The prognostic effects of cardiac rehabilitation (CR) are inconsistent in recent reports on heart failure (HF). Generally, participants in previous trials were relatively young and had HF with reduced ejection fraction. Herein, we examined the effects of CR on HF prognosis using a nationwide cohort study. This multicenter prospective cohort study included hospitalized patients with acute HF or worsening chronic HF. Patients who underwent CR once or more times weekly for 6 months after discharge were included in the CR group. The main study end point was a composite of all-cause mortality and HF rehospitalization during a 2-year follow-up period. We performed propensity score matching to compare the survival rates between the CR and non-CR groups. Of the 2,876 enrolled patients, 313 underwent CR for 6 months. After propensity score matching using confounding factors, 626 patients (313 pairs) were included in the survival analysis (median age: 74 years). CR was associated with a reduced risk of composite outcomes (hazard ratio [HR] 0.66; 95% confidence interval [CI] 0.48 to 0.91; p = 0.011), all-cause mortality (HR 0.53; 95% CI 0.30 to 0.95; p = 0.032), and HF rehospitalization (HR 0.66; 95% CI 47 to 0.92; p = 0.012). Subgroup analysis showed similar CR effects in patients with HF with preserved ejection fraction (≥50%) and HF with reduced ejection fraction (<40%). In the landmark analysis, CR did not reduce the aforementioned end points beyond 6 months after discharge (log-rank test: composite outcomes, p = 0.943; all-cause mortality, p = 0.258; HF rehospitalization, p = 0.831). CR is a standard treatment for HF regardless of HF type; however, further challenges may affect the long-term prognostic effects of CR.
Cardiac rehabilitation (CR) is a comprehensive disease management program that is highly recommended by heart failure (HF) guidelines. Meta-analyses have demonstrated the favorable effects mainly in young patients who have HF with reduced left ventricular ejection fraction (HFrEF). Still, available evidence regarding the effects of CR on the long-term prognosis of older patients or HF with preserved ejection fraction (HFpEF) is limited. Previous studies on HFpEF suggested that CR increases exercise capacity, which is a surrogate measure of clinical events. , Another retrospective cohort study demonstrated the prognostic effects of CR in both HFrEF and HFpEF. However, in that report, the effects of CR were analyzed in patients who underwent CR at least once within 3 months after discharge, which is less than the recommended frequency of effective CR. Additionally, the prognostic effects beyond 6 months remains uncertain. Therefore, this study aimed to examine the effects of CR for 6 months on the prognosis of HF in a cohort that included older patients with HFpEF.
This prospective observational study was part of a multicenter cohort study to develop F rai L ty-b A sed pro G no S tic criteria in H eart fa I lure P atients known as the FLAGSHIP study. The FLAGSHIP study’s design is described elsewhere. The present analysis included ambulatory patients hospitalized due to acute HF or worsening chronic HF. Patients who were able to walk for 20 meters with or without the assistance of a walking aid at hospital discharge were considered as “ambulatory”. The exclusion criteria included severe cognitive impairment, defined as a mini-mental state examination score <17 points; severe mental disorders; difficulty in answering questionnaires; and assumed short-term mortality (e.g., severe aortic valve stenosis without surgical indication or terminal-stage cancer). The FLAGSHIP study protocol was organized according to the Guidelines for the Epidemiological Research proposed by the Japanese Ministry of Health, Labour and Welfare. The study protocol was approved by the ethics committee of the Nagoya University School of Medicine (approval No. 2014–0421), and it complies with the principles of the Declaration of Helsinki. Ethical approval was also obtained from each participating hospital and each patient provided written informed consent.
