There are many methods available to fit random-effects meta-analysis. However, until 2024, the only option available in RevMan has been the DerSimonian and Laird random-effects method. This method is known to have poor statistical performance in meta-analyses with characteristics commonly found in Cochrane reviews (e.g., meta-analyses with few studies). To address this issue, Cochrane is implementing new random-effects methods in RevMan for intervention reviews. These include a new method for estimating the between-study (heterogeneity) variance, calculating the confidence interval for the summary effect, and adding prediction intervals to aid in interpreting random-effects meta-analysis findings.

The Methods Support Unit is running two web clinics to introduce, explain and demonstrate the new methods. Both clinics will be presented by Professor Jo McKenzie and Dr Areti Angeliki Veroniki, Co-Convenors of the Cochrane Statistical Methods Group. 

The first clinic, 'Introduction to new random-effects methods in RevMan', took place on Wednesday 23rd October. The recording is available here.

The second clinic will take place on Thursday 12th December 2024 at 20.00 UTC: Demonstration of new random-effects methods in RevMan. Sign-up here. 

Methods Support Unit web clinics are freely accessible to all. Don't worry if you can't make the live session - sign-up and you will be sent the recording shortly afterwards

Posted by Rachel Richardson