Establishing a nomogram model for predicting hospital length of stay in children with II-III degree scald burns based on admission assessment indicators.

BACKGROUND: Existing pediatric burn studies have primarily focused on risk factors or survival outcomes, while predictors of hospital length of stay (LOS) remain insufficiently explored. This study aimed to develop and validate a nomogram predicting LOS by incorporating the clinical characteristics of children with II-III degree scald burns. METHODS: This retrospective study collected cases of pediatric patients with II-III degree scald burns admitted to Children's Hospital, Zhejiang University School of Medicine, National Clinical Research Center for Children and Adolescents' Health and Diseases between July 2024 and August 2025. A generalized linear regression model for predicting LOS was established based on univariate Gamma regression, multicollinearity tests, and multivariate Gamma regression to screen variables, followed by construction of a nomogram. Model performance and calibration were internally validated using root mean square error (RMSE), mean absolute error (MAE), cross-validation, and 1,000-times bootstrap resampling. RESULTS: A total of 191 children (male: 64.9%; female: 35.1%) with II-III degree scald burns were included, with the majority aged between 1 and 3 years. Multivariate analysis showed that burn time before admission, total body surface area, albumin, blood glucose, white blood cell count, hemoglobin and skin condition were independent predictors of LOS in children with II-III degree scald burns. The model established had the lowest Akaike information criterion value, indicating the best fit. The Gamma generalized linear model demonstrated moderate predictive performance with a cross-validated R2 of 0.31, along with acceptable calibration and mild optimism in bootstrap internal validation. CONCLUSIONS: We constructed a nomogram to predict LOS in children with scald burns, which may help clinicians assess LOS early and optimize treatment.