心血管系イベントの予防を目的としたホモシステイン低下介入
アブストラクト
背景
冠動脈疾患、脳卒中およびうっ血性心不全をはじめとする心血管系疾患は、世界中で死因の第一位を占める。ホモシステインは、メチオニンの代謝で生物学的に機能するアミノ酸である。心血管系イベントと相関が認められる血中総ホモシステインの高値がリスク因子であると考えられている。患者の心血管系イベントに対するビタミンB6、B9、B12サプリメントのホモシステイン低下介入 の効果を検討する。本レビューは2009年と2013年発表のレビューの更新である。
目的
既往歴として心血管系イベントを有する患者および既往心血管系イベントの無い患者において、ホモシステイン低下介入が全死因死亡や心血管系イベントの予防に効果的であるかどうか確定し、その安全性を評価すること。
検索方法
コクラン・ライブラリのCochrane Central Register of Controlled Trials (CENTRAL、2014年第1号)、 MEDLINE(1950年~2014年1月第5週)、EMBASE(1980年~2014年第6週)およびLILACS(1986年~2014年2月)を検索した。さらに、Web of Science(1970年~2014年2月7日)を検索した。採用した論文の参考文献リストをハンドサーチした。当該分野の研究者に問い合わせも行った。検索では言語に制限を設けなかった。
選択基準
心血管系イベントの予防を目的に、ホモシステイン低下療法による介入の効果を評価したランダム化比較試験(RCT)のうち、フォローアップが1年以上のものを採用した。心筋梗塞および脳卒中を主要アウトカムとした。末期腎不全患者を対象にした研究は除外した。
データ収集と分析
研究の選択、バイアスのリスク評価およびデータの抽出を2名で独立して実施した。二値アウトカムにはリスク比(RR)を推定した。I2統計量を用いて統計学的異質性を測定した。 ランダム効果モデルを用いた。
主な結果
本コクラン・レビューの2回目の更新では、新規ランダム化比較試験の同定はなかった。したがって本新版に、総参加者数47,429例のランダム化比較試験12件を選択する。全体的に、75%(9/12)の試験はバイアスが低リスクであった。ホモシステイン低下介入はプラセボと比較して非致死的または致死的な心筋梗塞(1,743/23,590例 (7.38%) 対 1,247/20,190例 (6.17%); RR 1.02、 95% 信頼区間 (CI) 0.95 ~1.10、I2= 0%、質の高いエビデンス)、脳卒中(968/22,348例 (4.33%) 対974/18,957例(5.13%); RR 0.91、 95% CI 0.82~1.0、I2 = 11%、質の高いエビデンス)、全死因死亡(2,784/22,648例 (12.29%) 対2,502/19,250例(10.64%); RR 1.01、 95% CI 0.96 ~1.07、 I2= 6%、質の高いエビデンス)に有意に影響しなかった。 ホモシステイン低下介入はプラセボと比較して重篤な有害事象(癌)に重大 な影響を及ぼさなかった(1558/18,130例(8.59%) 対1334/14,739例(9.05%); RR 1.06、 95% CI 0.98 ~1.13; I2 = 0%、質の高いエビデンス)。
著者の結論
今回のコクラン・レビューの2回目の更新では、心血管系イベントを予防するには、ビタミンB6、B9またはB12のサプリメントを単独または併用で投与するホモシステイン低下療法による介入を実施すべきであるとするエビデンスは見つからなかった。さらに、ホモシステイン低下療法による介入が癌のリスクの増大と関係があることを示唆するエビデンスは認められなかった。
PICOs
一般語訳
心血管系イベントの予防を目的としたホモシステイン低下介入(ビタミンB複合体治療)
レビューの論点
心血管系イベントの予防を目的としたホモシステイン低下介入を評価した。
背景
心血管系疾患は、世界中で死因の第一位を占める。心血管系疾患の最もよくみられる原因で罹病や死亡に至るのが、虚血性心疾患、脳卒中およびうっ血性心不全である。心血管系疾患のある人の多くが無症状で、心筋梗塞、狭心症、脳卒中(虚血性、出血性またはその両方)の発症リスクが高い。心血管系疾患の「新興のリスク因子」が、確立されたリスク因子(糖尿病、高血圧、能動喫煙者、血中脂質の数値が不良)に最近加わった。このリスク因子のひとつが、血中総ホモシステインの高値である。ホモシステインはアミノ酸で、その血中濃度は、ビタミンB群のシアノコバラミン(B12)、葉酸(B9)およびピリドキシン(B6)の血中濃度に影響を受ける。高血漿総ホモシステイン値は動脈硬化性疾患(動脈内に プラークが蓄積する)のリスク増加と関係がある。
研究の特性
今回の2回目の更新では、12試験を採用した。試験参加者は47,429例で、食料による栄養の強化が義務付けられている国の人もいれば、義務付けられていない国の人もいた。これらの研究は、ビタミンB複合体(シアノコバラミン(B12)、葉酸(B9)、ピリドキシン(B6))の異なるレジメンをコントロールまたはその他と比較したものである。研究は2002年~2010年に発表されたものである。
主な結果
ビタミンB6、B9またはB12のサプリメントを単独または併用で摂取するホモシステイン低下療法による介入は、いかなる用量でも、プラセボや標準的なケアに比べて、心血管系疾患のリスクがある参加者や診断を受けた参加者の心筋梗塞や脳卒中を予防したり、総死亡を抑制したりするというエビデンスは見つからなかった。ホモシステイン低下療法による介入は、プラセボに比較して重篤な有害事象(癌)に有意な影響を及ぼさなかった。
エビデンスの質
選択した試験の質は高く、非常に多くの参加者において実施されたので、レビューの結果は信憑性の高いものである。
Authors' conclusions
Summary of findings
Homocysteine‐lowering interventions (folic acid, vitamin B6 and vitamin B12) compared with placebo or standard care for preventing cardiovascular events | ||||||
Patient or population: Adults at risk of or with established cardiovascular disease | ||||||
Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
Assumed risk | Corresponding risk | |||||
Placebo or standard care | Homocysteine‐lowering interventions (folic acid, vitamin B6 and vitamin B12) | |||||
Non‐fatal or fatal myocardial infarction | Study population | RR 1.02 | 43,290 | ⊕⊕⊕⊕ | ||
62 per 1000 | 64 per 1000 | |||||
Stroke | Study population | RR 0.91 | 40,815 | ⊕⊕⊕⊕ | ||
52 per 1000 | 47 per 1000 | |||||
Death from any cause | Study population | RR 1.01 | 41,898 | ⊕⊕⊕⊕ | ||
130 per 1000 | 131 per 1000 | |||||
Cancer | Study population | RR 1.06 | 32,869 | ⊕⊕⊕⊕ | ||
91 per 1000 | 96 per 1000 | |||||
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
GRADE Working Group grades of evidence | ||||||
1I² = 0%. |
Background
Description of the condition
The burden of cardiovascular disease
Cardiovascular disease is the number one cause of death worldwide (Jamison 2006; WHO 2002). The term cardiovascular disease covers a wide array of disorders, including diseases of the cardiac muscle and of the vascular system supplying the heart, brain and other vital organs (Gaziano 2006). The most common causes of cardiovascular disease‐related morbidity and mortality are ischaemic heart disease, stroke and congestive heart failure (Gaziano 2006).
