Article ID: CJ-23-0820
Background: This study aimed to clarify the effects of exercise-based cardiac rehabilitation (CR) on patients with heart failure.
Methods and Results: Patients were divided into groups according to intervention duration (<6 and ≥6 months). We searched for studies published up to July 2023 in Embase, MEDLINE, PubMed, and the Cochrane Library, without limitations on data, language, or publication status. We included randomized controlled trials comparing the efficacy of CR and usual care on mortality, prehospitalization, peak oxygen uptake (V̇O2), and quality of life. Seventy-two studies involving 8,495 patients were included in this review. It was found that CR reduced the risk of rehospitalization for any cause (risk ratio [RR] 0.80; 95% confidence interval [CI] 0.70–0.92) and for heart failure (RR 0.88; 95% CI 0.78–1.00). Furthermore, CR was found to improve exercise tolerance (measured by peak V̇O2 and the 6-min walk test) and quality of life. A subanalysis performed based on intervention duration (<6 and ≥6 months) revealed a similar trend.
Conclusions: Our meta-analysis showed that although CR does not reduce mortality, it is effective in reducing rehospitalization rates and improving exercise tolerance and quality of life, regardless of the intervention duration.
Heart failure (HF) is a complex syndrome caused by structural and/or functional cardiac abnormalities.1 It is one of the leading causes of morbidity and mortality, with a significant financial and social burden, and affects up to approximately 40 million individuals worldwide according to the Global Burden of Disease Study.2 The incidence of HF has increased dramatically over the past few decades and is expected to continue to rise over the next 2 decades.3,4
Patients with HF experience a substantial burden that includes exercise intolerance, poor health-related quality of life, mortality, increased hospital admissions, and higher healthcare costs.5 Cardiac rehabilitation (CR) plays an important role in managing patients with HF by improving physical function, reducing symptoms, and reducing the risk of future events. A meta-analysis by Cochrane showed that exercise-based CR for patients with HF may reduce their long-term all-cause mortality risk by 12% after a 12-month follow-up period.6 Moreover, the study revealed that this type of rehabilitation can also improve patients’ quality of life.6 However, most randomized controlled trials (RCTs) in that Cochrane review showed that exercise-based CR lasted only 3–6 months. In addition, in the context of undergoing CR under Japan’s healthcare insurance system, the prescribed threshold for the maximum duration is set at a period of 5 months. Therefore, the effects of longer-term exercise-based CR remain unclear. Belardinelli et al conducted a study on a 10-year exercise intervention and found that sustained exercise intervention reduced the risk of rehospitalization for HF by 36% compared with a non-exercise group while maintaining higher levels of quality of life and exercise tolerance.7 Although a Cochrane meta-analysis calculates the dose of exercise using a calculation that considers the total number of weeks of training, the mean number of sessions per week and the mean duration of sessions in minutes, all these factors are aggregated; therefore, the effect of long-term rehabilitation is unknown.
Thus, the aim of the present study was to evaluate the effects of exercise-based CR on patients with HF, who were divided into groups according to intervention duration (≥6 or <6 months).
This meta-analysis was registered at PROSPERO (Registration no. CRD 42022335271). The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) 2020 statement was as guide for this meta-analysis.8
Search Methods for Identifying StudiesWe conducted this study according to the PRISMA Statement.8 We searched for studies published up to July 2022 in the Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, and PubMed electronic databases. Based on discussions with an academic librarian, we finalized the search strategy and modified it appropriately for each database. The search strategy comprised a combination of free text words, words used in titles and/or abstracts and medical subject headings. The search strategy was limited to the inclusion of peer-reviewed publications that involved only human participants. To ensure that the search strategy covered all topic-related studies, we included different terms and spellings used worldwide. Furthermore, language restrictions were not applied.
Inclusion and Exclusion CriteriaWe included individual RCTs. Cluster RCTs were eligible for inclusion, but none were identified. We excluded cross-over trials and included full-text studies, studies for which only the abstract was published and unpublished data. We included participants aged ≥18 years diagnosed with HF. We planned that if a study included only a subset of eligible participants, we would ask the study authors to provide data on the subset of interest; if we could not obtain data for the subset of eligible participants, they would be excluded from the analysis. We excluded participants with an implantable ventricular assist device and those who underwent heart transplantation. When information on such histories was not included in the paper, the authors of the studies were contacted to obtain the missing information.
