Exercise-based rehabilitation for heart failure

This is a Cochrane systematic review and meta-analysis that pooled data from randomised controlled trials comparing exercise-based cardiac rehabilitation (exercise training alone or as part of a comprehensive rehabilitation programme) against usual medical care (no structured exercise programme) in adults with heart failure.

The interventions tested included:

Aerobic exercise training (walking, cycling, rowing) at moderate intensity (40–70% of heart rate reserve or peak oxygen uptake)

Resistance training added to aerobic exercise in some programmes

Programme durations ranged from 8 weeks to 12 months

Frequency: typically 3–5 sessions per week

Session length: 20–60 minutes per session

The comparators were:

Usual medical care (standard pharmacological treatment, dietary advice, and routine follow-up without structured exercise)

The primary outcome measures were:

All-cause mortality (death from any cause)

All-cause hospitalisation (admission to hospital for any reason)

Heart failure-specific hospitalisation (admission primarily due to worsening heart failure)

Health-related quality of life (measured by validated questionnaires)

Secondary outcomes included:

Exercise capacity (measured by peak oxygen uptake, 6-minute walk distance)

Left ventricular ejection fraction (a measure of heart pumping function)

Adverse events

The review included 33 randomised controlled trials with a total of 4,740 participants with heart failure.

Population characteristics:

All adults (mean age ranged from 51 to 81 years across studies)

74% male (range across studies: 50–100% male)

All had chronic heart failure with reduced ejection fraction (HFrEF) — meaning the heart's pumping ability was below normal

Left ventricular ejection fraction ≤ 40% in most studies (normal is ≥ 50%)

New York Heart Association (NYHA) class I–III (mild to moderate symptoms) — no studies included class IV (severe symptoms at rest)

Stable on optimal medical therapy (beta-blockers, ACE inhibitors, diuretics)

Excluded: recent heart attack (within 6 weeks), unstable angina, uncontrolled arrhythmias, severe valvular disease, or other conditions that would make exercise unsafe

Setting:

Hospital-based or clinic-based supervised exercise programmes

Some studies included a home-based component after initial supervision

Conducted in Europe, North America, Australia, and Asia

The review extracted data from individual trials that used the following measurement tools:

**All-cause mortality:** Confirmed by death certificates or hospital records

**Hospitalisation:** Confirmed by hospital admission records, reported as number of events per patient over follow-up

**Health-related quality of life:** Measured by validated disease-specific questionnaires:

- Minnesota Living with Heart Failure Questionnaire (MLHFQ, 0–105 scale, lower = better quality of life)

- Kansas City Cardiomyopathy Questionnaire (KCCQ, 0–100 scale, higher = better quality of life)

- Short Form-36 (SF-36, generic health survey, 0–100 scale, higher = better)

**Exercise capacity:**

- Peak oxygen uptake (VO₂ peak, measured in mL/kg/min during a maximal exercise test on a treadmill or cycle ergometer)

- 6-minute walk distance (metres walked in 6 minutes on a flat corridor)

**Left ventricular ejection fraction:** Measured by echocardiography or cardiac MRI (percentage of blood pumped out of the left ventricle with each beat)

**Adverse events:** Self-reported or clinician-reported during exercise sessions and follow-up

**Study design:** This is a systematic review and meta-analysis of randomised controlled trials. The authors searched multiple databases (CENTRAL, MEDLINE, Embase, CINAHL) up to July 2013, plus hand-searched conference proceedings and reference lists. Two reviewers independently screened studies, extracted data, and assessed risk of bias using Cochrane's Risk of Bias tool.

**Randomisation:** All included trials randomly assigned participants to exercise or control groups. The quality of randomisation varied: some used computer-generated sequences with concealed allocation (good), others did not clearly describe their methods (unclear risk).

**Blinding:** Blinding of participants and exercise trainers is impossible in exercise trials — participants know whether they are exercising. Some trials blinded outcome assessors (e.g., for exercise tests and echocardiography), but many did not. This is a major inherent limitation of exercise research.

**Duration:**

Exercise programmes lasted 8 weeks to 12 months

Follow-up after the programme ended ranged from 3 months to 5 years (median ~12 months)

The meta-analysis pooled outcomes at the longest available follow-up for each trial

**Statistical approach:**

Random-effects meta-analysis (DerSimonian and Laird method) — appropriate because exercise programmes varied across studies

Heterogeneity assessed using I² statistic (percentage of variation due to true differences rather than chance)

Subgroup analyses by: programme type (exercise only vs comprehensive), duration (<6 months vs ≥6 months), setting (hospital vs home), and heart failure severity

Sensitivity analyses excluding trials at high risk of bias

**What this design can and cannot prove:**

This design can prove that, across multiple well-conducted trials, exercise-based rehabilitation reduces hospitalisation and improves quality of life in people with stable heart failure. The meta-analytic approach increases statistical power and generalisability beyond any single trial.

