Power

  • We hope to release a dedicated module for this in the coming year.

  • In the meantime, please consider these alternatives

  • In Stata, a simulation based approach is available in the package ipdpower by Kontopantelis

  • This calculates power for complex mixed effects two-level data structures, especially focusing on a treatment-covariate interactions. However, note that it amalgamates within-trial and across-trial information, and so will give inflated estimates of power.

  • For examples, type 'help ipdpower' after installing

  • In R, the ipdmeta package by Kovalchik can be used to estimate the power of an IPD meta-analysis to detect a specified treatment-covariate interaction based on entered aggregate data

  • It applied to a binary or continuous outcomes and a binary or continuous covariate

  • As analytic solutions are used, it is very fast; however, the binary outcome results are only an approximation. 

  • For examples, see here.

  • An online calculator for an IPD meta-analysis with a continuous outcome, for estimating a treatment-covariate interaction with either a binary or continuous covariates, is given by Joie Ensor here. A video demonstration is available. This also implements the Kovalchick and Cumberland analytic approach, so has the aforementioned limitation for binary outcomes.