Introduction: In recurrent event a specific event occur repeatedly over time for a person. The frailty models take into account this correlation and provide efficient inferences. The frailties are assumed to be constant over time that it may be insufficient. Therefore time-varying frailty models are more realistic models. The aim of this study was to fit a time-dependent frailty model in the gap time between recurrent events.
Materials and methods: In this study, a time-dependent frailty model was introduced in the gap time between recurrent events, that was a generalization of the Wintrebert (2004) model in cluster data (center-effect). The parameters were estimated by Gaussian quadrature method. The model was applied to epilepsy data.
Results: The time-dependent frailty model fitted better in compare to shared frailty model. The observation time for IED on EEG in 56 patients (%73 male, %34 veteran status) with epilepsy was studied. Age and veteran status were the two risk factors in the gap time between IEDs. Variance of frailty was significant too.
Conclusion: The result of time-dependent frailty model was reliable when there were unknown time-dependent factors in medical data and make changes on times of occurring recurrent events. The Gaussian quadrature was an applied method to fit a time-dependent frailty model. The programming for this method was comfortable hence this method can cause time-dependent fraility models to be more practical in medical studies.
Hosseinzadeh S, Faghihzadeh S, Rahgozar M, Hajizadeh E, Hashemi Fesharaki S S, Gharakhani M. Time-dependent frailty model to gap times between recurrent events with application to epilepsy data. Koomesh 1395; 17 (3) :761-771 URL: http://koomeshjournal.semums.ac.ir/article-1-2917-en.html