Comparison of two groups of survival data using nonparametric predictive inference

The first purpose of this article is to analyze the data of patients who underwent the colectomy at the University Hospital of Ostrava in order to find the surgical technique that guarantees longer overall survival time. Nowadays there are used two basic surgical techniques for colectomy, either traditional (open) or minimally invasive (laparoscopic) and the patients are assigned to either of the surgical groups randomly. The second purpose is to present the use of nonparametric predictive inferential (NPI) method for modelling survival functions and comparison of groups of survival data containing censored data in the form of observed event times and right-censored data. The surgical techniques were compared separately for patients with resection of colon and for patients with resection of rectum. Results of the decision making process obtained by the NPI approach were confronted with the results obtained by the standard nonparametric approach represented by the log-rank test. Both approaches led to the same conclusion that the minimally invasive technique guarantees significantly longer survival time in comparison with the classical technique for patients who underwent resection of colon and that there is no statistically significant difference between survival times of patients operated by different surgical techniques who underwent resection of rectum.