USA Cox Multivariate and Logistic Regression Analyses

Cox multivariate and logistic regression analyses of the US 1996-2002 cases showed that factors have variable influence (relative risk or RR) on pancreas recipient and graft survival or rejection depending on the category (SPK, PAK, PTA).

Cox regressions were performed to assess variables affecting patient and graft survival; logistic regressions for variables affecting technical graft loss.

The variables included in all analyses were: duct management; donor and recipient age (older vs =45 years); donor cause of death (cerebro/cardiovascular cause of death vs traumatic causes); number of HLA mismatches at each locus (A, B, DR); recent panel reactive antibody (PRA) over vs =20%; increasing preservation time; recipient body mass index (BMI) above vs =25 kg/m2; use of depleting vs. non-depleting antibody for induction immunosuppression; the use of Tac and MMF for initial maintenance immunosuppression, and center size as previously defined. In the following Tables, relative risks (RR) are given for variables with a P value of =0.20; others were assumed to be non-significant. For patient survival models, statuses of the transplanted organs were included as time-dependent variables, since the status may change over the period analyzed. For pancreas graft status, distinctions were made between technical and rejection failures.