Underused Chronic Pain Interventions: Personal Value and Policy Value

A 2026 audit for adults with mixed chronic non-cancer pain.

Chronic pain is a domain where the personal and policy questions come apart. A motivated person can select a phenotype, track sleep and function, stop an intervention that fails, avoid unsafe providers, and make a local tradeoff. A health system has to deal with average implementation, billing incentives, provider drift, coercion, low adherence, and patients who were never good candidates.

I use a narrow frame: which interventions look better than current adoption suggests after correcting for weak controls, regression to the mean, placebo and context effects, functional unblinding, adherence problems, safety burden, patient selection, and chronic pain's long history of overextended treatments.

The pattern is uneven. Some boring interventions deserve more coverage. Some high-tech interventions are real for narrow refractory phenotypes and bad as general pain policy. Some reasonable personal trials should remain policy-light. Some attractive claims shrink as soon as you ask whether pain interference, function, sleep, work, medication use, and durability moved together.

Method

I used a mixed chronic-pain target population, then downgraded or upgraded by condition. Direct patient outcomes counted more than mechanism. Pain intensity counted, but not alone. Function, interference, sleep, mood separated from pain, disability, work participation, medication use, durability, dropout, and adverse events carried more weight than a one-point visual analog scale shift.

For evidence, I favored systematic reviews with heterogeneity checks, large RCTs, active comparators, credible shams, long follow-up, responder analyses, and pragmatic trials when the policy question was in view. I penalized waitlists, usual-care controls that add attention more than treatment, weak shams, functional unblinding, small pilot studies, short follow-up, per-protocol analyses after high dropout, enriched responder samples, industry device trials without independent replication, and outcomes that improve distress while leaving pain and function unchanged.

Grades are split. Person-level A means a reasonable default for many motivated people in the named subgroup. B means reasonable for selected people with one major uncertainty. C means a trial only under specific circumstances. D means the cost, risk, or attention load beats expected benefit. F means avoid outside research or special cases. Population-level A means broad guideline or reimbursement support. B means support with criteria. C means study more or allow with constraints. D means do not scale. F means scaling would cause harm, waste, or coercion in expectation.

Candidate Index

The index is split by practical role. Grades here are synthesis grades after the adversarial pass. The next section shows the harsher grades I would use if the strongest objection got full weight.

Broad or low-burden first moves

High-skill or phenotype-sensitive options

Bounded adjuncts and study-more candidates

Refractory or specialty-only options

Adversarial Regrading Pass

The skeptical grade is the grade I would use if I gave the strongest objection full weight. I did not mechanically replace every synthesis grade with the skeptical one; the gap marks where the claim is fragile.

Grades that hold under attack

Material downgrades under the skeptical read

Main Audit

Phenotyping before escalation

Before adding another treatment, chronic pain often needs a sorting step. Neuropathic pain, migraine, obesity-linked OA, inflammatory disease, nociplastic pain, CRPS, sleep-driven amplification, opioid-induced hyperalgesia, and medication-overuse headache do not fail in the same way. A drug, procedure, exercise plan, or psych intervention sensible for one bucket can be useless or harmful in another.

The best-tested version is stratified low-back care. STarT Back looked good in the UK trial, then traveled less cleanly into US and Danish implementations.[30][31][32] The evidence has a strong idea and fragile system implementation. Biomarker-heavy precision pain medicine has not earned the same status. History, exam, medication review, sleep assessment, migraine criteria, neuropathic features, inflammatory red flags, function goals, and diaries are more useful than most exotic tests.

For a motivated person, grade A is appropriate. Stop expanding the workup once new tests stop changing the next decision. For systems, grade B fits: reimburse structured assessment and triage, with guardrails against checkbox rationing, incidentaloma hunting, and permanent labels that block re-evaluation. Red flags come first: cancer, infection, fracture, progressive neurologic deficit, inflammatory disease, medication harm, suicidality, and unsafe opioid/sedative combinations.

Exercise, resistance training, and graded activity

Exercise is the boring intervention that gets recommended in a way that makes it easy to fail. "Move more" is not the intervention. Load selection, pacing, flare management, progression, and phenotype matching are the intervention.

