| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Epidemiology and Prevention and Internal Medicine (D.E.B.), Wake Forest University School of Medicine, Winston-Salem, North Carolina 27157; Fred Hutchinson Cancer Research Center (J.C.L.), Seattle, Washington 98109; Department of Epidemiology and Biostatistics (A.V.S.), University of California San Francisco, San Francisco, California 94105; Department of Epidemiology (E.S.S.), University of Pittsburgh, Pittsburgh, Pennsylvania 15213; Department of Internal Medicine (J.R.), University of California, Davis, Davis, California 95817; Department of Geriatric Medicine (B.L.R.), University of Hawaii at Manoa, Honolulu, Hawaii 96817; Department of Preventive Medicine (K.C.J.), University of Tennessee Health Science Center, Memphis Tennessee 38105; and HealthPartners Research Foundation (K.L.M.), Minneapolis, Minnesota 55440-1524
Address all correspondence and requests for reprints to: Denise E. Bonds, M.D., M.P.H., Department of Epidemiology and Prevention, Wake Forest University School of Medicine, Medical Center Boulevard, Winston-Salem, North Carolina 27157. E-mail: dbonds{at}wfubmc.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
Objective: The objective of the study was to determine the risk of fracture in postmenopausal women with type 2 diabetes and determine whether risk varies by fracture site, ethnicity, and baseline bone density.
Design, Setting, and Participants: Women with clinically diagnosed type 2 diabetes at baseline in the Women’s Health Initiative Observational Cohort, a prospective study of postmenopausal women (n = 93,676), were compared with women without diagnosed diabetes and risk of fracture overall and at specific sites determined.
Main Outcome Measures: All fractures and specific sites separately (hip/pelvis/upper leg; lower leg/ankle/knee; foot; upper arm/shoulder/elbow; lower arm/wrist/hand; spine/tailbone) were measured. Bone mineral density (BMD) in a subset also was measured.
Results: The overall risk of fracture after 7 yr of follow-up was higher in women with diabetes at baseline after controlling for multiple risk factors including frequency of falls [adjusted relative risk (RR) 1.20, 95% confidence interval (CI) 1.11–1.30]. In a subsample of women with baseline BMD scores, women with diabetes had greater hip and spine BMD. The elevated fracture risk was found at multiple sites (hip/pelvis/upper leg; foot; spine/tailbone) among black women (RR 1.33, 95% CI 1.00–1.75) and women with increased baseline bone density (RR 1.26, 95% CI 0.96–1.66).
Conclusion: Women with type 2 diabetes are at increased risk for fractures. This risk is also seen among black and non-Hispanic white women after adjustment for multiple risk factors including frequent falls and increased BMD (in a subset).
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
Diabetes ascertainment
Our primary definition of diabetes was an affirmative answer to the question asked at baseline: did a doctor ever say that you had sugar diabetes or high blood sugar when you were not pregnant? or the reported use of a medication to treat diabetes at baseline. Participants with type 1 diabetes, defined as those diagnosed before age 20 yr or who were ever hospitalized for a diabetic coma, were excluded from the analysis (n = 185) as were those individuals with missing baseline diabetes information (n = 87). After these exclusions there were 5285 women with probable type 2 diabetes. To account for potential undiagnosed diabetics at baseline, we compared the results obtained with our primary definition to results obtained using two alternative definitions of diabetes: 1) affirmative answer to the diabetes question or the baseline medication inventory included a drug to treat diabetes or a glucose greater than 125 mg/dl (in those with an available baseline blood glucose) or 2) self-reported use of insulin or pills to treat diabetes at any time during the study. For more restrictive criteria, we defined diabetes in sensitivity analysis as an affirmative answer to the diabetes question at baseline, and the baseline medication inventory included a drug to treat diabetes.
Fracture ascertainment
Questions on fractures were included in the annual questionnaire. Participants were asked whether they had had a fracture since they last completed the medical history questionnaire and to identify the location (vertebral, shoulder, upper arm, lower arm, wrist, hip, upper leg, lower leg, and foot). All hip fractures were adjudicated by centrally trained adjudicators. Nonhospitalized fractures were collected through self-report only. A validation study of self-reported fractures was conducted on the WHI-OS (27) and found good agreement between self-report and adjudicated fracture for hip (78%) and forearm/wrist (81%) but lower agreement for clinical spine fractures (51%) (27). Fracture data through August 31, 2004, were included in this analysis.
