Glycemic control in the intensive care unit assisted by Glycostat, a continuos central venous blood glucose measurement system with an algorithm giving advice on insulin infusion rate

Flowsion A/S (www.flowsion.dk) has developed the GlycostatTM system, intended
to continuously measure, record and track the blood glucose level (BGL) in
critically ill patients i.e., those patients under intensive-care treatment
within the operating rooms, emergency department, post-anaesthesia recovery
unit and an Intensive Care Unit (ICU). In addition Glycostat has an algorithm
which gives advice on the insulin infusion rate which -if adhered to- gives a
good glycemic control.
Hyperglycaemia, which in the ICU patient population occurs very frequently due
to an acquired stress induced insulin resistance, has long been known to be
indicative of severity and poor clinical outcome (both mortality and morbidity)
(1-4). Two large prospective randomized controlled studies have shown that not
only is hyperglycaemia indicative of poor outcome; treating these patients with
large insulin doses to bring the glucose levels down to the normal range
substantially reduces mortality and morbidity (5 and 6).
As a result of the two large studies by Greet Van Den Berghe (5 and 6) more
focus is being given to glucose control in the ICU setting. However, since the
brain is dependent solely on glucose for its energy production there is a major
concern about overdosing insulin and inducing hypoglycaemia, which may result
in brain damage and death. As a consequence the target level for blood glucose
is often increased above the normal range (i.e. 8.3 mM instead of 4.4 to 6.1
mM) (7) as a compromise between the benefits of reducing blood glucose and the
danger of hypoglycaemia.The recommendations regarding higher glucose
concentration were partly founded on the findings of the socalled NICE Sugar
study (17) in which patients were given large amounts of insulin without
frequent enough glucose concentration measurements. This resulted in increased
mortality in the tight glycemic control group compared to the higher glucose
concentration target of the control group.
Although the benefit of tight glycemic control has not been confirmed in more
recently published work, there is general agreement that glycemic control is
important and that continuous monitoring assisted by an algorithm would not
only be beneficial for patients but might answer some of the outstanding
questions within the area of tight glycemic control (8).
Regardless of the specific target range labour intensive frequent blood glucose
measurements and calculations of the needed insulin infusion rate are required
at the onset of insulin infusion therapy and until the patient is stabilized.
The most frequent manageable glucose monitoring in this early treatment window
is twice per hour. Once the patient is stabilized typically blood glucose
measurements are taken every four hours. Blood glucose is currently measured by
drawing blood samples (either arterial, capillary, or venous) from the patient
and using either a central laboratory analysis, a blood gas instrument or a
stat laboratory analyzer placed locally in the ICU, or a point-of care (POC)
glucose meter and strip (similar to the ones used by people with diabetes) to
measure the concentration of BG. Manual blood sampling and BG measurements are
usually continued 12 to 24 times per day over a several days period in order to
monitor for changes in the patient*s response to insulin therapy and to ensure
that insulin is administered in accordance with patient feeding.
Since Glycostat is able to measure BG continuously and accurate, and is able to
calculate the insulin infusion rate which will give the best achievable
glycemic control, Glycostat potentially can lower labour effort related to
glycemic control as well as potentially will lead to improved glycemic control
in the ICU.

Primary Objective:

To verify the accuracy and performance of the GlycostatTM System including its
ability to secure good glycemic control by means of its algorithm. The accuracy
of Glycostat, expressed by MARD (Mean Absolute Relative Difference), the mean
average of the relative deviation to the reference (Blood gas measurement) is
to be determined.

Secondary Objective(s):

To determine if the time in the desired glycemic range is improved by following
the insulininfusion rate advice from the Glycostat algorithm compared to a
control group in which the normal practice of the ICU is followed. The control
group data come from historical data on glycemic control of at least 40
patients extracted from EPIC the hospital`s Healthcare Information System.

3. STUDY DESIGN
The enrolled patients are all part of the study group:
- Control group: Historical patient data on glycemic control extracted from the
EPIC hospital`s Healthcare Information System, These patients have been
controlled glycemically using the normal procedure of the ICU.
- Study group: Patients enrolled will be glycemically controlled using the
Glycostat measurement values, and the insulin infusion rate of Glycostat will
be followed, as long as the ICU staff is of the opinion, that the insulin
infusion advice looks reasonable.
The aim is to carry out the study during 90 days in total. Each patient is
enrolled for 72 hours. The maximum number of blood samples needed for the study
is 40 blood samples over 72 hours.

