SEARCH

SEARCH BY CITATION

Keywords:

  • SERUM 25-HYDROXYVITAMIN D;
  • PARATHYROID HORMONE;
  • HIP FRACTURES;
  • NONSPINE FRACTURES

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

The effects of vitamin D and parathyroid hormone (PTH) levels on incident fracture remain uncertain. To test the hypothesis that increasing serum 25-hydroxyvitamin D [25(OH)D] and decreasing PTH levels are associated with decreased risk of hip and any nonspine fracture, we conducted a prospective cohort study among 2614 community-dwelling white and black participants, aged ≥70 years, from the Health, Aging and Body Composition (Health ABC) Study. Serum and plasma samples were drawn at year 2, which formed the baseline for this analysis. Serum 25(OH)D and intact PTH (1-84) were measured using radioimmunoassay with DiaSorin reagents and EDTA plasma with a two-site immunoradiometric assay kit, respectively. Incident fractures (hip and any nonspine) were assessed after year 2, every 6 months, by self-report and validated by radiology reports. The median (interquartile range) follow-up times for hip and any nonspine fractures were 6.4 (6.1–6.5) and 6.4 (5.5–6.5) years, respectively. Cox proportional hazards regression was used to estimate the hazard ratios (HR) with 95% confidence intervals (CI) for fracture. There were 84 hip and 247 nonspine fractures that occurred over the follow-up period. The multivariable adjusted HRs (95% CIs) of hip fracture for participants in the lowest (≤17.78 ng/mL), second (17.79 to 24.36 ng/mL), and third quartiles (24.37 to 31.94 ng/mL) of 25(OH)D were 1.92 (0.97 to 3.83), 0.75 (0.32 to 1.72) and 1.86 (1.00 to 3.45), respectively, compared with participants in the highest 25(OH)D quartile (>31.94 ng/mL) (p trend = 0.217). Additional adjustment for IL-6 (p = 0.107), PTH (p = 0.124), and hip areal bone mineral density (p = 0.137) attenuated HRs of hip fracture in the lowest quartile by 16.3%, 17.4%, and 26.1%, respectively. There was no evidence of an association between 25(OH)D and any nonspine fractures, or between PTH and hip or any nonspine fractures. We found limited evidence to support an association between calciotropic hormones and hip and nonspine fractures in older men and women. © 2012 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Serum 25-hydroxyvitamin D [25(OH)D] is the widely accepted indicator of vitamin D nutritional status.1 Older adults are less likely to have sufficient serum 25(OH)D.2 Low 25(OH)D in older adults has consequences for increased rates of bone loss and bone turnover, which subsequently lead to low bone mineral density (BMD).3 The optimal level of vitamin D sufficiency remains controversial. Recently published Institute of Medicine (IOM) guidelines recommended serum 25(OH)D levels ≥20 ng/mL for skeletal health in most US adults;4 lower than the level of ≥30 ng/mL recommended by some experts.5 The latter definition of 25(OH)D sufficiency was based on the level in which parathyroid hormone (PTH) values reached their nadir in relation to 25(OH)D. Identifying 25(OH)D inflection points based on fractures as outcome might be a more clinically relevant way to define 25(OH)D sufficiency. Evidence linking low vitamin D status to higher fracture risk is inconsistent,6–15 although prior cohort studies suggest an optimal 25(OH)D range of ≥12 to 24 ng/mL for decreasing risk of fracture.6–8, 10, 13–15 There is little evidence to suggest that PTH levels influence fracture risk,16 despite a robust inverse correlation with 25(OH)D and the known catabolic effects of PTH on increased bone turnover and decreased cortical BMD.3

Calciotropic hormones may influence fracture rates through various mediators in the causal pathway. Low levels of 25(OH)D have been linked independently to potential mediators such as falls, neuromuscular function, disability, and hip BMD.17–19 Cauley and colleagues reported that the association between elevated 25(OH)D and hip fracture in the Women's Health Initiative study was attenuated by 18% after adjusting for C-terminal telopeptide, a marker of bone resorption.6 Another article on 25(OH)D identified hip BMD and neuromuscular function measures as mediators of fracture risk.7

We conducted a prospective population-based cohort study of older adults to determine: 1) if low serum 25(OH)D or high PTH levels are associated with an increased risk of incident hip and any nonspine fracture in older adults; 2) the optimal threshold concentration of 25(OH)D based on fracture risk; and 3) if these associations are mediated by physical function, inflammatory markers, falls, kidney function, BMD, and serum calcium.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Study population

