Exercise‐based cardiac rehabilitation for adults after heart valve surgery
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
To assess the benefits and harms of exercise‐based intervention programmes (exercise‐based interventions alone or in combination with psycho‐educational components), compared to no intervention, or treatment as usual, in adults who have had heart valve surgery. In this review we will focus on programmes that include an exercise‐based intervention with, or without, another rehabilitation component (such as a psycho‐educational component).
Background
Description of the condition
Heart valve diseases account for one‐third of heart diseases, with increasing prevalence due to an ageing population and advances in treatment methods. Previously, heart valve diseases were typically caused by rheumatic heart disease, but now most are degenerative in nature (Nkomo 2006). The overall prevalence of heart valve diseases is widely discussed, as exact estimates do not exist, probably because studies have largely focused on hospitalised patients (Iung 2003), and due to the diagnostic inaccuracy of echocardiography (Nkomo 2006). Prevalence in the USA is 2.5% and is likely to be similar in Europe, although divergent counts exist worldwide, for example, in different countries prevalence for aortic regurgitation ranges from 0% to 39% (Supino 2006).
Heart valve disease is either left‐sided (aortic and mitral valve disease), which is more common, right‐sided (tricuspid and pulmonary valves), or a combination of the two. Heart valve disease is often asymptomatic initially, but, when they present, symptoms include dyspnoea (difficulty breathing), fatigue, fluid retention and decreased physical capacity. Symptomatic heart valve disease severely impacts quality of life and physical function, and is associated with significant morbidity and mortality (Ben‐Dor 2010). Treatment includes medical stabilisation with echocardiographic follow‐up (Vahanian 2012). The treatment of choice when severe symptoms present is valve surgery to repair existing valves, or replacement with biological or artificial valves (Nishimura 2008; Vahanian 2012).
The changing disease pattern and expected increase in healthcare burden of people after heart valve surgery seem to require a well‐established healthcare system, and an after‐care programme to support the patient's post‐surgical problems. These include both physical and psychological issues and the challenge of returning to work.
Before valve surgery, inactivity due to dyspnoea and physical incapacity is common. After surgery, people are often immobilised due to hospitalisation, possible post‐surgery complications, and restrictions designed to assist healing of the sternum can mean that physical capacity can decrease further. As open heart surgery is an extraordinary, and very stressful, life event (Karlsson 2010), quality of life may be affected (Hansen 2009), with some patients experiencing mental problems including depressive symptoms and anxiety (Fredericks 2012). A Cochrane review (Whalley 2011) showed that people who had undergone surgery for a coronary artery bypass graft might benefit from psychological interventions, though, the bias risk of the study was considered to be high (Whalley 2011). Little is known about the effects of psychological interventions in people after heart valve surgery.
In summary, the possible physical or mental recovery problems experienced after heart valve surgery may include affected quality of life, increased healthcare costs, readmission(s) to hospital, loss of earnings, and increased morbidity and mortality. Exercise‐based cardiac rehabilitation might affect all of these problems positively, but evidence for this is lacking.
Description of the intervention
Cardiac rehabilitation is a comprehensive complex intervention including components of exercise training, education, psychosocial management and a behaviour‐modification programme designed to improve the physical and emotional conditions of people with heart disease (Piepoli 2010). Cardiac rehabilitation can also include patient assessment, nutritional counselling, and risk factor management for lipids, blood pressure, weight, diabetes mellitus, and smoking cessation (Piepoli 2010).
European guidelines for people after heart valve surgery recommend the importance of rehabilitation that includes exercise training, anticoagulant therapy, and medical and echocardiographic follow‐up, but do not recommend psycho‐educational interventions as part of the rehabilitation programmes (Butchart 2005). American guidelines do not describe cardiac rehabilitation after heart valve surgery (Vahanian 2012).
No specific information exists about how physical exercise should be performed by people after heart valve surgery. The European Society of Cardiology recommends that physical exercise for people with cardiovascular disease should consist of 150 minutes per week (that is, two‐and‐a‐half hours), while others recommend three to four hours per week (Piepoli 2010). Further recommendations describe that low‐risk patients should perform 30 minutes of aerobic exercise daily in order to achieve a weekly expenditure of 1000 kcal, whereas high‐risk patients should have the amount of physical activity individually prescribed (Gianuzzi 2003). Preferably, the physical exercise should consist of submaximal endurance training (that is, starting at an intensity of 50% of maximum load), the intensity of which is increased over time, and the programme expanded to also include weight/resistance training. The uncertainty surrounding physical exercise for people after heart valve surgery is reflected in the very different training protocols employed in randomised clinical trials and observational studies that have investigated the effect of physical exercise for this group of people.