The primary outcome measures for this review were mortality and rehospitalization. Data were also collected on the following secondary outcomes: exercise capacity, quality of life, and adverse events. All quality of life assessment scales were included in the analysis.
Data ExtractionA data extraction form was used to extract study information and outcome data. Data were extracted by 4 authors divided into 2 teams (S.T. and Y.T.; and M.O. and M.S.). Disagreements within each team were resolved by discussion or by involving a third person (S.Y.). Two authors (S.T. and M.O.) transferred data from the data extraction form to the Review Manager file. The accuracy of data entry was double-checked by verifying the data presented in the systematic review against those presented in the research reports. Another author (S.Y.) randomly checked the extracted data and patient characteristics against their source to ensure accuracy. When the data were insufficient, the authors of the original paper were contacted and unpublished data were requested. Wherever possible, data were collected on an intention-to-treat basis.
Assessment of the Risk of Bias in the Included StudiesThe quality of each study was independently assessed by 4 authors (S.T., Y.T., M.O., and M.S.) using the Risk of Bias Tool 2 according to the Cochrane Handbook for Systematic Reviews of Interventions.9 The risk of bias was assessed based on the following criteria: bias arising from the randomization process, bias due to deviations from intended interventions, bias due to missing outcome data, bias in measuring the outcome and bias in selecting the reported result. In cases of differing assessments, a third author (S.Y.) was consulted for a final decision regarding the assessment.
Assessment of Evidence CertaintyTo assess the quality of a body of evidence on studies that contribute data to meta-analyses for prespecified outcomes, we used the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) tool (study limitations, consistency of effect, indirectness, imprecision, and publication bias). In the present study, we applied the methods and recommendations in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions,9 using GRADEpro GDT.
Data Synthesis and AnalysisPatients were divided into 2 groups: those participating in aerobic exercise or resistance training and those not participating in exercise therapy or resistance training. Data syntheses and analyses were performed using Review Manager version 5.3, STATA version 15, or R version 4.0.3. We expressed dichotomous data as risk ratios (RRs) with 95% confidence intervals (CIs). For continuous data, we expressed the results as mean differences (MDs) when the results were measured similarly in different studies. We expressed the results as the standardized MD (SMD) when the results obtained are conceptually the same but use different measurement scales.
Potential publication bias was assessed using funnel plots and Egger’s test. Heterogeneity was assessed among included studies both qualitatively using forest plots and quantitatively using the Chi-squared test of heterogeneity and I2 statistic. Data from each study were pooled using a random-effects model where appropriate. To examine the robustness of the results, we performed meta-analyses using fixed-effects models after attributing less weight to small trials. We used these meta-analyses only if their results differed from those of the random-effects models. When ≥10 studies were included in a meta-analysis, a funnel plot was created, and its asymmetry was visually examined to explore any publication bias. When an I2 score of >50% was obtained, heterogeneity was considered substantial, and subgroup analysis was performed according to the Cochrane Handbook for Systematic Reviews of Interventions.9
We performed subgroup analysis on all outcomes according to intervention duration (≤6 or >6 months). Furthermore, if I2 >50% was obtained, we conducted the following subgroup analyses to explore possible causes of heterogeneity for primary outcomes to increase the validity of the results of the analyses: (1) age (≤65 or >65 years); (2) body mass index (BMI; ≤28 or >28 kg/m2); (3) HF type (HF with reduced ejection fraction [HFrEF], defined as left ventricular ejection fraction [LVEF] ≤45%, or HF with preserved ejection fraction [HFpEF], defined as LVEF >45%); (4) peak V̇O2 at baseline (≤15 or >15 mL/kg/min); and (5) publication year (≤2000 vs. >2000).
We identified 20,243 records after the removal of duplicates and retrieved 19,411 records after the screening of titles and abstracts (Figure 1). We further excluded 65 records: 13 studies were non-RCTs; 16 had inappropriate study designs; 7 had inappropriate interventions; 1 included an inappropriate control; 6 had no relevant outcomes; 21 had no data; and 1 was a review article. Finally, we added 34 new studies10–43 to the previous Cochrane review.6 Seventy-two studies (118 references) involving 8,495 patients met the eligibility criteria of this review.
Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram. CENTRAL, Cochrane Central Register of Controlled Trials; RCT, randomized control trial.