The reason it keeps surviving skeptical review is that the corrected claim is not "exercise is a strong analgesic." Chronic pain erodes capacity through deconditioning, fear avoidance, weak sleep, metabolic load, and low confidence in movement. Better capacity can lower pain interference even when the pain score moves modestly. Cochrane evidence across chronic pain is favorable but often low-certainty, with sparse harms reporting.[1] Chronic low back pain has a stronger trial base: the 2021 Cochrane review included 249 trials and found exercise better than no treatment, usual care, or placebo, with smaller gains against active care.[2]

My grade is A/A for OA, chronic low back or neck pain, fibromyalgia, deconditioning, fear avoidance, and metabolic pain load. The personal trial should run 8-12 weeks and track walking tolerance, stairs, sleep, flare recovery, and interference alongside pain intensity. Screen progressive neurologic deficit, acute fracture, acute infection, unstable cardiopulmonary disease, inflammatory flares, severe PEM/ME-CFS pattern, syncope, joint instability, and osteoporosis risk. A flare lasting more than 24-48 hours is a dosing error, not a character fact.

CBT-I and sleep treatment

CBT-I belongs high on the list only when insomnia is active. It is not a general pain treatment. It treats a pain-amplifier system: short sleep, worse pain sensitivity, worse fatigue, worse mood regulation, and lower tolerance for the work demanded by other interventions.

The evidence is strongest for sleep. A 2024 systematic review and dose-response meta-analysis in chronic musculoskeletal pain found CBT-I improves sleep, with smaller and more variable effects on pain intensity and disability.[9] A 2022 review of non-pharmacological sleep interventions in chronic pain found a large sleep-quality signal in the CBT-I subset, while pain outcomes were less consistent.[10] The sleep signal supports an A at the person level when insomnia is present, and a B for coverage through digital or group programs. Sleep-hygiene pamphlets do not count.

The safety gates matter more than the index can show. Screen obstructive sleep apnea, restless legs, bipolar disorder, seizure disorder, parasomnias, shift-work constraints, sedatives, opioids, alcohol, and safety-sensitive driving or work. Stop or modify sleep restriction if there is dangerous daytime sleepiness, hypomania, seizure risk, or no sleep-efficiency or Insomnia Severity Index signal by 6-8 weeks.

Botulinum toxin A for chronic migraine

Botulinum toxin is misplaced if treated as a chronic-pain intervention. It is an A-grade chronic migraine intervention inside the right boundary. Chronic migraine is not generic head pain, and it gives you an unusually clean tracking variable: monthly headache days.

The Cochrane review found about 1.9 fewer headache days per month versus placebo, with more than 4000 participants and stronger evidence for chronic than episodic migraine.[24] NICE recommends it for adults with chronic migraine after failure of at least three preventives and with medication-overuse addressed.[25] The mean benefit is modest, injection placebo response is large, and repeated cycles create access burden. Those objections lower the hype, not the grade inside criteria.

Use requires headache diary documentation: at least 15 headache days per month, migraine features on at least 8 days, and inadequate response or intolerance to oral preventives. Trial two to three 12-week cycles. Stop if headache days or function do not move. The safety floor is a trained injector, medication-overuse plan, neuromuscular disorder screen, and instructions for dysphagia, respiratory symptoms, or spread-of-toxin weakness.

Interdisciplinary pain rehabilitation

Interdisciplinary rehab makes sense when pain has become a whole-life system. Severe disability often contains fear, sleep loss, medication effects, deconditioning, depression, family accommodation, work loss, and repeated failed procedures. Single-modality care can be too weak for that pattern.

Systematic reviews and health-technology assessments support multimodal or interdisciplinary programs for some long-term pain outcomes, with function and work participation often more persuasive than pain intensity.[7][8] The evidence is heterogeneous because the intervention is heterogeneous. Program quality dominates, and pre-post improvement can mix treatment effect with regression to the mean and selection.

The B/B grade applies to high-disability, high-utilization, medication-complex cases with a realistic ability to attend. It does not apply to generic wellness programs or mandatory gates before other appropriate care. A real program needs a coordinated medical, psychological, and physical plan, no forced rapid opioid discontinuation, psychiatric risk management, and escalation paths for withdrawal, suicidality, neurologic change, or uncontrolled flares.

CBT and ACT for chronic pain

CBT and ACT are easiest to misstate. They are useful adjuncts when the target is pain interference, fear avoidance, pacing, catastrophizing, values conflict, and distress. They are not evidence that pain is imaginary, and they do not replace medical evaluation.

