NeuroLogicEvidence levels

Pain and headache · AAPB chapter 13

Chronic pain

Chronic non-cancer pain

Chronic pain is not a single entity: the evidence level depends on site and mechanism. Biofeedback is recognised as a component of multimodal care, not a training that stands on its own, and is best established for noncardiac chest pain and chronic spinal pain. Neurofeedback, which AAPB left unrated, has since 2022 had a meta-analysis of randomised trials that supports an intermediate level.

Updated :

What the research shows

EEG neurofeedback — standard amplitude training

(alpha, SMR, theta/beta suppression) — chronic pain across sites

Ages
Adult (18+)
Techniques
Alpha training, Sensorimotor rhythm (SMR), Theta/beta ratio

AAPB0Not rated NeuroLogic3Probably efficacious

Level 3. AAPB assigns neurofeedback no level: its base is case series (Jensen 2007, n = 18, with no comparison against programme participants who did not receive neurofeedback; Jacobs & Jensen 2015, four cases) and Patel 2020's conclusion that high-quality controlled studies are lacking (AAPB ch. 13). Hesam-Shariati 2022 is the reason to rate it now: a systematic review of 10 RCTs and 13 non-randomised studies whose primary meta-analysis of the nine eligible RCTs found that EEG neurofeedback may have a clinically meaningful short-term effect on pain intensity at low certainty, rising to moderate certainty once high-risk-of-bias studies were removed; a post-hoc analysis restricted to conventional single-region protocols (alpha or SMR uptraining with theta and/or beta suppression) also gave a clinically meaningful effect estimate. Level 3 rather than 4: protocols and pain sites are heterogeneous, and no clinician-delivered amplitude protocol has been replicated across two independent settings in a diagnosed chronic-pain sample.

EEG neurofeedback — chronic non-specific low back pain

Ages
Adult (18+)

AAPB0Not rated NeuroLogic2Possibly efficacious

Level 2. One study is specific to this site: Yalfani 2025, a randomised trial of 60 women with chronic non-specific low back pain allocated to neurofeedback, cognitive functional therapy or control (20 per group) over eight weeks. Both active arms improved within group on pain intensity, disability and kinesiophobia (p < 0.05); cognitive functional therapy did better on vertical ground-reaction-force parameters (p < 0.05). The abstract does not report the neurofeedback-versus-control contrast and does not describe the training protocol. One adequately sized randomised study, unreplicated and with no protocol detail: Level 2 is the most that can be defended.

Other neurofeedback methods

Infra-low frequency (ILF) / infra-slow (ISF) with source localisation

Ages
Adult (18+)
Techniques
Infra-low frequency (ILF)

AAPB0Not rated NeuroLogic2Possibly efficacious

Level 2. Adhia 2023: double-blind, placebo-controlled randomised safety and feasibility trial, n = 60 with chronic low back pain, four arms of 15, 12 sessions of source-localised infraslow neurofeedback. The arm targeting the pregenual anterior cingulate cortex had the most favourable clinical outcomes, with the highest proportion of participants showing a clinically meaningful reduction in pain severity (53 %; MD −1.9, 95 % CI −2.7 to −1.0), interference (80 %; −2.3, −3.5 to −1.2) and disability (73 %; −4.5, −6.1 to −2.9) at one month. Adhia 2022, secondary analysis: effective connectivity from pgACC to left S1 increased versus sham at one month (p = 0.013) and correlated with pain-severity change (ρ = −0.630, p = 0.038). Mathew 2025 replicates the connectivity finding in a second sample (painful knee osteoarthritis). Level 2 rather than 3: the parent trial was powered for feasibility, arms of 15, and the clinical outcome is a responder proportion rather than a between-arm superiority test. These trials use source localisation only to target an infraslow band and are not a LORETA evidence base (no LORETA row).

Real-time fMRI neurofeedback (rostral anterior cingulate cortex)

Ages
Adult (18+)
Techniques
fMRI neurofeedback

AAPB2Possibly efficacious NeuroLogic2Possibly efficacious same vs AAPB

Level 2 (Level suggested in the AAPB text, not in the AAPB's rating headings.) AAPB states that "these studies suggest that there may be Level 2 – Possibly Efficacious empirical support for fMRI biofeedback focused on the rACC. However, more research is necessary to properly determine efficacy" (AAPB ch. 13). The base is deCharms 2005, eight chronic pain patients trained to modulate the rACC with reduced pain reports; and Guan 2015, a double-blind randomised study of 16 patients with postherpetic neuralgia, rACC-targeted fMRI neurofeedback versus a sham control, in which only the intervention group increased rACC activation, leading to changes in pain perception (AAPB ch. 13). No real-time fMRI trial appeared in the 2022-2026 window.