The burden of cardiovascular disease is significant and ischaemic heart disease is the single largest cause of death in developed countries. Moreover, it is one of the main contributors to death in developing countries (Appendix 1; Gaziano 2006). This knowledge is useful when developing strategies for reducing morbidity, mortality and costs (Kahn 2008; Math 2007).
The major risk factors for cardiovascular diseases include tobacco use, high blood pressure, high blood glucose, lipid abnormalities, obesity and physical inactivity (Epstein 1996; Gaziano 2006; Narayan 2006; Rodgers 2006; WHO 2002; Willet 2006). However, there are other risk factors for cardiovascular diseases called "emergent or new risk factors" (NACB 2009).
Homocysteine as a risk factor for cardiovascular disease
Homocysteine is an amino acid not used in protein synthesis (Blom 2011). Its role is to serve as an intermediate in methionine metabolism (Blom 2011; Hackam 2003; Humphrey 2008; NACB 2009; Selhub 2006), and several observational studies have shown that a raised blood homocysteine level is a risk factor for cardiovascular events (Casas 2005; Danesh 1998; Eikelboom 1999; Ford 2002; Guthikonda 2006; HSC 2002; Jacobsen 2005; Kardesoglu 2011; Refsum 1998; Splaver 2004; Stampfer 1992; Wald 2002; Wang 2005; Williams 2010; Wu 2013). The public significance of raised circulating blood homocysteine levels has been considered (Shelhub 2008). The risk of developing cardiovascular events could be explained by endothelial dysfunction between homocysteine and hydrogen sulfide (Pushpakumar 2014), or by an integration of the roles of homocysteine and folic acid in cardiovascular pathobiology, named methoxistasis (Joseph 2013).
In 1962, it was hypothesised that increased levels of total homocysteine may cause vascular disease: the homocysteine theory of arteriosclerosis (McCully 2005). The pathways through which total homocysteine levels may cause damage to endothelial cells and lead to atherosclerosis have been widely described (Ferretti 2006; Jacobsen 2006; Jakubowski 2000; Jakubowski 2004; Jakubowski 2008; Obeid 2009; Riksen 2005; Zhou 2009).
Circulating total homocysteine levels are composed of protein (albumin)‐homocysteine mixed disulfide, sulfhydryl form and low molecular weight disulfides (Mudd 2000). The normal levels of total homocysteine are close to 10 µmol/L (Mudd 2000). Hyperhomocysteinaemia is defined as the presence of an abnormally elevated concentration of plasma or serum total homocysteine levels (Mudd 2000). However, there is some controversy about the definition of the degree of hyperhomocysteinaemia. Fasting total homocysteine level concentrations between 12 and 30 µmol/L are termed mild or moderate, while intermediate hyperhomocysteinaemia includes levels between 31 to 100 µmol/L, and severe hyperhomocysteinaemia reflects values above 100 µmol/L (Maron 2006; Maron 2009). In the general population, the prevalence of hyperhomocysteinaemia is between 5% and 10% (Refsum 1998). However, rates may be as high as 30% to 40% in the elderly population (Selhub 1993). According to the results of population‐based studies, up to 10% of events due to coronary artery diseases may be attributable to elevated circulating total homocysteine levels (Boushey 1995).
Description of the intervention
B‐complex vitamins, cyanocobalamin (B12) (Fedosov 2012; Herrmann 2012; Kräutler 2012), folic acid (B9) (Crider 2011; Molloy 2012; Ohrvik 2011; Yetley 2011), and pyridoxine (B6) (di Salvo 2011; di Salvo 2012; Friso 2012; Mukherjee 2011), are given as a supplement. Figure 1 shows the role of B‐complex vitamins in homocysteine metabolism (Brustolin 2010).
How the intervention might work
The B‐complex vitamins are required for the transformation or excretion, or for both steps, in the total homocysteine level metabolism pathway (Castro 2006; Fowler 2005; Perła‐Kaján 2007; Ramakrishnan 2006). Supplementation with B‐complex vitamins reduces total homocysteine levels (Clarke 2007; HLTC 2005). There is some ambiguity regarding the function of pyridoxine (vitamin B6). Vitamin B6 supplementation has been shown to lower total homocysteine levels after methionine load, which occurs in an experimental situation and it is, as a result, believed to be a weak determinant of circulating total homocysteine levels. However, at least two studies have shown the contrary (Gori 2007; Sofi 2008). See Figure 1 for details.
Why it is important to do this review
This is the second update of this Cochrane Review and has been performed to identify the latest evidence, if available.
This review aims to address the following question. What are the benefits and harms of homocysteine‐lowering interventions compared with placebo or low‐dose vitamins B6, B9 and B12 for preventing cardiovascular events?
Objectives
To determine whether homocysteine‐lowering interventions, in those patients with and without pre‐existing cardiovascular disease:
-
are effective for preventing cardiovascular events or all‐cause mortality;
-
are safe;
-
differ in efficacy or safety.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials with a follow‐up period of one year or longer.
Types of participants
Adults (over 18 years) at risk of or with established cardiovascular disease. We excluded studies in patients with end‐stage renal disease.
Types of interventions
The interventions considered were vitamins B6, B9 or B12 given alone or in combination, at any dosage and via any administration route.
We made comparisons with placebo, or with differing regimens of vitamins B6, B9 or B12. When the included population was at risk of cardiovascular disease, we considered combinations of homocysteine‐lowering interventions with standard treatment (such as antihypertensives and statins) as long as the same standard treatment was given to the control group.
Types of outcome measures
Primary outcomes
-
Non‐fatal or fatal myocardial infarction.
-
Non‐fatal or fatal stroke (ischaemic or haemorrhagic stroke).
Secondary outcomes
-
First unstable angina pectoris episode requiring hospitalisation.
-
Hospitalisation for heart failure.
-
Death from any cause.
-
Serious or non‐serious adverse events.
We defined serious adverse events according to the International Conference on Harmonisation (ICH) Guidelines (ICH‐GCP 1997), as any event that leads to death, is life‐threatening, requires hospitalisation or prolongation of existing hospitalisation and/or results in persistent or significant disability. We considered all other adverse events non‐serious.
Search methods for identification of studies
Electronic searches
We reran the searches previously run in 2008 (Appendix 2) and 2012 (Appendix 3) on 12 February 2014 (Appendix 4).
