Angiotensin receptor blockers for heart failure

Balraj S Heran; Vijaya M Musini; Ken Bassett; Rod S Taylor; James M Wright

doi:10.1002/14651858.CD003040.pub2

. 2012 Apr 18;2012(4):CD003040. doi: 10.1002/14651858.CD003040.pub2

Angiotensin receptor blockers for heart failure

Balraj S Heran ^1,^✉, Vijaya M Musini ¹, Ken Bassett ¹, Rod S Taylor ², James M Wright ¹

Editor: Cochrane Heart Group

PMCID: PMC6823214 PMID: 22513909

Abstract

Background

Chronic heart failure (HF) is a prevalent world‐wide. Angiotensin receptor blockers (ARBs) are widely prescribed for chronic HF although their role is controversial.

Objectives

To assess the benefit and harm of ARBs compared with ACE inhibitors (ACEIs) or placebo on mortality, morbidity and withdrawals due to adverse effects in patients with symptomatic HF and left ventricular systolic dysfunction or preserved systolic function.

Search methods

Clinical trials were identified by searching CENTRAL, HTA, and DARE, (The Cochrane Library 2010 Issue 3), as well as MEDLINE (2002 to July 2010), and EMBASE (2002 to July 2010). Reference lists of retrieved articles and systematic reviews were checked for additional studies not identified by the electronic searches.

Selection criteria

Double blind randomised controlled trials in men and women of all ages who have symptomatic (NYHA Class II to IV) HF and: 1) left ventricular systolic dysfunction, defined as left ventricular ejection fraction (LVEF) ≤40%; or 2) preserved ejection fraction, defined as LVEF >40%.

Data collection and analysis

Two authors independently assessed risk of bias and extracted data from included studies.

Main results

Twenty two studies evaluated the effects of ARBs in 17,900 patients with a LVEF ≤40% (mean 2.2 years). ARBs did not reduce total mortality (RR 0.87 [95% CI 0.76, 1.00]) or total morbidity as measured by total hospitalisations (RR 0.94 [95% CI 0.88, 1.01]) compared with placebo.

Total mortality (RR 1.05 [95% CI 0.91, 1.22]), total hospitalisations (RR 1.00 [95% CI 0.92, 1.08]), MI (RR 1.00 [95% CI 0.62, 1.63]), and stroke (RR 1.63 [0.77, 3.44]) did not differ between ARBs and ACEIs but withdrawals due to adverse effects were lower with ARBs (RR 0.63 [95% CI 0.52, 0.76]). Combinations of ARBs plus ACEIs increased the risk of withdrawals due to adverse effects (RR 1.34 [95% CI 1.19, 1.51]) but did not reduce total mortality or total hospital admissions versus ACEI alone.

Two placebo‐controlled studies evaluated ARBs in 7151 patients with a LVEF >40% (mean 3.7 years). ARBs did not reduce total mortality (RR 1.02 [95% CI 0.93, 1.12]) or total morbidity as measured by total hospitalisations (RR 1.00 [95% CI 0.97, 1.05]) compared with placebo. Withdrawals due to adverse effects were higher with ARBs versus placebo when all patients were pooled irrespective of LVEF (RR 1.06 [95% CI 1.01, 1.12]).

Authors' conclusions

In patients with symptomatic HF and systolic dysfunction or with preserved ejection fraction, ARBs compared to placebo or ACEIs do not reduce total mortality or morbidity. ARBs are better tolerated than ACEIs but do not appear to be as safe and well tolerated as placebo in terms of withdrawals due to adverse effects. Adding an ARB in combination with an ACEI does not reduce total mortality or total hospital admission but increases withdrawals due to adverse effects compared with ACEI alone.

Plain language summary

Are angiotensin receptor blockers (ARBs) an effective treatment for heart failure?

Drugs called angiotensin receptor blockers (ARBs), such as losartan (brand name: Cozaar), candesartan (Atacand), eprosartan (Teveten), irbesartan (Avapro), telmisartan (Micardis) and valsartan (Diovan) are commonly used to treat heart failure. We asked whether ARBs reduced death, or severe disability as assessed by hospital admission for any reason versus an inert substance (placebo) or another class of drugs called ACE inhibitors, such as ramipril (Altace), captopril (Capoten), enalapril (Vasotec), fosinopril (Monopril), lisinopril (Prinivil, Zestril), and quinapril (Accupril). We also asked whether combining an ARB with an ACE inhibitor is more effective than an ACE inhibitor alone in reducing death, disability, or hospital admission for any reason. The scientific literature was searched to find all trials that had assessed these questions.

We found 24 trials that randomly assigned participants to take either an ARB or control substance (placebo or ACEI). These trials evaluated ARBs in 25,051 patients with heart failure and followed them for 2 years. ARBs were no better than placebo or ACE inhibitors in reducing the risk of death, disability, or hospital admission for any reason. However, more patients stopped treatment early with ARBs than with placebo due to side effects. Adding an ARB to an ACEI also did not reduce the risk of death, disability, or hospital admission for any reason as compared to ACEI alone, although more patients taking the combination stopped early due to side effects.

Background

Chronic heart failure (HF) is a serious condition associated with high morbidity and mortality rates. Therefore, the goal of any pharmacological therapy is to lower the risk of adverse clinical outcomes associated with this chronic disease.

It is a well‐known fact that excess activation of the renin‐angiotensin system (RAS) contributes to the pathophysiology of heart failure. Activation of the RAS results in increased production of angiotensin I (AI), which is converted to angiotensin II (AII) by angiotensin‐converting enzyme (ACE). AII is a potent vasoconstrictor and also stimulates aldosterone secretion, which increases sodium and water retention (Erhardt 2005). AII and aldosterone are also implicated in other potentially deleterious effects on the cardiovascular system, including endothelial damage, sympathetic activation, collagen formation and decreased nitric oxide production (Erhardt 2005). Together, these effects put a strain on the heart, which can eventually lead to heart failure. With the understanding that the RAS plays such a vital role in the progression of heart failure, two drug classes, ACE inhibitors and angiotensin receptor blockers (ARBs), were developed to inhibit the RAS and thus provide a potentially beneficial therapeutic approach for the treatment of heart failure.

ACE inhibitors are indicated as first‐line treatment for HF (ACCF/AHA 2009; CCS 2006; ESC 2008) because they have been conclusively demonstrated in major clinical trials evaluating morbidity and mortality to reduce mortality as well as rates of reinfarction and hospitalizations for heart failure (Flather 2000). ACE inhibitors achieve their favourable effects by blocking the production of AII, thereby inhibiting its biological effects, such as enhanced vasoconstriction and excessive sodium and water retention. The favourable effect of ACE inhibitors may also be partly attributed to the fact that ACE is identical to another enzyme called kininase II that is responsible for kinin degradation. By retarding degradation of bradykinin, ACE inhibitors prolong its beneficial vasodilatory and antitrophic effects (Jong 2002). However, the accumulation of bradykinin in the lung probably causes the side effect of dry cough and possibly other side effects that force some patients to discontinue treatment with ACE inhibitors (McMurray 2005).

Initially, it was believed that ACE is the only enzyme that will catalyze the production of AII from AI. However, it is unclear how complete the blockade by ACE inhibitors is and if there is continuing angiotensin II formation during chronic treatment with ACE inhibitors (Wolny 1997). Numerous studies have now shown that some patients on long‐term treatment with ACE inhibitors eventually have AII levels return to pretreatment levels, demonstrating that the blockade of the RAS by the ACE inhibitor is incomplete. This phenomenon is referred to as “ACE escape” and it may be the result of AII formation through non‐ACE‐dependent pathways. For example, chymase, a serine protease found in the human heart and other tissues, is able to form AII from AI and is not blocked by ACE inhibitors (Wolny 1997). The physiological significance of these non‐ACE‐dependent pathways for AII formation is not known at the present time.

In contrast, RAS blockade with ARBs is achieved by inhibiting the binding of AII to the angiotensin II type 1 (AT1) receptor, which is believed to mediate the harmful cardiovascular effects of angiotensin II. ARBs are believed to provide a more effective means of blockade of the RAS than is possible with ACE inhibitors because this blockade at the receptor level is independent of the pathway for AII formation (Erhardt 2005). In addition, this drug class allows the displaced AII to continue to bind to the angiotensin II type II (AT2) receptors that are not blocked by ARBs. Since AT2 receptors are believed to mediate favourable vasodilatory and anti‐trophic effects, this unopposed stimulation of the AT2 receptors may confer a theoretical advantage with ARBs over ACE inhibitors [Azizi 2004]. Furthermore, ARBs may be better tolerated since they do not interfere with the degradation of bradykinin that is responsible for cough and possibly other side effects of ACE inhibitors.

Despite their theoretical benefits, several clinical practice guidelines recommend ARBs only in ACE‐intolerant patients because clinical trial evidence of their effectiveness in heart failure patients is less robust yet ARBs are still widely prescribed (ACCF/AHA 2009; CCS 2006; ESC 2008). A systematic review of existing trial data may therefore provide new insights on the use of this drug class in patients with HF.

