netmeta R package
The netmeta R package helps researchers compare multiple treatments for the same condition by combining evidence from different studies, even when those treatments haven't been directly compared. It uses standard statistical methods and works within the R programming environment.
At a glance
Use when
Comparing multiple interventions using existing trial data, especially when direct comparisons are sparse; when frequentist methods are required or preferred; within evidence synthesis for HTA submissions.
Avoid when
Individual patient data is required; when complex hierarchical or patient-level modeling is needed; when Bayesian approaches with prior specification are desired.
Inputs
Aggregate outcome data from comparative studies (e.g., odds ratios, mean differences) in a contrast-based format; study design information; treatment labels; variance-covariance structure.
Outputs
Treatment effect estimates (with confidence intervals), inconsistency statistics, rank probabilities, forest plots, network geometry plots, and measures of heterogeneity.
How it works
netmeta is an R package that implements frequentist methods for network meta-analysis, enabling the synthesis of direct and indirect evidence across a network of interventions. It supports various outcome types and data formats, provides functions for model fitting, inconsistency assessment, and visualization (e.g., network plots, forest plots), and includes tools for ranking treatments and conducting sensitivity analyses.
- HTA domains
- Clinical Effectiveness
- Categories
- Evidence SynthesisIndirect Comparisons
- Assumptions
- The analysis assumes transitivity (comparability across treatment contrasts) and consistency (agreement between direct and indirect evidence); models are based on frequentist statistical theory.
- Strengths
- Open-source and freely available in R; supports multiple outcome types; provides comprehensive diagnostics and visualizations; well-documented with worked examples; uses transparent frequentist methods preferred in some regulatory settings.
- Limitations
- Requires statistical expertise in R and meta-analysis; limited to frequentist approaches (no Bayesian modeling); may be less flexible than Bayesian software for complex models.
- Also known as
- netmeta R package, netmeta package, R package netmeta
Questions this answers
- › Which treatment is most effective among multiple options for a given condition?
- › How do interventions compare when no head-to-head trials exist?
- › Is there inconsistency between direct and indirect evidence in the network?
- › What is the relative ranking of treatments based on efficacy or safety?
- › How robust are the results to inclusion or exclusion of specific studies?
- › What is the magnitude of treatment effects with confidence intervals?
References & sources
Similar by meaning
Beta record. Generated from the primary source via AI extraction and independent audit, pending final human review.