The Health Aging and Body Composition (Health ABC) Study enrolled 3075 women and men, aged 70 to 79 years, from two field centers, Pittsburgh, PA, USA, and Memphis, TN, USA, in 1997 to 1998. To be eligible to participate in Health ABC, participants had to report no difficulty walking at least 1/4 mile and/or climbing a flight of stairs. Participants were identified from a random sample of white Medicare beneficiaries and all age-eligible black community residents in designated zip code areas surrounding Pittsburgh and Memphis. Exclusion criteria included reported difficulty performing basic activities of daily living, obvious cognitive impairment, inability to communicate with the interviewer, intention of moving within 3 years, or participation in a trial involving a lifestyle intervention. Follow-ups occurred on an annual basis, and serum and plasma concentrations were collected during the first follow-up or year 2. Of the 2998 participants at year 2, we excluded 205 (6.9%) for not having 25(OH)D or PTH measurements; 6 with 25(OH)D values >187.5 ng/mL, based on statistical criteria (75th + 3 interquartile range [IQR]); 7 with PTH >250 pg/mL; and 140 for using osteoporosis medications. An additional 26 participants were excluded for not completing a visit after year 2, resulting in an analytic sample of 2614 participants. Among the 2614 eligible participants, 21.8% died over the study period and 5.3% were lost to follow-up. The institutional review board (IRB) at each center approved the study protocol, and written informed consent was obtained from all the participants.

Serum 25(OH)D and PTH

Both specimens (serum and plasma) were collected at one time point (year 2). Samples were drawn in the morning, after an overnight fast. After processing, the specimens were frozen at −70°C and shipped to the Core Laboratory at the University of Vermont for long-term storage. Serum 25(OH)D and plasma PTH levels were subsequently assayed in 2008 to 2009. Serum 25(OH)D was measured in serum samples using a two-step radioimmunoassay [25(OH)D 125I RIA Kit, DiaSorin, Stillwater, MN, USA]. The interassay coefficient of variation for 25(OH)D was 6.78% for log-transformed values. Intact PTH (1-84) was measured in EDTA plasma with a two-site immunoradiometric assay kit (N-tact PTHSP, DiaSorin). The interassay coefficient of variation for PTH was 8.6%.

Hip and any nonspine fractures

Participants were contacted every 6 months, alternating between clinic visits and telephone interviews. Only fractures that occurred after year 2 were included in the time to event analysis. Information on self-report of incident hip or any nonspine fractures was obtained by phone interview, in person, or from proxy respondents (ie, family or friends). All fractures were validated by radiology reports (pathological fractures, fractures of unknown etiology, and traumatic fractures were excluded). Adjudication of 84 hip and 247 nonspine fractures was complete through June 30, 2007 for the Pittsburgh clinic and through December 31, 2005 for the Memphis clinic. The median (IQR) follow-up times for hip and any nonspine fractures were 6.4 (6.1 to 6.5) and 6.4 (5.5 to 6.5) years, respectively. The follow-up times in person-years for hip and any nonspine fractures were 15136.4 and 14734.5, respectively.

Other measurements

Demographic variables included self-report of age, sex, race (whites and blacks), and education (<high school [HS] or ≥HS). Season of blood draw was coded as winter, December to February; spring, March to May; summer, June to August; fall, September to November. Weight was measured on a standard balance beam scale to the nearest 0.1 kg, and height was measured by a stadiometer to the nearest 0.1 cm with body mass index (BMI; kg/m2) calculated using the formula weight (kg)/height (m)2. Hip areal BMD (aBMD) was measured using dual-energy X-ray absorptiometry (DXA; QDR 4500A; software version 9.03; Hologic, Bedford, MA, USA). DXA quality-assurance procedures were conducted at both study sites and monitored by the study coordinating center, ensuring scanner reliability and identical scan protocols. An anthropometric spine phantom was scanned daily and a hip phantom once per week to assess longitudinal performance of the scanners.

Lifestyle factors included self-report of smoking in pack-years and alcohol consumption (none, <1 drink/week, 1 to 3 drinks/week, or ≥4 drinks/week). To assess supplementary intake for vitamin D and calcium, participants were asked to bring all prescription and over-the-counter medications, which were coded based on the Iowa Drug Information System.20 To estimate dietary intake of calcium, participants completed a 108-item interviewer-administered food frequency questionnaire (FFQ; Block Dietary Data Systems, Berkeley, CA, USA). Time spent walking (min/week) was determined by self-report. History of fracture after age 45 years was determined by self-report of doctor-diagnosed fracture. To better understand the effect of comorbidities, a composite clinical comorbidity index21 was created centering on self-report of seven chronic health conditions (cardiovascular disease, stroke, pulmonary disease, diabetes, kidney disease, arthritis, and depression). Respondents were asked to report the number of falls in the past 12 months. Physical function was determined by the Health ABC physical performance battery (HPPB) as described previously.22 Briefly, the performance scale (0 to 4) includes five repeated chair stands, 6-meter walk time, 6-meter narrow walk, and a balance test.

