American Health & Medical Research Association
Pain, fatigue & chronic conditions

Physical therapy matched surgery in several trials, with fine print worth reading

Trials pitting physical therapy against surgery and injections, what exercise therapy really delivers for back and joint pain, and how access and Medicare rules work in 2026.

By the Health Research USA editorial team

Published Last updated October 2026

Written from the sources listed below; links checked October 2026. It touches on medicines, doses or health conditions and has not yet been reviewed by a licensed clinician, so use it to prepare for a conversation with yours. How we research and review

Sources25, listed below Reading time16 minutes
On this page
  1. What US guidelines actually name
  2. Exercise therapy works, by less than many people think
  3. The surgery trials, and the people who crossed over
  4. Physical therapy versus a steroid shot
  5. Does seeing a therapist early save money?
  6. Getting in the door, and what Medicare really limits
  7. What a good first visit looks like, and what to ask
  8. When physical therapy is not the next step
  9. The single most useful takeaway

Her knee aches on the stairs. An MRI shows a torn meniscus, the rubbery cartilage cushion inside the knee. One option on the table is an arthroscopy, a keyhole operation to trim the tear. The other is eight weeks of physical therapy.

Dutch researchers put that exact choice to a test. They randomly assigned 321 people aged 45 to 70 with this kind of tear to surgery or to 16 sessions of exercise therapy. Over two years, physical therapy was statistically no worse than surgery at restoring knee function. And 29% of the physical therapy group had the operation anyway.

Both numbers are true, and the second one is the fine print this article is about. A run of randomized trials has found that physical therapy, built mostly on exercise, produced results similar to surgery for several common problems and beat a steroid injection for knee arthritis. The same trials show who crossed over to surgery, where surgery clearly won, and how modest the effect of exercise itself turns out to be.

The National Institute of Neurological Disorders and Stroke describes physical therapy as helping people recover after injury or surgery and manage chronic pain, with low-impact cardio, mobility, flexibility and strength work. The evidence below is mostly about that exercise core.

What US guidelines name

Start with a correction. Guidelines rarely recommend “physical therapy” as such. They name exercise, which physical therapists commonly deliver.

The Centers for Disease Control and Prevention’s 2022 opioid prescribing guideline says in its second recommendation, rated category A with evidence type 2, that nonopioid therapies are preferred for subacute and chronic pain. Its supporting text names exercise therapy, which it calls a prominent part of physical therapy, for back pain, fibromyalgia and hip or knee osteoarthritis. It adds that physical therapy can be especially helpful for people with limited access to safe public spaces, and that cost and access can be barriers to nonpharmacologic treatments, particularly for people who are uninsured or have limited income.

For chronic low back pain, the American College of Physicians’ 2017 guideline tells clinicians and patients to start with nonpharmacologic treatment, and lists exercise, multidisciplinary rehabilitation and motor control exercise among the choices, as a strong recommendation. How the rest of back pain care should go, from imaging to medicines, belongs to our back pain article.

Exercise therapy works, by less than many people think

The best evidence summaries agree that exercise helps. They part ways with the popular impression of how much.

For chronic low back pain, a 2021 Cochrane review led by Dalhousie University researchers pooled 249 trials, with data extracted through April 2018. Compared with no treatment, usual care or placebo, exercise lowered pain by about 15 points on a 100-point scale, right at the 15-point line the reviewers set for a difference that matters. Function improved by about 7 points, short of their 10-point threshold. Exercise beat education alone and physical therapy without exercise, such as electrical treatments, but did no better than hands-on manual therapy. Studies that reported side effects found at least one in 33% of exercise groups and 29% of comparison groups, mostly minor problems such as muscle soreness.

For knee osteoarthritis, Cochrane’s 2024 update pooled 139 trials and 12,468 people. Against usual care or no treatment, exercise improved pain by about 13 points and function by about 12.5 points on 100-point scales. Measured against thresholds for meaningful change, the reviewers concluded “these benefits were of uncertain clinical importance.” Most participants knew which group they were in, which may have inflated the results. Exercise may also raise the chance of adverse events compared with no treatment, on low-certainty evidence.

That is a shift. The 2015 version of the same review rated its pain evidence high quality and called the effect comparable with estimates for anti-inflammatory drugs. More trials made the picture more cautious, not less.

