Controlled breathing and heart rate variability biofeedback for substance use‐related outcomes: A systematic review and meta‐analysis

First published online July 29, 2026. Across 12 controlled studies involving 696 participants, breathing-based interventions were associated with a small reduction in craving compared with controls. Seven studies used HRV biofeedback and five used other breathing approaches, and the review found no evidence that one category produced a larger craving effect. Withdrawal results were not significant, and evidence for dependence came from only two heterogeneous trials. These practices may be useful as adjuncts, but the review does not show that they replace established substance-use treatment.

Roxburgh AD, Rowland B, Lubman DI, Kaur Bains S, Pek MA, Tso B, Manning V, Arunogiri S.
Addiction · Publication date: 2026-07-29
Added to site: 2026-11-07
DOI: 10.1111/add.70542

First published online July 29, 2026 and assigned to the November 2026 issue. Added to Neurofeedback.com November 7, 2026 after comparison with fresh and reserve candidates. This is a historical addition, not a newly published November 7 study.

Modalities

Heart rate variability biofeedback, Controlled breathing, Biofeedback

Clinical focus

Substance use disorders, Craving, Withdrawal and dependence

Population

People using alcohol, nicotine or other drugs represented in 20 studies

Sample

20 studies and 1,249 participants in the review; 15 controlled studies entered at least one meta-analysis

Intervention / method

Web of Science, PsycINFO, MEDLINE, CINAHL, Embase, PubMed, Scopus and Cochrane were searched through November 14, 2025 under PROSPERO CRD42024524235. The review included 13 randomized or cluster-randomized trials, four quasi-experimental studies and three uncontrolled studies. Only controlled studies entered meta-analysis.

Comparator

Active and passive comparators were eligible and were pooled across randomized and quasi-experimental designs; comparator subtype was not analyzed separately

Clinician summary

The pooled craving effect was small (Hedges' g 0.30, 95% CI 0.12 to 0.48; I2 23.5%) and rated low certainty because randomized and quasi-experimental studies were combined. The HRV-biofeedback subgroup estimate was g 0.34 (95% CI 0.00 to 0.68), but it did not differ from other controlled-breathing interventions (between-group p = 0.68). Withdrawal was null and highly heterogeneous, while the dependence estimate came from two trials and was also heterogeneous.

How to interpret the findings

The craving synthesis combined randomized and quasi-experimental designs and all comparator types, and one cluster trial could not be adjusted for clustering. Most included trials were not preregistered, blinding was often infeasible, subjective outcomes were common, direct measures of substance use were sparse and long-term follow-up was often absent. Intervention type, duration and delivery varied substantially. Withdrawal was not significant, dependence relied on only two studies, adverse-event reporting was not synthesized, and the review does not establish superiority of HRV biofeedback over other breathing approaches.

Equipment and disclosures

The review included varied clinician-led, self-directed, wearable and smartphone-assisted breathing approaches. It did not establish superiority for a particular device, app or commercial system, and no equipment is attributed to BrainMaster or another vendor.

The review received no direct funding. Two authors held NHMRC fellowships; the paper reports unrelated speaker honoraria, advisory fees, research grants and a former directorship/shareholding, with no declared interest related to this project. No correction, retraction or expression-of-concern relation appeared in the checked PubMed, Crossref, PMC and bounded DOI/title records on November 7, 2026. AI-assisted source review for Neurofeedback.com, not independent human clinical review.

Related evidence

Compare findings from the same clinical area, including mixed results.