Association of High Intakes of Vitamins B₆ and B₁₂ From Food and Supplements With Risk of Hip Fracture Among Postmenopausal Women in the Nurses’ Health Study

Key Points

Question  Are high intakes of vitamins B₆ and B₁₂ associated with an increased risk of hip fracture?

Findings  In a cohort study of 75 864 US postmenopausal women, 2304 had a hip fracture. A combined high intake of vitamins B₆ and B₁₂ was associated with an increased risk of hip fracture.

Meaning  These results add to the evidence suggesting that caution should be used in vitamin supplementation when there is no apparent deficiency.

Importance  Vitamin supplementation far exceeding recommended doses is popular in segments of the population. However, adverse effects can occur. In a previous secondary analysis of combined data from 2 double-blind randomized clinical trials (RCTs), an unexpected increased risk of hip fracture was found among those treated with high doses of vitamin B₆ in combination with vitamin B₁₂.

Objectives  To study if high intakes of vitamins B₆ and B₁₂ from food and supplements were associated with a risk of hip fracture in the Nurses’ Health Study and to investigate whether combined high intakes of both vitamins conferred a particularly increased fracture risk.

Design, Setting, and Participants  In this prospective cohort study, 75 864 postmenopausal women in the United States were followed up from June 1984 through May 2014. The dates of analysis were July 2016 to June 2018. Information on hip fracture and a wide range of potential confounders was collected at baseline and with biennial follow-up questionnaires. Extensive dietary information was collected approximately every 4 years with a semiquantitative food frequency questionnaire. Relative risks (RRs) were calculated by Cox proportional hazards regression, with cumulative average intakes of vitamins B₆ and B₁₂ as main exposures, adjusting for potential confounders.

Main Outcome and Measure  Hip fracture.

Results  During follow-up, 2304 of 75 864 women had a hip fracture. Among the women with hip fractures, the median (range) age at hip fracture was 75.8 (46.7-93.0) years and the mean (SD) body mass index (calculated as weight in kilograms divided by height in meters squared) was 24.3 (4.6). Median (interquartile range) cumulative average intakes of total vitamins B₆ and B₁₂ were 3.6 (4.8) mg/d and 12.1 (11.7) μg/d, respectively. Both vitamin B₆ (RR, 1.29; 95% CI, 1.04-1.59 for an intake of ≥35 vs <2 mg/d; P = .06 for linear trend) and vitamin B₁₂ (RR, 1.25; 95% CI, 0.98-1.58 for an intake of ≥30 vs <5 μg/d; P = .02 for linear trend) were associated with increased fracture risk. Risk was highest in women with a combined high intake of both vitamins (B₆ ≥35 mg/d and B₁₂ ≥20 μg/d), exhibiting an almost 50% increased risk of hip fracture (RR, 1.47; 95% CI, 1.15-1.89) compared with women with a low intake of both vitamins (B₆ <2 mg/d and B₁₂ <10 μg/d).

Conclusions and Relevance  In this cohort study, a combined high intake of vitamins B₆ and B₁₂ was associated with an increased risk of hip fracture. The intakes were far higher than the recommended dietary allowances. These findings add to previous studies suggesting that vitamin supplements should be used cautiously because adverse effects can occur.

Vitamin supplementation is widely popular. In the National Health and Nutrition Examination Survey (NHANES), approximately 50% of US adults reported taking at least 1 dietary supplement.¹ The prevalence of use increased by age and was highest among individuals of non-Hispanic white race/ethnicity and among those with the highest education. Among US nurses, 28% reported using at least 4 dietary supplements.²

Both insufficient and excess intakes of a nutrient may be harmful. According to randomized clinical trials (RCTs), high-dose vitamin supplementation may lead to unexpected adverse effects.^3-6 Based on the concept that high homocysteine concentrations could cause disease, several large RCTs have been performed. However, they failed to demonstrate a preventive effect of vitamin B supplementation on cardiovascular diseases and cancer,^7-10 and possible adverse effects have been reported.^9,11,12 Although hyperhomocysteinemia also may negatively influence bone quality by disturbing collagen cross-linkage and stimulating bone resorption,¹³ high doses of vitamin B₁₂ and folic acid supplementation have not shown a fracture-preventing effect in RCTs.¹⁴ In a previous secondary analysis of combined data from 2 double-blind RCTs with factorial design,¹⁵ an unexpected increased risk of hip fracture was found among those treated with high doses of vitamin B₆. The highest fracture risk was seen in the trial arm combining vitamin B₆ and vitamin B₁₂.

