A class of inference procedures for validating the generalized Koziol–Green model with recurrent events

The problem of validity of a model on the informativeness of the right-censoring random variable on the inter-event time with recurrent events is considered. The generalized Koziol–Green model for recurrent events has been used in the literature to account for informativeness in the estimation of the gap time distribution or the cumulative hazard rate function. No formal procedure for validating such assumption has been developed for a recurrent failure time data. In this manuscript, we propose procedures for assessing the validity of the assumed model with recurrent events. Our tests are based on the scaled difference of two competing estimators of the cumulative hazard rate possessing nice asymptotic properties. Large sample properties of the proposed procedures are presented. The asymptotic results are applied for the construction of χ2 and Kolmogorov–Smirnov type tests. Results of a simulation study on Type-I error probabilities and powers are presented. The procedures are also applied to real recurrent event data.