New model predicts critical admission in children with dengue

A recently developed model demonstrates its utility in predicting critical care entrance among children with dengue.

The development and internal validation of the predictive model involved a total of 1,385 patients. In temporal validation with 519 additional participants, the c-statistic was 0.82 (95 percent confidence interval [CI], 0.77–0.87), and the calibration slope was 0.98 (95 percent CI, 0.77–1.18).

The research team selected the 50th percentile of the distribution of predicted probability of critical care entrance (5 percent) as the threshold for increased alert at emergency admission. The result missed 10 percent of cases that needed to enter critical care (sensitivity, 90 percent, 95 percent CI, 82–95; specificity, 48 percent, 95 percent CI, 41–50).

“We recommend the validation and potential recalibration of our score in other clinical settings,” the researchers said. 

This retrospective cohort study used admissions from January 2019 to August 2021 at Hospital Infantil Napoleon Franco Pareja in Cartegena, Colombia. All children (aged 18 years or younger) with a positive immunoglobulin M or nonstructural protein 1 laboratory test and admitted for follow-up at the emergency department were included.

The research team selected variables retrospectively collected on emergency admission for feature selection and assessed discrimination and calibration in the development dataset, using 1,000 bootstrap replications for internal validation. They used data from 2019 to 2020 for development and 2021 for temporal validation.

The c-statistic for discrimination with 95 percent CIs and the calibration intercept and slope were estimated and reported.