PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Bone Miner Res. Author manuscript; available in PMC 2013 November 1.
Published in final edited form as:
PMCID: PMC3474871
NIHMSID: NIHMS389264

The Association of Concurrent Vitamin D and Sex Hormone Deficiency with Bone Loss and Fracture Risk in Older Men: The MrOS Study

E Barrett-Connor, MD,1 GA Laughlin, PhD,1 H Li, MD,2 CM Nielson, PhD,2 PY Wang, MS,2 TT Dam, MD,1,3 JA Cauley, PhD,4 KE Ensrud, MD, MPH,5 ML Stefanick, PhD,6 E Lau, MD,7 AR Hoffman, MD,6 and ES Orwoll, MD2, for the Osteoporotic Fractures in Men (MrOS) Research Group

Abstract

Low 25-hydroxyvitamin D (VitD), low sex hormones (SH), and high sex hormone binding globulin (SHBG) levels are common in older men. We tested the hypothesis that combinations of low VitD, low SH, and high SHBG would have a synergistic effect on bone mineral density (BMD), bone loss, and fracture risk in older men. Participants were a random subsample of 1468 men (mean age 74) from the Osteoporotic Fractures in Men Study (MrOS) plus 278 MrOS men with incident non-spine fractures studied in a case-cohort design. “Abnormal” was defined as lowest quartile for VitD (<20 ng/ml), bioavailable testosterone (BioT, <163 ng/dl), and bioavailable estradiol (BioE, <11 pg/ml); and highest quartile for SHBG (>59 nM).

Overall, 10% had isolated VitD deficiency; 40% had only low SH or high SHBG; 15% had both SH/SHBG and VitD abnormality, and 35% had no abnormality. Compared to men with all normal levels, those with both SH/SHBG and VitD abnormality tended to be older, more obese, and to report less physical activity. Isolated VitD deficiency, and low BioT with or without low VitD, was not significantly related to skeletal measures. The combination of VitD deficiency with low BioE and/or high SHBG was associated with significantly lower baseline BMD and higher annualized rates of hip bone loss than SH abnormalities alone or no abnormality. Compared to men with all normal levels, the multivariate-adjusted hazard ratio (95% CI) for incident non-spine fracture during 4.6 yr median follow-up was 1.2 (0.8–1.8) for low VitD alone; 1.3 (0.9–1.9) for low BioE and/or high SHBG alone; and 1.6 (1.1–2.5) for low BioE/high SHBG plus low VitD.

In summary, adverse skeletal effects of low sex steroid levels were most pronounced in older men with low VitD levels. The presence of low VitD in the presence of low BioE/high SHBG may contribute substantially to poor skeletal health.

Keywords: bone loss, fracture, older men, sex hormones, vitamin

INTRODUCTION

Circulating 25 hydroxyvitamin D [25(OH)D, VitD] levels decrease with age. In the MrOS cohort 26 % of community-dwelling men aged 65 and older had VitD deficiency defined as levels <20 ng/ml[1]. VitD is essential for bone health[2]. In previous publications from the MrOS group, VitD below 20 ng/ml was associated with greater rates of hip bone loss and hip fracture[1, 3] but observational studies have generally shown inconsistent results about VitD levels as an independent predictor of fracture[1, 47]; and the effectiveness of VitD alone in fracture prevention is unclear[6, 810].

Endogenous sex hormone levels also decrease with age, paralleling increasing levels of sex hormone binding globulin (SHBG). Evidence from cross-sectional studies conducted in the past several years indicates that estrogen plays a major, and likely dominant, role in bone metabolism in men[11]. In contrast to the contradictory data on endogenous VitD levels and fracture, most prospective studies of estradiol (E2) and testosterone (T) in men reported that E2, not T, was the best independent predictor of bone loss [12] [13], [14], [15].

Another study[16] reported that low total E2 was associated with increased risk of hip fracture; men with both low E2 and low T had the greatest risk; there were only 39 hip fractures. Low E2 but not T was associated with radiologically confirmed spine fracture in one prospective study[17], and low bioavailable estradiol (BioE) and high SHBG predicted increased non-spine and hip fracture risk in two studies[18, 19], but not in another with relatively few spine fractures [20]. In contrast, another study found that low T, but not low E2, was independently related to fracture risk[21]. Center and colleagues examined the influence of several hormonal factors including sex hormones, SHBG, and vitamin D and found that abnormalities in multiple factors greatly increased the risk of osteoporosis[22] and major fracture[23]. Most[1821, 23], but not all[16, 17], of these fracture studies adjusted for baseline bone mineral density (BMD).