The study is a single-site pragmatic, single-arm experimental.

All patients will receive a Glycostat intravenous probe placed in the distal
lumen of their central venous catheter (CVC) already in place after they have
been admitted to the ICU. All patients will be subject to glucose measurements
via a maximum of 40 blood gas measurements. Each blood sample will be maximum 5
ml. The Glycostat exposure will be 72 hours. The blood gas samples for the
study will be evenly distributed over the 72 hours.

The control group consists of historical patient data (blood glucose
concentration and insulin dosage) extracted from EPIC (the hospital`s
Healthcare Information System) and is needed in order to investigate if the
Glycostat algorithm is able to help control the glycemic level of the study
group in a way not inferior to the normal practice of the clinic.


8.1 Study parameters/endpoints
8.1.1 Main study parameter/endpoint

For each study subject the Glycostat probe will be in place 72 hours. In order
to determine the accuracy of Glycostat glucose measurements expressesd by MARD.
Glycostat measurement values will be compared to standard blood glucose values
from the standard ICU blood gas measurement instrument. The Glycostat signal is
proportional to the blood glucose value of the patient. In order to determine
the absolute blood glucose level of Glycostat it will be calibrated initially
after placement of the central venous probe. In order to compensate from
expected gradual biofouling of the probe Glycostat will be recalibrated once
every 24 hours at a predetermined time of day. A maximum of 40 blood gas
measurement of the patient blood glucose level will be performed per patient.
The main study parameter will be the MARD between the Glycostat measurements at
the time of blood gas measurement and the blood glucose value.

8.1.2 Secondary study parameters/endpoints (if applicable)

To determine if the time in the desired glycemic range is improved by following
the insulin infusion rate advice from the Glycostat algorithm compared to a
control group in which the normal practice of the ICU is followed. The control
group is represented with retrospective data (blood glucose concentration and
insulin dosage) from ICU patients extracted from the EPIC system.


8.1.3 Other study parameters (if applicable)
Glycostat needs information on the body mass of the subjects in order for the
algorithm to function. In adition Glycostat needs information on the insulin
infusion rate.
8.1.4 Randomisation, blinding and treatment allocation

There will be no randomization. But there are 2 study groups.
Control group: historical patient data on blood glucose concentration and
insulin dosage extracted from EPIC the hospital`s Health Information System
Control group patients have been controlled glycemically using the normal
procedure of the ICU.
Study group: All patients enrolled are part of the study group and will be
glycemically controlled using the Glycostat measurement values, and the insulin
infusion rate of Glycostat will be followed, as long as the ICU staff is of the
opinion, that the insulin infusion advice looks reasonable.


8.2 Study procedures

The procedure is described in the IFU (Instruction for Use) for Glycostat. The
startup sequence until calibration is shown on the Glycostat display. The
procedure can be summarized as follows:
The Glycostat application is started on the base unit by double clicking on the
Glycostat app icon on the touch screen dispaly.
A sound signal is heard. If the user hears it the user confirms on the touch
screen display
Glycostat then asks you to enter patient data as shown on the Patient ID
screen. Please enter patient ID by means of the touch screen keyboard according
to the procedure of your hospital
Next enter the patient*s body mass in kg and press Confirm on the touch screen
Next enter current insulin infusion rate in ml/hour and press confirm
Next it is time to install the Glycostat probe:
a. Open the Procedure Pack
b. Use the Glycostat Probe 20 or 16 cm package according to the length of the
Arrow CVC in place. Inspect the packaging of the probe and introducer prior to
opening to check integrity of the packaging.
c. If no damage to the package, unpack the Glycostat Probe 16 or 20 cm package.
Be aware that the probe and CVC introducer are provided sterile. Caution: Users
should employ aseptic techniques throughout the insertion.
d. This step shall be carried out by a physician. Insert guide wire into distal
lumen of the CVC all the way until the guide wire can be attached to the CVC
luer connection. Move the guide wire in and out a couple of times in order to
expand the CVC at the narrow places in the middle and at the distal tip.
Dispose of the guide wire.
e. Next insert the probe into distal lumen of Central Venous Catheter of the
patient by means of the CVC introducer. The probe must be introduced all the
way until the luer of the probe can be screwed onto the the luer connection of
the CVC distal lumen.
Caution: The microdialysis probe can be inserted provided that the CVC is
placed more than 2,5 cm from the atrium to avoid any risks to the patient.
f. Dispose of the CVC introducer.
g. Use the Consumable Kit package. Inspect the packaging prior to opening to
check integrity of the packaging.
h. If no damage to the package, unpack the Consumable Kit package.
i. Connect luer locks of the probe onto the satellite.
j. Place the Front-end in the satellite.
k. Close the lid of the satellite.
l. Attach satellite to the patient according to normal practice of your ICU.
m. Insert Reagent Cassette into the Base unit.
n. Close the lid.
When all tasks above have been performed: Press Confirm on the screen

