Association of Vitamin D or Calcium Supplementation with Cardiovascular Outcomes and Mortality: A Meta-Analysis with Trial Sequential Analysis

Abstract

Background

To exploring the role of vitamin D or calcium supplementation in reducing all-cause mortality and cardiovascular outcomes.

Methods

The search was restricted to systematic reviews or metaanalyses published from January 1, 2010, to July 7, 2019. An additional search was performed to identify recently published randomized controlled trials (from January 1, 2015, to July 7, 2019). Homogeneous results from different studies were pooled using Revman 5.3 software.

Results

Twenty-three studies involving 89,251 participants were ultimately included in this meta-analysis. No associations were observed between the supplementation and composite cardiovascular outcomes, consisting of all-cause mortality, cardiovascular mortality, myocardial infarction, and other MACEs.

Conclusions

Whether used alone or in combination, vitamin D and calcium supplementation do not exert meaningful effects on all-cause mortality, cardiovascular mortality, MACEs or MI among community-dwelling adults.

Introduction

Many cross-sectional and longitudinal studies have suggested that vitamin D deficiency is associated with an increased risk of cardiovascular disease (CVD), namely, myocardial infarction (MI), heart failure, and atrial fibrillation, and that it therefore plays an important role in regulating CVD (1, 2, 3, 4).A U-shaped relationship was also shown between 25 (OH) D3 levels and the cardiovascular risk in The Offspring Cohort of the Framingham Heart Study (5). In addition, preclinical studies have demonstrated that single nucleotide polymorphisms in the vitamin D receptor are potential risk factors for CVD (6). Vitamin D also regulates many genes involved in CVD, including those controlling cell proliferation and differentiation, apoptosis, oxidative stress, membrane transport, matrix homeostasis and cell adhesion (7). Furthermore, it downregulates pro-inflammatory cytokines, metalloproteinases, and natriuretic peptides and upregulates matrix Gla proteins, anti-inflammatory cytokines, and metalloproteinase inhibitors (8, 9).

With the new clinical guidelines and media attention, there is increased awareness that daily supplementation of vitamin D is necessary, especially considering that conventional vitamin D supplements are widely recommended by physicians and that recommendations for those supplements have increased in the US primary care setting (10, 11). Considering the link between cardiovascular disease and vitamin D deficiency, levels of 25(OH)D3 above 75 nmol/L may be beneficial for patients with risk factors or established CVD (12). Nonetheless, the evidence supporting the cardiovascular protective role of vitamin D hormones is almost entirely inferred from observational studies. To date, randomized trials have not consistently shown beneficial effects of vitamin D supplementation on subclinical or clinical cardiovascular outcomes (13).

Although the meta-analyses published to date have concluded that the association between vitamin D supplements and CVD risk is limited (14), recent large sample size trials have strengthened the evidence base (15, 16, 17). In addition, with the rapid development of an aging society, increasing attention has been paid to the primary prevention of cardiovascular health among community residents. Meanwhile, vitamin D supplementation is usually co-administered with calcium, as recommended by daily nutrition guidelines. Therefore, we performed a meta-analysis and trial sequential analysis (TSA) focused on the community-dwelling population to update the current recommendations with evidence of the benefits and harms of supplemental vitamin D and calcium, alone or in combination, with regard to cardiovascular health.

Materials and Methods

This study was registered at PROSPERO (No. CRD42019145593) and performed based on the preferred reporting items for systematic reviews and meta-analysis (PRISMA) criteria.

Type of studies

We included all prospective randomized control trials (RCTs) focusing on vitamin D, calcium, or combined vitamin D and calcium supplementation as a strategy for improving cardiovascular health. The follow-up duration lasted at least one year. Crossover trials, quasi-RCTs, animal experiments and other studies that were published repeatedly or that did not provide access to complete data were excluded.

Types of participants

The included studies enrolled community-dwelling adults older than 18 years. All participants were included in this review regardless of age, race and sex.

Types of interventions

Studies of interventions including vitamin D or calcium were included, regardless of duration or dosage. Trials with multiple interventions (e.g., coadministered calcium and vitamin D) were also eligible.

