Cargando…
Comparing the Predictive Power of Preoperative Risk Assessment Tools to Best Predict Major Adverse Cardiac Events in Kidney Transplant Patients
BACKGROUND: Patients undergoing kidney transplantation have increased risk of adverse cardiovascular events due to histories of hypertension, end-stage renal disease, and dialysis. As such, they are especially in need of accurate preoperative risk assessment. METHODS: We compared three different ris...
Autores principales: | , , , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
Hindawi
2019
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6446090/ https://www.ncbi.nlm.nih.gov/pubmed/31016227 http://dx.doi.org/10.1155/2019/9080856 |
Sumario: | BACKGROUND: Patients undergoing kidney transplantation have increased risk of adverse cardiovascular events due to histories of hypertension, end-stage renal disease, and dialysis. As such, they are especially in need of accurate preoperative risk assessment. METHODS: We compared three different risk assessment models for their ability to predict major adverse cardiac events at 30 days and 1 year after transplant. These were the PORT model, the RCRI model, and the Gupta model. We used a method based on generalized U-statistics to determine statistically significant improvements in the area under the receiver operator curve (AUC), based on a common major adverse cardiac event (MACE) definition. For the top-performing model, we added new covariates into multivariable logistic regression in an attempt to create further improvement in the AUC. RESULTS: The AUCs for MACE at 30 days and 1 year were 0.645 and 0.650 (PORT), 0.633 and 0.661 (RCRI), and finally 0.489 and 0.557 (Gupta), respectively. The PORT model performed significantly better than the Gupta model at 1 year (p=0.039). When the sensitivity was set to 95%, PORT had a significantly higher specificity of 0.227 compared to RCRI's 0.071 (p=0.009) and Gupta's 0.08 (p=0.017). Our additional covariates increased the receiver operator curve from 0.664 to 0.703, but this did not reach statistical significance (p=0.278). CONCLUSIONS: Of the three calculators, PORT performed best when the sensitivity was set at a clinically relevant level. This is likely due to the unique variables the PORT model uses, which are specific to transplant patients. |
---|