Hip osteoarthritis looks weaker still. Cochrane’s July 2026 update, 18 studies and 1,368 people, found exercise improved pain and function slightly compared with usual care, but judged the gains unlikely to be clinically meaningful. Against attention control or placebo, it may have little to no effect on pain. A 2014 Australian trial in JAMA gave 102 people with painful hip arthritis ten sessions of education, manual therapy, home exercise and a walking aid if needed, or a sham of switched-off ultrasound and inert gel. The authors concluded that “physical therapy did not result in greater improvement in pain or function compared with sham treatment, raising questions about its value for these patients.” Mild side effects hit 41% of the active group, against 14% with the sham.

The federal Agency for Healthcare Research and Quality’s 2020 review of noninvasive treatments for five chronic pain conditions sketched the same modest picture. Several treatments, exercise among them, brought small to moderate benefits that lasted beyond a month, with sparse evidence past a year. For chronic neck pain, exercise showed no clear gain in pain or function over the short or intermediate term. For knee osteoarthritis, therapeutic ultrasound did no better than a sham, and TENS, a skin-electrode nerve stimulator, did no better than a placebo device in one trial, both on low-strength evidence. No serious harms turned up, though harms were poorly reported.

On the library’s verdict scale, exercise for chronic back pain and knee arthritis lands in the first category, because US guidelines name it, while its measured average effect sits near or below the threshold patients notice. Those two facts do not conflict.

The surgery trials, and the people who crossed over

This is where physical therapy earned its reputation. The crossover rates belong in every retelling.

Degenerative meniscus tears

  • MeTeOR, a US multicenter trial funded by the National Institute of Arthritis and Musculoskeletal and Skin Diseases, randomly assigned 351 people 45 or older with a meniscus tear and mild to moderate arthritis to surgery plus therapy or to a standardized therapy program. At six months, function improved 20.9 points after surgery and 18.5 after therapy alone, a difference within chance. But 30% of the therapy group had surgery within those six months. At five years, pain had improved by similar amounts in both groups and stayed stable.
  • ESCAPE, the Dutch trial in the opening, required tears that did not lock the knee and excluded unstable knees, severe arthritis and a body mass index above 35. Therapy was noninferior over two years; 29% crossed over to surgery.
  • A Norwegian trial in the BMJ in 2016 gave 140 adults, average age 49.5, either 12 weeks of supervised exercise or surgery. Knee scores at two years differed by less than a point, and thigh strength was better in the exercise group at three months. The 19% who crossed over to surgery gained nothing extra from it.

Separately, a Finnish trial found meniscus surgery no better than a sham operation.

Lumbar spinal stenosis

A US trial funded by the National Institutes of Health enrolled 169 people 50 or older with spinal stenosis, a narrowing of the spaces around the spinal nerves, who were surgical candidates and had agreed to surgery. They were randomly assigned to decompression surgery or physical therapy. At two years, physical function had improved 22.4 points after surgery and 19.2 after therapy, with no difference between groups. But 57% of the therapy group crossed over to surgery. Because the trial had no untreated comparison group, the authors noted, it cannot show how much of either improvement came from the treatment itself. They called for shared decision-making with full disclosure of the evidence.

Shoulder impingement

The Finnish FIMPACT trial, in the BMJ in 2018, randomly assigned 210 people with shoulder impingement to decompression surgery, a diagnostic arthroscopy that served as placebo surgery, or exercise therapy. Decompression was no better than the placebo operation. Against exercise therapy it was statistically better by 7.5 to 12 points on a 100-point pain scale, below the 15-point threshold the researchers set for a meaningful difference, and the authors cautioned that this comparison was likely biased in surgery’s favor. Fourteen of the 71 people assigned to exercise, about one in five, later had the decompression operation.

Where surgery clearly won

A Danish trial in the New England Journal of Medicine in 2015 enrolled 100 people with moderate to severe knee arthritis who were eligible for a total knee replacement. Replacement plus 12 weeks of non-surgical care, including exercise, education, diet advice, insoles and pain medicine, beat the 12 weeks of non-surgical care alone: 32.5 points of improvement against 16.0. Surgery carried a price. There were 24 serious adverse events in the replacement group and 6 in the other. And 26% of the non-surgical group had a replacement within the year, which means most did not.

So “physical therapy is as good as surgery” is too broad. For degenerative meniscus tears and stenosis in surgical candidates, a course of therapy produced similar function on average, yet between about a fifth and more than half of the therapy groups went on to surgery. For shoulder impingement, surgery’s edge over exercise was real on paper but below what patients notice, and here too about a fifth of the exercise group later had the operation. For advanced knee arthritis, replacement did more. Whether trying therapy first changes how well a later operation works is not something these trials measured. Acute ACL tears in young adults have their own rehab-first trials, which our article on rehabilitation after surgery and injury takes up.