We have not identified any international consensus reports (for example from the World Health Organization), or guidelines from developing countries that describe detailed recommendations for physical exercise after heart valve surgery.
How the intervention might work
After heart valve surgery, patients often present with a low tolerance of exercise, and low physical capacity. These people have been inactive prior to surgery, then endured a period of bedrest and hospitalisation. At present the effect of cardiac rehabilitation with physical exercise on total mortality, morbidity, and quality of life after heart valve surgery remains uncertain. The existing evidence from both randomised clinical trials and observational studies indicates that exercise‐based interventions for people after heart valve surgery have a positive effect on physical recovery, blood pressure (decreases), New York Heart Association class (decreases) and left ventricular ejection fraction (increases) (Gohlke‐Bärwolf 1992; Landry 1984; Meurin 2005; Newell 1980; Sire 1987).
Exercise training for all cardiac patients ‐ as well as after heart valve surgery ‐ might have direct benefits on the heart and coronary vasculature, including myocardial oxygen demand, endothelial function, autonomic tone, coagulation and clotting factors, inflammatory markers, and the development of coronary collateral vessels (Clausen 1976; Hambrecht 2000). A study that included heart valve surgery patients as well as other cardiac patients, found that physical exercise had positive effects on exercise duration time, the intensity of exercise performed (measured by heart rate), and relative increase in oxygen uptake (Vanhees 2004).
As the evidence is sparse, we might expect to see the same type of effects for exercise after heart valve surgery as are seen in people with heart failure (HF) and ischaemic heart disease (IHD). Two Cochrane reviews have shown that exercise‐based cardiac rehabilitation has a number of positive effects (Davies 2010; Heran 2011). The Heran 2011 review investigated exercise‐based rehabilitation in people with IHD and indicated an overall reduction in all‐cause and cardiovascular mortality and hospital admissions in the shorter term (< 12 months follow‐up) compared with usual care. However, the risk of bias was high, as the reporting of the methodology and outcomes in many of the included trials was categorised as poor. Furthermore, the review found that cardiac rehabilitation did not reduce the risk of total myocardial infarction, coronary artery bypass graft or percutaneous transluminal coronary angiography. Meta‐analysis of seven out of the ten trials included in this Cochrane review showed that health‐related quality of life level significantly improved with exercise‐based cardiac rehabilitation compared to usual care that did not include an exercise‐based intervention (Heran 2011). Again, the risk of bias was high, as the reporting of the methodology and outcomes in many of the included trials was categorised as poor. Moreover, there are always risks of random errors that have not been accounted for in cumulative meta‐analyses (Thorlund 2011; Wetterslev 2008).
The other Cochrane review investigated people with heart failure (Davies 2010), and reported that exercise‐based cardiac rehabilitation seemed to be effective at reducing total and cardiovascular mortality in long‐term studies and hospital admissions in shorter‐term studies, but does not affect total myocardial infarction or revascularisation. The risk of bias in this review was judged to be high, as the overall quality of trials was reported to be poor. Furthermore, the risks of random errors were ignored (Thorlund 2009; Thorlund 2011; Wetterslev 2008).
Apart from the effects mentioned in the Cochrane reviews for people with IHD and HF that might be expected to apply to people after heart valve surgery, exercise‐based rehabilitation might also reduce the symptom burden, improve symptom and disease management, and decrease rates of anxiety and depression.
Possible harmful effects of physical exercise in people after heart valve surgery have not been investigated in randomised clinical trials as far as we know. We can expect valve surgery‐specific adverse events (e.g. arrhythmias, pericardial exudate, arterial embolism, death), as well as adverse events associated with valve disease (e.g. any arrhythmias, heart failure, death). A French prospective study of rehabilitation after cardiac surgery, that included coronary artery bypass grafting and valvular surgery, found a severe cardiac event rate of 1/49,565 patient‐hours of training, which the authors considered to be low (Pavy 2006).
Why it is important to do this review
National and international guidelines recommend physical rehabilitation after heart valve surgery (Butchart 2005; Vahanian 2012). Whilst randomised clinical trials (Landry 1984; Newell 1980; Sire 1987), and a narrative review of exercise‐based rehabilitation programmes for heart valve surgery patients have been conducted (Kiel 2011), we have not been able to identify any systematic reviews or meta‐analysis of these trials. Therefore, the benefits and harms of exercise‐based rehabilitation programmes for adults after heart valve surgery are unclear.