The Cochrane review included 75 studies and around 9400 adults. CBT had small benefits for pain, disability, and distress; ACT evidence was lower-certainty and smaller.[3] The skeptical read is straightforward: psychotherapy trials are unblinded, often use waitlists, and can be allegiance-heavy. The average effect is small. Distress reduction should not be renamed nociceptive pain reduction.

The result is a B for selected people with chronic primary pain, fibromyalgia, fear avoidance, high distress, or pain-related disability, and a B for coverage when offered as one option. The safety floor is framing: pain is real, danger appraisal and avoidance may be modifiable, red flags still count. Screen suicidality, mania or psychosis, severe dissociation, and whether the medical workup is adequate. Do not turn therapy attendance into a compliance ritual.

EAET, pain reprocessing therapy, and somatic tracking

EAET, PRT, and somatic tracking are the best example of person-level and population-level reasoning splitting apart. If a person's pain is plausibly maintained by threat prediction, fear, attention, emotion, and avoidance after tissue danger is low, a therapy that changes threat appraisal could beat generic coping.

The favorable evidence is too large to dismiss and too narrow to scale carelessly. The PRT trial in primary chronic back pain reported 66% pain-free or nearly pain-free at post-treatment versus 20% in open-label saline placebo.[4] EAET outperformed CBT in a 2024 trial of older veterans with chronic musculoskeletal pain.[5] The adversarial pass bites hard anyway: expectancy, therapist allegiance, phenotype selection, and limited independent replication.

My grade is B personally for selected primary or nociplastic pain with low structural threat, high fear, avoidance, or trauma/emotion links. It is C at population level. The safety floor is unusually important: adequate medical workup, no psychogenic certainty as a default, screening for dissociation, PTSD destabilization, psychosis, mania, suicidality, and substance instability, and open re-evaluation triggers. Update upward on independent multisite replications against credible active controls with phenotype moderators and functional outcomes.

Mindfulness and MBSR

Mindfulness is strongest when framed as interference training. Pain contains sensation, salience, aversion, vigilance, and narrative load. A person can suffer less interference even when pain intensity changes little.

In chronic low back pain, the Cherkin JAMA trial found MBSR and CBT improved function and pain bothersomeness versus usual care, with partial durability.[27] Broader reviews find modest, low-certainty benefits across chronic-pain conditions.[28] The familiar weaknesses apply: usual-care and waitlist comparators, adherence selection, shrinkage against active care, and unclear separation between mood, sleep, interference, and pain.

A B grade is reasonable for pain distress, rumination, arousal, low back pain, fibromyalgia, or insomnia-adjacent pain. Coverage is also reasonable because group and digital programs are cheap, if they are not used as a message that undertreated pain should be accepted. The safety floor is trauma-sensitive practice: eyes-open grounding, opt-out from long body scans, and escalation for panic, dissociation, mania, psychosis, or suicidality.

Pain neuroscience education

Pain neuroscience education is useful when it changes behavior. If every movement is interpreted as tissue damage, graded movement becomes threatening. If the model changes without the patient feeling dismissed, exposure and activity become easier.

Reviews in chronic musculoskeletal pain find small or inconsistent pain and disability effects, with a stronger case when PNE is combined with exercise or rehab.[6] The same slide deck can validate pain physiology or sound like "nothing is wrong." That makes the population case weak unless it is embedded in active care.

The personal grade is B as an amplifier for graded movement in fear-avoidant or nociplastic presentations. The population grade is C: reimburse it as part of active rehabilitation or psychologically informed care, not as a cheap substitute for diagnostics, medication, or accommodations. The required frame is additive: pain can be real, modulated, and still medically revisable.

Topical capsaicin and lidocaine

Topicals look better when the target is truly local. A focal peripheral generator can sometimes be treated at the skin, avoiding sedation, cognitive effects, falls, constipation, dependence, and many systemic interactions.

Cochrane finds high-concentration 8% capsaicin can provide moderate or substantial relief for a minority of adults with chronic neuropathic pain, especially localized peripheral neuropathies.[15] Lidocaine is safer and widely used, but Cochrane judged the RCT evidence insufficient and low quality.[16] Capsaicin burns and can unblind trials; lidocaine's popularity exceeds its trial certainty. Neither is a deep spine, widespread fibromyalgia, or central sensitization treatment.

The grade is B/B for localized postherpetic neuralgia, painful diabetic neuropathy patches, focal allodynia, or postsurgical neuropathic pain. N-of-1 testability is high. Safety centers on boundaries: intact skin only, no mucosa or eyes, no large-area or occluded lidocaine, no heating pads over lidocaine, no repeated application through dermatitis or burns.