Why it is important to do this review

In recent years, several meta‐analyses evaluating ARBs for HF have been published (Jong 2002; Lakhdar 2008; Lee 2004; Shah 2010; Sharma 2000; Shibata 2008). In fact, the authors who originally developed the protocol for this Cochrane review (Jong 2002b) decided to publish the results of their meta‐analysis elsewhere (Jong 2002). Subsequently a published meta‐analysis included all 17 studies from Jong 2002 as well as an additional five studies, totaling 22 studies evaluating ARBs in HF patients and LVEF ≤40% (Lee 2004). No new trials in patients with HF and LVEF ≤40% have been published since Lee 2004, which quantifies the effect of ARBs when compared with placebo (with and without background ACE inhibitors) and ACE inhibitors on all‐cause mortality and heart failure hospitalisations. This meta‐analysis concluded that there was a reduction in all‐cause mortality (RR 0.87 [95% CI 0.76 to 1.00]; ARR=7.1%, NNT=14, p=0.05) with ARBs and HF hospitalisations (RR 0.71 [95% CI 0.61 to 0.82]; ARR=7.9%, NNT=13, p<0.00001), as compared to placebo. For ARBs versus ACE inhibitors, all‐cause mortality (RR 1.05 [95% CI 0.91 to 1.22]) and HF hospitalisations (RR 0.96 [95% CI 0.83 to 1.11]) did not differ. For ARB plus ACE inhibitor combinations versus ACE inhibitors alone, all‐cause mortality was not reduced (RR 0.98 [95% CI 0.90 to 1.06]) but HF hospitalisations were reduced (RR 0.81 [95% CI 0.74 to 0.89]; ARR=4.4%; NNT=23; p<0.00001).

Of the 22 chronic HF trials included in the meta‐analysis, 17 were short‐term studies evaluating the effect of an ARB on cardiac haemodynamic and/or neurohormonal parameters. They were not designed to assess the long‐term impact of ARBs on mortality or morbidity and contribute very little (< 2% per trial) to the overall estimate of the effect of ARBs on all‐cause mortality and hospitalisations for HF.

Four large‐scale trials (CHARM‐Added 2003; CHARM‐Alternative 2003; ELITE II 2000; Val‐HeFT 2001) contribute nearly all of the data to the meta‐analysis. However, the Lee 2004 meta‐analysis is limited to only two outcomes (all‐cause mortality and HF hospitalisations), whereas each trial reports on other mortality and morbidity outcomes, as well as withdrawals due to adverse effects. Furthermore, a recent systematic review of two large‐scale studies evaluating ARBs in patients with HF and preserved systolic function (LVEF >40%) has also limited its meta‐analysis to all‐cause mortality and heart failure hospitalisations (Shah 2010).

Therefore, the purpose of this review is not only to provide an update on the available literature, but also to more completely evaluate the data from large‐scale clinical trials using a broader range of outcomes in order to get a more complete understanding of the benefit and harm of ARBs in the treatment of chronic heart failure.

Objectives

In patients with symptomatic HF (NYHA Class II to IV) and EF ≤40%, to assess the benefit and harm of:

- ARBs versus placebo, in addition to standard therapy, with or without an ACE inhibitor.
- ARBs versus ACEIs.
- ARB plus ACEI combination therapy versus ACEI alone.

In patients with symptomatic HF (NYHA Class II to IV) and EF >40% (i.e. preserved EF), to assess the benefit and harm of:

- ARBs versus placebo, in addition to standard therapy, with or without an ACE inhibitor.
- ARBs versus ACEIs.
- ARB plus ACEI combination therapy versus ACEI alone.

Methods

Criteria for considering studies for this review

Types of studies

Published double‐blind, randomised controlled trials enrolling patients with symptomatic HF with an ARB as the experimental intervention were considered.

Studies were included if:

Treatment assignments were randomised and administrated in parallel (i.e. no crossover). Studies with more than one ARB arm were permitted.
Mortality and/or morbidity rates were reported as either clinical or safety endpoints. Studies were counted even if no event of interest occurred during the study if this was explicitly reported. Endpoints were counted if they occurred outside of the period of randomised therapies. No maximum limit was imposed on the length of follow‐up.
Controlled interventions were either placebos or ACEIs. More than one control arm was allowed.
Duration of randomised therapy was at least four weeks (i.e. studies in which treatment consists of only a single one‐time dose of the ARB, such as in haemodynamic, pharmacodynamic dose‐response, or safety studies, were excluded).

Studies were excluded if:

Protocol included co‐administration of other non‐randomised investigational agents (e.g. angiotensin II, bradykinin).
Published only in abstract forms or non‐peer reviewed journals whereby no further or insufficient information could be procured from the authors.

Types of participants

For the purpose of this review, we relied on the investigators’ HF diagnosis. Men and women of all ages who had symptomatic HF with NYHA functional class II‐IV. Asymptomatic subjects with left ventricular dysfunction were excluded. Echocardiographic, radionuclide, or angiographic documentation of ventricular dysfunction was not required if the clinical diagnosis of HF has been established by the study investigators.

A LVEF cut‐off point of 40% was used to differentiate between HF patients with reduced EF from those with preserved EF.

Types of interventions

Experimental intervention with any ARB at any dose, including candesartan, eprosartan, irbesartan, losartan, olmesartan, telmisartan, valsartan, and other ARBs not currently marketed.

Studies with more than one control arm were allowed if at least one control intervention was either a placebo or an ACEI. If placebos were used as controls, both studies with and without background open‐label ACEI therapy were included.

The interventions could be administered as single agents, combination therapies (including ARB plus ACEI combo), and in fixed or stepped/titrated doses.

Types of outcome measures

All clinical events or other outcome measures reported post‐randomisation were included in this review. No maximum limit was imposed on the length of follow‐up.

Primary:

Total mortality
- Cardiovascular mortality
- Non‐cardiovascular mortality
Cardiovascular morbidity
- Myocardial infarction (MI)
- Stroke
Total hospitalisations
- Hospitalisations for HF (defined as a hospital admission for worsening signs or symptoms of HF, for complications relating to the treatment of HF, or for syncope or arrhythmias related to HF)
- Other hospitalisations

Secondary:

Withdrawals due to adverse effects (WDAE)

Search methods for identification of studies

The protocol for this review was first published in Issue 2, 2001 of the Cochrane Database of Systematic Reviews (Jong 2002b). The authors published the results of their meta‐analysis a year later in another medical journal (Jong 2002). However, the full Cochrane review has never been published. In 2009, we took over the protocol in order to complete the systematic review according to the rigorous quality standards of The Cochrane Collaboration.

The strategies included in the published Cochrane protocol to search the bibliographic databases for potentially relevant studies (Jong 2002b) have been listed in the appendices (Appendix 1; Appendix 2; Appendix 3; Appendix 4). It is not possible to know if these exact search strategies have been used for the review since the published report did not include the search strategies (Jong 2002). Nevertheless, other systematic reviews have also been published in recent years (Lakhdar 2008; Lee 2004; Shah 2010; Shibata 2008), which have served to independently verify that Jong 2002 identified all relevant studies evaluating ARBs for HF up to May 2001. Therefore, we searched for additional reports of studies published from May 2001 onwards using updated electronic database search strategies (Appendix 5; Appendix 6; Appendix 7).

Electronic searches

Randomized controlled trials have been identified from previously published systematic reviews. This list of studies has been updated by searching the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library 2010 Issue 3, MEDLINE (May 2001 to July 2010), and EMBASE (May 2001 to July 2010). Health Technology Assessment (HTA) and Database of Abstracts of Reviews of Effects (DARE) databases have been searched via The Cochrane Library 2010 Issue 3.

Search strategies were designed with reference to those of the previous systematic review and in accordance with the Cochrane Heart Group methods and guidance. Electronic databases were searched using a strategy combining a modified form of the Cochrane Highly Sensitive Search Strategy for identifying randomised trials in MEDLINE: sensitivity‐maximizing version (Lefebvre 2011) with selected MeSH terms and free text terms relating to angiotensin receptor blockers and heart failure. The MEDLINE search strategy was translated into the other databases using the appropriate controlled vocabulary as applicable.

Searches have been limited to randomised controlled trials and a filter applied to limit by humans. No language restrictions have been applied. Consideration was given to variations in terms used and spellings of terms in different countries so that studies were not missed by the search strategy because of such variations.

See appendices for a list of the search strategies used for this review (Appendix 5; Appendix 6; Appendix 7).

Searching other resources

Reference lists of retrieved articles and systematic reviews and meta‐analyses were checked for any studies not identified by the electronic searches.

Data collection and analysis

Selection of studies

The titles and abstracts of citations identified by the electronic searches from May 2001 onwards were examined for possible inclusion by two reviewers (BSH & VM) working independently. Full publications of potentially relevant studies were retrieved (and translated into English where required) and two reviewers (BSH & VM) then independently determined study eligibility using a standardised inclusion form. Any disagreements about study eligibility were resolved by discussion and, if necessary, a third reviewer (KB) was asked to arbitrate.

Data extraction and management

Data from included studies were extracted by one reviewer (BSH or VM) using standardised data extraction forms and checked by a second reviewer (VM or BSH). If data were presented numerically (in tables or text) and graphically (in figures), the numeric data were used because of possible measurement error when estimating from graphs. A second reviewer confirmed all numeric calculations and extractions from graphs or figures. Any discrepancies were resolved by consensus.