Physical therapy versus a steroid shot

A trial in US Military Health System primary care clinics, published in the New England Journal of Medicine in 2020, randomly assigned 156 people with knee arthritis, average age 56, to physical therapy or a glucocorticoid (steroid) injection. At one year, the therapy group scored 37.0 on a 240-point pain, stiffness and function scale where lower is better, against 55.8 for the injection group, a difference of 18.8 points favoring therapy. One patient fainted during an injection.

A 2022 cost analysis of the same trial found that although therapy can cost more up front, one-year knee-related costs came out nearly identical: $2,131 for therapy and $2,113 for injections, with better quality of life in the therapy group. The wider evidence on joint injections, from steroids to hyaluronic acid, is the subject of our arthritis coverage.

Does seeing a therapist early save money?

A widely cited 2012 study in Spine examined 32,070 employer-insured patients who saw a primary care clinician for new low back pain. Only 7% saw a physical therapist within 90 days. Those who started within 14 days had lower odds of later advanced imaging, surgery, injections and opioid prescriptions, and their back pain costs over 18 months were $2,736 lower than for people who started later.

That study was observational, so the people referred early may have differed in ways the data could not capture. Two randomized trials tested the idea directly. In a 2015 JAMA trial by a University of Utah team, with 220 adults whose back pain had begun less than 16 days earlier, four sessions of manipulation and exercise improved a disability score by 3.2 points at three months, about half the 6-point change the investigators considered meaningful. The difference was no longer significant at one year, pain intensity never differed, and neither did health care use. In a 2018 military trial of 119 service members, average age 27, early therapy helped at four weeks but not at one year, total costs were similar, and back pain costs were higher in the therapy group.

So the savings showed up in records, then failed to appear in trials. That does not make early therapy useless, only the cost argument weaker than the often-quoted figure suggests.

Getting in the door, and what Medicare really limits

A referral may not be needed at all. According to the American Physical Therapy Association’s July 2025 report, all 50 states, the District of Columbia and the US Virgin Islands allow some form of direct access to physical therapists. Twenty-one states allow it without restrictions; the rest set time or visit limits or require a referral for particular tests or treatments. Use was low in one look: in a 2016 Wisconsin survey the report cites, fewer than 10% of patients arrived through direct access at all but one facility offering it. Insurance plans can be stricter than state law, a gap our overview of rehabilitation explains.

Medicare Part B has not required a physician referral or visit to start outpatient physical therapy since 2005, the association reports. Patients must still be under a physician’s care, shown, for example, by a physician certifying the therapist’s plan of care.

Medicare’s old hard caps on outpatient therapy were repealed by the Bipartisan Budget Act of 2018. What remains, according to the Centers for Medicare & Medicaid Services, is a threshold that triggers paperwork. For 2026, once a patient’s yearly physical therapy and speech-language pathology expenses together pass $2,480, or occupational therapy passes a separate $2,480, claims must carry a billing code called the KX modifier, which confirms the services are medically necessary and documented in the record. Claims above the threshold without it are denied. Targeted medical review can begin at $3,000, a level CMS says stays put until 2028. Since January 2022, Medicare has paid 85% of the usual rate for services furnished in whole or in part by physical therapist assistants.

So “Medicare caps physical therapy at $2,480” is wrong. The number is a documentation line, not a limit on medically necessary care.

What a good first visit looks like, and what to ask

Is physical therapy just exercises you could copy from a video? Partly, and the evidence is candid about it. The 2024 knee arthritis review found no link between results and the type of exercise, the number of sessions prescribed, or how much live contact people had with a health professional. But the programs that matched surgery were structured and supervised, with 16 therapy sessions in ESCAPE and 12 weeks of supervised exercise in the Norwegian trial, and no video screens out the locking knee or the back pain that needs a doctor first. The assessment and the plan are what a visit adds.

So a first physical therapy visit for pain should feel like an assessment, not a class. Here is our own checklist, drawn from what the trials above suggest matters:

  • A history and examination that ask what you can’t do, not only where it hurts.
  • A goal in daily-life terms, such as climbing stairs without the rail or getting through a workday, with a date to check it.
  • An exercise program for between visits, since exercise is the part of therapy with the best evidence.
  • A reason for every passive treatment. Ultrasound and TENS did no better than shams for knee arthritis in the trials the federal review examined.
  • A plan for what happens if progress stalls.

Questions worth asking:

  1. What should I notice by week four, and how will we measure it?
  2. Which exercises matter most if I only have ten minutes a day?
  3. If this isn’t working, when do we reconsider, and who else should be involved?
  4. How many visits does my plan cover, and will my care go past Medicare’s documentation threshold?

