medicine and Community Health, University of Texas Medical Branch at Galveston. Background. Hip fracture in older adults is a significant medical, social, and economic concern to society. Little is known regarding diabetes as a risk factor for hip fracture in the Mexican American population. The objective of this study was to examine diabetes and other potential risk factors for hip fracture in a sample of community-dwelling, older, Mexican American adults. Methods. The study was a prospective cohort design involving the Hispanic Established Population for the Epidemiologic Study of the Elderly, a longitudinal study involving a weighted probability sample of Mexican American adults ( 65 years) living in the southwestern United States. Included in the study were 3050 older Mexican American subjects who were originally interviewed and tested at baseline and then followed with reassessment at 2, 5, and 7 years. Incidence of hip fracture was examined for subjects over 7-year follow-up. Results. At baseline, 690 subjects were identified with diabetes. One hundred and thirty-four subjects experienced a first-time hip fracture during follow-up. Cox proportional hazard regression revealed an increased hazard ratio for hip fracture in subjects with diabetes compared to those without diabetes (hazard ratio 1.57, 95% confidence interval [CI 95 ] 1.03, 2.39, p.04) when adjusted for age, body mass index, smoking, and previous stroke. The hazard ratio for Mexican Americans taking insulin was 2.84 (CI , 5.43, p.002) when adjusted for covariates. Conclusions. We found diabetes was associated with increased risk for a hip fracture in older Mexican Americans, particularly subjects taking insulin. Diabetes has not previously been considered a risk factor for hip fracture in older adults. The high incidence of type 2 diabetes in the Mexican American population highlights the need for increased research on risk factors in this ethnic group. H IP fracture is among the most common reasons for hospitalization and disability among older Americans (1 4). More than 300,000 adults in the United States experience a hip fracture each year (4). The incidence of hip fractures world-wide is projected to increase by 250% in the next 25 years (5). In addition to pain, disability, and mortality (6), hip fracture is significantly related to hospital costs (7,8). Sloan and colleagues (9) reported a 167% increase in Medicare payments for patients with hip fracture from 1994 through The cost of inpatient hospital care for persons with hip fracture in the United States is more than $6 billon per year and is expected to reach $16 billon by 2040 as the population ages (10). Risk factors for hip fracture include those related to falls and decreased bone density. Previous research has identified age, female gender, history of stroke, limitations in activities of daily living, decreased mobility, smoking, and impaired visual function as risk factors (1,3,6,7). Other variables associated with a lower risk of hip fracture include later age at menopause for women, use of thaizide, increased body mass index, and greater than average height and muscle strength (1,3 7). Some regional and ethnic variations have been reported, with higher rates of hip fracture in the southeastern United States and for non-hispanic white women (11). The prognosis for older adults who experience a hip fracture is not positive (3). Twenty percent die in the year following hip fracture, and 25% of those who survive require a higher level of long-term care (3). Those persons who do return to the community after a hip fracture have greater difficulty with activities of daily living, and only 60% will recover their prefracture walking ability by 6 months (6 9). Consequently, there is a need to improve health and functional outcomes for this population. An important step in understanding and improving functional outcomes is being able to identify and adjust for baseline characteristics that influence morbidity and functional loss. This is particularly true for minority and disadvantaged populations, where little research on outcomes has been conducted. In particular, we were interested in outcomes related to hip fracture in Mexican American older adults. Our interest in studying this population was based on the following factors: (i) the incidence of type 2 diabetes in this population is known to be high (12,13); (ii) recent studies in non-hispanic whites have suggested that diabetes might be a risk factor for hip fracture in some older adults (13); and (iii) we had access to a large population-based sample of Mexican American older adults that included comprehensive longitudinal data on health status and outcome (12). M648