Warning: Do not pull the probe when attached to the CVC
Warning: Do not use the USB port while inserting the probe.
Caution: Be aware to place tubing in such a way that the risk of entanglement
is reduced.
Caution: Control daily if leakage of reagents has occurred.

After confirmation, the system goes into Priming mode. This lasts 30-60 minutes

Next, the system informs that priming is complete. Press the button Initiate
calibration

The calibration screen step 1 is shown on the display. Take a blood gas sample
within 5 minutes to measure BGL. Press the Blood sample taken button.

Calibration screen 2 is shown on the display
Enter the blood gas glucose value in the shown units and press the Calibrate
button

Next the main screen with no BGL value is shown. Note: The blood BGL value
measured by Glycostat is shown when the delay of 12 * 20 minutes has elapsed.
Also note that due to this delay, the glucose value on the display is the value
from 12-20 minutes ago.

1. When the delay of about 12-20 minutes has elapsed the BGL value is shown.
2. The system will attempt to regulate patient BGL to a target value which is
the average of the upper and lower BGL control limits. This is obtained by
providing insulin infusion rate recommendations. The regulation system will
more aggressively drive the BGL towards the target value when the BGL is
outside the control limits.
The recommended insulin infusion rate is shown on the display if it deviates
more from the current insulin infusion rate than what is specified by
departments settings..
3. If the insulin infusion rate is adjusted, change it on the system as well by
pressing the Insulin Infusion Rate button. The screen will appear and the
insulin infusion rate can be entered. The recommended insulin infusion rate
displayed will disappear if it does not deviate more from the entered insulin
infusion rate than specified by department setting.
4. The system needs to be calibrated daily. The time for daily calibration is
specified by department setting. The count-down to calibration is shown in the
top of the screen, but the system can always be calibrated if desired.
When it is time for calibration the user will be notified and the calibration
is performed identically to the calibration during start-up.
If the system has not been calibrated in the last 26 hours the system will stop
showing BGL measurements.

5. If the patient becomes hypoglycemic or hyperglycemic, alarms will appear.
6. The consumable should be changed every third day. A count-down is visible in
the top of the screen. An alert and alarm will notify the user when it is time
to change the consumable. For change of consumable, remove the probe from the
patient and shut down the system by holding the power button depressed for ~5
seconds. When the screen is black you should disconnect the power supply from
mains. Next, remove the Cassette from the Base unit, and dispose of the
consumable according to department policy. If you want to continue using
Glycostat for this patient perform a complete start

8.3 Withdrawal of individual subjects
Subjects can leave the study at any time for any reason if they wish to do so
without any consequences. The investigator can decide to withdraw a subject
from the study for urgent medical reasons.

8.4 Replacement of individual subjects after withdrawal
New patients are to be added instead of the patients that have withdrawn from
the study to achieve the 20 patient enrollment in total.

