Greetings,

I am working on a meta-analysis examining how heat exposure influences neuromuscular function. We have several RCT's, but have found that some outcomes report data in each condition (between-groups; hot and temperate) with several timepoints for the within-groups comparison (e.g. mean and SD of outcome data at 10 sec, 20 sec... 120 sec).

My question is whether it is possible to appropriately determine an effect size that is representative of the entire data set.

One suggestion I have considered is to select a representative time point, based on reasoning, e.g. start, point of fatigue, end etc. However, I do not feel this sufficiently contrasts all data available.

Another option is to combine effect size data in comprehensive meta-analysis. This would involve establishing effect sizes between the conditions at each time point, inputting them to the software and then creating a 'combined' effect for the outcome. This has the potential to consider the "likely) correlated nature of the data.

Alternatively, data could be input as subgroups for each condition. This seems to be possible using the following calculator (https://www.campbellcollaboration.org/escalc/html/EffectSizeCalculator-SMD27.php). However, I am worried that this is designed for independent samples, whereas these measures are likely to be correlated (same participants), and hence, have implications on estimate precision.

Lastly, I have been investigating multilevel MA, and am wondering whether this may be suitable for outcomes such as this.

Any insights would be greatly appreciated.

Steve