Online Program

Simulating Real-World Data with Time-Varying Variables (303643)

 View Presentation

Keywords: simulation, correlated covariates, right-censored outcomes (or time-to-event outcomes), time-varying covariates, longitudinal studies

Simulation studies are useful for evaluating and developing statistical methods for the analyses of complex problems present in real scenarios. Generating sufficiently realistic data for this purpose, however, can be challenging. Our study of the comparative effectiveness (CE) of HIV protocols on the risk of cardiovascular disease – involving the longitudinal assessment of HIV patients – is such an example. The correlation structure across covariates and within subjects over time must be considered. A challenge in simulating the covariates is to incorporate a joint distribution for variables of mixed type – continuous or categorical. An additional challenge is incorporating within-subject correlation where some variables may vary over time and others may remain static. We have developed an R package for this purpose that extends the work of Demirtas and Doganay (2012) and which is used in the background of a website that allows the user to compute power for a CE study. In this poster, we describe our cohesive and user-friendly approach to simulate CE studies with right-censored outcomes that are functions of time-varying covariates and exemplify its use with a power calculation.