Predictive accuracy of EuroSCORE: is end-diastolic dysfunction a missing variable?

Objective: Left-ventricular function has been shown to be an important prognostic factor in estimating operative risk in cardiac surgery. As such, left-ventricular ejection fraction (LVEF) is included in the EuroSCORE. However, left-ventricular function is more comprehensively assessed by measures of both systolic and diastolic dysfunction. We hypothesised that end-diastolic dysfunction is an additional independent indicator for predicting outcome following coronary artery bypass grafting (CABG). Methods: We retrospectively assessed all patients undergoing isolated off-pump CABG between October 2000 and September 2004 by two surgeons. Left-ventricular end-diastolic pressure (LVEDP), measured during cardiac catheterisation, was used as a measure of left-ventricular diastolic dysfunction. Logistic regression was used to assess the association between LVEDP (a continuous and dichotomous variable) and mortality, while adjusting for EuroSCORE. Results: A total of 925 patients with complete LVEDP data were identified and stratified as follows: group 1 (LVEF >30% and LVEDP ≪20 mmHg), group 2 (LVEF ≪30% and LVEDP ≪20 mmHg), group 3 (LVEF >30% and LVEDP >20 mmHg) and group 4 (LVEF ≪30% and LVEDP >20 mmHg). Mortality increased progressively from group 2 (1.9%, odds ratio (OR) 1.22, RR 1.21, p 0.58) to group 3 (5.6%, OR 3.81, RR 3.66, p 0.07) and was highest in group 4 (7.4%, OR 5.18, RR 4.87, p 0.08). Receiver operating characteristic (ROC) curve c-characteristic improved from 0.7 to 0.78 when EuroSCORE was combined with LVEDP, identifying LVEDP as an independent predictor of mortality after adjusting for EuroSCORE. Logistic equation: odds of death = exp(−6.3283 + [EuroSCORE × 0.1813] + [EDP × 0.0954]). Conclusions: LVEDP as a marker of diastolic dysfunction seems an important variable in predicting patient-specific risk and should be considered for incorporation in future risk models.