Objectives
To assess the benefits and harms of exercise‐based intervention programmes (exercise‐based interventions alone or in combination with psycho‐educational components), compared to no intervention, or treatment as usual, in adults who have had heart valve surgery. In this review we will focus on programmes that include an exercise‐based intervention with, or without, another rehabilitation component (such as a psycho‐educational component).
Methods
Criteria for considering studies for this review
Types of studies
Randomised clinical trials irrespective of language of publication, publication year, publication type, and publication status will be eligible for inclusion in the review for assessment of benefits and adverse events. Observational studies that we identify in our searches for randomised clinical trials will be included for assessment of adverse events.
Types of participants
Adults (18 years or older), of both sexes and of any ethnicity, who have undergone heart valve surgery for any cause of heart valve disease (i.e. aortic valve disease; mitral valve disease; tricuspid or pulmonary valve disease, or both), and received either heart valve replacement, or heart valve repair (surgery to the valve and the related anatomical areas without valve replacement, e.g., mitraclips, mitral ring, chordae rupture) will be included.
Types of interventions
Experimental interventions
The experimental intervention must include an exercise‐based intervention for adults after heart valve surgery. Interventions that are 'exercise‐based' are defined as being a supervised, or unsupervised, programme conducted in an inpatient, outpatient, community, or home setting that includes any kind of exercise training. The intervention must include a physical exercise component that focuses on increasing exercise capacity, and it may include a psycho‐educational intervention that focuses on improving mental health and the patient's self‐management skills. Trial participants will engage in the exercise intervention before or after discharge from hospital for heart valve surgery. The rehabilitation programme must include a post‐surgical element and may include a pre‐surgical element in advance of surgery. There will be no restrictions in the length, intensity, or content of the training programme.
Control interventions
We will include the following control interventions:
-
treatment as usual (e.g. standard medical care, such as drug and anticoagulant therapy and medical follow‐up with echocardiography);
-
no intervention;
-
any other type of cardiac rehabilitation programme, as long as it does not include a physical exercise element.
Co‐interventions
We will include trials with co‐interventions other than rehabilitation of any kind, as long as these are identical and delivered equally in the experimental and the control groups. Co‐interventions can include anything other than the experimental intervention (e.g. drug delivery, surgery techniques (percutaneous versus transthoracic surgery), or dietary interventions).
Types of outcome measures
We will assess all outcomes at two time points:
-
end of the intervention (as defined by the trialists);
-
longest available follow‐up.
There will be no minimum length of follow‐up for the studies that are eligible for the review.
Primary outcomes
-
Mortality: all‐cause mortality and cardiovascular mortality.
-
Serious adverse events, defined as any untoward medical occurrence that is life threatening, results in death, or is persistent or leads to significant disability; or any medical event that has jeopardised the patient or required intervention to prevent it, or any hospital admission or prolongation of existing hospital admission (ICH‐GCP 1997).
-
Health‐related quality of life using generic or disease‐specific validated instruments, e.g. Short Form‐36, EQ‐5D, HeartQoL.
Secondary outcomes
-
Proportion of participants with symptoms measured by New York Heart Association (NYHA) class III‐IV.
-
Ejection fraction: mean values in the experimental and control groups will be compared.
-
Exercise capacity: any measure of exercise capacity including direct measurement of VO2 peak/VO2 max or indirect measures such as exercise time, walking distance etc.
Search methods for identification of studies
Electronic searches
We will search the following electronic databases from their inception to present to identify primary studies:
-
The Cochrane Central Register of Controlled Trials (CENTRAL);
-
The Database of Abstracts of Reviews of Effectiveness (DARE);
-
MEDLINE (Ovid);
-
EMBASE (Ovid);
-
CINAHL (EBSCO);
-
PsycINFO (Ovid);
-
LILACS (Bireme);
-
Conference Proceedings Citation Index‐S (CPCI‐S) on Web of Science (Thomson Reuters).
The preliminary search strategy for MEDLINE (Ovid) will be translated for use in the other databases (Appendix 1). The Cochrane sensitivity‐maximising RCT filter will be applied to MEDLINE (Lefebvre 2011), and adaptations of it to the other databases where applicable.