Types of comparisons

Eligible comparator groups were those receiving no treatment or a placebo.

Types of outcomes

The co-primary endpoints were all-cause mortality and CVD-related death. The secondary outcomes were a composite of major adverse cardiovascular events (MACEs) and MI. The longest available follow-up time was used for each trial in the analysis.

Information sources and search strategy

We searched the PubMed, Cochrane Library, and EMBASE databases from the inception dates to July 7, 2019, using the keywords “calcium”, “vitamin D”, and “cardiovascular” to identify published systematic reviews or meta-analyses evaluating the association between calcium, vitamin D, or combined calcium and vitamin D supplementation and CVDs. The search was restricted to systematic reviews or meta-analyses published in the last 10 years from January 1, 2010, to July 7, 2019, and excluded systematic reviews or meta-analyses that included only populations living in institutions. We identified the original RCTs included in the systematic reviews or meta-analyses. An additional search was performed to identify recently published RCTs (from January 1, 2015, to July 7, 2019) that met the inclusion criteria using the databases and keywords described above. There were no restrictions on the types of publications or the participants' characteristics, but only studies published in English were considered. Note Express 3.0 software was used for study management. The full list of the search strategy for PubMed is as follows:

• #1.(calcium[MeSH Terms]) OR calcium [Title/Abstract];
• #2.(((((vitamin d[MeSH Terms]) OR vitamin d[Title/Abstract]) OR ergocalciferols[MeSH Terms]) OR ergocalciferols[Title/Abstract]) OR cholecalciferol[MeSH Terms]) OR cholecalciferol[Title/Abstract];
• #3.((((((cardiovascular[MeSH Terms]) OR cardiovascular[Title/Abstract]) OR cardiac[MeSH Terms]) OR cardiac [Title/Abstract]) OR myocardial [MeSH Terms]) OR myocardial [Title/Abstract]) OR heart[MeSH Terms]) OR heart [Title/Abstract];
• #4.(#1) OR #2;
• #5.(#4) AND #3 Filters: Clinical Trial; Humans.

A modified search strategy was used for other electronic databases.

Study selection

Two independent reviewers scanned the retrieved records and determined whether they should be further assessed based on three domains: the title, abstract and keywords. If the information met the inclusion criteria, full articles were retrieved for further assessment. We retrieved the full content of each article if there was any doubt about these criteria from the information provided in the title and abstract. Authors were contacted to obtain relevant missing data if necessary and when resources allowed. Any disagreement was resolved by a third party (Y.W.).

Data extraction and study quality assessment

Data were extracted independently by two reviewers. The following items were extracted from the individual studies: general information (e.g., the name of first author, reference, and year of publication), characteristics of the population, sample size, details of intervention, length of follow-up, implementation country, outcomes specified above and any other outcomes assessed. The methodological quality of the RCTs was assessed independently per the Cochrane Handbook for Systematic Review of Interventions in seven domains, namely, sequence generation, allocation concealment, blinding of the participants and personnel, blinding of the outcome assessment, incomplete outcome data, selective outcome reporting and other bias. The quality was divided into three levels: ‘low risk', ‘high risk', or ‘unclear risk'. The evaluation of methodological quality was performed independently by the two reviewers, and discrepancies were solved through discussion until a consensus was reached.

Data analysis

RevMan 5.3 software provided by the Cochrane Collaboration was used for the meta-analytic calculations. The Q test was used to estimate the total percentage of variation in each study that was due to heterogeneity rather than chance, and the 12 statistic was used for quantification of the heterogeneity. The fixed effects model was chosen when the 12 statistic was less than 50% and the P value was more than 0.1, and the treatment effects were calculated with a random effects model when the P value was less than 0.1. The random effects model was also used when a subgroup analysis was performed and when the heterogeneity among the studies was significant.

The pooled relative risk (RR) with the 95% confidence interval (CI) was used as the effect measure. If the number of included studies was less than two or if heterogeneity was apparent, the result of our systematic review was narratively reported. If included studies had multiple arms, we identified the relevant intervention and control groups and combined the relevant groups into a single group before synthesizing the data.