To our knowledge, no previous studies have reported the impact of combined abnormalities in sex steroids and VitD deficiency on bone health outcomes in older men. The present study was designed to test the hypothesis that low VitD, low sex hormones, and high SHBG have a synergistic association with BMD, bone loss, and fracture risk in a prospective study of older men.

METHODS

Study Population

From March 2000 through April 2002, 5994 community-dwelling men at six clinical centers in the United States (Birmingham, Alabama; Minneapolis, Minnesota; Palo Alto, California; Monongahela Valley near Pittsburgh, Pennsylvania; Portland, Oregon; and San Diego, California) participated in the baseline examination for MrOS. Eligible men were at least 65 years of age, without bilateral hip replacements, and able to walk without the assistance of another person. Details of the MrOS design and cohort have been published[24, 25]. The Institutional Review Board (IRB) at each center approved the study protocol, and written informed consent was obtained.

For the prospective study, men were sent questionnaires tri-annually to report any falls or fractures during a median 4.6 year follow up. The median interval for repeat BMD and fracture follow up in the random cohort subsample was similar (4.7 years). All fractures were verified by medical records and confirmed by blinded central adjudicators[26]. Pathologic fractures were excluded. Spine fractures were not included in the primary analyses, based on the low proportion of radiographic spine fractures that are symptomatic. To assess new clinical vertebral fractures, all thoracic and lumbar spine fractures that were reported by participants were adjudicated centrally. To adjudicate, the radiological image (X-ray or, less often, MRI) used to diagnose the fracture was obtained from the participant’s physician in the community and sent to the Coordinating Center. The study radiologist confirmed the presence of a new vertebral fracture using these community-acquired images, comparing them to lateral thoracic and lumbar spine X-rays that had been obtained for all MrOS participants at baseline. A new clinical vertebral fracture was defined as a change in semiquantitative (SQ) grade of at least 1 between the baseline film and the follow-up image[27, 28].

Both high and low trauma fractures were included because studies in older adults show both types of fractures predict future fractures and are therefore likely to be osteoporotic fractures[29, 30]. Secondary analyses tested associations with new (within study) major osteoporotic fractures which included clinical fractures of the hip, arm, wrist, and lumbar and thoracic spine.

The overall study design is case-cohort(29); a flow chart describing the study population is presented in Figure 1. Of the 5994 MrOS participants, 1596 men were randomly selected for hormone and VitD assays and constituted the random sub-cohort. To increase statistical power for fracture analyses, an additional 310 MrOS men with incident non-spine fractures were included from outside the random sub-cohort; sex hormone and VitD assays were also performed in this case-cohort design[31]. After exclusion of 161 men who did not have sex hormone or VitD data, the final sample sizes were 1468 for the BMD analyses, and 1746 for the fracture analyses. A total of 374 fractures were included in the non-spine fracture analyses, 278 cases came from the case cohort and 96 from the random sub-cohort. A detailed list of non-spine fracture types and numbers is presented in Supplementary Table 1. A total of 134 fractures were included in the major osteoporotic fracture analyses, including 69 hip, 18 arm, 31 wrist and 16 lumbar and thoracic spine fractures as presented in Supplementary Table 2. The reason why the numbers of fractures differ by location in the supplementary tables is because the non-spine fracture analysis included all first non-spine fractures that occurred within the study; i.e. men who had clinically less important fractures as their first fracture (e.g., ribs, fingers, toes, and face) were included in the primary analysis. The secondary analysis included the first major validated clinical fracture occurring during the study (arm, wrist, spine, or hip).

Figure 1
Flowchart of MrOS participant selection.

