Abstract
Few studies have prospectively examined predictors of change in plasma concentrations of 25-hydroxyvitamin D [25(OH)D]. We sought to determine the predictors of 5-y change in 25(OH)D. Plasma 25(OH)D concentrations were assessed at baseline (1997–2000) and 5 y later (2002–2005) in 668 postmenopausal women enrolled in the Osteoporosis and Periodontal Disease Study. Baseline and changes in demographic, dietary, lifestyle, and health-related factors were tested as predictors of change in 25(OH)D concentrations by using multivariable linear regression. The mean 5-y change in 25(OH)D (mean ± SD) was 7.7 ± 0.7 nmol/L (P < 0.001). In our predictive model (n = 643), predictors explained 31% of the variance in change in 25(OH)D concentrations and included baseline 25(OH)D, baseline and change in vitamin D supplementation and physical activity, change in season of blood draw, BMI, whole-body T score, and baseline hormone therapy use. Baseline 25(OH)D and change in vitamin D supplementation explained the most variation (25%) in 25(OH)D. Exploratory analyses showed a borderline significant interaction between tertiles of baseline 25(OH)D and change in vitamin D supplementation over time (P = 0.06). The greatest mean increase in 25(OH)D (22.9 ± 16.8 nmol/L), with adjustment for other statistically significant predictors, occurred in women whose baseline 25(OH)D concentration was ≤51.0 nmol/L (tertile 1) and who increased supplementation use over time. These results confirm the importance of supplementation in increasing 25(OH)D concentrations in aging women, even after other statistically significant predictors are controlled for. These data also suggest that this is especially true among aging women with inadequate 25(OH)D (e.g., <50 nmol/L).
Introduction
The significance of vitamin D in bone metabolism is well established, and vitamin D is hypothesized by some to influence other chronic diseases as well (1). Blood concentrations of 25-hydroxyvitamin D [25(OH)D]7 above the range of 30 to 50 nmol/L are reported to be associated with optimal bone health (1). Therefore, a better understanding of predictors of change in 25(OH)D is needed to determine which modifiable and nonmodifiable factors may influence one’s status over time (i.e., years).
A number of cross-sectional studies have shown that factors such as vitamin D intake (both from diet and supplements), season of blood draw, sunlight exposure, age, race, adiposity, and physical activity predict 25(OH)D measured at one time point (2–9). However, few studies have prospectively examined predictors of change in 25(OH)D concentrations. Intraindividual variation in 25(OH)D over time has been shown by some to be moderate (10–14), and data suggest that 25(OH)D concentrations do in fact change over time (13–16). Obtaining a better understanding of the predictors of change in 25(OH)D might clarify how these and other factors contribute to the maintenance or achievement of vitamin D adequacy (≥50 nmol/L), as defined by the Institute of Medicine (17). In the few studies that prospectively examined predictors of change in 25(OH)D, predictors included measures of body size (13, 15), physical activity (13), vitamin D intake (13), and exogenous estrogen dose (18). Only 2 of these studies examined change over a period of years (13, 15).
Using data from the Osteoporosis and Periodontal Disease (OsteoPerio) Study, an ancillary study of the Women’s Health Initiative Observational Cohort Study (WHIOS) conducted in Buffalo, NY, we prospectively examined the predictors of 5-y change in plasma 25(OH)D in a sample of 668 postmenopausal women. This study contains extensive information on key variables for studying predictors of change in 25(OH)D and includes data on demographic characteristics, dietary and supplement intake, body composition, other lifestyle factors, and medication use. We hypothesized that change in 25(OH)D would be predicted by baseline measures of 25(OH)D and baseline and change measures of vitamin D intake from food and supplements; measures of sunlight exposure, body size, and bone mineral density; and other health and lifestyle factors that could affect vitamin D status.
Participants and Methods
Study sample
The WHIOS is a prospective cohort study that investigates the major causes of morbidity and mortality in 93,676 postmenopausal women (ages 50–79 y at enrollment) recruited from 40 clinical sites across the United States (1993–1998) (19–21). The OsteoPerio Study is an ancillary study of the WHIOS that investigated the association between osteoporosis and periodontal disease in postmenopausal women. Women enrolled in WHIOS at the Buffalo, NY, clinic center (n = 2249) were invited to participate in the OsteoPerio Study (1997–2000) (22). Of these, 338 were deemed ineligible [had <6 teeth (n = 162), history of bone disease (n = 2), both hips replaced (n = 16), cancer diagnosed in the past 10 y (n = 106), or other serious illnesses (n = 52)] and 549 were uninterested, lost to follow-up or died, leaving 1362 eligible women who completed the OsteoPerio Study baseline assessment. Of these, 934 women had sufficient baseline plasma samples for determination of 25(OH)D as well as complete baseline questionnaires and available data on periodontal disease [insufficient plasma samples (n = 407), missing baseline questionnaire data (n = 5), missing oral radiographs (n = 16); as previously described (14)]. Of these women, 673 also had sufficient plasma at OsteoPerio Study follow-up [details of follow-up exam previously described (23)]. Women were further excluded if they had colitis (a malabsorptive condition that could affect vitamin D absorption; n = 3). Two additional participants were excluded due to outlying 25(OH)D concentrations (530 nmol/L at baseline and 186 nmol/L at follow-up). The final sample used in analyses was composed of 668 women (see Supplemental Fig. 1 for the OsteoPerio Study participant flow diagram). The study was approved by the institutional review board at the University at Buffalo, and women provided written informed consent to participate in the WHIOS and OsteoPerio Study visits.
Data collection
Plasma vitamin D.
Fasting venous blood samples were obtained at OsteoPerio Study baseline and follow-up. Plasma samples were stored in liquid nitrogen at −196°C until the time of 25(OH)D assessment [as previously described (14)]. Samples were then removed from liquid nitrogen, put in a −80°C freezer, thawed, and divided into aliquots under standard protocol and placed in cryovials. Refrozen samples were shipped to the Heartland Assays, Inc., laboratory over a consecutive 4-mo period. Samples were assessed by competitive chemiluminescence immunoassay by using the DiaSorin LIAISON 25(OH)D assay. Although duplicate participant samples were analyzed in separate batches, the within-pair CV, determined by using the investigator’s blinded control samples nested in each batch, was 4.9%. Baseline samples for participants were analyzed first followed by samples collected at follow-up [as previously described (14)]. 25(OH)D concentrations (nmol/L) were defined using the following categories: deficiency (<30), inadequacy (30 to <50), and adequacy (2 categories: 50 to < 75 and ≥75) (17).
