Predictive regression models in a single-center series of double kidney transplantation.

Background: Double kidney transplantation (DKT) may be a useful approach to reduce the number of discarded marginal kidneys. In this study, we retrospectively evaluated the potential predictors for patient and graft survival in a single-center series of 59 DKT procedures performed in the period 1999-2008. Patients and Methods: The kidney recipients (mean age 63.27±5.17 years) included 16 women (27%) and 43 men (73%). The donors (mean age 69.54±7.48 years) included 32 women (54%) and 27 men (46%). The mean post-transplant dialysis time was 2.37±3.61 days. The mean hospitalization was 20.12±13.65 days. Serum creatinine (SCr) at discharge was 1.5±0.59 mg/dL. The proportional hazards assumption for each Cox regression model with P <0.05 was tested by using correlation coefficients between transformed survival times and scaled Schoenfeld residuals. Results: In Cox models for patient survival, the variables that reached statistical significance were donor SCr (P=0.007), donor creatinine cleararance (P=0.023), and recipient age (P=0.047). By entering these variables into a multivariate model for patient survival, no further significance was observed. In the univariate Cox models performed for graft survival, statistical significance was noted for donor SCr (P=0.027), SCr 3 months post-DKT (P=0.043), and SCr 6 months post-DKT (P=0.017). A final multivariate model retained SCr at 6 months (P=0.042) and donor SCr (P=0.090). Conclusions: SCr at 6 months seemed to emerge from both univariate and multivariate Cox models as a potential predictor of graft survival in DKT. Multicenter studies with larger double kidney recipient populations should be performed to confirm this finding.