B25-10 Contextualized Universal Vital Assessment Score Better Predicts In-Hospital Mortality

Abstract Rationale Illness severity scores guide triage, characterize patient populations, and predict mortality. The Universal Vital Assessment (UVA) Score employs data available at presentation, but its ability to predict mortality in hypoxemic patients in resource-constrained settings is only modest, with an area under the curve (AUC) of 0.69. The UVA score includes temperature, heart rate, respiratory rate, blood pressure, Glasgow Coma Score, HIV infection, and oxygen saturation. Contextualizing oxygenation with the imputed partial pressure of oxygen to fractional concentration of oxygen in inspired air ratio (iP:F) could improve prognostication. Methods We analyzed data from three East African sites in the Before Building Respiratory support in East Africa Through High flow versus standard flow oxygen Evaluation study (BREATHE, NCT05754034), a prospective cohort study of adults with hypoxemia. We restricted the analysis to 235 patients with complete data, comprising a 70% training cohort. To identify inflection points in the prognostic ability of iP:F on mortality, we employed decision tree analysis (rpart package in R) using the complexity parameter with smallest error by 10-fold cross-validation. We then built a logistic regression model using the original UVA score components, with iP:F replacing peripheral oxygen saturation. We assigned point values to iP:F severity categories by indexing their beta coefficient to the beta coefficient of the best-performing original UVA variable in this population. Results In this population, severe hypoxemia was found to be iP:F ≤100, which was associated with 60.4% absolute in-hospital mortality (odds ratio 9.20). Severe hypoxemia was assigned 4 points. Moderate hypoxemia was found to be iP:F 101-150, which was associated with 38.6% absolute in-hospital mortality (odds ratio 6.64). Moderate hypoxemia was assigned 3 points. The AUC for the contextualized UVA score was 0.75 (95% confidence interval [CI] 0.69-0.82), compared to 0.68 (95%CI 0.61-0.76) for the original UVA score in this population. By DeLong’s test, the contextualized UVA score demonstrated a 7.3% (3.9-10.6%, p-value<0.001) improvement in predictive ability over the original UVA score (Figure). Conclusions Replacing peripheral oxygen saturation with iP:F significantly improves the prognostic performance of the UVA score for patients with hypoxemia in East African hospitals. Future work will refine the model with data from two more hospitals and validate the new score in the remaining 30% of the cohort. This abstract is funded by: Wellcome Fund, NIH T32