Data on patient characteristics (e.g. age, sex, race, NYHA class), details of the intervention (including drug name and dose), ACEI background therapy, and length of follow up were also extracted.

Assessment of risk of bias in included studies

Two reviewers (BSH, JMHC) independently assessed the risk of bias in included studies using the Cochrane Collaboration's recommended tool, which is a domain‐based critical evaluation of the following domains: sequence generation; allocation concealment; blinding of outcome assessment; incomplete outcome data; and selective outcome reporting (Higgins 2011). Assessments of risk of bias are provided in the risk of bias table for each study.

Measures of treatment effect

Dichotomous outcomes for each comparison have been expressed as relative risks with 95% confidence intervals (CI). If it was statistically significant then absolute risk difference, the associated number needed to treat/harm was calculated.

Three treatment comparisons have been made: 1) ARBs versus placebo, without background ACEI therapy; 2) ARBs versus ACEIs; and 3) ARB plus ACEI combination therapy versus ACEI alone. The latter comparison of combination therapy with ARBs and ACEIs against ACEIs alone is methodologically similar to a comparison of ARBs against placebos where background (open‐label) ACEI therapy is given. Trials with these latter two designs have been categorized as making the same type of treatment comparison.

Dealing with missing data

If there were multiple reports of the same study, the duplicate publications were scanned for additional data. Outcome results have been extracted at all follow up points post‐randomisation. Study authors were contacted where necessary to provide additional information.

Assessment of heterogeneity

In absence of substantial heterogeneity as judged by the I² measure, a fixed effect model was used as the default model (Higgins 2011).

If there was substantial heterogeneity associated with an effect estimate, a random effects model was applied. This model provides a more conservative statistical comparison of the difference between intervention and control because a confidence interval around the effect estimate is wider than a CI around a fixed effect estimate. If a statistically significant difference was still present using the random effects model, the fixed effect pooled estimate and 95% CI have been reported because of the tendency of smaller trials, which are more susceptible to publication bias, to be over weighted with a random effects analysis (Heran 2008a; Heran 2008b).

Assessment of reporting biases

No language restrictions have been applied.

Data synthesis

Data have been processed in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). Data synthesis and analyses have been done using Review Manager 5.0 software.

Subgroup analysis and investigation of heterogeneity

Subgroup analysis

Symptomatic HF patients have been divided into 2 subgroups according to baseline LVEF: 1) patients with LVEF ≤40%; and 2) patients with LVEF >40%.

Heterogeneity

Heterogeneity amongst included studies was explored qualitatively (by comparing the characteristics of included studies) and quantitatively (using the I² statistic and chi‐squared test of heterogeneity). Where appropriate, data from each study have been pooled using a fixed effect model, except where substantial heterogeneity exists. The funnel plot and the Egger test were used to examine small study bias (Egger 1997).

Sensitivity analysis

Pooled effect estimates have been recalculated after exclusion of studies with a high risk of bias, i.e.: 1) unpublished studies; 2) studies in which participants with a previous intolerance of an ARB were excluded; or 3) studies with an ARB tolerability phase preceding randomisation. Each pooled treatment effect of ARBs will be considered qualitatively robust if the upper and lower confidence bounds for the pooled treatment effect remain unchanged in direction with respect to unity.

Results

Description of studies

Results of the search

Our search of the electronic databases yielded a total of 2622 records, and after de‐duplication 1911 records remained. After reviewing the titles and abstracts, we retrieved 55 full‐text articles for possible inclusion. A total of 21 studies (24 publications) were excluded: 10 reported no useful outcomes; five were not double‐blind; two were crossover studies; two studies included patients who were not all taking background ACE inhibitor therapy; one was not randomised; and one had an inappropriate control. Four non‐Cochrane systematic reviews were also excluded. Twenty four studies (27 publications) met the inclusion criteria and had extractable data to assess the effects of ARBs compared with ACE inhibitors or placebo on mortality and morbidity in patients with HF. The study selection process is summarized in the PRISMA flow diagram shown in Figure 1.

Included studies

The systematic review published in 2002 (Jong 2002) included a total of 17 studies, of which one was incorrectly judged to be a report of an independent study when, in fact, it was a review article (Weber 1997) that briefly summarized another included study (Crozier 1995). Thus, Weber 1997 has been listed as a duplicate publication of Crozier 1995 in this review.

In addition to the 16 studies (n=12,295) from Jong 2002 that have met the inclusion criteria of this review, eight studies (ARCH‐J 2003; CHARM‐Added 2003; CHARM‐Alternative 2003; CHARM‐Preserved 2003; HEAVEN 2002; I‐PRESERVE 2008; Mitrovic 2003; REPLACE 2001) have been identified by our search and have met the inclusion criteria. Thus, a total of 24 studies reporting data on a total of 25,051 patients with symptomatic HF have been included in this review. Twenty‐two studies randomised 17,900 patients with LVEF ≤40% and 2 studies randomised 7151 patients with LVEF >40%. Details of the studies included in the review are listed in the Characteristics of included studies table.

Eight duplicate publications of six included trials were identified. All 24 included studies were published in English, fifteen (63%) were multinational trials, and all but one (ADEPT 2001) were multicenter clinical trials. Twenty‐three (96%) of the included studies were industry‐sponsored while the remaining study did not report the source of funding.

Patients with LVEF >40%

In HF patients with preserved ejection fraction (EF) (i.e. >40%), only two placebo‐controlled studies have been published [CHARM‐Added 2003; I‐PRESERVE 2008). No studies comparing ARBs with ACE inhibitors or ARB plus ACE inhibitor combination with ACE inhibitor monotherapy have been identified. CHARM‐Added 2003 (candesartan) and I‐PRESERVE 2008 (irbesartan) randomised a total of 7151 patients (52% females) with mean age of 70 (range 67 to 72) years and LVEF 57% (range 54 to 60%) for a weighted mean duration of 176 (range 146 to 198) weeks.

Patients with LVEF ≤40%

In HF patients with LVEF ≤40%, eight were placebo‐controlled studies (ARCH‐J 2003; CHARM‐Alternative 2003; Crozier 1995; Mitrovic 2003; Sharma 2000, III‐Int'l; Sharma 2000, III‐US; SPICE 2000; STRETCH 1999), six studies compared ARBs with ACE inhibitors (Dickstein 1995; ELITE 1997; ELITE II 2000; HEAVEN 2002; Lang 1997; REPLACE 2001), and six studies compared ARB plus ACE inhibitor combination therapy with ACE inhibitor monotherapy (ADEPT 2001; CHARM‐Added 2003; Hamroff 1999; Tonkon 2000; V‐HeFT 1999; Val‐HeFT 2001). In addition, two were multi‐arm studies, of which one randomly assigned patients to an ARB, ACE inhibitor, and placebo (Mazayev 1998) and the other randomised patients to ARB plus ACE inhibitor combination therapy, ARB monotherapy, and ACE inhibitor monotherapy (RESOLVD 1999).

Of the 22 included trials, 18 were short‐term studies evaluating the effect of an ARB on cardiac haemodynamic and/or neurohormonal parameters. These studies were not designed to assess the long‐term impact of ARBs on mortality or morbidity and contribute very little (< 4% per trial) to the overall estimate of the effect of ARBs on total mortality and total hospitalisations. Four large‐scale trials (CHARM‐Added 2003; CHARM‐Alternative 2003; ELITE II 2000; Val‐HeFT 2001) contribute nearly all of the data to the meta‐analysis.

ARBs versus placebo

Nine studies with a weighted mean duration of 67 (range 4 to 135) weeks randomised a total of 4623 patients (29% females) with a mean age of 64 (range 53 to 65) years and LVEF 31% (range 23 to 35%). Trial sample sizes varied widely from 101 to 2028, with candesartan being the most studied ARB (5 studies; 3652 patients) followed by losartan (3 studies; 870 patients), and valsartan (1 study; 101 patients). The largest published trial was CHARM‐Alternative 2003 (n=2028), which contributes at least 90% of the data for this comparison of ARBs versus placebo.

ARBs versus ACEIs

Eight studies randomised 5201 patients (28% females) with a mean age of 70 (range 54 to 74) years and LVEF 30% (23 to 31%). The weighted mean duration of these studies was 56 (8 to 137) weeks. Sample sizes ranged from 90 to 3152, with losartan being the ARB most compared with an ACE inhibitor (4 studies; 4156 patients) followed by valsartan (2 studies; 231 patients), candesartan (1 study; 436 patients), and telmisartan (1 study; 378 patients). ELITE II 2000 (n=3152) contributes 80‐90% of the overall ARB versus ACE inhibitor data to the meta‐analysis.

ARB plus ACEI combination versus ACEI alone

Seven studies randomised a total of 8260 patients (20% females) with mean age of 63 (range 60 to 66) years and LVEF 27% (range 22 to 29%) for a weighted mean duration of 101 (range 4 to 144) weeks. Two large trials, CHARM‐Added 2003 (candesartan) and Val‐HeFT 2001 (valsartan), provide nearly all the data for this comparison.

Excluded studies

Twenty one studies (24 publications) were excluded for reasons listed in the Characteristics of excluded studies table, with the most common reason being a failure to report any of the pre‐specified outcomes of this review.

