Beta‐blockers for heart failure

Abstract

This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:

To assess the benefits and harms of beta‐blockers for adults heart failure.

Background

Description of the condition

Epidemiology

Cardiovascular diseases are the leading cause of death and account for an estimated 30% of all deaths worldwide (GBD 2015). Heart failure is the most rapidly growing cardiovascular condition in the world (Fonseca 2006; Heidenreich 2013; Ziaeian 2016), with an estimated prevalence of over 37.7 million people worldwide in 2010 (Vos 2012; Ziaeian 2016). The prevalence of heart failure in the USA alone exceeds 5.8 million people and over 550,000 new patients are diagnosed each year (Bui 2011; Levy 2006). It is estimated that by year 2030 more than eight million people (one in every 33) will be diagnosed with heart failure in the USA (Heidenreich 2013), with similar increasing trends in Asia (Sato 2015) and Europe (Maggioni 2015; Schocken 2008; Stewart 2003).

The growing prevalence of heart failure is thought to reflect a combination of population ageing, prolongation of the lives of cardiac patients by modern beneficial therapies, and epidemiological changes in the diseases underlying heart failure (Bui 2011; Dunlay 2014; Kannel 2000; Mahmood 2013; Ramani 2010). The prevalence also seems to be affected by ethnicity (Bui 2011; Heidenreich 2013; Rosamond 2008), sex (Heidenreich 2013; Levy 2002; Rosamond 2008), socioeconomic status (He 2001; Heidenreich 2013), and lifestyle (Artham 2008; Schocken 2008). Heart failure seems to be more prevalent in men compared to women, African‐Americans, and people without a high school education (Bahrami 2008; Bibbins‐Domingo 2009; Loehr 2008). The incidence of heart failure increases with age (Bleumink 2004; Bui 2011; Roger 2013; Sato 2015; Yancy 2013), and at 40 years of age the lifetime risk of developing heart failure is one in five (Bui 2011; Lloyd‐Jones 2002; Stewart 2003; Weir 2016). Nevertheless, most evidence indicates that the incidence of heart failure has plateaued and might even be decreasing in some groups, e.g. women and younger individuals. This is despite the growing prevalence, in part due to improved survival of patients living with heart failure (Bui 2011; Levy 2002; Roger 2004; Stewart 2003).

Heart failure represents a considerable burden to healthcare systems, costing over USD 30 billion annually in the USA alone or about 2% of the healthcare budget, and the costs are expected to increase to about USD 70 billion in 2030 in the USA (Cook 2014; Heidenreich 2013; Lloyd‐Jones 2010; Mazurek 2015). The total global heart failure costs in 2012 were estimated to be USD 108 billion per annum (Cook 2014), and accounts for approximately 1% to 2% of direct healthcare expenditure in Western industrialised countries (Neumann 2009). These high costs are partly due to high rates of hospitalisations, readmissions, and outpatient visits (Bui 2011; Jessup 2003; Ramani 2010; Schocken 2008).

Definition of heart failure

Heart failure is a clinical syndrome that arises from a multitude of cardiac diseases (Bui 2011; Jessup 2003; Yancy 2013). The American College of Cardiology Foundation (ACCF) and the American Heart Association (AHA) define heart failure as “a complex clinical syndrome that results from any structural or functional impairment of ventricular filling or ejection of blood” (Hunt 2005; Yancy 2013). For practical purposes, guidelines define heart failure as a clinical syndrome in which signs and symptoms include dyspnoea, fatigue, fluid retention, pulmonary congestion, and peripheral edema (Heidenreich 2013; Hunt 2005). Patients with heart failure often have elevated jugular venous pressure, hepatic enlargement, pulmonary crackles, a ventricular gallop, and a displaced apex beat (Fonseca 2006; McMurray 2012).

Heart failure can be classified according to symptoms or by left ventricular function, quantified by the proportion of blood ejected each heart beat (known as the ejection fraction). The most commonly used methods for categorising the severity of heart failure symptoms are either the New York Heart Association (NYHA) functional classification (Table 1; Dolgin 1994; Yancy 2013), or the ACCF/AHA staging system (Table 2; Hunt 2009; Yancy 2013). The NYHA classification focuses on exercise capacity and the symptomatic status of the patient (Table 1). The NYHA system assigns patients to one of four functional classes. The class is determined by the degree of effort needed to elicit symptoms ranging from a lack of symptoms during ordinary activity (class I) to the inability to perform any physical activity without symptoms and with symptoms at rest (class IV) (Table 1; Hunt 2005; Mazurek 2015). The ACCF/AHA stages of heart failure emphasise the development and progression of the disease and allows for preventive and treatment recommendations that are stage‐specific (Table 2). The ACCF/AHA encompasses four sequential stages of heart failure in which stages A and B, defined as asymptomatic, are considered as precursors to heart failure. Stages C and D represent the symptomatic phases of heart failure. In stage D patients develop marked symptoms at rest or with minimal activity despite optimal medical therapy (Heidenreich 2013). Most heart failure therapeutic interventions are targeted in patients with symptomatic heart failure in NYHA stage III (less than ordinary activity causes fatigue, palpitation, or dyspnoea), NYHA stage IV, ACCF/AHA stage C, or ACCF/AHA stage D.

Table 1.

NYHA classification

Stage	Description
I	No limitation of physical activity. Ordinary physical activity does not cause symptoms of HF.
II	Slight limitation of physical activity. Comfortable at rest, but ordinary physical activity results in symptoms of HF.
III	Marked limitation of physical activity. Comfortable at rest, but less than ordinary activity causes symptoms of HF.
IV	Unable to carry on any physical activity without symptoms of HF, or symptoms of HF at rest.

Abbreviations: HF: heart failure; NYHA: New York Heart Association Functional Classification.

Table 2.

ACCF/AHA stages of HF

Stage	Description
A	At high risk for HF but without structural heart disease or symptoms of HF
B	Structural heart disease but without signs or symptoms of HF
C	Structural heart disease with prior or current symptoms of HF
D	Refractory HF requiring specialized interventions

Abbreviations: HF: heart failure; ACCF/AHA: American College of Cardiology Foundation (ACCF) and the American Heart Association (AHA).

The heart failure guidelines differentiate between three types of heart failure depending on the level of the left ventricular ejection fraction (LVEF) (Yancy 2013; McMurray 2012; Ponikowski 2016).

• Heart failure with an LVEF of 40% or less is characterised as heart failure with reduced ejection fraction (HErEF).

• Heart failure with an LVEF of 50% or more is characterised as heart failure with preserved ejection fraction (HFpEF).