Subgroup analysis and sensitivity analysis

To address heterogeneity and perform the secondary analysis, a subgroup analysis was necessary. The analysis focused on types of intervention, gender and implementation country. To explore the potential sources of heterogeneity with regard to methodology, statistics and clinical characteristics, sensitivity analyses were performed. When the results of the clinical trials varied widely and when heterogeneity tests showed significant differences, we removed one trial that was significantly different from the other trials with regard to the clinical factors, methodological factors and other factors, and then, we pooled the remaining studies to compare the results before and after.

Trial sequential analysis and evidence level grading

TSA was applied to address the risk of random error associated with sparse data and/or multiple testing, which can affect a cumulative meta-analysis (18). TSA is performed to reduce or minimize the random error and is equivalent to an interim analysis in a randomized clinical trial. All TSAs were performed at the level of an overall 5% risk of a two-sided type I error and a power of 80%. We initially anticipated an intervention effect of a 10% RR reduction. TSA software (version 0.9.5.10 Beta, available from http://www.ctu.dk/tsa/) was used for these analyses.

We used the five GRADE considerations (19, 20), namely, study limitations, consistency of effect, imprecision, indirectness, and publication bias, to assess the quality of a body of evidence from studies that contributed data to the meta-analyses and generate narrative summaries for the pre-specified outcomes. Two reviewers (ZYL and LJJ) independently employed the GRADE system to interpret the findings. Any discrepancies in judgements were resolved through discussion with a third reviewer (WY).

Results

Search results and study characteristics

Three hundred thirty-three potentially eligible systematic reviews or meta-analyses were initially identified. The titles and abstracts of these records were screened for inclusion. Full texts of 90 records were read, and 15 met the inclusion criteria (Fig 1). Then, 62 RCTs from 15 included systematic reviews or meta-analyses were screened by reviewing the full texts, and 15 trials were included. The searches for recently published RCTs yielded 562 records, and 55 full texts of these records were reviewed. Of these RCTs, 8 trials met the inclusion criteria. Hence, 23 RCTs involving 89,251 participants were ultimately included in this meta-analysis (Fig 1).

Figure 1.

PRISMA flow diagram

In total, 44,757 of the participants received vitamin D, calcium, or combined vitamin D and calcium supplements, and 44,494 of the participants received a placebo. Follow-up durations were varied among the included trials (range, 1–7 years). Thirteen trials used vitamin D alone compared with a placebo, and the dosage of vitamin D3 ranged from 100 IU/d to 4,800 IU/d. Calcium alone was compared with a placebo in the other six trials. Seven trials compared the combined supplement (vitamin D plus calcium) with a placebo. The included trials were conducted in different countries, including the United States, the Netherlands, Denmark, Finland, France, the United Kingdom, Australia, New Zealand and Thailand. A detailed description of the characteristics of the 23 included trials is provided in Table 1.

Table 1.

