Table 4.
Prognostic Value of 2D and 3D Echocardiography in Assessing the Risk of Mitral Valve Reoperation Based on ROC Analysis and Decision-Tree Algorithms
| Mitral Stenosis and Mixed Disease Patients | ||
|---|---|---|
| Metric | Change in 2D Mean Gradient | Change in 3D-EOA |
| AUC (95% CI) | 0.676 (0.508-0.844) | 0.847 (0.723-0.970) |
| Best cutoff identified by the decision tree algorithm | <30% increase | <30% increase |
| Sensitivity | 88% (23/26) | 61% (11/18) |
| Specificity | 37% (7/19) | 95% (20/21) |
| PPV | 66% | 92% |
| NPV | 70% | 74% |
| Mitral Regurgitation and Mixed Disease Patients | ||
|---|---|---|
| Change in 2D-VCRA | Change in 3D-VCRA | |
| AUC (95% CI) | 0.720 (0.424-0.903) | 0.969 (0.916-0.999) |
| Best cutoff identified by the decision tree algorithm | <40% decrease | <40% decrease |
| Sensitivity | 47% (7/15) | 93% (13/14) |
| Specificity | 95% (21/22) | 94% (15/16) |
| PPV | 88% | 93% |
| NPV | 72% | 94% |
AUC curves are plotted and compared in the upper panel of the Central Illustration. Decision-tree algorithms are summarized in the lower panel of the Central Illustration.
3D-EOA = 3D effective orifice area; 3D-VCRA = 3D vena contracta regurgitant area; AUC = area under the curve; NPV = negative predictive value; PPV = positive predictive value; ROC = receiver operating characteristic.