THE PREDICTIVE VALUE OF THE NATIONAL EARLY WARNING SCORE 2 IS IMPROVED BY THE ADDITION OF INSPIRED OXYGEN FRACTION AS A WEIGHTED CATEGORICAL VARIABLE

ABSTRACT

TYPE Late Breaking TOPIC Critical Care PURPOSE: The National Early Warning Score (NEWS2) is validated for the prediction of acute deterioration, however, the binary grading of inspired oxygen (FiO2) may be a limitation. We tested whether incorporation of FiO2 as a weighted categorical variable improves the predictive validity of NEWS2. METHODS: A retrospective cohort study of 3704 adult ward admissions between 01/01/17 and 30/03/21, with a viral respiratory infection (SARS-CoV-2/Influenza). 143,157 longitudinal physiological observations were extracted. FiO2 was transformed into a weighted categorical variable and assigned 0-3 points, substituting the original 0/2 points. The primary outcome was a composite of cardiac arrest, unplanned critical care admission or death within 24 hours of the observation. Ethics approval was granted. RESULTS: Baseline characteristics age (mean±SD) 60.4±19.4 years, male gender n(%) 1949(52.6%), Charlson co-morbidity count (mean±SD) 1.6±2.3. The primary outcome occurred in 493(13.3%) patients and was strongly associated with the weighted FiO2 score (χ2, p=<0.001). In patients receiving supplemental oxygen, the area under the receiver operating characteristic curve was greater for NEWS-FiO2 (0.810 (95%CI 0.807-0.814)) versus NEWS2 (0.771 (95%CI 0.767-0.774)). This improvement persisted in the whole cohort. At the 5-point threshold, the positive predictive value (PPV) increased by 22.0% (number needed to evaluate 6.7) for only a 3.9% decrease in sensitivity. CONCLUSIONS: Incorporation of weighted FiO2 into NEWS2 improves the predictive validity for adverse events, particularly by improving the PPV. Confirmation with prospective analysis is required. CLINICAL IMPLICATIONS Weighted FiO2 ought to be considered in the next iteration of NEWS2. This can be operationalised automatically within electronic health records or manually with a reference table. DISCLOSURE No significant relationships. KEYWORD Early Warning Systems