Predictive ability of pretransplant comorbidities to predict long-term graft loss and death

Area of Science:

• Nephrology
• Transplantation
• Health Services Research

Background:

• Kidney allocation schemes are complex, with ongoing debate about incorporating additional pretransplant comorbidities beyond diabetes.
• Accurate prediction of graft and patient survival is crucial for equitable organ distribution.

Purpose of the Study:

• To investigate the predictive ability of multiple pretransplant comorbidities for kidney transplant graft and patient survival.
• To compare the predictive performance of different comorbidity data sources within kidney allocation models.

Main Methods:

• Retrospective cohort study of 25,270 first-kidney deceased donor transplant recipients (1995-2002) with Medicare primary payer status.
• Pretransplant comorbidities extracted from Medicare claims using Clinical Classifications Software (CCS), Charlson, and Elixhauser comorbidity measures.
• Cox regression models compared predictive performance (c-statistic) for graft loss, death with function (DWF), and death, using OPTN factors alone and augmented with comorbidity data.

Main Results:

• All models showed similar predictive value for graft loss (c-statistic ~0.61-0.63).
• Predictive performance for DWF and death improved slightly with CCS comorbidity data (c-statistic ~0.70).
• Pretransplant comorbidities from administrative claims did not significantly enhance graft outcome predictions beyond existing OPTN factors.

Conclusions:

• Administrative claims-derived comorbidities offer limited additional predictive value for graft outcomes in kidney transplantation.
• Current OPTN data appears adequate for predicting graft survival, suggesting no immediate need to incorporate extensive administrative comorbidity data into allocation schemes.
• Findings have implications for revising current kidney allocation policies.