2 DIABETES AS A RISK FACTOR FOR HIP FRACTURE M649 Diabetes mellitus has not been considered a significant risk factor for hip fracture in older adults (13). Researchers have reported inconsistent findings regarding bone density in persons with type 2 (adult onset) diabetes in comparison to control subjects (14), and few studies have examined the possible relation between diabetes and clinical outcomes of decreased bone density (15,16). Two recent large-sample cohort studies reported diabetes as a risk factor for fractures, including hip facture, in older women (13,17). These studies were limited to non-hispanic white women, and the baseline response rate to the survey questionnaire in one of the studies was low (42%). The prevalence of diabetes mellitus in the Hispanic population is high (12), and information regarding risk factors for hip fracture in the Hispanic population is limited. Espino and colleagues (18) examined the prevalence, incidence, and risk factors for hip fracture in community-dwelling, Mexican American older adults. They reported rates of 9.1 fractures per 1000 person-years for Mexican American women and 4.8 per 1000 person-years for Mexican American men. The hip fracture rates for the general population are 12.8 hip fractures per 1000 person-years for women and approximately 6.0 per 1000 person-years for men (19). Risk factors identified by Espino and colleagues (18) included age, female gender, living alone, previous stroke, and limitations in activities of daily living. Diabetes was not examined as an independent risk factor in this investigation. Given the prevalence of diabetes in the Hispanic population and the lack of previous research examining hip fracture in this ethnic group, we conducted a study examining whether diabetes was a risk factor for hip fracture in a large area probability sample of Mexican American older adults. Specifically, we examined whether diabetes was a risk factor for hip fracture after controlling for age, gender, body mass index (BMI), ever smoked, previous stroke, lower extremity functional ability, and distance vision. METHODS Sample Data are from the Hispanic Established Population for the Epidemiologic Study of the Elderly (H-EPESE) (20). The H-EPESE is an ongoing National Institute on Aging funded community-based study of 3050 older Mexican Americans. Mexican American subjects aged 65 and older were selected by area probability sampling techniques that involved selection of counties, census tracts, and households. A full description of the rationale, methods, and subject characteristics can be found elsewhere (12,20). The sample was designed to be representative of approximately 500,000 older Mexican Americans living in five southwestern states Texas, California, Arizona, Colorado, and New Mexico. The present study includes 2884 subjects at baseline (690 people with diabetes and 2194 people who did not have diabetes see detailed description below). Follow-up interviews were conducted 2 (n 2439), 5 (n 1979), and 7 years (n 1686) after baseline assessment. All interviews were done in the subject s home and were conducted in either Spanish or English. A description of the interview protocol is available in previous publications (12,20) and from the authors. Diabetes History of diabetes was assessed at the baseline interview. Subjects were asked if they ever had a physician diagnosis of diabetes. A total of 690 (23%) subjects reported a definite diagnosis of diabetes, 155 (5%) reported a borderline diagnosis of diabetes, and 2194 (72%) reported no history of diabetes. Subjects with a borderline diagnosis of diabetes were not included in the analysis. Subjects who reported using any medication during the baseline interview, including insulin, were requested to show the medication to the interviewer as a verification check. For the 690 subjects with a definite diagnosis of diabetes, 599 (87%) were taking medication for their diabetes, including 185 (27%) who were insulin-dependent. Hip Fracture Hip fracture was assessed over a 7-year follow-up period. Subjects were asked if they had a physician diagnosis of hip fracture since the last interview. At the 2-, 5-, and 7-year follow-up assessment interviews, 42, 48, and 44 first-time hip fractures were reported, respectively. Covariates Baseline sociodemographics included age and gender. Indicators of baseline health status were smoking history (ever smoked or nonsmoker), BMI (computed as weight in kilograms divided by the square of height in meters), history of stroke (yes or no), a summary performance measure of lower body function, and a test of distant visual acuity. The summary performance measure is comprised of three lowerbody activities, a timed 4-meter walk, rising from a chair five times, and a hierarchical standing balance task (21,22). Using previously established criteria (23), performance on each lower body activity was classified on a scale ranging from 0 to 4. When summed, the overall performance measure is scored from 0 to 12, where higher scores represent better lower-body functioning (23). A modified Snellen test using directional Es assessed distant visual acuity (24). Four visual categories were created and included subjects who were functionally blind ( 20/200), severely impaired ( 20/60 to 20/200), moderately impaired ( 20/40 to 20/ 60), and those with adequate vision ( 20/40). Statistical Analyses We examined demographic variables for all patients using descriptive and univariate statistics for continuous