Translating A Risk-Based Glycaemic Control Framework for Critically Ill Patients: STAR-Liège

peer reviewedGlycaemic control (GC) in the intensive care unit (ICU) has been widely debated over the last 20 years. While many studies showed benefits, many others failed to replicate the results, blaming the increased related risk of hypoglycaemia. Current ICU guidelines thus often suggest higher glycaemic target ranges, led by the fear of hypoglycaemia – permissive hyperglycaemia. However, recent studies have shown improved safety and performance in GC outcome, using model-based computerised methods. The Stochastic-Targeted (STAR) framework is a patient-specific risk-based dosing protocol modulating insulin and nutrition. This study presents recent intermediate results of the STAR-Liège clinical trial, targeting 4.4-8.0 mmol/L glycaemic band. Clinical data from patients controlled under STAR and STAR insulin only (STAR-IO) are compared to retrospective data under the standard protocol (SP), targeting higher 5.6-8.3 mmol/L glycaemic ranges. Overall, STAR performance was significantly higher (88% blood glucose measurements in the 4.4-8.0 mmol/L or 80-145 mg/dL target band) compared to STAR-IO (78%) and SP (55%). Incidence of hypoglycaemia was similar (1% below target), while hyperglycaemia was much higher for SP (31% above target) compared to STAR (9%) and STAR-IO (11%). The resulting lower median blood glucose (BG) levels in STAR (6.5 mmol/L), compared to STAR-IO (6.7 mmol/L) and SP (7.7 mmol/L), was achieved with less variability, but required higher clinical workload for STAR (12 measurements per day) compared to SP (7 measurements per day). Compliance to protocol was higher for STAR (98%) compared to STAR-IO (90%) and SP (79%). Although targeting lower glycaemic ranges, STAR provided better GC compared to the SP. Typically, the full version of STAR also modulating nutrition, was able to better control extremely insulin resistant patients, further improving glycaemic control results. The results of this clinical trial indicate the capability to provide the safe, effective control for all patients required to improve outcomes