Can HRV Biofeedback Actually Reduce Substance Use Cravings?

This piece covers what the clinical trial evidence shows about HRV biofeedback for craving and substance use outcomes specifically, not HRV biofeedback for anxiety, TBI, or cardiac rehab generally, and not the general relationship between drinking and HRV.

A phase 2 randomized clinical trial found that adding heart rate variability biofeedback to usual outpatient treatment was associated with significant reductions on measures including craving, compared to usual treatment alone, in adults with substance use disorder. That is a specific, controlled finding, not a general claim that biofeedback cures cravings for everyone. The trial used wearable HRV biofeedback technology and ecological momentary assessment to track outcomes over eight weeks of outpatient treatment.

The question people bring to this

Someone in recovery starts using a breathing app or a wearable that promises to calm the nervous system, and the pitch sounds almost too tidy: slow your heart rhythm, quiet the craving. The skepticism is fair. Cravings feel psychological, situational, tied to stress or triggers or boredom, not to the timing of a heartbeat. It is reasonable to wonder whether HRV biofeedback is doing anything specific to substance use, or whether it is just another relaxation exercise wearing a fitness-tracker costume.

There is also a narrower version of the doubt: even if slow-paced breathing calms someone down in the moment, does that translate into fewer cravings over weeks, or less actual use? A feeling of calm during a five-minute session is a different claim than a measurable change in drinking or drug use over an eight-week treatment window.

Why the body's rhythm was ever connected to cravings at all

The link is not just intuitive, it has a documented starting point in how alcohol and chronic substance use interact with the autonomic nervous system. A review of HRV in alcohol use found that resting HRV is reduced in people with alcohol use disorder, and that within that group, higher reactive HRV to stress was related to more craving, faster relapse, and more negative mood. That is the kind of finding that makes HRV biofeedback a plausible target rather than a random wellness trend: the autonomic signal researchers were already measuring in addiction populations before biofeedback trials existed showed a real association with craving and relapse risk.

Separately, research on acute alcohol consumption has shown it reduces resting HRV in the short term, and that heart rate alone is a less sensitive marker of alcohol's effects than variability measures. None of this proves biofeedback training changes craving by itself. It explains why someone would test the idea in a trial in the first place, and why the broader link between HRV and emotional regulation kept coming up in addiction research even before a wearable-based trial existed.

What the trial actually measured, and what it did not

The 2025 randomized trial split 120 treatment-seeking adults with substance use disorder into two groups: one received HRV biofeedback plus treatment as usual, the other received treatment as usual only, over eight weeks of outpatient care. Outcomes, including craving, negative affect, positive affect, and alcohol and other drug use, were tracked through ecological momentary assessment rather than a single post-treatment questionnaire, which means the measurement happened repeatedly in participants' daily lives rather than only in a clinic visit.

That design matters for how much weight the finding can carry. It is a controlled comparison against an active usual-treatment condition, not against doing nothing, and it used a form of HRV biofeedback described as second-generation wearable technology, distinct from earlier clinic-based versions. Whether the same pattern holds with different wearable hardware, a longer follow-up period, or a different treatment population is not something this single trial can answer on its own.

Why some people respond and others might not

One thing this research area keeps circling back to is that HRV biofeedback does not produce the same physiological response in everyone. A conceptual review on autonomic non-responsiveness describes situations where the expected engagement of the autonomic nervous system during biofeedback is weakened, absent, or fails to translate into a measurable benefit, and points to contributors like reduced autonomic flexibility, impaired baroreflex function, and disease burden. That review was not conducted in a substance use population specifically, but it is a useful caution against treating biofeedback as a uniform effect that works the same way for every person who tries it.

It is also worth separating this from related but distinct biofeedback research. A sham-controlled trial in people with traumatic brain injury found mixed results depending on which HRV metric was examined, with some measures favoring the biofeedback group and others favoring sham once prior scores were accounted for. A systematic review of biofeedback in cardiovascular rehabilitation similarly found the approach may support autonomic regulation and reduce psychological distress, while noting the evidence base is not yet fully established. Those findings come from different populations doing different things, and none of them are substitutes for the substance use trial, but together they suggest biofeedback's effects are consistently described as real but variable, not universal. For readers curious whether the same variability shows up in the anxiety literature, that question gets its own separate look.

Reading your own HRV number in this context

None of this trial evidence means a single low HRV reading on a wearable says anything about craving risk in the moment. HRV is shaped by age, sex, physical activity, smoking, and psychosocial stress, among other factors, according to a large cohort study of the general population, and it naturally shifts with things unrelated to substance use, which is a big part of why a single HRV number is hard to interpret in isolation. The craving-related finding here comes from a structured biofeedback training process studied over eight weeks, not from passively glancing at a daily HRV score. Someone wondering whether their wearable is even measuring HRV the way a research-grade device would is asking a fair and separate question from whether biofeedback training changes craving.

Common questions