Development and validation of the predictive score for pediatric COVID-19 pneumonia: A nationwide, multicenter study

Abstract

Background Due to the possibility of asymptomatic pneumonia in children with COVID-19 leading to overexposure to radiation and problems in limited-resource settings, we conducted a nationwide, multi-center study to determine the risk factors of pneumonia in children with COVID-19 in order to create a pediatric pneumonia predictive score, with score validation. Methods This was a retrospective cohort study done by chart review of all children aged 0–15 years admitted to 13 medical centers across Thailand during the study period. Univariate and multivariate analyses as well as backward and forward stepwise logistic regression were used to generate a final prediction model of the pneumonia score. Data during the pre-Delta era was used to create a prediction model whilst data from the Delta one was used as a validation cohort. Results The score development cohort consisted of 1,076 patients in the pre-Delta era, and the validation cohort included 2,856 patients in the Delta one. Four predictors remained after backward and forward stepwise logistic regression: age < 5 years, number of comorbidities, fever, and dyspnea symptoms. The predictive ability of the novel pneumonia score was acceptable with the area under the receiver operating characteristics curve of 0.677 and a well-calibrated goodness-of-fit test (p = 0.098). The positive likelihood ratio for pneumonia was 0.544 (95% confidence interval (CI): 0.491–0.602) in the low-risk category, 1.563 (95% CI: 1.454–1.679) in the moderate, and 4.339 (95% CI: 2.527–7.449) in the high-risk. Conclusion This study created an acceptable clinical prediction model which can aid clinicians in performing an appropriate triage for children with COVID-19. © 2022 Satdhabudha et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.