Physical measures.
Height (m) was measured with a stadiometer and weight (kg) with a calibrated balance beam; from these measures BMI (kg/m2) was calculated. DXA (Hologic QDR-4500A) was used to measure the percentage of fat mass of the trunk and total body. DXA was also used to measure hip and whole-body bone densities from which T scores were derived. T scores represent the number of SD below the mean bone density for healthy young adults, and lower T scores are indicative of lower bone density (24). Each of these physical measures was assessed at OsteoPerio Study baseline and follow-up clinic visits according to standard protocols.
Estimates of sunlight exposure.
At the WHIOS y 4 assessment (1997–2002; 1 y after baseline OsteoPerio Study enrollment), self-reported information regarding sunlight exposure was ascertained. Time spent outside (categorized as <30 min, 30 min to 2 h, and >2 h/d) during the summer and during other months (fall/winter/spring) for the past year was ascertained. Because these questions were asked only at WHIOS y 4, we also examined season of blood draw and physical activity, both assessed at OsteoPerio Study baseline and follow-up, also as proxy measures of sunlight exposure.
The season of blood draw was dichotomized into summer/fall (June–November) and winter/spring (December–May). On the basis of previous studies (25, 26), duration (min/wk) of recreational physical activity (sum of walking and mild, moderate, and strenuous exercise durations), as opposed to intensity, was chosen as a potential predictive variable. The duration of time spent outside in the sun while being physically active is thought to explain the frequently observed association between physical activity and 25(OH)D concentrations and thus is a reasonable proxy measure for individual sunlight exposure.
Other potential predictors.
At baseline, age (y), race (white/nonwhite), and education (high school or less, some education after high school, college degree or higher) were self-reported. Data for the following were collected at both OsteoPerio Study baseline and follow-up: vitamin D intake from foods and supplements (μg/d), alcohol intake (grams consumed in the past year), smoking status (never, past, or current), and current medication use (yes or no). Vitamin D intake in μg/d (1 μg = 40 IU) from foods was assessed with a validated FFQ that was specifically designed for the Women’s Health Initiative and described elsewhere (27). In addition, at OsteoPerio Study baseline and follow-up participants brought in any medications or supplements they were taking, and study personnel recorded the name, frequency of use, and dose. Both amount of vitamin D supplements (μg/d) taken and current use of medications (yes or no) were extracted from these data. On the basis of previous studies (2, 5), use of medications that could affect vitamin D status were examined, including hormone therapy, oral corticosteroids, and anticoagulant, anticonvulsant, antihypertensive, and osteoporosis-related medications (alendronate, miacalcin, raloxifene, and risedronate).
Statistical analysis
Among the 934 women with baseline plasma 25(OH)D samples, we compared baseline characteristics of those who were included (n = 668) and excluded (n = 266), as described above, from our analytic sample. Comparisons were made between age, education, and baseline measures of plasma 25(OH)D and intake of vitamin D, BMI, and duration of recreational physical activity. In addition, descriptive statistics were computed for baseline and follow-up measures of plasma 25(OH)D concentrations and potential predictors among the 668 women included in our analytic sample. Continuous variables (mean ± SD) were compared with paired t tests, and McNemar’s tests were used to compare frequencies of categorical variables.
Change in 25(OH)D was calculated by subtracting baseline from the 5-y follow-up 25(OH)D concentrations. Also determined were changes in the following potential predictor measures over 5 y (follow-up minus baseline): vitamin D intake from foods and supplements, duration of recreational physical activity, BMI, percentage of fat mass of the trunk and total body, hip and whole-body T scores, alcohol intake, smoking status, and medication use. For 25(OH)D and continuous predictors, negative change values indicate a decrease and positive change values indicate an increase in that variable from baseline to follow-up. For categorical predictors (smoking status and medication use), negative change values indicate that the behavior was present at baseline but ceased by follow-up, and positive change values indicate that the behavior was not present at baseline but commenced by follow-up. In addition, change in season of blood draw was examined as a predictor and was categorized as follows: 1) no change in season from baseline to follow-up, 2) blood draw during summer/fall at baseline and winter/spring at follow-up, and 3) blood draw during winter/spring at baseline and summer/fall at follow-up.
Simple linear regression was used to examine the crude associations between each potential predictor and change in 25(OH)D. To avoid multicollinearity between anthropometric measures, only BMI (baseline R2 = 0.001; change R2 = 0.015) was considered for inclusion in the multivariable model because BMI explained more of the variance (i.e., had the largest R2) in 25(OH)D change than did baseline waist circumference (R2 = 0.002) and percentage of fat mass of both the trunk (baseline R2 = 0.005; change R2 = 0.003) and total body (baseline and change R2 = 0.003) in univariate analyses.
Similarly, only vitamin D supplement intake (baseline R2 = 0.017; change R2 = 0.062) was considered for inclusion in the multivariable model because it explained more of the variance in change in 25(OH)D univariately than did vitamin D intake from foods (baseline and change R2 = 0.001) or total vitamin D intake from foods and supplements (baseline R2 = 0.012; change R2 = 0.044). Also, whole-body T score (baseline and change R2 = 0.005) was included in the multivariable model because it explained more of the variance in change in 25(OH)D univariately than did hip T score (baseline and change R2 = 0.001).
Next, stepwise linear regression was used to build the most predictive multivariable model for change in 25(OH)D. The entry criterion for predictors was P < 0.10, and the removal criterion was P > 0.05. The following variables were entered into the stepwise regression model: baseline 25(OH)D; baseline age, race, and education, time spent in the sun during summer and during fall/winter/spring at WHIOS y 4; both baseline and the change in vitamin D supplement intake, duration of recreational physical activity, BMI, whole-body T score, alcohol intake, smoking status, medication use (hormone therapy, cortical steroids, and anticoagulant, anticonvulsant, antihypertensive, and osteoporosis-related medications); and change in season of blood draw. The stepwise regression model was built in a sample size of n = 594. Any of the 668 participants who were missing potential predictor variables were dropped from the stepwise analysis (see Table 1 for noted missing data). The predictive variables identified from the stepwise regression were included in a multivariable model to determine the β-coefficients, respective P values, and R2 for the overall predictive model.
