Survival Analysis of Un-identical Recurrence Using Conditional I, Conditional II, and Marginal Method

9th Annual Basic Science International Conference 2019

自然科学（总论）

Kusthika Putri, Dita Ramadayanti^1 ; Sa'adah, Umu^1 ; Efendi, Achmad^1

Department of Statistics, Faculty of Sciences, Universitas Brawijaya, Malang, Indonesia^1

关键词: Backward elimination;    Explanatory variables;    Hypercholesterolemia;    Recurrence rates;    Recurrent events;    Secondary datum;    Stroke patients;    Survival analysis;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/546/5/052058/pdf
DOI  :  10.1088/1757-899X/546/5/052058

学科分类：自然科学（综合）

来源: IOP

【 摘 要 】

Survival analysis can be used to examine events that may occur more than once in one individual or is called a recurring event. This event can be seen in cases of stroke recurrence. The occurred recurrence shows a more severe recurrence stage than before (recurring events are not identical). The Stratified Cox model with Conditional I, Conditional II, and Marginal methods can be used to analyse recurring events, especially in recurrent events that are not identical. The used data are secondary data from the medical record archives of stroke patients at Chasbullah Abdulmadjid Hospital Bekasi City from March 1, 2016 to April 30, 2016. Based on the Wald test and the backward elimination steps, we found two explanatory variables that influence the recurrence rate of stroke which were hypertension and hypercholesterolemia. The model from the three methods is a model with a combination of variables of hypertension and hypercholesterolemia. Patients suffering from hypertension or hypercholesterolemia have the potential to experience a recurrence of stroke compared to patients who do not have hypertension or hypercholesterolemia. Based on the AIC value of each method, the best method for recurring data on stroke patients is the Conditional I method.