It is unlikely that new large trials of vitamin B₆ and vitamin B₁₂ will be carried out in the future. We aimed to study if high intakes of vitamins B₆ and B₁₂ from food and supplements were associated with a risk of hip fracture in the Nurses’ Health Study (NHS) and to investigate whether combined high intakes of both vitamins were associated with a particularly increased fracture risk.

The NHS was initiated in 1976 among 121 701 female registered nurses in the United States aged 30 to 55 years. They responded to a mailed questionnaire concerning medical history, lifestyle, and disease risk factors. Every 2 years, follow-up questionnaires have been mailed to update individual characteristics and to identify incident diagnoses. Postmenopausal women entered the present analysis in 1984, when vitamin B supplement use was first assessed; otherwise, they entered the questionnaire cycle when they reached menopause. As shown in the flowchart in the eFigure in the Supplement, the final study sample included 75 864 postmenopausal women for our primary analysis. The women were followed up from June 1984 through May 2014. The dates of analysis were July 2016 to June 2018. The follow-up rate for this study population was 92% over the period of our analysis.

Completion and return of the self-administered questionnaires constituted informed consent. The investigation was approved by the Institutional Review Board at Brigham and Women’s Hospital, Boston, Massachusetts. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.

On every biennial follow-up questionnaire, information about hip fracture (with the date of occurrence and a description of the circumstances) was requested. As nurses, they were expected to be capable of reporting hip fractures, as demonstrated in a small validation study¹⁶ in which all 30 self-reports were confirmed by medical records. Hip fractures were also identified from death records. For the primary cumulative average analysis, hip fractures were recorded in 2304 of 75 864 women during follow-up after exclusion of 117 fractures due to cancer or major traumatic events.

Diet was assessed with a semiquantitative food frequency questionnaire (FFQ) in 1984, 1986, and every 4 years thereafter until 2010. Participants reported their habitual frequency of consumption over the previous year for specified serving sizes of more than 130 foods. Daily energy and nutrient intakes were calculated from the total diet.

The FFQ requested information on current use of supplements, such as vitamin B₆, folic acid, vitamin B complex, vitamin B₁₂–only supplements (beginning in 1998), multivitamins, vitamin A, vitamin D, and calcium. For vitamin B₆ and vitamin A, information on daily dose was requested. For multivitamins, the respondents were asked to report the brand name and number of pills per week. Information on frequency and dose for other supplements was not obtained.

Vitamin intakes (except those from supplements only) were adjusted for total energy intake using regression analysis.¹⁷ In a validation study¹⁸ among 632 women in the NHS and the younger NHS2 cohort, correlations between intakes from the FFQ and 7-day diet records were 0.68 for total vitamin B₆ and 0.68 for total vitamin B₁₂. A correlation of 0.52 between total vitamin B₆ in the food and plasma vitamin B₆ (pyridoxal 5′-phosphate) and a corresponding correlation of 0.25 for vitamin B₁₂ have been reported in another substudy within the NHS.¹⁹

On all biennial follow-up questionnaires, the following measures were assessed: weight; hours per week spent in recreational activities; smoking status; menopausal status and use of postmenopausal hormone therapy; diagnoses of cancer, diabetes, cardiovascular disease, and osteoporosis; and use of thiazide diuretics, furosemide-like diuretics, and oral corticosteroids. We calculated total metabolic energy expenditure (metabolic equivalent [MET] hours per week) from the reported recreational activities²⁰ and body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared) from current weight and height on the initial cohort questionnaire.

Questions concerning difficulties with balance, climbing a flight of stairs, or walking 1 block were included on questionnaires beginning in 1992, and questions on falls and pernicious anemia were included from 1998 onward. Self-rated general health status was requested in 1992, 1996, and 2000.

Participants were followed up to the date of first self-reported hip fracture or death from hip fracture, last questionnaire response, or the end of follow-up in 2014. In our primary analyses, we used vitamin B₆, vitamin B₁₂, and other nutrient intakes that were cumulatively averaged over follow-up. At the beginning of each new FFQ cycle, the intakes were updated with the mean of all assessments up to that time. Intakes were carried forward 1 cycle to replace missing data, and we excluded person-time in cycles in which women failed to report their dietary intake on the 2 most recent FFQs.

