Online Program

Saturday, February 20
PS3 Poster Session 3 & Continental Breakfast sponsored by Capital One Sat, Feb 20, 8:00 AM - 9:15 AM
Ballroom Foyer

Software for Survival and Multistate Analysis in R (303262)

*Adam King, Cal Poly Pomona 

Keywords: R, software, suvival, multistate

Time-to-event outcomes are common in many different fields of study; medical researchers may study the timing of disease remission and relapse events, while labor economists study the timing of the unemployed finding jobs. Oftentimes, such modeling problems have complex features, such as clustering of subjects, recurrent events, competing risks, state transition events, and nonlinear relationships between predictors and event risks. However, the most commonly available software procedures and packages for analyzing survival outcomes in SAS, Stata, and R only typically allow a subset of these features to be incorporated into a model, due to limitations of likelihood inference procedures. We present an R software package for fitting survival and multistate models that allows any combination of the foregoing advanced features. Inferences are obtained using a Bayesian Markov Chain Monte Carlo (MCMC) technique, but the inference algorithms are automatic, requiring no special knowledge of Bayesian computing from the user. The software is illustrated with a public health application, in which we model recurrent episodes of drug use and incarceration in a population of illicit drug users.