 |
 |
Dear ALLSTAT, Most publications of crossover trial results report standard deviations for treatment groups, which relate to between-patient variability, whereas what is needed for meta-analysis is a measure of between-period variability for patients. It doesn't seem possible to extract the latter from the former, but I'm writing in the (forlorn?) hope that perhaps someone knows of a decent approximation that could be used, based on the commonly reported treatment group means and standard deviations. Regards, Sean McGuigan