PROLOGUE (PROgnostication using LOGistic regression model for Unselected adult cardiac arrest patients in the Early stages): Development and validation of a scoring system for early prognostication in unselected adult cardiac arrest patients

Background: Early prognostication after cardiac arrest would be useful. We aimed to develop a scoring model for early prognostication in unselected adult cardiac arrest patients. Methods: We retrospectively analysed data of adult non-traumatic cardiac arrest patients treated at a tertiary hospital between 2014 and 2018. The primary outcome was poor outcome at hospital discharge (cerebral performance category, 3-5). Using multivariable logistic regression analysis, independent predictors were identified among known outcome predictors, that were available at intensive care unit admission, in patients admitted in the first 3 years (derivation set, N = 671), and a scoring system was developed with the variables that were retained in the final model. The scoring model was validated in patients admitted in the last 2 years (validation set, N = 311). Results: The poor outcome rates at hospital discharge were similar between the derivation (66.0%) and validation sets (64.3%). Age <59 years, witnessed collapse, shockable rhythm, adrenaline dose <2 mg, low-flow duration <18 min, reactive pupillary light reflex, Glasgow Coma Scale motor score >= 2, and levels of creatinine <1.21 mg dl(-1), potassium <4.4 mEq l(-1), phosphate <5.8 mg dl(-1), haemoglobin >= 13.2 g dl(-1), and lactate <8 mmol l(-1) were retained in the final multivariable model and used to develop the scoring system. Our model demonstrated excellent discrimination in the validation set (area under the curve of 0.942, 95% confidence interval 0.917-0.968). Conclusions: We developed a scoring model for early prognostication in unselected adult cardiac arrest patients. Further validations in various cohorts are needed.