TABLE 1.
Characteristics assessed at baseline (1997–2000) and at 5-y follow-up (2002–2005): the Women’s Health Initiative OsteoPerio Study1
| Characteristic | Baseline | 5-y Follow-up | P value |
| Plasma 25(OH)D, nmol/L | 59.9 ± 22.2 | 67.6 ± 21.7 | <0.001 |
| Vitamin D status, n (%) | <0.001 | ||
| Deficient: <30 nmol/L | 58 (8.7) | 22 (3.3) | |
| Insufficient: 30 to <50 nmol/L | 157 (23.5) | 108 (16.2) | |
| Sufficient: 50–75 nmol/L | 303 (45.4) | 315 (47.2) | |
| ≥75 nmol/L | 150 (22.5) | 223 (33.4) | |
| Vitamin D intake2, μg/d | |||
| Foods | 5.0 ± 3.3 | 6.2 ± 6.1 | <0.001 |
| n3 | 658 | 666 | |
| Supplements | 8.3 ± 7.2 | 13.2 ± 8.9 | <0.001 |
| n3 | 668 | 664 | |
| Foods and supplements | 13.4 ± 7.9 | 19.4 ± 11.0 | <0.001 |
| n3 | 658 | 662 | |
| Estimates of sunlight exposure, n (%) | |||
| Time spent outside (summer)3,4 | — | ||
| <30 min/d | 158 (24.2) | — | |
| 30 min to 2 h/d | 327 (50.0) | — | |
| >2 h/d | 169 (25.8) | — | |
| Time spent outside (fall/spring/winter)3,4 | — | ||
| <30 min/d | 245 (37.6) | — | |
| 30 min to 2 h/d | 331 (50.8) | — | |
| >2 h/d | 75 (11.5) | — | |
| Season of blood draw, n (%) | 0.06 | ||
| Summer/fall | 371 (55.5) | 347 (52.0) | |
| Winter/spring | 297 (44.5) | 321 (48.1) | |
| Duration of recreational physical activity, min/wk5 | 209 ± 184 | 214 ± 196 | 0.53 |
| Anthropometric measures | |||
| BMI, kg/m2 | 26.6 ± 4.9 | 26.7 ± 5.1 | 0.36 |
| n3 | 668 | 667 | |
| Waist circumference6, cm | 84.2 ± 12.0 | — | — |
| n3 | 652 | — | |
| Trunk fat mass, % | 34.5 ± 7.2 | 34.0 ± 7.6 | 0.001 |
| n3 | 662 | 663 | |
| Total body fat mass, % | 36.7 ± 5.7 | 36.9 ± 6.0 | 0.07 |
| n3 | 662 | 663 | |
| T scores | |||
| Total hip | −0.60 ± 1.1 | −0.72 ± 1.1 | <0.001 |
| n3 | 657 | 662 | |
| Whole-body | −0.39 ± 1.3 | −0.52 ± 1.3 | <0.001 |
| n3 | 662 | 652 | |
| Alcohol intake, g | 13.3 ± 20.1 | 15.6 ± 26.1 | 0.008 |
| n3 | 661 | 658 | |
| Smoking status, n (%) | 0.92 | ||
| Never smoked | 369 (55.2) | 369 (55.2) | |
| Past smoker | 285 (42.7) | 287 (43.0) | |
| Current smoker | 14 (2.1) | 12 (1.8) | |
| Current medication use, n (%) | |||
| Hormone therapy use | <0.001 | ||
| No | 327 (49.0) | 424 (63.5) | |
| Yes | 341 (51.1) | 244 (36.5) | |
| Oral corticosteroids3 | 0.47 | ||
| No | 648 (98.2) | 645 (97.7) | |
| Yes | 12 (1.8) | 15 (2.3) | |
| Anticoagulant medications3 | <0.001 | ||
| No | 612 (92.5) | 272 (41.1) | |
| Yes | 50 (7.6) | 390 (58.9) | |
| Anticonvulsant medications3 | 0.039 | ||
| No | 646 (97.9) | 637 (96.5) | |
| Yes | 14 (2.1) | 23 (3.5) | |
| Antihypertensive medications3 | <0.001 | ||
| No | 443 (67.1) | 303 (45.9) | |
| Yes | 217 (32.9) | 357 (54.1) | |
| Osteoporosis-related medications3,7 | <0.001 | ||
| No | 598 (90.6) | 428 (64.9) | |
| Yes | 62 (9.4) | 232 (35.2) |
Values are means ± SD or n (%); n = 668. OsteoPerio, Osteoporosis and Periodontal Disease; 25(OH)D, 25-hydroxyvitamin D.
One microgram of vitamin D = 40 IU.
Total n does not add up to 668 due to missing data. For any one variable, 3% (n = 20) or fewer of the participants had missing data. For continuous variables with missing data, sample sizes are listed below values.
Assessed at the Women’s Health Initiative Observational Study 4-y follow-up.
Recreational physical activity included walking and mild, moderate, and strenuous exercise.
Waist circumference was measured only at the baseline clinic visit.
Osteoporosis-related medications included alendronate, miacalcin, raloxifene, and risedronate.
As an exploratory analysis, we explored whether a participant’s baseline vitamin D status might have influenced his or her response [i.e., change in 25(OH)D] to vitamin D supplementation. For this analysis, tertiles of baseline 25(OH)D and categories of change in vitamin D supplement intake (decreased, no change, or increased from baseline to follow-up) were used instead of the continuous data. An interaction between tertiles of baseline 25(OH)D and categories of change in vitamin D supplement intake was examined by adding an interaction term [baseline 25(OH)D × change in vitamin D supplement intake] to the final predictive model. Furthermore, ANOVA, with post hoc Tukey tests (28) for pairwise comparisons, was used to determine whether mean change in 25(OH)D over time differed by change in vitamin D supplement intake in each tertile of baseline 25(OH)D.
All statistical analyses were performed with SAS version 9.2 (29). All tests were 2-sided, and P < 0.05 was considered significant.
Results
Women with baseline 25(OH)D samples who were included (n = 668) compared with those who were excluded (n = 266) from our analytical sample were younger (mean age: 65.7 vs. 68.9 y; P < 0.0001) and had slightly lower mean intake of vitamin D from foods (5.0 vs. 5.5 μg/d, P < 0.04). There were no significant differences between women with respect to educational level and baseline measures of plasma 25(OH)D, vitamin D intake from supplements or food and supplements combined, BMI, or time spent in recreational physical activity (see Supplemental Table 1).
