Clinical syndromes caused by covid-19 and a bayesian model to predict severity

ABSTRACT

Background: This study aimed to identify the different syndromes presented in hospitalized children with SARS-CoV-2, to analyze if the clinical features and biomarkers confer different risk depending on the syndromes, and to create a predictive model to anticipate the probability of the need for critical care Methods: We conducted a multicenter, prospective study of children aged 0 to 18 years old with SARS-CoV-2 infection in 52 Spanish hospitals. The primary outcome was the need for critical care defined as the combined outcome of admission into a PICU, and/or need for respiratory support beyond nasal prongs. To understand the probability of needing critical care according to the diagnostic group and for each risk factor, a Bayesian multivariable model was applied. To build a predictive model of critical care, a naïve Bayes algorithm was implemented in a web app. Results: 292 children were hospitalized from March 12th, 2020 to July 1st, 2020;Of them, 214 (73.3%) were considered to have relevant COVID-19 (r-COVID-19). Among patients with r-COVID-19, 24.2% needed critical care. Out of 214 patients, 22.4% were admitted into a pediatric intensive care unit, 41.6% required respiratory support, and 38.8% presented complications (mostly cardiological). Four patients (1.8%) died, all of them had severe comorbidities. We identified 11 primaries diagnoses and grouped them into 4 large syndromes of decreasing severity MIS-C (17.3%), bronchopulmonary (51.4%), gastrointestinal (11.6%), and mild syndrome with complications (19.6%). In the predictive model, the predictors with higher relative importance were high C-reactive protein, anemia, lymphopenia, platelets <220 000/mm3, type of syndrome, high creatinine, and days of fever. The different risk factors increase the risk differently depending on the patient's syndrome the more severe the syndrome, the more risk the factor confers. We developed an online risk prediction tool to quantify the risk of critical disease (https//rserver.h12o. es/pediatria/EPICOAPP/, username user, password0000) Conclusion: We described the spectrum of r-COVID-19 in hospitalized children, consisting of 4 large syndromes of decreasing severity MIS-C, bronchopulmonary syndrome, gastrointestinal syndrome, and a mild syndrome with complications. The risk factors increase the risk differently depending on the syndrome. A Bayesian model was implemented in an online app to anticipate the individual risk of critical care.