Leuven I and II (2001 and 2006)
- Single-centre trials of intensive insulin therapy (target 4.4-6.1mmol/L) vs. conventional control
- In surgical ICU patients (Leuven I), tight control reduced mortality by 34%, as well as sepsis, critical illness polyneuropathy and renal failure
- Leuven II did not replicate these results in medical ICU patients
- Patients received early intravenous glucose and parenteral nutrition, which limits generalisability
- Intensive insulin therapy in severe sepsis did not affect organ failure or mortality
- Increased severe hypoglycaemia and serious adverse events
- Multicentre RCT of tight (4.5-6.0mmol/L) vs. conventional (<10mmol/L) control in over 6000 patients
- Tight control increased 90-day mortality (absolute increase 2.6%)
- Severe hypoglycaemia was ~13 times more common, and was itself associated with death, particularly in distributive shock
- Individualised targets based on the patient's pre-admission HbA1c vs. standard (<10mmol/L) control
- No survival benefit at 90 days, and a signal towards harm; stopped early
- Tight (4.4-6.1mmol/L) vs. liberal control in over 9000 patients, with early parenteral nutrition withheld
- No difference in duration of ICU dependency or 90-day mortality
- Suggests the benefit seen in Leuven I was linked to the hyperglycaemia caused by early parenteral feeding
Surviving Sepsis Campaign (2021)
- Start insulin when blood glucose is ≥10mmol/L (180mg/dl)
- Then target 8-10mmol/L
- Use a protocolised approach
International recommendations for glucose control in non-diabetic critically ill adults (2010)
- Aim <10mmol/L (7.8-10mmol/L acceptable)
- Avoid severe hypoglycaemia (<2.2mmol/L)
- Use arterial rather than capillary samples, analysed on a blood gas or laboratory analyser rather than a point-of-care meter
- Patients with chronic hyperglycaemia may tolerate higher glucose and be more vulnerable to 'relative hypoglycaemia', but individualised higher targets are not yet supported by trial evidence
- Capillary glucose meters are unreliable in shock, hypoperfusion, oedema and vasopressor use, and are affected by haematocrit and PO2 (see glucose and ketone meters)
- Arterial blood analysed on a blood gas analyser is preferred
- Measure 1-2 hourly while on insulin, and more frequently after changes or when values are falling
- Intravenous insulin infusion by protocol, as subcutaneous absorption is unpredictable
- Provide a carbohydrate source (feed or glucose) whenever insulin is running, and reduce or stop insulin if feeding is interrupted
- Monitor potassium, magnesium and phosphate, which fall with insulin
- Patients with type 1 diabetes must always have some insulin, even if glucose is normal
Non-insulin diabetes medicines
- Most are withheld in the critically ill:
- Metformin: risk of lactic acidosis with hypoperfusion and AKI
- Sulphonylureas: hypoglycaemia with impaired absorption and clearance
- SGLT2 inhibitors: euglycaemic DKA and volume depletion; monitor ketones
- GLP-1 receptor agonists: GI effects and unpredictable absorption; long half-lives mean effects persist after stopping
- Risk factors: tight targets, sepsis, renal and hepatic failure, interruption of feed, renal replacement therapy, insulin errors
- Signs are masked by sedation, so detection relies on regular measurement
- Treat promptly with intravenous glucose and review the cause