Most women were white (98.7%) and most had at least a college degree (48.2%). The mean age of participants was 65.7 y at baseline and 70.2 y at follow-up. Mean 25(OH)D increased from 59.9 nmol/L (range: 5.9–147 nmol/L) at baseline to 67.6 nmol/L (range: 13.0–145 nmol/L) at follow-up (P < 0.001) (Table 1). Furthermore, more women had adequate 25(OH)D (≥50 nmol/L) at follow-up (80.5%) compared with baseline (67.8%; P < 0.001). Over the 5-y follow-up period, vitamin D intake from foods and supplements, alcohol intake, and medication use (anticoagulant, anticonvulsant, antihypertensive, and osteoporosis-related medications) increased. Alternatively, women’s percentage of fat mass of the trunk, hip, and whole-body T scores, and hormone therapy use decreased. At baseline, 72% of women reported using vitamin D supplements and 89% reported use at follow-up (data not shown).
The final variables included in the predictive model are noted in Table 2. The final predictive model included a sample size of 643 because 25 subjects were missing data on the selected predictors. Both change in whole-body T score and baseline current hormone therapy use were included in the predictive model, although their associated P values for the multivariate model were >0.05. In the stepwise regression analysis, which was performed in a sample size of 594 women, these variables were selected on the basis of the previously noted inclusion and exclusion criteria. In the final predictive model, with a larger sample size of 643, they were no longer significant. We have included them in our multivariate model to be conservative.
TABLE 2.
Multivariable predictive model of change in 25(OH)D: the Women’s Health Initiative OsteoPerio Study1
| Predictor | Unadjusted β (SE) | P value for unadjusted β | Adjusted β (SE) | P value for adjusted β | Partial R2 | Model R2 |
| Baseline plasma 25(OH)D, nmol/L | −0.37 (0.03) | <0.001 | −0.40 (0.03) | <0.001 | 0.19 | 0.19 |
| Change in vitamin D supplement intake (2.5 μg/d) | 1.26 (0.19) | <0.001 | 1.34 (0.19) | <0.001 | 0.07 | 0.25 |
| Change in duration of recreational physical activity (60 min/wk) | 0.80 (0.23) | <0.001 | 0.82 (0.22) | <0.001 | 0.01 | 0.27 |
| Change in season of blood draw | 0.01 | 0.28 | ||||
| Blood drawn during same seasons at baseline and follow-up | Referent | Referent | ||||
| Blood drawn during summer/fall at baseline and during winter/spring at follow-up | −5.51 (2.10) | 0.009 | −4.74 (1.79) | 0.008 | ||
| Blood drawn during winter/spring at baseline and during summer/fall at follow-up | 4.10 (2.43) | 0.09 | 4.18 (2.06) | 0.043 | ||
| Change in BMI, kg/m2 | −1.21 (0.38) | 0.002 | −0.75 (0.33) | 0.023 | 0.01 | 0.29 |
| Baseline vitamin D supplement intake (2.5 μg/d) | −0.76 (0.25) | 0.003 | 0.55 (0.25) | 0.027 | 0.01 | 0.29 |
| Baseline duration of recreational physical activity (60 min/wk) | −0.34 (0.24) | 0.15 | 0.55 (0.23) | 0.016 | 0.01 | 0.30 |
| Change in whole-body T score | 2.32 (1.37) | 0.09 | 1.52 (1.17) | 0.20 | 0.01 | 0.30 |
| Baseline current hormone therapy use | 0.01 | 0.31 | ||||
| No | Referent | Referent | ||||
| Yes | 1.40 (1.46) | 0.34 | 2.10 (1.26) | 0.10 | ||
| Fully adjusted model R2 | 0.31 |
Changes in plasma 25(OH)D and potential predictors were calculated by subtracting the baseline (1997–2000) measure from the follow-up (2002–2005) measure. The final predictive model (shown above) was adjusted for all other covariates included in the model. Total n does not add up to 668 due to missing data on the final predictors, leaving a sample size of 643. OsteoPerio, Osteoporosis and Periodontal Disease; 25(OH)D, 25-hydroxyvitamin D.
In univariate analyses, increasing intake in vitamin D supplements, increasing duration of recreational physical activity, and increasing whole-body T score over time were positively associated with increasing 25(OH)D concentrations (Table 2). Baseline current hormone therapy use compared with no use was also associated with increasing 25(OH)D. Baseline 25(OH)D, baseline vitamin D supplement use, baseline duration of recreational physical activity, and increasing BMI over time were inversely associated with change in 25(OH)D. Change in 25(OH)D also differed by change in season of blood draw. Compared with women who had their blood drawn in the same seasons at baseline and at follow-up, 25(OH)D decreased over time among women who had their blood drawn in summer/fall at baseline and winter/spring at follow-up. The reverse was observed for women who had their blood drawn in winter/spring at baseline and summer/fall at follow-up.
The final multivariable predictive model explained 31% of the variance in change in 25(OH)D (Table 2). Of this, baseline 25(OH)D accounted for 18.6% and change in vitamin D supplement intake accounted for 6.6% of the variance in change in 25(OH)D. Other more modest, yet statistically significant, predictors included baseline and change in duration of recreational physical activity, change in season of blood draw, change in BMI, baseline vitamin D supplement intake, change in whole-body T score, and baseline hormone therapy use. However, together, these predictors accounted for only an additional 5.4% of the variance in 25(OH)D change over time.
All predictors were associated with change in 25(OH)D in the multivariate model in the same manner as observed in the univariate model except for baseline vitamin D supplement use and baseline duration of recreational physical activity, both of which accounted for a small percentage of variance (<1%) in 25(OH)D change. Both predictors were negatively associated with change in 25(OH)D in the univariate model but were positively associated with change in 25(OH)D in the multivariable model. Further investigation showed that the change in the β-coefficient for baseline supplement use became positive after adding both baseline plasma 25(OH)D and change in vitamin D supplement intake to the model. Change in the β-coefficient for baseline physical activity became positive after the addition of baseline plasma 25(OH)D to the model.