In alternate analyses, we used current intakes of vitamins and other nutrients (calculated as the mean of the 2 most recent FFQs) among 96 467 women with 2812 hip fractures (eFigure in the Supplement). Participants who had the 2 most recent dietary assessments missing for a cycle did not contribute person-time in that cycle.

Cross-sectional age-adjusted characteristics in the 2002 questionnaire cycle by categories of total vitamin B₆ and B₁₂ were recorded. We used Cox proportional hazards regression to compute hazard ratios (hereafter referred to as relative risks [RRs]) of hip fracture with 95% CIs according to predefined categories of vitamins B₆ and B₁₂. Cumulative average intakes of vitamins B₆ and B₁₂ were the main exposures. Separate analyses were performed for total vitamin intakes (from diet plus supplements) and intakes from supplements only. For vitamin B₆, the 5 categories of total intakes ranged from less than 2 mg/d to at least 35 mg/d, and the 5 categories of intakes from supplements only ranged from 0 mg/d (no supplement use) to at least 25 mg/d. For vitamin B₁₂, the 5 categories of total intakes ranged from less than 5 μg/d to at least 30 μg/d, and the 5 categories of intakes from supplements only ranged from 0 μg/d (no supplement use) to at least 25 μg/d. We also ran models with continuous intakes to test for linear trend. Based on the results from the previous RCT¹⁵ and from the categories used for vitamins B₆ and B₁₂, combined categories of vitamins B₆ and B₁₂ intakes were created.

All Cox proportional hazards regression models were stratified by age and questionnaire cycle to account for age and time. Relative risks were calculated from models using time-varying exposure and covariates (ie, person-time was assigned to the appropriate category for each variable at the beginning of every biennial follow-up questionnaire cycle). Multivariable RRs were computed from models that adjusted for potential dietary and nondietary confounding factors. For categorical covariates, missing data were assigned as a separate category. Less than 2% of the observations had missing data for BMI, physical activity, and smoking, and 5% of the observations had missing data for postmenopausal hormone therapy.

Multiplicative interactions between exposures were calculated using the Wald test for continuous data. The proportional hazards assumption was tested by including interaction terms between age and vitamin B₆ and vitamin B₁₂, respectively. In exploratory analyses, we also examined if the associations between vitamin B₆ and vitamin B₁₂ and risk of hip fracture differed by BMI and physical activity.

Data were analyzed using statistical software (SAS, version 9.4; SAS Institute Inc). Two-sided statistical tests were used, and P < .05 indicated statistical significance.

Among the 2304 fracture cases, the median age at hip fracture was 75.8 years (age range, 46.7-93.0 years). These women had a mean (SD) BMI of 24.3 (4.6), and median (IQR) cumulative average intakes of total vitamins B₆ and B₁₂ were 3.6 (4.8) mg/d and 12.1 (11.7) μg/d, respectively.

Height and BMI differed little across cumulative average intake categories of vitamin B₆ (Table 1) and B₁₂ (Table 2) in 2002, the approximate midpoint in follow-up, whereas physical activity increased and smoking prevalence decreased with higher intakes of both vitamins. Intakes of other micronutrients were also higher by increasing intakes of vitamins B₆ and B₁₂, whereas caffeine and alcohol consumption decreased. The reported prevalences of functional limitations, chronic diseases, and medication use tended to be lowest at low vitamin B₁₂ intakes but were similar at middle and high intakes. Although less apparent, a similar pattern was found for vitamin B₆. The Pearson product moment correlation coefficient between total intake of vitamins B₆ and B₁₂ was r = 0.51 (P < .001). Correlation coefficients between total vitamin B₆ intake and the other vitamins were 0.35 for folate (P < .001), 0.28 for vitamin D (P < .001), and 0.37 for retinol (P < .001). For total vitamin B₁₂ intake, the corresponding correlation coefficients were 0.43 for folate (P < .001), 0.33 for vitamin D (P < .001), and 0.36 for retinol (P < .001).