We updated the searches of the Cochrane Central Register of Controlled Trials (CENTRAL 2014, Issue 1), MEDLINE OVID (1950 to January week 5 2014), EMBASE OVID (1980 to 2014 week 6) and Web of Science (Thomson Reuters, 1970 to 7 February 2014). The search of LILACS was last run on 2 February 2012. In the previous version (Marti‐Carvajal 2009), we searched Allied and Complementary Medicine ‐ AMED (accessed through Ovid) and the Cochrane Stroke Group Specialised Register.
We used the Cochrane sensitive‐maximising RCT filters to search MEDLINE and EMBASE (Lefebvre 2011).
We imposed no language restrictions.
Searching other resources
We also checked the reference lists of all trials identified.
We also searched the World Health Organization International Clinical Trials Platform search portal (http://apps.who.int./trialsearch).
We contacted authors and researchers to obtain further details for published studies.
Data collection and analysis
We conducted data collection and analysis of data according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Selection of studies
Three authors (AMC, IS and DS) independently screened the results of the search strategy for potentially relevant trials and independently assessed them for inclusion based on the inclusion criteria.
Data extraction and management
Two authors (AMC and DS) carried out data extraction using a pre‐designed data extraction form that included publication details, patient population, randomisation, allocation concealment, details of blinding measures, description of interventions and results. We resolved discrepancies through discussion. We involved a third author (DEK) to check the data entered into the Review Manager software.
Assessment of risk of bias in included studies
All review authors independently assessed the risk of bias of the trials according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
We assessed the following domains, using the following definitions.
Generation of the allocation sequence
-
Low risk of bias, if the allocation sequence was generated by a computer or random number table, drawing of lots, tossing of a coin, shuffling of cards or throwing dice.
-
Unclear, if the trial was described as randomised but the method used for the allocation sequence generation was not described.
-
High risk of bias, if a system involving dates, names or admittance numbers was used for the allocation of patients. These studies are known as quasi‐randomised and we excluded them from the review when assessing beneficial effects.
Allocation concealment
-
Low risk of bias, if the allocation of patients involved a central independent unit, on‐site locked computer, identical‐appearing numbered drug bottles or containers prepared by an independent pharmacist or investigator, or sealed envelopes.
-
Unclear, if the trial was described as randomised but the method used to conceal the allocation was not described.
-
High risk of bias, if the allocation sequence was known to the investigators who assigned participants or if the study was quasi‐randomised. We excluded the latter from the review when assessing beneficial effects.
Blinding (or masking)
We assessed each trial (as low, unclear or high risk) with regard to the following levels of blinding.
-
Blinding of clinician (person delivering treatment) to treatment allocation.
-
Blinding of participant to treatment allocation.
-
Blinding of outcome assessor to treatment allocation.
Incomplete outcome data
-
Low risk of bias, if the numbers and reasons for dropouts and withdrawals in all intervention groups were described or it was specified that there were no dropouts or withdrawals.
-
Unclear, if the report gave the impression that there had been no dropouts or withdrawals but this was not specifically stated.
-
High risk of bias, if the number or reasons for dropouts and withdrawals were not described.
We further examined the percentage of dropouts overall in each trial and per randomisation arm and we evaluated whether intention‐to‐treat analysis was performed or could be performed from the published information.
Selective outcome reporting
-
Low risk of bias, if pre‐defined or clinically relevant and reasonably expected outcomes were reported on.
-
Unclear, if not all pre‐defined or clinically relevant and reasonably expected outcomes were reported on or were not reported on fully, or it was unclear whether data on these outcomes were recorded or not.
-
High risk of bias, if one or more clinically relevant and reasonably expected outcomes were not reported on; data on these outcomes were likely to have been recorded.
Other bias
-
Low risk of bias, the trial appeared to be free of other components that could put it at risk of bias.
-
Unclear, the trial may or may not be free of other components that could put it at risk of bias.
-
High risk of bias, there were other factors in the trial that could put it at risk of bias.
We considered low risk of bias trials to be those that adequately generated their allocation sequence, had adequate allocation concealment, adequate blinding, adequate handling of incomplete outcome data, were free of selective outcome reporting and were free of other bias.
We considered trials in which we could assess one of domains as high risk of bias or unclear risk of bias as trials with high risk of bias.
Two authors (AMC and IS) assessed the included studies and entered the information into tables; see Characteristics of included studies.
Measures of treatment effect
We pooled the risk ratios (RR) with 95% confidence interval (CI) for the following binary outcomes: non‐fatal or fatal myocardial infarction, non‐fatal or fatal stroke (ischaemic or haemorrhagic), first unstable angina pectoris episode requiring hospitalisation, hospitalisation for heart failure, mortality due to any cause and serious or non‐serious adverse events as recommended by Higgins 2011.
Dealing with missing data
For all included trials, we noted the levels of attrition. We contacted the first author of the paper if data were missing. We extracted data on the number of participants by allocated treatment group, irrespective of compliance and whether or not the participant was later thought to be ineligible or otherwise excluded from treatment or follow‐up. If we were not able to do so, we recorded for each study whether the results pertained to an intention‐to‐treat analysis or to available‐case analysis.
Assessment of heterogeneity
We quantified statistical heterogeneity using the I2 statistic, which describes the percentage of total variation across studies that is due to heterogeneity rather than sampling error (Higgins 2003). We considered statistical heterogeneity to be present if the I2 value was greater than 50% (Higgins 2011). When significant heterogeneity was detected (I2 > 50%), we attempted to identify the possible causes.
Assessment of reporting biases
We assessed publication bias for myocardial infarction, stroke and death from any cause using the Comprehensive Meta‐Analysis software (CMA 2005).
Data synthesis
We pooled the results from the trials using the Review Manager software (RevMan 2014). We summarised the findings using a random‐effects model.
Trial sequential analysis
Meta‐analysis of cumulative data may run the risk of random errors ('play of chance') due to sparse data and repetitive analyses of the same data (Brok 2008; Brok 2009; Thorlund 2010; Thorlund 2011; Wetterslev 2008; Wetterslev 2009). In order to assess the risks of random errors in our cumulative meta‐analyses, we conducted diversity‐adjusted trial sequential analyses based upon the proportion with the outcome in the control group, an a priori set relative risk reduction of 20%, an alpha of 5%, a beta of 20% and the diversity in the meta‐analysis (CTU 2011; Thorlund 2009; Thorlund 2011). We conducted sensitivity analysis of the trial sequential analysis to estimate the potential need for further trials.
Subgroup analysis and investigation of heterogeneity
We performed subgroup analysis according to the type of intervention.
Sensitivity analysis
We conducted a sensitivity analysis comparing the results using all studies and using only those with low risk of bias.
'Summary of findings' tables
We used the GRADE proposals to assess the quality of the body of evidence associated with the following outcomes: myocardial infarction, stroke, death from any cause and cancer (Guyatt 2011). We constructed summary of findings Table for the main comparison (SoF) using the GRADEpro software (GRADEpro 2008). GRADE classifies the quality of a body of evidence based on the extent to which one can be confident that an estimate of effect or association reflects the outcome being assessed (Balshem 2011; Brozek 2011; Guyatt 2011a; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2011g; Guyatt 2012; Guyatt 2012a).