In Japan, health insurance is mandatory and has covered CR for HF since 2007. CR is covered for 150 days in the Japanese health care system, and outpatient CR is covered for up to 60 minutes per session, 3 times a week. Participating hospitals provided in- and outpatient CR according to the guidelines of the Japanese Circulation Society. Based on the standard CR program recommended by the aforementioned guidelines, CR included exercise training, patient education, and patient counseling. Exercise training comprised preparatory exercises, aerobic exercise, and cool-down. Exercise intensity was set at each patient’s anaerobic threshold which was measured using cardiopulmonary exercise testing, the Karvonen formula, or an intensity of 11 to 13 on the Borg scale. Furthermore, resistance training was generally added to aerobic exercise. In this study, each patient’s CR participation frequency was assessed by physical therapists at each collaborating hospital based on medical records. We assessed CR frequency within 6 months of discharge because this was generally the same time as the final follow-up examinations in clinical practices in Japan. Patients who underwent CR once or more per week for 6 months after discharge were enrolled in the CR group. Hence, patients who did not complete 6-month CR or died within 6 months were included in the non-CR group for the present analysis. If outpatient CR was interrupted temporarily for any reason, including rehospitalization or canceling a session for illness but was resumed thereafter (≥1 session/week), the patient was included in the CR group.
Patient characteristics, including age, gender, body mass index, and clinical details (main HF etiology, co-morbidities, previous admission due to HF, and medications at discharge) were collected from medical records. Echocardiographic and biochemical data were collected just before discharge. Simpson’s method was used to calculate the left ventricular ejection fraction (LVEF) on 2-dimensional echocardiography. Biochemical data on brain natriuretic peptide (BNP), N-terminal pro-BNP (NT-proBNP), serum albumin, hemoglobin, estimated glomerular filtration rate, high-sensitivity C-reactive protein, and sodium levels were also collected. According to the statement from the Japanese Heart Failure Society, because several hospitals used NT-proBNP instead of BNP in their clinical practice, a categoric variable was created with BNP ≥200 pg/ml or NT-proBNP ≥900 pg/ml being defined as a high BNP level.
Physical frailty measures were also collected as covariates. In the FLAGSHIP study, physical frailty was assessed in each patient immediately before discharge using four domains: slowness, weakness, exhaustion, and low physical activity. Details on frailty assessments in the FLAGSHIP study are described elsewhere. Briefly, slowness and weakness were assessed based on the 10 meter usual walking speed and grip strength, respectively. Exhaustion was assessed using the performance measure for activity of daily living-8. Performance measure for activity of daily living-8 is a standardized questionnaire that assesses functional limitations in patients with HF. It comprises a list of 8 items that potentially require daily physical activity in patients with chronic HF and uses a 4-category response scale. It is scored from 8 to 32, with higher scores indicating more severe functional limitations. The score is strongly and negatively correlated with the peak oxygen uptake measured by the cardiopulmonary exercise test. Low physical activity was assessed using the Self-Efficacy for Walking-7 standardized questionnaire, which comprises 7 items with a 5-point Likert scale. The total score of this scale (7 to 35 points) has a moderate to strong correlation with accelerometer-measured step counts and a moderate correlation with vigorous physical activity. Global cognitive function was assessed using the mini-mental state examination, which is a standard 11-question test, scored from 0 to 30. Depression was assessed using a 5-item Geriatric Depression Scale questionnaire scored from 0 to 5. Depression was defined as a score ≥2 points.
The main study end point was a composite of HF rehospitalization and all-cause death within 2 years of discharge; secondary end points were independent HF rehospitalization and all-cause mortality. A follow-up survey for each patient was performed using the medical records of hospitals that managed the patient, and HF rehospitalization was determined by the cardiologists at each of the enrolling sites. If patients did not attend follow-ups at the enrolling hospitals, prognostic data were obtained from a survey mailed directly to the patients every 4 months. Follow-up was continued for each patient after HF rehospitalization until the patients or their families refused to attend follow-up or until the death of the patient. The follow-up period was defined as the time from discharge until the main end point occurred, until the patients refused to attend follow-ups or until the final follow-up.
Patients with missing data on most variables were excluded (n = 8). Continuous variables, with or without normal distribution, were described as either the mean ± SD or median (interquartile range [IQR]). Differences in patient characteristics between the CR group and the non-CR group were compared using t test, Mann–Whitney U test, or chi-square test, as appropriate. Because patients were not randomly assigned to the CR group, propensity score (PS) matching was performed to reduce the risk of bias in treatment selection and potential confounders. PSs for each patient were produced using logistic regression analysis with the CR group as the dependent variable and 33 variables as independent variables.