• If the LVEF ranges from 41% to 49%, it is characterised as heart failure with mid‐range ejection fraction (HFmrEF).

It is estimated that HRrEF and HFpEF each represent 40% to 45% of the patients with heart failure, while HFmrEF has a prevalence of 10% to 20% of heart failure patients (Mazurek 2015). However, these three groups exhibit different demographics and are associated with different proportions of co‐morbidities. In addition, whereas a number of therapies have been shown to reduce morbidity and mortality in HFrEF, the evidence of therapeutic benefit in HFpEF is much scarcer (Jessup 2003; McMurray 2012; Owan 2006). Studies have shown that patients with preserved ejection fraction are older, often female, are more likely to be obese, and usually have hypertension and diabetes (Jessup 2003). The characteristics, treatment patterns, and outcomes in patients diagnosed with HFmrEF seem to have features in common with both populations of patients with HFrEF and HFpEF but tend to have fewer clinical manifestations of heart failure compared to HFREF and HFpEF (Gómez‐Otero 2017; Nadruz 2016; Tsuji 2017).

Congestive heart failure describes acute or chronic heart failure with evidence of congestion (i.e. sodium and water retention) (McMurray 2012). Right ventricular heart failure is a hydraulic problem caused by impaired function of the pump, the valves, or the vessels and often occurring as a result of decompensated left ventricular heart failure (Hrymak 2017; Kapur 2017; Voelkel 2006).

Pathophysiology

Heart failure may be viewed as the final common stage of many diseases of the heart (Bui 2011; Jessup 2003). It may result from disorders of the pericardium (e.g. restrictive cardiomyopathy or chronic pericardial disease) (Garcia 2016; Gentry 2016), myocardium (e.g. idiopathic dilated cardiomyopathy) (Felker 2000), endocardium (e.g. infectious endocarditis) (Htwe 2012), cardiac valves (e.g. aortic stenosis or mitral regurgitation) (Jessup 2003), vasculature (e.g. ischaemic heart disease or hypertension) (Felker 2000), or from certain metabolic abnormalities (e.g. myocardial lipotoxicity or myocardial lipid‐deposition) (Felker 2000; Heidenreich 2013; Hunt 2005; Mazurek 2015; Yancy 2013). Ischaemic heart disease, valvular disease, hypertension, and dilated cardiomyopathy serve as the main causes of heart failure in most patients (He 2001; Hunt 2005; McMurray 2012; Ramani 2010).

The sympathetic nervous system activation in heart failure has many theoretical deleterious effects, including vasoconstriction, increased left ventricular afterload, cardiac remodeling, and fibrosis (Ferrario 2006; Jessup 2003; Mann 2005; Mazurek 2015). Left ventricular dysfunction can be triggered by an acute (e.g. myocardial infarction) or a chronic (e.g. hypertension or valvular heart disease) insult to the myocardium resulting in a series of neurohormonal responses (Hunt 2009; Kaye 2007; Schocken 2008). Despite not necessarily being subjected to repeatable injuries, left ventricular dysfunction is still viewed as a progressive condition (Hunt 2009; Kaye 2007). This progression results in a change in the structure of the left ventricle known as cardiac remodeling, which may lead to dilation and hypertrophy (Jessup 2003; Kaye 2007; Mazurek 2015; Schocken 2008). A neurohormonal activity, led by the sympathetic nervous system and neurohormonal cascade (mainly the renin‐angiotensin‐aldosterone system (RAAS), endothelin, neprilysin, and various inflammatory cytokines), is thought to be the root to these deleterious effects (Ferrario 2006; Mann 2005; Mazurek 2015). In vitro studies and animal studies suggest that the mechanism of action is through sodium retention, peripheral vasoconstriction, abnormalities in cellular signalling, increased myocyte apoptosis, necrosis, inflammation, direct toxic effects on cardiac cells, and initiation of myocardial fibrosis (Aukrust 2005; Hunt 2005; Mann 1992; Mazurek 2015; Moorjani 2009; Weber 1994). Cardiac remodelling seems to cause decreased contractility of the myocardium, decreased stroke volume, increased haemodynamic stress on the walls of the heart, and increased regurgitant flow through the mitral valve (Hunt 2005; Ramani 2010).

Diagnosis

It is crucial to identify the underlying cardiac pathophysiology, as the underlying cause of heart failure might guide the physician when choosing which specific intervention to use (e.g. valve surgery for valvular disease, specific pharmacological therapy for left ventricular systolic dysfunction, etc.) (McMurray 2012). The diagnosis of heart failure relies on a physical examination, where signs from examination and symptoms from the history of congestion and end‐organ hypoperfusion are used to make the diagnosis (Hunt 2005; Ramani 2010). Diagnostic tests such as electrocardiography, echocardiogram, and biomarkers, most notably B‐type natriuretic peptide (BNP) or N‐terminal‐proBNP, are considered helpful adjuncts (Erbel 1984; Hunt 2005; McMurray 2012; Ramani 2010; Yancy 2013). Two‐dimensional echocardiogram coupled with Doppler flow studies have been shown to be useful to determine whether abnormalities are restricted to myocardium, pericardium, or heart valves (Hunt 2009; McMurray 2012). However, cardiovascular magnetic resonance is now a rapidly growing noninvasive imaging technique to assess myocardial anatomy, regional and global function, tissue characterisation, and viability (Karamitsos 2009).

Prognosis

The risk of death in patients with heart failure is high, with an estimated annual mortality of approximately 21% in men and 17% in women; the risk increases to over 50% five years after the heart failure diagnosis has been made (Bui 2011; Dunlay 2014; Kannel 2000; Levy 2002; Mozaffarian 2016; Roger 2004; Roger 2013). Based on the Framingham Heart Study, 30‐day mortality is approximately 10%, one‐year mortality is 20% to 30%, and five‐year mortality is 45% to 60% from the time of the heart failure diagnosis (Levy 2002; Ziaeian 2016). The five‐year mortality is more than 75% after the first hospitalisation for heart failure (Goldberg 2007). However, several studies have indicated improvements in mortality since the late 1990s e.g. a study on a community‐based population showed that survival improved over a five‐year period from 43% during 1979 to 1984 to 52% during 1996 to 2000 (Barker 2006; Jhund 2009; Kotecha 2016; Levy 2002; Roger 2004; Yeung 2012).