Searching other resources
We will apply no language restrictions. Studies written in languages that the author group do not understand will be translated professionally. We will search for ongoing trials on:
-
ClinicalTrial.gov (www.clinicaltrials.gov);
-
The World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) search platform (http://apps.who.int/trialsearch/).
The reference lists of relevant publications will be checked for any unidentified randomised trials.
Data collection and analysis
Selection of studies
Two authors (KLS and LT) will independently read the titles and abstracts of potentially relevant papers retrieved by the searching activities described above. If in doubt about whether a paper is relevant, we will read the full article. We will retrieve full publications of potentially relevant studies and translate them where required, two authors (KLS and LT) will then determine study eligibility independently using a standardised inclusion form. We will exclude studies that do not meet the inclusion criteria. Full text copies of all potentially relevant studies will be retrieved and stored electronically. We will resolve any disagreements by discussion, between the two authors (KLS and LT), and where necessary, a third author (ADZ) will be asked to mediate.
Data extraction and management
Two authors (KLS and LT) will independently extract data from the identified abstracts and papers using standardised data extraction forms. Where data are presented numerically (in tables or text) and graphically (in figures), we will use numeric data, because of the possibility of making measurement errors when estimating from graphs. A third author (ADZ) will confirm all numeric calculations and extractions from graphs or figures. We will resolve any discrepancies by consensus.
We will extract the following data.
-
General information: publication status, title, authors' names, source, country, contact address, language of publication, year of publication, duplicate publication, financial conditions.
-
Study characteristics: design and duration.
-
Intervention: type of physical exercise, type of rehabilitation programme (does the programme/rehabilitation programme/intervention consist of only physical exercise or are there any other components included, for example, psycho‐educational intervention, diet‐intervention, behavioural intervention), setting (e.g., inpatient, outpatient, community, home setting, or a combination), time after hospitalisation, type of control intervention.
-
Participants: sampling method (e.g., convenience, random, etc.), inclusion and exclusion criteria, number of participants in intervention and control groups, participant demographics such as sex and age, baseline characteristics including type of valve affected and classification of heart valve disease, and number of participants lost to follow‐up.
-
Outcomes: mortality (all‐cause mortality, cardiovascular mortality), serious morbidity defined as any hospital readmission, adverse events, serious adverse events defined as above, health‐related quality of life using generic or disease‐specific validated instruments.
-
Risk of bias: please see Assessment of risk of bias in included studies below.
We will compare data from each intervention group of each parallel group trial.
If any cross‐over randomised clinical trials are identified, we will only use data from the first phase of the trial (i.e., before the cross‐over).
Assessment of risk of bias in included studies
Two authors (KLS and LT) will independently assess the risk of bias in the included studies using The Cochrane Collaboration's recommended tool for assessment of risk of bias (Higgins 2011b). As it is impossible to blind participants and trial staff for this intervention, when we interpret results from the domain 'Blinding of participants and personnel' we will take the risks of bias due to lack of blinding of participants and of personnel into consideration, and bear it in mind when we assess intervention effects (Savovic 2012; Wood 2008). We will provide assessments of risk of bias in the 'Risk of bias' table for each study.
Generaton of allocation sequence
Low risk of bias:allocation sequence generated using a computer random‐number generator or a table of random numbers. Drawing lots, tossing a coin, shuffling cards or envelopes, and throwing dice are all adequate, if performed by an independent adjudicator, or the method is unlikely to introduce selection bias.
Uncertain risk of bias: insufficient information provided to assess whether the method used could cause bias.
High risk of bias: the method used is improper and likely to be confounding (e.g. there is a non‐random component in the generation of the allocation sequence).
Allocation concealment
Low risk of bias: the method used will probably not cause bias on the final observed effect (e.g. allocation is controlled by a central and independent randomisation unit and the assignment cannot be foreseen).
Uncertain risk of bias:insufficient information about concealment of allocation provided to assess whether the method used could cause bias on the estimate of the effect.
High risk of bias: the method used will probably cause bias on the final observed effect (e.g. the allocation sequence is open and known to the investigators).
Blinding of outcome assessment
Low risk of bias: if the trial investigators performing the outcome assessments, analyses, and calculations are blinded to the treatment allocation, and this is described.
Uncertain risk of bias: if the procedure of blinding is insufficiently described.
High risk of bias:if blinding is not performed, or the procedure cannot be classified as 'low risk of bias'.