variables and contingency tables (chi-square) for categorical variables. Chi-square analyses were used to test for differences between patients with diabetes (n 690) and patients without diabetes (n 2194) at the baseline assessment interview, and hip fracture (n 134) versus no hip fracture (n 1443) at the follow-up assessment interviews. Cox proportional hazard models (25) in SAS (SAS Institute, Cary, NC) were used to estimate the hazard ratios (HRs) of hip fracture over 7 years by diabetes status at baseline interview, adjusting for relevant risk factors. We computed two Cox proportional hazard models. Model 1 examined the relationship between diabetes (definite diabetes vs absent) at baseline and new-onset hip fracture (present vs absent) at the three follow-up assessment interviews, adjusting for

3 M650 OTTENBACHER ET AL. age, gender, smoking status, BMI, and history of stroke. Model 2 examined the relationship between diabetes (definite diabetes vs absent) at baseline and new-onset hip fracture at the three follow-up assessment interviews, adjusting for the covariates described above, and a summary performance measure of lower body function and a test for distant vision. The two Cox proportional hazard models were reanalyzed using patients with diabetes who take insulin (definite diabetes and insulin-dependent vs absent) as a predictor variable. Gender diabetes and gender insulin diabetes interaction terms were created and were included in both models. A modified version of the Bonferroni correction was used to protect against type 1 errors in those cases where multiple univariate hypotheses were evaluated (26). In cases where univariate comparisons were made between unequal samples, a t test based on Levene s test that does not assume equal variances was calculated (27). Table 1. Selected Sociodemographic Characteristics of Subjects for the Overall Sample and by Diabetic Status at Baseline Interview Total Diabetes No Diabetes Selected Characteristics N % n % n % p Age Gender Men Women Ever Smoker No Yes Body Mass Index Stroke No Yes Summary Performance Measure Distant Vision Functionally blind Severely impaired Moderately impaired Adequate vision RESULTS The mean age of the sample at baseline was 71.8 years (SD 5.7). The sample included 1671 (58%) women and 1213 (42%) men. Six hundred and ninety subjects (23%) reported a definite diagnosis of diabetes at baseline, and 185 (27%) of these were taking insulin. Table 1 contains basic sociodemographic characteristics for persons with and without diabetes who had complete data for all variables included in the analysis at baseline. Persons with diabetes at baseline were younger (t 2.88, p.004), had a greater BMI (t 4.43, p.0001, see Figure 1), and were more likely to have had a stroke ( , p.0001). During the 7-year follow-up period, a total of 134 (5%) subjects reported a first-time hip fracture. The characteristics of persons reporting a hip fracture versus those reporting no hip fracture are presented in Table 2. The bivariate analysis of persons with and without diabetes versus those with and without hip fracture produced a statistically significant result ( , p.03, df 1) indicating the proportion of persons with hip fracture was higher in subjects with diabetes (Figure 2). We also computed a bivariate analysis using only those subjects with diabetes and taking insulin versus those without diabetes who experienced a hip fracture ( , p.01, df 1). The results of the Cox proportional hazard analyses are presented in Tables 3 and 4. The interaction term (gender diabetes) was not statistically significant in any of the models. Table 3 includes the proportional hazard results adjusted for age, gender, BMI, ever smoked, and previous stroke (Model 1), lower extremity functional ability, and distance vision (Model 2). Table 4 includes results for the same two Cox proportional hazard models except that the models include persons taking insulin rather that those only reporting a diagnosis of diabetes. The results reveal increased hazard ratios for hip fracture in persons with diabetes who are on insulin (HR 2.84, 95% confidence interval [CI 95 ] 1.49, 5.43, p.002) after adjusting for age, gender, BMI, ever smoked, and previous stroke. The hazard ratios remained statistically significant when potential explanatory variables such as lower extremity functional ability and distance vision were added to the models. As with any longitudinal study, there were missing values for some variables. For Table 3, 18 subjects who had a hip fracture had missing data in Model 1, and an additional six subjects had missing data in Model 2. In Table 4, 37 subjects who had a hip fracture had missing data in Model 1, and an additional six subjects who had a hip fracture had missing data in Model 2. For all variables except BMI, the proportion of missing data was evenly distributed across the diabetic and nondiabetic groups. Of the subjects with diabetes, 12.9% were missing BMI data, and 8.2% of the subjects without diabetes were Figure 1. Percentage of subjects with diabetes by body mass index (BMI) category.

