Derivation of the Pediatric Acute Gastroenteritis Risk Score to Predict Moderate‐to‐Severe Acute Gastroenteritis

• Published in: Journal of pediatric gastroenterology and nutrition Vol. 74; no. 4; pp. 446 - 453
• Main Authors: Levine, Adam C., O'Connell, Karen J., Schnadower, David, VanBuren, T. John M., Mahajan, Prashant, Hurley, Katrina F., Tarr, Phillip, Olsen, Cody S., Poonai, Naveen, Schuh, Suzanne, Powell, Elizabeth C., Farion, Ken J., Sapien, Robert E., Roskind, Cindy G., Rogers, Alexander J., Bhatt, Seema, Gouin, Serge, Vance, Cheryl, Freedman, Stephen B.
• Format: Journal Article
• Language: English
• Published: United States 01.04.2022
• Subjects: children; diarrhea; emergency department; prognosis
• ISSN: 0277-2116; 1536-4801
• DOI: 10.1097/MPG.0000000000003395

ABSTRACT Objectives: Although most acute gastroenteritis (AGE) episodes in children rapidly self‐resolve, some children go on to experience more significant and prolonged illness. We sought to develop a prognostic score to identify children at risk of experiencing moderate‐to‐severe disease after an index emergency department (ED) visit. Methods: Data were collected from a cohort of children 3 to 48 months of age diagnosed with AGE in 16 North American pediatric EDs. Moderate‐to‐severe AGE was defined as a Modified Vesikari Scale (MVS) score ≥9 during the 14‐day post‐ED visit. A clinical prognostic model was derived using multivariable logistic regression and converted into a simple risk score. The model's accuracy was assessed for moderate‐to‐severe AGE and several secondary outcomes. Results: After their index ED visit, 19% (336/1770) of participants developed moderate‐to‐severe AGE. Patient age, number of vomiting episodes, dehydration status, prior ED visits, and intravenous rehydration were associated with MVS ≥9 in multivariable regression. Calibration of the prognostic model was strong with a P value of 0.77 by the Hosmer‐Lemenshow goodness‐of‐fit test, and discrimination was moderate with an area under the receiver operator characteristic curve of 0.68 (95% confidence interval [CI] 0.65–0.72). Similarly, the model was shown to have good calibration when fit to the secondary outcomes of subsequent ED revisit, intravenous rehydration, or hospitalization within 72 hours after the index visit. Conclusions: After external validation, this new risk score may provide clinicians with accurate prognostic insight into the likely disease course of children with AGE, informing disposition decisions, anticipatory guidance, and follow‐up care.