Chronic Back Pain That Keeps Coming Back: What Exercise and Adjustments Actually Do

What this video covers

• Why chronic low back pain involves motor control, muscle endurance, and central sensitization, not just one structural lesion
• How exercise therapy works: progressive loading, graded exposure to feared movement, and a short-lived drop in pain sensitivity that is often blunted in long-standing pain
• What a spinal manipulation is thought to do, why its leading proposed mechanism comes from animal studies rather than human trials, and why the subluxation model is not supported
• What the evidence actually shows: versus no treatment or usual care, exercise improved pain by about 15 points on a 100-point scale, but function improved only about 7 points, below the level patients reliably notice, and versus other active conservative treatments the pain advantage shrinks to roughly 9 points and is no longer clinically important
• Who is a candidate, which red flags must be ruled out first, and why maintaining the program matters more than finishing it
• MEDICAL DISCLAIMER: This content is for educational purposes only and is not medical advice. It does not substitute for professional diagnosis or treatment. Always consult a licensed healthcare provider regarding your condition. Viewing this video does not establish a doctor-patient relationship.

You have had roughly thirty adjustments this year. For ten minutes afterward the world is different — a pop, a release across the lower back, the walk to the car convinced it took. By Thursday it is back. Or the other version: a hot pack, electrical stimulation, three sets of bridges, and a discharge letter when the visits run out.

Two years in, nobody has defined the problem, said what should measurably change in ninety days, or explained how they would know if it were not working. That is not therapy. That is a subscription. This article explains what is actually happening in a back that keeps flaring, what movement-based care genuinely does — and does not do — to it, and what the trial evidence shows when you read the numbers rather than the brochure.

“Nonspecific” is a conclusion, not a diagnosis

Chronic nonspecific low back pain — pain lasting past twelve weeks without an identified structural cause — is not one fault in one structure.

Your disc, your facet joints (the paired joints at the back of each spinal segment), and your sacroiliac joint can all generate nociception, the raw danger signal that a nerve ending sends when tissue is threatened. In most people no single one is ever confirmed as the sole source. Often more than one contributes. Sometimes none can be isolated.

“Nonspecific” describes the limit of our diagnostic reach. It does not mean nothing hurts, and it does not mean the search is always worth running.

The signal is also being amplified, and by three things at once:

• Stabilizers that recruit late. When the deep trunk muscles switch on behind schedule, load lands on passive tissue — ligament, capsule, disc — instead of on muscle.
• Fear-avoidance. Movement narrows to whatever has never hurt. Tolerance shrinks to match.
• Gain turned up in the dorsal horn. The relay in the spinal cord where pain signals are first processed can become more responsive over time, so an ordinary input produces a larger output.

That combination is why the pain does not track a single structure, and why a story built entirely around one bone rarely survives contact with the follow-up visit.

Why the pop feels like it worked — and why it wears off

Here is the part that explains the lived experience rather than the diagram.

A high-velocity, low-amplitude thrust — the classic adjustment — is real force applied to a real joint. Something genuinely happens. But the audible pop is cavitation: a gas bubble forming in the fluid of the facet capsule. It is not a bone returning to its correct address.

The leading proposed mechanism for what a manipulation does is altered discharge from paraspinal muscle spindles, the stretch sensors inside the muscles running alongside the spine. That work comes from anesthetized-animal recordings. It is not established in humans. The subluxation model — the idea that a vertebra slips out of alignment and gets put back — is unsupported.

Exercise has the same honesty problem in reverse. It works through progressive loading, motor control retraining, and graded exposure to the movements you have been avoiding. The short-lived drop in pain sensitivity you feel right after a session is commonly attributed to descending inhibition, the brain’s own pain-dampening pathways. That mechanism is unproven in humans and appears to be blunted in people with the longest-standing pain.

So the glow after a session — whichever kind of session — is not the treatment. The adaptation over twelve weeks is. That is precisely why relief that arrives in ten minutes and leaves by Thursday feels like failure. It is not failure. It was never the mechanism that was going to change anything.

What the numbers actually say

This is the part nobody quotes you.

Exercise therapy. A 2021 Cochrane review by Hayden and colleagues pooled 249 randomized trials in adults with chronic nonspecific low back pain. Against no treatment, usual care, or placebo, exercise reduced pain by about 15 points on a 100-point scale (mean difference −15.2; 95% CI −18.3 to −12.2), at moderate certainty. That is a real, clinically important effect.

