The short answer. Exercise, self-management and education programs, weight loss for people who are overweight, and anti-inflammatory medicines (NSAIDs) have the firmest support in the two main US guidelines for knee arthritis1,2. Cortisone injections give short-term relief1. For longer relief, hyaluronic acid (viscosupplementation), platelet-rich plasma (PRP) and bone marrow aspirate concentrate (BMAC) injections may help: in a 2024 analysis of 48 randomized trials, all three outperformed cortisone for pain and function at six months or longer3. In a review of BMAC trials, patients improved in pain and function starting one month after the injection4.
About 33 million US adults have osteoarthritis, and the knee is one of the joints it most often affects5. One study estimated that about 14 million people in the US have knee osteoarthritis with symptoms, and that more than half of them are younger than 656.
Two guidelines are the main reference points. The American Academy of Orthopaedic Surgeons (AAOS) guideline, published in 2021, covers knee osteoarthritis treatment short of knee replacement1. The American College of Rheumatology (ACR) and the Arthritis Foundation guideline, published in 2019, covers the hand, hip and knee2.
Both grade each recommendation by how strong the evidence is1,2. The ACR uses strong and conditional, and explains that conditional recommendations depend more on a patient's values and preferences and call for shared decision-making2. The ACR also says it does not rank one recommended option over another except by the strength of the recommendation2.
The table below lists the treatments both guidelines recommend. Hyaluronic acid, PRP and BMAC injections have their own section, based on published studies. This post is about evidence and the order of care; for national cost figures, see knee arthritis treatment options and what they cost.
These are recommended by both guidelines, and strongly by at least one:
| Treatment | AAOS 2021 | ACR 2019 |
|---|---|---|
| Exercise | Strong, for [1] | Strong, for [2] |
| Self-management and education programs | Strong, for [1] | Strong, for [2] |
| Weight loss, if overweight | Moderate, for [1] | Strong, for [2] |
| Topical NSAIDs (gels and creams) | Strong, for [1] | Strong, for [2] |
| Oral NSAIDs | Strong, for [1] | Strong, for [2] |
| Knee brace | Moderate, for [1] | Strong, for (tibiofemoral brace) [2] |
| Cane | Moderate, for [1] | Strong, for [2] |
| Cortisone injection | Moderate, for short-term relief [1] | Strong, for [2] |
The ACR also strongly recommends tai chi, a mind-body practice that combines meditation with slow, gentle movements2.
Exercise sits at the center. The ACR says all patients should be encouraged to make some form of exercise a central part of their treatment plan, and that programs tend to work better when supervised, often by a physical therapist2. A 2024 Cochrane review of 139 trials found that exercise probably improves pain, physical function and quality of life in the short term7.
Weight loss works better combined with exercise2. The ACR notes that losing 5% or more of body weight can be associated with improvement2. In the IDEA trial of 454 older adults who were overweight or obese, diet plus exercise led to less pain and better function after 18 months than diet or exercise alone8.
Physical therapy compared with cortisone. In a US trial of 156 patients, people who had physical therapy had less pain and disability at one year than those who had a cortisone injection9. In the studies the AAOS reviewed, cortisone relief lasted up to about three months1. As a safety point, the AAOS notes a potential risk that cortisone injections may speed up arthritis1.
Medicines. The ACR advises starting with treatments that have the least effect on the rest of the body, which is why it suggests topical NSAIDs before oral ones2. It also advises keeping oral NSAID doses as low, and courses as short, as possible2. Ask your doctor before starting any medicine, since other health conditions affect which ones are safe for you2.
Acetaminophen is recommended by both guidelines: strongly by the AAOS, to improve pain and function when it is not contraindicated, and conditionally by the ACR1,2. The ACR also conditionally recommends acupuncture for knee arthritis2.
A 2024 network meta-analysis pooled 48 randomized trials covering 9,338 knees, each with at least six months of follow-up. It compared cortisone, hyaluronic acid, PRP and BMAC injections. PRP, BMAC and hyaluronic acid were the treatments most likely to improve both pain and function, in that order, and all three outperformed cortisone3.
Hyaluronic acid (viscosupplementation). Hyaluronic acid was the injection studied most often in that analysis3. A separate claims study followed 1,567,024 people with knee arthritis in a large commercial insurance database over 4.5 years. Among patients who went on to have a knee replacement, the jump in costs from surgery came later for those treated with hyaluronic acid (around 16 to 17 months) than for those who were not (around 5 to 6 months). The authors say this reflects a longer time before knee replacement10.
PRP. PRP ranked highest of the injections in the 2024 analysis, ahead of BMAC, hyaluronic acid and cortisone. The authors concluded that at six months or longer, PRP significantly improved pain and function for people with knee arthritis3.
BMAC. A 2024 systematic review of 8 randomized trials with 937 patients found that people treated with BMAC improved significantly in pain and function scores, starting one month after the injection4. In the same review, BMAC compared favorably with hyaluronic acid for pain at 6 and 12 months, and no serious adverse events linked to BMAC were reported4. The review authors concluded that BMAC injections provided pain relief and functional improvement for people with knee arthritis4.
The ACR guideline does not set a single sequence and says the order will vary from patient to patient2. Read together, the ratings and studies suggest an order like this (our reading, not a sequence either guideline sets):
At our practice, physical therapy and viscosupplementation come first in the order of care. PRP, BMAC or the two combined may be considered when those have not worked. Not every patient is a candidate, and a proper orthopedic evaluation is essential.
For some people, the evidence suggests it can. In a trial of 100 people with moderate to severe knee arthritis who were eligible for a replacement, those assigned to 12 weeks of exercise, education, dietary advice, insoles and pain medicine improved, though less than those who also had surgery. Surgery also brought more serious adverse events (24 vs. 6)11. After two years, two out of three people in the non-surgical group had not had the operation12.
That is not a promise for any one knee. It does suggest a structured non-surgical program is a reasonable first step. Surgery stayed available: 16 people in the non-surgical group went on to have a replacement within two years12.
If you are in the Santa Barbara area, Dr. Scheinberg can examine your knee, review any imaging and talk through the options. See knee conditions or call (805) 682-1394.
Related reading: how to avoid knee replacement and arthritis treatment without surgery.
Short answers to the questions patients ask most.
Have a question about your own situation?
Call (805) 682-1394Yes. Both major US guidelines recommend a range of non-surgical treatments, led by exercise, self-management programs, weight loss for people who are overweight, and NSAIDs1,2. In one trial, two out of three people eligible for knee replacement had not had surgery two years after a 12-week non-surgical program12.
Studies have found that it can. In a 2024 network meta-analysis of 48 randomized trials, PRP ranked highest of the injections compared for improving pain and function at six months or longer, ahead of BMAC, hyaluronic acid and cortisone3. Not every patient is a candidate, and a proper orthopedic evaluation is essential.
Reviewed by Dr. Richard Scheinberg, MD, a board-certified orthopedic surgeon in Santa Barbara with 40 years of surgical experience.
