[en] Introduction:
Stochastic TARgeted (STAR) is unique patient-specific, model-based, and risk-based glycaemic control (GC)
framework providing safe, effective control for virtually all patients, targeting the 80-145 mg/dL range. STAR
uses a proven physiological model and a stochastic model to identify patient-specific insulin sensitivity (SI) from
clinical data and its likely future evolution in the next 1-3 hours. These predicted SI ranges can be used to predict
blood glucose (BG) outcomes for a given insulin and nutrition treatment to maximise nutrition and minimise (mild)
hypoglycaemic risk.
This study presents an initial safety and performance analysis of STAR using an enhanced 3D (STAR-3D) riskprediction
model from use as the new standard of care at Christchurch Hospital, New Zealand.
Methods:
STAR is unique as it modulates both insulin and nutrition inputs. In total, 200 GC episodes are analysed for
performance and safety. This data audit and analysis was approved by the New Zealand Health and Disability
Ethics Committee Upper South Regional Ethics Committee B (Ref: URB/07/15/EXP).
Results:
Results are presented in Table 1. STAR provided high GC efficacy with median [IQR] per-patient %BG in target
band of 81 [65 92]%, with minimal incidence of mild hypoglycaemia overall 0.3%BG<80mg/dL and no incidence of
severe hypoglycaemia (BG<40 mg/dL). The resulting median [IQR] 119 [112 130] mg/dL per-patient median BG is
achieved with median [IQR] 3.5 [2.5 5.0] U/h of median insulin and 97 [80 100] median % goal feed. Considering
results only once the target band is reached (145 mg/dL) to avoid bias from different initial starting BG, both
performance and safety are further improved (Table 1).
Conclusion:
STAR using this new 3D predictive model provides safe, effective control for all patients, with no incidence of
severe hypoglycaemia, despite targeting normoglycaemic ranges. Results are slightly better than the current
standard of care version of STAR.
Disciplines :
Endocrinology, metabolism & nutrition Anesthesia & intensive care Engineering, computing & technology: Multidisciplinary, general & others
Author, co-author :
Uyttendaele, Vincent ; Université de Liège - ULiège > In silico-Model-based therapeutics, Critical Care Basic Sc.
Knopp, Jennifer L.
Shaw, Geoffrey M.
Desaive, Thomas ; Université de Liège - ULiège > Département d'astrophys., géophysique et océanographie (AGO) > Thermodynamique des phénomènes irréversibles
Chase, J. Geoffrey
Language :
English
Title :
Improved Risk-Based Glycaemic Control for Critically Ill Patients: the First 200 Patients
Publication date :
2020
Event name :
e-ISICEM - International Symposium on Intensive Care & Emergency Medicine