Incomplete outcome data
Low risk of bias:the number and reasons for dropouts and withdrawals are properly described, and valid methods have been used to handle missing data.
Uncertain risk of bias:the study made the impression of having no dropouts or withdrawals, but this aspect was described insufficiently.
High risk of bias: the crude estimate of effects will be biased if the effects are concluded on missing or incomplete data (e.g. dropouts or withdrawals), or the methods being used to handle missing data were unsatisfactory.
Selective outcome reporting
Low risk of bias: all primary and clinically relevant outcomes of the trial have been reported.
Uncertain risk of bias:not all primary or clinically relevant outcomes are reported, or reported sufficiently, or whether these outcomes were recorded is unclear.
High risk of bias:not all primary or clinically relevant outcomes reported.
Performance bias
Low risk of bias:any co‐interventions are delivered equally across intervention and control groups.
Uncertain risk of bias:there is insufficient information to assess whether co‐interventions were present, or equally delivered across groups, and that could put the trial at a risk of bias.
High risk of bias:the co‐interventions are not delivered equally across intervention and control groups.
For‐profit‐bias
Low risk of bias: the trial appears to be free of industry sponsorship or other kind of for‐profit support that may manipulate the trial design, conduct, or results of the trial. Any co‐interventions are delivered equally across intervention and control groups.
Uncertain risk of bias: the trial may or may not be free of for‐profit bias, as no information about clinical trial support or sponsorship is provided.
High risk of bias:the trial is sponsored by a related industry, or has received other kind of for‐profit support.
Blinding of participants and personnel
Low risk of bias:any intervention is delivered blinded to the participants or personnel, or both, and neither the participants nor the personnel are aware of the group to which participants are allocated.
Uncertain risk of bias:there is insufficient information to assess whether the participants or personnel are blinded to the intervention.
High risk of bias: the patients and personnel are not blinded to the intervention.
Due to the type of intervention being investigated in this review, we will expect a high level of bias for this domain. This will apply to all studies, as it is impossible to blind participants when the intervention consists of physical exercise.
Overall risk of bias
A trial will be categorised as being at overall 'low risk of bias' if the trial is rated as being at 'low risk of bias' for all the risk of bias domains listed above. Likewise, a trial will be categorised as being at overall 'high risk of bias' if the risk of bias is rated as either 'uncertain risk of bias' or 'high risk of bias' for any domain listed above. Furthermore, as we expect all trials to be categorised as being at overall high risk of bias, because it is not possible to blind participants and personnel to this type of intervention, we will also categorise trials as being at overall 'lower risk of bias' and 'higher risk of bias'. A trial will be categorised as having an overall lower risk of bias if it is rated as having a low risk of bias in all the risk of bias domains listed above except for blinding of participants and personnel. All other trials will be categorised as being at overall 'higher risk of bias'.
Small study (publication) bias
Where possible, we will construct funnel plots for each outcome, to establish the potential influence of small study effects and potential publication bias. We will not use funnel plots for outcomes for which there are ten or fewer trials, or where all trials are of similar sizes (Sterne 2011).
Measures of treatment effect
Dichotomous data will be expressed as risk ratios (RR) with 95% confidence intervals (CI). When studies use continuous scales of measurement to assess the effects of the intervention, mean differences (MD) will be used, or, when studies use different scales or measurements, the standardised mean difference (SMD) will be used (Thompson 2002).
Unit of analysis issues
If any cluster‐randomised clinical trials are included, we will contact the trial authors to obtain an estimate of the intra‐cluster correlation (ICC) where appropriate adjustments for the correlation between participants within clusters have not been made, or impute it using estimates from the other included trials, or from similar external trials. We will inflate the trial standard errors.
Dealing with missing data
An attempt will be made to obtain missing data by contacting the authors of the trials. If data remain unavailable, the impact of the missing data will be discussed.
For dichotomous outcomes, analyses will be made according to the intention‐to‐treat method (Higgins 2011c), which includes all participants irrespective of compliance or follow‐up. For the primary analyses, we will assume that participants lost to follow‐up are alive, and have no serious adverse events. For continuous outcomes we will perform available patient analysis and include data only on those for whom results are known (Higgins 2011c). If it is not possible to obtain SDs either from authors or by calculation, the missing data will be imputed by using SDs from other included trials, specifically trials with a low risk of bias (Furukawa 2006). We will seek to undertake sensitivity analyses for binary outcomes to examine the impact of loss to follow‐up.