Results
Description of studies
The initial search in 2008 identified 4318 references. From these we excluded 4299 references after examining the title and abstract because they were not relevant. We obtained full reprints of the remaining 19 references for more detailed examination. This led to the identification of four further publications from their reference list. We included eight randomised controlled trials, reported in 11 articles. We subsequently included three identified ongoing studies (three references) in the 2012 update. We excluded nine studies (nine references).
The search in 2012 identified 1705 records, which resulted in 1393 unique references after duplicates were removed. After examining the titles and abstracts we excluded 1334 references. We obtained full reprints of the remaining 59 references for more detailed examination. Four randomised clinical trials, published in 21 references, met the inclusion criteria. We also identified 11 additional references to previously included studies. We excluded 25 new studies (27 references).
The search in February 2014 identified 888 records, which resulted in 653 unique references after duplicates were removed. After examining the titles and abstracts we excluded 636 references. We obtained full reprints of the remaining 17 references for more detailed examination. Ultimately, we were not able to find any new randomised clinical trials.
In total, this updated review includes 12 randomised clinical trials, published between 2002 and 2010, involving 47,429 participants. See Figure 2 for details.
![Study flow diagram for this update](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG02.png)
Study flow diagram for this update
These trials are described in Characteristics of included studies. The length of follow‐up ranged from one to 7.3 years. The trials varied in size, characteristics of participant populations, duration, drug dosage and experimental design.
Included studies
Eleven trials were conducted in patients with prior cardiovascular heart disease such as coronary artery disease, myocardial infarction, stable angina, unstable angina, stroke and intermittent claudication (BVAIT 2009; CHAOS 2002; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008); and a further trial explicitly included patients with history of non‐disabling cerebral infarction (VISP 2004).
Eleven trials included patients with at least one of the following known cardiovascular risk factors: diabetes mellitus, hypertension elevated total cholesterol, current smoking and low high density lipoprotein (HDL) cholesterol (BVAIT 2009; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). This aspect was unclear for CHAOS 2002. WAFACS 2008 included patients with three or more coronary risk factors. One trial explicitly excluded patients with previously known hyperhomocysteinaemia (total plasma homocysteine > 18 μmol/L) (FOLARDA 2004).
BVAIT 2009 included patients with hyperhomocysteinaemia without diabetes and cardiovascular disease. HOPE‐2 2006 included patients without a history of coronary heart disease (CHD). WAFACS 2008 only included female patients.
Nine trials included more than 1000 patients (CHAOS 2002; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008).
Nine trials were compared with placebo (BVAIT 2009; CHAOS 2002; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008), and two with standard care (FOLARDA 2004; GOES 2003), while one trial was a randomised controlled trial (VISP 2004), which compared doses of homocysteine‐lowering interventions.
The intervention assessed by most of the trials was a combination of vitamins B6, B9 and B12 (nine trials; BVAIT 2009; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). Three trials only included vitamin B9 as intervention (CHAOS 2002; FOLARDA 2004; GOES 2003). SU.FOL.OM3 2010 used 5‐methyltetrahydrofolate instead of folic acid.
FOLARDA 2004, GOES 2003, HOPE‐2 2006, NORVIT 2006, SEARCH 2010, WAFACS 2008 and WENBIT 2008 described lipid‐lowering drugs used as concomitant medication. SU.FOL.OM3 2010 reported omega 3 polyunsaturated fatty acids use as concomitant medication.
Three trials, BVAIT 2009, VISP 2004 and WAFACS 2008, were conducted in fortified population described as "...nutritional intervention programme with a specifically defined target, and fortified food products are expected to become a main source of the specific added nutrient" (Wirakartakusumah 1998). Two trials were performed in a mixed population (HOPE‐2 2006; VITATOPS 2010), and five were carried out in non‐fortified population (CHAOS 2002; FOLARDA 2004; GOES 2003; NORVIT 2006; WENBIT 2008). It was unclear for SEARCH 2010 and SU.FOL.OM3 2010.
Ten trials used composite outcomes in their analyses (CHAOS 2002; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008). Four trials included revascularisation or other vascular procedures (CHAOS 2002; GOES 2003; WAFACS 2008; WENBIT 2008). Eleven trials had stroke as the endpoint (BVAIT 2009; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). All trials assessed the impact of the intervention on myocardial infarction rates. None of the trials included pectoris angina as a component of composite outcomes.
Eleven studies reported the sample size calculation (BVAIT 2009; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). The trials used 80% or 90% power to detect between 20% and 50% reduction in endpoints.
Blood concentrations of total homocysteine blood levels at baseline were reported in 11 trials (BVAIT 2009; CHAOS 2002; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008;WENBIT 2008). Five trials reported the total homocysteine blood levels at the end follow‐up (CHAOS 2002; HOPE‐2 2006; NORVIT 2006; VISP 2004; WAFACS 2008). WENBIT 2008 described total homocysteine blood levels during the intervention: 10.8 (standard deviation (SD) 4.5) μmol/L at baseline versus to 7.6 (SD 2.2) μmol/L after one year of the intervention (B9 and B12 groups). There were no changes in the B6 and placebo groups. CHAOS 2002 did not report total homocysteine blood levels at baseline and end of follow‐up in the control arm. GOES 2003 reported total homocysteine blood levels at baseline and at the end follow‐up only for the intervention arm and not for the control. FOLARDA 2004 did not measure the circulating total homocysteine blood levels in either group.
Definitions used for defining myocardial infarction, stroke, unstable angina and death (any) are described in Appendix 5.
We identified several duplicate publications associated with eight randomised clinical trials (GOES 2003; HOPE‐2 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008).
Excluded studies
We excluded 44 studies in this review (25 in the prior versions and 19 in this update) by the following reasons: systematic review (Bazzano 2006; Clarke 2010; Holmes 2011; Huang 2012; Huo 2012; Jardine 2012; Ji 2013; Lee 2010; Mei 2010; Miller 2010; Pan 2012; Wang 2007; Wang 2012; Yang 2012; Zhang 2009; Zhang 2013; Zhou 2011), narrative review (Lonn 2007; Manolescu 2010; Méndez‐González 2010; Ntaios 2009; Vesin 2007; Wierzbicki 2007), randomised clinical trials with a follow‐up of less of one year (Earnest 2012; FINEST 2006; Imasa 2009; Lange 2004; PACIFIC 2002; Swiss 2002; Tighe 2011), observational study (Cui 2010; Mager 2009; Moghaddasi 2010; Rautiainen 2010), and miscellaneous (Deshmukh 2010; Durga 2011; Ebbing 2009; Ebbing 2009a; Green 2010; Ntaios 2010; Sharifi 2010; Shidfar 2009; Sudchada 2012; Zappacosta 2013). See table of Characteristics of excluded studies for details.