Risk of bias in included studies

Our judgements about each risk of bias item for each included study and about each risk of bias item presented as percentages across all included studies are summarized in Figure 2 and Figure 3, respectively.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Allocation

Nearly all the trial publications simply reported that the trial was "randomised" but did not provide any details. Only 2/24 (8%) studies (I‐PRESERVE 2008; STRETCH 1999) reported details of appropriate sequence generation and 4/24 (17%) studies (CHARM‐Added 2003; CHARM‐Alternative 2003; CHARM‐Preserved 2003; I‐PRESERVE 2008) reported appropriate concealment of allocation.

Given the fact that many investigators use the term “randomised” when it is not justified, such vague reporting is insufficient to be confident that the allocation sequence was properly randomised and adequately concealed.

Blinding

Nine (38%) trials simply reported that the trial was “double‐blind” but did not provide any details about the blinding method. Thirteen (54%) trial publications described the blinding method as using “matching” or "matched" placebo. One trial described the method of blinding as "double dummy" (HEAVEN 2002). Due to the different dosage intervals of the drugs in Dickstein 1995, the authors stated that placebo tablets were provided in addition to secure blinding to treatment. The success of blinding in patients or investigators was not assessed in any of the included trials.

Incomplete outcome data

It is unlikely that attrition bias would have had an impact on the systematic review since most of the patients randomised in each trial completed the double blind treatment period. Those patients who were lost to follow up or dropped out prematurely were usually included in the clinical outcome or safety analysis of each trial.

Selective reporting

Eighteen (75%) of the included studies were not designed to assess treatment group differences in mortality and morbidity (as these were not the primary outcomes of these trials). Therefore, our assessment of selective reporting bias has been limited to the six long‐term health outcome trials (CHARM‐Added 2003; CHARM‐Alternative 2003; CHARM‐Preserved 2003; ELITE II 2000; I‐PRESERVE 2008; Val‐HeFT 2001). There is a potential for selective reporting bias in Val‐HeFT 2001 since this study only reported hospitalisations for HF but not total hospitalisations.

Other potential sources of bias

Selection bias

One of the exclusion criteria reported in one large‐scale trial was participants with a previous intolerance of an ARB (I‐PRESERVE 2008). This suggests that investigators have knowledge of each participant’s prior experience with this drug class and, thus, may select for patients who have responded favourably or have been found to tolerate treatment with ARBs. It was possible for us to prove selection bias in terms of WDAE (Analysis 1.9). Two included trials compared ARBs with placebo in patients with HF and a preserved EF (CHARM‐Preserved 2003; I‐PRESERVE 2008). In the CHARM‐Preserved 2003 trial, which did not exclude patients who demonstrated a previous ARB intolerance, there was a statistically significant increase in WDAE in patients treated with candesartan as compared to placebo (RR 1.32 [95% CI 1.12, 1.56]). In contrast, the increase in WDAE with irbesartan I‐PRESERVE 2008 trial did not reach statistical significance compared with placebo (RR 1.15 [95% CI 0.99, 1.33]).

1.9 — Comparison 1: ARBs versus placebo, Outcome 9: WDAE

Two smaller studies (ARCH‐J 2003; Hamroff 1999) included an ARB tolerability phase and those patients with a confirmed intolerance of an ARB were excluded prior to randomisation. A sensitivity analysis was conducted by excluding these two studies and the results of the meta‐analysis were not affected.

Publication bias

Twenty three (96%) of the 24 included trials were sponsored by the manufacturer of the ARB being studied and one study did not report the source of funding. Given the fact that industry sponsored trials with positive results are selectively published, it is likely that this source of bias has had a significant impact on this review since it only included and appraised published trial evidence.

In order to test for the possibility of publication bias, a funnel plot was created for 9 of the placebo‐controlled studies (ARCH‐J 2003; CHARM‐Alternative 2003; Crozier 1995; Mazayev 1998; Mitrovic 2003; Sharma 2000, III‐Int'l; Sharma 2000, III‐US; SPICE 2000; STRETCH 1999) that reported total mortality in patients with EF ≤40%. CHARM‐Added 2003 and I‐PRESERVE 2008 looked at patients with preserved ejection (LVEF >40%) and were therefore excluded from the funnel plot analysis. There was no visual indication of funnel plot asymmetry and the Egger test was not statistically significant (Figure 4). However, due to the small number of studies the power was too low to reliably test for asymmetry.

Funnel plot of ARBs versus placebo for total mortality in HF patients with EF ≤40%

Despite this limitation, there was still evidence of publication bias. One publication, a meta‐analysis of the effect of losartan on mortality in patients with HF, reported on deaths that occurred in two multicenter, placebo‐controlled studies (Sharma 2000, III‐Int'l; Sharma 2000, III‐US). Published reports of the complete data for these two multicenter studies were not identified by our comprehensive search.

Since the source of funding for these trials was almost always the manufacturer, the risk of bias is high. Therefore, we believe that this meta‐analysis may overestimate the benefits and underestimate the harms of ARB therapy. All studies, regardless of the findings, need to be published and accessible for secondary analysis in order to accurately assess the benefits and harms associated with this class of drugs.

Effects of interventions

ARBs versus placebo

Mortality

Patients with LVEF >40%

In 2 (n=7151) of the included studies in patients with preserved EF, ARBs did not reduce total mortality, cardiovascular mortality, or non‐cardiovascular mortality (Analysis 1.1; Analysis 1.2; Analysis 1.3).

1.1 — Comparison 1: ARBs versus placebo, Outcome 1: Total mortality

1.2 — Comparison 1: ARBs versus placebo, Outcome 2: Cardiovascular mortality

1.3 — Comparison 1: ARBs versus placebo, Outcome 3: Non‐cardiovascular mortality

Patients with LVEF ≤40%

Nine (n=4643) of the included studies reported total mortality (Analysis 1.1). In these studies, the reduction in total mortality with ARB therapy was of borderline statistical significance at the 5% error level (RR 0.87 [95% CI 0.76, 1.00]). However, this estimate may be subject to bias in favour of ARBs as it includes two unpublished studies that have been summarized in a meta‐analysis sponsored by the manufacturer of losartan (Sharma 2000, III‐Int'l; Sharma 2000, III‐US). When the analysis is limited to the 7 trials with full reporting, the difference between ARBs and placebo is not statistically significant (RR 0.91 [95% CI 0.79 to 1.04]).

There were also no differences between ARBs and placebo for cardiovascular and non‐cardiovascular mortality (Analysis 1.2; Analysis 1.3).

Morbidity

Patients with LVEF >40%

None of the included studies reported stroke or MI.

Patients with LVEF ≤40%

Only 2 (n=2298) of the included trials studying candesartan reported total MI and stroke (Analysis 1.4; Analysis 1.5), of which CHARM‐Alternative 2003 contributes nearly all the data for these outcome measures. There was no statistically significant difference between candesartan and placebo for stroke, but candesartan increased the risk of MI (RR 1.44 [95% CI 1.03, 2.01]; ARI=1.9%; NNH=52) compared with placebo.

1.4 — Comparison 1: ARBs versus placebo, Outcome 4: MI

1.5 — Comparison 1: ARBs versus placebo, Outcome 5: Stroke

Hospitalisations

Patients with LVEF >40%

Two (n=7151) of the included studies reported total hospitalisations, hospitalisations for HF, and hospitalisations for other causes (Analysis 1.6; Analysis 1.7; Analysis 1.8). When this outcome was divided according to cause, the reduction in hospitalisations for heart failure with ARBs was of borderline statistical significance at the 5% error level (RR 0.90 [95% CI 0.81, 1.00]). This was offset by a non‐significant increase in other hospitalisations with ARB therapy (RR 1.05 [95% CI 0.99, 1.11]). Overall, total hospitalisations were not reduced with ARBs compared with placebo.

1.6 — Comparison 1: ARBs versus placebo, Outcome 6: Total hospitalisations

1.7 — Comparison 1: ARBs versus placebo, Outcome 7: Hospitalisations for heart failure

1.8 — Comparison 1: ARBs versus placebo, Outcome 8: Other hospitalisations

Patients with LVEF ≤40%

Two (n=2298), 3 (n=2590), and 2 (n=2298) of the included studies reported total hospitalisations, hospitalisations for HF, and hospitalisations for other causes, respectively (Analysis 1.6; Analysis 1.7; Analysis 1.8). All data for these outcomes were from 3 candesartan studies, of which the long‐term CHARM‐Alternative 2003 trial contributed 90‐97% of the data. All‐cause hospitalisations were not reduced with candesartan compared with placebo. When this outcome was divided according to cause, candesartan reduced the risk of hospital admissions for HF (RR 0.71 [95% CI 0.61, 0.82]; ARR=8.0%; NNT=13) but increased the risk of hospital admissions for other causes (RR 1.12 [95% CI 1.00, 1.25]). CHARM‐Alternative 2003 contributed 98% of the data and demonstrated a statistically significant increase in other hospitalisations with candesartan (RR 1.13 [95% CI 1.01, 1.27]; ARI=4.6%; NNH=22).

Since an increase in hospitalisations for other causes can be attributed to a harmful effect of ARB therapy that is unrelated to the exacerbation of heart failure, we combined the data for all HF patients irrespective of LVEF (Analysis 1.8). There was a significantly increased risk of hospitalisations for other causes compared with placebo (RR 1.06 [95% CI 1.01, 1.12]; ARI=1.4%; NNH=72).