8.5 Follow-up of subjects withdrawn from treatment
No follow up for patients withdrawn from the study foreseen.

8.6 Premature termination of the study
No premature termination of the study is foreseen.

The enrolled patients will give up to 40 blood samples @ 5 ml over the72 hours
duration of the enrolment. The associated infection risk and the blood loss are
main disadvantages to the enrolled patients.
The clinical evaluation of Glycostat has shown that the device is capable of
measuring BGL with an accuracy comparable to commonly used BGA devices. It
offers the additional benefit to patients that measurement is continuous, and
alarms are triggered as soon as a hyper- or hypoglycemic event is identified.
Thus, ICU staff can recognize and react to patient blood sugar being too high
or too low, making it more feasible for them to realize the clinical benefits
of glycemic control.
In addition to its measuring function, Glycostat calculates a recommended
insulin infusion rate to maintain patient BGL within upper and lower limits set
by clinical staff. The current state of the art for controlling blood glucose
concentration is done by looking at the latest manual blood glucose
measurements and, via either a paper-based or computerized protocol, manually
determine what insulin infusion rate should be applied. The algorithms are
typically implemented as written instructions, with calculations performed
bedside by ICU staff whenever a new glucose value is available (typically every
1-4 hours).
Computer-based algorithms aiming at providing the nursing staff insulin
infusion rate advices have become commercially available, either as standalone
software or as part of semi-automated insulin infusion devices. Studies have
shown that computerized protocols perform better in terms of keeping patient
BGL within defined limits than manual protocols.
By combining computerized insulin infusion rate advice with continuous blood
glucose monitoring, Glycostat offers patients the benefit of maintaining BGL
within defined limits, while avoiding hyper- or hypoglemic events, which are
known in the field to be associated with complications and longer ICU stays.