The mean follow-up time was 20.9 years (1 586 155 person-years of follow-up). Compared with the reference category of total vitamin B₆ less than 2 mg/d, an intake at least 35 mg/d was associated with an increased risk of hip fracture after adjusting for all covariates (RR, 1.29; 95% CI, 1.04-1.59; P = .06 for linear trend) (Table 3). For vitamin B₆ from supplements only, those consuming no vitamin B₆ supplements had the lowest risk compared with a similar increased risk in the other groups. For total vitamin B₁₂, intakes at least 30 μg/d were associated with a nonsignificant increased risk of hip fracture compared with intakes less than 5 μg/d (RR, 1.25; 95% CI, 0.98-1.58), and risk increased linearly with increasing intake (RR, 1.01; 95% CI, 1.00-1.03 per 10-μg/d increase in total intake; P for linear trend = .02). Similar results were found for vitamin B₁₂ intakes from supplements only (Table 3). The interaction term for the 2 vitamins on fracture risk was not significant. In analyses that included mutual adjustment for total B₆ and B₁₂ intakes, the associations were somewhat attenuated, with an RR of 1.19 (95% CI, 0.95-1.49) for vitamin B₆ at least 35 mg/d vs less than 2 mg/d and an RR of 1.22 (95% CI, 0.94-1.57) for vitamin B₁₂ at least 30 μg/d vs less than 5 μg/d. In fully adjusted models that included adjustment for intake from supplements, there was no clear association between vitamin B₆ from diet only and hip fracture (RR, 1.03; 95% CI, 0.91-1.16 per 1-mg/d increase in intake from food only; P for linear trend = .67) or between vitamin B₁₂ from diet only and hip fracture (RR, 1.01; 95% CI, 0.99-1.02 per 1-μg/d increase in intake from food only; P for linear trend = .54).

Women with a high intake of both vitamins had a significantly increased risk of hip fracture compared with the reference category of a low intake of both vitamins (RR, 1.47; 95% CI, 1.15-1.89) (Table 4). Among women in the medium-intake categories for both vitamins, risk was not significantly elevated (RR, 1.18; 95% CI, 0.98-1.42). Few women had low intakes of one vitamin and high intakes of the other.

There was a significant interaction between total vitamin B₆ intake and BMI on hip fracture risk, while the interaction between total vitamin B₁₂ intake and BMI was not significant. The results for total vitamins B₆ and B₁₂ intakes stratified by BMI are summarized in Table 5. The associations between vitamin B₆ and vitamin B₁₂ and hip fracture were stronger in women with BMI less than 25 compared with the overall analyses, whereas there were no clear associations in women with BMI at least 25. Among women with BMI less than 25, those with a high intake of both vitamins B₆ and B₁₂ had an RR of 1.71 (95% CI, 1.25-2.34) for hip fracture compared with those having a low intake of both vitamins. The corresponding RR in women with BMI at least 25 was 1.04 (95% CI, 0.66-1.64).

We observed no significant interactions between total vitamin B₆ or vitamin B₁₂ intakes and age, indicating that the proportional hazards assumption was met. Similarly, there was no significant interaction between total vitamin B₆ intake and physical activity for fracture risk. However, a significant interaction was found between total vitamin B₁₂ intake and physical activity. There was a significant linear trend with increasing intake among those with a physical activity level below the median (RR, 1.02; 95% CI, 1.00-1.03 per 10-μg/d increase in total intake; P = .02) but not among those with a higher physical activity level (RR, 1.01; 95% CI, 0.99-1.03 per 10-μg/d increase in total intake; P = .51). Stratified results are summarized in eTable 1 in the Supplement.

In additional analyses of the data in Table 3, the following had little influence on the estimates: adjustment for BMI 4 years before the last follow-up and later weight change, difficulty climbing a flight of stairs or walking 1 block, difficulty with balance, falls in the past year, self-reported general health status, and pernicious anemia (eTable 2 in the Supplement). Adjustment for total folate intake had only a small influence on the associations between vitamin B₆, vitamin B₁₂, and hip fracture (eTable 3 in the Supplement).

In sensitivity analyses censoring participants at age 80 years, the association for total vitamin B₆ intake was marginally stronger, whereas the results for total vitamin B₁₂ intake were similar to the findings of the main analyses. The association for combined intakes of vitamins B₆ and B₁₂ as analyzed in Table 4 was also strengthened, with an RR of 1.74 (95% CI, 1.30-2.32) for women with a high intake of both vitamins.