Weight loss, metabolic treatment, and GLP-1s

Metabolic treatment is a chronic-pain intervention only in a narrow sense. In obesity-linked knee OA, pain has mechanical and metabolic drivers: load, adipokines, inflammation, sleep apnea, low activity tolerance, and insulin resistance. That logic does not transfer to fibromyalgia, EDS pain, pelvic pain, CRPS, or migraine.

The STEP 9 NEJM trial randomized adults with obesity and moderate knee OA to semaglutide 2.4 mg or placebo plus diet/activity counseling for 68 weeks. WOMAC pain fell about 42 points with semaglutide versus 28 with placebo on a 0-100 scale.[26] Older weight-loss evidence also supports dose-related pain and function improvements in knee OA when weight loss is substantial.[33]

Grade B/B fits obesity plus knee or hip OA, metabolic disease, sleep apnea, or mechanical back load. It is a bad fit for normal BMI, frailty, eating-disorder risk, pancreatitis or gallbladder risk, severe gastroparesis, or inability to continue treatment. Policy should reimburse where obesity and OA/metabolic drivers are explicit, without turning weight loss into a gatekeeping ritual before analgesia, mobility aids, or surgery evaluation. Safety requires medical supervision, protein and resistance training to limit lean-mass loss, diabetes medication adjustment, pregnancy screening, and monitoring for dehydration, gallbladder disease, pancreatitis symptoms, severe GI intolerance, and compounded-product risk.

Opioid tapering and buprenorphine rotation

Opioid tapering is not one intervention. Voluntary, supported dose reduction and buprenorphine rotation are one thing. Forced administrative tapering is another, and it should not borrow the favorable evidence.

The biological case is real enough: long-term full agonist opioids can produce tolerance, sedation, constipation, endocrine effects, sleep-disordered breathing, falls, overdose risk, withdrawal cycling, and possibly opioid-induced hyperalgesia. Buprenorphine may preserve analgesia with a different safety profile for selected patients. Reviews still find inadequate evidence to balance benefits and harms, and better outcomes tend to come from voluntary, supported tapering instead of administrative tapering.[29] A patient-centered tapering cohort reduced dose without mean pain worsening, but it was uncontrolled and selected.[34] Evidence for buprenorphine rotation is low-quality but plausible.[35] CDC warns against abrupt discontinuation.[36]

The grade is B when harms exceed benefits, the patient wants change, and prescriber support exists. It is D for forced tapering or unsupported rotation. For policy, reimburse slow taper support, withdrawal care, behavioral support, naloxone, OUD assessment, and buprenorphine analgesic competence. Do not reward dose cuts as a quality metric. The safety rule is no abandonment: screen OUD, overdose history, suicidality, benzodiazepines, alcohol, gabapentinoids, sleep apnea, current MME, pregnancy, and respiratory disease; pause or reverse if function, sleep, mood, suicidality, or pain destabilizes.

Acupuncture, manual therapy, and dry needling

Acupuncture, manual therapy, and dry needling belong in the bounded-adjunct bucket. Touch, needling, expectation, local tissue input, short-term analgesia, reduced guarding, and a non-drug care context can help some people move and sleep better. That does not make the class a hidden high-effect treatment.

Vickers' individual patient data meta-analysis included 39 trials and 20,827 participants and found acupuncture superior to sham and no-acupuncture controls for back/neck pain, OA, headache/migraine, and shoulder pain, with small sham-adjusted effects.[21] Spinal manipulative therapy for chronic low back pain shows small effects similar to other recommended therapies.[22] Dry needling evidence is more mixed and short-term.[23] Sham controls are not inert, provider allegiance is large, and passive maintenance is easy to sell indefinitely.

The grade is C, or B-minus for a patient who likes it, can afford it, and uses it to enable activity. Trial 4-8 sessions. Stop if there is no functional signal after 4-6 sessions or the provider pushes indefinite maintenance. Cover short bounded courses for selected musculoskeletal pain, not open-ended maintenance. Safety requires a licensed practitioner, sterile single-use needles, anticoagulation and immune-risk screening, no high-velocity cervical manipulation with vascular or neurologic red flags, and escalation for dyspnea after thoracic needling.

