interactmeta

interactmeta implements contribution-adjusted meta-analysis (CAMS) for treatment effect modification from modifier-level aggregate data (Panaro, Röver & Friede 2025): Consistent Bayesian meta-analysis on subgroup specific effects and interactions.

cams() fits contribution-adjusted meta-analysis (CAMS) from modifier-level aggregate data. interactions(), modifiers(), and combined() report the main estimands; forestplot() and plot() and tidy() visualize and export a fitted model.

Start here	More
vignette("cams") — Getting started (estimands, assumptions, Bayesian & REML on steroids)	vignette("export-and-qc") — tables & sensitivity QC; vignette("worked-examples") — age, edss, and other datasets

Installation

# install.packages("remotes")
remotes::install_github("rvpanaro/interactmeta")

Documentation: https://rvpanaro.github.io/interactmeta/

CAMS supports Bayesian (default) and frequentist inference (method = "REML"). See the getting-started vignette for the full workflow.

Example

library(interactmeta)

st <- steroids[steroids$n > 0, ]
es <- escalc(
  measure = "OR",
  ai = treat.events,
  bi = treat.total - treat.events,
  ci = cont.events,
  di = cont.total - cont.events,
  data = st
)
both <- c("CAPE COVID", "COVID STEROID", "REMAP-CAP", "Steroids-SARI")
fit <- cams(es[es$study %in% both, ], reference = "NIV", method = "REML")
interactions(fit, exponentiate = TRUE)
modifiers(fit, exponentiate = TRUE)
forestplot(fit, exponentiate = TRUE)

Citation

@article{panaro2025cams,
  title   = {Consistent Bayesian meta-analysis on subgroup specific effects and interactions},
  author  = {Panaro, Renato and R{\"o}ver, Christian and Friede, Tim},
  journal = {arXiv preprint arXiv:2512.18785},
  year    = {2025},
  doi     = {10.48550/arXiv.2512.18785}
}