Serum calcium was measured with direct quantitative colorimetric determination using Stanbio Total Calcium LiquiColor Procedure No. 0500 (Stanbio Laboratory, Boerne, TX, USA). The interassay coefficient of variation was 2.2%. Interleukin (IL)-6 was measured in duplicate by enzyme-linked immunosorbent assay. The detectable limit for IL-6 was 0.10 pg/mL. Blind duplicate analyses (n = 50) for IL-6 showed an interassay coefficient of variation of 10.3%. Cystatin-C was measured using a BNII nephelometer (Dade Behring, Deerfield, IL, USA) that used a particle-enhanced immunonepholometric assay (N Latex Cystatin C). The assay range was 0.195 to 7.330 mg/L with an intra-assay coefficient of variation of 7.7%. The estimated glomerular filtration rate (eGRF) was then calculated using a validated cystatin C-based equation.23

Statistical analysis

Tests of trend were used to compare participant characteristics across quartiles of 25(OH)D and PTH. To further describe the relationship between PTH and 25(OH)D, a LOESS scatterplot was created with PTH plotted across values of 25(OH)D. Tests of threshold were conducted to find the 25(OH)D points where PTH values begin to reach their nadir. To evaluate participant characteristics by fracture status, two-sample t tests, ANOVA, chi-square tests, Wilcoxon rank-sum, and Kruskal-Wallis tests were used.

We assessed the association between 25(OH)D, PTH, and incident hip and nonspine fracture using semiparametric Cox proportional hazards models. Hazard ratios (HR) and 95% confidence intervals (95% CI) comparing quartile 4 (top quartile) with quartiles 1, 2, and 3 were calculated for 25(OH)D and PTH with tests for trend across the quartiles. There was a nonlinear association between 25(OH)D and incident hip fractures; consequently, we were unable to model 25(OH)D as a continuous variable. The proportional hazards assumption was confirmed graphically and formally using Schoenfeld residuals. The model fit was verified by plotting the Nelson-Aalen cumulative hazard of the Cox-Snell residuals.

To further understand the relationship between 25(OH)D and fracture, we performed spline analysis. Restricted cubic spline linear regression was used with knots for 25(OH)D at the 5th, 25th, and 75th percentile, and a reference group at the 95th percentile was set to create the spline plot. Threshold effects were evaluated by identifying potential inflection points on the spline and performing a test of equality to determine if the slopes above and below the cut point were equal.

For the base multivariable models, a variable would be considered for inclusion if it was associated with the exposure or outcome of interest. Backward elimination procedure (p = 0.15 entry and removal) was then used with age, sex, race, season of blood draw, and the exposure of interest forced into all base multivariable models. As a result, the final base multivariable model for hip fracture included the forced variables, BMI, alcohol use, and fracture after age 45 years. The final base multivariable model for nonspine fracture included the same variables. To investigate mechanisms by which 25(OH)D and PTH might be associated with hip and nonspine fractures, we added the following variables one at a time to the base model to determine if they mediated this association: number of falls, time to complete five chair stands, IL-6, serum calcium, hip aBMD, and estimated glomerular filtration rate (eGFR). We then adjusted for all variables simultaneously in the full multivariable model.

Sex, race, age (<75 versus ≥75 years), and season did not modify the associations between the calciotropic hormones and incident hip or nonspine fracture for all models (p interaction > 0.05), and thus we refrained from performing stratified analyses.

We conducted secondary analyses to determine if 25(OH)D is associated longitudinally with potential mediators (incident falls and BMD loss) in the causal pathway leading to fracture. Hip aBMD measurements were taken at baseline, years 3, 5 or 6, 8, and 10. Linear mixed effects models accounted for repeated measures of hip aBMD. We excluded participants with ≤1 measurement of hip BMD (analytic sample n = 2439). We included a random intercept for each subject and a random slope for time to account for within-person correlation. Change in aBMD was estimated through an interaction of time with covariates of interest. We estimated the average annual change in hip aBMD by 25(OH)D quartiles. Incident number of falls for each participant was aggregated using data from years 3 to 10. The association between 25(OH)D quartiles and incident falls was then evaluated using Poisson regression with a robust variance estimator to account for overdispersion.

All statistical analyses were performed using the Statistical Analysis System (SAS, version 9.2; SAS Institute, Cary, NC, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

Table 1 shows the descriptive baseline characteristics by 25(OH)D and PTH quartiles. Lower 25(OH)D levels were associated with women, blacks, <high school education, greater BMI, winter, lower alcohol consumption, lower dietary and supplemental calcium intake, lower supplemental vitamin D intake, lower time spent walking, higher clinical comorbidity index, lower physical function, greater IL-6, lower hip aBMD, and greater eGFR. For PTH, the direction of these associations was reversed, with the exception of eGFR and hip aBMD, which were not significantly associated with PTH. Serum 25(OH)D was inversely associated with PTH. PTH concentrations began to reach their nadir at approximately 15 ng/mL of 25(OH)D (Fig. 1).

Table 1. Descriptive Baseline Characteristics by 25(OH)D and PTH Quartiles
 25(OH)DPTH
Q1Q2Q3Q4Q1Q2Q3Q4
(n = 660)(n = 660)(n = 660)(n = 660)(n = 659)(n = 659)(n = 661)(n = 660)
  1. 25(OH)D quartile cut points are 17.777 ng/mL, 24.364 ng/mL, and 31.9375 ng/mL.

