Abstract
<title>Abstract</title> <p> <bold>Objective</bold> : Glycaemic control (GC) in the neonatal intensive care unit (NICU) is debated due to significantly increased risk of hypoglycaemia. This study reviews the safety and performance of 15-years of model-based GC in the Christchurch Women’s Hospital NICU. <bold>Study Design</bold> : Retrospective and model-based protocols (STAR; STAR-GRYPHON) are compared in 2-3 year epochs, and overall. Safety is assessed via severe hypoglycaemia, and performance via time in 4.0-8.0mmol/L. Cohort evolution is assessed by the evolution of model-based insulin sensitivity (SI), birthweight (BW), and gestational age (GA). <bold>Results</bold> : Model-based control was safer than Retrospective sliding scale control with better performance. Performance declined over time with increasing low SI values concomitant with lower BW and GA. <bold>Conclusion</bold> : STAR-GRYPHON GC is safe and effective. Performance is increasingly limited by cohort evolution creating trade-offs between increasing insulin dose and risk required to achieve normoglycaemia in the face of demographically decreasing gestational age and birthweight. </p>