| |
Vitamin D and Cardiovascular Outcomes: A Systematic Review and Meta-Analysis
|
| |
| |
Download the PDF here
Published online before print June 15, 2011
The Journal of Clinical Endocrinology & Metabolism
Mohamed B. Elamin, Nisrin O. Abu Elnour, Khalid B. Elamin, Mitra M. Fatourechi, Aziz A. Alkatib, Jaime P. Almandoz, Hau Liu, Melanie A. Lane, Rebecca J. Mullan, Ahmad Hazem, Patricia J. Erwin, Donald D. Hensrud, Mohammad Hassan Murad, and Victor M. Montori
Knowledge and Evaluation Research Unit (M.B.E., N.O.A.E., M.M.F., A.A.A., M.A.L., R.J.M., A.H., P.J.E., M.H.M., V.M.M.), Mayo Clinic, Rochester, Minnesota 55905; Department of Medicine (K.B.E.), Case Western Reserve University, Metrohealth Medical Center, Cleveland, Ohio 44109; Division of Endocrinology, Diabetes, Metabolism, Nutrition (J.P.A., V.M.M.), Mayo Clinic, Rochester, Minnesota 55905; Division of Endocrinology and Metabolism (H.L.), Santa Clara Valley Medical Center, San Jose, California 95128; and Division of Preventive, Occupational, and Aerospace Medicine (A.H., D.D.H., M.H.M.), Mayo Clinic, Rochester, Minnesota 55905
Abstract
Context: Several studies found association between vitamin D levels and hypertension, coronary artery calcification, and heart disease.
Objective: The aim of this study was to summarize the evidence on the effect of vitamin D on cardiovascular outcomes.
Design and Methods: We searched electronic databases from inception through August 2010 for randomized trials. Reviewers working in duplicate and independently extracted study characteristics, quality, and the outcomes of interest. Random-effects meta-analysis was used to pool the relative risks (RR) and the weighted mean differences across trials.
Results: We found 51 eligible trials with moderate quality. Vitamin D was associated with nonsignificant effects on the patient-important outcomes of death [RR, 0.96; 95% confidence interval (CI), 0.93, 1.00; P = 0.08], myocardial infarction (RR, 1.02; 95% CI, 0.93, 1.13; P = 0.64), and stroke (RR, 1.05; 95% CI, 0.88, 1.25; P = 0.59). These analyses were associated with minimal heterogeneity. There were no significant changes in the surrogate outcomes of lipid fractions, glucose, or diastolic or systolic blood pressure. The latter analyses were associated with significant heterogeneity, and the pooled estimates were trivial in absolute terms.
Conclusions: Trial data available to date are unable to demonstrate a statistically significant reduction in mortality and cardiovascular risk associated with vitamin D. The quality of the available evidence is low to moderate at best.
Ecological evidence suggests an association between vi- tamin D status and cardiovascular disease (prevalent and incident heart disease, stroke, and risk factors for heart disease such as hypertension and coronary artery calcification) (1-6). Several mechanisms have been pro- posed including endothelial dysfunction, vascular compli- ance, inflammation, and effects relating to PTH, renin- angiotensin system, and others (1, 7-12). When this evidence is summarized by meta-analyses of observational studies, an inverse association between 25-hydroxyvita- min D and cardiovascular risk is suggested (13, 14).
Despite this observational and epidemiological evidence, it is unclear whether vitamin D in interventional studies would affect cardiovascular risk. The Endocrine Society assembled a task force of experts to develop clinical practice guidelines regarding the supplementation of vitamin D. To assist in formulating these guidelines, we conducted a systematic review of the literature to quan- titatively and qualitatively summarize the available evi- dence regarding the possible cardiovascular harms and benefits of vitamin D.