Characteristics of the included trials

ID	Population	Sample size		Treatment		Control	Study Follow-up, y	Country
		EG	CG	Vitamin D Type and Dosage	Calcium Type and Dose
Malluche et al., 1988 (21)	Women with postmenopausal osteoporosis between the ages of 50 and 80 years	12	15	Vitamin D3,400 IU/d	NA	placebo	2	United States
Ott et al., 1989 (22])	All women were postmenopausal, between the ages of 50 and 80 years	43	43	Vitamin D3, 0.5 µg/d	NA	placebo	2	United States
Chapuy et al., 1992 (23)	Healthy ambulatory women (mean age 84±6 years)	1,634	1,636	Vitamin D3 800 IU/d	1,200 mg/d elemental calcium in the form of tricalcium phosphate	placebo	1.5	France
Ooms et al., 1995 (24)	Elderly people aged 70 yrs and older	177	171	Vitamin D3,400 IU/d	NA	placebo	2	The Netherlands
Lips et al., 1996 (25)	Adults (≥70yrs) recruited from general practitioners or from apartment houses or homes for elderly persons	1,291	1,287	Vitamin D3,400 IU/d	NA	placebo	Median,3.5	The Netherlands
Bæksgaard et al., 1998 (26])	Caucasian background, age of 58–67 years, good general health and postmenopausal status defined as the cessation of menstrual bleeding for at least 6 months.	65	63	Vitamin D3, 14 µg (560 IU)/d	1,000 mg/elemental calcium in the form of calcium carbonate	placebo	2	Denmark
Komulainen et al., 1999 (27)	Women in early postmenopause who were non-osteoporotic and aged 47- to 56 years	112	115	Vitamin D3, 300 IU/d (the Vitamin D3 dosage was lowered to 100 IU/day after 4 years of treatment because of adverse lipid changes noticed during the first years of the trial)	93 mg/d elemental calcium in the form of calcium lactate	placebo	5	Finland
Chapuy et al., 2002 (28)	639 patients (mean age 85 years, range 64–99 years) living in 55 apartment houses for elderly people, 610 of whom were randomized in the study	194	190	Vitamin D3, 800 IU/d	1,200 mg/delemental calcium in the form of tricalcium phosphate	placebo	2	France
Trivedi et al., 2003 (29)	Participants aged 65–85 years living in the general community	1,345	1,341	Vitamin D3, 100,000 IU every 4 months for 5 years	NA	placebo	5	United Kingdom
Prince et al., 2006 (30)	1460 women recruited from the population and older than 70 years (mean age, 75 years)	730	730	NA	1,200 mg/calcium carbonate tablet	placebo	5	Australia
Reid et al., 2006 (31)	Subjects were aged more than 55 years, were not receiving therapy for osteoporosis or taking calcium supplements, and were free of major ongoing disease	732	739	NA	1,000 mg/d calcium in the form of calcium citrate	placebo	5	New Zealand
Hsia et al., 2007 (32)	Postmenopausal women 50 to 79 years of age	18,176	18,106	Vitamin D 400 IU/d	1,000 mg/d elemental calcium in the form of calcium carbonate	placebo	7	United States
Lyons et al., 2007 (33)	All residents were invited to participate in the study	1,670	1,673	2.5 mg vitamin D2 tablets 3 times a year for 3 years	NA	placebo	3	United Kingdom
Reid et al., 2008 (34)	Men aged at least 40 years in good general health	107	108	NA	Two doses: 600 mg/d elemental calcium and 1,200 mg/d elemental calcium	placebo	2	New Zealand
Zhuet al., 2008 (35)	Community-dwelling women aged 70–80 years who were ambulatory	39	81	Vitamin D2, 1,000 IU/d	1200mg elemental calcium in the form of calcium carbonate	placebo	5	Australia
Prince et al., 2008 (36)	Women aged 70–90 years who were ambulatory	151	151	Vitamin D2, 1,000 IU/d	1,000 mg/d elemental calcium in the form of calcium citrate	placebo	1	Australia
Sanders et al., 2010 (37)	Community-dwelling women, aged 70 years or older	1,131	1,125	Vitamin D3, 500,000 IU each year for 3 to 5 years	NA	placebo	1	Australia
Chailurkit et al., 2010(38)	Healthy postmenopausal women without osteoporosis	175	161	NA	500 mg/d elemental calcium in the form of calcium carbonate	placebo	2	Thailand
Gallagher et al., 2012(39)	Healthy, white, postmenopausal women aged 57 to 90 years who were at least 7 years postmenopausal	142	21	Vitamin D3 (400, 800, 1,600, 2,400, 3,200, 4,000, and 4,800 IU/d)	NA	placebo	1	United States
Witham et al., 2013 (40)	Patients 70 years and older with isolated systolic hypertension and baseline 25-hydroxy vitamin D levels less than 30ng/mL	80	79	Vitamin D3, 100,000IU every 3 months for 1 year	NA	placebo	1	United Kingdom
Baron et al., 2015 (41)	Participants aged 45–75 years who had ≥1 colorectal adenoma removed within 120 d before enrollment and no remaining polyps after a complete colonoscopy	1,130	1,129	Vitamin D3, 1,000 IU/d	1,200 mg/d elemental calcium in the form of calcium carbonate	placebo	3	United States
Scragg et al., 2017 (42)	Adults invited from family practices and 163 from community groups	2,558	2,552	Vitamin D3, in an initial dose of 200,000 IU, followed a month later by monthly doses of 100,000 IU for a median of 3.3years	NA	placebo	3	New Zealand
Manson et al., 2018 (43)	Participants had no history of cancer (except nonmelanoma skin cancer) or cardiovascular disease at trial entry	12,927	12,944	Vitamin D3, 2,000 IU/d	NA	placebo	Median, 5.3 (3.8 to 6.1)	United States