Risk of bias in included studies
The risk of bias in the included trials is summarised in Figure 3 and Figure 4, and detailed in the Characteristics of included studies tables.
![Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG03.png)
Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies
![Methodological quality summary: review authors' judgements about each methodological quality item for each included study](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG04.png)
Methodological quality summary: review authors' judgements about each methodological quality item for each included study
Allocation
Random sequence generation
The risk of bias arising from the method of generation of the allocation sequence was low in seven trials (BVAIT 2009; GOES 2003; HOPE‐2 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008). Five trials had an unclear risk for this domain (CHAOS 2002; FOLARDA 2004; NORVIT 2006; VISP 2004; WENBIT 2008).
Allocation concealment
We rated the risk of bias arising from the method of allocation concealment as low in nine trials (BVAIT 2009; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). Three trials showed an unclear risk for this domain (CHAOS 2002; FOLARDA 2004; GOES 2003).
Blinding
We rated the risk of bias arising from lack of blinding of participants and personnel as low in nine trials (BVAIT 2009; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). The risk of bias from blinding was unclear in one trial (CHAOS 2002). We rated the risk of bias arising from lack of blinding as high in two trials (FOLARDA 2004; GOES 2003).
Incomplete outcome data
We rated the risk of attrition bias as low in seven trials (BVAIT 2009; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010). We rated the risk of attrition bias as high in two trials (FOLARDA 2004; WENBIT 2008). We rated the risk of bias as unclear in three trials (CHAOS 2002; VISP 2004; WAFACS 2008).
Selective reporting
All trials had a low risk of bias in this domain.
Other potential sources of bias
Nine trials had a low risk of bias (BVAIT 2009; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VISP 2004; VITATOPS 2010; WAFACS 2008; WENBIT 2008). Three trials had an unclear risk of bias (CHAOS 2002; FOLARDA 2004; GOES 2003).
Effects of interventions
Results were based on 47,429 participants (summary of findings Table for the main comparison).
Primary outcomes
Non‐fatal or fatal myocardial infarction
Homocysteine‐lowering interventions compared with placebo or conventional care
Meta‐analysis of 11 randomised clinical trials did not show a difference in non‐fatal or fatal myocardial infarction between homocysteine‐lowering interventions and placebo or conventional care groups (1747/23,590 (7.40%) versus 1247/20,190 (6.17%); risk ratio (RR) 1.02, 95% confidence interval (CI) 0.95 to 1.10; P value = 0.58, I2 = 0%; high quality evidence) (CHAOS 2002; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008) (Analysis 1.1). Trial sequential analysis for myocardial infarction suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials (Figure 5). Figure 6 shows the funnel plot for myocardial infarction.
![Trial sequential analysis on myocardial infarction in 11 trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on myocardial infarction based on the diversity‐adjusted required information size (DARIS) of 10,888 patients. This DARIS was calculated based upon a proportion of patients with myocardial infarction of 6.17% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) does not cross the conventional alpha of 5%. After the fourth trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG05.png)
Trial sequential analysis on myocardial infarction in 11 trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on myocardial infarction based on the diversity‐adjusted required information size (DARIS) of 10,888 patients. This DARIS was calculated based upon a proportion of patients with myocardial infarction of 6.17% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) does not cross the conventional alpha of 5%. After the fourth trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.
![Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing myocardial infarction The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are upper and closer to the pooled estimate. The effect sizes of the smaller studies are more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG06.png)
Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing myocardial infarction
The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are upper and closer to the pooled estimate. The effect sizes of the smaller studies are more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.
Homocysteine‐lowering interventions (high‐dose) compared with homocysteine‐lowering interventions (low‐dose)
One trial found no significant difference in non‐fatal or fatal myocardial infarction between intervention and control groups (72/1841 (3.91%) versus 81/1835 (4.41%); RR 0.90, 95% CI 0.66 to 1.23, P value = 0.50) (VISP 2004). See Analysis 1.1.
Non‐fatal or fatal stroke
Homocysteine‐lowering interventions compared with placebo
Meta‐analysis of nine trials did not show a difference in non‐fatal or fatal stroke between homocysteine‐lowering interventions and placebo groups (968/22,348 (4.33%) versus 974/18,957 (5.13%); RR 0.91, 95% CI 0.82 to 1.00, P value = 0.06, I2 = 11%, high quality evidence) (FOLARDA 2004; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008) (Analysis 1.2.). Trial sequential analysis for stroke suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials (Figure 7). Figure 8 shows the funnel plot for stroke.
![Trial sequential analysis on stroke in nine trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on stroke based on the diversity‐adjusted required information size (DARIS) of 17,679 patients. This DARIS was calculated based upon a proportion of patients with stroke of 5.13% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 26%. The cumulative Z‐curve (blue line) temporally crosses the conventional alpha of 5%, but reverts to insignificant values. The cumulative Z‐curve never crosses the trial sequential alpha‐spending monitoring boundaries. After the third trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG07.png)
Trial sequential analysis on stroke in nine trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on stroke based on the diversity‐adjusted required information size (DARIS) of 17,679 patients. This DARIS was calculated based upon a proportion of patients with stroke of 5.13% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 26%. The cumulative Z‐curve (blue line) temporally crosses the conventional alpha of 5%, but reverts to insignificant values. The cumulative Z‐curve never crosses the trial sequential alpha‐spending monitoring boundaries. After the third trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.
![Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing stroke The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG08.png)
Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing stroke
The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.
Homocysteine‐lowering interventions (high‐dose) compared with homocysteine‐lowering interventions (low‐dose)
One trial found no significant difference in non‐fatal or fatal myocardial infarction between intervention and control groups (152/1814 (8.37%) versus 148/1835 (8.06%); RR 1.04, 95% CI 0.84 to 1.29, P value = 0.73) (VISP 2004). See Analysis 1.2.
Secondary outcomes
First unstable angina pectoris episode requiring hospitalisation
Meta‐analysis of four trials found no difference between intervention and placebo groups (910/8015 (11.35%) versus 468/4629 (10.11%); RR 0.98, 95% CI 0.80 to 1.21, P value = 0.87, I2 = 66%) (FOLARDA 2004; HOPE‐2 2006; NORVIT 2006; WENBIT 2008) (Analysis 1.3).
Hospitalisation for heart failure
One trial found no significant difference in hospitalisation for heart failure between intervention and placebo groups (202/2758 (7.32%) versus 174/2764 (6.69%); RR 1.16, 95% CI 0.96 to 1.41, P value = 0.13) (HOPE‐2 2006).
Death from any cause
Homocysteine‐lowering interventions compared with placebo
Meta‐analysis of 10 trials found no difference between intervention and placebo in mortality from any cause (2784/22,648 (12.29%) versus 2502/19,250 (12.99%); RR 1.01, 95% CI 0.96 to 1.07, P value = 0.39, I2 = 6%, high quality evidence) (BVAIT 2009; FOLARDA 2004; GOES 2003; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008) (Analysis 1.4). Trial sequential analysis for stroke suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials (Figure 9). Figure 10 shows the funnel plot for death from any cause.