In analyses conducted using current intakes of total vitamins B₆ and B₁₂, a similar pattern was seen. However, it was somewhat weaker compared with the analysis using cumulative average intakes (eTable 4 and eTable 5 in the Supplement).

We found that high intakes of vitamins B₆ and B₁₂ were associated with increased hip fracture risk among postmenopausal women in the Nurses’ Health Study. The risk was highest in women with a combined high intake of both vitamins, exhibiting an almost 50% increased risk of hip fracture compared with women with a low intake of both vitamins. High intakes were due to use of supplements. These findings add to previous studies suggesting that vitamin supplements should be used cautiously because adverse effects can occur.

Our findings are in line with the results of a secondary analysis of combined data from 2 Norwegian double-blind RCTs with an identical intervention (6837 participants) in which an unexpected increased risk of hip fracture was found in those treated with high doses of vitamin B₆ during an extended follow-up.¹⁵ Both RCTs had a 2 × 2 factorial design: the 4 groups were given (1) folic acid (0.8 mg) plus vitamin B₁₂ (0.4 mg) and vitamin B₆ (40 mg), (2) folic acid (0.8 mg) plus vitamin B₁₂ (0.4 mg), (3) vitamin B₆ alone (40 mg), or (4) placebo. Risk of hip fracture was highest in the first group (hazard ratio, 1.49; 95% CI, 1.05-2.11 compared with placebo).

We are not aware of other RCTs studying fractures that have supplemented with vitamin B₆ alone, but several trials¹⁴ have combined vitamins B₆ and B₁₂. As summarized in eTable 6 in the Supplement, the results from the RCTs are divergent, as are the doses used and fracture type studied, making it difficult to draw firm conclusions. In the Women’s Antioxidant and Folic Acid Cardiovascular Study (WAFACS),²¹ there was no significant effect of the intervention (50 mg/d of vitamin B₆ and 1 mg/d of vitamin B₁₂ plus 2.5 mg/d of folic acid) on nonvertebral fractures (RR, 1.08; 95% CI, 0.88-1.34). A statistical interaction with baseline plasma B₆ concentration was observed; when excluding women in the lower quartile of baseline vitamin B₆, there was a significantly increased risk of nonvertebral fractures (RR, 1.39; 95% CI, 1.01-1.91). The Heart Outcomes Prevention Evaluation (HOPE) 2 trial²² intervened with 50 mg/d of vitamin B₆, 1 mg/d of vitamin B₁₂, and 2.5 mg/d of folic acid. There was a nonsignificant hazard ratio of 1.07 (95% CI, 0.85-1.33) for nonvertebral fractures. The Vitamins to Prevent Stroke (VITATOPS) trial²³ gave a lower dose of vitamin B₆ (25 mg/d) in addition to 0.5 mg/d of vitamin B₁₂ plus 2.0 mg/d of folic acid. The authors reported no significant effect on any osteoporotic fracture (n = 145) (hazard ratio, 0.86; 95% CI, 0.62-1.18) or hip fracture (n = 70) (hazard ratio, 0.94; 95% CI, 0.59-1.50).

A possible biological explanation for the findings in the present study is not clear. The magnitude of intakes of vitamins B₆ andB₁₂ associated with an increased risk of hip fracture in our study far exceeded the recommended dietary allowances (RDAs) (1.3-1.7 mg/d for vitamin B₆ and 2.4 μg/d for vitamin B₁₂).¹⁵

Possible adverse effects of high-dose vitamin B₆ supplementation have previously been suggested.^24,25 High doses (≥500 mg/d) might increase the risk of falling because neurological symptoms, including ataxia, neuropathy, and decreased muscle tone, have been reported, and milder neurological symptoms have been observed at doses of approximately 100 mg/d as adverse effects.²⁶ Preliminary work suggested that high vitamin B₆ concentrations might accelerate bone loss by counteracting the modulating influence of estrogens on steroid receptors.²⁷ A recent paradox theory proposes that large doses of pyridoxine, the inactive form of vitamin B₆ included in supplements and found in foods, inhibits the active form pyridoxal phosphate.²⁸

We do not have an explanation for the mechanism by which vitamin B₁₂ may contribute to increased fracture risk. However, as summarized in Table 4 and in eTable 5 in the Supplement, a high intake of vitamin B₁₂ and a low intake of vitamin B₆ were not associated with increased risk, which is in agreement with a meta-analysis¹⁴ of RCTs giving vitamin B₁₂ and/or folic acid alone (without vitamin B₆).