Study Characteristics
The characteristics of the included studies, including participant information, are presented in Table 1. The mean age of most participants was 60 years, which accounted for approximately half the studies. Studies involving participants with a mean age ≥70 years accounted for only approximately 15% of studies. Several studies consisted of only male participants. Furthermore, many participants of the included studies had HFrEF, and only 4 studies were identified in which the participants had HFpEF. The intervention duration tended to be divided into 2 categories: short-term interventions, lasting 3–3.9 months (36% of all studies), and long-term interventions, lasting 6–8.9 months (32% of all studies). Three studies included resistance training only, whereas studies including only aerobic or combined aerobic/resistance training accounted for 80% of all studies. In addition, exercise intensity varied between studies, and was determined using different methods. Trials that were able to perform cardiopulmonary exercise tests typically aimed for 60–80% of peak oxygen uptake (V̇O2), whereas the Borg rating of perceived exertion (RPE) scale was commonly used to assess perceived exertion, aiming for an RPE of 11–15.
Characteristics of the Included Studies, Including Participant Information
| No. studies (%) | |
|---|---|
| Publication year | |
| 1990–1999 | 5 (7) |
| 2000–2009 | 21 (29) |
| 2010–2019 | 32 (44) |
| ≥2020 | 14 (19) |
| Duration of follow-up (months) | |
| <4 | 13 (18) |
| 4–6 | 24 (33) |
| 7–9 | 1 (1) |
| 10–12 | 10 (14) |
| >12 | 12 (17) |
| Unclear | 12 (17) |
| Sample size (no. participants) | |
| <31 | 27 (38) |
| 31–50 | 17 (24) |
| 51–100 | 13 (18) |
| 101–500 | 13 (18) |
| 501–1,000 | 1 (1) |
| >1,000 | 1 (1) |
| Population characteristics | |
| Age categories (years) | |
| 40–49 | 1 (1) |
| 50–59 | 23 (32) |
| 60–69 | 35 (49) |
| 70–79 | 8 (11) |
| 80–89 | 2 (3) |
| Not reported | 3 (4) |
| Sex | |
| Male only | 15 (21) |
| ≥50%f male | 49 (68) |
| >50% female | 7 (10) |
| Female only | none |
| Not reported | 1 (1) |
| BMI (kg/m2) | |
| <25 | 3 (4) |
| 25–29.9 | 27 (38) |
| 30–34.9 | 13 (18) |
| ≥35 | 1 (1) |
| Not reported | 28 (39) |
| LVEF (%) | |
| <40 | 42 (58) |
| 40–50 | 8 (11) |
| >50 | 4 (6) |
| Not reported | 17 (24) |
| Intervention characteristics | |
| Exercise type | |
| Aerobic training only | 33 (46) |
| Resistance training only | 3 (4) |
| Aerobic and resistance training | 24 (33) |
| Interval training | 7 (10) |
| Others | 5 (7) |
| Dose of intervention | |
| Duration (months) | |
| <3 | 8 (11) |
| 3–3.9 | 26 (36) |
| 4–5.9 | 7 (10) |
| 6–8.9 | 23 (32) |
| 9–11.9 | 2 (3) |
| 12–23.9 | 4 (6) |
| ≥24 months | 2 (3) |
| Frequency | |
| <3 times/week | 23 (32) |
| 2–3.9 times/weeks | 35 (49) |
| ≥4 weeks | 13 (18) |
| Not reported | 1 (1) |
| Setting | |
| Center based only | 41 (57) |
| Home based only | 10 (14) |
| Combination of center and home | 21 (29) |
BMI, body mass index; LVEF, left ventricular ejection fraction.
Quality and Bias Assessment in the Included Studies
We found no evidence of funnel plot asymmetry for all-cause mortality, HF-related mortality, all-cause rehospitalization, HF-related hospitalizations, peak V̇O2, or 6-minute walking distance (6MWT; Egger’s test P>0.05 for all). However, we found evidence of asymmetry for quality of life (Egger test P<0.05). Figure 2 summarizes the risk of bias for the outcomes of death and rehospitalization. The risk of bias was high for more than half the outcomes of death and rehospitalization.
Risk of bias summaries of (A) mortality and (B) rehospitalization in each study and risk of bias graphs of (C) mortality and (D) rehospitalization in each domain.
