A new procedure for the statistical evaluation of intrauterine contraception

Abstract

An improved statistical method for assessing the efficacy of intrauterine contraception is presented. This method, called the log-rank method, is based on a daily life table, which yields the best estimates of the cumulative net termination probabilities. It provides a sensitive chi-square test statistic to detect differences among these probabilities for two or more contraceptive devices. When three or more devices are compared, the chi-square statistic may be partitioned into additive components representing comparisons among groups of devices. The log-rank method also permits adjusting for differences among subgroups by controlling for factors such as parity or age. Although calculations may be performed on a desk calculator, a FORTRAN-IV computer program is made available to interested researchers.

PIP: A new procedure for the statistical evaluation of IUDs, the log-rank method, is presented. The log-rank method is based on a daily life table. Comparisons of 2 or more devices are made by examining cumulative net probabilities of termination for any reason. The data required for analysis include patient identification, IUD type, date of insertion, date the patient was last seen or date of termination, and the reason for termination. Steps for calculating the cumulative net termination probabilities are presented. The method provides a sensitive chi-square test to detect differences among net terminiation probabilities. When there are more than 3 devices to compare, the chi-square statistic may be partitioned into additive components representing comparisons among groups of devices. It is also possible to adjust the method for such subgroup differences as parity and age. 2 applications of this method, performed with the FORTRAN-4 computer are presented. 3 IUDs are compared: TCu-300, the Dalkon Shield, and SCS-D. Results from the original study correlated with results using the log-rank method. In the 2nd application the log-rank analysis of 589 women using either the Lippes Loop or the Dalkon Shield is given. This statistical method is also useful in comparing other methods of contraception.