BMD ascertainment
BMD was performed at three (n = 6,384) of the 40 clinical sites in the Women’s Health Initiative using dual-energy x-ray absorptiometry (QDR 2000, 2000+, or 4500W; Hologic, Inc., Bedford, MA). BMDs were obtained at baseline and yr 3, 6, and 9 (with fewer performed in yr 9). One set of spine, hip, and linearity phantoms was circulated during 1995 for measurement on the five QDR2000 scanners (two at Pittsburgh, two at Tucson, one at Birmingham). The variability in scanners did not warrant correction factors across sites. Multivariate models including BMD have been adjusted for scanner used to account for the slight differences found. For longitudinal changes in each scanner, correction factors were derived from spine and hip phantoms measured throughout the study (28). In Tucson, a QDR2000 scanner was replaced with a QDR4500W in 1999; the same upgrade was made in Birmingham in 2003. Before the upgrades, measurements were obtained on the same subjects on both the QDR2000 and QDR4500 (50 subjects in Tucson and 25 in Birmingham). Linear regression was used to derive correction factors from these in vivo cross-scanner measurements to allow conversion of the QDR4500 to QDR2000 values.
Other variables
Several possible confounding variables were obtained. Ethnicity, educational level, previous fractures, and history of osteoporosis were obtained from self-report at baseline. Participants were asked whether they were current, past, or never smokers. Alcohol use was also determined through survey, and the average number of drinks per week was determined at baseline. Physical activity was reported as the expenditure of energy from recreational physical activity reported at baseline (including walking, mild, moderate, and strenuous physical activity) and computed in metabolic equivalent task scores (29). Total calcium and vitamin D intake was obtained by combining the dietary intake as determined through the food frequency questionnaire administered at baseline and the daily intake from supplements. BMI was calculated using height and weight measured by clinic staff. To account for the increased risk of fracture in women with frequent falls, a time-dependent variable defined as three falls in the past 12 months at baseline or two or more falls in the last 6 months during follow-up was included.
Statistical analysis
Descriptive statistics, including means, frequencies, and percentages, were used to both describe the study population and look at differences between those participants with type 2 diabetes at baseline and those without (Table 1
). Comparisons for the continuous variables were done by running ANOVA models with a response of the continuous covariate of interest and an explanatory variable of type 2 diabetes at baseline (yes/no). Means and SD values are presented. For the categorical covariates, a 
2 test was used to evaluate differences in the distribution of the categorical levels of the covariate of interest between the diabetic and nondiabetic women. Frequencies and percentages are presented.
|
). A participant’s eligible follow-up was defined as the time from baseline to event if they had the fracture of interest and time from baseline to the date of death, loss to follow-up, or August 31, 2004 (whichever occurred first) for those without an event. Comparisons between the incidence rates of those with and without diabetes at baseline were done by running an unadjusted Cox proportional hazards model with a response of the fracture site of interest and explanatory variable of diabetes at baseline. A participant was counted once at the time of their first fracture for the any fracture outcome. Individuals could be included in several site-specific outcomes (i.e. wrist and hip) but only once for any particular site.
|
). For each fracture outcome, the first model was adjusted for age. Next, a second model was run, also adjusting for baseline height, weight, and falls as a time-dependent variable. Finally, the primary outcome was run, a fully adjusted model, adjusting for the same covariates as the first two with the addition of ethnicity, alcohol use, smoking, hormone use, physical activity, calcium/vitamin D intake, moderate to severe trouble seeing, history of fracture, history of osteoporosis, bisphosphonate use at baseline, steroid use, selective estrogen receptor modulator (SERM) use at baseline, insulin use, and thyroid hormone use. The hazard ratios for baseline diabetes along with its 95% confidence limits are presented for each model run. The Cox proportional hazards assumption was checked graphically, whereas collinearity was checked by analyzing a table of correlations between the various covariates in the model. All analyses were done in the SAS System for Windows (version 9.1; SAS Institute, Cary, NC).
|
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
), were less likely to be white, reported lower energy expenditure, and had more trouble seeing. They were equally likely to have a history of osteoporosis and slightly less likely to have a previous fracture but were more likely to have a history of a fall and be a current smoker. They also reported less calcium, vitamin D, and bisphosphonate use but were more likely to have used oral steroid hormones. Women with diabetes were shorter at baseline and heavier (Table 1) and, for the subset that underwent BMD measurements at baseline, had a higher hip and spine BMD (Table 2).
|
). When fractures were broken down by location, women with diabetes had a higher rate of hip/pelvis/upper leg, lower leg/ankle/knee, foot, upper arm/shoulder/elbow, and spine/tailbone fractures. There was an equal rate of fracture of the lower arm/wrist/hand reported by both groups.
Women with diabetes were 29% more likely to have suffered a fracture during the follow-up period (Table 4). This increased risk remained after adjustment for other baseline differences in the multivariate adjusted model, our primary outcome. When fractures by region were compared, women with diabetes had a significantly increased risk of the hip/pelvis/upper leg, foot, and spine/tailbone fracture (Table 4). There was also an increased risk of lower leg/ankle/knee and upper arm/shoulder/elbow fractures that did not meet statistical significance.