Non-invasive neuromodulation: TENS, tDCS, rTMS

Non-invasive neuromodulation has a good mechanism story and weak practical certainty. Pain is electrically and network-mediated; segmental gating, endogenous opioid release, and cortical excitability modulation are plausible. The evidence justifies a cheap TENS trial and does not justify broad brain-stimulation policy.

The Cochrane overview of TENS could not draw reliable efficacy conclusions because trials were small and inconsistently reported.[17] tDCS reviews show small short-term signals with heavy protocol heterogeneity and sham concerns.[18] rTMS has signals in some pain reviews, with durability and protocol selection still unsettled. Device interventions inherit weak sham controls, functional unblinding, industry enthusiasm, and selective-responder problems.

TENS is a C-level personal experiment for localized flares over 1-2 weeks. Stop if there is no reproducible benefit. tDCS and rTMS belong in specialty or research settings. Screen implanted pacemakers, ICDs, neurostimulators, seizure history, pregnancy when relevant, skin integrity, and cortical-stimulation contraindications. No TENS over carotid sinus, anterior neck, chest, damaged skin, or implanted-device fields without clearance.

Implanted neuromodulation: SCS and PNS

Implanted neuromodulation is the device category where condition boundaries decide everything. Some refractory peripheral or spinal pain states are driven by aberrant afferent signaling and dorsal horn network behavior; an implant can modulate that input when medication fails. Nonspecific chronic low back pain is a different reference class.

The SENZA-PDN trial of 10-kHz spinal cord stimulation in refractory painful diabetic neuropathy showed a large responder advantage over conventional medical management, with longer-term follow-up in implanted and crossover cohorts.[19] For chronic low back pain, Cochrane found no meaningful improvement in pain, function, or quality of life versus placebo at six months.[20] Peripheral nerve stimulation evidence is condition-specific and weaker overall.[37]

Person-level B applies to refractory painful diabetic neuropathy and selected focal neuropathic, radicular, postsurgical, or CRPS-like pain after full workup. Policy stays at C: strict criteria, registry follow-up, functional outcomes, complication reporting, and no drift into poorly phenotyped centralized pain. The surgical burden is part of the intervention: infection, lead migration, revision, explant, MRI/procedure constraints, and device management. Do not implant without a trial showing at least 50% pain relief or clear functional gain.

Low-dose naltrexone

Low-dose naltrexone has the right profile for personal experimentation and the wrong evidence profile for policy. The mechanism story is neuroimmune modulation, glial/TLR4 effects, and opioid-system rebound. The practical appeal is low cost and low systemic burden in an opioid-free patient.

Reviews and meta-analyses in fibromyalgia find small and fragile signals across small trials.[11][12] The larger 6 mg randomized double-blind placebo-controlled fibromyalgia trial did not find superiority for pain.[13] A cheap generic with no sponsor can be under-studied and still weak. The current evidence is compatible with a responder subgroup, but it has not defined one.

My grade is C/C. A reasonable trial is fibromyalgia or nociplastic pain, no opioids, acceptance of off-label uncertainty, and 8-12 weeks of pain, fatigue, sleep, and function tracking. The hard gate is opioid exposure: no current opioids, buprenorphine, methadone, tramadol, kratom, or expected opioid need without a plan. Review liver disease, pregnancy, compounding quality, insomnia/vivid dreams, and emergency analgesia. A large independent RCT with responder analyses and function/fatigue endpoints would move this.

Ketamine and NMDA approaches

Ketamine is where dramatic short-term experience can outrun durable evidence. NMDA antagonism is relevant to central sensitization and may help some CRPS or refractory neuropathic pain states. Its antidepressant effect can also matter in severe comorbid depression, a separate claim from chronic-pain analgesia.

Guidelines and reviews put the strongest chronic-pain support around CRPS or refractory neuropathic pain, with short-term analgesic signals and weak durability.[14] CADTH's 2023 update found limited, heterogeneous evidence and persistent safety/monitoring concerns.[14] Functional unblinding is unavoidable, protocols vary, follow-up is short, and private-clinic incentives are bad.

The person-level grade is C for severe refractory CRPS or neuropathic pain in a specialist protocol, and D for routine mixed chronic pain or wellness-clinic dose chasing. Population grade is D. The safety floor is non-negotiable: monitored setting, blood pressure and cardiovascular screen, psychosis/mania/substance-use screen, liver and urinary monitoring for repeated use, sedation interaction review, and emergency response capacity. Durable controlled evidence with function, opioid use, cognition, urinary, hepatic, and misuse outcomes would be required for a real upgrade.