![Trial sequential analysis on death from any cause in 10 trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on death from any cause based on the diversity‐adjusted required information size (DARIS) of 10,419 patients. This DARIS was calculated based upon a proportion of death from any cause out of 13% in the control group; a RRR of 15% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 16%. After the third trial, the cumulative Z‐curve (blue line) crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 15% relative risk reduction. Smaller risk reductions might still require further trials.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG09.png)
Trial sequential analysis on death from any cause in 10 trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on death from any cause based on the diversity‐adjusted required information size (DARIS) of 10,419 patients. This DARIS was calculated based upon a proportion of death from any cause out of 13% in the control group; a RRR of 15% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 16%. After the third trial, the cumulative Z‐curve (blue line) crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 15% relative risk reduction. Smaller risk reductions might still require further trials.
![Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing death from any cause This figure shows a low risk of publication bias. The circles show the point estimates of the included randomised clinical trials. The pattern of distribution simulates an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG10.png)
Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing death from any cause
This figure shows a low risk of publication bias. The circles show the point estimates of the included randomised clinical trials. The pattern of distribution simulates an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.
Homocysteine‐lowering interventions (high‐dose) compared with homocysteine‐lowering interventions (low‐dose)
One trial found no significant difference in mortality from any cause between intervention and control groups (99/1814 (5.45%) versus 117/1835 (6.37%); RR 0.86, 95% CI 0.66 to 1.11, P value = 0.24) (VISP 2004). See Analysis 1.4.
Serious or non‐serious adverse events
Meta‐analysis of seven trials assessing cancer found no difference in cancer between intervention and placebo groups (1558/18,130 (8.59%) versus 1334/14,739 (9.05%); RR 1.06, 95% CI 0.98 to 1.13, P value = 0.13, I2 = 0%, high quality evidence) (BVAIT 2009; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; WENBIT 2008). See Analysis 1.5. Trial sequential analysis for adverse events suggests that no more trials may be needed to disprove an intervention effect of 13% relative risk reduction (Figure 11).
![Trial sequential analysis on adverse events (cancer) in seven trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on adverse events (cancer) based on the diversity‐adjusted required information size (DARIS) of 17,676 patients. This DARIS was calculated based upon a proportion of patients developing cancer of 9% in the control group; a RRR of 13% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) crosses the trials sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials are needed to disprove an intervention effect of 13% relative risk reduction.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG11.png)
Trial sequential analysis on adverse events (cancer) in seven trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on adverse events (cancer) based on the diversity‐adjusted required information size (DARIS) of 17,676 patients. This DARIS was calculated based upon a proportion of patients developing cancer of 9% in the control group; a RRR of 13% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) crosses the trials sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials are needed to disprove an intervention effect of 13% relative risk reduction.
Sensitivity analysis
Heterogeneity for all main outcomes was low, as conveyed by the I2 values. Moreover, very few studies were included to allow subgroup analysis. Therefore, we did not pursue extensive investigation as previously planned. Due to the lack of important heterogeneity fixed‐effect and random‐effects models gave the same results for the main outcomes. Sensitivity analysis included trials with low risk of bias (BVAIT 2009; HOPE‐2 2006; NORVIT 2006; SEARCH 2010; SU.FOL.OM3 2010; VITATOPS 2010; WAFACS 2008; WENBIT 2008).
Primary outcomes
Non‐fatal or fatal myocardial infarction
Homocysteine‐lowering interventions compared with placebo
Meta‐analysis of seven trials found no significant difference in non‐fatal or fatal myocardial infarction between intervention and placebo groups (1707/21,960 (7.77%) versus 1219/18,572 (6.56%); RR 1.02, 95% CI 0.95 to 1.09, P value = 0.85, I2 = 0%) (Analysis 2.1).
Non‐fatal or fatal stroke
Homocysteine‐lowering interventions compared with placebo
Meta‐analysis of seven trials found no significant difference in non‐fatal or fatal stroke between intervention and placebo groups (967/21,960 (4.40%) versus 972/18,572 (5.23%); RR 0.91, 95% CI 0.81 to 1.01, P value = 0.28, I2 = 20%) (Analysis 2.2).
Secondary outcomes
First unstable angina pectoris episode requiring hospitalisation
Homocysteine‐lowering interventions compared with placebo
Meta‐analysis of three randomised clinical trials found no significant difference in first unstable angina pectoris episode requiring hospitalisation between intervention and placebo groups (904/7875 (11.47%) versus 460/4486 (10.25%); RR 0.99, 95% CI 0.79 to 1.24, P value = 0.95, I2 = 76%) (Analysis 2.3).
Death from any cause
Homocysteine‐lowering interventions compared with placebo
Meta‐analysis of eight trials found no significant difference in mortality from any cause between intervention and placebo groups (2246/22,208 (10.11%) versus 1953/18,814 (10.38%); RR 1.03, 95% CI 0.95 to 1.12, P value = 0.17, I2 = 33%) (Analysis 2.4).
Discussion
Summary of main results
This updated Cochrane Review of homocysteine‐lowering interventions (B vitamins) for preventing cardiovascular events identified 12 randomised controlled trials incorporating 47,429 participants. Trials reported different combinations of homocysteine‐lowering interventions compared with different control interventions. Overall, the trials had a low risk of bias and were adequately powered. Participants differed somewhat in cardiovascular risk levels (with established cardiovascular disease (CVD) or at high risk of CVD), baseline total homocysteine blood levels, access to foods fortified with folic acid or not, different dosages of vitamin and control groups, and treatment periods varying from two to seven years. We did not find significant differences on the incidence of myocardial infarction (fatal or non‐fatal), stroke (fatal or non‐fatal), death from any cause or adverse events (cancer).
Overall completeness and applicability of evidence
This updated review found evidence suggesting that homocysteine‐lowering interventions (vitamins B6, B12 and folic acid (B9)) are not useful for preventing cardiovascular events. We conducted a sensitivity analysis restricted to trials with low risk of bias for myocardial infarction, stroke and death from any cause. These results show consistency and are based on data from trials that included a broad range of patients with different co‐morbidities who received different treatment approaches. Although these aspects could be considered as a threat to applicability, the consistency in the results derived from our analyses shows that the included trials may represent a broad picture of patients with a high risk of cardiovascular events.
This updated Cochrane Review found no new trials assessing homocysteine‐lowering interventions for preventing cardiovascular events. Therefore, this update version shows the same findings as Martí‐Carvajal 2013. It showed that supplementary vitamin B6, B12 and folic acid administration could not prevent cardiovascular events in patients with or without pre‐existing cardiovascular disease. The trial sequential analysis for the same outcomes suggested that no more randomised trials are needed to assess the benefits and harms of homocysteine‐lowering interventions for preventing cardiovascular events (Figure 5; Figure 7; Figure 9; Figure 11). Martí‐Carvajal 2013 found a null effect of vitamin B‐complex supplementation on cancer (Figure 11).
