Table 4.
Multivariable logistic regression of clinically relevant variables associated with poor neurological outcomes.
| Adjusted OR (95% CI) | p Value | |
|---|---|---|
| Malignant EEG patterns | 36.43 (4.632–1013.711) | <0.001 |
| First monitored rhythm | ||
| Asystole | 1 | Reference |
| Pulseless electrical activity | 0.84 (0.046–11.674) | 0.894 |
| Shockable rhythm (VT or VF) | 0.19 (0.011–1.622) | 0.135 |
| CPR duration (min) | 1.06 (1.009–1.138) | 0.020 |
| Use of sedative or analgesic | 0.26 (0.023–1.845) | 0.188 |
OR: odd ratio; CI: confidence interval; EEG: electroencephalography; VT: ventricular tachycardia; VF: ventricular fibrillation; CPR: cardiopulmonary resuscitation. Variables with p values less than 0.2 in univariate analyses and clinically relevant variables were subjected to a stepwise multiple logistic regression model to obtain statistically meaningful predictor variables. They were EEG groupings by its pattern, age, target temperature management, first monitored rhythm, CPR duration, Glasgow Coma Scale on EEG scan, and use of sedative or analgesic. To alleviate the small sample issue, we applied the Firth’s correction.