A Comparison of Kaplan-Meier Product Limit Estimation and Markov Process in Estimating Survival Time of A Cohort Of Dogs with Rabies

Abstract

In this research two methods were used to compute survival time of a cohort of 84 dogs with rabies. In carrying out Kaplan-Meier analysis, euthanasia was equated with death. Death (including subjects) recorded were treated as occurring prior to the exact time they occurred while censored (subjects lost to follow-up) were treated as occurring later than they time they actually occurred. Survival probability estimates and variance were calculated. In Markov process method, a 5 state time homogeneous Markov chain was used, the fundamental matrix which was obtained  by summing counts recorded based on the number of dogs making respective transition was used to obtain probability matrix. The N matrix was calculated as well as its variance. The variances of the two methods computed served as a basis for comparing the efficiency of the two methods. Hence, the Kaplan-Meier product limit estimation with the smaller variance is more efficient in estimating survival time

Keywords: Transition probability; Efficiency; Censored; Euthanasia; Fundamental matrix