Quality of the evidence
We conducted GRADE assessments on outcomes using the meta‐analysed trials. Overall, the included trials had a low risk of bias (Figure 3; Figure 4).
summary of findings Table for the main comparison shows the quality of evidence for homocysteine‐lowering interventions compared with placebo or standard care for preventing cardiovascular events. The evidence available in this setting can be considered high quality due to the consistency of the results of the 12 trials for the main outcomes assessed (myocardial infarction, stroke and death from any cause), the precision in the pooled estimates, and the design and execution of these trials, which can be judged to be free of major threats to their validity.
Potential biases in the review process
In a systematic review process, there are a group of biases called significance‐chasing biases, such as publication bias and selective outcome reporting bias (Ioannidis 2010). Selective outcome reporting bias operates through suppression of information on specific outcomes and has similarities to study publication bias in that 'negative' results remain unpublished (Ioannidis 2010). This Cochrane Review found that overall the included randomised trials had a low risk of attrition bias and a low risk of selective outcome reporting bias (Figure 3; Figure 4).
Agreements and disagreements with other studies or reviews
Our results are similar to the other non‐Cochrane reviews (Clarke 2010; Huang 2012; Huo 2012; Ji 2013). These four reviews differ in their eligibility criteria resulting in the following: i) the inclusion by Clarke 2010, Huang 2012, Huo 2012 and Ji 2013 of the HOST trial (Jamison 2007), designed to assess the effects of homocysteine in patients with kidney or renal disease, which is out of our scope; ii) Clarke 2010 and Huo 2012 included all the trials in their pooled analysis (whereas we preferred to present the results from trials controlled with placebo separately from the results of the trials that compared different doses of homocysteine‐lowering drugs (VISP 2004)); iii) it can be concluded from the Clarke 2010 publication that the authors had access to some additional data from CHAOS 2002, which we had to extract from an abstract; and finally iv) our systematic review includes five additional trials not considered in Clarke 2010, with 12,031 more patients that allowed us to obtain more accurate estimates for our outcomes of interest (BVAIT 2009; FOLARDA 2004; GOES 2003; SU.FOL.OM3 2010; VITATOPS 2010). Despite these differences, both reviews have similar results for the most relevant outcomes, which did not show significant effects derived from homocysteine‐lowering drugs on cardiovascular events or overall mortality.
On the other hand, two randomised controlled trials (Jamison 2007; Vianna 2007), and one systematic review (Jardine 2012; Pan 2012) involving patients with end‐stage renal disease, found no effects of homocysteine‐lowering interventions for preventing cardiovascular events.
Regarding cancer, this Cochrane Review shows similar results to a recent meta‐analysis involving data on 50,000 individuals (Vollset 2013). Both meta‐analyses found no increased risk of cancer associated with homocysteine‐lowering interventions.
![Homocysteine metabolism (Reproduced with Dr Félix TM's permission from Brustolin 2010)](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG01.png)
Homocysteine metabolism (Reproduced with Dr Félix TM's permission from Brustolin 2010)
![Study flow diagram for this update](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG02.png)
Study flow diagram for this update
![Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG03.png)
Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies
![Methodological quality summary: review authors' judgements about each methodological quality item for each included study](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG04.png)
Methodological quality summary: review authors' judgements about each methodological quality item for each included study
![Trial sequential analysis on myocardial infarction in 11 trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on myocardial infarction based on the diversity‐adjusted required information size (DARIS) of 10,888 patients. This DARIS was calculated based upon a proportion of patients with myocardial infarction of 6.17% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) does not cross the conventional alpha of 5%. After the fourth trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG05.png)
Trial sequential analysis on myocardial infarction in 11 trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on myocardial infarction based on the diversity‐adjusted required information size (DARIS) of 10,888 patients. This DARIS was calculated based upon a proportion of patients with myocardial infarction of 6.17% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) does not cross the conventional alpha of 5%. After the fourth trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.
![Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing myocardial infarction The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are upper and closer to the pooled estimate. The effect sizes of the smaller studies are more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG06.png)
Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing myocardial infarction
The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are upper and closer to the pooled estimate. The effect sizes of the smaller studies are more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.
![Trial sequential analysis on stroke in nine trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on stroke based on the diversity‐adjusted required information size (DARIS) of 17,679 patients. This DARIS was calculated based upon a proportion of patients with stroke of 5.13% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 26%. The cumulative Z‐curve (blue line) temporally crosses the conventional alpha of 5%, but reverts to insignificant values. The cumulative Z‐curve never crosses the trial sequential alpha‐spending monitoring boundaries. After the third trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG07.png)
Trial sequential analysis on stroke in nine trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on stroke based on the diversity‐adjusted required information size (DARIS) of 17,679 patients. This DARIS was calculated based upon a proportion of patients with stroke of 5.13% in the control group; a RRR of 20% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 26%. The cumulative Z‐curve (blue line) temporally crosses the conventional alpha of 5%, but reverts to insignificant values. The cumulative Z‐curve never crosses the trial sequential alpha‐spending monitoring boundaries. After the third trial, the cumulative Z‐curve crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 20% relative risk reduction. Smaller risk reductions might still require further trials.
![Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing stroke The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG08.png)
Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing stroke
The circles show the point estimates of the included randomised clinical trials. The pattern of distribution resembles an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.
![Trial sequential analysis on death from any cause in 10 trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on death from any cause based on the diversity‐adjusted required information size (DARIS) of 10,419 patients. This DARIS was calculated based upon a proportion of death from any cause out of 13% in the control group; a RRR of 15% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 16%. After the third trial, the cumulative Z‐curve (blue line) crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 15% relative risk reduction. Smaller risk reductions might still require further trials.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG09.png)
Trial sequential analysis on death from any cause in 10 trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on death from any cause based on the diversity‐adjusted required information size (DARIS) of 10,419 patients. This DARIS was calculated based upon a proportion of death from any cause out of 13% in the control group; a RRR of 15% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 16%. After the third trial, the cumulative Z‐curve (blue line) crosses the trial sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials may be needed to disprove an intervention effect of 15% relative risk reduction. Smaller risk reductions might still require further trials.
![Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing death from any cause This figure shows a low risk of publication bias. The circles show the point estimates of the included randomised clinical trials. The pattern of distribution simulates an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG10.png)
Funnel plot of data from the meta‐analysis of the effects of homocysteine‐lowering interventions for preventing death from any cause
This figure shows a low risk of publication bias. The circles show the point estimates of the included randomised clinical trials. The pattern of distribution simulates an inverted funnel. Larger trials are closer and upper to the pooled estimate. The effect sizes of the smaller trials are lower and more or less symmetrically distributed around the pooled estimate. This figure shows a low risk of publication bias.
