Scolaris Content Display Scolaris Content Display

Cochrane Database of Systematic Reviews Protocol - Intervention

Implantable cardiac defibrillators for patients with non‐ischaemic cardiomyopathy

This is not the most recent version

Collapse all Expand all

Abstract

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

To evaluate the effects of implantable cardiac defibrillators on patients with non‐ischaemic cardiomyopathy compared to optimal medical therapy, and the adverse events associated with using these devices.

Background

Description of the condition

Heart failure (HF) is a public health concern and its burden has increased over the past decades. It is estimated that as of 2016, more than 37.7 million people are affected by this disease globally (Ziaeian 2016). The annual global economic burden of the disease was estimated at USD 108 billion per annum in 2012, with direct costs to healthcare systems accounting for USD 65 billion and indirect costs (through lost productivity, morbidity, premature mortality, etc.) accounting for USD 43 billion (Ambrosy 2014). While the discovery of several evidence‐based drug and device therapies has improved the outcomes for ambulatory HF patients, postdischarge mortality and readmission rates of hospitalised HF patients have not changed in the past two decades (Ambrosy 2014).

HF is a clinical syndrome that is characterised by an impaired systemic delivery of blood, whereby the heart fails to pump enough blood to meet the body's metabolic demand (McMurray 2012). HF can result from many cardiac diseases, which had led to a multitude of classifications for this syndrome based on anatomy (disease of the myocardium, pericardium, or endocardium), pathophysiological mechanism (systolic vs diastolic dysfunction), and etiology (ischaemic vs non‐ischaemic), with the latter being the most commonly used in the clinical setting (Maron 2006). Ischaemic HF is caused by myocardial ischaemia due to an underlying coronary artery disease (Ho 1993). Various potential causes leading to non‐ischaemic cardiomyopathy (NICM) are described in the literature, which are classified into four main categories: dilated cardiomyopathy (DCM) (idiopathic, viral illness, alcoholism, etc.), hypertrophic cardiomyopathy (genetic factors and hypertension), restrictive cardiomyopathy (amyloidosis, sarcoidosis, haemochromatosis, etc.), and arrhythmogenic right ventricular dysplasia (Wu 2007). DCM is characterised by systolic dysfunction leading to a reduced Left Ventricular Ejection Fraction (LVEF) (Daughenbaugh 2007). While not all patients with DCM display signs of HF, it is believed that long‐standing DCM eventually leads to HF (McNally 2013; Dass 2015). In fact, patients with DCM can be either asymptomatic (New York Heart Assocation (NYHA) class I), or can display signs and symptoms of HF (e.g. dyspnoea, fatigue, fluid retention) (NYHA class II‐IV depending on severity) in which case they are considered to have non‐ischaemic systolic HF (Yancy 2013; Ponikowski 2016).

HF patients, whether ischaemic or non‐ischaemic, are at increased risk of developing ventricular arrhythmias due to underlying structural and mechanical factors (e.g. increased inward sodium and calcium currents due to myocardial stretch) (Wu 2007). These can range from non‐sustained ones, such as premature ventricular beats, to life‐threatening ones such as ventricular fibrillation (VF) that leads to sudden cardiac death (SCD). SCD is defined as a rapid and fatal cardiovascular decompensation in a previously haemodynamically stable patient that happens within one hour, and its risk increases as the LVEF falls below 40% (Yap 2005; Shiga 2009; Mosterd 2001). SCD has long been considered the leading cause of death in HF patients (Bardy 2005; Packer 2009). However, new evidence suggests that this has changed over the last few years. In fact, a recent study (CERTITUDE) has shown that progressive HF is now the leading cause of death in cardiac resynchronisation therapy (CRT) patients (Marijon 2015). Other studies have also shown that mortality due to progressive HF and to non‐cardiovascular causes were more common than SCD in HF patients (Henkel 2008; Pons 2010). Nevertheless, current medical and device therapies are directed towards preventing and terminating arrhythmic events before their progression to fatal VF and SCD (Derfler 2004).