In exploratory analyses, we examined whether change in 25(OH)D was predicted by an interaction between change in vitamin D supplement intake and baseline 25(OH)D after adjustment for statistically significant predictors of change in 25(OH)D including duration of recreational physical activity, season of blood draw, BMI, whole-body T score, and hormone therapy use. The adjusted interaction between tertiles of baseline 25(OH)D and categories of change in vitamin D supplement intake (decreased, no change, or increased) was borderline significant (P = 0.06).
Mean change in 25(OH)D differed significantly by change in supplement use category (Fig. 1). This was true in the lowest (P < 0.001), middle (P = 0.034), and highest (P = 0.043) tertiles of baseline 25(OH)D. In each tertile of baseline plasma 25(OH)D, women who increased their vitamin D supplement intake had significantly greater changes in plasma 25(OH)D than did women who decreased their vitamin D supplement intake. The same was true when women who increased their vitamin D supplement intake were compared with those who did not change their vitamin D supplement intake [within the same tertile of baseline 25(OH)D]. This increase was most pronounced among women in the lowest tertile of baseline 25(OH)D (mean ± SE: 22.9 ± 1.8 nmol/L). Vitamin D supplement intake (mean ± SE) increased similarly for women in the lowest (10.0 ± 6.1 μg/d), middle (11.1 ± 9.5 μg/d), and highest (10.1 ± 6.3 μg/d) tertiles of baseline 25(OH)D.
FIGURE 1.
Boxplots of changes over 5 y in plasma 25(OH)D concentrations (nmol/L) by categories of change in vitamin D supplement intake of women in the lowest (5.9–51.0 nmol/L; n = 215), middle (51.1–67.0 nmol/L; n = 217), and highest (67.2–147 nmol/L; n = 211) tertiles of baseline 25(OH)D. Values are means ± SEM. Means were adjusted for significant predictors of change in 25(OH)D, including physical activity, season of blood draw, BMI, whole-body T score, and estrogen use. Within each tertile, means with unlike letters differ, P < 0.05. T, tertile; 25(OH)D, 25-hydroxyvitamin D.
Discussion
Plasma 25(OH)D concentrations increased by 7.7 nmol/L and the proportion of vitamin D adequacy increased in this sample of postmenopausal women over 5 y. Predictors of change in 25(OH)D explained only 31% of the variation in change in 25(OH)D and included baseline 25(OH)D, vitamin D supplement intake, duration of recreational physical activity, season of blood draw, BMI, whole-body T score, and hormone therapy use. This proportion is greater than or similar to previous prospective studies, which explained 1.3–29% of the variance in change in 25(OH)D (13, 15, 16).
Baseline 25(OH)D and change in vitamin D supplement intake explained the most variance in change in 25(OH)D (25%). Lower baseline 25(OH)D concentrations were associated with an increase in 25(OH)D after 5 y. Baseline plasma 25(OH)D concentrations were unknown to women in this study at the baseline exam because the vitamin D assays were conducted in 2010. Therefore, we cannot assume that direct knowledge of their blood 25(OH)D concentrations prompted this change, but perhaps medical conditions associated with low vitamin D status (e.g., poor bone health) did.
Exploratory analyses showed that change in 25(OH)D by supplement use change over time was modified by women’s baseline 25(OH)D. Women in the lowest tertile of baseline 25(OH)D (<51.0 nmol/L) and who increased their vitamin D supplement intake over time had a greater increase in 25(OH)D than did other women in the lowest tertile. Women in the lowest tertile of baseline 25(OH)D and who increased their vitamin D supplement intake over time also had a larger increase in 25(OH)D compared with women in other tertiles who increased their supplement intake. This suggests that a woman’s blood 25(OH)D response to supplementation depends on her baseline blood 25(OH)D.
In addition, change in time spent engaged in recreational physical activity and change in season of blood draw, both proxy measures of sunlight exposure, also explained a substantial proportion of variance in change in 25(OH)D (3%). This is not surprising because many previous studies have found that sunlight exposure predicts higher 25(OH)D (3–5, 30). Although duration of physical activity was examined as a proxy measure of sunlight exposure, it should be noted that location was not ascertained. Thus, the positive association we observed between change in 25(OH)D and change in duration of recreational physical activity may have been attenuated by not being able to exclude indoor activity from our estimate of duration of physical activity.
Changes in BMI and whole-body T score and baseline hormone therapy use together explained an additional 2% of the variance in change in 25(OH)D. Similar to Ding et al. (15), we found an inverse association between changes in BMI and 25(OH)D, independent of other factors (e.g., physical activity). Mason et al. (31) recently conducted a randomized controlled trial that found that weight loss through diet modification and exercise was associated with increases in 25(OH)D. Cross-sectional studies have also found greater BMI to be predictive of lower 25(OH)D (2–5, 7, 8), and it has been hypothesized that this is the case because vitamin D becomes sequestered in fat tissue (32). However, change in BMI was a slightly stronger predictor of change in 25(OH)D in this study than was percentage of fat mass of the total body and trunk. Likewise, several previous studies have found BMI, but not percentage of fat mass, to be related to 25(OH)D (33–37). The reason for this association is not clear. Perhaps BMI also captures aspects of individual behavior, such as physical activity, sun exposure, and diet, that also predict vitamin D status.
Change in whole-body T score was also a predictor of change in 25(OH)D such that 25(OH)D increased with increasing T score in our sample. Women in the OsteoPerio Study received the results of their DXA scan for bone mineral density. Thus, women with a low whole-body T score at baseline, which is indicative of poor bone health, might have been advised by their doctors to increase their calcium intake (some calcium-rich foods are also fortified with vitamin D), supplementation (calcium supplements often contain vitamin D to increase calcium absorption), and weight-bearing physical activity (if performed outside would increase sun exposure), which could lead to increases in both T score and 25(OH)D. Randomized controlled trials have found that bone density increases after supplementation with vitamin D (38, 39). Furthermore, it has been suggested that bone density might be a marker for optimal 25(OH)D required to sustain bone health (39).
Although we found that both hormone therapy users and nonusers increased their 25(OH)D status, baseline hormone therapy users had a greater increase in 25(OH)D. Sowers et al. (40) found that exogenous estrogen use in women was associated with higher 25(OH)D concentrations, which might have been the case because estrogen use has been shown to increase vitamin D–binding protein in the blood (12) which might lead to an increase in 25(OH)D. An alternative explanation is that estrogen and vitamin D compete with each other to bind with megalin, a cellular uptake membrane receptor (41).