Hormone assays

Fasting morning blood was collected; serum was stored at −70°C until thawed for assays. A combined gas chromatographic negative ionization tandem mass spectrometry and liquid chromatographic electrospray tandem mass spectrometry bioanalytical method was used to measure T and E2 (Taylor Technology, Princeton NJ); the limit of detection was 0.625 pg/ml for E2 and 2.5 ng/dl for T. The intra-assay and inter-assay coefficients of variation (CV), respectively, were 2.5 and 6% for T and 6.4 and 10.1% for E2. SHBG concentration was determined on an Immulite analyzer with chemiluminescent substrate (Diagnostic Products Corp., Los Angeles, CA). The standard curve ranged from 0.2–180 nmol/l with intra-assay CV of 4.6% and inter-assay CV of 5.8%. Calculation of the bioavailable fractions of T and E2l used the methods described by Södergård and colleagues[32]. Measures of 25(OH) VitD2 and 25(OH) VitD3 were performed at the Mayo Clinic using liquid chromatography-tandem mass spectrometry as previously described [33]. 25(OH) D2 and 25(OH) D3 were quantified, reported individually and summed for total 25(OH) D (VitD). The minimum detectable limit was 4 ng/ml for 25(OH) D2 and 2 ng/ml for 25(OH) D3. The inter-assay CV was 4.4%; intra-assay CV, 4.9%. VitD “deficiency” was defined as total 25(OH) D <20 ng/ml[34]. Estrogen and T therapy were not considered, and seem unlikely given the highest circulating levels observed here.

Measurement of Bone Mineral Density and Body Composition

Bone mineral density (BMD) (g/cm2) of the total hip was measured using dual-energy x-ray absorptiometry (DXA) (QDR 4500W, Hologic Inc., Bedford, Mass). Standardized procedures for participant positioning and scan analysis were used by centrally certified DXA operators. Densitometry technicians at the Coordinating Center reviewed a random sample of all the scans to assure adherence to standardized techniques. Percent body fat was measured using whole body DXA scans. Hip phantom scan results were assessed for longitudinal and cross-sectional quality control. The intra-clinic CV for hip phantoms (0.37% to 0.58%) were within acceptable limits. The inter-clinic CV was 0.9%, and the maximum difference between means was 2.2%. To adjust for inter-clinic differences, statistical models included indicator variables for the individual scanners[25].

Other Covariates

Characteristics assessed at baseline included age, race/ethnicity, history of fracture, falls in the past year, smoking, self-rated health, alcohol consumption, and weight at age 25. Trained, certified clinical staff measured weight and height and obtained blood samples. The Physical Activity Score for the Elderly (PASE)[35] was self-administered. Height (cm) was measured on Harpenden stadiometers, and weight (kg) on regularly calibrated balance beam or digital scales using standard protocols, with participants wearing light clothing without shoes. Body mass index (BMI) was calculated as kg/m2. Estimated glomerular filtration rate (eGFR) was calculated by the Modification of Diet in Renal Disease equation for Caucasians[36]: eGFR (mL/min/1.73m2) = 186 × (serum creatinine [mg/dL]) −1.154 × (Age[years]) −0.203.

VitD and calcium intake over the year before baseline were examined using a modified Block Food Frequency questionnaire which included diet and supplement intake (Block Dietary Data Services, Berkeley, CA)[37]. Participants were not excluded or treated any differently if they reported supplement use or had higher serum 25(OH)D2 levels. Instead, total serum 25(OH)D measures were used in the analysis. This decision to include supplements was based in part on the very small percent of the circulating D levels explained by supplements in the context of the very large percent explained by sunlight exposed skin; an earlier MrOS study reported that those who reported supplement use had only slightly higher total 25(OH)D levels but more often had detectable levels of 25(OH)D2[38]. Values for participants who reported < 400 kcal per day were recorded as missing. At baseline, participants were asked to bring in all medications used within the last 30 days. All prescription medications recorded by the clinics were stored in an electronic medications inventory database. Each medication was matched to its ingredient(s) based on the Iowa Drug Information Service (IDIS) Drug Vocabulary (College of Pharmacy, University of Iowa, Iowa City, IA)[39].

Statistical Analyses

Most analyses including baseline characteristics, BMD associations, and bone loss were examined in the randomly selected sub-cohort only. The risk of non-spine fracture and major osteoporotic fracture was evaluated in the case-cohort sample, designed to increase statistical power. Baseline characteristics, hip BMD profile and fractures were examined among mutually exclusive groups based on normal/abnormal VitD and sex hormone levels. We defined quartiles on the basis of the distribution in the random sub-cohort; “abnormal” was defined as the lowest quartile for VitD (<20 ng/ml), BioT (<163 ng/dl), and BioE (<11 pg/ml) and the highest quartile for SHBG (>59 nM). The counterparts of these cut points were considered as normal levels. Groups were further stratified based on specific sex hormone/SHBG abnormalities for the BMD analyses.