HF patients can also present with sudden severe symptoms (acute decompensated HF) or with an insidious onset (chronic HF). The two conditions are managed differently: acute HF therapy is mainly managed pharmacologically (diuretics, vasodilators, etc.) while chronic HF is managed through both device and medical therapy (Ponikowski 2016). Since our analysis focuses on device therapy, our target population are patients with chronic HF.

Description of the intervention

The standard of care for the treatment of HF patients is defined by a guideline recommended approach (primarily Class I), which consists of optimal medical therapy (OMT) as a first‐line treatment, either with or without device therapy (Yancy 2013; Ponikowski 2016). OMT is a multidimensional approach to managing people with HF, which consists of pharmacological treatment of HF and co‐morbidities (e.g. hypertension) and lifestyle modifications, all of which serve to improve patients' outcome by reducing the risk of HF hospitalisation and death. Treatment with Angiotensin Converting Enzyme inhibitors (ACE‐I) or Angiotensin Receptor Blockers (ARBs) in addition to a beta‐blocker (Class I level A) is the mainstay for the pharmacological treatment of HF with reduced ejection fraction (i.e. LVEF ≤ 40%) (Yancy 2013; Ponikowski 2016). Studies have linked the use of these drugs with a decreased morbidity and mortality in patients with HF, by decreasing the workload on the heart (Heran 2012; Chatterjee 2013). If symptoms persist despite treatment with ACE‐I and a beta‐blocker, a mineralocorticoid receptor antagonist (MRA) is recommended (Class 1 Level A) to reduce the risk of HF hospitalisation and death (Yancy 2013; Ponikowski 2016). Long‐term control of weight, blood pressure, glucose and lipid levels, and reduction in smoking and alcohol consumption also improves patients' survival and reduces the risk of HF in at‐risk patients (McMurray 2012).

Implantable Cardioverter Defibrillators (ICDs) are implantable devices that deliver electrical energy to the heart during episodes of ventricular arrhythmias, thus ending the life‐threatening event and resetting the sinus rhythm. They have been effective in reducing mortality in patients with tachycardia, fibrillation, and other arrhythmic anomalies (Kuck 2000; Young 2003; Josephson 2004). The very first guidelines that discussed the use of ICD in clinical practice were based on the Antiarrhythmics versus Implantable Defibrillators (AVID) study, a large multicenter trial which showed a reduction in all‐cause mortality in cardiac arrest survivors (AVID investigators 1997). Hence, a Class I recommendation was established by the American College of Cardiology (ACC)/American Heart Association (AHA) for the use of ICD in secondary prevention in patients meeting enrolment criteria for AVID (Poole 2014). The Canadian Implantable Defibrillator Study (CIDS) and the Cardiac Arrest Study Hamburg (CASH) trials supported the results of this trial. While these studies lacked statistical significance, a pattern of mortality reduction in ICD‐treated patients started to emerge (Connolly 2000; Kuck 2000).

It was not until after the MADIT II trial that guidelines upgraded ICD from a secondary to a primary and secondary prevention therapy (Moss 2002). Furthermore, guidelines were once again updated in 2006 based on the results of the SCD‐Heft trial, assigning Class I recommendation for ICD implantation in non‐ischaemic HF patients as well. As a matter of fact, the recommendations have not changed since: ICD implantation is a Class I recommendation for the primary and secondary prevention of SCD due to VT/VF in more severe cases of non‐ischaemic HF (LVEF ≤ 35% despite ≥ three months of optimal medical therapy) as an adjunct to medical therapy (Yancy 2013; Ponikowski 2016).

The recommendations for the prophylactic use of ICDs in non‐ischaemic HF were based on subgroup analyses and small trials of low quality evidence. Indeed, the DANISH study, which is the largest randomised control on patients with non‐ischaemic HF, showed that there is no long‐term beneficial effect from ICD implantation in terms of mortality reduction (Kober 2016).

