The short version
- Network meta-analyses rank PRP first among injectables for knee osteoarthritis, pooling many small trials.
- The largest rigorous placebo-controlled trial found no benefit on pain, on cartilage, or on 29 of its 31 secondary outcomes.
- When many small trials and one large careful trial disagree, the large careful trial usually wins. That is not a rule of thumb; it is what happens repeatedly across medicine.
- It is expensive and rarely covered, which matters more when the evidence is this contested.
Platelet-rich plasma is made from your own blood. A sample is drawn, spun to concentrate the platelets, and the concentrate is injected into the joint. The rationale is that platelets carry growth factors, and delivering them in concentration might reduce inflammation or support repair.
It is the most enthusiastically marketed injection in musculoskeletal medicine, and the one where the gap between marketing and evidence is widest.
Why does PRP rank so well in pooled analyses?
Because a lot of small trials report positive results, and network meta-analyses aggregate them.
One network meta-analysis pooling 35 studies and 3,104 patients ranked PRP first among the injectables it compared, ahead of hyaluronic acid and corticosteroid, and found no increase in treatment-related adverse events. Another, covering 79 trials and 8,761 patients, placed it favourably as well.
If that were the whole picture, PRP would be the obvious first choice. It is not the whole picture.
What did the largest rigorous trial find?
Nothing. That is worth stating plainly, because it is rarely mentioned on the pages selling the procedure.
The RESTORE trial randomised 288 people with symptomatic mild-to-moderate knee osteoarthritis to three weekly injections of leukocyte-poor PRP or saline placebo. Participants, injectors and assessors were all blinded, follow-up ran twelve months, and 93% completed the trial.
| Outcome at 12 months | PRP vs. placebo |
|---|---|
| Change in knee pain | −2.1 vs −1.8 points. Difference −0.4, not significant. |
| Medial tibial cartilage volume | −1.4% vs −1.2%. Difference −0.2%, not significant. |
| Prespecified secondary outcomes | 29 of 31 showed no significant difference between groups. |
The authors' conclusion was unambiguous: these findings do not support the use of PRP for the management of knee osteoarthritis.
Why do those two answers disagree?
This is the part worth understanding, because it applies well beyond PRP.
Pooling many small trials is not the same as running one large one. Small trials are more vulnerable to inadequate blinding, to selective reporting of whichever outcomes happened to reach significance, and to publication bias — positive results get written up and negative ones quietly do not. A meta-analysis inherits every one of those problems from its inputs, and can produce a confident-looking summary of a biased literature.
A single large trial with blinded participants, blinded injectors, blinded assessors, prespecified outcomes and a saline control is designed specifically to defeat those failure modes. When the two disagree, that is usually the direction the truth lies.
The fair counter-arguments
RESTORE tested one preparation — leukocyte-poor PRP, three weekly injections — in mild-to-moderate knee osteoarthritis. PRP is not a standardised product; platelet concentration, leukocyte content and preparation method vary enormously between systems, and a different formulation might perform differently.
It also studied the knee. Evidence for PRP in tendon problems is a separate literature with its own conclusions, and nothing here speaks to it.
Those are real limitations. What they are not is a reason to treat a well-conducted negative trial as though it did not happen.
How I discuss it
I tell people what is above: that the pooled evidence is favourable, that the best single trial was negative, and that I weight the second more heavily. Some still want to try it, and if they can afford it and understand what they are buying, that is a reasonable decision to make with open eyes.
That is a position on ambiguous evidence rather than a finding. What I try not to do is let enthusiasm around this treatment substitute for telling someone that the strongest study of it did not work.
The cost question
PRP is typically not covered by insurance for osteoarthritis and is usually paid out of pocket, generally as a series rather than a single injection.
That changes the calculation in a way it would not for a cheap treatment. A modest chance of modest benefit is a reasonable gamble at low cost. At high cost the same odds deserve a harder look — particularly if the money would otherwise go toward supervised strengthening, which has better evidence and more durable effects.
Sources
- Bennell KL, Paterson KL, Metcalf BR, et al. Effect of intra-articular platelet-rich plasma vs placebo injection on pain and medial tibial cartilage volume in patients with knee osteoarthritis: the RESTORE randomized clinical trial. JAMA 2021;326(20):2021–2030. 288 participants, fully blinded, 12-month follow-up. Retrieved via PubMed.
- Qiao X, Yan L, Feng Y, et al. Efficacy and safety of corticosteroids, hyaluronic acid, and PRP and combination therapy for knee osteoarthritis: a systematic review and network meta-analysis. BMC Musculoskeletal Disorders 2023;24(1):926. Retrieved via PubMed.
- Anil U, Markus DH, Hurley ET, et al. The efficacy of intra-articular injections in the treatment of knee osteoarthritis: a network meta-analysis of randomized controlled trials. The Knee 2021;32:173–182. Retrieved via PubMed.