Physiological-social score (PMEWS) vs. CURB-65 to triage pandemic influenza: A comparative validation study using community-acquired pneumonia as a proxy

Background. An influenza pandemic may increase Emergency Department attendance 7-fold. In the absence of a validated "flu score" to assess severity and assist triage decisions from primary into secondary care, current UK draft management recommendations have suggested the use of CURB-65 and chest X-ray as a proxy. We developed the Pandemic Medical Early Warning Score (PMEWS) to track and triage flu patients, taking into account physiological and social factors and without requiring laboratory or radiology services. Methods. Validation of the PMEWS score against an unselected group of patients presenting and admitted to an urban UK teaching hospital with community acquired pneumonia. Comparison of PMEWS performance against CURB-65 for three outcome measures: need for admission, admission to high dependency or intensive care, and inpatient mortality using area under ROC curve (AUROC) and the Hanley-McNeil method of comparison. Results. PMEWS was a better predictor of need for admission (AUROC 0.944) and need of higher level of care (AUROC 0.83) compared with CURB-65 (AUROCs 0.881 and 0.640 respectively) but was not as good a predictor of subsequent inpatient mortality (AUROC 0.663). Conclusion. Although further validation against other disease datasets as a proxy for pandemic flu is required, we show that PMEWS is rapidly applicable for triage of large numbers of flu patients to self-care, hospital admission or HDU/ICU care. It is scalable to reflect changing admission thresholds that will occur during a pandemic.