Overall baseline characteristics and distribution of sex hormone abnormality and VitD deficiency in random sub-cohort were summarized using frequency/percentage for categorical variables and mean/standard deviation for continuous variables. Characteristics were compared across sex hormone/VitD groups using ANOVA for continuous variables and Chi-square test for categorical variables. Variables with P < 0.05 in univariate analysis or known to be covariates of bone loss and fracture from published literature (age, race/ethnicity, BMI, study site, season of blood draw, ever smoked, 7+ alcohol drinks/wk, physical activity (PASE), self-rated health, eGFR, and history of diabetes) were included in multivariate analyses of BMD and fracture risk.

Two measures of BMD were used, and the difference was annualized over the follow-up period. Baseline BMD and its annualized-percentage change (percent difference between BMD at the second and first clinical visits divided by years of follow up) were compared across groups using ANOVA, and age-adjusted and multivariate-adjusted least square means were estimated. The potential effect of regression to the mean was not considered since it should not influence group comparisons. For the fracture analyses, some categories in mutually exclusive groups of VitD and sex hormone were combined based on results from BMD analysis. The time from baseline to first within-study non-spine fracture among groups was evaluated using Cox proportional hazard models using a weighting method to accommodate the sampling in case-cohort design[40]; men without a fracture event were censored at the time of last visit. Covariates were selected as described above for the BMD analyses, and baseline total hip BMD was included as an additional adjustment in secondary fracture analyses. Age-adjusted and multivariate-adjusted hazard ratios and 95% confidence limits were estimated based on a robust sandwich method[41]. For those exposure groups that showed statistically significant associations with fracture, we calculated percent attributable risk in the exposed.

All analyses were performed using SAS statistical software (Version 9.2; SAS Institute Inc, Cary, NC). All reported P values are 2-sided with a value of P < 0.05 considered statistically significant.

RESULTS

At the baseline visit, the 1468 MrOS men in the random sub-cohort had a mean (SD) age of 74; 91% were white. A majority (85%) rated their health as good or excellent. A total of 12% had at least mild chronic kidney disease based on the glomerular filtration rate <60 ml/min/1.73m2 cut-point suggested by the National Kidney Foundation[42]. One quarter of men had a low VitD of <20 ng/dl; the mean level in VitD-deficient men was 15.5 ng/ml.

Table 1 presents baseline characteristics for the random sub-cohort stratified into 4 groups based on normal versus abnormal sex hormone and/or VitD levels. One out of three men had normal levels of VitD and sex hormones, almost 40% had one or more abnormal sex hormone levels but normal VitD levels, 10% had low VitD only, and 15% had both low VitD and sex hormone abnormalities. Several factors were associated with low VitD, abnormal sex hormones, or both (Table 1) and were considered in multivariate analyses along with variables known to be associated with osteoporosis (alcohol use, physical activity, and smoking history).

Table 1
Baseline Characteristics by Vitamin D and Sex Hormone Groups from Random Sub-Cohort Sample

For the BMD analyses, groups were further stratified based on specific sex hormone/SHBG abnormalities. Among all the men, 12% had low BioT alone, 10% had low BioE alone, 16% had high SHBG alone, and 17% had more than 1 abnormal sex hormone (regardless of VitD status). Table 2 shows the age and multivariate adjusted baseline hip BMD and hip BMD change by these mutually exclusive groups. Compared to men with all normal levels, men with low VitD had similar hip BMD at baseline, while men with low BioE or high SHBG (regardless of low VitD) had low BMD at baseline. After adjusting for age, ethnicity, and other covariates, the annualized percentage BMD loss was significantly higher only for those who had low VitD plus low BioE or high SHBG. The rates of annualized BMD loss at the hip were twice as high for men with the concurrence of low D with low BioE or high SHBG, than for men with only low BioE or high SHBG.

Table 2
Hip BMD by Sex hormone/VitD group: random sub-cohort sample

Based on these results, men with low BioE and/or high SHBG were combined and analyses were repeated for hip BMD and hip BMD change (Figure 2). Compared to men with all normal levels, BMD and BMD change did not differ in men with low VitD only, or in those with low BioT with or without low VitD. Men with low BioE and/or high SHBG without low VitD had significantly lower baseline BMD (Figure 2A) and higher percent hip bone loss (Figure 2B) compared to men with all normal levels. However, the lowest baseline BMD and greatest percent hip bone loss was observed in men with low BioE and/or high SHBG concurrent with low VitD.