This study has several limitations. Data were collected only twice over 5 y. Many health and lifestyle changes that could have affected 25(OH)D status might have occurred during this timeframe, which might not have been fully captured by the questionnaires. Also, measurement error, such as inaccurate recall, might have led to inaccurate assessment of the predictors. In addition, our sample specifically included white postmenopausal women residing in a northern region of the United States. The participants of the WHIOS are relatively well educated, and it appears that the subset of women included in our current analysis may be slightly younger than other WHIOS women in our broader OsteoPerio Study baseline sample. In our sample, >90% of women reported using vitamin D supplements at follow-up (2002–2005), which is greater than the 56.3% of women aged >60 y who reported use in the 2003–2006 NHANES (42). This suggests that our participants likely engage in health-conscious behaviors, such as supplement use, more so than the general population. Our findings might not be generalizable to males, other age groups, races, or geographic locations, and our results should be interpreted with this limitation in mind.
Despite these potential limitations, our study is, to our knowledge, one of very few prospective studies that have examined predictors of change in 25(OH)D. These data were useful for determining which factors are most responsible for maintaining or increasing 25(OH)D to obtain vitamin D adequacy, as opposed to assessing cross-sectional associations between vitamin D status and predictors. In fact, the prediction of change in 25(OH)D in this study was better than in most cross-sectional studies that explained 19–39% of the variance in a single measure of 25(OH)D (43–49). An additional strength of our study was that 25(OH)D measurements were available for the majority of our sample at both study visits (69%) and analyzed by the same lab using the same assay for plasma from both time points, which likely minimizes measurement error.
The WHIOS’s detailed database allowed us to examine a comprehensive list of potential predictors including demographic, dietary, health-related, and other lifestyle factors. Despite this wealth of data, we were able to explain only 31% of the variation in change in 25(OH)D. There might be other predictors that we did not have measures of or could not account for in our study. Other studies have found that leptin (15) and albumin (16) concentrations explained a substantial proportion of variation in change in 25(OH)D. Race has commonly been shown to be a predictor of 25(OH)D status (3, 5, 7); however, our sample was predominantly white. Furthermore, there is some evidence that genetic factors, such as single nucleotide polymorphisms on the vitamin D–binding protein gene, predict 25(OH)D status (9, 50, 51).
In conclusion, we identified several modifiable and nonmodifiable factors that contributed to the prediction of 5-y change in 25(OH)D, including baseline 25(OH)D, vitamin D supplement intake, duration of physical activity, season of blood draw, BMI, whole-body T score, and hormone therapy use. Increases in blood 25(OH)D associated with change in vitamin D supplementation appear to be influenced by vitamin D status prior to changes in supplementation. Women with low blood 25(OH)D had the greatest response to vitamin D supplementation. Integrating modifiable predictors into one’s lifestyle, especially vitamin D supplement intake, could help to maintain or achieve adequate 25(OH)D over time.
Supplementary Material
Acknowledgments
M.A.K. and A.E.M. contributed to the design of the research, performed statistical analyses, interpreted the analyses, and wrote the manuscript; and J.W.-W., M.E.P., C.A.D., and K.M.H. contributed to the design of the research, interpreting the analyses, and writing of the manuscript. A.E.M. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript.
Footnotes
Supported by NIH grants 1R21DE020918 (awarded to A.E.M.) and 1R01DE13505 (awarded to J.W.-W.) from the National Institute of Dental and Craniofacial Research and a grant awarded to J.W.-W. from the Department of Defense (DAMD179616319). The Women’s Health Initiative (WHI) program is supported by contracts from the National Heart, Lung, and Blood Institute, NIH (N01WH32122). The WHI program is funded by the National Heart, Lung, and Blood Institute, NIH, U.S. Department of Health and Human Services through contracts N01WH22110, 24152, 32100-2, 32105-6, 32108-9, 32111-13, 32115, 32118-32119, 32122, 42107-26, 42129-32, and 44221.
Supplemental Figure 1 and Supplemental Table 1 are available from the “Online Supporting Material” link in the online posting of the article and from the same link in the online table of contents at http://jn.nutrition.org.
Abbreviations used: 25(OH)D, 25-hydroxyvitamin D; OsteoPerio Study, Osteoporosis and Periodontal Disease Study; WHIOS, Women’s Health Initiative Observational Study.
Literature Cited
- 1.Holick MF. Vitamin D deficiency. N Engl J Med. 2007;357:266–81 [DOI] [PubMed] [Google Scholar]
- 2.Nesby-O'Dell S, Scanlon KS, Cogswell ME, Gillespie C, Hollis BW, Looker AC, Allen C, Doughertly C, Gunter EW, Bowman BA. Hypovitaminosis D prevalence and determinants among African American and white women of reproductive age: third National Health and Nutrition Examination Survey, 1988–1994. Am J Clin Nutr. 2002;76:187–92 [DOI] [PubMed] [Google Scholar]
- 3.Looker AC, Pfeiffer CM, Lacher DA, Schleicher RL, Picciano MF, Yetley EA. Serum 25-hydroxyvitamin D status of the US population: 1988–1994 compared with 2000–2004. Am J Clin Nutr. 2008;88:1519–27 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Macdonald HM, Mavroeidi A, Barr RJ, Black AJ, Fraser WD, Reid DM. Vitamin D status in postmenopausal women living at higher latitudes in the UK in relation to bone health, overweight, sunlight exposure and dietary vitamin D. Bone. 2008;42:996–1003 [DOI] [PubMed] [Google Scholar]
