Mei Chung, A. Tang, Zhuxuan Fu
Dec 20, 2016
Citations
3
Influential Citations
79
Citations
Quality indicators
Journal
Annals of Internal Medicine
Abstract
Calcium is a nutrient essential for maintaining bone health. A small proportion of total body calcium (less than 1%) also regulates vascular contraction and vasodilation, muscle function, nerve transmission, intracellular signaling, and hormonal secretion. Vitamin D promotes calcium absorption in the gut and maintains adequate serum calcium and phosphate concentrations, enabling normal bone mineralization and preventing hypocalcemic tetany (1). Although adequate calcium and vitamin D intake is critical for maintaining bone health, the role of calcium and vitamin D supplementation in older adults is unclear. Some systematic reviews showed that combined calcium and vitamin D supplementation reduced the risk for fractures in older adults (2, 3), whereas more recent systematic reviews reported inconsistent effects for fractures across randomized, controlled trials (4, 5). Experts have raised concerns about a potential effect of a high intake of calcium (with or without vitamin D) from foods and supplements on cardiovascular disease (CVD) outcomes (68). A meta-analysis of both study- and patient-level data from randomized trials showed that calcium with or without vitamin D supplementation increased the risk for myocardial infarction (pooled relative risk, 1.24 [95% CI, 1.07 to 1.45]) and stroke (pooled relative risk, 1.15 [CI, 1.00 to 1.32]) (9, 10). However, a more recent meta-analysis showed that calcium with or without vitamin D supplementation had no statistically significant effects on coronary heart disease events (pooled relative risk, 1.02 [CI, 0.96 to 1.09]) or mortality (pooled relative risk, 1.04 [CI, 0.88 to 1.21]) (11). Many researchers have questioned the strength of the body of evidence linking supplemental calcium intake with CVD risk, noting that cardiovascular outcomes have not been the primary end point of any trial investigating calcium or calcium and vitamin D supplementation to date (12, 13). To inform a joint position statement from the National Osteoporosis Foundation (NOF) and American Society for Preventive Cardiology, NOF commissioned a focused update and reanalysis of 2 broader evidence reports examining the effects of calcium and vitamin D on a wide range of clinical and intermediate outcomes (5, 14). This update addresses the effects of calcium intake (from dietary or supplemental sources), alone or in combination with vitamin D, on CVD risk in generally healthy adults. Methods This systematic review implemented the same methodology as the 2009 evidence report examining the effects of calcium and vitamin D (alone or in combination) on 17 health outcomes across all life stages that was produced to inform the Institute of Medicine committee charged with updating the dietary reference intake values for calcium and vitamin D (14). In 2014, the Agency for Healthcare Research and Quality commissioned an update of the 2009 evidence report focusing on studies of vitamin D alone or in combination with calcium (5). The effects of calcium intake (from foods or supplements) alone on CVD were not updated in the 2014 evidence report. Methodological details for the reviews were described in a protocol (15). Data Sources and Searches MEDLINE, the Cochrane Central Register of Controlled Trials, and Scopus (including EMBASE) were searched from 2009 to July 2016 for prospective cohort or nested casecontrol (or casecohort) studies reporting an association between calcium intake (dietary or supplemental) and risk for incident CVD (cardiac, cerebrovascular, or peripheral vascular events and new hypertension), and for randomized, controlled trials on the effect of increasing calcium intake (by food or supplements) on the same outcomes. Analyses of combinations of calcium and micronutrients other than vitamin D that could not isolate the independent effects of calcium with or without vitamin D were not included. Studies or analyses that did not quantify the amount of calcium in the interventions or exposures also were excluded. The literature search strategy was adapted from the 2009 evidence report (14) but focused on calcium exposures and CVD outcomes. Unpublished data were not sought. Study Selection Two reviewers performed abstract and full-text screening to identify peer-reviewed, English-language studies of generally healthy adults in which no more than 20% of participants had known CVD. Studies involving participants with hypertension or elderly populations (>60 years of age) were included, whereas those restricted to pregnant women, persons with diabetes, or those receiving dialysis were excluded. Reference lists of relevant systematic reviews were cross-checked with lists of included studies to ensure that no relevant studies were missed. All cardiovascular event or mortality outcomes (defined by the original authors) were included. Data Extraction and Risk-of-Bias (Quality) Assessment All extracted data in the 2009 and 