Development and validation of a predictive tool for adverse drug reactions in neonates under intensive care.

Abstract

Neonates hospitalized in neonatal intensive care units (NICUs) commonly experience adverse drug reactions (ADRs). Thus, we aimed to develop and validate a tool for predicting ADRs in neonates hospitalized in NICUs. A nested case-control study in an open cohort with neonates admitted to the NICU of a maternity hospital in Natal, Brazil was conducted from January 2019 to January 2022 [Correction added on 4 December 2023, after first online publication: 2023 has been changed to 2019 in the preceding sentence.]. Neonates with ADR were randomly paired with 2 controls. For the development of the tool, a multivariate logistic regression was applied on 2/3 of the sample (cases with respective controls). The model's fit was evaluated using the Hosmer-Lemeshow test for calibration and the Brier score for performance assessment. Validation of the tool was performed by determining the area under the receiver operating characteristic curve with bootstrap adjusted c-statistics. In all, 450 neonates (150 cases and 300 controls) were included in the study. We identified 5 independent risk factors for ADR, 4 related to the neonate (current mechanical ventilation, heart rate ≥178 beats/min, intravenous medications, ≥5 prescription medications) and 1 to the mother (gestational hypertension). The tool had a classification cut-off point of ≥15, and its total score ranged from 0 to 34. In validation, the tool had an area under the receiver operating characteristic curve of 0.74 (95% confidence interval [CI] 0.66-0.81) with sensitivity of 52.02% (95% CI 47.40-56.64) and specificity of 81.35% (95% CI 77.75-84.95). The tool demonstrated adequate discriminative ability and utilized 5 commonly monitored variables in the NICU.