Description of the intervention

The recommended pharmacological management for patients with heart failure depends on the severity (defined by the above‐mentioned stages (see Table 2)) and often includes a combination of one or more of: an angiotensin converting enzyme (ACE) inhibitor, angiotensin II receptor blocker (ARB), beta‐blockers, angiotensin receptor‐neprilysin inhibitor (ARNI), mineralocorticoid receptor antagonist (MRA), digitalis, diuretics, hydralazines, and nitrates (Hunt 2009; McMurray 2012).

The beta‐receptor is an adrenergic Gs heterotrimeric G‐protein coupled receptor located throughout the body (Bristow 2011). There are three types of beta‐receptors: beta₁, beta₂, and beta₃.

The beta₁‐receptor is mainly located in the heart, where it induces positive chronotropic effects (increases heart rate) and positive inotropic effects (increases contractility of the myocardium). In the kidneys, activation of the beta₁‐receptor results in an increased release of renin which in turn increases blood pressure, among other effects (Golan 2011; Marlin 1975; Singh 1975). The beta₂‐receptor is mainly located in smooth muscle cells, where it promotes relaxation; in skeletal muscle cells, where it promotes tremor and increased glycogenolysis; and in the liver, where it increases glycogenolysis (Bristow 2011; Golan 2011). The beta₃‐receptor is mainly located in adipose tissue where it primarily induces lipolysis (Bristow 2011; Golan 2011).

The beta‐receptors are stimulated by the sympathetic nervous system with catecholamines epinephrine (adrenaline) and norepinephrine (noradrenaline) as their primary endogenous agonists resulting in an increase in Ca₂₊ concentration through cAMP/PKA‐dependent pathway (Baker 2014; Bristow 2011; Golan 2011). Stimulation of beta receptors by the sympathetic nervous system leads to increased heart rate and energy demands, adverse remodelling, interstitial fibrosis, arrhythmia provocation, and RAAS activation (Chatterjee 2002; McMurray 2012). Inhibition of adrenergic activity with the use of beta‐blockers is associated with attenuation of all these effects (Baker 2014; Chatterjee 2002; Farrell 2002; McMurray 2012).

Beta‐blockers are classified as non‐selective beta‐blockers or selective beta‐blockers according to their selectivity for one of the three subtypes of beta‐receptors (Golan 2011). Beta‐blockers may be administered both intravenously and orally. Three different classes of beta‐blockers exist and are as follows.

• The first generation non‐selective beta‐blockers (e.g. propanolol, oxprenolol, sotalol, and timolol).

• The second generation selective beta₁‐blockers (e.g. metoprolol, bisoprolol, acebutolol, atenolol, and esmolol).

• The third generation beta‐blockers which are combined alpha‐blocking and non‐selective beta‐blocking (e.g. carvedilol) (Golan 2011).

Beta‐blockers used to be contraindicated in patients with heart failure (Bristow 2000; Bristow 2011; Nayler 1969), but selective beta₁‐blockers or non‐selective combined alpha‐ and beta‐blockers are now a part of the standard treatment of heart failure since a number of trials showed beneficial effects (Chatterjee 2013; Dargie 1999; Hjalmarson 1999; Packer 2001; Ponikowski 2016; Yancy 2013). The ACCE/AHA as well as the European Society of Cardiology guidelines specifically recommend metoprolol succinate, bisoprolol, or carvedilol for treatment of heart failure (Ponikowski 2016; Yancy 2013).

The adverse effects of beta‐blockers are both cardiac adverse effects and non‐cardiac adverse effects (Kiel 2015). Among the most serious cardiac adverse effects are symptomatic hypotension, unacceptable fatigue, and exacerbation of heart failure in patients with acute decompensated heart failure (Barron 2013; Chatterjee 2013; Taylor 1982).

Case‐studies have suggested that depression, fatigue, dizziness, bradycardia, and sexual dysfunction are among the beta‐blocker induced non‐cardiac adverse effects (Greenblatt 1974; Ko 2004; Waal 1967; Warren 1977). However, a meta‐analysis comparing beta‐blockers versus placebo showed no difference on depressive symptoms and only a minor increase in sexual dysfunction and fatigue in patients randomised to beta‐blockers compared with placebo (Ko 2002).

Beta‐blockers for other conditions

Beta‐blockers are also considered an option in the treatment of hypertension, but are rarely used as first‐line treatment (Mancia 2013). A recent Cochrane Review found that beta‐blockers were inferior when compared with other anti‐hypertensive drugs (Wiysonge 2012). Non‐selective beta‐blockers are used in the treatment of anxiety due to their effect on decreasing tremor and tachycardia (Turner 1994).

Beta‐blockers used to be contraindicated in patients with chronic obstructive pulmonary disease (COPD). However, cardioselective beta‐blockers have been shown to reduce mortality and COPD exacerbations even in COPD patients without cardiovascular disease (Short 2011).

Perioperative beta‐blockers have been used in patients having non‐cardiac surgery. However, systematic reviews have shown that evidence does not support the use of beta‐blockers for the prevention of perioperative clinical outcomes in patients having non‐cardiac surgery (Bangalore 2008; Blessberger 2014; Cole 2014). Perioperative beta‐blockers are still recommended for patients undergoing cardiac surgery (Blessberger 2014).

How the intervention might work

The theoretical justification of using beta‐blockers in patients with heart failure rests on their inhibition of the chronotropic and inotropic effects of the beta₁‐receptors, a possible effect on regression in myocardial mass, and a possible normalisation in ventricular shape. Beta‐blockers are thought to reverse cardiac remodelling (Jessup 2003). Beta‐blockers are also thought to provide heart rate control in those with atrial fibrillation and anginal relief in patients with ischaemic heart disease (McMurray 2012; Hunt 2009).

Why it is important to do this review

Heart failure is the most rapidly growing cardiovascular condition globally (Fonseca 2006; Heidenreich 2013; Ziaeian 2016). Substantial increases in heart failure prevalence are expected in the near future (Heidenreich 2013; Schocken 2008). An analysis from Scotland predicted that if the current trends in heart failure prevalence and mortality persists, an increase in prevalence of 17% in women and 31% in men will exist by the year 2020 (Stewart 2003). If the above‐mentioned changes become reality, heart failure is likely to have an even more profound economic impact with a subsequent rise in hospitalisations and outpatient visits (Heidenreich 2013; Stewart 2003; Vos 2012). An optimal treatment might reduce morbidity and mortality associated with heart failure.

Current guidelines from ACCF/AHA recommend the use of beta‐blockers in all patients with reduced LVEF and heart failure corresponding to stage B or C, or both (ACCF/AHA) regardless of whether they are with or without a history of myocardial infarction (see Table 2) (Dunlay 2014; Hunt 2009; Smith 2011; Yancy 2013), or in patients classified with mild to moderate heart failure (class II‐IV) according to NYHA (see Table 1) (Chatterjee 2002; Hunt 2009; McMurray 2012; Yancy 2013).