Function is where the story turns. It improved only about 7 points (−6.8; 95% CI −8.3 to −5.3) — below the threshold the reviewers set in advance for a change patients reliably notice. And against other active conservative treatments, the pain advantage shrank to about 9 points (−9.1; 95% CI −12.6 to −5.6), also below that threshold.

Exercise and manual therapy came out equivalent — one point apart (MD 1.0; 95% CI −3.1 to 5.1). This is not a fight between physical therapy and chiropractic. The trials do not support one.

Harms from exercise. The same review counted them: a median of 0.14 adverse events per participant in the exercise groups (IQR 0.01 to 0.57) against 0.12 in the controls (IQR 0.02 to 0.32) — almost all of it muscle soreness. Read that carefully. Exercise will make you sore, and soreness is not injury. A flare in week three is a scheduling problem, not evidence the program is wrong.

Spinal manipulation. Rubinstein and colleagues, BMJ, 2019: 47 trials, 9,211 participants. Against guideline-recommended care, manipulation produced no statistically significant pain difference — −3.2 points, 95% CI −7.9 to +1.5. Function favored it slightly (standardized mean difference −0.25; 95% CI −0.41 to −0.09), both at moderate quality. In about half the trials examining harms, the reviewers could not tell whether adverse events were recorded systematically — so the true harm rate is uncertain rather than reassuringly small.

The guideline. The 2017 American College of Physicians guideline, by Qaseem and colleagues, strongly recommends starting chronic low back pain with nonpharmacologic care, and then lists thirteen options without ranking them. Exercise sits on moderate-quality evidence. Spinal manipulation sits on low-quality evidence. Thirteen options, no hierarchy — which is a guideline’s way of saying the evidence does not yet tell us who should get what.

What the punch-card version misses

None of this is an argument against movement, and none of it is a criticism of the clinicians delivering it. Loading a deconditioned spine is the right instinct, and a thrust inside a loading program is a legitimate adjunct. Symptom relief has value on the day you get it.

What the subscription model misses is everything that makes the evidence work:

• No named problem. “Nonspecific” was accepted as a starting assumption rather than reached as a conclusion, so red flags and inflammatory patterns were never formally cleared.
• No dose. A real program specifies load, sets, tempo, and progression, and moves them on a schedule. A handout of three bridges is not that.
• No date. Nothing was reassessed against a function you can name, so there was never a point at which the plan had to change.
• No endpoint honesty. Benefit is largest early and attenuates over the long term. It tracks with continued training. No trial sets an expiry date, and nobody can promise you one.

A program without a problem, a dose, and a date is not rehabilitation. It is a punch card.

What a real back program looks like

First, clear the things that are not therapy problems. Fever, unexplained weight loss, a cancer history, night pain, saddle numbness, bowel or bladder change, progressive weakness — those are urgent imaging problems and they get resolved before anyone lays a hand on your spine.

Then separate axial pain from radicular pain and from neurogenic claudication, and look for the inflammatory pattern: onset before forty, insidious rather than sudden, morning stiffness lasting over half an hour, pain that improves with movement and does not improve with rest, and night pain in the second half of the night that drives you out of bed. Axial spondyloarthritis is one of the most consistently delayed diagnoses in medicine, mistaken for mechanical back pain for years. An inflammatory spine does not belong under a thrust.

Second, localize — honestly. Extension and rotation that reproduce your pain raise the question of a facet source. They do not answer it. Physical examination performs poorly here, and any clinician who says the exam alone made the diagnosis is guessing.

The answer comes from comparative medial branch blocks under live fluoroscopy with contrast: two blocks, different anesthetics, on separate days, each requiring at least 80% relief lasting as long as that drug should last. As Dr. Padda notes in the video, a single uncontrolled block is wrong somewhere between a quarter and nearly half the time — which is why two are done, and why you should be told when they disagree. Sacroiliac pain sits below L5, is tested with provocation maneuvers, and is confirmed under image guidance. Discogenic pain worsens with sustained flexion and prolonged sitting — that is a pattern, not a test; the only confirmatory procedure is provocative discography, which carries its own controversy.

Naming the generator changes what you train and what you avoid — but only when the naming is honest about how sure we are.

Third, dose the exercise like a drug. Specified load, sets, and tempo. Progressed on a schedule. Reassessed at four and twelve weeks against a function you can name: how far you walk, whether you can lift your grandchild. If those numbers do not move, the plan changes. That is what separates rehabilitation from a punch card, and it is the whole substance of how a physical therapy program is built, dosed, and reassessed.

