interactionRCS

A tool to calculate and plot Hazard Ratios, Odds Ratios or linear estimates in a simple or restricted cubic splined interaction model

Version 1.1 (February 25, 2022)

Description

interactionRCS facilitates interpretation and presentation of results from a regression model (linear, logistic, Cox) where an interaction between the main predictor of interest X (binary or continuous) and another continuous covariate Z has been specified. In particular, interactionRCS allows for basic interaction assessment (i.e. log-linear/linear interaction models where a product term between the two predictors is included) as well as settings where the second covariate is flexibly modeled with restricted cubic splines. Confidence intervals for the predicted effect measures (beta, OR, HR) can be calculated with either bootstrap or the delta method. Lastly, interactionRCS produces a plot of the effect measure over levels of the other covariate.

Installation

To install the latest version of interactionRCS, type the following lines in a web-aware R environment.

if(!"devtools" %in% rownames(installed.packages())){
  install.packages("devtools")
}
devtools::install_github("https://github.com/gmelloni/interactionRCS.git")
# or alternative devtools::install_git("https://github.com/gmelloni/interactionRCS.git")
library(interactionRCS)

Usage

After estimating a regression model (linear, logistic, Cox) such as model<-glm(y~ ...) estimate and plot interactions with:

int<-estINT(model=model, ...)

plotINT(int, ...)

For a detailed introduction to interactionRCS and code examples please refer to this vignette

Authors

Giorgio Melloni, Hong Xiong, Andrea Bellavia

TIMI study group, Department of Cardiovascular Medicine, Brigham and Womens Hospital / Harvard Medical School