4 DIABETES AS A RISK FACTOR FOR HIP FRACTURE M651 Table 2. Selected Baseline Sociodemographic Characteristics of Subjects by Hip Fracture Status at Follow-up (N 1577 ) Total Hip Fracture No Hip Fracture Selected Characteristics N % n % n % p Age Gender Men Women Ever Smoker No Yes Body Mass Index Stroke No Yes Summary Performance Measure Distant Vision Functionally blind Severely impaired Moderately impaired Adequate vision Due to missing values, sample size ranges from 1482 to missing BMI information. We compared subjects with missing and nonmissing BMI values on selected baseline characteristics. No statistically significant differences were observed for gender (p.49), marital status (p.42), or education (p.42). Subjects with missing BMI values were older (p.001) and were more likely to be diabetic (p.001). Table 3. Multivariate Analysis Assessing Risk of Hip Fracture by Diabetic Status Over a 7-Year Follow-up Period, Adjusting for Relevant Risk Factors Model 1 (N 2592) Cases 116 Model 2 (N 2521) Cases 110 Baseline Characteristics HR (95% CI) p HR (95% CI) p Diabetes (yes vs no) 1.57 ( ) ( ).07 Age (continuous) 1.06 ( ) ( ).004 Female versus Male 1.86 ( ) ( ).03 Ever Smoker 1.03 ( ) ( ).99 Body Mass Index 0.96 ( ) ( ).02 Stroke 1.21 ( ) ( ).93 Summary Performance Measure 0.91 ( ).003 Distant Vision 1.14 ( ).38 Note: HR hazard ratio; CI confidence interval. DISCUSSION Information on incidence and prevalence of hip fracture has been obtained primarily from the non-hispanic white population (28) and indicates the lifetime risk of hip fracture is about 18% in women and 6% in men (4). Hip fractures increase substantially with age: By 90 years, one-third of women and one-sixth of men will experience a hip fracture (29,30). Diabetes has not been considered a risk factor for hip fracture in older adults until recently (13). Adult-onset diabetes is generally associated with being overweight, and the protective effect of obesity against osteoporosis is presumed to reduce the risk of hip fracture (5,8). Multiple studies have demonstrated the relationship between low bone mass and increased fracture risk, mainly in the non-hispanic white population (1 3). Previous research, again in the non-hispanic white population, has reported relatively normal bone mass in persons with type 2 diabetes (15,16). We found increased risk of hip fracture in older Mexican Americans with self-reported diabetes when compared to hip fracture rates in Mexican American subjects without diabetes. The hazard ratio for experiencing a hip fracture for Mexican Americans taking insulin was particularly high Table 4. Multivariate Analysis Assessing Risk of Hip Fracture by Insulin Diabetic Status Over a 7-Year Follow-up Period, Adjusting for Relevant Risk Factors Model 1 (N 2151) Cases 97 Model 2 (N 2095) Cases 91 Baseline Characteristics HR (95% CI) p HR (95% CI) p Figure 2. Percentage of cases with and without diabetes who experienced a hip fracture over the 7-year follow-up period. Insulin Diabetes (yes vs no) 2.84 ( ) ( ).007 Age (continuous) 1.06 ( ) ( ).02 Female versus Male 1.70 ( ) ( ).09 Ever Smoker 1.12 ( ) ( ).82 Body Mass Index 0.95 ( ) ( ).01 Stroke 1.49 ( ) ( ).71 Summary Performance Measure 0.92 ( ).009 Distant Vision 1.05 ( ).77 Note: HR hazard ratio; CI confidence interval.