Biofeedback

Respiratory biofeedback (tidal CO2, thoracic-abdominal feedback) — noncardiac chest pain

Ages
Adult (18+)
Techniques
Respiratory / capnometry

AAPB4Efficacious NeuroLogic4Efficacious same vs AAPB

Level 4 maintained. The AAPB Level 4 was printed jointly with chronic low back and chronic neck pain (this row is a disaggregation by sub-indication; the AAPB level shown is the level of the printed grouping). DeGuire 1992: 41 patients with noncardiac chest pain randomly assigned to guided breathing retraining alone, retraining with thoracic and abdominal respiratory feedback, retraining with tidal CO2 feedback, or control; all three retraining arms improved respiration and chest pain, and three-year follow-up showed the gains held (DeGuire 1996) (AAPB ch. 13). Nothing new in the 2022-2026 window. The level rests on a single research group, which is its main weakness, but there is no specific reason to lower it.

EMG biofeedback (pressure/imaging feedback as adjunct) — chronic low back pain

Ages
Adult (18+)
Techniques
EMG biofeedback

AAPB4Efficacious NeuroLogic4Efficacious same vs AAPB

Level 4 maintained. The AAPB Level 4 was printed jointly with noncardiac chest pain and chronic neck pain (this row is a disaggregation by sub-indication). Sielski 2017: meta-analysis of 21 studies, mostly EMG, with significant reductions in pain intensity, disability, depression, coping and muscle tension. Newton-John 1995: waitlist versus cognitive-behavioural therapy versus EMG biofeedback, both active arms improving and holding gains at six months — one of the few tests of biofeedback on its own. Flor & Birbaumer 1993: only the biofeedback arm held its gains at 24 months in chronic musculoskeletal pain. In the window, Lazaridou 2023: eight weeks of virtual EMG biofeedback versus usual care, with lower pain intensity in the biofeedback arm (mean group difference 0.9, 95 % CI −1.07 to −0.32, p ≤ 0.01) and no group-by-time effects on interference, disability or pressure-pain thresholds.

EMG and postural biofeedback (one HRV trial) — chronic neck pain (samples usually neck and shoulder)

Ages
Adult (18+)
Techniques
EMG biofeedback, HRV — resonance-frequency breathing

AAPB4Efficacious NeuroLogic4Efficacious same vs AAPB

Level 4 maintained. The AAPB Level 4 was printed jointly with noncardiac chest pain and chronic low back pain (this row is a disaggregation by sub-indication). AAPB base: Bissett 1985 (EMG versus waitlist, reduced muscle activity and pain), portable myofeedback trials against exercise or conventional care, pressure-biofeedback-enhanced exercise trials, and Hallman 2011 (HRV biofeedback versus no treatment in stress-related chronic neck pain) (AAPB ch. 13). The counterweight is Campo 2021's meta-analysis, which concludes that biofeedback does not affect reports of pain but has small-to-moderate effects on disability. In the 2022-2026 window: Sabir 2026 (n = 44, cervical traction with or without EMG biofeedback over six weeks — pain and range of motion significantly better in the biofeedback arm, p < 0.05, with no difference on disability or muscle tension) and Szeto 2026 (three arms, muscle or postural biofeedback versus stretching, two hours daily for six weeks — pain and NDI reduced within group in both active arms and not in the control; the abstract gives neither n nor between-group statistics).

EMG-guided diaphragmatic biofeedback — gastro-oesophageal reflux disease

Ages
Adult (18+)
Techniques
EMG biofeedback, Respiratory / capnometry

AAPB3Probably efficacious NeuroLogic3Probably efficacious same vs AAPB

Level 3 maintained. The AAPB Level 3 was printed jointly with five other sub-indications (this row is a disaggregation). Sun 2016: 40 patients with gastro-oesophageal reflux disease randomly assigned to a proton-pump inhibitor or to diaphragm biofeedback plus the inhibitor; both groups improved symptomatically, but oesophageal manometry improved only in the biofeedback arm, and at six months 82.3 % of the biofeedback group versus 6.2 % of the medication group had discontinued treatment (AAPB ch. 13). One trial from one group, but with an objective physiological outcome and a marked reduction in medication use: Level 3 is consistent, and nothing in the 2022-2026 window changes it.

