Sample size calculation for meta-epidemiological studies

Meta-epidemiological studies are used to compare treatment effect estimates between randomized clinical trials with and without a characteristic of interest. To our knowledge, there is presently nothing to help researchers to a priori specify the required number of meta-analyses to be included in a meta-epidemiological study. We derived a theoretical power function and sample size formula in the framework of a hierarchical model that allows for variation in the impact of the characteristic between trials within a meta-analysis and between meta-analyses. A simulation study revealed that the theoretical function overestimated power (because of the assumption of equal weights for each trial within and between meta-analyses). We also propose a simulation approach that allows for relaxing the constraints used in the theoretical approach and is more accurate. We illustrate that the two variables that mostly influence power are the number of trials per meta-analysis and the proportion of trials with the characteristic of interest. We derived a closed-form power function and sample size formula for estimating the impact of trial characteristics in meta-epidemiological studies. Our analytical results can be used as a 'rule of thumb' for sample size calculation for a meta-epidemiologic study. A more accurate sample size can be derived with a simulation study.