Several randomised clinical trials that assessed the long‐term effects of beta‐blockers showed an improvement in systolic function and a reversal of cardiac remodeling (Bristow 2000; Groenning 2000). These trials suggest that adding beta‐blockers to conventional treatment may result in an approximately 24% to 35% relative risk reduction in mortality, may improve heart failure symptoms, and may reduce the risk of heart failure hospitalisations regardless of age and sex (Dargie 1999; Hjalmarson 1999; Kotecha 2016; Mazurek 2015; Packer 2001). Beta‐blockers may also reduce the risk of arrhythmia, improve LVEF, improve symptoms of heart failure, and may control ventricular rate (Chatterjee 2013; Dargie 2001).

Several existing meta‐analyses have compared the effects of beta‐blockers versus placebo or no intervention in participants with heart failure. These meta‐analyses showed that beta‐blockers were associated with a decreased risk of death, improved NYHA class, decreased hospitalisation rates, and improved some types of exercise tolerance tests (Abdulla 2006; Al‐Gobari 2013; Brophy 2001; Chatterjee 2013; Doughty 1997; Heidenreich 1997; Kotecha 2016; Lechat 1998; McAlister 2009). However, these meta‐analyses did not systematically assess the following

• The risks of systematic errors (bias), design errors, and random errors (play of chance) (Higgins 2011a; Keus 2010; Jakobsen 2014; Thorlund 2011).

• The outcome quality of life (QoL) (Jakobsen 2014; Garattini 2016).

• Trials irrespective of outcome, follow‐up duration, number of participants.

• Outcomes at several time points and take into account the variability in follow‐up.

• The validity of the evidence with GRADE (Guyatt 2008).

This Cochrane Review will be the first to use Cochrane methodology to assess the effects of beta‐blockers in patients with heart failure.

Objectives

To assess the benefits and harms of beta‐blockers for adults heart failure.

Methods

Criteria for considering studies for this review

Types of studies

Randomised clinical trials (both individual and cluster‐randomised trials) irrespective of publication type, publication status, publication date, and language.

Types of participants

We will include all adults (≥ 18 years of age) participants with any diagnosis of heart failure (as defined by the trial authors).

Types of interventions

Experimental group

We will include any type of beta‐blockers, both administered as intravenous therapy and as oral administration, as experimental intervention (non‐selective beta‐blockers (propranolol, oxprenolol, sotalol, timolol); selective beta₁‐blockers (metoprolol, bisoprolol, acebutolol, atenolol, esmolol); and beta‐blockers that are combined alpha‐ and non‐selective beta‐blockers (carvedilol)).

Control group

We will accept placebo, usual care, or no intervention as control interventions.

Co‐interventions

We will accept any co‐intervention provided they are intended to be delivered similarly to the experimental and the control group. Assuming no interaction, the effects of the co‐interventions will ‘even out’ in both groups so the possible effects of beta‐blockers will be reflected in the results.

Types of outcome measures

Primary outcomes

• All‐cause mortality.

• Serious adverse event defined as any untoward medical occurrence that: resulted in death, was life threatening, was persistent, led to significant disability, jeopardised the participant, led to hospitalisation or prolonged hospitalisation (ICH‐GCP 1997).

• Major adverse cardiovascular event defined as a composite outcome consisting of either cardiovascular mortality (defined by trial authors) or myocardial infarction (defined by trial authors).

Secondary outcomes

• Quality of life measured on any valid scale, such as the Short Form Health Survey (SF‐36) (Ware 1992).

• Cardiovascular mortality.

• Myocardial infarction.

• Hospitalisation for any cause.

• Change in ejection fraction (continuous outcome).

We will estimate all outcomes at maximum follow‐up according to randomised intervention group.

Search methods for identification of studies

Electronic searches

We will identify trials by searching the following bibliographic databases: Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library, MEDLINE (Ovid), Embase (Ovid), LILACS (Bireme), BIOSIS Citation Index (Web of Science, Thomson Reuters), and Science Citation Index Expanded (Web of Science, Thomson Reuters) (Royle 2003). We will also conduct a search of ClinicalTrials.gov (www.ClinicalTrials.gov), the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search portal (http://apps.who.int/trialsearch/), Turning Research Into Practice (TRIP), and Google Scholar, in order to search for unpublished and ongoing trials.

We will search all databases from their inception to the present and we will impose no restriction on language of publication. If we identify any papers in a language not known by the author group, we will seek assistance. This will be acknowledged in the published review. The preliminary search strategy for MEDLINE (Ovid) is given in Appendix 1, which will be adapted for use in the other databases. We will apply the Cochrane sensitivity and precision maximising RCT filter (Lefebvre 2011) to MEDLINE (Ovid) and adaptations of it to the other databases, except CENTRAL.

We will not perform a separate search for adverse effects of interventions used for the treatment of heart failure. We will consider adverse effects described in included studies only.

Searching other resources

We will check the bibliographies of review articles and identified trials for additional references.

We will search the websites of the European Medicines Agency (EMA) (www.ema.europa.eu), the US Food and Drug Administration (FDA) (www.fda.gov), and pharmaceutical company sources (www.astrazeneca‐us.com/; http://lifepharmaceuticalcompany.com/; http://pharma.bayer.com/; and http://pl.gsk.com/) for ongoing or unpublished trials.

We will also examine any relevant retraction statements and errata for included studies.

Data collection and analysis

We will perform the review following the recommendations of Cochrane (Higgins 2011a). The analyses will be performed using Review Manager 5 (RevMan 5) (RevMan 2014), STATA 15 (STATA 2015), and Trial Sequential Analysis version 0.9.5.5 Beta (TSA) (CTU 2011; Thorlund 2011).

Selection of studies

Two review authors (SS and NJS) will independently assess each identified trial. We will resolve any disagreements through discussion. If no agreement can be reached, a third review author (JCJ) will resolve the discussion.

Data extraction and management

We will use a data collection form for trial characteristics and outcome data that has been piloted on at least one trial in the review. One review author (SS) will extract trial characteristics and outcome data from all the included trials. The role of second assessor will be shared between the four other review authors (SKK, EEM, NJS, or JF), who will independently extract trial characteristics and outcome data from their own subset of studies. We will resolve disagreements by consensus or by involving a third review author (JCJ). One review author (SS) will transfer data into RevMan 5 (RevMan 2014). We will double‐check that data is entered correctly by comparing the data presented in the systematic review with the trial reports. A second review author will spot‐check trial characteristics for accuracy against the trial report. We will extract the following trial characteristics.

