Multi-State Survival Analysis with Simultaneous Events Followed by a Consecutive Event

Non-fatal cardiovascular diseases including myocardial infarction, stroke, angina, congestive heart failure, or coronary heart disease may occur simultaneously and recurrently. Survival analysis is widely used for modeling the primary event of interest. To model the concurrence and recurrence of events while accounting for missing data that is often present in this type of data, we propose a multi-state model dealing with both concurrence and recurrence of events. This consists of taking into account nine possible combinations of events, for example, where a single or simultaneous events are followed by another event. In addition, we propose an imputation algorithm for missing covariates in survival data. The imputation approach uses the closest predictor imputation method in the first step, then implements imputation from a derived multivariate distribution for multiple missing covariates. This method is applied to data for non-fatal cardiovascular diseases.