Provides customized forest plots for network meta-analysis incorporating direct, indirect, and NMA effects. Includes visualizations of evidence contributions through proportion bars based on the hat matrix and evidence flow decomposition.
NMAforest is an R package for generating detailed forest plots in network meta-analysis (NMA). It visualizes direct, indirect, and network meta-analysis treatment effects, along with study- and path-level contribution proportions. The visualization is based on the evidence flow decomposition method by Papakonstantinou et al. (2018).
This package relies on key infrastructure from the netmeta, igraph, and ggplot2 R packages.
It also adapts methods and code presented by Papakonstantinou et al. (2018) for evidence flow decomposition in network meta-analysis, including the comparisonStreams() function from the flow_contribution GitHub repository.
The stable release of NMAforest can be installed from CRAN:
install.packages("NMAPropForest")
The following columns are required (either with these names or specified via function arguments):
| Column | Required For | Description | Type |
|---|---|---|---|
treat |
All analyses | Treatment label for each arm | character or factor |
event |
Binary outcomes | Number of events in the arm | numeric |
n |
All analyses | Sample size in each arm | numeric |
mean, sd |
Continuous outcomes | Mean and standard deviation | numeric |
study |
All analyses | Study label or grouping variable | character or numeric |
study_id |
Optional | Unique numeric study identifier (auto-generated if missing) | integer |
Note: We recommend that users include an explicit study_id column where each value uniquely corresponds to a study label in the study column.
If the study_id column is not present in the dataset, the function will automatically generate one and return the updated data frame with this column added.