  2. PTH quartile cut points are 25.17 pg/mL, 33.81 pg/mL, and 45.92 pg/mL.

  3. Bold indicates p for trend < 0.05.

Age, mean ± SD (years)74.7 ± 2.974.6 ± 2.974.6 ± 2.974.7 ± 2.874.5 ± 2.874.6 ± 2.974.6 ± 2.874.9 ± 2.9
Female gender (%)61.245.944.643.945.147.250.253.0
Black race (%)69.147.429.218.331.437.841.053.8
Education <high school (%)34.526.521.416.822.623.425.228.1
BMI, mean ± SD (kg/m2)28.6 ± 5.627.7 ± 5.127.2 ± 4.226.0 ± 3.826.2 ± 4.027.0 ± 4.627.7 ± 5.128.4 ± 5.2
Season of blood draw (%)
 Winter31.028.224.219.122.524.928.127.1
 Spring33.233.530.630.129.731.732.433.5
 Summer11.417.121.119.919.018.216.315.9
 Fall24.421.224.130.928.825.223.223.5
Pack-years exposure to cigarettes, median (IQR)3.0 (0.0–27.0)6.0 (0.0–33.0)3.0 (0.0–32.0)5.5 (0.0–32.5)4.0 (0.0–32.0)8.0 (0.0–33.0)3.0 (0.0–27.0)3.0 (0.0–31.0)
Current alcohol consumption (%)
 None70.468.463.258.762.163.464.071.1
 <1 drink/week4.75.66.57.17.56.14.55.8
 1–3 drinks/week15.917.522.724.021.119.522.317.3
 4+ drinks/week9.08.57.610.29.311.09.25.8
Dietary calcium intake, median (IQR) (mg/d)650 (433–897)731 (517–971)725 (533–978)766 (556–1036)769 (547–1034)718 (527–955)719 (511–965)679 (466–950)
Supplemental calcium intake (% yes)6.213.623.830.427.220.415.011.4
Supplemental vitamin D intake (% yes)2.05.611.215.012.38.87.45.3
Time spent walking, median (IQR) (min/week)0 (0–90)40 (0.0–175)60 (0–180)70 (0–210)60 (0–210)45 (0–180)38 (0–180)20 (0–120)
Any fracture after age 45 years (% yes)21.923.020.319.421.121.720.221.5
Clinical Comorbidity Index (0–7), mean ± SD0.91 ± 0.980.78 ± 0.880.76 ± 0.850.73 ± 0.810.70 ± 0.800.77 ± 0.880.78 ± 0.870.93 ± 0.96
Any falls in past 12 months (% yes)26.121.423.822.122.222.425.023.7
Health ABC Performance Score (0–4), mean ± SD2.0 ± 0.62.2 ± 0.52.3 ± 0.52.3 ± 0.52.3 ± 0.52.3 ± 0.52.2 ± 0.62.1 ± 0.6
IL-6, median (IQR) (pg/mL)2.8 (1.7–4.7)2.4 (1.5–4.2)2.3 (1.5–3.7)2.2 (1.4–3.7)2.1 (1.4–3.6)2.3 (1.5–4.0)2.6 (1.6–4.3)2.7 (1.8–4.5)
Hip aBMD, mean ± SD (g/cm2)0.90 ± 0.170.91 ± 0.170.89 ± 0.160.88 ± 0.160.89 ± 0.160.90 ± 0.170.90 ± 0.170.90 ± 0.17
Glomerular filtration rate, mean ± SD74.6 ± 17.173.8 ± 16.672.3 ± 15.869.7 ± 14.674.6 ± 15.874.0 ± 14.873.5 ± 14.668.2 ± 18.3
Serum calcium, mean ± SD (mg/dL)8.9 ± 0.48.9 ± 0.48.9 ± 0.48.9 ± 0.48.9 ± 0.48.9 ± 0.48.8 ± 0.48.9 ± 0.5
25(OH)D, median (IQR) (ng/mL)    29.1 (22.3–36.1)25.7 (19.5–32.3)22.9 (17.2–30.2)19.3 (13.8–26.3)
PTH, median (IQR) (pg/mL)41.9 (31.7–58.3)35.2 (26.7–46.2)31.1 (24.4–41.0)28.6 (21.4–38.2)    
thumbnail image

Figure 1. LOESS scatterplot showing PTH and 95% confidence limits (shaded area) across values of 25(OH)D.

Download figure to PowerPoint

Table 2 shows the descriptive characteristics by hip and any nonspine fracture status. Participants with hip and any nonspine fracture were significantly more likely to be older, women, white, have lower BMI, have a history of fracture after age 45 years, have a higher clinical comorbidity index, and have lower aBMD. Additional factors associated with any nonspine fracture were: more likely to have a high school education, have higher supplemental calcium and vitamin D intake, more likely to have fallen in the past 12 months, and have lower physical function and eGFR.

Table 2. Descriptive Baseline Characteristics by Hip and Nonspine Fracture Status
 Hip fractureAny nonspine fracture
No (n = 2556)Yes (n = 84)No (n = 2393)Yes (n = 247)
  1. Bold indicates p < 0.05.