- 5.Millen AE, Wactawski-Wende J, Pettinger M, Melamed ML, Tylavsky FA, Liu S, Robbins J, LaCroix AZ, LeBoff MS, Jackson RD. Predictors of serum 25-hydroxyvitamin D concentrations among postmenopausal women: the Women's Health Initiative Calcium plus Vitamin D clinical trial. Am J Clin Nutr. 2010;91:1324–35 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Giovannucci E, Liu Y, Rimm EB, Hollis BW, Fuchs CS, Stampfer MJ, Willett WC. Prospective study of predictors of vitamin D status and cancer incidence and mortality in men. J Natl Cancer Inst. 2006;98:451–9 [DOI] [PubMed] [Google Scholar]
- 7.Holick MF, Siris ES, Binkley N, Beard MK, Khan A, Katzer JT, Petruschke RA, Chen E, de Papp AE. Prevalence of Vitamin D inadequacy among postmenopausal North American women receiving osteoporosis therapy. J Clin Endocrinol Metab. 2005;90:3215–24 [DOI] [PubMed] [Google Scholar]
- 8.Rizzoli R, Eisman JA, Norquist J, Ljunggren O, Krishnarajah G, Lim SK, Chandler J. Risk factors for vitamin D inadequacy among women with osteoporosis: an international epidemiological study. Int J Clin Pract. 2006;60:1013–9 [DOI] [PubMed] [Google Scholar]
- 9.Engelman CD, Fingerlin TE, Langefeld CD, Hicks PJ, Rich SS, Wagenknecht LE, Bowden DW, Norris JM. Genetic and environmental determinants of 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D levels in Hispanic and African Americans. J Clin Endocrinol Metab. 2008;93:3381–8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Platz EA, Leitzmann MF, Hollis BW, Willett WC, Giovannucci E. Plasma 1,25-dihydroxy- and 25-hydroxyvitamin D and subsequent risk of prostate cancer. Cancer Causes Control. 2004;15:255–65 [DOI] [PubMed] [Google Scholar]
- 11.Hofmann JN, Yu K, Horst RL, Hayes RB, Purdue MP. Long-term variation in serum 25-hydroxyvitamin D concentration among participants in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial. Cancer Epidemiol Biomarkers Prev. 2010;19:927–31 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Rejnmark L, Lauridsen AL, Brot C, Vestergaard P, Heickendorff L, Nexo E, Mosekilde L. Vitamin D and its binding protein Gc: long-term variability in peri- and postmenopausal women with and without hormone replacement therapy. Scand J Clin Lab Invest. 2006;66:227–38 [DOI] [PubMed] [Google Scholar]
- 13.Jorde R, Sneve M, Hutchinson M, Emaus N, Figenschau Y, Grimnes G. Tracking of serum 25-hydroxyvitamin D levels during 14 years in a population-based study and during 12 months in an intervention study. Am J Epidemiol. 2010;171:903–8 [DOI] [PubMed] [Google Scholar]
- 14.Meng JE, Hovey KM, Wactawski-Wende J, Andrews CA, Lamonte MJ, Horst RL, Genco RJ, Millen AE. Intraindividual variation in plasma 25-hydroxyvitamin D measures 5 years apart among postmenopausal women. Cancer Epidemiol Biomarkers Prev. 2012;21:916–24 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ding C, Parameswaran V, Blizzard L, Burgess J, Jones G. Not a simple fat-soluble vitamin: changes in serum 25-(OH)D levels are predicted by adiposity and adipocytokines in older adults. J Intern Med. 2010;268:501–10 [DOI] [PubMed] [Google Scholar]
- 16.Nakamura K, Nishiwaki T, Ueno K, Yamamoto M. Age-related decrease in serum 25-hydroxyvitamin D concentrations in the frail elderly: a longitudinal study. J Bone Miner Metab. 2007;25:232–6 [DOI] [PubMed] [Google Scholar]
- 17.Institute of Medicine Dietary Reference Intakes for calcium and vitamin D. Washington, DC: The National Academies Press, 2011 [PubMed] [Google Scholar]
- 18.Nelson ML, Blum JM, Hollis BW, Rosen C, Sullivan SS. Supplements of 20 μg/d cholecalciferol optimized serum 25-hydroxyvitamin D concentrations in 80% of premenopausal women in winter. J Nutr. 2009;139:540–6 [DOI] [PubMed] [Google Scholar]
- 19.The Women's Health Initiative Study Group Design of the Women's Health Initiative clinical trial and observational study. Control Clin Trials. 1998;19:61–109 [DOI] [PubMed] [Google Scholar]
- 20.Anderson GL, Manson J, Wallace R, Lund B, Hall D, Davis S, Shumaker S, Wang CY, Stein E, Prentice RL. Implementation of the Women's Health Initiative study design. Ann Epidemiol. 2003;13:S5–17 [DOI] [PubMed] [Google Scholar]
- 21.Hays J, Hunt JR, Hubbell FA, Anderson GL, Limacher M, Allen C, Rossouw JE. The Women's Health Initiative recruitment methods and results. Ann Epidemiol. 2003;13:S18–77 [DOI] [PubMed] [Google Scholar]
- 22.Wactawski-Wende J, Hausmann E, Hovey K, Trevisan M, Grossi S, Genco RJ. The association between osteoporosis and alveolar crestal height in postmenopausal women. J Periodontol. 2005;76:2116–24 [DOI] [PubMed] [Google Scholar]
- 23.Bole C, Wactawski-Wende J, Hovey KM, Genco RJ, Hausmann E. Clinical and community risk models of incident tooth loss in postmenopausal women from the Buffalo Osteo Perio Study. Community Dent Oral Epidemiol. 2010;38:487–97 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Nelson HD, Helfand M, Woolf SH, Allan JD. Screening for postmenopausal osteoporosis: a review of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med. 2002;137:529–41 [DOI] [PubMed] [Google Scholar]
- 25.Scragg R, Camargo CA., Jr Frequency of leisure-time physical activity and serum 25-hydroxyvitamin D levels in the US population: results from the Third National Health and Nutrition Examination Survey. Am J Epidemiol. 2008;168:577–86; discussion 87–91 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Kluczynski MA, Lamonte MJ, Mares JA, Wactawski-Wende J, Smith AW, Engelman CD, Andrews CA, Snetselaar LG, Sarto GE, Millen AE. Duration of physical activity and serum 25-hydroxyvitamin D status of postmenopausal women. Ann Epidemiol. 2011;21:440–9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Patterson RE, Kristal AR, Tinker LF, Carter RA, Bolton MP, Agurs-Collins T. Measurement characteristics of the Women's Health Initiative food frequency questionnaire. Ann Epidemiol. 1999;9:178–87 [DOI] [PubMed] [Google Scholar]
- 28.O'Rourke N, Hatcher L, Stepanski EJ. Using SAS for univariate and multivariate statistics. 2nd ed. Cary (NC): SAS Institute and Wiley; 2005 [Google Scholar]
- 29. SAS Institute. Statistical Analysis Software. Version 9.2. Cary (NC): SAS Institute, 2009.