Primary Outcomes
All-Cause Mortality Forty-five trials involving 7,355 participants reported all-cause mortality (Figure 3). We found that the effect of CR on all-cause mortality was uncertain compared with non-CR (RR 0.90; 95% CI 0.79–1.03; I2=0%) and assessed the evidence to be of low quality via the GRADE method. We performed a subgroup analysis for intervention duration ≤6 months (RR 0.91; 95% CI 0.72–1.15; I2=0%) or >6 months (RR 0.90; 95% CI 0.76–1.05; I2=0%), which showed no difference between the groups (test for subgroup differences: P=0.904). We did not perform other subgroup analyses because an I2 >50% was not obtained.
Forrest plots of the effect of cardiac rehabilitation (CR) compared with usual care on all-cause mortality. CI, confidence interval.
HF-Related Mortality Six trials involving 1,592 participants reported HF-related mortality (Figure 4). We found that the effect of CR on HF-related mortality was uncertain compared with non-CR (RR 1.00; 95% CI 0.68–1.48; I2=0%) and assessed the evidence to be of low quality via the GRADE method. We performed a subgroup analysis for the intervention duration of ≤6 months (RR 1.03; 95% CI 0.70–1.52; I2=0%) or >6 months (unavailable because only 1 trial was identified), which showed no difference between the groups (test for subgroup differences: P=0.296). We did not perform other subgroup analyses because an I2 >50% was not obtained.
Forrest plots of the effect of cardiac rehabilitation (CR) compared with usual care on heart failure-related mortality.
All-Cause Rehospitalization Twenty-six trials involving 5,716 participants reported all-cause rehospitalization (Figure 5). We found that participants who underwent CR had a significantly lower risk of all-cause rehospitalization than those who did not undergo CR (RR 0.80; 95% CI 0.70–0.92; I2=43.2%) and assessed the evidence to be of moderate quality via the GRADE method. We performed a subgroup analysis for the intervention duration of ≤6 months (RR 0.90; 95% CI 0.82–0.99; I2=0%) or >6 months (RR 0.72; 95% CI 0.54–0.96; I2=64.2%), which showed no difference between the groups (test for subgroup differences: P=0.154). We did not perform other subgroup analyses because an I2 >50% was not obtained.
Forrest plots of the effect of cardiac rehabilitation (CR) compared with usual care on all-cause rehospitalization.
HF-Related Rehospitalization Twenty-one trials involving 5,139 participants reported HF-related rehospitalization (Figure 6). We found that participants who underwent CR had a significantly lower risk of HF-related rehospitalization than those who did not undergo CR (RR 0.88; 95% CI 0.78–1.00; I2=7.1%) and assessed the evidence to be of low quality via the GRADE method. We performed a subgroup analysis for the intervention duration of ≤6 months (RR 0.89; 95% CI 0.73–1.09; I2=18.1%) or >6 months (RR 0.86; 95% CI 0.69–1.06; I2=3.3%), which showed no difference between the groups (test for subgroup differences: P=0.778). We did not perform other subgroup analyses because an I2 >50% was not obtained.
Forrest plots of the effect of cardiac rehabilitation (CR) compared with usual care on heart failure-related rehospitalization.
Secondary Outcomes
Peak V̇O2 Thirty-four trials involving 2,235 participants reported peak V̇O2 (Table 2). We found that participants who underwent CR had a significant improvement in peak V̇O2 than those who did not (MD 3.18; 95% CI 2.22–4.15; I2=94.9%). We performed a subgroup analysis for the intervention duration of ≤6 months (MD 3.31; 95% CI 2.11–4.51; I2=83.2%) or >6 months (MD 3.09; 95% CI 1.66–4.52; I2=97.2%), which showed no difference between the groups (test for subgroup differences: P=0.818). We performed other subgroup analyses because an I2 >50% was obtained. We performed our planned subanalyses but were unable to resolve the heterogeneity. Furthermore, we performed a meta-regression analysis, which showed a significant negative correlation between the effect size of peak V̇O2 and age (P<0.001).