Figure 2Figure 2
BMD and BMD loss by sex hormone/VitD groups: random-cohort sample. Panel A. Baseline Total Hip BMD by sex hormone/VitD group. Panel B. Total Hip BMD loss by sex hormone/VitD group. Adjusted for age, race, latitude of study site, season of blood draw, ...

Figure 3A shows results of the analysis of non-spine fracture risk using the case-cohort sample, which included the subset of MrOS men with fractures from outside the random cohort, using the same VitD/SH/SHBG groupings as those in Figure 2. Risk estimates are adjusted for age, race, latitude of study site, physical activity (PASE), season of blood draw, BMI, ever smoked, alcohol drinks per week, self-rated health condition, kidney function (eGFR), and history of diabetes. Fracture risk for men with isolated VitD deficiency, or those with low BioT with or without low VitD, did not differ from risk for men without VitD deficiency or SH/SHBG abnormality. Risk of fracture for men with normal VitD but low BioE and/or high SHBG was higher, but differences between this group and the “normal” group were not statistically significant (P=0.08). Significantly higher fracture risk was detected in the men with low BioE and/or high SHBG concurrent with a low VitD (adjusted HR, 95% CI: 1.62, 1.05–2.51).

Figure 3Figure 3
Risk of non-spine fracture (Panel A) and major osteoporotic fracture (Panel B) by sex hormone/VitD groups: case-cohort sample. Adjusted for age, race, latitude of study site, season of blood draw, BMI, ever smoked, alcohol drinks per week, self-rated ...

Figure 3B shows results of the same analyses with validated new within-study major osteoporotic fracture as the outcome. In general, results were similar to those for any non-spine clinical fracture, although the risk of fracture for men with normal VitD but low BioE and/or high SHBG was statistically significant (adjusted HR, 95% CI: 1.89, 1.04–3.46) and the risk of fracture in men with low BioE and/or high SHBG plus low VitD was stronger (adjusted HR, 95% CI: 3.94, 2.05–7.57) than for non-spine fractures. The addition of spine fractures is unlikely to explain these results because only 16 men had a new spine fracture during the study.

Adjustment for baseline hip BMD did not materially change results for either non-spine fractures or major osteoporotic fractures (data not shown).

DISCUSSION

In previous publications from the MrOS group, VitD below 20 ng/ml was associated with greater rates of hip bone loss and hip fracture[1, 3] but observational studies have generally shown inconsistent results about VitD levels as an independent predictor of fracture [1, 4, 5]; and the effectiveness of VitD alone in fracture prevention is unclear [610]. In this analysis of older community-dwelling men followed for a median of 4.6 years, VitD deficiency (< 20 ng/ml) without a concomitant sex steroid hormone abnormality was not associated with low BMD, BMD loss, or non-spine fracture, but only 10% of these men had isolated VitD deficiency. This may reflect several characteristics of these men; they were community-dwelling and ambulatory and relatively healthy. Also, they were largely white and not obese -- two factors associated with higher VitD levels. The absent association of isolated VitD deficiency with adverse bone outcomes persisted after appropriate adjustments.

More than 40% of these older men had low BioE and/or high SHBG levels; as previously reported [3, 19]; these men showed an increased prevalence of lower BMD at baseline, a higher rate of bone loss, and increased risk for fracture. Men with the concurrence of low VitD with low BioE and/or high SHBG had significantly lower hip BMD, higher rate of hip BMD loss/year, and higher risk for fracture than men with only low BioE and/or high SHBG (shown in Figures 2 and and3).3). The risk associated with endocrine deficiencies was stronger for new major clinical fractures than non-spine fractures. Neither isolated VitD deficiency nor low bioT with or without low VitD were significantly associated with BMD, bone loss, or fracture risk. The effect of combined endocrine deficiency on fracture risk was not materially altered by adjustment for baseline BMD, suggesting that multiple endocrine deficiencies may influence fracture risk through mechanisms in addition to BMD or BMD loss.

To our knowledge this is the first study to show that the combination of VitD deficiency and low estrogen and/or high SHBG may improve the prediction of low bone density, bone loss, and fracture risk. The combination of low estrogen and/or high SHBG with low vitamin D occurred in 7% of our cohort, and thus represents an important segment of older men. Since MrOS is a volunteer cohort, the fraction of men affected may even be larger in the general population.