- 30.Zgaga L, Theodoratou E, Farrington SM, Agakov F, Tenesa A, Walker M, Knox S, Wallace AM, Cetnarskyj R, McNeill G, et al. Diet, environmental factors, and lifestyle underlie the high prevalence of vitamin D deficiency in healthy adults in Scotland, and supplementation reduces the proportion that are severely deficient. J Nutr. 2011;141:1535–42 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Mason C, Xiao L, Imayama I, Duggan CR, Bain C, Foster-Schubert KE, Kong A, Campbell KL, Wang CY, Neuhouser ML, et al. Effects of weight loss on serum vitamin D in postmenopausal women. Am J Clin Nutr. 2011;94:95–103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Wortsman J, Matsuoka LY, Chen TC, Lu Z, Holick MF. Decreased bioavailability of vitamin D in obesity. Am J Clin Nutr. 2000;72:690–3 [DOI] [PubMed] [Google Scholar]
- 33.Rodríguez-Rodríguez E, Navia-Lomban B, Lopez-Sobaler AM, Ortega RM. Associations between abdominal fat and body mass index on vitamin D status in a group of Spanish schoolchildren. Eur J Clin Nutr. 2010;64:461–7 [DOI] [PubMed] [Google Scholar]
- 34.Foo LH, Zhang Q, Zhu K, Ma G, Trube A, Greenfield H, Fraser DR. Relationship between vitamin D status, body composition and physical exercise of adolescent girls in Beijing. Osteoporos Int. 2009;20:417–25 [DOI] [PubMed] [Google Scholar]
- 35.Blum M, Dallal GE, Dawson-Hughes B. Body size and serum 25 hydroxy vitamin D response to oral supplements in healthy older adults. J Am Coll Nutr. 2008;27:274–9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Rueda S, Fernandez-Fernandez C, Romero F, Martinez de Osaba J, Vidal J. Vitamin D, PTH, and the metabolic syndrome in severely obese subjects. Obes Surg. 2008;18:151–4 [DOI] [PubMed] [Google Scholar]
- 37.McGill AT, Stewart JM, Lithander FE, Strik CM, Poppitt SD. Relationships of low serum vitamin D3 with anthropometry and markers of the metabolic syndrome and diabetes in overweight and obesity. Nutr J. 2008;7:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Dawson-Hughes B, Harris SS, Krall EA, Dallal GE, Falconer G, Green CL. Rates of bone loss in postmenopausal women randomly assigned to one of two dosages of vitamin D. Am J Clin Nutr. 1995;61:1140–5 [DOI] [PubMed] [Google Scholar]
- 39.Ooms ME, Roos JC, Bezemer PD, van der Vijgh WJ, Bouter LM, Lips P. Prevention of bone loss by vitamin D supplementation in elderly women: a randomized double-blind trial. J Clin Endocrinol Metab. 1995;80:1052–8 [DOI] [PubMed] [Google Scholar]
- 40.Sowers MR, Wallace RB, Hollis BW, Lemke JH. Parameters related to 25-OH-D levels in a population-based study of women. Am J Clin Nutr. 1986;43:621–8 [DOI] [PubMed] [Google Scholar]
- 41.Hammes A, Andreassen TK, Spoelgen R, Raila J, Hubner N, Schulz H, Metzger J, Schweigert FJ, Luppa PB, Nykjaer A, et al. Role of endocytosis in cellular uptake of sex steroids. Cell. 2005;122:751–62 [DOI] [PubMed] [Google Scholar]
- 42.Gahche J, Bailey R, Burt V, Hughes J, Yetley E, Dwyer J, Picciano MF, McDowell M, Sempos C. Dietary supplement use among U.S. adults has increased since NHA NES III (1988–1994). NCHS Data Brief, no 61. Hyattsville, MD. National Center for Health Statistics. 2011:1–8 [PubMed] [Google Scholar]
- 43.Jacques PF, Felson DT, Tucker KL, Mahnken B, Wilson PW, Rosenberg IH, Rush D. Plasma 25-hydroxyvitamin D and its determinants in an elderly population sample. Am J Clin Nutr. 1997;66:929–36 [DOI] [PubMed] [Google Scholar]
- 44.Bhattoa HP, Bettembuk P, Ganacharya S, Balogh A. Prevalence and seasonal variation of hypovitaminosis D and its relationship to bone metabolism in community dwelling postmenopausal Hungarian women. Osteoporos Int. 2004;15:447–51 [DOI] [PubMed] [Google Scholar]
- 45.Brustad M, Alsaker E, Engelsen O, Aksnes L, Lund E. Vitamin D status of middle-aged women at 65–71 degrees N in relation to dietary intake and exposure to ultraviolet radiation. Public Health Nutr. 2004;7:327–35 [DOI] [PubMed] [Google Scholar]
- 46.Andersen R, Molgaard C, Skovgaard LT, Brot C, Cashman KD, Chabros E, Charzewska J, Flynn A, Jakobsen J, Karkkainen M, et al. Teenage girls and elderly women living in northern Europe have low winter vitamin D status. Eur J Clin Nutr. 2005;59:533–41 [DOI] [PubMed] [Google Scholar]
- 47.Burgaz A, Akesson A, Oster A, Michaelsson K, Wolk A. Associations of diet, supplement use, and ultraviolet B radiation exposure with vitamin D status in Swedish women during winter. Am J Clin Nutr. 2007;86:1399–404 [DOI] [PubMed] [Google Scholar]
- 48.Lucas JA, Bolland MJ, Grey AB, Ames RW, Mason BH, Horne AM, Gamble GD, Reid IR. Determinants of vitamin D status in older women living in a subtropical climate. Osteoporos Int. 2005;16:1641–8 [DOI] [PubMed] [Google Scholar]
- 49.Lappe JM, Davies KM, Travers-Gustafson D, Heaney RP. Vitamin D status in a rural postmenopausal female population. J Am Coll Nutr. 2006;25:395–402 [DOI] [PubMed] [Google Scholar]
- 50.Davies JR, Chang YM, Snowden H, Chan M, Leake S, Karpavicius B, Haynes S, Kukalizch K, Randerson-Moor J, Elliott F, et al. The determinants of serum vitamin D levels in participants in a melanoma case-control study living in a temperate climate. Cancer Causes Control. 2011;22:1471–82 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 51.Simon KC, Munger KL, Kraft P, Hunter DJ, De Jager PL, Ascherio A. Genetic predictors of 25-hydroxyvitamin D levels and risk of multiple sclerosis. J Neurol. 2011;258:1676–82 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.