Use of GEE for modeling censored correlated data: application to the studyof risk factors for withdrawal of totally implantable vascular access devices in cystic fibrosis

Background: The proportional hazards model proposed by Cox for modeling cen sored data is not suited for correlated delays, for instance when several e vents can be observed on each subject. Methods: To analyze correlated delays, we propose to use a log-linear margi nal model equivalent to Cox model, Correlations are taken into account thro ugh the use of Liang and Zeger's Generalized Estimating Equations (GEE) and of their robust variance estimator An advantage of this method is that it can be implemented through the SAS(R) GENMOD procedure. When ties are obser ved, we propose to use multiple imputations, creating M data sets without t ies from the original one. Results: This method is applied to a retrospective survey opt the risk of w ithdrawing totally implantable vascular access devices (TIVAD) because of c omplication in cystic fibrosis patients: 265 TIVAD implanted ill 200 patien ts were observed. Risk factors were characteristics of the device or of the patient. Results obtained,with the robust variance estimator and ten imput ations show that the use of the device for taking blood (vs exclusive perfu sion of antibiotics), polyurethane catheter (vs. silicon), use of counterpr essure for upkeeping and pulmonary colonization by Pseudomonas Aeruginosa a re significantly associated to withdrawal. Under the Cox model which does n ot account for the correlations, some conclusions differ because the robust variance of the estimators is smaller than the variance obtained under the working assumption of independent delays. Conclusion: This approach allows the modeling of correlated survival data w ith SAS(R) software. Our results illustrate the necessity of accounting for existing correlations.