Table 5. Predictors of in-hospital mortality using multivariable logistic regression with stepwise backward variable selection to minimize the AIC value.
| Variable | Adjusted OR | 95% CI | P value |
|---|---|---|---|
| Demographics & comorbidities | |||
| Age (per 10 years) | 1.300 | 1.116–1.514 | 0.0008 |
| Female sex | 0.713 | 0.492–1.034 | 0.0741 |
| Charlson Comorbidity Index (per point) | 1.127 | 1.056–1.202 | 0.0003 |
| History of diabetes mellitus | 0.652 | 0.436–0.975 | 0.04 |
| Year of admission (per year) | 0.904 | 0.843–0.970 | 0.005 |
| Admission diagnoses | |||
| Respiratory failure | 2.525 | 1.661–3.836 | <0.0001 |
| Sepsis | 2.106 | 1.294–3.429 | 0.0027 |
| Severity of illness | |||
| APACHE score (per 10 points) | 1.223 | 1.096–1.364 | 0.0003 |
| Day 1 SOFA score (per point) | 0.890 | 0.813–0.974 | 0.01 |
| Admission Braden Skin Score (per point) | 0.900 | 0.844–0.960 | 0.0013 |
| Shock severity | |||
| SCAI shock stage (per stage) | 1.243 | 1.024–1.508 | 0.03 |
| Inodilators in first 24 hours | 1.864 | 1.054–3.296 | 0.03 |
| Procedures and therapies | |||
| Angiogram without PCI vs. no angiogram | 0.726 | 0.473–1.112 | 0.14 |
| PCI vs. no angiogram | 0.424 | 0.267–0.673 | 0.0003 |
| PCI vs. angiogram without PCI | 0.583 | 0.348–0.980 | 0.04 |
| pVAD or ECMO | 6.021 | 1.621–22.365 | 0.0073 |
| Dialysis | 2.005 | 0.826–4.866 | 0.12 |
| CRRT | 3.272 | 1.001–10.698 | 0.05 |
| Cardiac arrest | |||
| VF CA vs. no CA | 3.036 | 1.795–5.135 | <0.0001 |
| Non-VF CA vs. no CA | 6.878 | 4.112–11.505 | <0.0001 |
| VF CA vs. non-VF CA | 2.266 | 1.246–4.121 | 0.007 |
| IHCA | 3.372 | 1.740–6.534 | 0.02 |
| LVSWI (per 10 J/m2) | 0.664 | 0.564–0.782 | <0.0001 |
Data are displayed as adjusted odds ratio (OR) and 95% confidence interval (CI) values. The final model C-statistic value was 0.927 for discrimination of in-hospital mortality. Candidate variables which were not selected for the model included APACHE-IV predicted hospital mortality; race; invasive and noninvasive ventilator use; history of myocardial infarction, heart failure, chronic kidney disease, dialysis and stroke; peak VIS and NEE in first 24 hours; use of vasopressors in first 24 hours; IABP and PAC use; blood transfusion; CardShock score; LVEF; and admission diagnosis of heart failure or acute coronary syndrome.