Linear and Nonlinear Dose-Response Meta-Regression

Overview

MetaDose provides a suite of functions to perform linear and nonlinear dose-response meta-regression on study-level data. It supports both continuous (mdcont()) and binary (mdbin()) outcomes, with visualization and S3 methods for easy inspection of results.

The workflow is:

1. Model: Use mdcont() for continuous outcomes or mdbin() for binary outcomes to estimate linear or nonlinear dose-response relationships, including restricted cubic spline modeling.
2. Visualize: Use the returned dose object’s plot() method to generate publication-ready dose-response plots, and print() to inspect the model summaries.
3. Interact: For users who prefer a graphical interface, MetaDose provides an interactive Shiny application. The app allows uploading data, performing linear or nonlinear dose-response meta-regression, and visualizing results without writing R code.

The Shiny app is hosted online and can be accessed here: MetaDose Shiny App

This approach helps researchers understand the relationship between dose and outcome in a meta-analytic context, providing both numerical and graphical summaries.

Installation

Install the development version of MetaDose from GitHub with:

# install.packages("remotes")
remotes::install_github("asmpro7/MetaDose")

Example Usage

Continuous Outcome Example

# Perform linear and nonlinear dose-response meta-regression
cont_results <- mdcont(
  data = study_data,
  mean.e = mean_e,
  sd.e = sd_e,
  n.e = n_e,
  mean.c = mean_c,
  sd.c = sd_c,
  n.c = n_c,
  dose = dose,
  measure = "MD"
)
# Print both linear and nonlinear model summaries
print(cont_results, model = "both")
# Plot the dose-response curves
plot(cont_results, model = "both")

Binary Outcome Example

# Perform linear and nonlinear dose-response meta-regression
bin_results <- mdbin(
  data = study_data,
  event.e = event_e,
  n.e = n_e,
  event.c = event_c,
  n.c = n_c,
  dose = dose,
  measure = "RR"
)
# Print model summaries
print(bin_results, model = "both")
# Plot the dose-response curves
plot(bin_results, model = "both")