When physical therapy is not the next step

  • A knee that truly locks, catching so it can’t straighten. ESCAPE enrolled only tears that did not lock the knee, so its results do not describe this situation, and it calls for an orthopedic evaluation.
  • Back pain with new weakness or numbness in a leg, or a change in bladder or bowel control. These are medical questions before they are therapy questions, and our back pain article sets out which warning signs mean same-day or emergency care.
  • No meaningful gain after a fair trial. The stenosis and meniscus trials built in the option of surgery for people who did not improve. Whether to continue, change course or consider an operation belongs with the clinician following the case, ideally with the therapist’s notes in hand.
  • Hip arthritis pain that is not budging with therapy. In the sham-controlled hip trial, ten sessions did no better than a sham, which makes stalled progress a reason to reassess the plan rather than simply extend it.

The single most useful takeaway

It helps to frame a course of physical therapy as a time-limited test with a named goal. The trials support it as a reasonable first move for degenerative meniscus tears and spinal stenosis, show it beating a steroid shot at one year for knee arthritis, and back its exercise core for back pain. They also show that many people in the surgery trials went on to an operation, that knee replacement did more for advanced arthritis, and that exercise’s average effect is modest. The useful version sets a goal and a review date with the therapist at the start. Any decision to continue, change course or consider surgery then belongs with the clinician following the case, and the fine print starts working in your favor.

