Biofeedback, nearly always combined with breathing or relaxation work, lowers blood pressure at a good evidence level. It sits within a lifestyle-medicine approach and does not mean stopping medication. The support around it seems to matter: slow-breathing devices used alone at home have not confirmed their early results.
Updated :
What the research shows
Biofeedback
Assisted relaxation (skin conductance, thermal, EMG) and direct blood-pressure feedback
AAPB4EfficaciousNeuroLogic4Efficacioussame vs AAPB
Level 4 maintained. Yucha 2001, a meta-analysis of 23 studies from 1975-1996, found no difference in blood-pressure reduction between biofeedback and other active treatments such as relaxation and meditation, but a significantly greater reduction than inactive controls (SBP 6.7 mmHg, DBP 4.8 mmHg). Nakao 2003, 22 randomised studies and 905 patients, found biofeedback superior to non-intervention controls (SBP 7.3 mmHg, DBP 5.8 mmHg) but not different from other behavioural interventions; only relaxation-assisted biofeedback reduced both pressures. Linden and Moseley 2006, reviewing more than 100 randomised trials, report modest reductions, larger than in inactive controls and largest when baseline pressure is high and treatment individualised. Elavally 2020: nurse-led home skin-conductance biofeedback lowered mean SBP from 140.8 to 136.9 while controls rose, and 14 % of stage 1 patients moved to prehypertension against 3.3 % of controls who moved to stage 2. Xu 2007 (n = 49, prehypertension): greater falls with EMG biofeedback and relaxation, held at three months, 70.4 % returning to normal pressure against 22.7 %. Jenkins 2024 (20 trials) gives the current pooled estimate: SBP −4.52 mmHg (95 % CI −8.35 to −0.69), DBP −5.19 (−8.07 to −2.32). Direct pressure feedback is the weaker member: superior to a measurement-only control (Nakao 2000) and to sham after baseline adjustment (Tsai 2007), but Wang 2016 (n = 59) found real and sham finger-pressure feedback equal. Level 5 is not reached: no trial shows superiority over a bona fide active treatment.
Heart rate variability biofeedback (prehypertension and essential hypertension)
Ages
Adult (18+)
Techniques
HRV — resonance-frequency breathing
AAPB4EfficaciousNeuroLogic4Efficacioussame vs AAPB
Level 4 (a disaggregation of the pooled AAPB rating for biofeedback in hypertension). HRV biofeedback carries the clearest specificity signal in this indication because its trials use slow breathing alone as the comparator. Lin 2012 randomised 43 young adults with prehypertension to HRV biofeedback, slow abdominal breathing or control: pressure fell from 131.7/79.3 to 118.9/71.9 mmHg with increased HRV and baroreflex sensitivity and decreased skin conductance, still present at three months, while slow breathing produced similar but non-significant changes. Chen 2016 randomised 32 students with prehypertension to HRV biofeedback, breathing training or no treatment: both active groups lowered SBP and DBP after training, but only the HRV group held the reduction at three months and only it modulated cardiac autonomic activity better during cold-pressor and mental-arithmetic challenge. Wang 2010 randomised 22 postmenopausal women with prehypertension to EMG biofeedback with slow abdominal breathing or breathing alone, with greater falls in both pressures maintained at three months. Nolan 2010, behavioural neurocardiac training in Toronto, found HRV feedback compared with slow breathing or relaxation alone produced significant pressure reductions with correlated HRV increases. Level 4 rather than 5: the samples are small and mostly prehypertensive, and Lin 2012, Chen 2016, Wang 2010 and Xu 2007 all come from one research group, leaving Nolan 2010 as the only independent replication.
