|
Input parameter | Definition | Value | Comments |
|
Starting trial sample size | Sample size sufficient for standard statistical techniques | 30 | As recommended in a prior publication [20] |
Large trial sample size | 90th percentile of largest trial size included in published meta-analyses | 256 | Rounded up to nearest multiple of 2 |
Population effect size | Median absolute effect size among published meta-analyses | 0.77 | On a previous systematic review, median effect size for quantitative outcomes was 0.8 [21] |
SD of effect size | Median of the 95% CI half-width divided by 1.96 | 0.27 | Similar to previously published IQR of effect sizes for quantitative outcomes (0.6–1.0) [21] |
Minimum effect size deemed clinically relevant | 10th percentile of the absolute effect sizes among published meta-analyses | 0.11 | 0.10–0.34 is considered a small effect size; 0.35–0.64 is considered a medium effect size; 0.65–1.19 is considered a large effect size [22] |
Number of trials needed for earliest meta-analysis | 10th percentile of the number of trials among published meta-analyses | 4 | |
Number of trials needed for next meta-analysis | SD of number of trials, among published meta-analyses | 6 | |
Number of trials needed to reach saturation of the literature | 90th percentile of the number of trials among published meta-analyses | 16 | On a previous systematic review, the median meta-analysis included 16 trials (IQR: 10, 28) [23] |
|