Start with the data
Analysis quality depends on extraction quality. Confirm outcome definitions, arm labels, effect measures, units, timepoints, and missingness before running statistical workflows.
Supported analysis areas
- Pairwise continuous and binary meta-analysis.
- Generic, correlation, proportion, rate, survival hazard-ratio, and diagnostic-test-accuracy analyses where data fit the required shape.
- Saved analysis runs for reproducibility.
- Forest plots, funnel plots, diagnostic plots, leave-one-out, cumulative analysis, and meta-regression where supported.
- Frequentist network meta-analysis for eligible premium projects.
- GRADE table generation and human-reviewable GRADE profiles where enabled.
Run an analysis
- Open Analyze and confirm the project analysis configuration.
- Select the outcome and model family that match your extracted data.
- Review continuity corrections, direction settings, sorting, model selection, and plot options.
- Run the analysis and inspect warnings, heterogeneity, figures, and generated text.
- Save the run when the settings/results should be reused or exported.
NMA expectations
NMA is available to eligible premium projects configured for network analysis. Open Analyze to confirm that the workflow is available for the active project.
Practical tips
- Check that treatment nodes are consistently named.
- Inspect network geometry before interpreting rankings.
- Treat rankings as one part of the evidence, not a substitute for certainty assessment.
GRADE expectations
GRADE outputs should be reviewed by a qualified human. Coventra can help assemble tables and suggestions from saved analysis runs, but certainty judgments require methodological review.