In sensitivity analyses, we reran the above analyses using our alternative definition of diabetes and found no significant variation in the results. We also used a cutoff of age 39 yr for diagnosis of diabetes instead of 20 yr and found no variation in the results. We then included only women who had baseline BMD measurements and added baseline hip BMD to the model. The addition of baseline hip BMD had little effect on the risk of any fracture and increased the risk of hip/pelvis/upper leg fractures despite the higher mean BMD for women with diabetes. We next ran the any-fracture model for those with type 2 diabetes who used insulin at baseline and found an increased risk of any fracture [relative risk (RR) 1.58, 95% confidence interval (CI) 1.18–2.11]. Finally, we stratified women by race [non-Hispanic white (NHW), black] to determine whether the risk varied. Using the same multivariate model described above, we found a somewhat higher risk of any fracture for black women with diabetes (RR 1.33, 95% CI 1.00–1.75), compared with that for NHW women with diabetes (RR 1.18, 95% CI 1.08–1.29).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
Our results support the findings of previous studies. When older women with diabetes were compared with women without diabetes in the Study of Osteoporotic Fractures, a 30–39% increased risk of a nonvertebral fracture was found (noninsulin user: RR 1.30, 95% CI 1.10–1.53; insulin user: RR 1.39, 95% CI 0.97–1.98) (19). Specific fracture types that were found to be increased in the Study of Osteoporotic Fractures included hip, proximal humerus, foot, and ankle fractures; as in our study, distal forearm fractures were not increased. In the Iowa Women’s Health Study, women with type 2 diabetes had a higher risk of hip fracture (RR 1.70, 95% CI 1.21–2.38) than women without diabetes after adjustment for multiple risk factors (11). Similar findings were seen in the Health ABC study, which found a 23% increased risk of hip fracture (RR 1.23, 95% CI 0.82–1.86) (21). This increased risk has also been demonstrated among Hispanic women (20) and Norwegian women (9, 10).
The underlying mechanism is not clear. Women with diabetes often suffer from neuropathy and retinopathy and thus are at greater risk of falls and the fractures that may result from these falls. In the WHI-OS cohort, more women with diabetes reported falls at baseline and during follow-up (44 vs. 32%), and more women with diabetes reported moderate or severe trouble seeing at baseline (11 vs. 5%). However, women with diabetes remained at greater risk for a fracture, even after adjustment for falls and difficulty seeing. It is possible that the women with diabetes experienced more severe falls or falls resulting from a different mechanism and thus have a higher risk of injury in any given fall or that they experienced a greater load on their bones from their increased weight. WHI-OS does not collect information on the severity or the mechanism of the fall. In a study on the mechanism of arm fractures, wrist fractures were more likely to result from falls that were obliquely forward, whereas upper arm injuries were associated equally with forward and lateral falls (30). Although the women with diabetes in our study had more falls, it is possible that the mechanism of the fall was such that they were less able to break their fall with their hand, thus decreasing their risk of wrist fractures.
Our study and others have reported increased BMD in women with diabetes, perhaps caused by the increased body weight during adolescence and early 20s, the years of peak bone formation. Thus, the increased risk does not appear to be due to an increased risk of osteoporosis. It is possible that the BMD of women with diabetes may be overestimated due to measurement error caused by the increased BMI (31, 32). We did find an increase in spine BMD over time in both groups. This is a common finding attributable to an increase in aortic calcifications, osteophytes, and other degenerative changes rather than an actual gain in bone mass (33, 34). Animal models have indicated that although the bone density is greater in diabetes, the bone structure is more fragile, with fractures occurring under a smaller load and the bones exhibiting reduced mechanical indices (35). The higher serum level of glucose in women with diabetes may result in a larger concentration of advanced glycation end products in collagen-containing tissues such as bone (36). Advanced glycation end products have been associated with decreased strength in human cadaver femurs (37). Thus, women with diabetes may be more likely to suffer a fracture from a fall or minor trauma due to the decreased bone strength. There may be increased bone loss in women with diabetes due to lower levels of IGF-I (38), hypercalciuria secondary to elevated glucose in the urine (39), or increased inflammation (40). In the subsample of women with diabetes who underwent bone density testing, we found a similar increased risk of any fracture despite the increased bone density of the women with diabetes at baseline.
It is not clear what can be done to prevent the increased fracture rate seen among women with diabetes. Observational studies have found a greater risk of fractures among women with diabetes that have higher fasting glucose levels (22), suggesting that better control of blood glucose may reduce the risk. One study that examined the effect of alendronate on BMD found it to be effective in increasing bone density in women with diabetes (41). However, this study was limited to women who had a low bone density at baseline. Additional studies testing methods to prevent fractures in postmenopausal diabetic women are needed, including studies that assess biochemical markers of bone turnover.