The proportion of missing data were 0% to 3% for most variables. To avoid bias caused by excluding patients with missing data, multiple imputation was conducted according to the Mitra and Reiter’s report. First, we performed 20-fold multiple imputations using chained equations. Second, we estimated the PSs of each patient using each of the 20 imputed complete datasets. Finally, we took the average of the 20 estimated PSs for each patient and then we matched PSs once, based on each patient’s average PS. We performed 1:1 nearest available PS matching with a caliper width of 0.2 of the PS standard deviation. Standardized mean differences for all covariates of <0.1 were considered to indicate successful PS matching. In the matched cohort, survival curves were estimated using the Kaplan-Meier method and were compared using the log-rank test for each study outcome. Hazard ratios and 95% confidence intervals were calculated using a Cox proportional hazards model. The proportional hazards assumption was checked using the Schoenfeld residuals test. We also performed a sensitivity analysis in which patients who died within 6 months were excluded and PS matching was performed using the methods described previously. Another sensitivity analysis was performed based on LVEF values: HFrEF (LVEF <40%), HF with midrange ejection fraction (LVEF = 40% to 49%), and HFpEF (LVEF ≥50%).
To evaluate events beyond 6 months, we also conducted landmark analyses starting at 6 months. Patients with individual end point events before 6 months were excluded from the landmark analysis.
All statistical analyses were performed with Stata 15 (Stata Corporation, College Station, Texas) software. A p value of <0.05 was considered statistically significant.
A total of 2,876 patients who were hospitalized due to acute HF or worsening chronic HF were included in this study ( Figure 1 ). Study subjects’ characteristics before and after PS matching are presented in Table 1 . During 4,331.0 person-years of follow-up, 733 patients were rehospitalized due to HF exacerbation and 305 died. The median period from discharge to the time of rehospitalization was 137 days (IQR 45 to 336 days), and the median period to the time of all-cause mortality was 349 days (IQR 180 to 516 days). Participants’ characteristics and prognosis according to HF types are presented in Supplementary Table 1 and Supplementary Figure 1. PS matching identified 313 matched pairs. The C-statistic of PS predicting CR participation was 0.689 (95% confidence interval 0.657 to 0.719). PS distributions after PS matching were well balanced between the groups ( Supplementary Figure 2 ). Standardized mean differences for all characteristics were <0.1; this threshold was considered to indicate sufficient balance ( Supplementary Figure 2 ). During 1,006.1 person-years of follow-up, 137 patients were rehospitalized due to HF exacerbation and 50 died in the matched cohort (n = 626). The median period from discharge to the time of rehospitalization was 157 days (IQR 46 to 285 days) and the median period to the time of all-cause mortality was 410 days (IQR 220 to 583 days).
| Variable | Before Matching | After Matching | |||
|---|---|---|---|---|---|
| Non-CR (n = 2,563) | CR (n = 313) | p | Non-CR (n = 313) | CR (n = 313) | |