Fourth, be honest about adjuncts and about risk. Manipulation should be avoided or modified with osteoporosis, spinal cord compression, inflammatory instability such as rheumatoid involvement of the neck, or anticoagulation. Serious events are rare but real: cauda equina syndrome, disc herniation, osteoporotic fracture. Cervical arterial dissection has been reported after neck manipulation — the evidence discussed here concerns the low back, and a neck adjustment is a separate conversation worth having before anyone thrusts your neck. Estimates put cauda equina syndrome or a new disc herniation at roughly one per several million lumbar manipulations; treat that as a literature estimate rather than a figure from the three reviews above, particularly given how unsystematically harms were recorded.

Low-level laser therapy — that modality specifically, not consumer red-light panels — appears in the ACP list on low-quality evidence. The 660-to-850-nanometer photobiomodulation marketed for back pain has not been graded at all. Adjunctive and emerging. Never the plan.

What twelve weeks actually looks like

What follows is a composite — a picture assembled from many patients with this condition, not one person’s chart.

An adult two years past the original episode, punched card in hand. Red flags cleared; no inflammatory pattern. Two comparative medial branch blocks, separate days, different anesthetics, under live fluoroscopy — each relieved the pain for exactly as long as that drug should last. Hours, not weeks. That told us the facet joints were carrying part of the signal. It did not treat anything, and nobody pretended it had. What it bought was a decision: what to load, and what to leave alone.

Week three brought a flare after a hinge progression that cost a session and a good deal of trust.

By twelve weeks the pain change was modest — and it matters which number applies. This person had already had two years of adjustments and therapy, so the honest comparator is not the 15-point figure against no treatment. It is the 9-point figure against other active care, which sits below the threshold patients reliably notice. The change was real to this person. The trials would not call it clinically important. Both are true, and anyone who tells you only the first half is selling.

The functional gain stayed smaller than the pain gain — which is also the shape of the pooled trial data, about 7 points on function against about 15 on pain. Those are trial averages, not this person’s measured scores. And that is what Cochrane measured: size, not sequence. It says nothing about which one moves first. In this composite, walking moved first — that is a clinical observation from practice, not a trial finding. Individual results vary.

Then the program stopped over a holiday stretch, and ground was lost. The benefit tracks with continued training.

Frequently asked questions

Key takeaways

• Chronic nonspecific low back pain is rarely one fault in one structure: disc, facet, and sacroiliac sources overlap, and the signal is amplified by late-recruiting stabilizers, fear-avoidance, and increased gain in the dorsal horn.
• The pop of an adjustment is cavitation, not realignment; the subluxation model is unsupported, and manipulation’s leading proposed mechanism comes from anesthetized-animal work and is not established in humans.
• Exercise beats no treatment on pain by about 15 points out of 100 at moderate certainty, but its functional gain of roughly 7 points — and its roughly 9-point advantage over other active care — fall below the threshold patients reliably notice; exercise and manual therapy performed equivalently.
• Harms are mostly muscle soreness (a median of 0.14 events per exercise participant versus 0.12 in controls); serious manipulation-related events are rare but real, and about half the manipulation trials did not record harms systematically.
• Ask for the three things a subscription never provides: a named problem, a specified dose, and a date by which you will both know whether it worked.

Medically reviewed by Dr. Gurpreet Singh Padda, MD, MBA, MHP — Board Certified in Anesthesiology, Pain Medicine, Interventional Pain Management, Addiction Medicine, and Obesity Medicine. Last reviewed July 2026.

This article is educational and is not a substitute for evaluation, diagnosis, or treatment by a physician. Individual results vary. Do not start, stop, or change any medication without consulting your physician. To be evaluated, call (314) 481-5000 or text (314) 886-5902.

References

1. Hayden JA, Ellis J, Ogilvie R, Malmivaara A, van Tulder MW. Exercise therapy for chronic low back pain. Cochrane Database Syst Rev. 2021. PMID 34580864. PubMed
2. Rubinstein SM, de Zoete A, van Middelkoop M, Assendelft WJJ, de Boer MR, van Tulder MW. Benefits and harms of spinal manipulative therapy for the treatment of chronic low back pain: systematic review and meta-analysis of randomised controlled trials. BMJ. 2019. PMID 30867144. PubMed
3. Qaseem A, Wilt TJ, McLean RM, Forciea MA; Clinical Guidelines Committee of the American College of Physicians. Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians. Ann Intern Med. 2017. PMID 28192789. PubMed

Dr. Gurpreet Singh Padda, MD, MBA, MHP