There are several limitations to this study. There was no confirmation of the self-reported diabetes diagnosis with medical records, as was also the case with most previous studies. Based on baseline fasting glucose levels greater than 125 mg/dl in women who did not self-report diabetes, 3% of women in the WHI-OS had undiagnosed diabetes. We tested a variety of definitions of diabetes in sensitivity analysis and had similar results, regardless of the definition used. The women with diabetes who participate in the WHI-OS may be healthier than the general population and thus not representative of all women with diabetes. There may be under- or overreporting of fractures. To minimize bias in the comparison of fracture rates, it is most important to identify false-positive reports and achieve a high specificity in the adjudication of fractures. With 100% specificity, underascertainment will not attenuate RR estimates (42). For hip fractures, central adjudication eliminated virtually all false-positive reports, and we can therefore be most confident of a lack of bias in this outcome. As described in Subjects and Methods, a validation study of fracture was conducted in the WHI-OS (27). Whereas this study did not examine results separately for women with and without diabetes, there is no reason to believe that there would be a systematic difference. Women with diabetes may see physicians more frequently than women without and consequently may have clinical vertebral fractures diagnosed more frequently. Whereas we did see a higher rate of vertebral fractures in women with diabetes, the consistently higher rates of fractures at all sites make this less likely. Finally, no measurements of visual acuity or neuropathy were available, and it was necessary to rely on self-report. Diabetic women may underreport decreased visual acuity, which is a known risk factor for fractures (43).
|
Conclusion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Disclosure statement: The authors have nothing to disclose.
First Published Online June 27, 2006
Abbreviations: BMD, Bone mineral density; BMI, body mass index; CI, confidence interval; NHW, non-Hispanic white; RR, relative risk; SERM, selective estrogen receptor modulator; WHI-OS, Women’s Health Initiative Observational Study.
Received March 20, 2006.
Accepted June 15, 2006.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionConclusionReferences |
|---|
1 genotypes and bone turnover in premenopausal women with diabetes mellitus. Diabetologia 41:1314–1320[CrossRef][Medline]Find additional patient-related information at:
This article has been cited by other articles:
 |
|
 |
|
  Y. K. Loke MBBS MD, S. Singh MD MPH, and C. D. Furberg MD PhD Long-term use of thiazolidinediones and fractures in type 2 diabetes: a meta-analysis Can. Med. Assoc. J., January 6, 2009; 180(1): 32 - 39. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. J. Melton III, B. L. Riggs, C. L. Leibson, S. J. Achenbach, J. J. Camp, M. L. Bouxsein, E. J. Atkinson, R. A. Robb, and S. Khosla A Bone Structural Basis for Fracture Risk in Diabetes J. Clin. Endocrinol. Metab., December 1, 2008; 93(12): 4804 - 4809. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Tao, J.-M. Liu, H.-Y. Zhao, L.-H. Sun, W.-Q. Wang, X.-Y. Li, and G. Ning Differences between Measurements of Bone Mineral Densities by Quantitative Ultrasound and Dual-Energy X-Ray Absorptiometry in Type 2 Diabetic Postmenopausal Women J. Clin. Endocrinol. Metab., May 1, 2008; 93(5): 1670 - 1675. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Kahn, B. Zinman, J. M. Lachin, S. M. Haffner, W. H. Herman, R. R. Holman, B. G. Kravitz, D. Yu, M. A. Heise, R. P. Aftring, et al. Rosiglitazone-Associated Fractures in Type 2 Diabetes: An analysis from A Diabetes Outcome Progression Trial (ADOPT) Diabetes Care, May 1, 2008; 31(5): 845 - 851. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Meier, M. E. Kraenzlin, M. Bodmer, S. S. Jick, H. Jick, and C. R. Meier Use of Thiazolidinediones and Fracture Risk Arch Intern Med, April 28, 2008; 168(8): 820 - 825. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Monami, B. Cresci, A. Colombini, L. Pala, D. Balzi, F. Gori, V. Chiasserini, N. Marchionni, C. M. Rotella, and E. Mannucci Bone Fractures and Hypoglycemic Treatment in Type 2 Diabetic Patients: A case-control study Diabetes Care, February 1, 2008; 31(2): 199 - 203. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Grey, M. Bolland, G. Gamble, D. Wattie, A. Horne, J. Davidson, and I. R. Reid The Peroxisome Proliferator-Activated Receptor-{gamma} Agonist Rosiglitazone Decreases Bone Formation and Bone Mineral Density in Healthy Postmenopausal Women: A Randomized, Controlled Trial J. Clin. Endocrinol. Metab., April 1, 2007; 92(4): 1305 - 1310. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
