KIDNEY - ACUTE GRAFT INJURY

N.B. Cross¹, P. Macaskill², A. Hayen¹, A. Webster³, G. Wong⁴, P.J. O'connell⁵, J.R. Chapman⁶, B. Nankivell⁷, J.C. Craig^4
1School Of Public Health, University of Sydney, Sydney/AUSTRALIA, ², University of Sydney, Sydney/AUSTRALIA, ³School Of Public Health, University of Sydney, Sydney/NSW/AUSTRALIA, ⁴Centre For Kidney Research, The Children's Hospital at Westmead, Westmead/AUSTRALIA, ⁵Centre Of Transplant And Renal Research, Westmead Hospital, Westmead/NSW/AUSTRALIA, ⁶Renal Unit, Westmead Hospital, Westmead/AUSTRALIA, ⁷, Westmead Hospital, Sydney/AUSTRALIA

Body: Introduction
After kidney transplantation, serum creatinine is monitored routinely. Rising serum creatinine, relative to expected values, is used an indicator of complications, but the optimal method for determining expected values in the first post-transplant month is unknown.

Methods
Using retrospective data, we compared five methods for defining individual patients’ expected creatinine values - daily group mean (mean creatinine by postoperative day in patients without complications), nadir (lowest prior creatinine), delta (previous day’s creatinine) and two methods using a multivariate prediction model which adjusts daily expected values for patient, donor and transplant characteristics known at time of transplant (day zero predictions) and a more complex model that also included individuals’ prior observed creatinine values (daily predictions). For each method, we estimated area under the receiver operating characteristic curve (AUC) for percentage rise in creatinine relative to expected creatinine for detection of biopsy confirmed, clinically diagnosed complications in the first posttransplant month. We also assessed when complications would have been detected for the day zero predictions and daily predictions methods, relative to when complications were detected clinically.

Results
Among 157 included patients, 56 had complications in the first month after transplantation. Group mean (AUC 0.86 95%CI 0.82-0.91), nadir (AUC 0.68, 95%CI 0.61-0.75) and delta methods (AUC 0.79, 95%CI 0.73-0.85) did not perform as well as day zero (AUC 0.90, 95%CI 0.87-0.94) or daily predictions (AUC 0.94, 95%CI 0.93-0.96). The prediction methods detected complications at least one day earlier than was achieved clinically in 80% of patients.

Conclusions
Rises in creatinine relative to predicted creatinine, based on a mixed model, provided better test performance than other methods studied.

Disclosure: All authors have declared no conflicts of interest.