Abbreviation: NA, not available

Methodological quality

Fourteen trials that used a random number table or other methods, and these studies mentioned the method of allocation concealment. Ten study protocols were available via various registration platforms. With the exception of five studies, the implementation of blinding in the remaining studies was satisfactory. Furthermore, almost all trials reported the drop-out or withdrawal of patients. We also considered some included trials with an ‘unclear' rating for the other sources of bias, such as the sample size (Fig 2).

Figure 2.

Risk of bias graph

The effects of therapy

The study was divided into three subgroups based on the choice of different interventions: vitamin D versus placebo (subgroup 1), calcium versus placebo (subgroup 2) and vitamin D plus calcium versus placebo (subgroup 3). Among these subgroups, we separately assessed the primary and secondary outcomes.

Primary outcomes

In the three subgroups, no significant differences were observed in all-cause mortality (subgroup 1: 44,239 patients, RR=0.98 [95% CI, 0.93–1.04], P = 0.45,12 = 0%; subgroup 2: 3,651 patients, RR=0.95 [95% CI, 0.68 – 1.32], P = 0.66,12 = 0%; subgroup 3: 37,525 patients, RR=0.92 [95% CI, 0.76–1.11], P = 0.51, 12 = 0%) between the experimental group and the control group (Fig 3).

Figure 3.

Meta-analysis of all-cause mortality

Cardiovascular mortality was only assessed in subgroup 1, and vitamin D supplements were not significantly association with cardiovascular mortality (33,785 patients, RR=1.00 [95% CI, 0.85–1.18], P = 0.43,12 = 0%) (Fig 4).

Figure 4.

Meta-analysis of cardiovascular mortality

Furthermore, the results of subgroup analyses performed after stratification by factors such as gender and the country were not statistically significant (Table 2).

Table 2.

The results of all outcomes for subgroups stratified by gender and implementation country

				Primary outcomes
		All-cause mortality				Cardiovascular mortality
Subgroup	Number of studies	Participants	Risk ratio (95% CI)	I2(%)	Number of studies	Participants	Risk ratio (95% CI)	I2(%)
Gender
Female	11	41,686	0.92 [0.76, 1.11]	0	2	120	0.33 [0.01, 7.96]	0
Male	1	323	0.99 [0.09, 10.80]	NA	0
Female and Male	10	3,713	0.98 [0.91, 1.05]	26	3	33,665	1.00 [0.85, 1.18]	13
Location
America	6	63,969	1.06 [0.95, 1.19]	0	3	25,991	1.09 [0.87, 1.38]	0
Europe	12	17,486	0.93 [0.85, 1.02]	0	2	7,794	0.90 [0.71, 1.14]	0
Australia and Thailand	4	4,233	0.80 [0.58, 1.10]	0	0
				Secondary outcomes
			MACE				MI
Subgroup	Number of studies	Participants	Risk ratio (95% CI)	I2(%)	Number of studies	Participants	Risk ratio (95% CI)	F(%)
Gender
Female	6	4,537	1.19 [0.85, 1.66]	0	7	3,066	1.21 [0.48, 3.04]	0
Male	1	323	3.96 [0.50, 31.28]		1	323	3.52 [0.18, 68.86]
Female and Male	3	33,665	0.97 [0.89, 1.06]	0	5	36,083	0.95 [0.83, 1.10]	0
Location
America	3	26,068	0.97 [0.85, 1.11]	0	5	28,291	0.96 [0.79, 1.18]	0
Europe	4	8,344	0.98 [0.90, 1.07]	7	5	8,503	0.96 [0.82, 1.13]	0
Australia and Thailand	3	4,133	1.18 [0.84, 1.66]	0	3	2,678	1.31 [0.44, 3.92]	0