![Trial sequential analysis on adverse events (cancer) in seven trials investigating homocysteine‐lowering interventions versus placebo Trial sequential analysis of homocysteine‐lowering interventions versus placebo on adverse events (cancer) based on the diversity‐adjusted required information size (DARIS) of 17,676 patients. This DARIS was calculated based upon a proportion of patients developing cancer of 9% in the control group; a RRR of 13% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) crosses the trials sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials are needed to disprove an intervention effect of 13% relative risk reduction.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-AFig-FIG11.png)
Trial sequential analysis on adverse events (cancer) in seven trials investigating homocysteine‐lowering interventions versus placebo
Trial sequential analysis of homocysteine‐lowering interventions versus placebo on adverse events (cancer) based on the diversity‐adjusted required information size (DARIS) of 17,676 patients. This DARIS was calculated based upon a proportion of patients developing cancer of 9% in the control group; a RRR of 13% in the experimental intervention group; an alpha (α) of 5%; a beta (β) of 20%; and a diversity of 0%. The cumulative Z‐curve (blue line) crosses the trials sequential beta‐spending monitoring boundary, showing that the area of futility has been reached. This suggests that no more trials are needed to disprove an intervention effect of 13% relative risk reduction.
![Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 1 Myocardial infarction.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-001-01.png)
Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 1 Myocardial infarction.
![Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 2 Stroke.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-001-02.png)
Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 2 Stroke.
![Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 3 First unstable angina pectoris episode requiring hospitalisation.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-001-03.png)
Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 3 First unstable angina pectoris episode requiring hospitalisation.
![Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 4 Death from any cause.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-001-04.png)
Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 4 Death from any cause.
![Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 5 Serious adverse events (cancer).](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-001-05.png)
Comparison 1 Homocysteine‐lowering treatment versus other (any comparisons), Outcome 5 Serious adverse events (cancer).
![Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 1 Myocardial infarction.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-002-01.png)
Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 1 Myocardial infarction.
![Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 2 Stroke.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-002-02.png)
Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 2 Stroke.
![Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 3 First unstable angina pectoris episode requiring hospitalisation.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-002-03.png)
Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 3 First unstable angina pectoris episode requiring hospitalisation.
![Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 4 Death from any cause.](https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD006612.pub4/media/CDSR/CD006612/rel0004/CD006612/image_n/nCD006612-CMP-002-04.png)
Comparison 2 Homocysteine‐lowering treatment versus other (Sensitivity analysis), Outcome 4 Death from any cause.
Homocysteine‐lowering interventions (folic acid, vitamin B6 and vitamin B12) compared with placebo or standard care for preventing cardiovascular events | ||||||
Patient or population: Adults at risk of or with established cardiovascular disease | ||||||
Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of participants | Quality of the evidence | Comments | |
Assumed risk | Corresponding risk | |||||
Placebo or standard care | Homocysteine‐lowering interventions (folic acid, vitamin B6 and vitamin B12) | |||||
Non‐fatal or fatal myocardial infarction | Study population | RR 1.02 | 43,290 | ⊕⊕⊕⊕ | ||
62 per 1000 | 64 per 1000 | |||||
Stroke | Study population | RR 0.91 | 40,815 | ⊕⊕⊕⊕ | ||
52 per 1000 | 47 per 1000 | |||||
Death from any cause | Study population | RR 1.01 | 41,898 | ⊕⊕⊕⊕ | ||
130 per 1000 | 131 per 1000 | |||||
Cancer | Study population | RR 1.06 | 32,869 | ⊕⊕⊕⊕ | ||
91 per 1000 | 96 per 1000 | |||||
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
GRADE Working Group grades of evidence | ||||||
1I² = 0%. |
Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
1 Myocardial infarction Show forest plot | 12 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
1.1 Homocysteine‐lowering versus placebo | 11 | 43780 | Risk Ratio (M‐H, Random, 95% CI) | 1.02 [0.95, 1.10] |
1.2 Homocysteine‐lowering treatment at high dose versus low dose | 1 | 3649 | Risk Ratio (M‐H, Random, 95% CI) | 0.90 [0.66, 1.23] |
2 Stroke Show forest plot | 10 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
2.1 Homocysteine‐lowering treatment versus placebo | 9 | 41305 | Risk Ratio (M‐H, Random, 95% CI) | 0.91 [0.82, 1.00] |
2.2 Homocysteine‐lowering treatment at high dose versus low dose | 1 | 3649 | Risk Ratio (M‐H, Random, 95% CI) | 1.04 [0.84, 1.29] |
3 First unstable angina pectoris episode requiring hospitalisation Show forest plot | 4 | 12644 | Risk Ratio (M‐H, Random, 95% CI) | 0.98 [0.80, 1.21] |
4 Death from any cause Show forest plot | 11 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
4.1 Homocysteine‐lowering treatment versus placebo | 10 | 41898 | Risk Ratio (M‐H, Random, 95% CI) | 1.01 [0.96, 1.07] |
4.2 Homocysteine‐lowering treatments at high dose versus low dose | 1 | 3649 | Risk Ratio (M‐H, Random, 95% CI) | 0.86 [0.66, 1.11] |
5 Serious adverse events (cancer) Show forest plot | 7 | 32869 | Risk Ratio (M‐H, Random, 95% CI) | 1.06 [0.98, 1.13] |
Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
1 Myocardial infarction Show forest plot | 7 | 40532 | Risk Ratio (M‐H, Random, 95% CI) | 1.02 [0.95, 1.09] |
1.1 Trials with low risk of bias (mixed populations) | 6 | 35090 | Risk Ratio (M‐H, Random, 95% CI) | 1.02 [0.95, 1.10] |
1.2 Trials with low risk of bias (only women included) | 1 | 5442 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.63, 1.22] |
2 Stroke Show forest plot | 7 | 40532 | Risk Ratio (M‐H, Random, 95% CI) | 0.91 [0.81, 1.01] |
2.1 Trials with low risk of bias (mixed populations) | 6 | 35090 | Risk Ratio (M‐H, Random, 95% CI) | 0.89 [0.81, 0.98] |
2.2 Trials with low risk of bias (only women included) | 1 | 5442 | Risk Ratio (M‐H, Random, 95% CI) | 1.14 [0.83, 1.57] |
3 First unstable angina pectoris episode requiring hospitalisation Show forest plot | 3 | 12361 | Risk Ratio (M‐H, Random, 95% CI) | 0.99 [0.79, 1.24] |
4 Death from any cause Show forest plot | 8 | 41022 | Risk Ratio (M‐H, Random, 95% CI) | 1.03 [0.95, 1.12] |
4.1 Trials with low risk of bias (mixed populations) | 7 | 35580 | Risk Ratio (M‐H, Random, 95% CI) | 1.05 [0.95, 1.15] |
4.2 Trials with low risk of bias (only women included) | 1 | 5442 | Risk Ratio (M‐H, Random, 95% CI) | 0.98 [0.83, 1.15] |