Withdrawals due to adverse effects

Patients with LVEF >40%

When the 2 large‐scale studies (n=7151) were pooled (Analysis 1.9), significantly more patients treated with ARBs withdrew due to adverse effects compared with placebo (RR 1.22 [95% CI 1.09, 1.36]; ARI=3%; NNH=33).

Patients with LVEF ≤40%

Six (n=3766) of the included studies reported WDAE (Analysis 1.9). When these trials were pooled, more patients in the ARB group discontinued therapy due to an adverse effect but the increase did not reach statistical significance (RR 1.14 [95% CI 0.97, 1.33]).

We pooled the data for all symptomatic HF patients irrespective of LVEF (Analysis 1.8) and WDAE were significantly higher in patients treated with ARBs (RR 1.19 [95% CI 1.09, 1.30]; ARI=1.4%; NNH=72).

ARBs versus ACE inhibitors

Mortality

Eight (n=5201), 4 (n=4131), and 4 (n=4131) of the included studies reported total mortality, cardiovascular mortality, and non‐cardiovascular mortality, respectively (Analysis 2.1; Analysis 2.2; Analysis 2.3). There was no difference between ARBs and ACEIs with regard to total mortality, cardiovascular mortality, or non‐cardiovascular mortality.

2.1 — Comparison 2: ARBs versus ACE inhibitors, Outcome 1: Total mortality

2.2 — Comparison 2: ARBs versus ACE inhibitors, Outcome 2: Cardiovascular mortality

2.3 — Comparison 2: ARBs versus ACE inhibitors, Outcome 3: Non‐cardiovascular mortality

Morbidity

MI and stroke were reported in two studies (n=3874) and 1 study (n=3152), respectively. There was no significant difference between treatment groups for these outcomes (Analysis 2.4; Analysis 2.5).

2.4 — Comparison 2: ARBs versus ACE inhibitors, Outcome 4: MI

2.5 — Comparison 2: ARBs versus ACE inhibitors, Outcome 5: Stroke

Hospitalisations

Four (n=4310) of the included studies reported total hospitalisations, hospitalisations for HF, and hospitalisations for other causes (Analysis 2.6; Analysis 2.7; Analysis 2.8). When these four studies were pooled, total hospitalisations, HF hospitalisations, and other hospitalisations did not differ between treatment groups.

2.6 — Comparison 2: ARBs versus ACE inhibitors, Outcome 6: Total hospitalisations

2.7 — Comparison 2: ARBs versus ACE inhibitors, Outcome 7: Hospitalisations for heart failure

2.8 — Comparison 2: ARBs versus ACE inhibitors, Outcome 8: Other hospitalisations

Withdrawals due to adverse effects

Six (n=3511) of the included studies reported WDAE (Analysis 2.9), which were significantly lower in patients treated with ARBs (RR 0.63 [95% CI 0.52, 0.76]; ARR=6.0%; NNT=17).

2.9 — Comparison 2: ARBs versus ACE inhibitors, Outcome 9: WDAE

ARB plus ACEI versus ACEI alone

Mortality

Seven (n=8260), 2 (n=7558), and 2 (n=7558) of the included studies reported total mortality, cardiovascular mortality, and non‐cardiovascular mortality, respectively (Analysis 3.1; Analysis 3.2; Analysis 3.3). With respect to total mortality, cardiovascular mortality, or non‐cardiovascular mortality, there were no differences between combination therapy and ACEI monotherapy groups.

3.1 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 1: Total mortality

3.2 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 2: Cardiovascular mortality

3.3 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 3: Non‐cardiovascular mortality

Morbidity

MI and stroke were reported in only 1 large‐scale study, CHARM‐Added 2003 (n=2548). There was no significant difference between treatment groups for stroke (Analysis 3.5), but MI was reduced with combination therapy compared with ACEI alone (RR 0.64 [95% CI 0.44, 0.92]; ARR=2.0%; NNT=50) [Analysis 3.4].

3.5 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 5: Stroke

3.4 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 4: MI

Hospitalisations

Two (n=2989), 4 (n=8108), and 2 (n=2989) of the included studies reported total hospitalisations, hospitalisations for HF, and hospitalisations for other causes, respectively (Analysis 3.6; Analysis 3.7; Analysis 3.8). Total hospitalisations were not reduced with ARBs compared with placebo. When this outcome was divided according to cause, combination therapy reduced the risk of hospital admissions for HF (RR 0.81 [95% CI 0.74, 0.89]; ARR=4.4%; NNT=23) but this benefit was offset by a non‐significant increase in the risk of hospital admissions for other causes (RR 1.07 [95% CI 0.98, 1.18]).

3.6 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 6: Total hospitalisations

3.7 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 7: Hospitalisations for heart failure

3.8 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 8: Other hospitalisations

Withdrawals due to adverse effects

Four (n=7703) of the included studies reported WDAE (Analysis 3.9). When the data from these four trials were pooled, WDAE were significantly higher in patients receiving combination therapy (RR 1.34 [95% CI 1.19, 1.51]; ARI=3.7%; NNH=27).

3.9 — Comparison 3: ARB plus ACEI versus ACEI alone, Outcome 9: WDAE

Discussion

Summary of main results

Jong 2002 included a total of 16 studies that randomly allocated 12,295 patients to either an ARB or control (placebo or ACEI). This review has allowed analysis of an increased number of patients from an additional eight studies published from 2001 to 2008, totaling 24 studies that enrolled 25,051 patients with symptomatic HF. A total of 17,900 patients with a LVEF ≤40% were studied in 22 studies (with a weighted mean follow‐up of 2.2 years in four large‐scale studies (CHARM‐Added 2003; CHARM‐Alternative 2003; ELITE II 2000; Val‐HeFT 2001) and 7151 patients with a LVEF >40% were studied in two large‐scale studies for a weighted mean duration of 3.7 years (CHARM‐Preserved 2003; I‐PRESERVE 2008).

ARBs versus placebo

Treatment with ARBs in symptomatic HF patients with a LVEF ≤40% did not significantly reduce all‐cause mortality, stroke, or all‐cause hospital admission as compared to placebo. CHARM‐Alternative 2003, a large‐scale trial in which HF patients who had demonstrated intolerance to ACE inhibitors were treated with candesartan or placebo for a median duration of 33.7 months, contributed 90% or more of the data to the outcome of hospital admission. In this trial, candesartan reduced the risk of hospital admission for HF (RR 0.73 [95% CI 0.62, 0.85]; ARR=7.8%; NNT=13) but increased the risk of hospital admission for other causes (RR 1.13 [95% CI 1.01, 1.27]; ARI=4.6%; NNH=22). Candesartan also increased the risk of MI (RR 1.57 [95% CI 1.10, 2.23]; ARI=2.7%; NNT=37) compared with placebo, which may partly explain the significant increase in other hospitalisations observed in this trial. A recently published systematic review evaluating the effect of ARBs in all major trials reporting MI did not demonstrate an increased risk of MI with ARBs as compared to other drugs or placebo (Volpe 2009). However, we cannot be certain all unpublished trials are accounted for and data not reported in existing published trials are needed to confirm the findings of this meta‐analysis.

There is no available RCT evidence that demonstrates that any individual ARB is more or less effective than another in the treatment of HF. In fact, the data included in this review showed no reduction in total mortality or total hospitalizations when candesartan or valsartan studies were pooled. Elevated blood pressure is in most cases the predominant contributor to the development and progression of HF. ARBs or any other antihypertensive agents have not been shown to provide any benefit beyond BP lowering. The evidence (46 studies, n=13,451) from a Cochrane review of the BP lowering efficacy of ARBs for primary hypertension suggests that there are no clinically meaningful BP lowering differences between available ARBs (Heran 2008b).

In symptomatic HF patients with a LVEF >40%, ARB therapy did not significantly reduce all‐cause mortality or all‐cause hospital admission compared with placebo. A statistically marginal reduction in hospitalisations for HF with ARBs was counteracted by a non‐significant increase in other hospitalisations. Rates of stroke and MI were not reported.

We pooled the data for all HF patients, irrespective of LVEF, to fully examine the effect of ARBs on hospital admission for other causes and there was a significantly increased risk of hospitalisations for other causes compared with placebo (RR 1.06 [95% CI 1.01, 1.12]; ARI=1.4%; NNH=72). We also pooled WDAE data from all placebo‐controlled studies and significantly more patients in the ARB group stopped treatment early due to adverse effects (RR 1.19 [95% CI 1.09, 1.30]; ARI=1.4%; NNH=72). These findings are in contrast with the widely held belief in the literature that ARBs have a safety and tolerability profile similar to that of placebo (Mancia 2009; Smith 2008; White 2011).

ARBs versus ACEIs

We pooled trials that compared ARBs with ACE inhibitors in symptomatic HF patients with a LVEF ≤40% and there appears to be no difference between the two drug classes in total mortality, cardiovascular mortality, stroke, MI, total hospitalisations, or hospitalisations for HF. ARBs were found to be superior to ACE inhibitors in tolerability with a significantly lower rate of withdrawals due to adverse effects (RR 0.63 [95% CI 0.52, 0.76]; ARR=6.0%; NNT=17).