Age, mean ± SD (years)74.6 ± 2.876.4 ± 2.874.6 ± 2.875.3 ± 3.0
Female gender (%)48.560.746.770.5
Black race (%)41.428.642.328.3
Education <high school (%)24.825.025.617.0
BMI, mean ± SD (kg/m2)27.4 ± 4.825.8 ± 4.327.4 ± 4.826.7 ± 4.8
Season of blood draw (%)
 Winter25.529.825.428.3
 Spring32.123.832.129.2
 Summer17.319.017.317.4
 Fall25.127.425.225.1
Pack-years exposure to cigarettes, median (IQR)4.0 (0.0–30.0)4.0 (0.0–36.5)4.5 (0.0–30.0)1.5 (0.0–30.0)
Current alcohol consumption (%)
None65.071.165.165.8
 <1 drink/week6.06.06.14.9
 1–3 drinks/week20.118.119.623.9
 4+ drinks/week8.94.89.25.4
Dietary calcium intake, median (IQR) (mg/d)717 (515–973)736 (532–995)719 (517–978)716 (501–940)
Supplemental calcium intake (% yes)18.325.017.428.7
Supplemental vitamin D intake (% yes)8.313.18.112.2
Time spent walking, median (IQR) (min/week)40 (0–180)60 (0–195)40 (0–180)36 (0–150)
Any fracture after age 45 years (% yes)20.733.320.032.4
Clinical Comorbidity Index (0–7), mean ± SD0.79 ± 0.881.06 ± 0.970.78 ± 0.870.95 ± 0.99
Any falls in past 12 months (% yes)23.128.922.728.9
Health ABC Performance Score (0–4), mean ± SD2.2 ± 0.52.1 ± 0.62.2 ± 0.52.1 ± 0.6
IL-6, median (IQR) (pg/mL)2.4 (1.5–4.1)2.7 (1.7–4.4)2.4 (1.5–4.1)2.4 (1.6–4.0)
Hip aBMD, mean ± SD (g/cm2)0.90 ± 0.160.74 ± 0.140.91 ± 0.160.78 ± 0.14
Glomerular filtration fate, mean ± SD72.7 ± 16.170.2 ± 16.672.9 ± 16.269.8 ± 15.5
Serum calcium, mean ± SD (mg/dL)8.9 ± 0.438.9 ± 0.428.9 ± 0.438.9 ± 0.41
25(OH)D, median (IQR) (ng/mL)24.3 (17.8–31.9)26.3 (15.9–31.9)24.3 (17.9–31.8)24.7 (16.5–32.2)
PTH, median (IQR) (pg/mL)33.8 (25.2–45.9)34.3 (24.8–46.9)33.8 (25.2–45.7)34.3 (24.9–47.0)

25(OH)D and fracture

Table 3 shows the associations between 25(OH)D and hip and any nonspine fracture. The HR (95% CIs) of hip fracture for participants in the lowest quartile (≤17.78 ng/mL) was 1.92 (0.97 to 3.83) after adjusting for age, sex, race, season of blood draw, BMI, alcohol use, fracture after age 45 years, and the clinical comorbidity index compared with participants in the highest 25(OH)D quartile (≥31.93 ng/mL). There was no association (0.75 [0.32 to 1.72]) between 25(OH)D and hip fracture among participants in the second quartile (>17.78 to 24.36 ng/mL); however, participants in the third quartile (>24.36 to <31.93 ng/mL) had an elevated risk (1.86 [1.00 to 3.45]) of fracture. The overall trend was not statistically significant (p trend = 0.217). Adjusting individually for IL-6 (p = 0.107), PTH (p = 0.124), and hip aBMD (p = 0.137) in addition to the aforementioned covariates attenuated these associations further by 16.3%, 17.4%, and 26.1%, respectively.

Table 3. Hazard Ratios (95% Risk Limits) of Hip and Nonspine Fracture Across 25(OH)D Quartiles
ModelNo.Q1Q2Q3Q4p Trend
  • Quartile cut points are 17.777 ng/mL, 24.364 ng/mL, and 31.9375 ng/mL.

  • a

    Base MV (multivariable) model adjusted for age, sex, race, season of blood draw, BMI, current drinking, fracture after age 45 years, and Clinical Comorbidity Index.