Sources

  1. Centers for Disease Control and Prevention. “Dowell D, Ragan KR, Jones CM, Baldwin GT, Chou R. CDC Clinical Practice Guideline for Prescribing Opioids for Pain, United States, 2022. MMWR Recommendations and Reports, 2022;71(3):1-95.” cdc.gov
  2. Annals of Internal Medicine (American College of Physicians). “Qaseem A, Wilt TJ, McLean RM, Forciea MA, et al. "Noninvasive Treatments for Acute, Subacute, and Chronic Low Back Pain: A Clinical Practice Guideline From the American College of Physicians." Ann Intern Med, 2017;166(7):514-530. PubMed. DOI: 10.7326/M16-2367.” doi.org
  3. National Institute of Neurological Disorders and Stroke. “Back Pain (last reviewed March 13, 2026).” ninds.nih.gov
  4. Agency for Healthcare Research and Quality. “Skelly AC, Chou R, Dettori JR, et al. Noninvasive Nonpharmacological Treatment for Chronic Pain: A Systematic Review Update. Comparative Effectiveness Review No. 227, AHRQ Pub. No. 20-EHC009, April 2020 (summary).” effectivehealthcare.ahrq.gov
  5. Cochrane Database of Systematic Reviews. “Hayden JA, Ellis J, Ogilvie R, Malmivaara A, van Tulder MW. "Exercise therapy for chronic low back pain." Cochrane Database Syst Rev, 2021;9(9):CD009790. PubMed. DOI: 10.1002/14651858.CD009790.pub2.” doi.org
  6. Cochrane Database of Systematic Reviews. “Lawford BJ, Hall M, Hinman RS, et al. "Exercise for osteoarthritis of the knee." Cochrane Database Syst Rev, 2024;12(12):CD004376. PubMed. DOI: 10.1002/14651858.CD004376.pub4.” doi.org
  7. Cochrane Database of Systematic Reviews. “Fransen M, McConnell S, Harmer AR, Van der Esch M, Simic M, Bennell KL. "Exercise for osteoarthritis of the knee." Cochrane Database Syst Rev, 2015;1:CD004376 (superseded by the 2024 update). PubMed. DOI: 10.1002/14651858.CD004376.pub3.” doi.org
  8. Cochrane Database of Systematic Reviews. “Hall M, Lawford BJ, Hinman RS, et al. "Exercise for osteoarthritis of the hip." Cochrane Database Syst Rev, 2026;7(7):CD007912. PubMed. DOI: 10.1002/14651858.CD007912.pub3.” doi.org
  9. JAMA. “Bennell KL, Egerton T, Martin J, et al. "Effect of physical therapy on pain and function in patients with hip osteoarthritis: a randomized clinical trial." JAMA, 2014;311(19):1987-1997. PubMed. DOI: 10.1001/jama.2014.4591.” doi.org
  10. New England Journal of Medicine. “Katz JN, Brophy RH, Chaisson CE, et al. "Surgery versus physical therapy for a meniscal tear and osteoarthritis." N Engl J Med, 2013;368(18):1675-1684. PubMed. DOI: 10.1056/NEJMoa1301408.” doi.org
  11. Arthritis & Rheumatology. “Katz JN, Shrestha S, Losina E, et al. "Five-Year Outcome of Operative and Nonoperative Management of Meniscal Tear in Persons Older Than Forty-Five Years." Arthritis Rheumatol, 2020;72(2):273-281. PubMed. DOI: 10.1002/art.41082.” doi.org
  12. JAMA. “van de Graaf VA, Noorduyn JCA, Willigenburg NW, et al. "Effect of Early Surgery vs Physical Therapy on Knee Function Among Patients With Nonobstructive Meniscal Tears: The ESCAPE Randomized Clinical Trial." JAMA, 2018;320(13):1328-1337. PubMed. DOI: 10.1001/jama.2018.13308.” doi.org
  13. BMJ. “Kise NJ, Risberg MA, Stensrud S, Ranstam J, Engebretsen L, Roos EM. "Exercise therapy versus arthroscopic partial meniscectomy for degenerative meniscal tear in middle aged patients: randomised controlled trial with two year follow-up." BMJ, 2016;354:i3740. PubMed. DOI: 10.1136/bmj.i3740.” doi.org
  14. New England Journal of Medicine. “Sihvonen R, Paavola M, Malmivaara A, et al. "Arthroscopic partial meniscectomy versus sham surgery for a degenerative meniscal tear." N Engl J Med, 2013;369(26):2515-2524. PubMed. DOI: 10.1056/NEJMoa1305189.” doi.org
  15. Annals of Internal Medicine. “Delitto A, Piva SR, Moore CG, et al. "Surgery versus nonsurgical treatment of lumbar spinal stenosis: a randomized trial." Ann Intern Med, 2015;162(7):465-473. PubMed. DOI: 10.7326/M14-1420.” doi.org
  16. BMJ. “Paavola M, Malmivaara A, Taimela S, et al. "Subacromial decompression versus diagnostic arthroscopy for shoulder impingement: randomised, placebo surgery controlled clinical trial." BMJ, 2018;362:k2860. PubMed. DOI: 10.1136/bmj.k2860.” doi.org
  17. New England Journal of Medicine. “Skou ST, Roos EM, Laursen MB, et al. "A Randomized, Controlled Trial of Total Knee Replacement." N Engl J Med, 2015;373(17):1597-1606. PubMed. DOI: 10.1056/NEJMoa1505467.” doi.org
  18. New England Journal of Medicine. “Deyle GD, Allen CS, Allison SC, et al. "Physical Therapy versus Glucocorticoid Injection for Osteoarthritis of the Knee." N Engl J Med, 2020;382(15):1420-1429. PubMed. DOI: 10.1056/NEJMoa1905877.” doi.org
  19. JAMA Network Open. “Rhon DI, Kim M, Asche CV, Allison SC, Allen CS, Deyle GD. "Cost-effectiveness of Physical Therapy vs Intra-articular Glucocorticoid Injection for Knee Osteoarthritis: A Secondary Analysis From a Randomized Clinical Trial." JAMA Netw Open, 2022;5(1):e2142709. PubMed. DOI: 10.1001/jamanetworkopen.2021.42709.” doi.org
  20. Spine. “Fritz JM, Childs JD, Wainner RS, Flynn TW. "Primary care referral of patients with low back pain to physical therapy: impact on future health care utilization and costs." Spine, 2012;37(25):2114-2121. PubMed. DOI: 10.1097/BRS.0b013e31825d32f5.” doi.org
  21. JAMA. “Fritz JM, Magel JS, McFadden M, et al. "Early Physical Therapy vs Usual Care in Patients With Recent-Onset Low Back Pain: A Randomized Clinical Trial." JAMA, 2015;314(14):1459-1467. PubMed. DOI: 10.1001/jama.2015.11648.” doi.org
  22. Spine. “Rhon DI, Miller RB, Fritz JM. "Effectiveness and Downstream Healthcare Utilization for Patients That Received Early Physical Therapy Versus Usual Care for Low Back Pain: A Randomized Clinical Trial." Spine, 2018;43(19):1313-1321. PubMed. DOI: 10.1097/BRS.0000000000002619.” doi.org
  23. American Physical Therapy Association. “State of Direct Access to Physical Therapist Services (report, July 2025).” apta.org
  24. Centers for Medicare & Medicaid Services. “Therapy Services (last modified March 10, 2026).” cms.gov
  25. Centers for Medicare & Medicaid Services. “Transmittal 13437, Change Request 14252: 2026 Annual Update of Per-Beneficiary Threshold Amounts (issued October 30, 2025; effective January 1, 2026).” cms.gov