What, why and how to monitor blood glucose in critically ill patients

Table 1 Risk factors for developing hyperglycemia and hypoglycemia in intensive care unit patients

Risk factors for hyperglycemia	Risk factors for hypoglycemia
Release of stress hormones: Corticosteroids and catecholamines	Targeting tight glucose control with insulin infusions
Release of proinflammatory mediators	Use of bicarbonate-containing fluids
Administration of exogenous drugs: Corticosteroids, vasopressors, ascorbic acid	Interruption of nutritional support
Parenteral solutions containing dextrose	Infection, sepsis
Stress-induced hyperglycaemia	Drugs e.g. Octreotide, anti-glycaemic agents, betablockers, antibiotics (levofloxacin, quinine, trimethoprim-sulfamethoxazole)
Use of commercial dietary feeds or supplements	Use of vasopressors
	Liver failure
	Dialysis support

Table 2 Comparison between arterial and capillary monitoring of glucose

	Arterial	Capillary
Accuracy	As accurate as laboratory testing	Accuracy affected by poor perfusion states, pH, anaemia, renal failure, and high oxygen tension levels (old generation glucose oxidase based glucometers)
		Overestimation in all glucose range, especially in hypoglycaemic range
Sample volume	0.25-1 mL (can be more depends on method)	Minimal
Other variables	Simultaneous measurement of electrolytes, haemoglobin, and blood gases (partial pressure of oxygen and carbon dioxide, pH)	Single variable measured is sugar
Pain	Arterial sampling required	Repeated pin prick may cause patient discomfort
	Convenient in patients with indwelling arterial line
Need of expertise	Needs arterial line or arterial sampling which needs expertise	Simple finger stick, no expertise needed

Table 3 Advantages and disadvantages of continuous glucose monitoring

Advantages	Disadvantages
Real-time interstitial glucose	Lag time of 15 min from blood glucose, in transdermal and subcutaneous devices (Caution if levels are fluctuating rapidly)
Deviation from arterial blood glucose is less than 20%	Direct vascular sampling continuous monitoring devices are still evolving
Provides long-term day-to-week blood glucose levels	Frequent calibration (2-3 times per day)
Reduced hypoglycaemic events	Biosensors have limited life (around 7 d)
Less labour intensive	Limited glycaemic range 40-400 mg/dL
Can reduce contact of care-givers reducing cross infections and risk to care-givers	Evolving clinical evidence (especially in critically ill patients)
	Invasive device, risk of infection when using intravenous devices

Table 4 Suggested targets for various glycemic indices in critically ill patients

Glycemic indices	Suggested targets
Blood glucose	140-180 mg/dL
Time in range	More than 70%
Glycaemic gap	Less than 25.89 mg/dL in type 2 diabetics
	Less than 40 mg/dL in community acquired pneumonia
Glycaemic lability	Below median (40 mmol/L2/h/week)
Stress hyperglycaemia ratio	Less than 1.14 in sepsis patients
Mean amplitude of glycaemic excursions	Less than 65 mg/dl in sepsis patients
Coefficient of variation	Less than 36%

Table 5 Possible critical care applications of artificial intelligence in diabetes management

Potential applications	Clinical examples
Blood glucose monitoring and prediction of adverse glycaemic events	Early detection of hypoglycaemia and hyperglycaemias e.g., MD-Logic controller
Blood glucose control strategies	Software-based algorithms for insulin dosing e.g., proportional-integral-derivative models, Glucose Regulation for Intensive Care Patients, and Model predictive controls
Insulin bolus calculators and advisory systems	CGM regulated insulin infusion system predicting hypoglycaemia and regulating insulin doses
	Artificial intelligence based artificial pancreas
Risk and patient stratification	Prediction of sepsis and risk of nosocomial infections
	Risk of renal and cardiac complications like acute kidney injury and myocardial infarction
	Need for ICU admission
	ICU mortality

CGM: Continuous glucose monitoring, ICU: Intensive care unit.