5 M652 OTTENBACHER ET AL. (HR 2.84, CI , 3.43, p.002) even when we controlled for factors normally associated with increased (or decreased) risk of hip fracture. These factors included age, gender, BMI, previous history of smoking, and previous stroke. The relationship remained statistically significant when possible explanatory variables were added to the models including summary lower extremity function and distance vision scores. Why did the older Mexican Americans with diabetes have more hip fractures than subjects without diabetes? Many factors associated with diabetes could lead to an increase in falls and associated fractures. These include visual impairment related to diabetic retinopathy, diminished proprioception in the lower extremities, poor balance, and unsteady gait caused by diabetic neuropathy and/or peripheral vascular disease. Any combination of these factors might increase the risk of falls in older adults and subsequent hip fracture (31). Type 1 diabetes has been associated with low bone density; however, most studies of type 2 diabetes have reported normal bone mass (14,15). The importance of bone quality, however, is relatively unexamined in type 2 diabetes. The vascular changes associated with diabetes could, theoretically, have a negative impact on bone remodeling and, therefore, on bone quality. The combination of poor bone quality and frequent falls might increase the risk of fracture, independent of bone mass (32). Falling is a risk factor for hip fracture, and poor depth perception has been reported as an independent risk factor for lower extremity fractures (33). There is a logical link between diabetic retinopathy, poor depth perception, falls, and fractures in persons with diabetes. Finally, failure to adequately manage glucose and medication levels in older persons taking insulin may also result in an increased risk of falls and subsequent fracture. These are not all the potential mediating mechanisms that link diabetes to increased risks of falls and hip fracture in Mexican American older adults, but they are areas where additional research is needed. We consider these findings preliminary, but believe they strongly suggest the need for additional research on diabetes as a risk factor for hip fracture, with a particular need for increased research on risk factors in the Hispanic population. The importance of our findings is based on the following strengths of the current investigation. We collected longitudinal information from a large, well-defined sample representative, at baseline, of 500,000 Mexican Americans in the southwestern United States (12,20). The reliability and consistency of the data collection procedures in the H-EPESE investigation are well-established (12,18). Our sample included both men and women, whereas most previous studies on risk factors for hip fracture have focused on non-hispanic white women. The H-EPESE sample represents a population (Mexican American) that is known to have a high rate of diabetes and related complications (12,18), but has not been comprehensively studied. Relatively little is known regarding health disparities and risk factors associated with hip fracture in this population (18). The limitations of the investigation include the fact that several key variables including hip fracture and diabetes were obtained by self-report. Previous researchers have reported good validity for diabetes self-reports confirmed by physician diagnosis (34,35). Individuals in this communitydwelling sample should have good recall of a major health event associated with hospitalization, such as a hip fracture. Receiving a diagnosis of diabetes, however, is a less dramatic health event. As noted previously, if a subject reported using insulin or other medications, the interviewer requested the subject present those medications. Thus, there was some interviewer verification of the diabetes diagnosis for persons taking insulin or other diabetic medications. A large number of the subjects with diabetes at baseline (87%) reported taking some type of diabetic medication. Those participants identified as not having diabetes probably included some individuals with undiagnosed diabetes. This misclassification would tend to weaken any association between diabetes and fracture. Insulin use can be considered as a marker for disease severity in subjects with diabetes. Duration of diabetes is also a potential marker for disease severity and a risk factor for hip fracture in this population. We did not have consistent information on duration of diabetes for the subjects in the H-EPESE sample, nor did we have any measure of glycemic control for the participating subjects. This study highlights the need for a more aggressive focus on identifying and addressing risk factors for hip fracture in the Hispanic population. Our findings suggest that diabetes may potentially be a more significant risk factor in older Mexican Americans than in other ethnic or racial groups. Research is needed concerning how diabetes and its complications contribute to increased risk of hip fracture in all populations, but specifically in Mexican Americans. The results of this research must then be incorporated into diabetes education and prevention programs, particularly those aimed at the Mexican American population. Acknowledgments This research was supported by Grants R01-AG10939, R01-AG17638, and P60-AG17231 from the Public Health Service, Department of Health and Human Services. Address correspondence to K. J. Ottenbacher, UTMB, 301 University Blvd., Route 1137, Galveston, TX Kottenba@utmb.edu References 1. Gillespie WJ. Clinical review: hip fracture. BMJ. 2001;322: Cummings SR, Rubin SM, Black DM. The future of hip fractures in the United States. Clin Orthop. 