Results
The initial search of the literature yielded 5584 citations, of which 51 eligible studies were selected with a good inter-reviewer agreement (k=0.80) (Supplemental Fig. 1, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org). Contact of all authors was attempted to verify data collected from their publication as well as to request additional or missing information. If an author did not respond to our initial request, a second request was attempted. We successfully contacted around 75% of the primary or secondary au- thors. Table 1 summarizes the characteristics of the in- cluded studies. Most studies recruited elderly women and coadministered calcium with vitamin D, and none used sun exposure or interventions other than vitamin D sup- plementation. Most studies effectively concealed the ran- dom allocation and blinded caregivers and patients. The methodological quality of included studies is summarized in Supplemental Table 1.
Meta-analyses
Mortality
Most of the included studies reported on mortality (n 30). Pooling across studies showed a nonsignificant and potentially trivial reduction in mortality that was consis- tent across studies (RR, 0.96; 95% CI, 0.93, 1.00; P = 0.08; I2 =0%) (Fig. 1).
MI, stroke, and peripheral vascular disease
Six studies reported the outcome of MI, and six re- ported on stroke. Meta-analyses showed no significant effect of vitamin D on MI (RR, 1.02; 95% CI, 0.93, 1.13; P = 0.64; I2 = 0%) or stroke (RR, 1.05; 95% CI, 0.88, 1.25; P =0.59; I2 =15%) (Fig. 1). Five studies reported the outcome of peripheral vascular disease but with no events in either study arm.
Serum lipids, blood pressure, and blood glucose
Table 2 shows pooled data for the effect of vitamin D on blood lipids, blood glucose, and blood pressure mea- surements. Vitamin D did not significantly affect any of the cardiovascular risk factors. However, the direction of vitamin D effect was consistent with reduction of all parameters measured except an increase in high-density lipoprotein cholesterol. Results were inconsistent across studies, and the pooled estimates were trivial in absolute terms.
Subgroup and sensitivity analyses
The planned subgroup analyses did not show any sig- nificant subgroup-effect interactions (Table 3). This also includes the subgroup of studies reporting on vitamin D supplementation in vitamin D-deficient patients, in which we found no significant decrease in mortality, MI, or stroke (P > 0.05 for all outcomes).
The use of a fixed-effect model instead of a random- effects model did not change study conclusions about any outcome. Excluding a study reporting death of a patient irrelevant to the intervention (21) did not change the over- all pooled mortality data (RR, 0.96; 95% CI, 0.93, 1.00; P = 0.07; I2 = 0%). Excluding a study in which the effects of vitamin D were possibly confounded by other interventions (22) did not change the overall mortality estimate (RR, 0.97; 95% CI, 0.93, 1.01; P = 0.67; I2 = 0%). Ex- cluding a study, in which mortality was possibly con- founded by the population's comorbidity (23) did not change the overall results (RR, 0.96; 95% CI, 0.93, 1.00; P = 0.84; I2 = 0%). We also excluded a study in which the intervention was a one-time high dose of vitamin D given im (24), which also did not affect the overall conclusion about mortality (RR, 0.96; 95% CI, 0.93, 1.00; P = 0.66; I2 = 0%). The same was true when we excluded a study that was heavily weighted in the analysis and used low vitamin D doses with poor ad- herence (25) (RR for mortality, 0.97; 95% CI, 0.93, 1.02; P = 0.69; I2 = 0%; RR for MI, 0.99; 95% CI, 0.86, 1.15; P = 0.64, I2 = 0%; RR for stroke, 1.16; 95% CI, 0.92, 1.46; P = 0.41; I2 = 0%). In general, analyses conducted in adherent patients (compliers) provided consistent trend of reduction in mortality, although results were nonsignificant.
We also conducted a sensitivity analysis in which we only included studies that increased vitamin D baseline levels as opposed to studies that administered low doses of vitamin D, which did not actually raise the blood level regardless of baseline. The pooled estimates were: mor- tality (16 studies)ÑRR, 0.95; 95% CI, 0.87, 1.03; P = 0.22; I2 = 6%; MI (three studies)ÑRR 0.96; 95% CI, 0.81, 1.13; P = 0.51; I2 = 0%; and stroke (three studies)Ñ RR, 1.04; 95% CI, 0.80, 1.34; P = 0.99; I2 = 0%. When we only analyzed the studies that reported vitamin D repletion in vitamin D-deficient patients, we still did not see any statistically significant results (RR, 1.06; 95% CI, 0.81, 1.41; P = 0.66; I2 = 3%, in six studies for mortality; RR, 1.26; 95% CI, 0.20, 7.98; P = 0.81; I2 = 23%, in two studies for MI; and RR, 1.01; 95% CI, 0.25, 3.97; P = 0.99; I2 = 0%, in two studies for stroke). The latter CIs are clearly wide, suggesting underpowered analyses.
