An Application of Extended Cox Proportional Hazard Approach for Survival Analysis for Recurrent Events

• Activity Number: 527
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Abdul Salam*+ and Abdul K. Hussein and Peter Senior and James Shapiro
• Companies: NGHA and University of Windsor and University of Alberta and University of Alberta
• Keywords: Type I Diabetes ; Islet Transplant ; Recurrent event ; Extended Cox-proportional model ; Log rank test ; Kaplan Meier

Abstract

Two or more islet transplants are often required to achieve insulin independence in patients with type 1 diabetes. Conventional outcome measures in clinical islet transplantation (CIT) are based on graft survival either from time to first event (TTFE), or time to graft failure post last transplant (TTGFPLT). Consequently imprecise estimates of graft survival result from this failure to consider multiple transplants. The objective is to compare extended Cox-based model with classical method for graft survival. Type I diabetes patients who received multiple islet transplants between March 1999 and December 2006 were used to examine the relationship between pre transplant panel reactive antibodies (PRA) on graft survival. Pre transplant PRA of >15% was highly associated with graft failure. Hazard ratio (95% CI) were obtained from Cox-based models for recurrent event AG: 4.3 (1.8 -10.4), PWP-TT: 4.7 (2 -11), PWP-GT: 5(2.3 -12), LWA: 4.3 (1.9 -10) and using classical model TTFE: 3.92 (0.86 -17.9), TTGFPLT: 4.9 (1.7 -13.8). Extended Cox-based models are potentially important in CIT because estimates of hazard ratio are consistent and precise compared with classical methods.