Assessment of heterogeneity
Clinical heterogeneity will be explored by comparing the population, experimental intervention and control intervention. Statistical heterogeneity will be observed in the trials both by visual inspection of a forest plot, and by using a standard Chi2 value with a significance level of P = 0.10. Heterogeneity will be assessed by the I2 statistic. An I2 estimate greater than or equal to 50% with a statistically significant value for Chi2, will be interpreted as evidence of a substantial problem with heterogeneity (Higgins 2011a). If this is the case, we will explore reasons for heterogeneity. If there is high inconsistency, and clear reasons for this are found, we will present data separately.
Assessment of reporting biases
Different forms of reporting bias will be handled according to recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). Funnel plots will be used to give a visual assessment of whether intervention effects are associated with the size of the study.
Data synthesis
Data synthesis will be performed according to recommendations in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), using Review Manager (RevMan 2011), and Trial Sequential Analysis (TSA 2011), for statistical analysis. Meta‐analyses will be made using both a random‐effects model and fixed‐effect model (Deeks 2011; DeMets 1987; DerSimonian 1986). If there are discrepancies between results from the two models, both sets of results will be presented, otherwise we will report the results from the random‐effects model only. Should it be inappropriate, or not possible, to do a meta‐analysis, we will perform a narrative assessment.
Trial sequential analysis
We will undertake 'trial sequential analysis’ to assess the risk of random errors, because cumulative meta‐analyses are at risk of producing random errors due to sparse data and repetitive testing on the accumulating data (Thorlund 2009; Thorlund 2011; TSA 2011; Wetterslev 2008). The underlying assumption of trial sequential analysis is that testing for significance may be performed each time a new trial is added to the meta‐analysis. We will add the trials according to their year of publication, and, if more than one trial has been published in a year, trials will be added alphabetically according to the last name of the first author (Wetterslev 2008).To minimise random errors, calculation of the required information size is planned (that is, the number of participants needed in a meta‐analysis to detect or reject a certain intervention effect) (Wetterslev 2008). The information size calculation should also account for the diversity present in the meta‐analysis (Wetterslev 2009).
In our meta‐analysis, the required information size for binary outcomes will be based on the assumption of a plausible RR reduction of 20% from the proportion with the outcome in the control group, or on the RR reduction observed in the included trials with low risk of bias (Wetterslev 2008). For continuous outcomes, we will test a difference of 0.5 SD using the SD from the control groups. As default, we will use a type I error of 5%, a type II error of 20%, and adjust information size for diversity, unless otherwise stated (Wetterslev 2008; Wetterslev 2009).
Trial sequential monitoring boundaries can be constructed on the basis of the required information size and the risks for type I (5%) and type II (20%) errors (Thorlund 2011; Wetterslev 2008). These boundaries will determine the statistical inference that can be drawn regarding the cumulative meta‐analysis that has not reached the required information size: if the trial sequential monitoring boundary is crossed before the required information size is reached, it is possible that firm evidence may be established and further trials may turn out to be superfluous. On the other hand, if the boundary is not surpassed, it is most probably necessary to continue conducting trials in order to detect or reject a certain intervention effect.
Subgroup analysis and investigation of heterogeneity
We will perform subgroup analysis to analyse the primary outcomes, using stratified meta‐analysis, according to the following:
-
trials at overall low risk of bias compared to trials at overall high risk of bias; if no trials are categorised as being at overall low risk of bias, we will perform subgroup analysis on trials at overall lower risk of bias compared to trials at overall higher risk of bias;
-
trials including women only versus trials including men only;
-
trials including younger patients only versus trials including older patients only;
-
trials with an exercise intervention only, compared to trials with an exercise intervention plus any other co‐intervention, such as a psycho‐educational intervention.
Sensitivity analysis
For the primary outcomes, we plan to perform the following sensitivity analyses.
Binary outcomes
Best/worse‐case scenario: for this analysis we will assume that all participants lost to follow‐up in the experimental group have survived, and have had no serious adverse events; and all those with missing outcomes in the control group have not survived, and have had serious adverse events.
Worst/best‐case scenario: for this analysis it will be assumed that all participants lost to follow‐up in the experimental group have not survived, and have had serious adverse events; and all those with missing outcomes in the control group have survived, and have had no serious adverse events.
Continous data
Assumptions for lost data: where assumptions have to be made for lost data (see Dealing with missing data), we will compare the findings from our assumptions with data only from those participants who completed the trials.
