[Comparison of the effectiveness of various types of therapy]

The paper deals with the analysis of one of the most characteristic problems in medical statistics,--comparison of the efficiency of various therapies (including those with different drugs) from clinical evidence; it gives a detailed account of the applied issues--identification of homogeneous groups and obtaining the estimates of distributed characteristics) of the required accuracy. Standardly, the best agent is identified, by comparing the mean values of the distributed characteristics. Such an approach cannot be considered satisfactory for some reasons. To arrive at a conclusion that it is advisable to introduce a new drug or a new treatment, the appropriate distributions should be more fully compared. A complete analysis of all these aspects has been made on a common basis and a general scheme of their practical decision has been identified.