Table 2. Predictive Value in Univariable and Multivariable Analysisa.
| Predictor Variable | Univariable | Multivariable (nā=ā552) |
||
|---|---|---|---|---|
| HR (95% CI) | P Value | HR (95% CI) | P Value | |
| Prior palliative shunt | 2.44 (1.26-4.75) | .009 | 1.77 (0.88-3.56) | .11 |
| QRS duration ā„180 ms | 1.94 (0.59-6.34) | .27 | 0.67 (0.15-3.10) | .61 |
| Ventriculotomy incision | 2.13 (0.99-4.56) | .05 | 1.67 (0.76-3.66) | .20 |
| Previous VT | 5.02 (1.77-14.2) | .002 | 3.98 (1.13-14.0) | .03 |
| LV EF (per %)b | 0.92 (0.89-0.95) | <.001 | 0.95 (0.92-0.99) | .01 |
| RV EF (per %)b | 0.94 (0.91-0.97) | <.001 | 0.96 (0.92-0.99) | .01 |
| LV EF <45%c | 4.77 (2.42-9.39) | <.001 | 3.23 (1.57-6.65) | .001 |
| RV EF <30%c | 5.28 (2.62-10.6) | <.001 | 3.90 (1.84-8.26) | <.001 |
Abbreviations: EF, ejection fraction; HR, hazard ratio; LV, left ventricle; RV, right ventricle; VT, ventricular tachycardia.
Predictive value of the variables in the original Khairy et al risk model and the cardiovascular magnetic resonance imaging thresholds, which were only included in the adapted model. Predictive value for adverse outcomes in univariable and multivariable Cox regression analysis are described as HR with 95% CIs and P values.
Multivariable analysis was first performed with RV EF and LV EF as continuous variables.
The HRs reported for RV EF less than 30% and LV EF less than 45% were based on a multivariable model including the noninvasive component of the Khairy et al risk model but without RV EF and LV EF as continuous variables.