A possible explanation for the interaction between total vitamin B₆ intake and BMI is not clear. At low BMI, fracture risk was higher, and a larger number of incident hip fractures thus occurred, yielding higher statistical power. We speculate that the possible mechanism of excessive vitamin B₆ exposure increasing fall risk through neurological symptoms could particularly aggravate fracture risk in women with low BMI, who are more prone to fracturing their hip when experiencing a fall. However, adjustment for falls had little influence on the estimates, and the association between falls and fracture risk was not altered by adjustment for vitamins B₆ and B₁₂ intakes. It also could be that the possible interaction between vitamin B₆ and the steroid receptor might be most influential in lean women, who have a reduced capacity for production of adipose-derived estrogens.²⁹

Our study has strengths and limitations. We were able to follow up a large cohort of women with repeated detailed assessments of diet, supplement use, and other possible confounding factors. However, we cannot exclude the possibility that individuals started taking supplements due to ill health. In addition, all information on vitamin intake and confounders was collected by questionnaires, with their inherent limitations. Although residual confounding could be present, our results were not substantially influenced by adjustment for indicators of frailty or disease or a long list of other possible confounders. Another limitation was the self-report of hip fractures. However, in sensitivity analyses censoring participants at age 80 years (thus omitting the oldest, in whom underreporting of fracture could be an issue), the association for total vitamin B₆ intake was marginally stronger. An additional limitation is that the findings may be applicable only to women of white race/ethnicity.

The results for supplemental vitamin B₆ are puzzling because all categories above the reference category have similar increases in risk. Also, intakes of different supplements are correlated, making it challenging to disentangle specific associations. Nevertheless, our results were adjusted for intake of calcium, vitamin D, and retinol, and the differences between the model adjusted for age and questionnaire cycle and the fully adjusted models were modest.

We did not control for multiple hypothesis testing. However, our analyses were based on the RCT results, in which the highest fracture risk was found in those treated with high doses of both vitamins.¹⁵ A low proportion of the women were in the low-intake category of both vitamins. Yet, compared with the group with a medium intake of both vitamin B₆ (2 to <35 mg/d) and vitamin B₁₂ (10 to <20 μg/d), the risk was still significantly increased in those with a high intake of both vitamins (RR, 1.25; 95% CI, 1.03-1.51).

The RDAs are established to meet the nutritional requirements of almost the entire population. Despite that, use of high-dose vitamin supplementation far exceeding the RDAs is common, often without any definite indication and in the absence of clear evidence of benefit.

Our results are in line with several reports suggesting that unexpected adverse effects can occur with high-dose vitamin supplementation. For example, high-dose beta-carotene supplementation increased the risk of lung cancer in smokers,³ and high-dose vitamin E supplementation may increase all-cause mortality.⁴ Higher risk of fracture was reported in 2 RCTs^5,6 after treatment with annual megadoses of vitamin D, and possible adverse effects of homocysteine-lowering treatment with B vitamins have been observed,^9,11 including a potentially increased risk of cancer.¹² Although we acknowledge the limitations of our cohort design, the findings herein add to the body of literature that suggests caution should be used in vitamin supplementation when there is no apparent deficiency.

In this large prospective cohort study of postmenopausal women in the Nurses’ Health Study, we found that a combined high intake of vitamins B₆ and B₁₂ was associated with an increased risk of hip fracture. These findings add to previous studies suggesting that vitamin supplements should be used cautiously because adverse effects can occur.

Accepted for Publication: March 22, 2019.

Published: May 10, 2019. doi:10.1001/jamanetworkopen.2019.3591

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2019 Meyer HE et al. JAMA Network Open.

Corresponding Author: Haakon E. Meyer, MD, PhD, Department of Community Medicine and Global Health, University of Oslo, PO Box 1130 Blindern, 0318 Oslo, Norway (h.e.meyer@medisin.uio.no).

Author Contributions: Drs Meyer and Feskanich had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Meyer, Willett, Feskanich.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Meyer, Holvik.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Meyer, Willett, Feskanich.

Obtained funding: Willett.

Administrative, technical, or material support: Willett.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by grants CA186107 and AG030521 from the National Institutes of Health.

Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.