Discussion
We conducted a systematic review and meta-analysis to summarize the best available research evidence regarding the effect of vitamin D on patient-important cardiovascular events and other cardiovascular risk factors. Previous systematic reviews of observational studies found significant associations between low vitamin D levels and the risk of cardiovascular disease (of variable definitions across the studies) and overall mortality (13, 14). Our analysis of randomized trials in which vitamin D was given as an intervention, as opposed to a blood level, did not demonstrate a significant effect on death, stroke, MI, lipid fractions (except a trivial increase in high-density lipoprotein), blood pressure, and blood glucose values. Our estimate for the mortality outcome, although nonsignificant, is in the same direction (i.e. reduction in risk) of that re- ported in another systematic review by Grandi et al. (13).
The limitations of this review stem from the fact that many of the included studies were not designed to evaluate cardiovascular outcomes; therefore, if the ascertainment of these endpoints was systematically different between the intervention and the control groups, which might have occurred in the 18% of the studies that were unblinded, results could be biased. Publication and reporting biases cannot be ruled out in any systematic review, although we attempted to contact study authors to reduce the effect of these biases. Lastly, the heterogeneity in some of the analyses makes the overall evidence to be of low to moderate quality. There remains the possibility that potential car- diovascular benefit of vitamin D remains undetected due to confounding baseline cardiac risk factors that randomization failed to correct or due to the coadministration of calcium that may have a detrimental cardiovascular effect (26). It is also important to note that randomized trials are likely to enroll participants without severe vitamin D deficiency who are less likely to benefit from vitamin D, which would drive the results toward the null. The strengths of this review relate to the comprehensive liter- ature search and the bias protection measures undertaken during the conduct of the systematic review (i.e. selecting studies and evaluating outcomes and quality by blinded independent pairs of reviewers).
The effect of vitamin D on all-cause mortality remains unclear. Our analysis did not find an association, whereas a previous meta-analysis (27) found that vitamin D was associated with decreased all-cause mortality (RR, 0.93; 95% CI, 0.87-0.99). Our meta-analysis includes more trials (51 vs. 18). Nevertheless, it is obvious that the choice of which trials to include in the meta-analysis is affecting the inference; hence, inference regarding mortality is not robust to the inclusion of evidence. It is plausible that vitamin D affects certain disease-specific mortalities such as cancer mortality (28, 29), for example; and when data are aggregated, the noise-to-signal ratio hides such effect. It is also plausible that the current data, when restricted to studies with adequate protection of bias, sufficient follow-up, and documented increase in vitamin D level, become underpowered to detect benefits in cardiovascular outcomes. The answer to the mortality question will likely require a very large trial with long follow-up in which disease-specific mortality is measured and ascertained as a primary endpoint. Trials with factorial design similar to the Randomized Evaluation of Calcium or Vitamin D (RECORD) trial (30) in which patients can be randomized to differing doses of vitamin D with and without calcium will be needed to determine the optimal dose and the nonskeletal effects of these interventions.
The practice implications of this systematic review indicate that recommending vitamin D to patients to reduce cardiovascular risk is not consistent with the current evidence. Individuals will require the age- and sex-appropriate daily intake of vitamin D and may require additional supplementation for other indications such as bone health, but not for cardiovascular risk reduction. The accompanying guideline document developed by the task force of the Endocrine Society will provide additional practical advice and detailed recommendations regarding vitamin D supplementation (75).
Conclusion
Trial data available to date are unable to demonstrate a statistically significant reduction in mortality and cardiovascular risk associated with vitamin D. The quality of the available evidence is low to moderate at best.
| |
| |
| |
|
|
|