Strengths of this study include the well-characterized cohort and excellent central laboratory assessment for VitD and for sex hormones. Our definition of vitamin D deficiency is the same as that recommended recently by the Institute of Medicine [7]. Tri-annual mailers with a more than 99% response rate (data not shown) likely improved recall of incident clinical fractures, which were validated by medical review in all cases included here. Their Medicare eligible age and socioeconomic status likely improved both access to health care and medical literacy, reducing confounding due to variability in these factors. Baseline data were collected before the increasingly heavy media attention to multiple putative benefits of VitD therapy, making reverse causality a less likely explanation for these findings. Analyses were adjusted for age, race/ethnicity, BMI, study site, season of blood draw, ever smoked, 7+ alcohol drinks/wk, physical activity (PASE), self-rated health, estimated glomerular filtration rate (eGFR), and history of diabetes--characteristics known to be associated with bone health.

The limitations of this study are the homogeneous SES/education and race/ethnic characteristics of the cohort, such that results may not be generalizible to other populations. It is plausible that men with low VitD levels are sicker, therefore exposed to less sunlight, with poorer bone health. We were able to control for this to some extent by adjusting for physical activity. Nevertheless, study participants with isolated low VitD levels had the poorest self-rated health, as shown in Table 1. Although it would be interesting to consider whether the effects of combined endocrine deficiency were more pronounced in men with low bone mass or osteoporosis, the study population was not large enough to isolate those groups for analysis. In addition, multiple comparisons were performed and we cannot exclude the possibility that some of our results may be due to chance, although the consistency of findings across outcomes argues against this possibility.

In conclusion, although VitD is essential for bone health, observational studies have shown inconsistent results about VitD levels as an independent predictor of fracture (1, 4, 5); and clinical trial results have also been inconsistent [810]. We found that adverse skeletal effects of VitD deficiency were apparent only in the presence of low E2 or high SHBG. This observation is based on the very few men who had VitD deficiency with normal levels of sex steroids. Published inconsistent associations of VitD and bone health should be re-examined in light of these findings, and given the wide divergence in the prevalence of VitD deficiency in the literature [43]. These results suggest a multifactorial approach may yield optimal estimates of endocrine influences on bone health.

Supplementary Material

Supp Table S1-S2

Acknowledgments

Funding Support: The Osteoporotic Fractures in Men (MrOS) Study is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), the National Institute on Aging (NIA), and the National Center for Research Resources (NCRR) and the NIH Roadmap for Medical Research through grants U01 AR45580, U01 AR45614, U01 AR45632, U01 AR45647, U01 AR45654, U01 AR45583, U01 AG18197, U01 AG027810, and UL1 RR024140.

Dr. Barrett-Connor has also received grant support from NIH (National Institutes of Health/National Institute on Aging) grants AG07181 and AG028507 and the National Institute of Diabetes and Digestive and Kidney Diseases, grant DK31801.

Footnotes

Conflict of Interest: All authors state that they have no conflicts of interest.

Authors’ roles: Study design: EBC, GAL, JAC, KEE, ESO, and TTD. Study conduct: EBC, GAL, JAC, KEE, and ARH. Data collection: EBC and JAC. Data analysis: EBC, GAL, CMN, PYW, JAC, and HL. Data interpretation: EBC, GAL, CMN, PYW, JAC, KEE, and HL. Drafting manuscript: EBC, GAL, and HL. Revising manuscript content: EBC, GAL, CMN, PYW, JAC, KEE, and ESO. Approving final version of manuscript: EBC, GAL, CMN, PYW, JAC, KEE, JAC, ARH, ESO, TTD, MLS, and EL. EBC, GAL, and ARH take responsibility for the integrity of the data analysis.

