Life-Table Analysis of IUDS: Problems and Recommendations

Publisher Website
Google Scholar
Abstract
Two major prospective studies of copper-bearing IUDs showed substantial changes in termination rates when calculated at different dates. One of the studies employed the Tietze life-table method and showed progressive increases in termination rates with the passage of time. The other study used the Potter life-table method and showed sharply decreased termination rates at later assessments. In investigating the reasons for these changes, it was noted that the data collected for periodic analyses during the studies violated the assumptions underlying the life-table technique. It was discovered that as a result of these violations the Potter and the Tietze life-table calculations based on the same set of data produced markedly different estimates of IUD termination rates during the course of a study. Underlying these differences, and the reason for the apparent changes in termination rates, was a large set of incomplete observations. Neither of the two life-table methods was able to deal adequately with the biases arising from these incomplete observations. The "anniversary method," devised to overcome the perceived problems of the Potter and Tietze methods, also proved inadequate to deal with the incomplete observations. Only vigorous and active follow-up, together with ample time to complete data collection, editing, coding, and key punching, is likely to reduce the proportion of women with incomplete observations and is likely to minimize the biases attendant upon incomplete or partial observations of acceptors in prospective clinical studies of IUDs.

This publication has 0 references indexed in Scilit: