Discussion
This systematic review shows that when compared to the no-exercise control, exercise interventions alone or as a component of comprehensive CR programme does not reduce or increase short-term (up to 12 months follow-up) all-cause mortality. We saw reductions in the risk of hospitalisation due to HF (RR reduction: 25%, 95% CI 8% to 38%) and improvements in health-related quality of life following exercise interventions. In trials reporting the Minnesota Living with Heart Failure questionnaire, those undertaking exercise were on an average 5.8 points higher than controls. A difference of four points or larger on the Minnesota Living with Heart Failure questionnaire has been shown to represent a clinically important, meaningful difference for patients.59 While the majority of included participants in this review were HFREF and NYHA class II and III, recent trials have recruited those who with HFPEF and NHYA IV, and a greater proportion of females and older patients. We found the benefits of exercise-CR appear to be independent of type of exercise CR (exercise only vs comprehensive CR, aerobic exercise only vs aerobic and resistance exercise, average dose of exercise intervention), and trial characteristics (ie, length of follow-up, overall risk of bias, publication date).
Many trials included in this review have been conducted in the era of contemporary medical therapy for HF. For example, in the large multicentre HF-ACTION trial, 94% of patients were receiving β-blockers and angiotensin-receptor blocker or ACE inhibitors, and 45% had an implantable cardioverter defibrillator or implanted biventricular pacemaker at the time of enrolment.35 Given the proven survival advantage of these medical treatments, it might be expected that any incremental all-cause mortality benefit with exercise is likely to be small. Nevertheless, there was a trend (p=0.09) towards a reduction in all-cause mortality with exercise training in the six trials reporting outcomes beyond 12 months.
The improvements in health-related quality of life with exercise training seen in this review are in accordance with the previous systematic review of van Tol et al60 but not with that of Chien et al,61 which focused on home-based exercise training and concluded that exercise training compared with usual care did not improve the health-related quality of life of patients with HF. However, the review by Chien et al was limited to three trials in 198 patients. Our metaregression analysis showed no difference in the magnitude of the reduction in hospitalisations and improvement in health-related quality of life with exercise training in those studies based in a hospital setting compared with those based in a home setting.
Study limitations
The general lack of reporting of methods in the included RCT reports made it difficult to assess their methodological quality and thereby, judge their risk of bias. There appeared to be improvement in the quality of reporting in recent trials. Funnel plot asymmetry for health-related quality of life outcomes indicated evidence of small study bias and therefore, possible publication bias. Future trials need to also provide fuller details of the interventions and controls in accordance with Consolidated Standards of Reporting Trials (CONSORT) extension for trials assessing non-pharmacological treatments.62 A specific goal of this update review was to clarify the impact of exercise training programmes on clinical events; many included trials were relatively small and of short-term follow-up so that the number of deaths and hospitalisations reported by the majority of trials was small. The majority of trials reported low numbers of deaths and hospitalisations. There was evidence of larger treatment effect for health-related quality of life outcomes in studies judged to be at higher risk of bias compared with lower risk of bias studies, suggesting that risk of bias may be a major driver of the substantive statistical heterogeneity seen across trials in this outcome. The majority of trials in this review have investigated exercise training as a single intervention and against a no-exercise control. However, in practice, exercise-based CR is often an adjunct to other HF management interventions, such as specialist HF nurse support or disease management programmes. While trials have demonstrated the benefits of such HF management interventions alone, few trials have compared such interventions with and without adding a structured exercise training programme.63 ,19 This is an important question for the future design of HF services because the addition of an exercise CR programme can add considerably to staffing and equipment costs. An individual participant data meta-analysis (ExTraMATCH II), using the RCTs identified in this review, is currently underway to clarify the patient and intervention characteristics that may drive variation in outcomes with exercise-based CR for HF.64
In conclusion, this updated Cochrane review shows that improvements in hospitalisation and health-related quality of life with exercise-based CR appear to be consistent across both patients with HF regardless of type of CR programme and may reduce mortality in the longer term. Individual participant data meta-analysis is needed to provide confirmatory evidence of the importance of patient subgroup and exercise programme characteristics on outcomes.
In addition to improving the quality of reporting, future clinical trials of exercise-based interventions in HF need to consider the generalisability of trial populations (women, older people and people with HFPEF remain under-represented in trial populations) and interventions to enhance the long-term maintenance of exercise-based CR programmes, as well as the outcomes, costs and cost-effectiveness of programmes delivered exclusively in a home-based setting.