Transplant Trial Watch

Predictive Score for Posttransplantation Outcomes.

Molnar MZ, Nguyen DV, et al.

Transplantation 2016; 07: 07.


Aims
To develop a novel score to predict posttransplant outcomes in kidney transplantation based on variables which are available before, or at the time of transplantation.

Interventions
Data of all kidney transplant recipients listed in the Scientific Registry of Transplant Recipients was linked to a list of individuals with end-stage renal disease who underwent maintenance hemodialysis treatment from July 2001 to June 2006. Prediction scores were developed using Cox models for mortality, allograft loss and combined death or transplant failure, and performance was tested in the validation data set.

Participants
15,125 hemodialysis patients who underwent first deceased transplantation, randomly divided into a two thirds set for model development (N = 10 083) and a one third set for validation (N = 5042).

Outcomes
Primary measured outcomes included mortality, allograft loss and combined death or transplant failure. Other measured outcomes included recipients' age, cause and length of end-stage renal disease, hemoglobin, albumin, selected comorbidities, race and type of insurance as well as donor age, diabetes status, extended criterion donor kidney, and number of HLA mismatches.

Follow-up
Median follow-up time was 794 days

CET Conclusions
This study uses a large US dataset to develop predictive models for patient and graft survival from recipient and donor information available prior to transplant. The model developed is tested in a random subset of the dataset. The resulting model incorporates 10 predictor variables, and demonstrates good concordance for mortality (C-statistic 0.7) and moderate concordance for graft survival (C-statistic 0.63). Whilst this predictive model may have a role to play in helping clinician’s decision making when considering an organ offer, there are some limitations. Predictive accuracy, as with all models of this type, is relatively limited and so decisions should not be taken using the model in isolation. The model has not been validated in an external cohort, either in the US or outside, and so may not be representative of risk in other populations with different characteristics.

Quality notes
Quality assessment not appropriate

Trial registration
None

Funding source
Non-industry funded