Subanalyses and Quality of Evidence of Peak Oxygen Uptake, 6MWT, and QOL
| Outcome measures | Peak V̇O2 | MD (95% CI) |
I2 (%) | 6MWT | MD (95% CI) |
I2 (%) | QOL: All measurements |
SMD (95% CI) |
I2 (%) | QOL: MLHFQ | MD (95% CI) |
I2 (%) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| All results | 34 RCTs, n=2,235 | 3.18 (2.22, 4.15) | 94.9 | 24 RCTs, n=2,925 | 33.89 (23.18, 44.60) | 96.0 | 37 RCTs, n=5,194 | 0.79 (0.49, 1.09) | 95.4 | 24 RCTs, n=3,567 | 9.88 (5.38, 14.38) | 97.0 |
| Quality of evidence (GRADE) |
⊕⊖⊖⊖ Very low |
⊕⊖⊖⊖ Very low |
⊕⊖⊖⊖ Very low |
⊕⊖⊖⊖ Very low |
||||||||
| Publication year | ||||||||||||
| <2000 | 5 RCTs, n=184 | 3.44 (2.04, 4.83) | 89.1 | 1 RCT, n=29 | 26.21(−14.76, 67.19) | 0 | 1 RCT, n=94 | 0.65 (0.23, 1.07) | 0 | 1 RCT, n=94 | 0.65 (0.23, 1.07) | 0 |
| ≥2000 | 29 RCTs, n=2,051 | 3.13 (1.89, 4.37) | 95.4 | 23 RCTs, n=2,914 | 34.20 (23.27, 45.14) | 96.2 | 36 RCTs, n=5,100 | 0.80 (0.49, 1.10) | 95.5 | 23 RCTs, n=3,473 | 0.88 (0.53, 1.22) | 94.1 |
| Duration of intervention (months) | ||||||||||||
| ≤6 | 18 RCTs, n=1,359 | 3.31 (2.11, 4.51) | 83.2 | 18 RCTs, n=2,047 | 34.00 (21.35, 46.66) | 90.3 | 20 RCTs, n=2,071 | 1.02 (0.51, 1.53) | 95.5 | 14 RCTs, n=798 | 11.07 (4.46, 17.68) | 92.7 |
| >6 | 16 RCTs, n=876 | 3.09 (1.66, 4.52) | 97.2 | 6 RCTs, n=878 | 36.24 (1.02, 71.46) | 90.4 | 17 RCTs, n=3,123 | 0.56 (0.16, 0.95) | 94.9 | 10 RCTs, n=2,769 | 8.46 (1.71, 15.20) | 97.0 |
| Age (years) | ||||||||||||
| ≤65 | 26 RCTs, n=1,951 | 3.90 (3.13, 4.68) | 89.4 | 12 RCTs, n=1,794 | 44.28 (21.15, 67.42) | 81.6 | 21 RCTs, n=3,553 | 1.02 (0.66, 1.38) | 94.4 | 15 RCTs, n=2,486 | 14.45 (8.48, 20.43) | 95.6 |
| >65 | 7 RCTs, n=257 | 0.23 (−2.29, 2.76) | 94.8 | 10 RCTs, n=1,011 | 33.27 (16.53, 50.01) | 98.3 | 14 RCTs, n=1,523 | 0.49 (−0.12, 1.10) | 96.3 | 7 RCTs, n=963 | 2.31 (−5.14, 9.76) | 96.2 |
| BMI (kg/m2) | ||||||||||||
| ≤28 | 8 RCTs, n=346 | 4.53 (3.43, 5.62) | 72.8 | 6 RCTs, n=487 | 81.79 (48.55, 115.03) | 44.7 | 9 RCTs, n=675 | 1.93 (0.78, 3.08) | 96.5 | 7 RCTs, n=558 | 20.09 (16.80, 23.38) | 67.2 |
| >28 | 11 RCTs, n=1,192 | 2.52 (0.48, 4.57) | 93.8 | 8 RCTs, n=1,684 | 18.47 (−2.12, 39.06) | 89.6 | 12 RCTs, n=1,615 | 0.71 (−0.02, 1.44) | 96.8 | 6 RCTs, n=307 | 8.53 (8.88, 20.98) | 94.0 |
| Type of heart failure | ||||||||||||
| HFrEF | 28 RCTs, n=2,002 | 3.19 (2.14, 4.24) | 95.6 | 13 RCTs, n=1,818 | 20.82 (5.79, 35.85) | 77.8 | 23 RCTs, n=3,820 | 0.71 (0.40, 1.03) | 93.5 | 14 RCTs, n=2,636 | 10.83 (4.10, 17.55) | 96.8 |
| HFpEF | 3 RCTs, n=147 | 0.43 (−2.57, 3.42) | 58.9 | 4 RCTs, n=480 | 73.19 (34.58, 111.79) | 66.0 | 4 RCTs, n=508 | 1.06 (0.53, 1.58) | 82.4 | 3 RCTs, n=435 | 16.07 (13.77, 18.37) | 0.0 |
| Peak V̇O2 at baseline (mL/kg/min) | ||||||||||||
| ≤15 | 8 RCTs, n=501 | 3.57 (2.34, 4.80) | 93.3 | 3 RCTs, n=287 | 41.81 (−5.66, 89.26) | 78.5 | 4 RCTs, n=373 | 1.87 (−0.48, 4.22) | 98.5 | 3 RCTs, n=284 | 18.78 (11.98, 25.57) | 86.9 |
| >15 | 26 RCTs, n=1,734 | 3.06 (1.65, 4.47) | 95.2 | 5 RCTs, n=995 | 45.83 (10.84, 80.82) | 65.6 | 13 RCTs, n=1,337 | 0.96 (0.58, 1.34) | 84.0 | 6 RCTs, n=244 | 16.81 (12.09, 21.52) | 52.5 |