EMG biofeedback — other musculoskeletal pain (patellofemoral, mixed)

Ages
Adult (18+)
Techniques
EMG biofeedback

AAPB3Probably efficacious NeuroLogic3Probably efficacious same vs AAPB

Level 3 maintained. The AAPB Level 3 was printed jointly with five other sub-indications (this row is a disaggregation). The AAPB rating rests on small exercise-with-or-without-EMG-biofeedback trials in patellofemoral pain (Qi & Ng 2007, n = 26, better muscle activity in the biofeedback arm; Kim 2016, n = 30, three arms, biofeedback-assisted closed-kinetic-chain exercise better than exercise alone on Q angle and quadriceps activation) and on Flor & Birbaumer 1993, in which only the EMG biofeedback arm held its gains at 24 months in chronic musculoskeletal pain (AAPB ch. 13). Small samples, often physiological outcomes; nothing new in the window for this subgroup.

EMG-assisted relaxation and HRV biofeedback — cancer and cancer-treatment-related pain

Ages
Adult (18+)
Techniques
EMG biofeedback, HRV — resonance-frequency breathing

AAPB3Probably efficacious NeuroLogic3Probably efficacious same vs AAPB

Level 3 maintained. The AAPB Level 3 was printed jointly with five other sub-indications (this row is a disaggregation). Tsai 2007: 24 patients with advanced cancer in a palliative care unit randomly assigned to EMG biofeedback-assisted relaxation or typical care, with significantly less reported pain. Crowley 2020 (Burch et al.): four to six sessions of HRV biofeedback in cancer survivors did not improve self-rated pain but did improve sleep (AAPB ch. 13). Two small randomised trials, one of them null on pain: Level 3 is the defensible ceiling, and nothing in the 2022-2026 window moves it.

Thermal biofeedback (amputation-site warming) and EMG — phantom limb pain

Ages
Adult (18+)
Techniques
Thermal biofeedback, EMG biofeedback

AAPB3Probably efficacious NeuroLogic3Probably efficacious same vs AAPB

Level 3 maintained. The AAPB Level 3 was printed jointly with five other sub-indications (this row is a disaggregation). Harden 2005: a time-series analysis in amputees taught over seven sessions to raise temperature at the amputation site, most reporting at least a 20 % reduction in pain intensity; AAPB adds Sherman's review of case studies of EMG biofeedback for post-amputation cramping (AAPB ch. 13). The evidence is case-series in quality and Level 3 is generous; in keeping with the framework it is held rather than lowered, since there is no specific reason to lower it — and no new data appeared between 2022 and 2026.

Pelvic-floor EMG biofeedback — pelvic pain (chronic prostatitis, vulvodynia)

Ages
Adult (18+)
Techniques
Pelvic-floor EMG

AAPB3Probably efficacious NeuroLogic3Probably efficacious same vs AAPB

Level 3 maintained. The AAPB Level 3 was printed jointly with five other sub-indications (this row is a disaggregation). AAPB base: Yang 2017 (biofeedback plus electrostimulation, better on pain, symptoms and quality of life in refractory male chronic pelvic pain syndrome), Danielsson 2006 (n = 46, pelvic-floor biofeedback versus topical lidocaine gel in vulvar vestibulitis, both arms improving) and Bergeron 2008 (n = 51, biofeedback versus cognitive-behavioural therapy versus vestibulectomy, gains held at two and a half years, lowest pain after surgery) (AAPB ch. 13). In the 2022-2026 window: Wagner 2022, a systematic review of 37 studies, reports "tentative evidence" that biofeedback-assisted training has a positive effect on pain reduction, overall symptom relief and quality of life for non-anorectal phenotypes, with heterogeneity preventing pooling.

Anorectal / pelvic-floor EMG biofeedback — constipation and pain on defecation

Ages
Adult (18+)
Techniques
Pelvic-floor EMG

AAPB3Probably efficacious NeuroLogic3Probably efficacious same vs AAPB

Level 3 maintained. The AAPB Level 3 was printed jointly with five other sub-indications (this row is a disaggregation). AAPB cites Simón 2019 and Turnbull & Ritvo 1992 for painful chronic functional constipation in women, with reduced evacuation difficulty and lower pain grade on defecation in dyssynergic defecation (AAPB ch. 13). Wagner 2022 strengthens this row: for anorectal disorders "several landmark studies demonstrate the efficacy of biofeedback", making it the best-established pelvic phenotype in that review. Level 3 is consistent with a small number of small trials and no replicated credible active comparator.