However, this design cannot prove:

That exercise is safe for all heart failure patients (the trials excluded high-risk patients)

That the benefits persist beyond the follow-up period (most trials followed patients for only 12 months)

That exercise is better than other interventions (e.g., dietary changes, medication optimisation) — because the comparator was usual care, not an active alternative

That the results apply to women or elderly patients equally (the sample was predominantly male and younger than the typical heart failure population)

That the mechanism is purely physiological (exercise may improve mood, social support, and medication adherence, which also reduce hospitalisation)

**Major methodological weaknesses:**

Inability to blind participants or trainers (performance bias)

Many trials had small sample sizes (median ~50 participants per trial)

Publication bias possible (small negative trials may not have been published)

Variable quality of reporting across trials (some did not clearly describe randomisation or attrition)

Only 10 of 33 trials reported adequate concealment of allocation

**Primary outcomes (pooled across all trials):**

**All-cause mortality:** No statistically significant reduction with exercise compared to control.

- Risk ratio (RR) 0.93, 95% confidence interval (CI) 0.75 to 1.15, p = 0.50

- This means a 7% relative risk reduction, but the confidence interval includes the possibility of no effect or even a small harm

- Absolute risk: 12.5% died in the exercise group vs 13.5% in the control group over ~12 months (not statistically different)

**All-cause hospitalisation:** Statistically significant reduction with exercise.

- RR 0.75, 95% CI 0.62 to 0.92, p = 0.005

- This is a 25% relative risk reduction

- Absolute risk: 28% hospitalised in the exercise group vs 37% in the control group over ~12 months

- Number needed to treat (NNT): 11 — meaning 11 people need to exercise to prevent one hospitalisation

**Heart failure-specific hospitalisation:** Statistically significant reduction.

- RR 0.61, 95% CI 0.46 to 0.80, p < 0.001

- This is a 39% relative risk reduction

- Absolute risk: 12% hospitalised for heart failure in the exercise group vs 20% in the control group

**Health-related quality of life (MLHFQ):** Statistically significant improvement with exercise.

- Mean difference: -5.8 points (95% CI -9.2 to -2.4), p < 0.001

- The MLHFQ scale runs 0–105, lower = better. A 5.8-point improvement is considered a moderate clinically meaningful change (minimal important difference is ~5 points)

**Secondary outcomes:**

**Peak oxygen uptake (VO₂ peak):** Statistically significant improvement.

- Mean difference: +2.2 mL/kg/min (95% CI 1.5 to 2.9), p < 0.001

- This represents a ~15–20% improvement from baseline (typical baseline ~12–15 mL/kg/min in heart failure patients)

**6-minute walk distance:** Statistically significant improvement.

- Mean difference: +41 metres (95% CI 27 to 55), p < 0.001

- The minimal clinically important difference for heart failure is ~30 metres

**Left ventricular ejection fraction:** Small but statistically significant improvement.

- Mean difference: +2.7 percentage points (95% CI 1.4 to 4.0), p < 0.001

- Clinical significance is debated — a 2.7% absolute increase is modest

**Adverse events:** No significant difference between exercise and control groups.

- Exercise-related adverse events (musculoskeletal injuries, falls, arrhythmias) were rare and not systematically reported across all trials

**Subgroup analyses:**

Benefits for hospitalisation were consistent across exercise-only and comprehensive programmes

Benefits appeared greater in programmes lasting ≥6 months compared to shorter programmes

No clear difference between hospital-based and home-based programmes

Benefits were similar across age groups and heart failure severity (NYHA I–III)

Translating these numbers into plain English:

**Hospitalisation risk:** If you have heart failure and do not exercise, about 37 out of 100 people will be hospitalised for any reason within a year. If you start a supervised exercise programme, that drops to about 28 out of 100 — meaning 9 fewer hospitalisations per 100 people. For heart failure specifically, the drop is from 20 to 12 per 100 — 8 fewer heart failure hospitalisations per 100 people.

**Quality of life:** The 5.8-point improvement on the MLHFQ is roughly equivalent to going from "quite bothered" by shortness of breath during daily activities to "moderately bothered" — a noticeable but not dramatic shift. It's similar to the improvement seen with starting a beta-blocker.

**Exercise capacity:** A 2.2 mL/kg/min increase in peak oxygen uptake is roughly equivalent to being able to walk up one additional flight of stairs without stopping, or walking about 40 metres further in 6 minutes (about half a city block). For a typical heart failure patient, this can mean the difference between needing to rest after walking to the mailbox versus being able to walk to the corner store.