6MWT, 6-min walk distance test; BMI, body mass index; CI, confidence interval; GRADE, Grading of Recommendations, Assessment, Development, and Evaluation; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; MD, mean difference; MLHFQ, Minnesota Living With Heart Failure Questionnaire; QOL, quality of life; RCTs, randomized control trials; SMD standardized mean difference; V̇O2, oxygen uptake.
6MWT Twenty-four trials involving 2,925 participants used the 6MWT (Table 2). We found that participants who underwent CR had a significant improvement in 6MWT than those who did not (MD 33.89; 95% CI 23.18–44.60; I2=96.0%). We conducted a subgroup analysis for the intervention duration of ≤6 months (MD 34.00; 95% CI 21.35–46.66; I2=90.3%) or >6 months (MD 36.24; 95% CI 1.02–71.46; I2=90.4%), which showed no difference between the groups (test for subgroup differences: P=0.907). We performed other subgroup analyses because an I2 >50% was obtained. We performed our planned subanalyses and meta-regression analysis but were unable to resolve the heterogeneity.
Quality of Life Quality of life was assessed using the Minnesota Living with HF Questionnaire (MLHFQ), the Kansas City Cardiomyopathy Questionnaire, EuroQol-5 dimension (EQ-5D), or the 36-item Short Form Health Survey (SF-36). When multiple quality of life assessments were performed in a single study, the MLHFQ, which was used most frequently in this review, was used in the analysis.
First, all quality of life scores using all measurements were included in the analysis. Thirty-five trials involving 5,628 participants reported the quality of life of the participants (Table 2). We found that participants who underwent CR had a significant improvement in quality of life compared with those who did not undergo CR (SMD 0.79; 95% CI 0.49–1.09; I2=95.4%). We performed a subgroup analysis for the intervention duration of ≤6 months (SMD 1.02; 95% CI 0.51–1.53; I2=95.5%) or >6 months (SMD 0.56; 95% CI 0.16–0.95; I2=94.9%), which showed a difference between the groups (test for subgroup differences: P=0.159). This SMD size is considered to have a moderate to large effect according to the Cohen criteria, as cited in the Cochrane Handbook for Systematic Reviews of Interventions.9
Second, only quality of life as measured by the MLHFQ was included in the analysis. Twenty-four trials involving 3,567 participants reported the MLHFQ (Table 2). We found that participants who underwent CR had a significant improvement in quality of life compared with those who did not undergo CR (MD 9.88; 95% CI 5.38–14.38; I2=97.0%). We performed a subgroup analysis for the intervention duration of ≤6 months (MD 11.07; 95% CI 4.46–17.68; I2=92.7%) or >6 months (MD 8.46; 95% CI 1.71–15.20; I2=97.0%), which showed no difference between the groups (test for subgroup differences: P=0.587). We performed our planned subanalyses and meta-regression analysis but were unable to resolve the heterogeneity.