ARB plus ACEI versus ACEI alone

In symptomatic HF patients with a LVEF ≤40%, the addition of an ARB to an ACE inhibitor was not effective in reducing total mortality or cardiovascular mortality compared with ACE inhibitor therapy alone. Combination therapy reduced hospitalisations for HF (RR 0.81 [95% CI 0.74, 0.89]; ARR=4.4%; NNT=23) but a non‐significant increase in hospital admissions for other causes (RR 1.07 [95% CI 0.98, 1.18]) was also observed, resulting in no significant difference between combination and ACE inhibitor monotherapy in all‐cause hospital admissions. The risk of stroke was not reduced but the risk of MI (RR 0.64 [95% CI 0.44, 0.92]; ARR=2.0%; NNT=50) was reduced with combination therapy. More patients receiving ARB plus ACE inhibitor combination had to stop treatment prematurely because of adverse effects (RR 1.34 [95% CI 1.19, 1.51]; ARI=3.7%; NNH=27).

Our search did not identify any studies that compared ARB plus ACE inhibitor combination therapy with ACE inhibitor alone in HF patients with a LVEF >40%.

Overall completeness and applicability of evidence

There are important differences between our review and previously published meta‐analyses (Lee 2004; Shah 2010; Sharma 2000; Shibata 2008), including the non‐Cochrane meta‐analysis published by the authors who originally developed this protocol (Jong 2002). The major limitation with these other reviews is that they have limited their analyses to total mortality and hospital admission for worsening HF. Our review also analysed other clinical outcomes, such as cardiovascular mortality, total hospitalisations, hospitalisations for other causes, stroke, MI, and withdrawals due to adverse effects, in an attempt to fully elucidate the benefit and harm of ARB treatment in HF patients based on published data from large‐scale clinical trials. In particular, all‐cause hospital admission is a useful measure of total morbidity and was reported in three of four large‐scale studies. Other useful indexes of morbidity would be total length of hospital admission or days in ICU or on a ventilator; however, these outcomes were not reported in any of the studies.

By expanding our analysis, we have observed some effects of ARBs that have not been reported in other meta‐analyses. For instance, the reduction in hospitalisations for HF with ARBs compared with placebo has been reported previously. However, the increase in hospitalisations for other causes with ARBs observed in our review has not been reported before. Given that in these trials the majority of total hospital admissions is to due to reasons other than worsening symptoms of HF, and the fact that total hospitalisations for any cause are not reduced, this suggests that there is no net health benefit of ARBs. If this is true the use of ARBs in this setting is called into question.

It is commonly accepted that ACEI reduce mortality and morbidity and therefore have become standard therapy in HF patients (ACCF/AHA 2009; CCS 2006; ESC 2008). We therefore decided to review the findings of a meta‐analysis of data from individual patients with HF or left‐ventricular dysfunction that is frequently cited to support that claim (Flather 2000). The authors of this systematic overview collected individual patient data from three large‐scale studies (n=5966) that enrolled acute MI patients (AIRE 1993; SAVE 1992; TRACE 1995) and two studies (n=6797) that enrolled patients with chronic HF (SOLVD treatment 1991) or LV dysfunction (SOLVD prevention 1992) who were randomly assigned to an ACEI or placebo and continued treatment for at least a year.

When the analysis was limited to the SOLVD treatment and prevention trials of enalapril in chronic HF, the ACEI group had lower rates of mortality (OR 0.87 [95% CI 0.78, 0.98]), reinfarction (OR 0.78 [95% CI 0.65, 0.92]) and readmission for HF (OR 0.63 [95% CI 0.56, 0.72]) than with placebo. The Flather 2000 meta‐analysis did not report total hospitalisations so we retrieved the published reports of SOLVD treatment 1991 and SOLVD prevention 1992 for these data. When these 2 studies were pooled, total hospitalisations (RR 0.96 [95% CI 0.92, 0.99]) and hospitalisations for HF (RR 0.69 [95% CI 0.63, 0.77]) were significantly reduced with enalapril (Figure 5; Figure 6). However, hospitalisations due to other causes were significantly increased (RR 1.10 [95% CI 1.04, 1.16]) with enalapril as compared to placebo (Figure 7), a finding consistent with what he have observed with ARBs.

Forest plot of ACE inhibitor enalapril versus placebo for total hospitalisations.

Forest plot of ACE inhibitor enalapril versus placebo for hospitalisations for heart failure.

Forest plot of ACE inhibitor enalapril versus placebo for other hospitalisations.

Given this information, the finding in the present review that ARBs are not effective in terms or mortality and morbidity versus placebo is surprising and inconsistent with the finding that ACE inhibitors and ARBs are not different when compared directly. We cannot completely resolve this inconsistency; however, it is possible that we may have missed a modest benefit of ARBs compared to placebo in patients with HF and reduced ejection fraction. The mortality effect for ARBs (RR 0.87 [95% CI 0.76, 1.00] versus ACE inhibitors (OR 0.87 [95% CI 0.78, 0.98] and the total hospitalisations effect for ARBs (RR 0.94 [95% CI 0.88, 1.01] versus ACE inhibitors (RR 0.96 [0.92, 0.99] is not necessarily that different.

Where other meta‐analyses also failed to include WDAE as a safety outcome, we pooled all studies that reported this outcome and observed a higher rate in patients taking ARBs as compared to placebo. We also observed a higher rate of WDAE in patients taking ARB plus ACEI combination therapy versus an ACE inhibitor alone.

Overall, given the lack of proven benefit with ARBs in terms of total mortality and total hospitalisations as compared to placebo and a proven benefit of ACE inhibitors as compared to placebo, it is clear that ACE inhibitor should be the first choice class of drugs in HF patients.

Quality of the evidence

This review has revealed several significant limitations in the available RCT evidence, most notably the poor reporting of methodology in many trial publications. The method of randomisation, allocation concealment, and blinding was rarely described. Although the quality of reporting tended to be better in long‐term health outcome studies, details on the methodology was still incomplete.

All but one of the 24 included studies were sponsored by the manufacturer making a high risk of publication bias and other biases likely. It was not possible for us to determine the existence of publication bias by analysing funnel plots for asymmetry due to an insufficient number of included studies. However, we were able to find evidence of unpublished studies during our search for relevant studies. Sharma 2000 was an industry‐sponsored meta‐analysis of losartan for HF that included 2 unpublished placebo‐controlled, multicenter studies. The full reports of these 2 studies have not been published elsewhere. Therefore, it is possible that our inability to identify unpublished studies with negative results may have led to overestimation of treatment effects in our review (Egger 1997).

Furthermore, this meta‐analysis may under‐estimate the harms associated with ARB therapy as a result of patient selection bias. In one of two large‐scale trials in HF patients with preserved systolic function, participants with a previously documented intolerance of ARBs were excluded (I‐PRESERVE 2008). This source of bias clearly influenced the outcome of WDAE since in I‐PRESERVE 2008 trial there was no increase in WDAE with irbesartan compared with placebo, yet in the other trial (CHARM‐Preserved 2003) WDAE were significantly higher with candesartan, which did not select for patients who were tolerant of ARBs. Overall, significantly more patients treated with ARBs discontinued prematurely due to an adverse effect (RR 1.22 [95% CI 1.09, 1.36]; ARI=3%; NNH=33).

Authors' conclusions

Implications for practice.

ARBs versus placebo

ARBs do not reduce total mortality or all‐cause hospitalisations compared with placebo in the treatment of HF patients irrespective of LVEF. In HF patients with LVEF ≤40%, the benefit observed with ARBs in terms of a reduction in hospitalisations for HF was mitigated by an increase in hospitalisations for other causes. More patients treated with ARBs stopped treatment early due to adverse effects compared with placebo.

ARBs versus ACE inhibitors

ARBs do not reduce total mortality or all‐cause hospitalisations compared with ACE inhibitors in the treatment of symptomatic HF patients with LVEF ≤40%. ARBs reduced WDAE as compared to ACE inhibitors.

ARB plus ACE inhibitor combination versus ACE inhibitor alone

Adding an ARB to an ACE inhibitor does not reduce total mortality or all‐cause hospitalisations over an ACE inhibitor alone in the treatment of HF patients with LVEF ≤40%. Adding an ARB to an ACE inhibitor increases WDAE as compared to an ACE inhibitor alone in the treatment of HF patients with LVEF ≤40%.

Implications for research.

In placebo‐controlled trials, the higher rate of hospitalisations for other causes in patients treated with ARBs is a key observation. The majority of total hospital admissions in the included studies is to due to reasons other than worsening symptoms of HF. This potentially harmful effect of ARBs needs to be investigated further by exploring the causes of hospital admission using individual patient data.

RCTs are needed comparing ARBs and ACE inhibitors in HF patients with LVEF >40%.