Cannabinoids

Cannabinoids occupy the uncomfortable middle: plausible for selected people, weak as policy. CB1/CB2 systems modulate nociception, sleep, affect, and reward. Some patients also want less NSAID, gabapentinoid, or opioid exposure and tolerate cannabinoids better than those drugs.

The BMJ systematic review found small improvements in pain, sleep, and physical functioning, with increased adverse events.[38] The linked BMJ guideline made a weak recommendation for a trial of non-inhaled medical cannabis or cannabinoids when standard care is insufficient.[39] Psychoactive unblinding is large, products vary, and THC adds cognition, anxiety, psychosis, driving, dependence, and cardiovascular concerns. CBD-only analgesia claims remain ahead of evidence.

Person-level C fits selected neuropathic or mixed pain with insomnia/appetite issues and low psychosis or substance-use risk. Trial 2-4 weeks, then stop if dose escalates without function gain, cognition worsens, anxiety rises, or driving/work safety is affected. Population-level D means allow clinician-supervised trials where legal, while avoiding broad promotion or reimbursement for a variable commercial product class. Avoid pregnancy/lactation, psychosis history, unstable bipolar disorder, cannabis use disorder, high-THC products in high cardiovascular risk, driving while impaired, and unsafe sedative/opioid/alcohol stacking.

Cross-Cutting Patterns

The interventions most likely underrated because they are boring are exercise/resistance training, CBT-I, topical agents for focal neuropathic pain, and basic phenotyping. None promises a dramatic cure. All can reduce disability or avoid worse care when matched well.

The interventions underrated because of institutional incentives are interdisciplinary rehabilitation, careful opioid taper support, buprenorphine rotation, and perhaps low-dose naltrexone. These require time, coordination, or generic prescribing. That does not prove effect. It explains why evidence can remain thinner than clinical plausibility.

The interventions overrated by chronic pain's evidentiary environment are ketamine for generic pain, broad cannabinoid use, consumer neuromodulation, unbounded manual/bodywork maintenance, and any therapy that converts "central sensitization" into a universal explanation. Chronic pain creates high regression to the mean, high placebo/context response, high desperation, and many outcomes that can improve because the patient received attention and hope.

The reference-class adjustment is harshest for devices, procedures, and psychological cures. Devices inherit the history of weak shams and industry enthusiasm. Psychological therapies inherit waitlist inflation and the risk of patient-blaming. Supplements and generic drugs inherit the mechanism-story failure mode. Opioid policy inherits both sides of a disaster: overconfident scaling of full agonists, followed by overconfident administrative tapering.

Stacking

Stacking is plausible only when mechanisms differ, risks do not compound, and adherence remains realistic. A good stack for many chronic-pain patients is not ten interventions. It is phenotyping, one active capacity intervention, one amplifier, and one safety review.

Examples: chronic low back pain with insomnia might use phenotyping, graded resistance/walking, CBT-I, and optional CBT/ACT or MBSR if fear or distress blocks function. Obesity-linked knee OA might use weight loss or GLP-1 when indicated, resistance training, sleep apnea treatment if present, and topical or medication support during the activity ramp. Chronic migraine should not be folded into generic pain; it needs headache-day tracking, medication-overuse review, and migraine-specific prevention such as botulinum toxin or CGRP-pathway treatment when indicated.

Interference matters. CBT-I sleep restriction can temporarily worsen pain and undermine exercise. Aggressive exercise can sabotage a PRT-style safety-learning frame. Cannabis or ketamine can blur whether sleep, mood, cognition, or pain changed. Opioid tapering during a major rehab push may help some people and destabilize others. The common chronic-pain failure mode is that the protocol becomes another full-time job.

Person-Level Recommendations

Reasonable person-level defaults for the right subgroup: phenotyping before escalation, graded exercise/resistance, CBT-I when insomnia is present, botulinum toxin for true chronic migraine, and topical capsaicin/lidocaine for focal peripheral neuropathic pain.

Reasonable selected options: interdisciplinary pain rehab for high disability, CBT/ACT for pain interference and fear avoidance, EAET/PRT for well-selected primary or nociplastic pain, MBSR for distress/interference, metabolic/GLP-1 treatment for obesity-linked knee OA, and voluntary opioid taper or buprenorphine rotation when opioid harms exceed benefits.