Menstruation-related distress, whiplash

Ages
Adult (18+)
Techniques
EMG biofeedback, Thermal biofeedback

AAPB2Possibly efficacious NeuroLogic2Possibly efficacious same vs AAPB

Level 2 maintained; this row is kept as AAPB prints it. Menstruation-related distress: Hart 1981 compared thermal (hand-warming) and frontal EMG biofeedback in primary dysmenorrhoea, both improving somatic and affective symptoms over two months; AAPB adds small series on sequenced TENS/EMG/thermal training and one trial in which EMG biofeedback did better than relaxation or no intervention (AAPB ch. 13). Whiplash: Voerman 2006, 11 patients, improved muscle activation and reduced pain and disability after four weeks of ambulatory myofeedback (AAPB ch. 13); AAPB notes that chronic neck pain samples frequently include whiplash patients. Nothing new in the 2022-2026 window for either sub-indication.

Refractory complex regional pain syndrome, refractory myofascial pain

Ages
Adult (18+)
Techniques
Thermal biofeedback, Other biofeedback

AAPB1Not empirically supported NeuroLogic1Not empirically supported same vs AAPB

Level 1 maintained; this row is kept as AAPB prints it. Grunert 1990 (thermal self-regulation in reflex sympathetic dystrophy syndrome) and subsequent work on refractory CRPS report improvement within multidisciplinary programmes that included biofeedback, without ever isolating the biofeedback component. Sorrell & Flanagan 2003: 52 patients with refractory myofascial pain treated with a protocol combining biofeedback, physical therapy and trigger-point injections, with reduced pain and improved functioning — the same limitation (AAPB ch. 13). No controlled trial appeared in the 2022-2026 window: Level 1 marks an absence of evidence for refractory presentations, not evidence of absence of effect.

Resonance-frequency HRV biofeedback — chronic pain across sites

Ages
Adult (18+)
Techniques
HRV — resonance-frequency breathing

AAPB0Not rated NeuroLogic3Probably efficacious

Level 3. AAPB treats HRV biofeedback inside its sub-indications rather than as a row of its own: Hallman 2011 (randomised against no treatment in stress-related chronic neck pain, with improved pain, vitality and social functioning), Berry 2014 (veterans with mixed chronic pain, better perceived pain, stress and disability than a control group) and Crowley 2020 (null on pain, positive on sleep) (AAPB ch. 13). In the window, Chadwick 2026: n = 73 with co-occurring post-traumatic stress disorder and chronic pain, six weeks of HRV biofeedback versus waitlist — pain interference improved by 24.9 % (d = −1.14), with no between-group difference in pain intensity or pain disability. Level 3: waitlist- and no-treatment-controlled positives without a replicated superiority over an active comparator.

Skin conductance (EDA) biofeedback — chronic pain across sites

Ages
Adult (18+)
Techniques
Skin conductance (EDA)

AAPB0Not rated NeuroLogic2Possibly efficacious

Level 2. One study: Chrousos & Boschiero 2019, a clinical validation of a non-invasive electrodermal biofeedback device in which patients with various chronic pain conditions trained to reduce electrodermal activity reported reduced pain and showed fewer inflammatory indicators, with no change in a placebo group (AAPB ch. 13). AAPB gives it no level of its own; this row is new. One adequately sized study with a placebo condition supports Level 2, but the absence of replication and of a validated pain measure as primary outcome rules out anything higher.

In short

Clinical reading

The AAPB rating excludes bruxism, TMD, fibromyalgia, headache, IBS, arthritis and Raynaud's, covered separately. The AAPB levels apply to biofeedback alone, by sub-indication: Level 4 for noncardiac chest pain, chronic low back pain and chronic neck pain; Level 3 for six further sub-indications; Level 2 for menstruation-related distress and whiplash; Level 1 for refractory presentations. NeuroLogic holds every one of those levels and splits the bundled sub-indications so that the techniques become readable. Amplitude neurofeedback is rated 3 (Hesam-Shariati 2022), infraslow neurofeedback 2 (Adhia 2023), fMRI stays at the Level 2 the AAPB text suggests, and two biofeedback rows are added: HRV at 3 and EDA at 2. Adult indication only.

Protocols

SEMG on the relevant muscle groups, respiratory feedback and capnometry for chest pain, pelvic-floor EMG for pelvic pain and constipation, thermal biofeedback for phantom limb pain, resonance-frequency HRV, skin conductance; in neurofeedback, alpha or SMR uptraining with theta and/or beta suppression on one region at a time. Combined with cognitive-behavioural therapy and graded activity.