• Methods: duration of the trial, details of any 'run in' period, trial characteristics (i.e. cluster‐randomised controlled trial (RCT), multicentre RCT etc.), and date of publication.

• Participants: number randomised, number analysed, number completing trial or withdrawing, mean age, sex, inclusion criteria, and exclusion criteria.

• Interventions: intervention, comparison, concomitant medications, and excluded medications.

• Outcomes: primary and secondary outcomes specified and collected, and time points reported.

• Notes: funding of trial, and notable conflicts of interest of trial authors.

Assessment of risk of bias in included studies

We will use the instructions given in the Cochrane Handbook for Systematic Reviews of Interventions in our evaluation of the methodology and the risk of bias of the included trials (Higgins 2011a). One review author (SS) will assess risk of bias of all the included trials. The included trials will be shared amongst the other four review authors (SKK, EEM, NJS, and JF), with each independently assessing risk of bias in their own subset of studies. We will resolve any disagreements by discussion or by involving another review author (JCJ). We will evaluate the risks of bias according to the following domains.

• Random sequence generation.

• Allocation concealment.

• Blinding of participants and personnel.

• Blinding of outcome assessment.

• Incomplete outcome data.

• Selective outcome reporting.

• For‐profit bias.

• Other bias sources.

This is done because these domains enable classification of randomised clinical trials with low risk of bias and of randomised clinical trials with unclear or high risk of bias. The latter trials overestimate benefits and underestimate harms (Gluud 2006; Kjaergard 2001; Lundh 2017; Moher 1998; Savović 2012a; Savović 2012b; Schulz 1995; Wood 2008). For additional details on how we will assess the risk of bias in included studies, see Appendix 2.

Overall risk of bias

• Low risk of bias: we will classify the outcome result as at overall 'low risk of bias' only if all of the bias domains described in the above paragraphs are classified as at low risk of bias.

• High risk of bias: we will classify the outcome result as at 'high risk of bias' if any of the bias risk domains described above are classified as at 'unclear' or 'high risk of bias'.

We will assess the domains 'Blinding of outcome assessment', 'Incomplete outcome data', and 'Selective outcome reporting' for each outcome. Thus, we will be able to assess the risk of bias for each outcome in addition to each trial.

Assessment of bias in conducting the systematic review

We will conduct the review according to this published protocol and report any differences from it in the 'Differences between protocol and review' section of the review.

Measures of treatment effect

Dichotomous outcomes

We will calculate risk ratios (RRs) or odds ratios (OR) depending on the event rate in the control group. If the event rate in the control group is below 5% we will use the OR with 95% confidence intervals (CIs) for dichotomous outcomes (Sweeting 2004). If the event rate is, however, equal to or higher than 5% we will use risk ratios (RR) with 95% CI for dichotomous outcomes (Sweeting 2004). We will calculate OR and RR with 95% CIs and Trial Sequential Analysis (TSA)‐adjusted CI (Thorlund 2011). We will report TSA‐adjusted CI, instead of 95% CIs, if the cumulative Z‐curves does not reach the futility area or passes the diversity‐adjusted required information size (DARIS).

Continous outcomes

We will calculate the mean differences (MD) with 95% CI and TSA‐adjusted CI (Thorlund 2011) for continuous outcomes. We will secondly consider to use the standardised mean difference when the trials all assess the same outcome but measure it in a variety of ways (e.g. different scales) (Higgins 2011a). We will report TSA‐adjusted CI, instead of 95% CIs, if the cumulative Z‐curves does not reach the futility area or passes the diversity‐adjusted required information size (DARIS).

Unit of analysis issues

If we find any cross‐over trials, we will only include data from the first treatment period (Elbourne 2002).

If trials compare more than two intervention groups, we will divide the participants in the control group into two or more groups. In this way, we will avoid double‐counting participants in the control group, as a serious unit‐of‐analysis problem arises if the same group of participants is included twice in the same meta‐analysis.

We will include both individual and cluster‐randomised clinical trials and analyse the results separately (Higgins 2011a). Methods have been developed so results of individual and cluster‐randomised clinical trials may be meta‐analysed together. However, as participants from different clusters will, almost always, have different baseline characteristics and because it has repeatedly been shown that it is not possible with certainty to adjust the statistical analysis for such baseline differences we will analyse results of individual and cluster‐randomised clinical trials separately (Deeks 2003).

Dealing with missing data

We will contact all trial authors for missing or unclear data.

Dichotomous outcomes

If the included trials have used rigorous methodology (e.g. reporting on outcomes for all patients or multiple imputation to deal with missing data), we will use these data in our primary analysis. We will not impute missing values for any outcomes in our primary analysis.

Continous outcomes

If the included trials have used rigorous methodology (e.g. reporting on outcomes for all patients or multiple imputation to deal with missing data), we will use these data in our primary analysis. We will not impute missing values for any outcomes in our primary analysis. If standard deviations (SD) are not reported, the SDs will be calculated using data from the trial if possible. We will not use intention‐to‐treat data if the original report did not contain such data.

In our sensitivity analysis for dichotomous and continuous outcomes, we will impute data, see below and 'Sensitivity analysis'.

Best‐worst and worst‐best case scenarios

To assess the potential impact of the missing data for dichotomous outcomes, we will perform the following sensitivity analyses.

• ‘Best‐worst‐case’ scenario: we will assume that all participants lost to follow‐up in the experimental group survived, had no serious adverse event, major adverse cardiovascular event, reinfarction, or were hospitalised for any cause; and all those with missing outcomes in the control group have not survived, had a serious adverse event, major adverse cardiovascular event, reinfarction, or were hospitalised for any cause.

• ‘Worst‐best‐case’ scenario: we will assume that all participants lost to follow‐up in the experimental group did not survive, had a serious adverse event, major adverse cardiovascular event, reinfarction, or were hospitalised for any cause; and all those with missing outcomes in the control group survived, had no serious adverse event, major adverse cardiovascular event, reinfarction, or were hospitalised for any cause.

We will present results from both scenarios in the review.

When analysing continuous outcomes regarding missing data, a ‘beneficial outcome’ will be the group mean plus two SDs (and one SD) of the group mean, and a ‘harmful outcome’ will be the group mean minus two SDs (and one SD) of the group mean (Jakobsen 2014).

To assess the potential impact of missing SDs for continuous outcomes, we will perform the following sensitivity analysis: where SDs are missing and not possible to calculate, we will impute SDs from trials with similar populations and low risk of bias. If no such trials can be found, we will impute SDs from trials with a similar population. As the final option, we will impute SDs from all included trials.