Discuss with a clinician before trialing: GLP-1s, botulinum toxin, low-dose naltrexone, buprenorphine rotation, implanted neuromodulation, ketamine, cannabinoids if psychiatric/cardiovascular/substance-use risk is present, and any intervention requiring medication changes.

Trial or refractory-case options: ketamine for CRPS/refractory neuropathic pain, implanted SCS/PNS outside painful diabetic neuropathy or clear focal neuropathic indications, rTMS/tDCS for pain, low-dose naltrexone if the aim is population-level proof, and most precision-biomarker matching tools.

Ignore or deprioritize for now: CBD-only analgesia claims, consumer tDCS, indefinite passive bodywork maintenance, SCS for nonspecific chronic low back pain, psychedelics for chronic pain without direct pain-outcome evidence, and any "central sensitization" protocol that closes medical re-evaluation.

Population-Level Recommendations

Reimburse across common use cases: graded exercise/resistance programs with pacing and flare management; CBT-I for chronic pain with insomnia; chronic migraine botulinum toxin inside criteria; bounded topical trials for focal neuropathic pain; structured chronic-pain assessment that separates nociceptive, neuropathic, nociplastic, migraine, sleep, metabolic, medication, and psychiatric risk patterns.

Reimburse with criteria: interdisciplinary pain rehab for high-disability cases; pain-focused CBT/ACT; EAET/PRT with phenotype discipline and provider training; metabolic/GLP-1 treatment for obesity-linked OA where metabolic indication is present; voluntary opioid taper support and buprenorphine rotation; acupuncture/manual therapy as bounded adjuncts; implanted neuromodulation for refractory painful diabetic neuropathy and well-selected neuropathic pain.

Allow without broad promotion: cannabinoids, low-dose naltrexone, TENS self-trials, dry needling, and rTMS/tDCS outside research or specialty contexts.

Needs better trials: PRT/EAET replication, LDN responder subgroup work, ketamine durability and harms, stimulation protocols, PNS indications, tapering versus buprenorphine rotation, and pragmatic phenotyping that changes decisions.

Do not build into broad guidelines: ketamine infusion clinics for mixed chronic pain, SCS for nonspecific low back pain, cannabis as general chronic-pain replacement therapy, consumer brain stimulation, mandatory psych treatment before medical care, and forced opioid tapering.

Limitations

Chronic pain is not one disease. A grade for mixed chronic non-cancer pain can hide true effects in small subgroups and null effects elsewhere. Fibromyalgia, chronic low back pain, migraine, CRPS, pelvic pain, EDS-related pain, neuropathic pain, and post-surgical neuropathic pain should not be pooled without condition-level checks.

Publication bias is likely. Weak positive trials survive better than negative trials. Psychotherapy, device, and complementary-medicine trials are hard to blind. Functional unblinding is not a footnote when the endpoint is subjective pain. Regression to the mean is large because people often seek care during flares.

Distress reduction counts, and analgesia needs separate evidence. A patient who sleeps better, fears movement less, or feels less trapped has a meaningful outcome. The policy error is laundering that into "the pain generator has been treated" or "medical care is no longer needed."

Expert-provider trials overestimate routine practice. PRT, EAET, interdisciplinary rehab, manual therapy, acupuncture, and graded activity can all degrade when delivered by average providers under billing pressure. I may still underweight implementation failure because the trial literature underreports it.

The patient-blaming risk is not theoretical. Behavioral and psychological pain protocols can be used to deny diagnostics, accommodations, analgesia, or disability support. That risk sharply lowers population grades for interventions that may still be reasonable for a self-directed person with good screening and a trustworthy clinician.

Personal experimentation is more defensible than population recommendation when the intervention is reversible, cheap, self-monitorable, and has a clear stopping rule. It is less defensible when the safety floor is serious, the provider can exploit hope, or the intervention changes future options.