Limits

Adult indication only: AAPB is written for adults and paediatric pain is a separate indication — no child or adolescent band is created here. The specific biofeedback effect is rarely separable from the wider programme. Refractory presentations are unsupported. The only adequately sized sham-controlled neurofeedback trial (Rice 2024) was home-based with no clinician and found no difference — which, under the Parsons 2026 framework, leaves the question open rather than settled. Recent trials of ultrasound-imaging or movement-sensor feedback added to exercise are null on pain and belong to a different construct from physiological self-regulation. HRV biofeedback improves interference, vitality and sleep more than pain intensity itself.

Study base

Broad, mature biofeedback literature organised by pain site, with moderate effect sizes and largely self-reported outcomes; a small, protocol-heterogeneous neurofeedback base dominated by uncontrolled reports. 2022-2026 base: 87 records returned by the search, 18 kept (12 biofeedback, 5 neurofeedback, 3 HRV) and 20 rows indexed in the archive, 5 with a PDF on file.

Brendan's perspective

Neurofeedback gets a level here for the first time — 3, on Hesam-Shariati 2022. The part I would point at is the post-hoc analysis restricted to conventional single-region protocols: alpha or SMR up, theta and beta down, one region at a time. That is amplitude training doing the work. People will quote Rice 2024 against it. Read it first: 116 participants, home-based, self-directed, no clinician adjusting a threshold, and about a quarter concordant rewards in the sham arm. That is not a null against clinical neurofeedback; it is a null against unsupervised neurofeedback. The biofeedback side is the mature literature in the AAPB base and I treat it that way — surface EMG on the muscles that guard, respiratory work and capnometry for noncardiac chest pain, resonance-frequency HRV, inside graded activity and CBT rather than instead of them. The honest limit: what moves most reliably is interference, sleep and function. Pain intensity moves less, and Alkhawajah 2024 is a reminder that HRV work is not always the best tool in the room. One thing worth borrowing from next door: repetitive transcranial magnetic stimulation is having its moment in chronic pain, and the reason is instructive. What rTMS demonstrates is that modulating cortical activity changes pain — active neuromodulation, imposed from outside. Neurofeedback is the passive counterpart of the same idea: the same cortical territory, reached by learning rather than by induction. If the active route moves pain, it would be odd if the passive one had no part to play. That is an argument by analogy and I offer it as one, but it is the analogy I would want the field to take seriously, and it points at trials that have not been run.

Read next on the NeuroBLOG

Brendan Parsons, Ph.D., BCN — Founder of NeuroLogic, neurofeedback practitioner and trainer in Nice, AAPB board member

Explore this condition in the tool Compare every condition

References cited

  1. Sielski, Rief & Glombiewski (2017) Efficacy of biofeedback in chronic back pain: A meta-analysis doi:10.1007/s12529-016-9572-9
  2. Campo et al. (2021) The effectiveness of biofeedback for improving pain, disability and work ability in adults with neck pain: A systematic review and meta-analysis doi:10.1016/j.msksp.2021.102317
  3. DeGuire et al. (1992) Hyperventilation syndrome and the assessment of treatment for functional cardiac symptoms doi:10.1016/0002-9149(92)90211-G
  4. Hesam-Shariati et al. (2022) The analgesic effect of electroencephalographic neurofeedback for people with chronic pain: A systematic review and meta-analysis doi:10.1111/ene.15189
  5. Rice et al. (2024) Home-based EEG Neurofeedback for the Treatment of Chronic Pain: A Randomized Controlled Clinical Trial doi:10.1016/j.jpain.2024.104651
  6. Adhia et al. (2023) Exploring electroencephalographic infraslow neurofeedback treatment for chronic low back pain: a double-blinded safety and feasibility randomized placebo-controlled trial doi:10.1038/s41598-023-28344-2
  7. Patel et al. (2020) Effects of neurofeedback in the management of chronic pain: A systematic review and meta-analysis of clinical trials doi:10.1002/ejp.1612
  8. Lazaridou et al. (2023) Biofeedback EMG alternative therapy for chronic low back pain (the BEAT-pain study) doi:10.1177/20552076231154386
  9. Crowley et al. (2020) Symptom management among cancer survivors: Randomized pilot intervention trial of heart rate variability biofeedback doi:10.1007/s10484-020-09462-3
  10. Danielsson et al. (2006) EMG biofeedback versus topical lidocaine gel: A randomized study for the treatment of women with vulvar vestibulitis doi:10.1080/00016340600883401