Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
In a recent meta-analysis by Laudy et al,1 the authors used a fixed-effects model when there was significant heterogeneity for Western Ontario and McMaster Universities Osteoarthritis (WOMAC) knee pain (refer to figure 11 of the article, χ2=17.12, I2=94%, p<0.0001) and WOMAC physical function (refer to figure 13 of the article, χ2=16.40, I2=94%, p<0.0001), and stated that platelet-rich plasma (PRP) reduced pain and improved function more effectively than hyaluronic acid (HA).
Fixed-effects and random-effects models are the most commonly used statistical methods for meta-analysis, and systematic reviewers often need to choose one of the two models. In the presence of small heterogeneity, the fixed-effects model is more appropriate, while it inclines to the random-effects model when heterogeneity is large. Thus, dealing with heterogeneity is one of the most challenging tasks in meta-analysis.
Although I2 statistics are considered good indicators for assessing heterogeneity,2 it has been found that the test has poor …
Contributors ZW and YL conducted the study, analysed the data, and wrote and submitted the manuscript. ML planned the study, analysed the data and corrected the manuscript.
Competing interests None declared.
Provenance and peer review Commissioned; externally peer reviewed.