More than two treatments, one analysis
Pairwise meta-analysis answers one question: A versus B. When your field has three, five, or ten competing treatments, you need a network meta-analysis to compare them all at once, including pairs no trial ever compared directly. That analysis has mostly lived in R packages and Bayesian software. Covexe runs it in the browser.
What you get
- A network diagram of your evidence, drawn from your comparisons
- A league table with every treatment-versus-treatment estimate
- P-score rankings, the frequentist analog of SUCRA
- Inconsistency assessment, comparing direct against indirect evidence
- Bucher indirect comparisons when a full network model is not possible
How it works
- Paste your pairwise comparisons (treatment names that differ only in case or spacing are merged automatically, so "Placebo" and "placebo" stay one node).
- Check the network diagram: you see immediately whether your network is connected.
- Read the league table and P-scores, and export the results.
Validated against netmeta
The engine is checked against R's netmeta package, the standard for frequentist network meta-analysis. Every method shows its formula and citation, and R replication code is available so anyone can reproduce your numbers.
The review around the network
Screening, data extraction, risk of bias, GRADE, and the PRISMA 2020 flow diagram live in the same free project, so the network meta-analysis is a step in your workflow rather than a separate tool with separate files.