| Age (years) | 76 (67 to 83) | 74 (68 to 79) | <0.001 | 74 (66 to 82) | 74 (68 to 79) |
| Men | 60.8% | 59.1% | 0.570 | 60.1% | 59.1% |
| BMI (kg/m 2 ) | 21.8 (19.5 to 24.6) | 22.4 (20.0 to 24.8) | 0.028 | 22.2 (19.7 to 25.7) | 22.4 (20.0 to 24.8) |
| Prior HF hospitalization | 30.2% | 26.5% | 0.180 | 26.8% | 26.5% |
| SBP (mm Hg) | 112 (100 to 125) | 109 (98 to 123) | 0.029 | 111 (99 to 122) | 109 (98 to 123) |
| DBP (mm Hg) | 63 (55 to 73) | 64 (55 to 70) | 0.520 | 64 (55 to 72) | 64 (55 to 70) |
| HR (bpm) | 73 (64 to 82) | 75 (66 to 85) | 0.005 | 75 (65 to 85) | 75 (66 to 85) |
| Principal etiology | |||||
| Ischemic | 26.2% | 32.3% | 0.010 | 31.0% | 32.3% |
| Arrhythmic | 18.3% | 12.1% | 16.3% | 12.1% | |
| Valvular | 16.9% | 17.6% | 16.6% | 17.6% | |
| Cardiomyopathy | 13.7% | 15.7% | 15.0% | 15.7% | |
| Hypertensive | 11.0% | 7.0% | 8.0% | 7.0% | |
| Others | 13.9% | 15.3% | 13.1% | 15.3% | |
| LVEF categories | |||||
| <40% | 37.2% | 43.6% | 0.013 | 46.0% | 43.6% |
| 41% to 49% | 17.5% | 19.9% | 17.5% | 19.9% | |
| ≥50% | 45.3% | 36.5% | 37.5% | 36.5% | |
| Atrial fibrillation | 35.3% | 31.9% | 0.240 | 32.9% | 31.9% |
| Co-morbidities | |||||
| Diabetes mellitus | 34.6% | 37.4% | 0.320 | 34.8% | 37.4% |
| Past cardiac surgery | 12.1% | 12.8% | 0.710 | 14.4% | 12.8% |
| Cancer | 7.4% | 7.0% | 0.810 | 6.4% | 7.0% |
| COPD | 5.9% | 4.2% | 0.200 | 4.5% | 4.2% |
| Orthopedic disease | 5.0% | 2.2% | 0.031 | 3.2% | 2.2% |
| Stroke | 1.3% | 1.0% | 0.590 | 1.3% | 1.0% |
| High BNP level | 61.3% | 61.6% | 0.920 | 63.8% | 61.6% |
| eGFR (ml/min/1.73 m 2 ) | 47 (34 to 61) | 48 (36 to 61) | 0.330 | 48 (36 to 59) | 48 (36 to 61) |
| Albumin (g/100 ml) | 3.6 (3.3 to 3.9) | 3.6 (3.4 to 4.0) | 0.001 | 3.6 (3.3 to 3.9) | 3.6 (3.4 to 4.0) |
| Hemoglobin (g/100 ml) | 12.2±2.2 | 12.5±1.9 | 0.009 | 12.5±2.2 | 12.5±1.9 |
| hs-CRP (mg/100 ml) | 0.29 (0.10 to 0.84) | 0.23 (0.09 to 0.67) | 0.050 | 0.25 (0.10 to 0.70) | 0.23 (0.09 to 0.67) |
| Medications | |||||
| Beta blocker | 73.5% | 79.2% | 0.029 | 81.8% | 79.2% |
| ACEi/ARB | 62.7% | 68.1% | 0.062 | 70.0% | 68.1% |
| MRA | 41.4% | 47.3% | 0.045 | 43.1% | 47.3% |
| Diuretic | 80.8% | 85.0% | 0.074 | 82.1% | 85.0% |
| Oral inotropic agent | 10.3% | 16.0% | 0.002 | 17.6% | 16.0% |
| Statin | 36.6% | 41.9% | 0.067 | 40.3% | 41.9% |
| Anticoagulant | 47.9% | 49.5% | 0.580 | 49.5% | 49.5% |
| MMSE (points) | 27 (25 to 29) | 28 (26 to 30) | <0.001 | 28 (26 to 30) | 28 (26 to 30) |
| Depression | 41.5% | 34.9% | 0.027 | 34.9% | 34.9% |
| Usual walking speed (m/s) | 0.95±0.28 | 1.05±0.24 | <0.001 | 1.03±0.26 | 1.05±0.24 |
| Grip strength (kg) | 23.6 (17.2 to 31.6) | 25 (19.2 to 33.3) | 0.001 | 26.1 (18.3 to 34.6) | 25 (19.2 to 33.3) |
| PMADL-8 (points) | 20 (15 to 24) | 19 (15 to 24) | 0.170 | 19 (15 to 23) | 19 (15 to 24) |
| SEW-7 (points) | 18 (13 to 24) | 21 (15 to 25) | 0.002 | 20 (14 to 25) | 21 (15 to 25) |
| Living alone | 22.4% | 14.7% | 0.002 | 12.5% | 14.7% |
| Frequency of CR | <0.001 | ||||
| ≥1 session/wk | 0% | 100% | 0% | 100% | |
| 1 session /2 wk | 3.5% | 1.5% | |||
| 1 session/mo | 7.0% | 6.1% | |||
| Dropout within 6 mo | 1.3% | 1.8% | |||
| None | 88.2% | 90.6% | |||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