**Mortality:** The 7% relative reduction in death was not statistically significant, meaning we cannot be confident exercise reduces mortality. However, the confidence interval (0.75 to 1.15) includes the possibility of a 25% reduction — so a small benefit cannot be ruled out, but neither can no effect.

**Heart function:** The 2.7 percentage point increase in ejection fraction is modest. If your heart pumps at 35% of normal, exercise might raise it to 38% — still below normal (≥50%), but a small improvement in pumping efficiency.

**What the authors acknowledge:**

Most trials had small sample sizes and short follow-up (median 12 months)

Inability to blind participants to exercise allocation

Variability in exercise programmes (type, intensity, duration, supervision)

Limited data on long-term outcomes beyond 12 months

Few trials reported adverse events systematically

Most trials excluded patients with severe heart failure (NYHA class IV), recent decompensation, or significant comorbidities

**What a critical reader would note:**

**Population bias:** 74% male, mean age ~60–65 years. Real-world heart failure patients are older (median age at diagnosis ~75), more often female, and have more comorbidities (diabetes, kidney disease, COPD). Results may not generalise to older, frailer patients or women.

**Publication bias:** The funnel plot for mortality was asymmetrical, suggesting small negative trials may be missing. This could overestimate the benefit of exercise.

**Attrition bias:** Dropout rates ranged from 5% to 30% across trials. People who dropped out of exercise groups may have been sicker or less motivated, potentially inflating the apparent benefit.

**Confounding by adherence:** People who stick with exercise programmes may also take their medications more reliably, eat better, and have better social support. The trials could not separate the effect of exercise from these correlated behaviours.

**No active comparator:** Usual care is a weak control. A more rigorous test would compare exercise to another active intervention (e.g., dietary counselling, stress management) to isolate the specific effect of exercise.

**Short follow-up:** Most trials followed patients for only 12 months. Heart failure is a chronic condition with a median survival of 5 years. We do not know if the benefits persist, diminish, or reverse over longer periods.

**Safety data are sparse:** Only 8 of 33 trials reported adverse events. Exercise-related cardiac events (arrhythmias, heart attacks) during supervised sessions were rare, but the true rate in unsupervised settings is unknown.

**Industry funding:** Several trials were funded by device manufacturers or pharmaceutical companies (e.g., Medtronic, Pfizer). While the review assessed risk of bias, industry-funded trials tend to report more favourable results.

For someone running their own n=1 experiment (with medical clearance from your cardiologist first — do not start an exercise programme for heart failure without physician approval):

### What to test

**Intervention:** Moderate-intensity aerobic exercise (brisk walking, stationary cycling, or elliptical machine) at 40–70% of your heart rate reserve. If you don't know your heart rate reserve, use the "talk test" — you should be able to speak in full sentences but not sing.

**Dose:** 30 minutes per session, 5 days per week (150 minutes/week total). Alternatively, 20 minutes daily if you're deconditioned.

**Programme duration:** Minimum 6 months to see meaningful changes in hospitalisation risk and quality of life. The meta-analysis found greater benefits with programmes lasting ≥6 months.

### Minimum meaningful duration

**For quality of life:** You may notice improvements in 8–12 weeks (some trials showed MLHFQ changes by 3 months)

**For exercise capacity:** Measurable changes in 6-minute walk distance or VO₂ peak typically appear by 3–6 months

**For hospitalisation risk:** The trials followed patients for 12 months — this is the minimum timeframe to assess whether you are staying out of hospital

**For ejection fraction:** Changes, if they occur, take 6–12 months of consistent training

### What to measure (specific metrics)

**Primary metric:** Number of hospitalisations (any cause and heart failure-specific) over 12 months. Track dates, reasons, and duration of each admission.

**Secondary metrics:**

- 6-minute walk distance (test monthly on a flat, measured course — mark a 30-metre stretch and walk back and forth for 6 minutes)

- Resting heart rate and blood pressure (measure weekly, same time of day, after 5 minutes seated rest)

- Quality of life: Complete the Minnesota Living with Heart Failure Questionnaire (free online) every 4 weeks

- Daily symptom diary: Rate shortness of breath, fatigue, and swelling on a 0–10 scale each evening

- Weight (daily, morning after urinating, before eating — rapid weight gain >2 kg in 3 days signals fluid retention)

**Safety metric:** Track any chest pain, palpitations, severe shortness of breath, or dizziness during or after exercise

### Key confounds to control for

**Medication adherence:** Take all heart failure medications at the same times