Patients who meet the criteria for ICD implantation may present with a pre‐existing device therapy: a cardiac re‐synchronisation therapy device (CRT). CRT is recommended for symptomatic patients with HF in sinus rhythm with a QRS interval — the deflections seen on an electrocardiogram corresponding to the depolarisation of the right and left ventricles of the heart — duration ≥ 150 msec (normal 80 to 100 msec) and left bundle branch block (LBBB) QRS morphology and with LVEF ≤ 35% despite OMT in order to improve symptoms and reduce morbidity and mortality (Yancy 2013; Ponikowski 2016). In fact, ICDs can be implanted alone, or in conjunction with a CRT device, in which case the device is upgraded to a CRT defibrillator (CRT‐D), maintaining its initial function as a pacemaker, and also taking on a new function which is that of the defibrillator. In other words, if these patients to whom a pacemaker (CRT‐P) was already implanted meet the criteria for ICD implantation, their device is upgraded to a CRT‐D.

ICD implantation is associated with a series of adverse events that are both physical (e.g. haemothorax, pneumothorax) and psychological (Rosenqvist 1998). ICD is only effective in patients who are at high risk of SCD; otherwise, patients are exposed to unnecessary adverse events from device implantation that reduce their quality of life (QoL) without benefiting from its action. While the AVID trial showed that ICD reduces all‐cause mortality in cardiac arrest survivors, patients randomised to the ICD arm of the study showed significant decline in QoL manifested by a decrease in physical functioning and mental well‐being and increased anxiety (Schron 2002). ICDs deliver unexpected shocks to the patients without warning, which creates distress to them and their family members (Dunbar 1993; Dougherty 1995). Patients live in a state of anxiety, constantly anticipating an electric shock, which results in an elevated incidence of anxiety disorders and panic attacks (Sears 2002). Finally, Post‐traumatic stress disorder has also been reported to be elevated in patients with ICD, causing them to re‐experience the events of shock and manifest emotions associated with them (Hamner 1999; Ladwig 2008).

Hence, careful evaluation of patients' profile and ensuring they meet the guidelines' criteria for implantation is critical in preventing patients' exposure to unnecessary harm through overprescription of ICD therapy.

How the intervention might work

Beta‐blockers suppress the firing of these action potentials by blocking the sodium and calcium channels. However, ICDs work through a different mechanism: while beta‐blockers prevent the abnormal firing of cardiomyocytes, ICDs work to stop the chain reaction created by the abnormal pacemaker cells. During episodes of arrhythmias, they deliver an electric shock that depolarises all the cardiac cells simultaneously, thereby halting the progression of the arrhythmia. Thus, the primary purpose of ICDs is to end life‐threatening ventricular arrhythmias and reduce SCD or arrhythmia‐related complications. Therefore, we can only expect a relevant reduction in mortality rate in subgroups of HF patients with high incidence of sustained ventricular arrhythmias leading to cardiac arrest if not terminated.

In non‐ischaemic HF, currents that are abnormally generated in the ventricle can be immediately stopped through the delivery of a therapeutic dose of electric energy from an ICD device.

Why it is important to do this review

Several studies have assessed the effectiveness of ICD treatment in NICM patients (Kadish 2000; Desai 2004); however, these studies had conflicting results, with some showing significant improvement in primary outcome (e.g. COMPANION trial) while others showed no benefit (e.g. CAT and AMIOVIRT trials) (Bansch 2002; Strickberger 2003; Bristow 2004).

A meta‐analysis of 1854 patients from five ICD primary prevention trials concluded that ICD treatment in NICM patients significantly improves mortality rate when compared to medical therapy alone (Desai 2004). Moreover, the ACC and Euopean Society of Cardiology (ESC) guidelines for device‐based therapy recommend ICD implantation in patients with non‐ischaemic HF with LVEF ≤ 35% (Yancy 2013; Ponikowski 2016).

