Multivariate linear regression on the relative increase in pVO2 after AVR
| β | SE | p Value | |
|---|---|---|---|
| All with p<0.10 at univariate analysis* (adjusted R2=0.33) | |||
| Preoperative mean gradient (per mm Hg) | 0.35 | 0.16 | 0.046 |
| Preoperative BNP>ULN | −5.31 | 5.03 | 0.30 |
| Hypertension | −1.41 | 4.89 | 0.78 |
| Ln days in hospital | −1.76 | 4.16 | 0.68 |
| Post-AVR pacemaker | −8.45 | 5.34 | 0.13 |
| Change in Hb (mmol/L) | 0.59 | 4.16 | 0.89 |
| Post-AVR β-blocker | −3.18 | 4.96 | 0.53 |
| Post-AVR E/e’ | −0.47 | 0.41 | 0.27 |
| Final model (adjusted R2=0.33) | |||
| Preoperative mean gradient (per mm Hg) | 0.50 | 0.14 | 0.001 |
| Preoperative BNP>ULN | −11.62 | 4.15 | 0.009 |
*For BNP, the preoperative BNP>ULN was used because we regard preoperative predictors as most important and BNP>ULN gave slightly higher adjusted R2 in the final model than lnBNP/ULN BNP ratio.
If only all predictors with p<0.05 at univariate analysis were forced into a multivariate model, the mean gradient was also the single predictor with p<0.10 with p=0.047 (adjusted R2 for model=0.39).
AVR, aortic valve replacement; BNP>ULN, brain natriuretic peptide > upper level of normal; E, early diastolic inflow velocity; e’, early lateral mitral annulus velocity; Hb, haemoglobin; pVO2, peak oxygen consumption.