Interpreting the Meta-analysis of Platelet-Rich Plasma Injections for the Treatment of Knee Osteoarthritis With Caution: Further Research Is Needed: Letter to the Editor

Dear Editor:

We read with great interest the meta-analysis by Bensa et al ¹ on platelet-rich plasma (PRP) injections for the treatment of knee osteoarthritis. We appreciate the effort to synthesize the available evidence, but we would like to raise several critical issues that warrant clarification and reconsideration of the study’s conclusions.

First, while the authors conclude that high-platelet PRP provides clinically significant pain relief, the confidence intervals for the pooled effects often include or barely exceed the minimal clinically important difference (MCID) threshold. For example, for visual analog scale scores at 12 months, the reported mean difference (−1.95) has a confidence interval of −3.29 to −0.60, which overlaps the MCID threshold (1.37). Therefore, the observed benefit cannot be conclusively considered clinically meaningful.

Second, although the study protocol was registered in PROSPERO, the authors performed several analyses that were not prespecified in the original protocol, such as those based on injection schedule or a meta-regression on PRP-injected volume. These unplanned analyses should have been clearly identified as post hoc and properly justified. Importantly, the authors omit a crucial analysis that is only presented in the supplementary appendix: the subgroup analysis based on the risk of bias of selected studies. According to this interesting analysis, trials with a low risk of bias found no statistically significant difference between PRP and placebo, whereas trials with a high or moderate risk of bias did. This omission is problematic, particularly in light of Cochrane’s strong recommendations to interpret findings in relation to study quality. ²

Beyond these major concerns, we identify additional methodological issues. First, the MCID values were applied uniformly across all studies, regardless of differences in outcome scales, population characteristics, or clinical context. Applying a single MCID to heterogeneous data sources risks misrepresenting clinical relevance. Second, the analysis does not incorporate stratification by study quality, despite including trials with substantial risk of bias. A more appropriate approach would have been to present pooled estimates restricted to low-risk studies. Third, clinical heterogeneity across trials is substantial (eg, PRP leukocyte content, number of injections, and osteoarthritis severity), yet the meta-analysis only explores platelet concentration and injected volume. Other key factors such as patient characteristics, including age, sex, body weight, and baseline functional status, were reported descriptively but not examined through subgroup analyses or meta-regression. Fourth, methodological details (eg, injection protocol, leukocyte content, risk of bias, and injection volume) are buried in supplementary materials and not adequately summarized or discussed in the main text, which limits the reader’s ability to assess the external validity of the findings. Additionally, the article lacks a clear explanation of the methods used to identify the presence and extent of statistical heterogeneity, such as the I² statistic or Cochran’s Q test. This omission is concerning, especially given the very high heterogeneity (I² > 75%) found in multiple meta-analyses (only reported in the appendices), which demands careful interpretation.

In summary, while the authors provide a valuable synthesis, the conclusions overstate the evidence by affirming clinically meaningful benefits without properly qualifying the important methodological limitations and heterogeneity. In its current form, the conclusion may mislead readers about the true strength and applicability of the evidence.