What's new

Date	Event	Description
9 February 2021	Review declared as stable	The one aspect of uncertainty within this topic is covered with a new review (CD012721). This review is therefore not planned to be updated.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Total mortality	11		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.93, 1.12]
1.1.2 Patients with LVEF ≤40%	9	4643	Risk Ratio (M‐H, Fixed, 95% CI)	0.87 [0.76, 1.00]
1.2 Cardiovascular mortality	6		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.2.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	1.02 [0.90, 1.14]
1.2.2 Patients with LVEF ≤40%	4	3382	Risk Ratio (M‐H, Fixed, 95% CI)	0.88 [0.75, 1.03]
1.3 Non‐cardiovascular mortality	6		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.3.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.85, 1.25]
1.3.2 Patients with LVEF ≤40%	4	3382	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.70, 1.54]
1.4 MI	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.4.1 Patients with LVEF >40%	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	Not estimable
1.4.2 Patients with LVEF ≤40%	2	2298	Risk Ratio (M‐H, Fixed, 95% CI)	1.44 [1.03, 2.01]
1.5 Stroke	2		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.5.1 Patients with LVEF >40%	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	Not estimable
1.5.2 Patients with LVEF ≤40%	2	2298	Risk Ratio (M‐H, Fixed, 95% CI)	0.83 [0.54, 1.27]
1.6 Total hospitalisations	4		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.6.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.97, 1.05]
1.6.2 Patients with LVEF ≤40%	2	2298	Risk Ratio (M‐H, Fixed, 95% CI)	0.94 [0.88, 1.01]
1.7 Hospitalisations for heart failure	5		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.7.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	0.90 [0.81, 1.00]
1.7.2 Patients with LVEF ≤40%	3	2590	Risk Ratio (M‐H, Fixed, 95% CI)	0.71 [0.61, 0.82]
1.8 Other hospitalisations	4	9449	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [1.01, 1.12]
1.8.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	1.05 [0.99, 1.11]
1.8.2 Patients with LVEF ≤40%	2	2298	Risk Ratio (M‐H, Fixed, 95% CI)	1.12 [1.00, 1.25]
1.9 WDAE	8	10917	Risk Ratio (M‐H, Fixed, 95% CI)	1.19 [1.09, 1.30]
1.9.1 Patients with LVEF >40%	2	7151	Risk Ratio (M‐H, Fixed, 95% CI)	1.22 [1.09, 1.36]
1.9.2 Patients with LVEF ≤40%	6	3766	Risk Ratio (M‐H, Fixed, 95% CI)	1.14 [0.97, 1.33]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Total mortality	8	5201	Risk Ratio (M‐H, Fixed, 95% CI)	1.05 [0.91, 1.22]
2.2 Cardiovascular mortality	4	4131	Risk Ratio (M‐H, Fixed, 95% CI)	1.08 [0.91, 1.28]
2.3 Non‐cardiovascular mortality	4	4131	Risk Ratio (M‐H, Fixed, 95% CI)	0.94 [0.66, 1.34]
2.4 MI	2	3874	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.62, 1.63]
2.5 Stroke	1	3152	Risk Ratio (M‐H, Fixed, 95% CI)	1.63 [0.77, 3.44]
2.6 Total hospitalisations	3	4310	Risk Ratio (M‐H, Fixed, 95% CI)	1.00 [0.92, 1.08]
2.7 Hospitalisations for heart failure	3	4310	Risk Ratio (M‐H, Fixed, 95% CI)	0.96 [0.83, 1.11]
2.8 Other hospitalisations	3	4310	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.92, 1.15]
2.9 WDAE	6	3511	Risk Ratio (M‐H, Random, 95% CI)	0.63 [0.52, 0.76]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
3.1 Total mortality	7	8260	Risk Ratio (M‐H, Fixed, 95% CI)	0.98 [0.90, 1.06]
3.2 Cardiovascular mortality	2	7558	Risk Ratio (M‐H, Fixed, 95% CI)	0.93 [0.84, 1.03]
3.3 Non‐cardiovascular mortality	2	7558	Risk Ratio (M‐H, Fixed, 95% CI)	1.05 [0.90, 1.23]
3.4 MI	1	2548	Risk Ratio (M‐H, Fixed, 95% CI)	0.64 [0.44, 0.92]
3.5 Stroke	1	2548	Risk Ratio (M‐H, Fixed, 95% CI)	1.14 [0.76, 1.72]
3.6 Total hospitalisations	2	2989	Risk Ratio (M‐H, Fixed, 95% CI)	0.99 [0.94, 1.05]
3.7 Hospitalisations for heart failure	4	8108	Risk Ratio (M‐H, Fixed, 95% CI)	0.81 [0.74, 0.89]
3.8 Other hospitalisations	2	2989	Risk Ratio (M‐H, Fixed, 95% CI)	1.07 [0.98, 1.18]
3.9 WDAE	4	7703	Risk Ratio (M‐H, Fixed, 95% CI)	1.34 [1.19, 1.51]

*Study characteristics*
Methods	Single‐center, prospective DBRCT (double blind randomised control trial) conducted in Glasgow, UK Follow‐up: 8 wks Background ACEI? Yes
Participants	N=36 patients ≥18 years of age with CHF, caused by IHD or nonischemic dilated cardiomyopathy, receiving ACEI therapy for ≥4 wks Mean age: 63.5 y Females: 19.4% NYHA Class: II–IV LVEF: ≤35%
Interventions	Eprosartan 200 mg BID to target 400 mg BID (n=18) Placebo (n=18)
Outcomes	Primary: LVEF Secondary: haemodynamics; neurohormones
Notes	Funding source: Smithkline Beecham Pharmaceuticals
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Reasons for withdrawal from study (6/36 patients) have been described.
Other bias	High risk	Funding source is manufacturer of eprosartan.

*Study characteristics*
Methods	Multicenter, prospective DBRCT conducted in Japan Follow‐up: 6 mos Background ACEI? No
Participants	N=305 patients ≥20 years of age with CHF due to previous MI, hypertensive heart disease, dilated cardiomyopathy or valvular disease N=292 patients included in full analysis Mean age: 63.7 y Females: 22.5% NYHA Class: II–III LVEF: ≤45%
Interventions	Candesartan 8 mg OD (n=155; n=148 included in full analysis) Placebo (n=150; n=144 included in full analysis)
Outcomes	Primary: Composite outcome, 'confirmed progression of CHF', which included the following: 1) hospitalisation for management of CHF; or 2) addition of, or increase in, any medication(s) administered specifically for management of CHF in response to apparent aggravation of its manifestations Secondary: occurrence of CV event, including progression of CHF, cardiac death, life‐threatening arrhythmias, MI, coronary artery disease (defined as angina, coronary artery intervention or revascularization), stroke and transient Ischaemic attack
Notes	Funding source: Takeda Chemical Industries, Ltd
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias) All outcomes	High risk	Missing outcome data, i.e. total hospitalisations Patients excluded from full analysis [13/305 patients; candesartan (n=7), placebo (n=6)] are described in Table 1.
Other bias	High risk	Patient selection bias: Prior to randomisation, patients received 4‐mg test dose of candesartan to confirm tolerance of single dose of study drug. Eight patients were not randomised after test dose due to adverse event caused by test dose (n=7), such as marked hypotension, and one patient exited study immediately after having received test dose. Funding source is manufacturer of candesartan.

*Study characteristics*
Methods	Multinational, multicenter, prospective DBRCT Follow‐up: 3 y (mean); 41 mos (median) Background ACEI? Yes
Participants	N=2548 patients ≥18 years of age with symptomatic CHF, due to IHD (62.4%), idiopathic (26.2%), hypertensive cause (6.5%), or other cause (4.9%), and treated with ACEI at constant dose for ≥30 days Mean age: 64.1 y Females: 21.3% Race (white/black/other): 91%/5%/4% NYHA Class: II–IV (if class II, patients had to have admission to hospital for cardiac reason in previous 6 mos LVEF: ≤40%
Interventions	Candesartan, target dose 32 mg OD (n=1276) Placebo (n=1272)
Outcomes	Primary: CV death or unplanned admission to hospital for management of worsening CHF Secondary: CV death, hospital admission for CHF, or non‐fatal MI; CV death, hospital admission for CHF, non‐fatal MI, or non‐fatal stroke; CV death, hospital admission for CHF, non‐fatal MI, non‐fatal stroke, or coronary revascularization; all‐cause mortality or hospital admission for CHF; development of new diabetes
Notes	Funding source: AstraZeneca R&D
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Low risk	Quote: "We randomly assigned patients, in a double‐blind way, candesartan or matching placebo, which could be started at 4 or 8 mg once daily (figure 1), the assignment code being held at an independent centre and by the data safety monitoring board."
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "We randomly assigned patients, in a double‐blind way, candesartan or matching placebo, which could be started at 4 or 8 mg once daily (figure 1), the assignment code being held at an independent centre and by the data safety monitoring board."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "The analysis was done on an intention‐to‐treat basis and included all randomised patients."
Other bias	High risk	Funding source is manufacturer of candesartan.

*Study characteristics*
Methods	Prospective DBRCT conducted at 4 centers in US and France Follow‐up: 6 mos Background ACEI? Yes
Participants	N=33 patients with severe symptomatic CHF, due to IHD (30%) or non‐ischemic heart disease (70%), despite treatment with maximally recommended or tolerated doses of ACEI for ≥3 months Mean age: 60.8 y Females: 51.5% NYHA Class: III–IV
Interventions	Losartan 50 mg OD (n=16) Placebo (n=17)
Outcomes	Primary: Peak VO₂; NYHA functional class Secondary: Laboratory safety parameters; doses of concomitant background medications
Notes	Funding source not reported.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All randomised patients were included in safety analysis.
Other bias	High risk	Patient selection bias: All patients tolerated losartan 50 mg/day during single‐blind tolerability phase.

