ASPECTS OF USING SURVIVAL ANALYSIS IN MED ICINE REPRESENTED BY LONG-TERM FOLLOW-UP OF 544-MULLER CURVED TOTAL HIP PROTHESES

The survivorship-analysis is the best method of evaluating long term f ollow-up studies in joint replacement by endoprostheses. The rate of s urvival (i.e. to have no failure of operation) at a time-interval (lif e table method) or a time-point (Kaplan-Meier method) is estimated, wh ereby homogenous subgroups should be respected. Regressionmodels (such as by Cox) select influencing parameters (covariates) and allow indiv idual prognoses. The methods respect censured data, but don't separate different kinds (death or lost to follow-up). As a high rate of losse s may be misleading, these cases have to be analysed by mathematical f unctions, graphical methods or the worst case method. These procedures are described and applied to our long term study (9-15 years) of 544 Muller curved stem total hip prostheses (completeness of data in 98 %) . 10 year survival rate was 98,2 % (revision or imminent revision for aseptic loosening or deep infection). Age at operation and size of ace tabular cup were selected as covariates but not sex, disease and activ e stabilization of acetabular floor.