Assessment of heterogeneity

We will primarily investigate forest plots to visually assess any sign of heterogeneity. We will secondly assess the presence of statistical heterogeneity by Chi² test (threshold P < 0.10) and measure the quantities of heterogeneity by the I² statistic (Higgins 2002; Higgins 2003).

We will follow the recommendations for threshold of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a).

• 0% to 40%: might not be important.

• 30% to 60%: may represent moderate heterogeneity.

• 50% to 90%: may represent substantial heterogeneity.

• 75% to 100%: may represent considerable heterogeneity.

We will investigate possible heterogeneity through subgroup analyses. Ultimately, we will consider not to conduct the overall meta‐analysis if the subgroup analysis shows different effects and the overall meta‐analysis shows substantial statistical heterogeneity (assessed by visual inspection of forest plots and I² statistic) (Higgins 2011a).

Assessment of reporting biases

We will use a funnel plot to assess reporting bias if ten or more trials are included. Using the asymmetry of the funnel plot, we will assess the risk of bias. For dichotomous outcomes, we will test asymmetry with the Harbord test if τ² is less than 0.1 (Harbord 2006), and with the Rücker test if τ² is more than 0.1 (Rücker 2008).

For continuous outcomes, we will use the regression asymmetry test (Egger 1997).

Data synthesis

Meta‐analysis

We will accept both end‐scores and change from baseline scores when analysing continuous outcomes. If both end‐scores and change from baseline scores are reported then we will use end‐scores. If only change values are reported the results will be analysed together with end‐scores by using unstandardized mean differences (Higgins 2011a).

We will undertake this systematic review according to the recommendations stated in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), Keus 2010, and Jakobsen 2014. We will use RevMan 5 to meta‐analyse data (RevMan 2014). We will use STATA in case of zero‐event trials (STATA 2015), where RevMan 5 zero‐event handling (replacing zero with a constant of 0.5) is not sufficient (e.g. in cases with skewed number of participants between groups, which we will handle with reciprocal zero‐event handling according to Sweeting 2004, and in case meta‐regression (post hoc) is needed).

If we identify trials with zero events in one or both intervention groups, then we will perform a supplementary analysis using beta‐binominal regression (Cheng 2016; Kuss 2015; Sharma 2017). If the result of the beta‐binominal regression differs from the traditional meta‐analysis result, then we will discuss the implications thoroughly (Cheng 2016; Kuss 2015; Sharma 2017).

We will assess our intervention effects with both random‐effects model meta‐analyses (DerSimonian 1986), and fixed‐effect model meta‐analyses (Demets 1987). We will use the more conservative point estimate of the two (Jakobsen 2014). The more conservative point estimate is the estimate closest to no effect (highest P value). If the two estimates are equal, we will use the estimate with the widest CI. We will use three primary outcomes and we will therefore consider a P value less than P ≤ 0.025 as statistically significant (Jakobsen 2014). We will consider maximum follow‐up as the primary time point. We will use the eight‐step procedure to assess if the thresholds of significance are crossed or not (Jakobsen 2014). We will use five secondary outcomes and we will therefore consider a P value less than P ≤ 0.016 for the secondary outcomes (Jakobsen 2014). We will use the eight‐step procedure to assess if the thresholds for significance are crossed or not (Jakobsen 2014).

We will present a table describing the types of serious adverse events in each trial.

Trial Sequential Analysis

Cumulative meta‐analyses are at risk of producing random errors due to sparse data and multiple testing of accumulating data (Brok 2008; Brok 2009; Higgins 2011a; Pogue 1997; Thorlund 2009; Wetterslev 2008; Wetterslev 2017). Trial Sequential Analysis (TSA), CTU 2011, can be applied to control these random errors and to assess the risks of imprecision (http://www.ctu.dk/tsa/) (Jakobsen 2014; Thorlund 2011). Similar to a sample size calculation in a randomised controlled trial, TSA calculates the required information size for the meta‐analysis (that is the number of participants needed in a meta‐analysis to detect or reject a certain intervention effect) in order to minimise random errors (Wetterslev 2009; Wetterslev 2017). The required information size takes into account the event proportion in the control group, the assumption of a plausible risk ratio (RR) reduction, and the heterogeneity of the meta‐analysis (Turner 2013; Wetterslev 2009). TSA enables testing for significance to be conducted each time a new trial is included in the meta‐analysis. On the basis of the required information size, trial sequential monitoring boundaries can be constructed. This enables one to determine the statistical inference concerning cumulative meta‐analysis that has not yet reached the required information size (Wetterslev 2008; Wetterslev 2017).

Firm evidence for benefit or harms may be established if the trial sequential monitoring boundary is crossed before reaching the required information size, in which case further trials may turn out to be superfluous. In contrast, if the boundary is not surpassed one may conclude that it is necessary to continue with further trials before a certain intervention effect can be detected or rejected. Firm evidence for lack of the postulated intervention effect can also be assessed with TSA. This occurs when the cumulative Z‐score crosses the trial sequential monitoring boundaries for futility.

For dichotomous outcomes, we will estimate the required information size based on the proportion of patients with an outcome in the control group, a relative risk reduction of 25%, an alpha of 2.5% for primary outcomes and 1.67% for secondary outcomes, a beta of 10%, and a variance suggested by the trials in a random‐effects meta‐analysis (diversity‐adjusted required information size) (Jakobsen 2014; Kotecha 2016; Wetterslev 2009). In case there is some evidence of effect of the intervention, a supplementary TSA will use the limit of the CI closest to 1.00 as the anticipated intervention effect (Jakobsen 2014). Additionally, we will calculate TSA‐adjusted CI.

For continuous outcomes, we have not identified valid previous data on effect sizes on quality of life so we have chosen to use SD/2 as anticipated intervention effect. Hence, we will estimate the required information size based on the SD observed in the control group of trials with low risk of bias or lower risk of bias and a minimal relevant difference of the observed SD/2, an alpha value of 2.5% for primary outcomes and 1.67% for secondary outcomes, a beta value of 10%, and a diversity suggested by the trials in the meta‐analysis (Jakobsen 2014; Wetterslev 2009). In case there is some evidence of effect of the intervention, as a supplementary TSA will use the limit of the CI closest to 0.00 as the anticipated intervention effect (Jakobsen 2014). Additionally, we will calculate TSA‐adjusted CI.

Subgroup analysis and investigation of heterogeneity

We will perform the following subgroup analyses.

• Comparison of intervention effects based on trials assessing the effects of different specific beta‐blockers.

• Comparison of intervention effects based on trials using different treatment periods:

  • less or equal to six months;

  • between six months and 12 months;

  • between one year and three years;

  • more than or equal to three years.

• Comparison of the intervention effects of beta‐blockers in participants with different underlying causes of heart failure (Yancy 2013):

  • idiopathic dilated cardiomyopathy;

  • ischaemic heart disease;

  • valvular heart disease;

  • hypertension;

  • tachycardia;

  • infectious endocarditis;

  • metabolic abnormalities;

  • mix causes of heart failure.

• Comparison of the effects of beta‐blockers for adults with heart failure according to dose (over or equal to or under the observed median dose) of beta blockers.

• Comparison of the effects of beta‐blockers for adults with heart failure according to severity of heart failure (namely HFrEF, HFmrEF, HFpEF and mixed or not specified group).

• Comparison of trials with different clinical trial registration status:

  • pre‐registration;

  • post‐registration;

  • no registration.

Sensitivity analysis

To assess the potential impact of bias, we will perform a sensitivity analysis where we exclude trials at overall 'high risk of bias'.

To assess the potential impact of the missing data for dichotomous outcomes, we will perform best‐worst and worst‐best case scenarios (see 'Dealing with missing data' section).

We will assess statistical heterogeneity by visual inspection of the forest plots and I² statistic values (Jakobsen 2014). Underlying reasons behind statistical heterogeneity in meta‐analyses will be investigated by assessing trial characteristics.

'Summary of findings' table

We will use GRADE to assess the quality of the body of evidence associated with each of the primary outcomes (all‐cause mortality, serious adverse events, and major adverse cardiovascular events), and some of our secondary outcomes (quality of life, cardiovascular mortality, and myocardial infarction). We will construct 'Summary of findings' tables using GRADEpro GDT (GRADEpro GDT 2015), and the standard CI for statements of precision or imprecision (Guyatt 2008; Jakobsen 2014). The TSA‐adjusted CI will be reported in the 'comments' field. We will use methods and recommendations described in Chapter 11 (Section 11.5) (Higgins 2011b), and Chapter 12 (Schünemann 2017), of the Cochrane Handbook for Systematic Reviews of Interventions. The GRADE approach appraises 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 item being assessed. We will assess the GRADE levels of evidence as high, moderate, low, or very low and will downgrade the evidence by one or two levels depending on the following quality measures: within study risk of bias, the directness of the evidence, heterogeneity of the data, precision of effect estimates, and risk of publication bias (Guyatt 2008). Two review authors (SS and NJS) will assess the quality of evidence independently and decide on downgrading. If no agreement can be reached, a third review author (JCJ) will resolve the discussion. We will justify all decisions to downgrade the quality of the evidence using footnotes and we will provide comments to aid the reader's understanding of the review where necessary. We plan to only create one overall 'Summary of findings' table for our main analysis of beta‐blockers versus placebo or no intervention.

We will include all studies in our analyses, and conduct a sensitivity analysis with studies at low risk of bias. If the results are similar we will base our primary 'Summary of findings' table and primary conclusions on the overall analysis.

Appendices

Appendix 1. Preliminary MEDLINE (Ovid) search strategy

1 exp Heart Failure/

2 ((heart or cardiac or myocard*) adj2 (fail* or insufficien* or decomp*)).tw.

3 1 or 2

4 exp Adrenergic beta‐Antagonists/

5 betablock*.tw.

6 (beta adj3 (antagonist* or adrenergic* or block*)).tw.

7 acebutolol*.tw.

8 adaprolol*.tw.

9 adimolol*.tw.

10 afurolol*.tw.

11 alprenolol*.tw.

12 amosulalol*.tw.

13 ancarolol*.tw.

14 arnolol*.tw.

15 arotinolol*.tw.

16 atenolol*.tw.

17 befunolol*.tw.

18 betaxolol*.tw.

19 bevantolol*.tw.

20 bisoprolol*.tw.

21 bometolol*.tw.

22 bopindolol*.tw.

23 bornaprolol*.tw.

24 brefonalol*.tw.

25 bromoacetylalprenololmenthane*.tw.

26 bucindolol*.tw.

27 bucumolol*.tw.

28 bufetolol*.tw.

29 bufuralol*.tw.

30 bunitrolol*.tw.

31 bunolol*.tw.

32 bupranolol*.tw.

33 butofilolol*.tw.

34 butoxamine*.tw.

35 carazolol*.tw.

36 carpindolol*.tw.

37 carteolol*.tw.

38 carvedilol*.tw.

39 celiprolol*.tw.

40 cetamolol*.tw.

41 cicloprolol*.tw.

42 cinamolol*.tw.

43 cloranolol*.tw.

44 dexpropranolol*.tw.

45 diacetolol*.tw.

46 draquinolol*.tw.

47 ecastolol*.tw.

48 epanolol*.tw.

49 ericolol*.tw.

50 esatenolol*.tw.

51 esmolol*.tw.

52 exaprolol*.tw.

53 falintolol*.tw.

54 flestolol*.tw.

55 flusoxolol*.tw.

56 idropranolol*.tw.

57 indenolol*.tw.

58 indopanolol*.tw.

59 iprocrolol*.tw.

60 isoxaprolol*.tw.

61 labetalol*.tw.

62 landiolol*.tw.

63 levobetaxolol*.tw.

64 levobunolol*.tw.

65 levomoprolol*.tw.

66 medroxalol*.tw.

67 mepindolol*.tw.

68 metipranolol*.tw.

69 metoprolol*.tw.

70 moprolol*.tw.

71 nadolol*.tw.

72 nadoxolol*.tw.

73 nebivolol*.tw.

74 nifenalol*.tw.

75 nipradilol*.tw.

76 oxprenolol*.tw.

77 pacrinolol*.tw.

78 pafenolol*.tw.

79 pamatolol*.tw.

80 pargolol*.tw.

81 penbutolol*.tw.

82 penirolol*.tw.

83 pindolol*.tw.

84 pirepolol*.tw.

85 practolol*.tw.

86 primidolol*.tw.

87 prizidilol*.tw.

88 procinolol*.tw.

89 propranolol*.tw.

90 ridazolol*.tw.

91 ronactolol*.tw.

92 soquinolol*.tw.

93 sotalol*.tw.

94 spirendolol*.tw.

95 talinolol*.tw.

96 tazolol*.tw.

97 tertatolol*.tw.

98 tienoxolol*.tw.

99 tilisolol*.tw.

100 timolol*.tw.

101 tiprenolol*.tw.