References

1. Cauley JA, Parimi N, Ensrud KE, et al. Serum 25-hydroxyvitamin D and the risk of hip and nonspine fractures in older men. J Bone Miner Res. 2010;25(3):545–53. [PMC free article] [PubMed]
2. DeLuca HF. Overview of general physiologic features and functions of vitamin D. Am J Clin Nutr. 2004;80(6 Suppl):1689S–96S. [PubMed]
3. Ensrud KE, Lewis CE, Lambert LC, et al. Endogenous sex steroids, weight change and rates of hip bone loss in older men: the MrOS study. Osteoporos Int. 2006;17(9):1329–36. [PubMed]
4. Roddam AW, Neale R, Appleby P, Allen NE, Tipper S, Key TJ. Association between plasma 25-hydroxyvitamin D levels and fracture risk: the EPIC-Oxford study. Am J Epidemiol. 2007;166(11):1327–36. [PubMed]
5. Looker AC, Mussolino ME. Serum 25-hydroxyvitamin D and hip fracturerisk in older U.S. white adults. J Bone Miner Res. 2008;23(1):143–50. [PubMed]
6. Ross AC, Manson JE, Abrams SA, et al. The 2011 report on dietary reference intakes for calcium and vitamin D from the Institute of Medicine: what clinicians need to know. J Clin Endocrinol Metab. 96(1):53–8. [PMC free article] [PubMed]
7. IOM; Medicine Io. Dietary reference intakes for calcium and vitamin D. Washington, DC: National Academies Press; 2011.
8. Cranney A, Horsley T, O’Donnell S, et al. Effectiveness and safety of vitamin D in relation to bone health. Evid Rep Technol Assess (Full Rep) 2007;(158):1–235. [PubMed]
9. Bischoff-Ferrari HA, Willett WC, Wong JB, Giovannucci E, Dietrich T, Dawson-Hughes B. Fracture prevention with vitamin D supplementation: a meta-analysis of randomized controlled trials. JAMA. 2005;293(18):2257–64. [PubMed]
10. Avenell A, Gillespie WJ, Gillespie LD, O’Connell D. Vitamin D and vitamin D analogues for preventing fractures associated withinvolutional and post-menopausal osteoporosis. Cochrane Database Syst Rev. 2009;(2):CD000227. [PubMed]
11. Khosla S, Melton LJ, 3rd, Riggs BL. Clinical review 144: Estrogen and the male skeleton. J Clin Endocrinol Metab. 2002;87(4):1443–50. [PubMed]
12. Khosla S, Melton LJ, 3rd, Atkinson EJ, O’Fallon WM. Relationship of serum sex steroid levels to longitudinal changes in bone density in young versus elderly men. J Clin Endocrinol Metab. 2001;86(8):3555–61. [PubMed]
13. Gennari L, Merlotti D, Martini G, et al. Longitudinal association between sex hormone levels, bone loss, and bone turnover in elderly men. J Clin Endocrinol Metab. 2003;88(11):5327–33. [PubMed]
14. Van Pottelbergh I, Goemaere S, Kaufman JM. Bioavailable estradiol and an aromatase gene polymorphism are determinants of bone mineral density changes in men over 70 years of age. J Clin Endocrinol Metab. 2003;88(7):3075–81. [PubMed]
15. Cauley JA, Ewing SK, Taylor BC, et al. Sex steroid hormones in older men: longitudinal associations with 4. 5-year change in hip bone mineral density--the osteoporotic fractures in men study. J Clin Endocrinol Metab. 2010;95(9):4314–23. [PubMed]
16. Amin S, Zhang Y, Felson DT, et al. Estradiol, testosterone, and the risk for hip fractures in elderly men from the Framingham Study. Am J Med. 2006;119(5):426–33. [PubMed]
17. Barrett-Connor E, Mueller JE, von Muhlen DG, Laughlin GA, Schneider DL, Sartoris DJ. Low levels of estradiol are associated with vertebral fractures in older men, but not women: the Rancho Bernardo Study. J Clin Endocrinol Metab. 2000;85(1):219–23. [PubMed]
18. LeBlanc ES, Nielson CM, Marshall LM, et al. The effects of serum testosterone, estradiol, and sex hormone binding globulin levels on fracture risk in older men. J Clin Endocrinol Metab. 2009;94(9):3337–46. [PubMed]
19. Mellstrom D, Vandenput L, Mallmin H, et al. Older men with low serum estradiol and high serum SHBG have an increased risk of fractures. J Bone Miner Res. 2008;23(10):1552–60. [PubMed]
20. Goderie-Plomp HW, van der Klift M, de Ronde W, Hofman A, de Jong FH, Pols HA. Endogenous sex hormones, sex hormone-binding globulin, and the risk of incident vertebral fractures in elderly men and women: the Rotterdam Study. J Clin Endocrinol Metab. 2004;89(7):3261–9. [PubMed]