Hip fractures
 Base MV modela25011.92 (0.97, 3.83)0.75 (0.32, 1.72)1.86 (1.00, 3.45)Ref0.217
 Base MV modela + falls24261.99 (0.99, 3.96)0.68 (0.29, 1.61)1.86 (1.01, 3.46)Ref0.230
 Base MV modela + Health ABC Performance Score (0–4)24681.89 (0.93, 3.81)0.79 (0.34, 1.84)2.04 (1.09, 3.83)Ref0.112
 Base MV modela + IL-624771.77 (0.88, 3.58)0.78 (0.34, 1.79)1.74 (0.93, 3.26)Ref0.269
 Base MV modela + serum calcium24921.89 (0.94, 3.78)0.74 (0.32, 1.70)1.83 (0.98, 3.40)Ref0.245
 Base MV modela + hip aBMD24801.68 (0.85, 3.34)0.66 (0.29, 1.51)1.49 (0.80, 2.79)Ref0.632
 Base MV modela + eGFR25011.95 (0.97, 3.90)0.76 (0.33, 1.77)1.88 (1.01, 3.50)Ref0.198
 Base MV modela + PTH25001.76 (0.86, 3.60)0.74 (0.32, 1.70)1.82 (0.98, 3.39)Ref0.210
 Full MV model (base MV modela + all mediators)23471.73 (0.80, 3.75)0.74 (0.31, 1.79)1.60 (0.83, 3.10)Ref0.430
Nonspine fractures
 Base MV modela24941.21 (0.83, 1.75)1.01 (0.68, 1.49)1.12 (0.78, 1.60)Ref0.752
 Base MV modela + falls24191.17 (0.80, 1.71)0.98 (0.66, 1.45)1.12 (0.78, 1.61)Ref0.739
 Base MV modela + Health ABC Performance Score (0–4)24611.31 (0.89, 1.93)1.09 (0.73, 1.63)1.20 (0.83, 1.74)Ref0.522
 Base MV modela + IL-624701.21 (0.83, 1.75)1.02 (0.69, 1.52)1.11 (0.77, 1.59)Ref0.777
 Base MV modela + serum calcium24851.21 (0.83, 1.75)0.99 (0.66, 1.46)1.12 (0.78, 1.60)Ref0.774
 Base MV modela + hip aBMD24731.16 (0.79, 1.68)1.02 (0.69, 1.52)1.11 (0.77, 1.59)Ref0.743
 Base MV modela + eGFR24941.26 (0.87, 1.83)1.06 (0.71, 1.57)1.16 (0.81, 1.67)Ref0.606
 Base MV modela + PTH24931.16 (0.79, 1.70)1.00 (0.68, 1.49)1.11 (0.78, 1.60)Ref0.727
 Full MV model (base MV modela + all mediators)23401.36 (0.90, 2.07)1.16 (0.76, 1.77)1.23 (0.85, 1.80)Ref0.430

The risk of any nonspine fracture was highest among those in the lowest 25(OH)D quartile; however, there was no significant association in any model. There was no evidence of a threshold (p for test of threshold > 0.05; Figs. 2 and 3) or trend for serum 25(OH)D concentrations and incident hip or nonspine fracture.

thumbnail image

Figure 2. Restricted cubic-spline Cox regression plot showing the hazard ratios and 95% confidence limits (dotted lines) of incident hip fractures by serum 25(OH)D. Regression was adjusted for age, sex, race, season of blood draw, BMI, alcohol use, fracture after age 45 years, Clinical Comorbidity Index, falls, Health ABC performance score, IL-6, serum calcium, hip aBMD, eGFR, and PTH.

Download figure to PowerPoint

thumbnail image

Figure 3. Restricted cubic-spline Cox regression plot showing the hazard ratios and 95% confidence limits (dotted lines) of incident nonspine fractures by serum 25(OH)D. Regression was adjusted for age, sex, race, season of blood draw, BMI, alcohol use, fracture after age 45 years, Clinical Comorbidity Index, falls, Health ABC performance score, IL-6, serum calcium, hip aBMD, eGFR, and PTH.

Download figure to PowerPoint

PTH and fracture

Table 4 shows the associations between PTH and hip and any nonspine fracture. In the base multivariable model, the adjusted HR (95% CI) of hip fracture was 0.68 (0.35 to 1.31) among participants in the lowest quartile of PTH compared with those in the highest quartile of PTH (p trend = 0.942). Similarly, there were also no significant trends or differences across quartiles for the association of PTH with nonspine fracture.

Table 4. Hazard Ratios (95% Risk Limits) of Hip and Nonspine Fractures Across PTH Quartiles
ModelNo.Q1Q2Q3Q4p Trend
  • Quartile cut points are 25.17 pg/mL, 33.81 pg/mL, and 45.92 pg/mL.

  • a

    Base MV (multivariate) model adjusted for age, sex, race, season of blood draw, BMI, current drinking, fracture after age 45 years, and Clinical Comorbidity Index.

