HRV training is feasible and well tolerated in coronary patients, with benefits for anxiety, quality of life and symptoms — and encouraging signals on hospitalisations. One pilot trial even suggests an effect on blood flow to the heart muscle under mental stress. It comes on top of cardiac rehabilitation, never instead of it.
Updated :
What the research shows
Biofeedback
Heart rate variability biofeedback in cardiac rehabilitation and secondary prevention
Ages
Adult (18+)
Techniques
HRV — resonance-frequency breathing
AAPB3Probably efficaciousNeuroLogic4Efficacioushigher vs AAPB
Level 4, raised from the AAPB Level 3, which pools all biofeedback for coronary artery disease and predates three further randomised trials. Randomised superiority is now replicated across independent settings. Del Pozo 2004: 63 patients randomised to conventional care or six sessions of resonance-frequency HRV biofeedback — HRV improved at weeks 6 and 18, with no change in controls. Nolan 2005: 46 patients randomised against an active stress-management control — only the biofeedback group improved HRV during a stress-reactivity challenge, the change correlating with improvements in depression and perceived stress. Lin 2015: 154 patients randomised against a waitlist — HRV rose in the trained group and fell in controls, with hostility also decreasing. Yu 2018: multicentre trial, 222 patients, HRV biofeedback vs standard care — lower respiration rate, depression and hostility and higher HRV after six sessions, and at one year fewer all-cause hospital admissions and emergency-room visits, the cardiac-specific reductions being non-significant. The 2022-2026 window adds three randomised trials: Lin 2022 (n = 84, waitlist-controlled; lower HRV reactivity during an anger-recall task), Limmer 2022 (n = 46 post-infarction, self-guided mobile biofeedback vs usual care; effects on short-term HRV only) and Shah 2025 (pilot, n = 21; mental-stress myocardial flow reserve on PET favoured biofeedback by 0.10, 95 % CI 0.01 to 0.19). Not Level 5: no sham-controlled trial, and only Nolan 2005 used an active comparator.
Relaxation and breathing therapy with EMG feedback, skin-conductance feedback and direct heart-rate feedback
Ages
Adult (18+)
Techniques
EMG biofeedback, Skin conductance (EDA), Other biofeedback
AAPB3Probably efficaciousNeuroLogic3Probably efficacioussame vs AAPB
Level 3 maintained (a disaggregation of the pooled AAPB rating). Van Dixhoorn randomised 156 patients in cardiac rehabilitation to conventional exercise-based rehabilitation or the same plus weekly relaxation and breathing therapy with EMG feedback over six weeks: less anxiety and greater well-being (1990), lower respiratory and heart rates and increased respiratory sinus arrhythmia in a subset of 76 (1998), and at five years 13 % fewer cardiac events and 31 % fewer hospitalisations, with no change in HRV (1999). Cowan 2001 randomised 133 survivors of cardiac arrest, 80 % of them with coronary disease, to standard care or a package of HRV and EMG biofeedback with cognitive behavioural therapy and education over eleven 90-minute sessions: an 86 % reduction in cardiovascular death at two years, holding after adjustment for other mortality predictors, with no difference in non-fatal events or HRV. Dath 1997 (n = 5) trained skin-conductance regulation with counselling: lower skin conductance, anxiety and symptoms at eight months; Johnston and Lo 1983 (n = 7) reported less angina and longer exercise tolerance. Direct heart-rate feedback failed: coronary patients regulated heart rate less well than age-matched or student controls (Lang 1975), and three of four post-infarction patients could not change it at all (Palomba 1982). Two large randomised trials of multicomponent packages with striking clinical outcomes, but neither isolates the biofeedback contribution; the rest is case series.
In short
Clinical reading
The AAPB gives a pooled Level 3. NeuroLogic raises HRV biofeedback to Level 4: randomised superiority replicated across independent settings (Del Pozo 2004; Nolan 2005 against an active control; Lin 2015, n = 154; Yu 2018, multicentre n = 222, with fewer all-cause hospital admissions and emergency-room visits at one year), plus three randomised trials from 2022-2026 (Lin 2022, Limmer 2022, Shah 2025: mental-stress myocardial flow reserve). Relaxation and breathing therapy with EMG feedback, skin-conductance feedback and heart-rate feedback stay at Level 3 (van Dixhoorn 1990-1999; Cowan 2001; multicomponent programmes in which the biofeedback contribution cannot be isolated).
Protocols
Resonance-frequency HRV on the Lehrer protocol, 5-10 sessions embedded in cardiac rehabilitation, with daily practice; relaxation and slow breathing with EMG feedback as the historical variant.
Limits
Few studies target the biological manifestations of the disease itself — none has shown regression of atherosclerotic plaque; outcomes are mostly physiological and psychological; there is no sham-controlled trial and only one active-comparator trial; the cardiac-specific reductions in hospital use were not significant (Yu 2018); direct heart-rate feedback failed in coronary patients; recent app-based trials give weaker or null results (Limmer 2022, Faasse 2026); there are no paediatric data.
Study base
Seven randomised trials in the AAPB base, including a multicentre trial of 222 patients; 2022-2026 base: 6 publications indexed in the archive (4 trials, 1 cohort, 1 systematic review), all on HRV biofeedback.
Brendan's perspective
Raising HRV biofeedback from 3 to 4 is defensible and I would have done the same. Randomised superiority replicates across independent groups, Yu 2018 carries 222 patients across several centres with fewer all-cause admissions at a year, and Shah 2025, small as it is, measured mental-stress myocardial flow reserve on PET — an instrument that could have returned nothing, which is exactly what I want from a primary outcome. What holds it below 5 is real: only Nolan 2005 used an active comparator, there is no sham, and the cardiac-specific reductions in hospital use were not significant. Note also where the weak results sit — the self-guided app trials. The clinician in the loop is not a detail. This is not my caseload, but the protocol is the ordinary Lehrer resonance-frequency one: five to ten supervised sessions with daily practice, inside cardiac rehabilitation and never instead of it. Nothing here shows an effect on the plaque, and a patient should hear that in those words.
Climov et al. (2014) Biofeedback on heart rate variability in cardiac rehabilitation: practical feasibility and psycho-physiological effects doi:10.1080/ac.69.3.3027833
Del Pozo et al. (2004) Biofeedback treatment increases heart rate variability in patients with known coronary artery disease doi:10.1016/j.ahj.2003.08.013
Nolan et al. (2005) Heart rate variability biofeedback as a behavioral neurocardiac intervention to enhance vagal heart rate control doi:10.1016/j.ahj.2005.03.015
Lin et al. (2015) Randomized controlled trial of heart rate variability biofeedback in cardiac autonomic and hostility among patients with coronary artery disease doi:10.1016/j.brat.2015.05.001
Yu et al. (2018) One year cardiovascular prognosis of the randomized, controlled, short-term heart rate variability biofeedback among patients with coronary artery disease doi:10.1007/s12529-017-9707-7
van Dixhoorn & Duivenvoorden (1999) Effect of relaxation therapy on cardiac events after myocardial infarction: a 5-year follow-up study doi:10.1097/00008483-199905000-00005
Shah et al. (2025) Heart Rate Variability Biofeedback and Mental Stress Myocardial Flow Reserve: A Randomized Clinical Trial doi:10.1001/jamanetworkopen.2025.38416