AAPB4EfficaciousNeuroLogic3Probably efficaciouslower vs AAPB
Level 3, below the pooled AAPB 4, for a stated reason: the pooled rating rests on clinician-delivered relaxation-assisted biofeedback, whereas this row covers unattended home devices their developers decline to call biofeedback, and the largest pooled estimate for them is null. The AAPB's positive basis was Sharma 2011, summarising four studies (two matched case-control, two randomised), all favourable on breathing rate and on clinically meaningful SBP and DBP reductions, together with the American Heart Association's Class IIA, Level of Evidence B rating for device-guided breathing (Brook 2013), and Adler 2019, in which a single 15-minute session lowered pressure in 22 normotensive adults. Against this, de Freitas Gonçalves 2022, a systematic review and meta-analysis of 22 randomised trials with 17,214 participants in the quantitative analysis, found device-guided slow breathing did not significantly reduce SBP (mean difference −2.13 mmHg, 95 % CI −12.71 to 8.44) or DBP (−0.90) against usual care, on 288 individuals for that comparison and with heterogeneity of 93 %. Breathing exercises in general still pool positively (Garg 2024, 15 RCTs: SBP −7.06 mmHg, 95 % CI −10.20 to −3.92; DBP −3.43), and acute within-session effects are reproducible (Mengden 2023, uncontrolled, n = 44: self-measured SBP 5 mmHg lower after each 10-minute session). Multiple positive controlled studies against a null pooled estimate for the device-guided form is Level 3.
In short
Clinical reading
The AAPB gives a pooled Level 4 for biofeedback in essential hypertension and prehypertension. NeuroLogic separates three rows: relaxation-assisted biofeedback and direct pressure feedback Level 4 (Yucha 2001; Nakao 2003; Linden and Moseley 2006; Elavally 2020; Jenkins 2024: SBP −4.52 mmHg, DBP −5.19), HRV biofeedback Level 4 (Lin 2012, Chen 2016, Wang 2010, Nolan 2010 — superiority over slow breathing alone at three-month follow-up), unattended device-guided slow breathing lowered to 3 (de Freitas Gonçalves 2022 meta-analysis not significant against usual care). In clinical practice biofeedback is almost never used alone: paced breathing, mindful breathing or whole-body relaxation form part of the intervention.
Protocols
Guided slow breathing (around 6 breaths/min), HRV, thermal, skin conductance or EMG depending on the team; 8-12 supervised sessions plus daily practice, with home blood-pressure measurement.
Limits
Long-term stability of blood-pressure reductions is poorly documented — AAPB judges data beyond one year too sparse to analyse; mechanisms of change are uncertain; the responder profile is unidentified; biofeedback does not outperform active relaxation or meditation; most HRV trials come from one research group; heterogeneity reaches 93 % in the device-guided breathing meta-analysis; there are no paediatric data.
Study base
Extensive RCT literature, including a reference review of behavioural treatments for hypertension and four meta-analyses (Yucha 2001, Nakao 2003, Costa Vital 2021, Jenkins 2024). 2022-2026 base: 3 publications indexed in the archive (2 meta-analyses, 1 cohort).
Brendan's perspective
Two rows stay at 4 and one comes down, and the one that comes down is the interesting one. Unattended device-guided slow breathing goes from 4 to 3: its developers decline to call it biofeedback, and de Freitas Gonçalves 2022 pools to nothing. Take the clinician out of the loop and you are no longer delivering the intervention the rest of this literature studied. The HRV row is the opposite case. Lin 2012, Chen 2016 and Nolan 2010 use slow breathing alone as the comparator, which strips out the pacing and the relaxation and leaves the feedback, and the separation shows up at three months. That is good design, and why I would not lump the two rows together. Hypertension is not my caseload, but resonance-frequency training is, and I would deliver it supervised, with home readings, alongside the physician. The limits, plainly: four to five mmHg, no advantage over active relaxation or meditation, nothing dependable past a year, and nobody stops medication on the strength of it.
Nakao et al. (2003) Blood pressure-lowering effects of biofeedback treatment of hypertension: a meta-analysis of randomized controlled trials doi:10.1291/hypres.26.37
Jenkins et al. (2024) Effectiveness of biofeedback on blood pressure in patients with hypertension: systematic review and meta-analysis doi:10.1038/s41371-024-00937-y
Lin et al. (2012) Heart rate variability biofeedback decreases blood pressure in hypertensive subjects by improving autonomic function and baroreflex doi:10.1089/acm.2010.0607
de Freitas Gonçalves et al. (2022) Device and nondevice-guided slow breathing to reduce blood pressure in hypertensive patients: A systematic review and meta-analysis doi:10.1002/hsr2.636
