27 Risk of requiring intravenous antibiotics among paediatric patients with cellulitis: external validation of the Melbourne ASSET Score

Abstract Background Cellulitis is a prevalent condition leading to emergency department (ED) visits, but a lack of evidence-based consensus exists regarding the classification of cellulitis severity and the criteria for selecting between intravenous (IV) and oral (PO) antibiotic regimens in moderate-to-severe cases. The Melbourne ASSET score was designed to aid clinicians in identifying patients who may require IV treatment; however, its external validity has yet to be thoroughly examined in diverse clinical settings. Objectives To validate the Melbourne ASSET score in a paediatric cohort with cellulitis presenting to another ED. Design/Methods We prospectively enrolled patients aged 6 months to 18 years diagnosed with cellulitis at a tertiary care paediatric center. Patients were excluded if they had orbital cellulitis, were immunosuppressed/compromised, showed signs of toxicity, or were unable to take oral antibiotics. The primary outcome was the proportion of patients for whom the Melbourne ASSET score accurately predicted the route of antibiotic administration at 24 hours, compared to the treatment decisions made by the attending physician. We assessed the overall performance of the score using a Receiver Operating Characteristic curve, and inter-rater reliability among physicians was evaluated using the kappa statistic. Results From January 2022 to January 2024, a total of 230 patients were enrolled, with a median age of 5.8 years and 3 months (IQR: 2.9, 9.5). The cohort was predominantly male (61.7%), and 53.9% of patients had a Melbourne ASSET score of 4 or higher (maximum score of 7). At the 24-hour mark, 155 patients (67%) received PO antibiotics, while 75 patients (33%) were administered IV antibiotics. Using a threshold of 4 points to initiate IV treatment, the Melbourne ASSET score demonstrated a poor discriminatory capacity, with an area under the curve of 0.64 (95%CI: 0.58, 0.71). The score exhibited a sensitivity of 73%, specificity of 56%, positive predictive value of 43%, and negative predictive value of 82%. Additionally, among the 47 patients evaluated by two physicians, inter-rater reliability was found to be 0.28 (95%CI: 0.10, 0.45). Conclusion The Melbourne ASSET score was found to be insufficient in accurately predicting which paediatric patients with moderate-to-severe cellulitis required IV antibiotics when compared to clinical judgment by attending physicians. Consequently, reliance on this score alone may increase unnecessary hospital admissions or IV antibiotic prescriptions rather than providing a reliable framework for treatment decisions in other settings.