Table 3.
Multivariable Model for Mortality in Relation to Baseline BNP, Aldosterone, and Fluid Management Strategy
| Characteristic | Complete Cases (n = 513) |
Multiple Imputations (n = 625) |
||||
|---|---|---|---|---|---|---|
| OR | 95% CI | P Value | OR | 95% CI | P Value | |
| Age, y | 1.96 | 1.36-2.81 | < .001 | 1.87 | 1.32-2.50 | < .001 |
| Female sex | 0.60 | 0.37-0.99 | .048 | 0.69 | 0.44-1.07 | .10 |
| Race (non-white to white) | 1.23 | 0.73-2.06 | .43 | 1.29 | 1.07-2.67 | .03 |
| BMI, kg/m2 | 0.67 | 0.45-1.00 | .03 | 0.78 | 0.55-1.20 | .12 |
| APACHE III score | 3.28 | 2.15-4.99 | < .001 | 3.33 | 2.26-4.84 | < .001 |
| Shock | 1.94 | 1.16-3.23 | .01 | 1.72 | 1.12-2.85 | .02 |
| Central venous pressure, mm Hg | 0.86 | 0.59-1.25 | .43 | 0.88 | 0.62-1.26 | .35 |
| Fluid strategy (liberal to conservative) | 0.95 | 0.58-1.55 | .82 | 1.148 | 0.74-1.78 | .54 |
| Conservative group | ||||||
| BNP, log-transformed | 0.78 | 0.42-1.43 | .42 | 0.80 | 0.47-1.37 | .42 |
| Aldosterone, log-transformed | 1.22 | 0.86-1.75 | .26 | 1.22 | 0.88-1.67 | .23 |
| Liberal group | ||||||
| BNP, log-transformed | 0.71 | 0.38-1.33 | .28 | 0.89 | 0.50-1.57 | .69 |
| Aldosterone, log-transformed | 0.68 | 0.43-1.08 | .10 | 0.64 | 0.43-0.96 | .03 |
| BNP × fluid strategy | … | … | .83 | … | … | .79 |
| Aldosterone × fluid strategy | … | … | .04 | … | … | .01 |
Of the 625 patients with both biomarkers available on day 0, 112 were missing central venous pressure or BMI data and were either excluded (complete cases) or had these values imputed (multiple imputations). Sensitivity analysis adding baseline creatinine and blood urea nitrogen to the model did not change aldosterone’s modification of the effect of fluid management on mortality (P value for interaction = .01). The odds of 60-d in-hospital mortality are given for the 75th percentile relative to the 25th percentile of age, BMI, APACHE III score, central venous pressure, log-transformed BNP, and log-transformed aldosterone. See Table 1 legend for expansion of abbreviations.