The tools in this package are intended to help researchers assess multiple treatment-covariate interactions with data from a parallel-group randomized controlled clinical trial. The methods implemented in the package were proposed in Kovalchik, Varadhan and Weiss (2013)
README file for R project ANOINT ANOINT is a collection of R tools that were created to help trialists evaluate effect modification with data from a single clinical trial, when it is hypothesized that multiple factors could influence responsiveness to treatment. The package performs a number of testing procedures for the presence of any effect modification among a set of candidate treatment response factors. The methods are applicable to glm or Cox proportional hazards models. The tools allow for a lasso selection procedure for prognostic factors, prior to the assessment of effect modification, which is conducted on the control data only. One of the models in the ANOINT methods is a proportional interactions model. This is a generalized version of Tukey's 1-df model which was first described by Follmann and Proschan. It is a special case of multiple interaction which supposes that the effect modification is proportional (a constant multiple) of prognostic effects. This model has been under-utilized in assessments of treatment-covariate interactions in a clinical trial. Plotting methods include a standard forest plot for each of the candidate treatment response factors and a diagnostic plot for proportional interaction. For more details, see the paper: Kovalchik SA, Varadhan R, Weiss CO. Assessing heterogeneity of treatment effect in a clinical trial with the proportional interactions model. Stat Med. 2013 Dec 10;32(28):4906-23. doi: 10.1002/sim.5881. Epub 2013 Jun 21. PMID: 23788362.