*Study characteristics*
Methods	Multicenter, prospective DBRCT in Sweden Follow‐up: 12 wks Background ACEI? Yes
Participants	N=141 patients ≥18 years of age with symptomatic CHF, due to IHD (61%) or non‐ischemic heart disease (39%), stabilized on an ACEI for ≥3 mos and able to perform 6‐minute walk test Mean age: 67.5 y Females: 25.5% NYHA Class: II–III LVEF: ≤45%
Interventions	Valsartan 160 mg OD (n=70) Enalapril 5 mg BID (n=71)
Outcomes	Primary: exercise capacity (6‐minute walk test) Secondary: QoL (Minnesota Living with Heart Failure Questionnaire); LVEF; dyspnea‐fatigue index score; left ventricular end diastolic diameter
Notes	Funding source: Novartis Pharma
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "double‐dummy"
Incomplete outcome data (attrition bias) All outcomes	Low risk	All randomised patients were included in safety analysis.
Other bias	High risk	Funding source is manufacturer of valsartan.

*Study characteristics*
Methods	Multicenter (15 centers from US and 1 from Canada), prospective DBRCT Follow‐up: 12 wks Background ACEI? No
Participants	N=116 patients with symptomatic CHF, primarily due to IHD (47%) or dilated cardiomyopathy (41%), previously treated with stable doses of ACEI and diuretic agents for a minimum of 6 weeks and 2 weeks, respectively, with or without concurrent digitalis and other vasodilators. Mean age: 57.8 y Females: 22% Race (white/black/hispanic/oriental): 71%/22%/5%/3% NYHA Class: II–IV LVEF: ≤45%
Interventions	Losartan 25 mg OD (n=38) Losartan 50 mg OD (n=40) Enalapril 10 mg BID (n=38)
Outcomes	Primary: Symptom‐limited treadmill exercise duration; 6‐minute walk test; dyspnea‐fatigue index; signs and symptoms of heart failure and functional class Secondary: Clinical and laboratory adverse events
Notes	Funding source: Merck Research Laboratories Study "Phase II‐US" in Sharma 2000 meta‐analysis
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All randomised patients were included in safety analysis.
Other bias	High risk	Funding source is manufacturer of losartan.

*Study characteristics*
Methods	Multicenter, prospective DBRCT in Russia Follow‐up: 4 wks Background ACEI? No
Participants	N=116 patients aged 18‐80 years with stable CHF previously untreated with ACEI and who had pulmonary capillary wedge pressure ≥15 mm Hg Mean age: 56.0 y Females: 17.2% Race: All white NYHA Class: II–IV
Interventions	Valsartan 40 mg BID (n=24) Valsartan 80 mg BID (n=24) Valsartan 160 mg BID (n=27) Lisinopril 5 mg OD for 7 days followed by 10 mg OD for 3 wks (n=15) Placebo (n=26)
Outcomes	Primary: Change from baseline in mean pulmonary capillary wedge pressure Secondary: Changes from baseline in cardiac output and systemic vascular resistance
Notes	Funding source: Novartis Pharma
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "Patients in the lisinopril group received active drug plus placebo, in order to have a twice daily regimen identical to that of valsartan. To maintain blindness, a matching technique was used so all three medications (valsartan, lisinopril, placebo) were identical."
Incomplete outcome data (attrition bias) All outcomes	Low risk	103/116 patients completed double‐blind treatment period; all randomised patients were included in safety analysis.
Other bias	High risk	Funding source is manufacturer of valsartan.

*Study characteristics*
Methods	Multicenter, prospective DBRCT in Europe and South Africa Follow‐up: 12 wks Background ACEI? No
Participants	N=218 patients aged 18‐75 years with symptomatic CHF (principal causes were coronary heart disease and/or cardiomyopathy) and pulmonary capillary wedge pressure ≥13 mm Hg at baseline Mean age: 54.0 y Females: 14.7% Race: All white NYHA Class: II–III LVEF: ≤40%
Interventions	Candesartan 2 mg OD (n=45) Candesartan 4 mg OD (n=46) Candesartan 8 mg OD (n=39) Candesartan 16 mg OD (n=44) Placebo (n=44)
Outcomes	Primary: pulmonary capillary wedge pressure; systemic vascular resistance; cardiac index Secondary: pulmonary arterial pressure; mean right arterial pressure; mean arterial blood pressure; heart rate; neurohormones; symptom scores for breathlessness, fatigue and ankle swelling; physicians' overall efficacy score; NYHA classification; QoL using validated measure of subjective health status SF‐36 questionnaire
Notes	Funding source: Takeda Europe R&D Centre Ltd
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias) All outcomes	Low risk	189/218 patients completed double‐blind treatment period; all randomised patients were included in safety analysis.
Other bias	High risk	Funding source is manufacturer of candesartan.

*Study characteristics*
Methods	Multicenter, prospective DBRCT Follow‐up: 12 wks Background ACEI? No
Participants	N=385 patients with symptomatic CHF who had never received ACEI or who had discounted ACEI for ≥12 wks Mean age: 60.3 y Females: 30% Race (white/hispanic/black/other): 59%/31%/5%/5% NYHA Class: II–IV LVEF: ≤40%
Interventions	Losartan, target dose 50 mg (n=254) Placebo (n=131)
Outcomes	Primary: Exercise capacity (maximal treadmill exercise time)
Notes	Funding source: Merck Research Laboratories Phase IIII‐Int'l study information and results reported in retrospective meta‐analysis conducted and published by manufacturer.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "...patients were randomised to losartan or matching placebo in a 2:1 ratio."
Incomplete outcome data (attrition bias) All outcomes	Low risk	All randomised patients were included in safety analysis.
Other bias	High risk	Funding source is manufacturer of losartan.

*Study characteristics*
Methods	Multicenter, prospective DBRCT in US Follow‐up: 12 wks Background ACEI? No
Participants	N=351 patients with symptomatic CHF who had never received ACEI or who had discounted ACEI for ≥6 wks Mean age: 65.0 y Females: 33% Race (white/hispanic/black/other): 83%/1%/15%/1% NYHA Class: II–IV LVEF: ≤40%
Interventions	Losartan, target dose 50 mg (n=237) Placebo (n=114)
Outcomes	Primary: Exercise capacity (maximal treadmill exercise time)
Notes	Funding source: Merck Research Laboratories Study information and results reported in retrospective meta‐analysis conducted and published by manufacturer
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not described.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding (performance bias and detection bias) All outcomes	Low risk	Quote: "...patients were randomised to losartan or matching placebo in a 2:1 ratio."
Incomplete outcome data (attrition bias) All outcomes	Low risk	All randomised patients were included in safety analysis.
Other bias	High risk	Funding source is manufacturer of losartan.

Study	Reason for exclusion
Belousov 2005	No useful outcomes reported.
Berezin 2001	No useful outcomes reported.
Blanchet 2005	No useful outcomes reported.
De Tommasi 2003	No useful outcomes reported.
Ellis 2002	Not double‐blind.
Guazzi 1999	Crossover study.
HEAAL 2009	No parallel control (placebo or ACEI) arm.
Hikosaka 2002	Not double‐blind. No useful outcomes reported.
Houghton 2000	No useful outcomes reported.
Kasama 2003	Not double‐blind. No useful outcomes reported.
Kasama 2005	No useful outcomes reported.
Ki 2008	No useful outcomes reported.
Nakamura 2002	All patients were not receiving background ACEI therapy. No useful outcomes reported.
Parthasarathy 2009	All patients were not receiving background ACEI therapy.
SADKO‐CHF 2005	No useful outcomes reported.
Tang 2003	No useful outcomes reported.
Tsutamoto 2000	Not double‐blind. No useful outcomes reported.
Vaile 2001	Crossover study.
Vizir 2000	Not double‐blind.
Vizir 2002	Not randomised; control arm comprised patients who refused therapy with ARB instead of being randomised to appropriate control.
Ye 2005	No useful outcomes reported. Non‐equivalent doses studied; benazepril 20 mg vs. benazepril 10 mg plus valsartan 80 mg combination vs. valsartan 160 mg.

PERMALINK

Angiotensin receptor blockers for heart failure

Balraj S Heran

Vijaya M Musini

Ken Bassett

Rod S Taylor

James M Wright

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Background

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary:

Secondary:

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Subgroup analysis

Heterogeneity

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Patients with LVEF >40%

Patients with LVEF ≤40%

ARBs versus placebo

ARBs versus ACEIs

ARB plus ACEI combination versus ACEI alone

Excluded studies

Risk of bias in included studies

2.

3.

Allocation

Blinding

Incomplete outcome data

Selective reporting

Other potential sources of bias

Selection bias

1.9. Analysis.

Publication bias

4.

Effects of interventions

ARBs versus placebo

Mortality

Patients with LVEF >40%

1.1. Analysis.

1.2. Analysis.

1.3. Analysis.

Patients with LVEF ≤40%

Morbidity

Patients with LVEF >40%

Patients with LVEF ≤40%

1.4. Analysis.

1.5. Analysis.

Hospitalisations