A recently published large randomised control trial — the DANISH study — defied current guidelines and the 2004 meta‐analysis, supporting the evidence that ICD use in NICM patients does not reduce the all‐cause mortality rate (Kober 2016). Hence, the purpose of this Cochrane Review is to assess the benefit of ICD use in NICM patients using a meta‐analysis of trials tackling this subject, including the Danish study, and thus provide a framework for future guideline updates.

While several systematic reviews have been published on this topic after the publication of the DANISH results, the measured outcomes were all mortality‐related (SCD mortality, all‐cause mortality, etc.) (Al‐Khatib 2017; Golwala 2017; Luni 2017; Shun‐Shin 2017; Wolff 2017). None of these studies assessed the adverse events versus benefits of ICD implantation and the cost of this therapy. Moreover, none of these reviews assessed the benefits of ICDs in different patient subgroups (e.g. age, gender, baseline ejection fraction, etc.). This is of critical importance since the DANISH study showed that there was a significant treatment‐by‐subgroup interaction for age, with all‐cause mortality being significantly lower in younger patients on ICD, compared to older patients who did not derive any benefit from ICD implantation (Kober 2016). Finally, some of the studies performed a ʽRisk of bias' assessment; however, none assessed the quality of the evidence.

To this end, this Cochrane Review will be the first to: measure mortality‐related outcomes in addition to adverse events vs added benefits and cost; assess the risk of bias using the Cochrane tool and the quality of evidence using the GRADE approach; and perform an analysis on different patient subgroups in a single study.

Finally, we will also search for unpublished studies which may add evidence so far unknown.

Objectives

To evaluate the effects of implantable cardiac defibrillators on patients with non‐ischaemic cardiomyopathy compared to optimal medical therapy, and the adverse events associated with using these devices.

Methods

Criteria for considering studies for this review

Types of studies

We will include cluster‐randomised controlled trials (cluster‐RCTs), parallel randomised controlled trials (RCTs), and economic evaluation studies published after 1975 in any language. The first ICD was implanted in February 1980 at Johns Hopkins Hospital by Dr. Levi Watkins (Mirowski 1980). Therefore, all RCT and economic evaluation studies are supposedly published after this date. Nevertheless, to be on the safe side, we will give a safety margin of five years (1975). We will include studies reported either as full‐text or as abstracts, in addition to unpublished studies.

Types of participants

We will include all adults (aged 18 years or older) with a diagnosis of chronic non‐ischaemic cardiomyopathy due to a left ventricular systolic dysfunction and with ejection fraction ≤ 35% (NYHA type I ‐ IV).

Left ventricular dysfunction includes both systolic and diastolic ventricular dysfunctions. The left ventricular systolic dysfunction is defined by a LVEF, whereas the left ventricular diastolic dysfunction is defined by an abnormal filling of left ventricle accompanied by an elevated filling pressure (Wu 2007).

The focus of our review will be on the primary prevention of SCD in HF patients.

Many published studies include both ischaemic and non‐ischaemic participants. In such situations, we will contact the authors of the studies for data on the non‐ischaemic participants.

Types of interventions

Intervention

ICD or CRT‐D, in addition to OMT.

Control

OMT alone for studies with ICD as the intervention, or OMT with CRT‐P for studies with CRT‐D as the intervention.

We will include CRT studies that include a defibrillator arm (CRT‐D) (intervention group) and a cardiac resynchronisation pacing only arm (CRT‐P) (control group), since the control group (CRT‐P) will be appropriately balanced for the anti‐arrhythmic effect of CRTs. We will be, in fact, comparing ICD vs no‐ICD. We will exclude studies that compare CRT‐D to ICD since both treatment arms will have ICDs. Also, we will exclude studies that compare CRT‐D to no device.

Types of outcome measures

Primary outcomes

  • All‐cause mortality.

  • Cardiovascular mortality.

  • Sudden cardiac death.

  • Adverse events of using CRT‐D or ICD (adverse events include serious device infection, pneumothorax, haemothorax, inappropriate shocks, lead displacement, dissection and tamponade, lead‐failure/fracture).

Secondary outcomes

  • Non‐cardiovascular death.

