VS10P DOES THE IRAD SCORE PREDICT IN‐HOSPITAL MORTALITY AFTER SURGICAL TREATMENT OF ACUTE TYPE A AORTIC DISSECTION?

Abstract

Purpose: A risk model for predicting in‐hospital mortality after surgical repair of acute type A aortic dissection (AAAD) was recently developed from data from the International Registry of Acute Aortic Dissection (IRAD). We sought to determine whether the IRAD risk model would accurately predict in‐hospital mortality in our series of patients with AAAD.Methodology: We retrospectively reviewed peri‐operative data from 88 patients who underwent AAAD repair on our service between 2003 and 2008. Data were collected for each of the variables in the proposed IRAD risk model for predicting in‐hospital mortality; the IRAD model was fit to these data. The IRAD risk score was calculated for each patient, and then patients were stratified into 8 groups according to their score. We compared the observed and predicted mortalities and used the likelihood ratio chi‐square test to evaluate model fit.Results: The observed in‐hospital mortality was 18% (n = 16). The variables used in the IRAD model were not significantly associated with mortality in our patients. Furthermore, in groups 2 through 8, agreement was poor between predicted and observed mortality rates: respectively, 5% vs. 10% for group 2, 10% vs. 21% for group 3, 15% vs. 33% for group 4, 25% vs. 17% for group 5, 35% vs. 50% for group 6, 45% vs. 29% for group 7, and 60% vs. 33% for group 8. The likelihood ratio chi‐square test confirmed that the IRAD model fit our data poorly (chi‐square = 10.68; degrees of freedom, 7; p = 0.15).Conclusion: The IRAD risk model did not accurately predict in‐hospital mortality in our AAAD patients. Prospective validation studies are needed before the IRAD score can be considered for adoption as a tool for making clinical decisions.