Hip fractures
 Base MV modela25000.68 (0.35, 1.31)0.73 (0.38, 1.37)0.96 (0.52, 1.79)Ref0.942
 Base MV modela + falls24250.65 (0.34, 1.27)0.72 (0.38, 1.35)0.90 (0.48, 1.68)Ref0.798
 Base MV modela + Health ABC Performance Score (0–4)24670.71 (0.36, 1.40)0.78 (0.41, 1.49)1.05 (0.56, 1.98)Ref0.814
 Base MV modela + IL-624760.69 (0.35, 1.36)0.73 (0.38, 1.39)0.91 (0.48, 1.73)Ref0.810
 Base MV modela + serum calcium24910.68 (0.35, 1.33)0.73 (0.39, 1.38)0.95 (0.51, 1.78)Ref0.927
 Base MV modela + hip aBMD24790.84 (0.43, 1.64)0.86 (0.45, 1.64)1.03 (0.54, 1.95)Ref0.930
 Base MV modela + eGFR25000.68 (0.35, 1.32)0.73 (0.38, 1.38)0.96 (0.52, 1.80)Ref0.956
 Base MV modela + 25(OH)D25000.76 (0.38, 1.51)0.79 (0.41, 1.51)1.00 (0.54, 1.88)Ref0.956
 Full MV model (base MV modela + all mediators)23470.94 (0.44, 2.02)0.94 (0.47, 1.90)1.02 (0.51, 2.06)Ref0.951
Nonspine fractures
 Base MV modela25000.99 (0.68, 1.42)0.85 (0.58, 1.23)0.95 (0.66, 1.36)Ref0.588
 Base MV modela + falls24251.00 (0.69, 1.44)0.85 (0.58, 1.23)0.94 (0.65, 1.35)Ref0.534
 Base MV modela + Health ABC Performance Score (0–4)24670.95 (0.66, 1.38)0.83 (0.57, 1.20)0.90 (0.63, 1.31)Ref0.444
 Base MV modela + IL-624761.01 (0.70, 1.47)0.87 (0.59, 1.26)0.96 (0.67, 1.39)Ref0.644
 Base MV modela + serum calcium24911.00 (0.69, 1.45)0.86 (0.59, 1.24)0.96 (0.67, 1.39)Ref0.641
 Base MV modela + hip aBMD24791.17 (0.81, 1.71)1.00 (0.68, 1.46)1.05 (0.72, 1.52)Ref0.980
 Base MV modela + eGFR25001.01 (0.70, 1.46)0.90 (0.62, 1.31)0.98 (0.68, 1.41)Ref0.782
 Base MV modela + 25(OH)D25001.01 (0.69, 1.49)0.86 (0.59, 1.25)0.96 (0.67, 1.38)Ref0.614
 Full MV model (base MV modela + all mediators)23471.29 (0.85, 1.94)1.11 (0.74, 1.66)1.07 (0.72, 1.58)Ref0.944

Secondary analyses

Lower 25(OH)D was associated with greater aBMD loss (p trend = 0.024). Participants in the top quartile (−0.55%; 95 CI −0.48 to −0.62) of 25(OH)D had significantly lower annualized hip aBMD loss compared with those in the lowest quartile (−0.65%; 95 CI −0.58 to −0.72). Serum 25(OH)D levels were not associated with incident falls (p trend = 0.284). The multivariable adjusted HRs (95% CIs) of falls among participants in the lowest, second, and third quartiles of 25(OH)D were 0.91 (0.77 to 1.08), 0.96 (0.81 to 1.13), and 0.98 (0.83 to 1.15), respectively, compared with participants in the top 25(OH)D quartile. Both secondary analyses adjusted for age, sex, race, BMI, and season.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

In our prospective cohort study, we found limited evidence to support an association between 25(OH)D and hip fracture and no evidence of an association with nonspine fracture. The lack of a clear trend complicates the interpretation of these data. We were also unable to identify an optimal threshold concentration of 25(OH)D based on fracture risk. In addition, PTH concentrations were not associated with hip or nonspine fractures.

The lack of a clear trend between 25(OH)D and hip fracture has implications for the cause and effect relationship, as well as potential biological plausibility. Although we observed that subjects with the lowest 25(OH)D and those in the third quartile had a borderline increased risk of fracture, there was no evidence of a clear trend. Thus, it is possible that some of the hazard rate variations of hip fracture across 25(OH)D quartiles are a result of random error or chance.

Our mostly null finding for 25(OH)D and hip fracture supports the results from a prior prospective study9 among participants from the Study of Osteoporotic Fractures (SOF). However, comparing these two studies is difficult, mainly as a result of the SOF cohort's use of a much older assay that has been shown to misclassify vitamin D status.24

Previous studies showed an increased risk of hip fracture among older adults with low serum 25(OH)D levels.6–8, 14 The National Health and Nutrition Examination Survey (NHANES III) reported the highest risk of hip fracture was at levels ≤17.2 ng/mL of 25(OH)D.8 Similarly, Cauley and colleagues reported that men7 and women6 with estimated 25(OH)D values of <19 ng/mL were at the greatest risk of hip fracture. The Cardiovascular Health Study showed that serum concentrations of <15 ng/mL were associated with a 61% greater risk of hip fracture.14 The number of hip fractures (n = 84) in our study has implications for power (estimated post hoc power when comparing Q4 and Q1 was 54%) and an increased chance of any type II errors. However, the MrOS study7 found a dose-response relationship with an even lower number of hip fractures (n = 81).

We found no association between 25(OH)D and any incident nonspine fracture consistent with two prospective studies in older men7 and postmenopausal women.11 Fractures in bones with a larger proportion of trabecular bone (ie, wrist and ankle fractures) may explain this consistent null association for nonspine fractures. The association between 25(OH)D and trabecular BMD has been shown to be largely null,25, 26 and excess PTH may actually help maintain or even have anabolic effects on trabecular bone.27 In addition, the association between hip aBMD and nonspine fracture was slightly weaker than for hip fractures in our population.