References

1. Geneen et al., Cochrane 2017. Supports the broad but modest evidence for physical activity/exercise across chronic pain.
2. Hayden et al., Cochrane 2021. Supports exercise therapy for chronic non-specific low back pain, with small-to-moderate effects depending on comparator.
3. Williams et al., Cochrane 2020. Supports small benefits of psychological therapies for chronic non-headache pain and the limits of CBT/ACT evidence.
4. Ashar et al., JAMA Psychiatry 2022. Supports the large but replication-sensitive PRT signal in selected primary chronic back pain.
5. Yarns et al., JAMA Network Open 2024. Supports EAET outperforming CBT in older veterans with chronic musculoskeletal pain.
6. Watson et al., Journal of Pain 2019. Supports PNE as a plausible adjunct with limited standalone analgesic evidence.
7. SBU assessment, 2021/2022 record. Supports multimodal and interdisciplinary intervention evidence for long-term pain.
8. Interdisciplinary multimodal pain treatment review, 2022. Supports functional and symptom improvements while illustrating pre-post and heterogeneity limits.
9. Bilterys et al., 2024. Supports CBT-I for sleep disturbance in chronic musculoskeletal pain, with pain effects smaller than sleep effects.
10. Non-pharmacological sleep interventions review, 2022. Supports CBT-I improving sleep quality in chronic pain and weaker pain-function evidence.
11. LDN fibromyalgia systematic review/meta-analysis, 2025. Supports the fragile pooled LDN signal and adverse-effect profile.
12. LDN fibromyalgia review, 2023. Supports the conclusion that LDN evidence remains preliminary.
13. Bruun et al., Lancet Rheumatology 2024. Supports the negative larger naltrexone 6 mg fibromyalgia RCT.
14. CADTH ketamine update, 2023. Supports ketamine's limited chronic non-cancer pain evidence and safety burden.
15. Cochrane capsaicin review. Supports high-concentration capsaicin for a minority of localized neuropathic pain responders.
16. Cochrane topical lidocaine review. Supports the low-certainty status of topical lidocaine for neuropathic pain.
17. Gibson et al., Cochrane TENS overview 2019. Supports the conclusion that TENS evidence for chronic pain is inconclusive.
18. tDCS chronic pain review, 2022. Supports small short-term tDCS signals and protocol limitations.
19. Petersen et al., JAMA Neurology 2021. Supports 10-kHz SCS for refractory painful diabetic neuropathy while noting open-label/device limitations.
20. Cochrane SCS low-back-pain review, 2023. Supports not scaling SCS for nonspecific chronic low back pain.
21. Vickers et al., Journal of Pain 2018. Supports small but persistent acupuncture effects in several chronic pain conditions.
22. Rubinstein et al., BMJ 2019. Supports small effects of spinal manipulative therapy for chronic low back pain.
23. Hu et al., dry needling review 2018. Supports short-term dry needling signals and weak durability evidence.
24. Herd et al., Cochrane botulinum toxin migraine review. Supports botulinum toxin A for chronic migraine with modest mean headache-day reduction.
25. NICE TA260. Supports coverage criteria for botulinum toxin in chronic migraine.
26. Bliddal et al., NEJM 2024. Supports semaglutide 2.4 mg improving weight and WOMAC pain in obesity-linked knee OA.
27. Cherkin et al., JAMA follow-up. Supports MBSR and CBT improving chronic low-back-pain function and bothersomeness versus usual care.
28. Hilton et al., mindfulness chronic pain review 2017. Supports modest and low-certainty mindfulness benefits across chronic pain.
29. Mackey et al., opioid dose reduction rapid review 2020. Supports caution about tapering evidence and harms reporting.
30. Hill et al., STarT Back trial 2011. Supports stratified low-back-pain care in the original UK implementation.
31. Cherkin et al., MATCH trial 2018. Supports the weaker/null US implementation of risk-stratified low-back-pain care.
32. Danish stratified low-back-pain trial. Supports reduced resource use without clear superior patient outcomes.
33. Christensen et al., weight reduction knee OA meta-analysis. Supports dose-related pain/function benefit from substantial weight loss in knee OA.
34. Darnall et al., patient-centered opioid tapering 2018. Supports voluntary tapering feasibility without mean pain worsening in a selected cohort.
35. Powell et al., JAMA Network Open 2021. Supports low-quality but plausible evidence for buprenorphine rotation in chronic pain on long-term opioids.
36. CDC opioid prescribing guideline, 2022. Supports collaborative opioid risk-benefit review and warnings against abrupt discontinuation.
37. Xu et al., peripheral nerve stimulation review 2021. Supports condition-specific and still-limited PNS evidence.
38. Wang et al., BMJ cannabinoids chronic pain review 2021. Supports small average benefits and increased adverse events.
39. BMJ Rapid Recommendation, 2021. Supports only a weak recommendation for non-inhaled cannabis/cannabinoids after standard care is insufficient.