  • Health‐Related Quality of life (HRQoL) (measured on any validated scale such as the Chronic Heart Failure Questionnaire (Guyatt 1989), Quality of Life Questionnaire for Severe Heart Failure (Wiklund 1987), the Kansas City Cardiomyopathy Questionnaire (Green 2000), the Left Ventricular Dysfunction questionnaire (O'Leary 2000), and the Minnesota Living with Heart Failure Questionnaire (Rector 1987)).

  • Cost (measured by physician visits, number of hospital readmissions after implant, length of hospital stay during readmissions, or direct cost of medical resources (laboratory tests, cost of device, and medications)).

  • HF hospitalisation.

  • First ICD‐related hospitalisation.

We will include studies that have a follow‐up time of at least 12 months, since only long‐term trials may obtain sufficient and reliable results for the mortality‐related outcomes of interest (e.g. SCD, all‐cause mortality, etc.).

Reporting one or more of the outcomes listed here in the trial is not an inclusion criterion for the review.

Search methods for identification of studies

Electronic searches

We will identify trials through systematic searches of the following bibliographic databases:

  • Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library

  • MEDLINE (Ovid, from 1946 onwards)

  • Embase (Ovid, from 1980 onwards)

  • Web of Science Core Collection.

We will adapt the search strategy developed for MEDLINE (Ovid) (Appendix 1) for use in the other databases. We will apply the Cochrane sensitivity‐maximising RCT filter (Lefebvre 2011) to MEDLINE (Ovid) and adaptations of it to the other databases, except CENTRAL.

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/), and the ISRCTN registry (https://www.isrctn.com/) for ongoing or unpublished trials.

We will search all databases from 1975 to the present. The first ICD was implanted in February 1980 at Johns Hopkins Hospital by Dr. Levi Watkins, therefore all RCT studies are likely to be published after this date. However, we will allow a five‐year safety margin and search from 1975, in case there are any earlier publications. We will impose no restriction on language of publication or publication status.

We will conduct a separate search of the NHS Economic Evaluation Database in the Cochrane Library (from inception to 31 March 2015) to identify economic evaluation studies. As this database has not been updated since 2015, we will also search MEDLINE (Ovid, from 2015 onwards) and Embase (Ovid, from 2015 onwards). We will adapt the review search strategy, removing the RCT filter and applying an adaptation of the Centre for Reviews and Dissemination (CRD) economics evaluation search filter instead (Centre for Reviews and Dissemination) to MEDLINE (Ovid) (Appendix 2) and Embase (Ovid).

We will not perform a separate search for adverse effects of interventions used for the treatment of non‐ischaemic HF. We will consider adverse effects described in included studies only.

Searching other resources

We will check reference lists of all included studies and any relevant systematic reviews identified for additional references to trials. We will also examine any relevant retraction statements and errata for included studies.

Data collection and analysis

Selection of studies

Two review authors (ME, JN) will independently screen the titles and abstracts of all citations identified by the search for potential eligibility (eligible or unclear eligibility). We will retrieve the full texts of citations judged as potentially eligible by at least one of the two review authors. Two review authors (ME, JN) will both independently screen the full texts for eligibility. They will resolve any disagreement through discussion or, if required, with the help of a content expert (MR).

We will identify and exclude duplicate articles and will collate several reports of the same study so that each study, rather than each report, is the unit of interest in the review. We will complete a PRISMA flow diagram to record the selection process. We will also complete a ʽCharacteristics of excluded studies' table to record reasons for exclusion of the ineligible studies.

Data extraction and management

Two review authors (JN, ME) will independently extract data regarding the study characteristics from included studies. We will develop and pilot test a data collection form for study characteristics and outcome data. The two review authors (JN, ME) will resolve any disagreement through discussion or, if required, with the help of a content expert (MR).

We will extract data the following study characteristics.

  • Methods: study design, total duration of study, details of any 'run in' period, number of study centres and location, study setting, withdrawals, and date of study.