21. Meier C, Nguyen TV, Handelsman DJ, et al. Endogenous sex hormones and incident fracture risk in older men: the Dubbo Osteoporosis Epidemiology Study. Arch Intern Med. 2008;168(1):47–54. [PubMed]
22. Center JR, Nguyen TV, Sambrook PN, Eisman JA. Hormonal and biochemical parameters in the determination of osteoporosis in elderly men. J Clin Endocrinol Metab. 1999;84(10):3626–35. [PubMed]
23. Center JR, Nguyen TV, Sambrook PN, Eisman JA. Hormonal and biochemical parameters and osteoporotic fractures in elderly men. J Bone Miner Res. 2000;15(7):1405–11. [PubMed]
24. Blank JB, Cawthon PM, Carrion-Petersen ML, et al. Overview of recruitment for the osteoporotic fractures in men study (MrOS) Contemp Clin Trials. 2005;26(5):557–68. [PubMed]
25. Orwoll E, Blank JB, Barrett-Connor E, et al. Design and baseline characteristics of the osteoporotic fractures in men (MrOS) study--a large observational study of the determinants of fracture in older men. Contemp Clin Trials. 2005;26(5):569–85. [PubMed]
26. Lewis CE, Ewing SK, Taylor BC, et al. Predictors of non-spine fracture in elderly men: the MrOS study. J Bone Miner Res. 2007;22(2):211–9. [PubMed]
27. Wang XSA, Cawthon PM, Palermo L, Jekir M, Christensen J, Ensrud KE, Cummings SR, Orwoll E, Black DM, Keaveny TM. Osteoporotic Fractures in Men (MrOS) Research Group. Prediction of new clinical vertebral fractures in elderly men using finite element analysis of CT scans. J Bone Miner Res. 2012;27(4):808–816. [PMC free article] [PubMed]
28. Genant HK, Jergas M. Assessment of prevalent and incident vertebral fractures in osteoporosis research. Osteoporos Int. 2003;14 (Suppl 3):S43–55. [PubMed]
29. Sanders KM, Pasco JA, Ugoni AM, et al. The exclusion of high trauma fractures may underestimate the prevalence of bone fragility fractures in the community: the Geelong Osteoporosis Study. J Bone Miner Res. 1998;13(8):1337–42. [PubMed]
30. Mackey DC, Lui LY, Cawthon PM, et al. High-trauma fractures and low bone mineral density in older women and men. JAMA. 2007;298(20):2381–8. [PubMed]
31. Prentice RL. A case-cohort design for epidemiologic cohort studies and disease prevention trials. Biometrika. 1986;73:1–11.
32. Sodergard R, Backstrom T, Shanbhag V, Carstensen H. Calculation of free and bound fractions of testosterone and estradiol-17 beta to human plasma proteins at body temperature. J Steroid Biochem. 1982;16(6):801–10. [PubMed]
33. Singh RJ, Taylor RL, Reddy GS, Grebe SK. C-3 epimers can account for a significant proportion of total circulating 25-hydroxyvitamin D in infants, complicating accurate measurement and interpretation of vitamin D status. J Clin Endocrinol Metab. 2006;91(8):3055–61. [PubMed]
34. Holick MF. Vitamin D deficiency. N Engl J Med. 2007;357(3):266–81. [PubMed]
35. Washburn RA, Smith KW, Jette AM, Janney CA. The Physical Activity Scale for the Elderly (PASE): development and evaluation. J Clin Epidemiol. 1993;46(2):153–62. [PubMed]
36. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130(6):461–70. [PubMed]
37. Block G, Hartman AM, Naughton D. A reduced dietary questionnaire development and validation. Epidemiology. 1990;1(1):58–64. [PubMed]
38. Orwoll E, Nielson CM, Marshall LM, et al. Vitamin D deficiency in older men. J Clin Endocrinol Metab. 2009;94(4):1214–22. [PubMed]
39. Pahor M, Chrischilles EA, Guralnik JM, Brown SL, Wallace RB, Carbonin P. Drug data coding and analysis in epidemiologic studies. Eur J Epidemiol. 1994;10(4):405–11. [PubMed]
40. Barlow WE, Ichikawa L, Rosner D, Izumi S. Analysis of case-cohort designs. J Clin Epidemiol. 1999;52(12):1165–72. [PubMed]
41. Lin DY, Wei LJ. The Robust Inference for the Cox Proportional Hazards Model. Journal of the American Statistical Association. 1989;84(408):1074–1078.
42. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis. 2002;39(2 Suppl 1):S1–266. [PubMed]
43. Mithal A, Wahl DA, Bonjour JP, et al. Global vitamin D status and determinants of hypovitaminosis D. Osteoporos Int. 2009;20(11):1807–20. [PubMed]