CLINICAL RESULTS FROM A GLYCAEMIC CONTROL
PROTOCOL IMPLIMENTATION IN NEONATAL CARE
JL Dickson1, AM Lynn2, AJ LeCompte1, T Desaive3, B Benyó4, JG Chase1
1University
of Canterbury, Department of Mechanical Engineering, New Zealand
2Chrischurch
Women's Hospital, Neonatal Intensive Care Unit, New Zealand
3University
of Liege, GIGA-Cardiovascular Sciences, Belgium
4Budapest
University of Technology and Economics, Hungary
Abstract
Hyperglycaemia is associated with increased
mortality and morbidity in neonatal intensive
care. STAR-GRYPHON is a model-based
glycaemic control protocol, which uses insulin
sensitivity to tailor insulin treatments to patients
and changing patient condition. It is implemented
as a standard of care in a neonatal intensive
care unit in Christchurch, New Zealand.
Keyword(s): physiological modelling - clinical
engineering
1. Introduction
High blood glucose (BG), hyperglycaemia, in the
neonatal intensive care unit (NICU) affects 3070% of extremely premature infants (gestational
age ≤ 27 weeks), and is associated with
increased
risk
of
infection,
ventilator
dependence, hospital length of stay and
mortality. Increasingly, evidence suggests that
controlling the blood sugar levels to a normal
range can prevent and reduce the extent of
morbidities, thus improving patient outcomes and
reducing the costs of care.
STAR-GRYPHON
(Stochastic
TARgeted
Glycaemic Regulation sYstem to Prevent Hyperand hypO-glycaemia in Neonates) is a modelbased protocol for dosing insulin in very
premature infants. It is currently being used as a
standard of care in Christchurch (CCH), New
Zealand (NZ). It is also being implemented in
Miskolc, Hungary, and as part of a randomized
trial in Auckland (NZ). Presented here are clinical
results from its use in CCH.
2. Model based control
Mathematical
models
of
glucose-insulin
physiology are used. Inter-patient variability is
managed directly by identifying insulin sensitivity
from data available at the bedside. Prediction of
future sensitivity values using cohort based
statistical modelling allows likely BG outcomes to
be forecast for a given treatment. Typically,
treatments are selected such that the risk of
BG<4.4 mmol/L is 5%, maximising likelihood of
outcomes in the target range of 4.0-8.0 mmol/L.
2.3 Implementation
STAR-GRYPHON is implemented on a tablet
computer interface, designed in conjunction with
clinical staff. Starting criteria for insulin is two
consecutive BG>10 mmol/L in 4-6 hrs.
3. Results
STAR-GRYPHON achieves more BG within the
clinically targeted range than retrospective data,
with much less hypoglycaemia, which can also
be dangerous.
Table 1: Clinical results, with comparison to
retrospective data.
Retrospective
STAR-GRYPHON
Patient Cohort
# Episodes
25
13
GA (weeks)
27 [25-27]
26 [25-27]
Weight (g)
845 [800-900]
925 [750-1030]
PNA (days)
N/A
3 [1.5-3.5]
Clinical interventions
Meas. intrvl. (hr)
3.2 [2.6 - 3.9]
4.0 [3.9-4.1]
Insulin (U/kg/hr)
0.03 [0.02-0.06]
0.05 [0.04 -0.07]
Performance
Median BG
7.8 [6.6- 9.1]
6.9 [6.2-7.8]
% BG in range
51.9
79.4
% BG>10
16.3
4.0
Safety
% BG<4.0
2.1
0.1
% BG<2.6
0.1
0
# episodes with
1 (4)
0 (0)
BG <2.6
All BG units are mmol/L. Data given as median [IQR]. GA is
gestational age, PNA is post-natal age.
4. Summary
STAR-GRYPHON is a model-based glycaemic
control protocol for neonatal intensive care.
Results show it to be very safe and effective.
13th Belgian Day on Biomedical Engineering – joint meeting with IEEE EMBS Benelux Chapter
November 28, 2014