1990;252: Zuckerman JD. Review article: hip fracture. N Engl J Med. 1996;334: Brainsky A, Glick H, Lydick E, et al. The economic cost of hip fractures in community-dwelling older adults: a prospective study. J Am Geriatr Soc. 1997;45: Gullberg B, Johnell O, Kanis JA. World-wide projections for hip fracture. Osteoporos Int. 1997;7: Hannan E, Magaziner J, Wang JJ, et al. Mortality and locomotion 6 months after hospitalization for hip fracture: risk factors and riskadjusted hospital outcomes. JAMA. 2001;285: Boyce WJ, Vessey MP. Rising incidence of fracture of the proximal femur. Lancet. 1985;1: Wolinsky FD, Fitzgerald JF, Stump T. The effect of hip fracture on mortality, hospitalization, and functional status. Am J Public Health. 1997;87: Sloan FA, Taylor DH, Picone G. Costs and outcomes of hip fracture and stroke, 1984 to Am J Public Health. 1999;89: Hospital Inpatient Statistics Washington, DC: Agency for Health Care Policy and Research; AHCPR publication

6 DIABETES AS A RISK FACTOR FOR HIP FRACTURE M Jacobson SJ, Goldberg J, Miles T, et al. Regional variation in the incidence of hip fracture. JAMA. 1990;264: Black SA, Ray LA, Markides KS. The prevalence and health burden of self-reported diabetes in the Mexican American elderly: findings from the Hispanic EPESE. Am J Public Health. 1999;89: Nicodemus KK, Folsom AR. Type 1 and type 2 diabetes and incident of hip fractures in postmenopausal women. Diabetes Care. 2001;24: Kayath MJ, Tavares EF, Bid SA, Vieria JG. Prospective bone mineral density evaluation in patients with insulin-dependent diabetes mellitus. J Diabetes Complications. 1998;12: Meema HE, Meema S. The relationship of diabetes mellitus and body weight to osteoporosis in elderly females. Can Med Assoc J. 1967;96: Heath H, Melton LJ, Chu CP. Diabetes mellitus and risk of skeletal fracture. N Engl J Med. 1980;303: Schwartz AV, Sellmeyer DE, Ensrud KE, et al. Older women with diabetes have an increased risk of fracture: a prospective study. J Clin Endocrinol Metab. 2001;86: Espino DV, Palmer RF, Miles TP, et al. Prevalence, incidence, and risk factors associated with hip fractures in community-dwelling older Mexican Americans: results of the Hispanic EPESE. J Am Geriatr Soc. 2000;48: Kellie SE, Brody JA. Sex-specific and race-specific hip fracture rates. Am J Public Health. 1990;80: Markides KS, Rudkin L, Angel RJ, Espino DV. Health status of Hispanic elderly. In: Martin LJ, Soldo B, eds. Racial and Ethnic Differences in the Health of Older Americans. Washington, DC: National Academy Press; 1997: Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-extremity function in persons over the age of 70 years as a predictor of subsequent disability. N Engl J Med. 1995;332: Ostir GV, Markides KS, Black SA, Goodwin JS. Lower body functioning as a predictor of subsequent disability among older Mexican Americans. J Gerontol Med Sci. 1998;53A:M491 M Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol Med Sci. 1994;49:M85 M Salive ME, Guralnik J, Christen W, Glynn RJ, Colsher P, Ostfeld AM. Functional blindness and visual impairment in older adults from three communities. Ophthalmology. 1992;99: Cox DR. Regression models and life tables. J R Stat Soc B. 1972;34: Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc B. 1995; 57: SPSS. Statistical Package for the Social Sciences, Version Chicago, IL: SPSS Systems; Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. N Engl J Med. 1995;332: Hochberg MC, Williamson J, Skinner EA, et al. The prevalence and impact of self-reported hip fracture in elderly community-dwelling women: the Women s Health and Aging Study. Osteoporos Int. 1998; 8: Cameron I, Crotty M, Currie C, et al. Geriatric rehabilitation following hip fractures in older people: a systematic review. Health Technol Assess. 2000;4: Miller DK, Lui LY, Perry HM, Kaiser FE, Morley JE. Reported and measured physical functioning in older inner-city diabetic African Americans. J Gerontol Med Sci. 1999;54A:M230 M Nelson DA, Jacbober SJ. Editorial: why do older women with diabetes have an increased fracture risk? J Clin Endocrinol Metab. 2001;86: Seeley DG, Kelsey J, Jergas M, Nevitt MC. Predictors of ankle and foot fractures in older women. The Study of Osteoporotic Fractures research group. J Bone Miner Res. 1996;11: Kaye SA, Folsom AR, Sprafka JM, Prineas RJ, Wallace RB. Increased incidence of diabetes mellitus in relation to abdominal adiposity in older women. J Clin Epidemiol. 1991;44: Midthjell K, Holmen J, Bjorndal A, Lund-Larsen F. Is questionnaire information valid in the study of chronic disease such as diabetes? The Nord-Trondelg diabetes study. J Epidemiol Community Health. 1992; 46: Received February 12, 2002 Accepted May 1, 2002