On the methods calculating mean survival time in 51Cr-platelet survival study

The International Committee for Standardization in Hematology recommended three methods for calculating objectively the mean platelet survival time, ie, weighted mean of linear estimate and logarithmic estimate, truncated exponential model, and gamma model. For determining which method is better, data of the survival study were fitted to each of three methods, as well as the Paulus polynomial and exponential polynomial models, and the mean survival time and the mean square error were compared. Mean survival obtained by each method was always longest in healthy controls and shortest in ITP. In 3rd degree of polynomial estimate and 2nd and 3rd degrees of logarithmic estimate, several cases showed practically outlying mean survival time of more than 12 days. In healthy controls, highest mean square error was observed in 1st degree logarithmic estimate, and in cases with short mean survival time the highest one was obtained in 1st degree polynomial estimate. Mean square error in ITP pre- and postsplenectomy and in aplastic anemia was smallest in the gamma model. Among nine methods recommended by the International Committee for Standardization in Hematology and by Dr. Paulus, the gamma model so far provides the best method for calculating the mean survival time of 51Cr-platelets.