Model-Based Approach to Improve Clinical Outcomes in Neonates With Opioid Withdrawal Syndrome Using Real-World Data.

Abstract

At least 60% of the neonates with opioid withdrawal syndrome (NOWS) require morphine to control withdrawal symptoms. Currently, the morphine dosing strategies are empiric, not optimal and associated with longer hospital stay. The aim of the study was to develop a quantitative, model-based, real-world data-driven approach to morphine dosing to improve clinical outcomes, such as reducing time on treatment. Longitudinal morphine dose, clinical response (Modified Finnegan Score (MFS)), and baseline risk factors were collected using a retrospective cohort design from the electronic medical records of neonates with NOWS (N=177) admitted to the University of Maryland Medical Center. A dynamic linear mixed effects model was developed to describe the relationship between MFS and morphine dose adjusting for baseline risk factors using a split-sample data approach (70% training: 30% test). The training model was evaluated in the test dataset using a simulation based approach. Maternal methadone and benzodiazepine use, and race were significant predictors of the MFS response. Positive autocorrelations of 0.56 and 0.12 were estimated between consecutive MFS responses. On an average, for a 1,000μg increase in the morphine dose, the MFS decreased by 0.3units. The model evaluation showed that observed and predicted median time on treatment were similar (13.0 vs. 13.8days). A model-based framework was developed to describe the MFS-morphine dose relationship using real-world data that could potentially be used to develop an adaptive, individualized morphine dosing strategy to improve clinical outcomes in infants with NOWS.