Development and validation of a clinical predictive model for severe and critical pediatric COVID-19 infection.

<h4>Introduction</h4>Children infected with COVID-19 are susceptible to severe manifestations. We aimed to develop and validate a predictive model for severe/ critical pediatric COVID-19 infection utilizing routinely available hospital level data to ascertain the likelihood of developing severe manifestations.<h4>Methods</h4>The predictive model was based on an analysis of registry data from COVID-19 positive patients admitted to five tertiary pediatric hospitals across Asia [Singapore, Malaysia, Indonesia (two centers) and Pakistan]. Independent predictors of severe/critical COVID-19 infection were determined using multivariable logistic regression. A training cohort (n = 802, 70%) was used to develop the prediction model which was then validated in a test cohort (n = 345, 30%). The discriminative ability and performance of this model was assessed by calculating the Area Under the Curve (AUC) and 95% confidence interval (CI) from final Receiver Operating Characteristics Curve (ROC).<h4>Results</h4>A total of 1147 patients were included in this analysis. In the multivariable model, infant age group, presence of comorbidities, fever, vomiting, seizures and higher absolute neutrophil count were associated with an increased risk of developing severe/critical COVID-19 infection. The presence of coryza at presentation, higher hemoglobin and platelet count were associated with a decreased risk of severe/critical COVID-19 infection. The AUC (95%CI) generated for this model from the training and validation cohort were 0.96 (0.94, 0.98) and 0.92 (0.86, 0.97), respectively.<h4>Conclusion</h4>This predictive model using clinical history and commonly used laboratory values was valuable in estimating the risk of developing a severe/critical COVID-19 infection in hospitalized children. Further validation is needed to provide more insights into its utility in clinical practice.

Abstract licence: CC BY

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.