Seminar at SMU Delhi
April 30, 2014 (Wednesday) ,
3:30 PM at Webinar
Carmel College, Thrissur, Kerala
On testing problems in current status competing risks data
Abstract of Talk
In survival or reliability studies, current status censoring is common where the exact life time of patients (objects) is unobservable, but one can only observe a monitoring time and whether the event of interest has happened or not before the monitoring time. In addition, when patients (objects) are exposed to the risk of failure due to two or more causes, we also observe the cause of failure for relapsed patients. Competing risks data with current status censoring arise frequently from cross sectional studies in demography, epidemiology and reliability experiments. In the present work, we develop a non parametric test procedure for comparing cumulative incidence functions of current status competing risks data. We also extend this procedure to test whether time to failure and cause of failure are independent in a current status competing risks set up. Asymptotic distribution of the test statistic is derived in both cases. Simulation studies are conducted to assess the finite sample behavior of the test statistics. The practical utility of the procedure is well demonstrated using a real life data set on menopausal history of 2423 women given in Jewell et al., 2003.
Key Words: Chernoff’s distribution, Chi-square test, Competing risks data, Current Status censoring.