Pediatric Traumatic Brain Injury and Healthcare-associated Infections: A Prospective Study on Determinants of Mortality Using Least Absolute Shrinkage and Selection Operator Regression.

Abstract

Background and aims: Healthcare-associated infections (HAIs) represent an important secondary insult in pediatric traumatic brain injury (TBI); data from low- and middle-income country (LMIC) settings remain limited, and clinically applicable models using routinely available variables to predict mortality are lacking. We aimed to characterize the burden and microbiological profile of HAIs and to identify determinants of mortality in pediatric TBI using routine clinical variables.

Patients and methods: A prospective observational study was conducted at a level 1 trauma center in New Delhi, India, from July 2023 to December 2024. The study enrolled 160 pediatric patients (6 months to 15 years) with moderate to severe pediatric TBI [Glasgow Coma Scale (GCS): 3-12] requiring mechanical ventilation for at least 24 hours. Baseline variables included age, sex, injury mechanism, Injury Severity Score (ISS), GCS, and Marshall CT classification. HAIs were identified through active surveillance, with pathogen profiles and antibiotic susceptibility analyzed. Hospital courses, mortality, and duration of hospital stay were noted. Functional outcomes were assessed at 3 months using the Glasgow Outcome Scale Extended (GOSE).A least absolute shrinkage and selection operator (LASSO) regression model was used for finding the determinants of mortality using age, sex, ISS, surgical intervention, GCS, ventilator days, hospital days, and HAI status. Model performance was evaluated using accuracy, area under the curve (AUC), sensitivity, specificity, and Kappa statistic.

Results: Healthcare-associated infections occurred in 35% of patients, primarily respiratory (61.1%), with Acinetobacter baumannii (40.3%) as the leading pathogen. Healthcare-associated infection-positive patients had longer ventilator, neurotrauma intensive care unit, and hospital stays (p < 0.001). Among other determinants, ventilator days [odds ratio (OR): 1.93] and HAI (OR: 1.55) emerged as important predictors within the multivariable LASSO analysis. The LASSO model achieved 97.5% accuracy, 0.995 AUC, 99.2% sensitivity, and 92.1% specificity, with ventilator days, HAI status, and admission GCS as key predictors of mortality.

Conclusions: Healthcare-associated infections were associated with prolonged critical care burden in mechanically ventilated pediatric patients with moderate to severe TBI. A LASSO model using routinely available clinical variables offers a robust, scalable tool for determinants of mortality in LMIC neurocritical care settings. However, these findings should not be generalized to non-ventilated or mild TBI populations and warrant external validation.

How to cite this article: Ahmed MN, Bindra A, Sahu P, Thakur AK, Mohamed Samsudeen MS, Goyal K, et al. Pediatric Traumatic Brain Injury and Healthcare-associated Infections: A Prospective Study on Determinants of Mortality Using Least Absolute Shrinkage and Selection Operator Regression. Indian J Crit Care Med 2026;30(8):656-663.