2014 evidence reports (5, 14) are accessible to the public on PubMed and PubMed Health. Relevant data in the 2 evidence reports were extracted from their evidence tables (Appendix C of the evidence reports) and are included in this update. Data from studies published after the 2 evidence reports were extracted by 1 reviewer and confirmed by at least 1 other using the same data extraction form. The risk of bias in randomized, controlled trials and that of observational studies was assessed separately, with the same assessment tools used in the 2009 and 2014 evidence reports (15). However, to be consistent with the current methodology recommended in the Cochrane Handbook for Systematic Reviews of Interventions, we did not assign an overall quality grade for each study in this update (16). Two reviewers did the risk-of-bias assessments independently; disagreements were discussed until consensus was reached. Data Synthesis We synthesized trials and cohort studies separately but based our conclusions on the total body of evidence. We did not perform a meta-analysis of trial data, because trials reported outcomes with heterogeneous definitions. For cohort studies, we charted doseresponse curves by using adjusted results and did doseresponse metaregressions if 4 or more studies reported analyses of similar exposureoutcome relationships. If more than 1 analysis model was reported in a study, we focused on the model that adjusted for the most potential confounders. Many cohort studies had several analyses reporting different calcium exposures or cardiovascular outcomes of interest. We planned our doseresponse metaregressions carefully to ensure that study populations did not overlap in each analysis. We performed linear and nonlinear doseresponse metaregressions to examine the associations between calcium intake levels and the risk for CVD by using a 2-stage hierarchical regression model, implemented in the dosresmeta R package (17, 18). The method, first formalized by Greenland and Longnecker (19), uses estimates of the covariance matrix to account for the within-study correlations across dose levels and incorporates them into the estimation of the linear trend by using generalized least-squares regression. In addition, we applied a method developed by Hamling and colleagues (20) that allowed reconstruction of a table of cell counts (effective counts) from reported adjusted risk estimates and CIs. We used this method to facilitate doseresponse metaregressions and recalculate risk estimates comparing calcium dose categories greater than 1000 mg/d with those less than 1000 mg/d, the recommended dietary allowance for healthy adults (1). See the Appendix for details of these procedures. Analyses were conducted by using SAS, version 9.3 (SAS Institute), and R, version 3.2.5 (R Foundation for Statistical Computing). All P values were 2-tailed, and a P value less than 0.05 was considered statistically significant. Role of the Funding Source This research was supported by an unrestricted educational grant from the NOF through Pfizer Consumer Healthcare. The authors were blind to the corporate funder until the final manuscript was submitted to the NOF. The funder reviewed the evidence synthesis for drafting the position statement but had no role in study selection, quality assessment, data analysis, or writing the manuscript. Results Search Results We included 4 randomized, controlled trials (in 10 publications [10, 2129]), 1 nested casecontrol study (30), and 26 cohort studies (29, 3155). One publication contained data from a randomized trial and a cohort study (29). Appendix Figure 1 shows the summary of literature searches and study selection flow for this update. Appendix Figure 1. Summary of evidence searches and study selection flow. Cardiovascular death includes death from ischemic heart disease, myocardial infarction, coronary heart disease, and any cardiovascular death. RCTs = randomized, controlled trials. * Total of 4 unique RCTs in 10 publications. Randomized, Controlled Trials Two trials (reported in 8 publications) examined the effects of calcium plus vitamin D supplementation (10, 2126, 29), whereas 3 looked at the effects of calcium supplementation alone (21, 27, 28). Of these 5 trials, 1 (RECORD [Randomised Evaluation of Calcium or Vitamin D]) was a 22 factorial design of calcium and vitamin D that contributed to both comparisons (calcium vs. placebo, calcium plus vitamin D vs. placebo) (21). Cardiovascular disease outcomes were secondary end points in all trials (Appendix Table 1). The overall risk of bias of the trials was low, although concerns were raised regarding poor adherence to the interventions in all trials (Appendix Table 2). None of the trials reported levels of total calcium intake from dietary and supplemental sources. Appendix Table 1. Characteristics of Randomized, Controlled Trials Examining the Effects of Calcium With or Without Vitamin D Supplementation on CVD Outcomes Appendix Table 2. Risk-of-Bias Assessment, Background Calcium Intake Levels, and Adherence in t