This review included 72 RCTs (118 references) involving 8,495 participants. This review has included the highest number of published RCTs, approximately 30 more than the Cochrane review published in 2019.6 The meta-analysis included participants with HFrEF and HFpEF; however, most participants had HFrEF and were male. The findings showed that CR reduced the risk of rehospitalization for any cause by 20% (10% for <6 months and 28% for >6 months) and the risk of rehospitalization for HF by 12%. CR was also found to improve exercise tolerance (measured by peak V̇O2 and the 6MWT) and quality of life. However, no statistically significant difference in the reduction of all-cause or HF-related mortality was observed between participants who underwent CR and those who did not.
Similar to several previous systematic reviews, the present study did not find that CR was associated with a reduction in all-cause or HF-related mortality.6 However, CR was found to reduce the risk of all-cause rehospitalization by 20% and HR-related rehospitalization by 12%. In addition to the previous Cochrane review, the Exercise Training for Chronic Heart Failure [ExTraMATCH II] trial also supports the finding that exercise therapy for patients with HF is not associated with a reduction in the risk of all-cause mortality.44 However, the trial did show a reduction in HF-related hospitalization rates with exercise training These findings suggest that although exercise-based CR does not directly affect the risk of mortality in patients with HF, it still has important benefits in terms of reducing the burden of hospitalization. One possible reason for the lack of difference in mortality rates could be the poor adherence to exercise programs among patients with HF. In the HF-ACTION trial, which tested the effects of exercise training in a large cohort of patients with HF, the overall exercise adherence rate was only 65%,45 suggesting that patients have difficulty maintaining regular exercise habits. ExTraMATCH II, an RCT of exercise-based CR for patients with HF, also reported poor adherence to exercise programs, with only 38% of patients attending more than half the scheduled exercise sessions.44 This poor adherence may have limited the potential benefits of exercise training and contributed to the lack of difference in mortality rates between the intervention and control groups. In contrast, CR was shown to reduce the overall risk of rehospitalization by 20%, with a 10% risk reduction for <6 months CR and a 28% risk reduction for ≥6 months CR. CR was also found to reduce the risk of HF-related rehospitalization by 12%, although no statistically significant difference was observed between <6 and ≥6 months CR. A possible explanation for the lack of a statistically significant difference in HF-related rehospitalizations could be the wide CIs and the lack of hazard ratios. Furthermore, it is likely that conducting an individual participant data meta-analysis may yield different results.46
Another important finding was that CR significantly improved key outcomes, such as peak V̇O247 and 6MWT,48 which are strongly associated with the occurrence of events. CR has also been shown to significantly improve the quality of life of participants. Because of the high heterogeneity of these results, subanalyses and meta-regression analyses were performed. This explained why younger age was associated with greater improvements in peak V̇O2 but did not resolve the heterogeneity in other analyses. In our meta-analysis, regardless of the intervention duration, BMI, LVEF, or peak V̇O2, the findings revealed that CR unequivocally improved peak V̇O2, 6MWT, and quality of life.
This study has several limitations. First, most participants had a BMI ≥25 kg/m2. It has been reported that CR interventions have a greater effect on exercise tolerance improvement in individuals with lower BMI.49,50 Considering the lower BMI in Asian populations, data may differ between Asia and Europe. Second, the prevalence of HFrEF was high. Although the subanalysis based on LVEF showed no differences in effectiveness between HFpEF and HFrEF in this study, increasing the number of cases may yield different results. Finally, the analysis of peak V̇O2 and quality of life did not resolve the heterogeneity. Other factors may have influenced the results, which were not included in the planned subanalysis.
Although CR does not significantly reduce mortality, it can effectively reduce rehospitalization rates when continued for more than 6 months, and improves exercise tolerance and quality of life regardless of the intervention duration.
The authors thank all those who contributed to this review.
This work was supported by Health and Labour Sciences Research Grants (22FA1020) from the Ministry of Health Labour and Welfare of Japan.
Y.J.A. is a member of Circulation Journal’s Editorial Board. The other authors declare no conflicts of interest.
S.Y., K.A., Y.J.A., and K.K. conceived and designed the analysis; M.O., S.T., M.S., Y.T., and Y.K. collected the data; S.Y. and Y.J.A. performed the analyses; and S.Y., Y.J.A., A.N., S.M., and M.I. wrote the paper.
The datasets generated and/or analyzed during the present study are not publicly available because the data used are from papers published by different publishers. However, the data are available from the corresponding author on reasonable request.