Abbreviation: NA, not applicable

We conducted sensitivity analyses by sequentially removing each study, but did not observe a significant change compare to the original result, indicating that results of the meta-analysis of outcomes were robust according to the sensitivity analyses. In addition, we performed a TSA for two primary outcomes. In the TSA of all-cause mortality, the cumulative Z curves did not cross either conventional boundaries nor did they reach the optimal size (Fig 5). This finding must be explained or emphasized based on the number of studies published. For cardiovascular mortality, the boundary required for the information size was ignored due to the availability of too little information (incidence of events), and it did not cross the conventional boundaries.

Figure 5.

Trial sequential analysis of all-cause mortality

The blue Z-curve measures the treatment effect for the cumulative meta-analyses. Horizontal green dotted lines represent a Z score of +1.96 and −1.96, indicating a conventional significant P value = 0.05. The area inside the red lines on the far right indicates the futility region. The optimal size indicates the calculated optimum sample size for statistical inference, and N indicates the number of participants in the meta-analysis. The cumulative Z-curve did not cross any of the boundaries, and with only approximately 30% of the optimal size being reached suggests the true effect, although it is unlikely to be clinically important and is not conclusively known

Secondary outcomes

Only subgroup 1 and subgroup 2 reported MI and MACEs. By pooling studies with a low level of heterogeneity, neither vitamin D supplementation nor calcium supplementation was significantly associated with the two outcomes in these two subgroups (Table 3). No factor accounted for the interstudy heterogeneity (P>0.05), consistent with the result of subgroup analyses (Table 2).

Table 3.

The results of all outcomes for the three subgroups

Subgroups	Outcomes	Number of studies	Participants	Risk ratio [95% CIJ	P	P for heterogeneity	F(%)
Vitamin D versus Placebo	MACE	7	36,345	0.98 [0.91, 1.05]	0.56	0.91	0
	MI	12	39,149	0.96 [0.85, 1.09]	0.55	0.99	0
Calcium versus Placebo	MACE	3	2,180	1.21 [0.85, 1.72]	0.30	0.31	15
	MI	1	323	3.52 [0.18, 68.86]	0.41	NA	NA

Abbreviation: NA, not applicable

Overall quality of evidence according to the GRADE system

We graded the overall quality of available evidence with the GRADEpro Guideline Development Tool (GDT) at https://gradepro.org/. The quality of evidence for all outcomes was graded as high, which was based on the rigorous evaluation of the five items for “Decrease quality of evidence” (risk of bias, inconsistency, indirectness, imprecision and publication bias) and three items for “Increased quality of evidence” (large effect, plausible confounding would change the effect and dose-response effect).

Discussion

In this secondary analysis of long-term (≥12 months) trials enrolling 89,251 community-dwelling participants taking vitamin D, calcium or combined vitamin D and calcium supplements, no associations of supplementation with all-cause mortality and composite cardiovascular outcomes, consisting of cardiovascular mortality, MI, and other MACEs, were observed. Although this updated meta-analysis specifically focused on community-dwelling participants and included both vitamin D and calcium supplements, it reached similar conclusions as those presented in the most recent meta-analysis (44), supporting and extending earlier findings (45, 46). Moreover, evidence was obtained in this updated meta-analysis to draw conclusions regarding the effect of calcium supplementation alone on cardiovascular health; interventions with a calcium supplement alone were not included in the previous review.