102 tolamolol*.tw.

103 tribendilol*.tw.

104 trigevolol*.tw.

105 xibenolol*.tw.

106 xipranolol*.tw.

107 4 or 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 or 32 or 33 or 34 or 35 or 36 or 37 or 38 or 39 or 40 or 41 or 42 or 43 or 44 or 45 or 46 or 47 or 48 or 49 or 50 or 51 or 52 or 53 or 54 or 55 or 56 or 57 or 58 or 59 or 60 or 61 or 62 or 63 or 64 or 65 or 66 or 67 or 68 or 69 or 70 or 71 or 72 or 73 or 74 or 75 or 76 or 77 or 78 or 79 or 80 or 81 or 82 or 83 or 84 or 85 or 86 or 87 or 88 or 89 or 90 or 91 or 92 or 93 or 94 or 95 or 96 or 97 or 98 or 99 or 100 or 101 or 102 or 103 or 104 or 105 or 106

108 3 and 107

109 randomized controlled trial.pt.

110 controlled clinical trial.pt.

111 randomized.ab.

112 placebo.ab.

113 clinical trials as topic.sh.

114 randomly.ab.

115 trial.ab.

116 109 or 110 or 111 or 112 or 113 or 114 or 115

117 exp animals/ not humans.sh.

118 116 not 117

119 108 and 118

Appendix 2. 'Risk of bias' assessment details

We will classify each trial according to the domains below for each outcome.

Random sequence generation

• Low risk: if sequence generation is achieved using computer random number generator or a random numbers table. Drawing lots, tossing a coin, shuffling cards and throwing dice are also be considered adequate if performed by an independent adjudicator.

• Unclear risk: if the method of randomisation is not specified.

• High risk: if the allocation sequence is not randomised or only quasi‐randomised.

Allocation sequence concealment

• Low risk: if the allocation of patients is performed by a central independent unit, on‐site locked computer, identical looking numbered sealed opaque envelopes, drug bottles or containers prepared by an independent investigator. There must be no risk of the investigator knowing the sequence.

• Unclear risk: if the trial is classified as randomised but the allocation concealment process is not described.

• High risk: if the allocation sequence is known to the investigators who assigned participants.

Blinding of participants and personnel

• Low risk: if the participants and the personnel are blinded to treatment allocation and this is described.

• Unclear risk: if the procedure of blinding is insufficiently described or not described at all.

• High risk: if blinding of participants and personnel is not performed.

Blinding of outcome assessment

• Low risk: if the trial investigators performing the outcome assessments, analyses and calculations are blinded to the intervention.

• Unclear risk: if the procedure of blinding is insufficiently described or not described at all.

• High risk: if blinding of outcome assessment is not performed.

Incomplete outcome data

• Low risk: (1) there are no dropouts or withdrawals for all outcomes, or (2) the numbers and reasons for the withdrawals and dropouts for all outcomes are clearly stated, can be described as being similar in both groups, and the trial handles missing data appropriately in intention‐to‐treat analysis using proper methodology, e.g. multiple imputations*. As a general rule the trial is judged as at a low risk of bias due to incomplete outcome data if the number of dropouts is less than five per cent. However, the 5% cut off is not definitive.

• Unclear risk: the numbers and reasons for withdrawals and dropouts are not clearly stated.

• High risk: the pattern of dropouts can be described as being different in the two intervention groups or the trial uses improper methodology in dealing with the missing data, e.g. last observation carried forward.

* Multiple imputation is a general approach to the problem of missing data. It aims to allow for the uncertainty about the missing data by creating several different plausible imputed data sets and appropriately combining results obtained from each of them. The first stage is to create multiple copies of the dataset, with the missing values replaced by imputed values. These are sampled from their predictive distribution based on the observed data, thus multiple imputation is based on a Bayesian approach. The imputation procedure must fully account for all uncertainty in predicting the missing values by injecting appropriate variability into the multiple imputed values. The second stage is to use standard statistical methods to fit the model of interest to each of the imputed datasets. The estimated associations from the imputed datasets will differ and are only useful when averaged together to give overall estimated associations. Valid inferences are obtained because we are averaging over the distribution of the missing data given the observed data (Sterne 2009).

Selective outcome reporting

• Low risk: a protocol is published before or at the time the trial is begun and the outcomes called for in the protocol are reported on. If there is no protocol or the protocol is published after the trial has begun, reporting of the primary outcomes will grant the trial a grade of low risk of bias.

• Unclear risk: if there is no protocol and the primary outcomes are not reported on.

• High risk: if the outcomes which are called on in a protocol are not reported on.

Other bias risk

• Low risk of bias: the trial appears to be free of other components (e.g. academic bias or for‐profit bias) that could put it at risk of bias.

• Unclear risk of bias: the trial may or may not be free of other components that could put it at risk of bias.

• High risk of bias: there are other factors in the trial that could put it at risk of bias (e.g. authors have conducted trials on the same topic, for‐profit bias etc).

Overall risk of bias

• Low risk of bias: the outcome result will be classified as overall 'low risk of bias' only if all of the bias domains described in the above paragraphs are classified as low risk of bias.

• High risk of bias: the outcome result will be classified 'high risk of bias' if any of the bias risk domains described in the above are classified as 'unclear' or 'high risk of bias'.

Contributions of authors

Sanam Safi (SS), Steven Kwasi Korang (SKK), and Emil E. Nielsen (EEN) conceived, designed and drafted the protocol.

Joshua Feinberg (JF), Naqash J Sethi (NJS), Janus C. Jakobsen (JCJ), and Christian Gluud (CG) provided general advice and revised the protocol.

All authors agreed on the final protocol version.

Sources of support

Internal sources

• No sources of support supplied

External sources

• This project was supported by the National Institute for Health Research, via Cochrane Infrastructure to the Heart Group. The views and opinions expressed therein are those of the authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR, NHS or the Department of Health., UK.

Declarations of interest

The performance of this review is free of any real or perceived bias introduced by receipt of any benefit in cash or kind, on any subsidy derived from any source that may have or be perceived to have an interest in the outcomes of the review.

Sanam Safi (SS): no conflict of interest.

Steven Kwasi Korang (SKK): no conflict of interest.

Emil E Nielsen (EEN): no conflict of interest.

Naqash J Sethi (NJS): no conflict of interest.

Joshua Feinberg (JF): no conflict of interest.

Christian Gluud (CG): member of The Copenhagen Trial Unit's task force for developing TSA methods, manual, and software available free of charge at www.ctu.dk/tsa.

Januc C Jakobsen (JCJ): no conflict of interest.

New