  • Participants: inclusion criteria, and exclusion criteria, number of participants randomised, number of participants lost to follow‐up/withdrawn, number of participants analysed, and general characteristics (e.g. mean age, age range, gender, severity of condition, diagnostic criteria, baseline lung function, smoking history).

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

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

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

One review author (ME) will transfer data into the Review Manager 5 (RevMan 5) file (RevMan 2014). We will double‐check that data is entered correctly by comparing the data presented in the systematic review with the study reports. A second review author (JN) will spot‐check study characteristics for accuracy against the trial report.

Assessment of risk of bias in included studies

Two review authors (ME, JN) will both independently assess the risk of bias for each included study. They will resolve any disagreements by discussion or by consulting a third review author (GI).

We will assess the risk of bias using the criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

  • Random sequence generation.

  • Allocation concealment.

  • Blinding of participants and personnel.

  • Blinding of data collectors and outcome adjudicators.

  • Incomplete outcome data.

  • Selective outcome reporting.

  • Other bias.

We will grade each potential source of bias as either high, low, or unclear and will provide a quote from the study report, together with a justification for our judgment, in the ʽRisk of bias' table. We will summarise the ʽRisk of bias' judgements across different studies for each of the domains listed. Where information on risk of bias relates to unpublished data or correspondence with a trial author, we will note this in the ʽRisk of bias' table.

When considering treatment effects, we will consider the risk of bias for the studies that contribute to that outcome.

Assessment of bias in conducting the systematic review

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

Measures of treatment effect

We will analyse dichotomous data as risk ratios with 95% confidence intervals (CIs) and continuous data as mean difference (MD) or standardised mean difference (SMD) values with 95% CIs. We will use the MD when we pool data from studies that used the same tool or scale to assess the outcome of interest. We will use the SMD when pooling data from studies that used different tools or scales to assess the outcome of interest. We will enter data presented as a scale with a consistent direction of effect.

We will narratively describe skewed data reported as medians and interquartile ranges.

Unit of analysis issues

If we come across cluster‐RCTs that did not make adjustments for correlation between cluster participants, and that meet the inclusion criteria, we will contact the corresponding authors of these studies for information about the correlation between participants of the individual clusters.

Dealing with missing data

We will contact investigators or study sponsors in order to verify key study characteristics and obtain missing numerical outcome data where possible (e.g. when we identify a study as an abstract only). Where this is not possible, and we consider that the missing data may introduce serious bias, we will explore the impact of including such studies in the overall assessment of results by a sensitivity analysis (Akl 2013; Ebrahim 2013; Ebrahim 2014).

Assessment of heterogeneity

Clinical heterogeneity

We will assess clinical heterogeneity by comparing the population, experimental intervention, and control intervention.

Statistical heterogeneity

We will use the standard Chi² test value (P = 0.1) and a visual analysis of a forest plot to explore statistical heterogeneity in the included trials. Additionally, we will use the I² statistic to measure heterogeneity among the trials in each analysis. We will consider a statistically significant value for the Chi² test along with an I² statistic value of ≥ 50% as substantial statistical heterogeneity (Higgins 2011). If so, we will report it and explore possible causes by prespecified subgroup analysis.

Methodological heterogeneity

We will explore methodological heterogeneity by comparing the quality and study design of the included trials.

Assessment of reporting biases

If we are able to pool more than 10 trials, we will create and examine a funnel plot to explore possible small study biases for the primary outcomes. If not, the power of the tests becomes too low to distinguish chance variation from real asymmetry (Higgins 2011).

Data synthesis

We will undertake meta‐analyses only where this is meaningful i.e. if the treatments, participants and the underlying clinical question are similar enough for pooling to make sense. We will use a random‐effects model for pooling of treatment effects since the studies will differ in the mixes of participants and in the implementations of interventions, etc. We will present all results with the corresponding 95% CIs. We will conduct all analyses according to the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011 ), and with the statistical components of RevMan 5 (RevMan 2014). We will include outcome measures only if it was the intention of the study to perform the necessary assessments in all included patients (i.e. not an optional outcome or only performed in some centres). When the results of a particular outcome measure are available for fewer than 50% of the patients in a study, we will not report the results of this outcome measure, which may be at high risk of attrition bias. We will pool results only if treatment groups are comparable, including the outcome definitions that were used.