This review included evidence on 4 outcomes: all-cause mortality, cardiovascular mortality, MACEs and MI. The strength of evidence for the three types of intervention and all outcomes was graded as high according to the GRADE system. Additionally, the TSA confirmed that the evidence that vitamin D supplementation did not result in meaningful clinical benefits for the primary outcomes (all-cause mortality and cardiovascular mortality) was reliable. Currently, the cumulative Z line has not yet reached the optimal size and has not crossed the conventional boundaries. Thus, further trials with large samples might be needed to support or, likely, to alter the conclusions. From another perspective, if the results from future large-scale trials are significantly different from the data from the current trials, the combined results will substantially increase the heterogeneity of the trial results, which will in turn reduce the weight of the new large-scale trials in the combined analysis. Therefore, an increase in the number of positive results from large RCTs may only exert a small effect on the current outcomes and are unlikely to alter the conclusions of the existing meta-analyses.

In addition, this meta-analysis differs from previous studies by focusing on community residents, particularly the elderly population. The widespread fascination with vitamin D as a panacea for most illnesses, including CVD, is responsible for nearly a 100-fold increase in vitamin D testing and oral supplementation over the last decade (47). The popularity of vitamin D or calcium supplements is at least partly due to the misunderstanding of the impressive epidemiological associations between supplements use and a wide range of health indicators (48]). At the same time, due to the strong advocacy in commercial advertisements, middle-aged and elderly residents in the community often take vitamin D or calcium as a “health care product” daily. We also expect that the secondary analysis based on evidence-based medicine with a focus on community residents might cautiously provide evidence that vitamin D or calcium supplementation is not effective as a primary prevention strategy for cardiovascular health in community residents.

The strengths of the current analyses are that they are comprehensive, include all available data from previous systematic reviews and the most recent original studies, and concomitantly assess the hard endpoints of cardiovascular health. The analyses are based on substantially more trials, more participants, and a more representative population than previous analyses, including that they have greater power, the effect estimates have greater precision, and the TSAs are able to examine efficacy at lower RR thresholds. The TSAs are important because they provide estimates of the reliability of current evidence and the likelihood of future trials to change the current conclusions. At the same time, we performed quality evaluations of the evidence though the GDT for each outcome in this study, indicating that the bias and potential for error in the current conclusions are further reduced. This meta-analysis also provides valuable information for the development of clinical practice guidelines in the future. In addition, the protocol of this study was registered in PROSPERO, which may increase the transparency and quality of this meta-analysis.

Our findings also should be interpreted in the context of several limitations. Only three databases were accessed, and studies related to our topic that were published in English were considered. All relevant RCTs may not have been included. Additionally, some of the included trials did not include prespecified mortality or cardiovascular outcomes as the primary endpoint, which are designed to evaluate effects of supplement intake on bone density, fracture or other incidence of diseases. Therefore, inaccuracy may exist. In addition, several meta-analyses displayed moderate heterogeneity in the trial results, generally because a few small-to-moderate-sized studies reported positive results that were not observed in larger trials. The heterogeneity of the sample size, study designs, and follow up duration was also an issue that may explain the differences between the findings.

Contributors

Authors' contributions: Funding acquisition, Wei Wang and Yong Wang; Methodology, Yili Zhang and Xu Wei; Project administration, Wei Wang and Yong Wang; Resources, Yili Zhang and Nannan Tan; Software, Yuan Li; Visualization, Yili Zhang and Yuan Li; Writing - original draft, Yili Zhang, Yuan Li, Junjie Liu and Xu Wei; Writing - review & editing, Nannan Tan, Jian Zhang, Huihui Zhao and Yong Wang.

Funding

This work was funded by Beijing «Double First-rate» Personnel Department-High-level-Scientific Research funding Project for Wangyong's team [No. 1000041510165 (18–19)],the Nation Natural Science Foundation of China (No.81822049, 81673712), Fok Ying Tung Education Foundation (No. 151044), Beijing Nova program (Z171100001117028) and the National Key R&D Program of China (No. 2017YFC1700100, 2017YFC1700102). The funders had no role in the study design, data collection and analysis, the decision to publish, or the preparation of the manuscript.

Conflict of Interest

The authors declare that they have no conflict of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.