Serum 25(OH)D concentrations may affect fracture risk through various mediators in the causal pathway. Individual adjustment for IL-6, PTH, and hip aBMD attenuated HRs of hip fracture in the lowest quartile by 16.3%, 17.4%, and 26.1%, respectively. Low 25(OH)D has been linked to increased inflammatory activity in vivo28 and in a population-based cohort,29 and high levels of inflammatory markers (ie, IL-6) have been shown to increase fracture risk.30, 31 In our study, there was a robust inverse association between 25(OH)D and IL-6. PTH was also inversely correlated with 25(OH)D in our population. This is expected because low serum 25(OH)D causes increased stimulation of the parathyroid gland and subsequent release of excess PTH.5 The effects of 25(OH)D on incident hip fracture were mainly explained by hip aBMD. This is not surprising given the robust positive trend we observed between 25(OH)D and hip aBMD, and the large difference in hip aBMD (0.16 g/cm2) between hip fracture cases and noncases. Our cross-sectional findings were also supported by our longitudinal analysis, which showed that lower 25(OH)D was associated with higher hip aBMD loss.

Plasma PTH levels were not associated with hip or nonspine fracture, which is consistent with a recent prospective cohort study.14 There was a robust inverse correlation between PTH levels and 25(OH)D; however, PTH was not associated with hip aBMD, which may explain our null findings. Our findings are also consistent with a meta-analysis of 905 hip fracture cases and 924 controls from population-based case-control studies.16 Of the 10 case-control studies identified, only two showed significantly higher PTH levels in hip fracture patients compared with controls.32, 33

Although we were unable to identify a statistically significant optimal 25(OH)D threshold for fracture risk, there was a marginal elevated risk of hip fracture among participants with a 25(OH)D level of ≤17.8 ng/mL. This is consistent with the current IOM recommendations for 25(OH)D sufficiency4 and results from other longitudinal studies on 25(OH)D and hip fracture.6–8 Optimal 25(OH)D concentrations were previously defined based on the nadir of PTH levels,34 leading to a wide range of optimal 25(OH)D thresholds and thus resulting in a lack of a consensus regarding a defined level of sufficiency. In our study, we found that PTH values reached their lowest point of 25(OH)D at 15 ng/mL, consistent with our findings on 25(OH)D and incident hip fracture.

Strengths of our study include the large sample size, long follow-up, and thus the ability to reliably assess temporality, control for many potential confounders, exploration of several mechanisms potentially underlying these associations, and the use of a valid and reliable 25(OH)D assay. Our study also has several limitations. Blacks are more likely to be in the lowest quartiles of 25(OH)D; however, most of the fractures occurred among whites. We previously showed that the relationship between 25(OH)D and fracture differed in black women,35 but we had too few hip fractures in blacks to carry out race-specific analyses. Second, serum 25(OH)D was measured at one point and may not reflect vitamin D status over the follow-up period. Nonetheless, a strong correlation (r = 0.7) was reported for 25(OH)D values measured after 3 years.36 Third, even though the blood test was taken during visit 2, the actual assay of 25(OH)D and PTH occurred around 10 years later. Hence, participants wouldn't have known their vitamin D status, and thus, their behavior regarding vitamin D and calcium supplementary intake would be unaffected by the serum samples taken at year 2. Nevertheless, they still could have been tested by their own doctors and initiated vitamin and calcium supplementation. Fourth, we considered many potential confounders and mediators in our analyses; however, there may be residual confounding by unmeasured factors. Finally, we did not have measures of bone turnover that have been shown to mediate the relationship between 25(OH)D and fracture.6

In summary, we found limited evidence to support an association between calciotropic hormones and hip and any nonspine fractures in a community-based sample of older men and women.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References

The Health Aging and Body Composition Study (Health ABC) includes the contract numbers N01-AG-6-2101, N01-AG-6-2103, N01-AG-6-2106, RO1-AG028050, RO1-NR012459, and WFUHS11200. This research was supported in part by the Intramural Research Program of the NIH, National Institute on Aging.

Authors' roles: Study design: JAC, KEB, SBK, DKH. Study conduct: SBK, DKH, JAC, FAT, TBH, SRC. Data collection: JAT, SBK, DKH, JAC, FAT, TBH, SRC. Data analysis: KEB, YS, RB, TP. Data interpretation: JAT, SBK, DKH, JAC, FAT, TBH, SRC, KEB, RB, YS, TP, DCB. Drafting manuscript: KEB, JAC. Revising manuscript content: JAT, SBK, DKH, JAC, FAT, TBH, SRC, KEB, RB, YS, TP, DCB. Approving final version of manuscript: JAT, SBK, DKH, JAC, FAT, TBH, SRC, KEB, RB, YS, TP, DCB. KEB takes responsibility for the integrity of the data analysis.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References