We will descriptively summarise the studies for which pooling of results is not possible.

ʽSummary of findings' table

We will create a ʽSummary of findings' table using the following outcomes: all‐cause mortality, cardiovascular mortality, sudden cardiac death, adverse effects of using CRT‐D or ICD, non‐cardiovascular death, HRQoL, and first hospital admission after ICD implant. We will use the five GRADE considerations (study limitations, consistency of effect, imprecision, indirectness and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes (Balshem 2011; Guyatt 2011a; Guyatt 2011b; Guyatt 2011c; Guyatt 2011d; Guyatt 2011e; Guyatt 2011f; Guyatt 2013a; Guyatt 2013b; Guyatt 2013c). We will use methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a; Schünemann 2011) using GRADEpro software (GRADEpro GDT 2015). We will justify all decisions to downgrade the quality of studies using footnotes and we will make comments to aid reader's understanding of the review where necessary.

Subgroup analysis and investigation of heterogeneity

We plan to perform the following subgroup analyses.

  • Mean age (< 65 years vs ≥ 65 years).

  • Pre‐existing pacemaker (yes vs no).

  • CRT therapy (yes vs no).

  • NYHA functional class (NYHA I‐II vs III‐IV).

  • Gender (male vs female).

  • Mean estimated Glomerular Filtration Rate (GFR) (< 60 mL/min/1.73 m² vs ≥ 60 mL/min/1.73 m²).

  • Mean baseline left ventricle ejection fraction (measured by echocardiography) (LVEF ≥ 25% vs LVEF < 25%).

  • Mean HF duration (< 18 months vs ≥ 18 months).

  • Diabetes mellitus (yes vs no)

  • Follow‐up duration of studies (< 24 months vs 24 to 36 months vs > 36 months).

  • Year of publication of RCT (studies published before or in 2003 vs studies published from 2004 onwards)

The paradigm of HF treatment has shifted over time, and the introduction of beta blockers to the OMT regimen is largely due to the publication of large RCTs in the early 2000s that showed improved mortality benefit with the use of these drugs. Since randomisation of patients in ICD trials usually takes two to three years, we will consider 2003 as the time point of comparison. Hence, trials that assessed ICDs vs OMT that were published before or in 2003 did not consider beta‐blockers as part of the OMT. On the other hand, trials published from 2004 onwards considered beta‐blockers as part of the OMT.

We will contact the corresponding authors of studies for information about subgroups whenever they are not well defined and data are not available separately.

The outcomes evaluated in the subgroup analysis are the same ones used for the general group.

We will use the formal test for subgroup interactions in RevMan 5 (RevMan 2014).

Sensitivity analysis

We will perform a sensitivity analysis to assess the potential impact of bias. We will perform a ʽRisk of bias' assessment using the Cochrane ʽRisk of bias' assessment tool and we will exclude all studies at high risk of bias (Higgins 2011). This is to ensure that a sensitivity analysis is conducted on studies that are at low risk of bias from any source.

The nature of the intervention makes it difficult to blind patients and healthcare personnel, albeit not impossible. In fact, some trials that also studied the effect of device therapy, though not necessarily ICD, versus pharmacological therapy blinded patients with regards to the treatment received by having the device switched on or switched off (Young 2003). Also, in some studies that compared CRT‐Ds to CRT‐Ps, patients were blinded to the type of device they had received (Higgins 2003).

Also, we will perform a sensitivity analysis to assess the potential impact of missing data using recent GRADE guidance (Guyatt 2017).

Reaching conclusions

We will base our conclusions only on findings from the quantitative or narrative synthesis of included studies for this review. We will avoid making recommendations for practice and our implications for research will suggest priorities for future research and outline what the remaining uncertainties are in the area.