ANALYSIS OF THE RENAL-TRANSPLANT WAITING LIST - APPLICATION OF A PARAMETRIC COMPETING RISK METHOD

Background. The strong competition for scarce renal. graft resources j eopardizes an individual patient's chances of a transplantation within a reasonable time scale. This study was undertaken to quantify these chances of receiving a transplant. Methods. All patients registered fo r their first renal allograft between January 1980 and December 1993 ( n=40,636) in Eurotransplant(4) were selected, The influence of patient characteristics, such as age, HLA phenotype frequency, % panel-reacti ve antibodies, period of registration, and AB0 blood group, on the wai ting list outflow was studied. The competing risk method was applied a nd Poisson models were built to estimate the risk factor effects. Resu lts. The chance of transplantation within 10 years after registration was overestimated by Kaplan-Meier (84%); using the competing risk meth od it was only 74%. The predicted chance for death on the waiting List was overestimated by 33% (45% Kaplan-Meier vs. 12% competing risk). A time-varying covariate effect on the chances of waiting list outflow was observed, Favorable factors for quick transplantation, such as blo od group AB or a common HLA phenotype, were no longer seen to be drivi ng forces for transplantation once 5 to 6 years of waiting time had be en accrued. Conclusion. When multiple outcomes exist, Kaplan-Meier est imates should not be interpreted as survival rates, while competing ri sk estimates yield appropriate chances. A significantly decaying effec t of the usual allocation parameters is observed with ongoing waiting time. This phenomenon is the statistical basis for redesigning allocat ion strategies, Organ exchange algorithms should have the potential to adapt to these time-varying effects.