Meta-Analysis in the Screening Tool

What the synthesis module does

Click Meta-analysis in the sidebar (Pro, on a cloud-saved review) to pool your included studies without leaving the tool. You define named outcomes, enter per-study data, and get pooled effects, heterogeneity statistics, subgroup analysis, forest and funnel plots, and copy-ready methods text. If you already ran data extraction, study rows can prefill from the extraction grid.

Effect measures and models

For binary outcomes, enter events and totals per arm and choose risk ratio, odds ratio, or risk difference. For continuous outcomes, enter mean, standard deviation, and sample size per arm and choose mean difference or standardized mean difference; the standardized mean difference is computed as Hedges g with the small-sample bias correction.

Two models are available: Random effects (DerSimonian and Laird), the usual default when studies differ in populations and settings, and Fixed effect, which uses Mantel-Haenszel weighting for binary outcomes and inverse variance weighting for continuous outcomes.

Heterogeneity, subgroups, and small-study effects

Every analysis reports Cochran's Q with its degrees of freedom and p value, I squared (the percentage of variability due to heterogeneity rather than chance), and tau squared (the between-study variance). Add a subgroup label to any study row to unlock subgroup analysis: each subgroup is pooled separately and a between-subgroups Q test tells you whether the subgroup effects genuinely differ.

The funnel plot displays effect against standard error for a visual asymmetry check, and Egger's regression test quantifies it, with the usual caution that the test is underpowered below about ten studies.

Outputs

1. Create an outcome, pick its effect measure, and enter or prefill the study data. Leave a row blank to exclude it from the pool.
2. Choose the model and read the pooled estimate with its confidence interval and heterogeneity line.
3. Download the forest plot and funnel plot with their SVG and PNG buttons.
4. Click Export CSV for the per-study effects, weights, and pooled results.
5. Copy the generated methods text: a ready statistical methods paragraph naming the measure, model, heterogeneity statistics, and software, written to paste into your manuscript.