2.1 Summary of known and potential risks and benefits
2.1.1 Clinical Background and Medical benefit
Glycostat is intended for use in critically ill adults in Intensive Care Units
(ICUs), where it provides quantitative continuous monitoring of blood glucose
level (BGL) and insulin infusion rate advice. The Glycostat Clinical Evaluation
Report. includes a detailed summary of the clinical background for glucose
monitoring and glycemic control in this patient population.
Routine patient management in ICUs involves avoiding complications of
hyperglycemia (high blood glucose) and hypoglycemia (low blood glucose), which
are common in critically ill patients. Hyper- and hypoglycemia are associated
with a high risk for adverse clinical outcomes. Control of blood glucose in ICU
patients to maintain it within prescribed boundaries is therefore crucial to
avoid further complications.
Currently, there are several commercially available technologies to measure BGL
in the ICU: Central Laboratory Devices (CLD), Handheld Point-of-Care (POC)
devices and Blood Gas Analyzers (BGAs). CLDs represent the gold standard for
accuracy, but have a longer turnaround time for results, as they are not
located bedside. Handheld POCs and BGAs are therefore commonly used to measure
patient BGL in the ICU. Of these devices, BGAs are generally more accurate.
However, they all suffer from the drawback that glucose measurement is
intermittent, not continuous, and increased measurement frequency results in
increased nurse workload. Hyper- or hypoglycemic events that occur between
measurement points may not be recognized.
The advantage of continuous BGL measurement is that both hypo- and
hyperglycemia events may not be missed and the nurse workload may be reduced.
There are also several different commercially available systems for Continuous
Glucose Monitoring (CGM), applying various technologies ranging from more
invasive systems placed inside a blood vessel to less invasive subcutaneous
systems. Subcutaneous systems have been tested in clinical studies with varying
results. Their accuracy has been shown to be poor for patients with
microcirculatory impairment.
Intravascular microdialysis, the method applied by the Glycostat device allows
continuously blood glucose measurements without blood sampling. The method was
first experimentally described in 1996 and has been studied as part of the
Eirus medical device (Maquet Critical Care, Solna, Sweden). This method is more
invasive, as it requires a probe to be placed inside a blood vessel. Glycostat
and Eirus both utilize the central venous catheter (CVC) which is routinely
placed in intensive care patients.
The clinical evaluation of Glycostat has shown that the device is capable of
measuring BGL with an accuracy comparable to commonly used BGA devices. It
offers the additional benefit to patients that measurement is continuous, and
alarms are triggered as soon as a hyper- or hypoglycemic event is identified.
Thus, ICU staff can recognize and react to patient blood sugar being too high
or too low, making it more feasible for them to realize the clinical benefits
of glycemic control.
In addition to its measuring function, Glycostat calculates a recommended
insulin infusion rate to maintain patient BGL within upper and lower limits set
by clinical staff. The current state of the art for controlling blood glucose
concentration is done by looking at the latest manual blood glucose
measurements and, via either a paper-based or computerized protocol, manually
determine what insulin infusion rate should be applied. The algorithms are
typically implemented as written instructions, with calculations performed
bedside by ICU staff whenever a new glucose value is available (typically every
1-4 hours).
Computer-based algorithms aiming at providing the nursing staff insulin
infusion rate advices have become commercially available, either as standalone
software or as part of semi-automated insulin infusion devices. Studies have
shown that computerized protocols perform better in terms of keeping patient
BGL within defined limits than manual protocols.
By combining computerized insulin infusion rate advice with continuous blood
glucose monitoring, Glycostat offers patients the benefit of maintaining BGL
within defined limits, while avoiding hyper- or hypoglemic events, which are
known in the field to be associated with complications and longer ICU stays.
2.1.2 Justification and/or risk benefit analysis for individual risks
During risk assessment it was possible to implement risk control measures that
lowered the risk of harm to an acceptable level as defined by Flowsion.
However, for the following hazards this was not possible. In the following
table justification for proceeding with the design and production of Glycostat
despite residual risks being in the *medium* category is provided.
HAZARD# DESCRIPTION RESIDUAL RISK (RPN2) JUSTIFICATION
RA-42 Insufficient cleaning and bad hygiene can cause cross contamination and
thereby infection of patients and operators. 6 (medium) This is not different
from any other equipment used in the ICU environment. There is no indication
that Glycostat causes greater risk than any other equipment introduced via a
CVC.
RA-60 Wrong and too high glucose value entered during calibration. This can
cause undetected severe hypoglycemia during BGL monitoring. 8 (medium) Two risk
control measures are implemented to reduce risk as far as possible. Benefit of
device exceeds residual risk.
RA-157 Wrong and too low glucose value entered during calibration. This can
cause undetected severe hyperglycemia during BGL monitoring. 8 (medium) Two
risk control measures are implemented to reduce risk as far as possible.
Benefit of device exceeds residual risk.
RA-176 Patient has very high BGL and the operator increase the insulin infusion
rate beyond the recommendation provided by Glycostat. 8 (medium) Device does
not control the insulin infusion rate pump and therefore cannot prevent the
operator to make wrong decisions.
RA-151 Patient/staff trips over the wires and tubing from base unit to
satellite resulting in bruises/fractures/dislocations. 6 (medium) Hazard it is
not limited to the Glycostat device. Patients/staff will generally move
carefully around the patient in ICU settings.
RA-152 Patient/staff trips over the power adapter cable resulting in
bruises/fractures/dislocations. 6 (medium) Hazard it is not limited to the
Glycostat device. Patients/staff will in generally move carefully around the
patient in ICU settings.
RA-182 Operator does not notice visual alarm. 8 (medium) Risk control measure
ensures that alarms are both visual and auditory. The alarm will only be
overheard if department is understaffed or attending more important situations.
URA-21 Distractions in the ICU result in the user not noticing alarm. 8
(medium) Risk control measure ensures that alarms are both visual and auditory.
The alarm will only be overheard if department is understaffed or attending
more important situations.
2.2 Acceptability of overall residual risk
The residual risk associated with the Glycostat device has been reduced as much
as possible. Some risks remain that cannot be further reduced. As described in
section 6.4.2 above, some of these are inherent to devices in the ICU and would
thus also be present if another device than Glycostat was used.
Residual risks in the low category will not result in any significant health
issues or the probability of these are only theoretical and unlikely to occur
in the product lifetime. Post Market Surveillance will provide input to the
iterative process of risk management. In this connection the probability of
occurrence might be reevaluated.
The clinical evaluation of Glycostat did not identify any side effects
associated with the clinical use of the device or equivalent devices. It also
did not identify any hazards that have not been considered in the risk
assessment and reduced as far as possible through the risk management process.
The overall residual risk is outweighed by the Medical Benefit. For this
reason, the Overall Residual Risk is acceptable, and there is no single risk
that precludes the release or the continued production of the device.
Information on residual risks has been disclosed to the user in the patient
consent form.