User guide

Create a project

One project represents one review. Invite collaborators and set participation for each stage.

• Creator: manages the project, including deletion.
• Lead: manages settings, assignments and reconciliation.
• Collaborator: completes assigned work.

An extraction supervisor can inspect and reconcile extraction sheets. Select Import demo project to explore a private example.

Import records

Search PubMed or upload RIS, NBIB, CSV, XLSX, PubMed XML or EndNote files. Export Zotero or Mendeley records as RIS or CSV.

Imports retain source counts. Review suspected duplicates in Duplicates before confirming them.

Screen titles and abstracts

Set the reviewer threshold and assignment mode.

• Smart: distributes work by outstanding workload.
• All to all: assigns each record to all participants.
• Manual: lets leads assign records.

Record inclusion or exclusion decisions. Leads enable the Conflicts screen under Review visibility. It is hidden by default, including for leads.

Optional ranking uses previous decisions by each reviewer to order remaining records. It does not make screening decisions.

Assess full-text reports

Included records move to full-text assessment. The application attempts open-access retrieval when a DOI is available. Upload other reports individually or in batches.

Assess eligibility beside the PDF. Record a reason for each exclusion.

Extract and reconcile data

Define fields and sheets in Manage fields. Use multiple rows for outcomes, arms or time points. Match Label aligns rows across reviewers.

Complete extractions independently. A lead or supervisor resolves discrepancies in Comparison.

Prepare the dataset

Reconciled data transfer to the analysis dataset. Dataset edits preserve the original extraction entries.

For missing studies, check unresolved discrepancies and pending confirmation of AI-extracted numbers.

Calculate and analyse effect sizes

Select a measure and map its required inputs to dataset columns. Direct effect-size inputs require a standard error, not a variance.

metaDETECT flags potential data inconsistencies. Verify these against the source report.

R packages metafor and clubSandwich provide fixed-effect, random-effects and multivariate analyses. Account for within-study dependence when studies contribute multiple effect sizes.

Outputs depend on the model and include estimates, intervals, heterogeneity, forest plots and diagnostics.

Report and export

PRISMA reporting distinguishes records, reports and studies. Review counts and enter source totals that cannot be derived automatically.

Download the flow diagram and study lists in Word. Review the draft selection-methods text before using it. Export review data as CSV or Excel where available.