SPECIAL ARTICLE
Society of Critical Care Medicine Guidelines on Glycemic Control for Critically Ill Children and Adults 2024: Executive Summary
Honarmand, Kimia; Sirimaturos, Michael; Hirshberg, Eliotte L.; et al
Critical Care Medicine 52(4):p 649-655, April 2024.
KEY RECOMMENDATIONS
Adult Target
Question:
Should insulin infusion therapy be titrated to achieve INT glucose levels, 4.4–7.7 mmol/L (80–139 mg/dL) or CONV glucose levels, 7.8–11.1 mmol/L (140–200 mg/dL) for unselected (mixed) critically ill adults or any patient subgroups?
Good Practice Statement:
Clinicians should use glycemic management protocols and procedures that demonstrate a low risk of hypoglycemia among critically ill adults and should treat hypoglycemia without delay.
Recommendation:
Based on available randomized controlled trial (RCT) data, in critically ill adults, we suggest against titrating an insulin infusion to a lower blood glucose (BG) target INT, 4.4–7.7 mmol/L (80–139 mg/dL) as compared with a higher BG target range, CONV 7.8–11.1 mmol/L (140–200 mg/dL) to reduce the risk of hypoglycemia (Conditional recommendation; moderate certainty of evidence).
Comments
- Analysis of data from neurologic or cardiac surgery ICUs yielded comparable findings and these patients should be managed like unselected patients.
- For other specific subsets of critically ill patients (e.g., cardiac, medical, surgical, trauma, etc.) data were inadequate to perform subgroup analyses and thus patients should be managed like unselected patients.
- For the subset of patients with preexisting diabetes mellitus (DM) or preadmission hyperglycemia there is insufficient evidence from RCTs to make a recommendation regarding personalized targets for glycemic control.
Research Statement:
Observational data suggest a potential benefit of personalized glucose targets that more closely match chronic prehospital glycemic control. We recommend high-quality interventional trials of individualized glycemic targets in critically ill adults, stratified by prior glycemic control (such as indicated by glycosylated hemoglobin).
Rationale
Clinical benefits of INT have not been consistently demonstrated in the RCTs included in our meta-analysis; specifically no effect is shown on mortality among mixed populations of ICU patients. However, INT targets were associated with increased frequency of severe hypoglycemia, less than 2.2 mmol/L (40 mg/dL) compared with CONV targets, although there was a reduced infection risk, and lower ICU length of stay (LOS) with INT vs. CONV targets (5–42). In neurologic and cardiac surgery subsets, INT targets were associated with increased risk of severe hypoglycemia and although the cardiac surgery subset had a lower ICU mortality and lower critical illness polyneuropathy (both from a single clinical trial) there were no other outcome benefits (hospital mortality, any infection) (5,25–31,40–44).
A large RCT of insulin infusion targeting tight glucose control without early parenteral nutrition (TGC-Fast) comparing insulin titrated to INT vs. a higher target than the CONV range in this guideline, 10–11.9 mmol/L (180–215 mg/dL) was published after our literature review but similarly found no difference in outcomes (time to discharge alive from ICU or 90-d mortality) despite low rates of hypoglycemia in both groups (45). As a result, the upper limit for a glycemic target with insulin infusion is not well defined with current literature. Further, it appears that lower targets may be acceptable for selected patients if the risk of hypoglycemia is documented to be negligible when using a safe and effective protocol. Although observational data suggest a potential role for personalized glucose targets relative to a history of DM, the TGC-Fast trial showed no benefit of INT targets despite 80% of the patients having no history of DM (45–53). The panel recommends prospective randomized clinical trials using individualized targets for insulin titration, which will inform the need to revise this recommendation in the future.
Pediatric Target
Question:
Should insulin therapy be titrated to achieve INT glucose levels, 4.4–7.7 mmol/L (80–139 mg/dL) or CONV glucose levels, 7.8–11.1 mmol/L (140–200 mg/dL) for unselected (mixed) critically ill children?
Good Practice Statement:
Clinicians should use glycemic management protocols and procedures that demonstrate a low risk of hypoglycemia among critically ill children and should treat hypoglycemia without delay.
Recommendation:
We recommend against INT BG control, 4.4–7.7 mmol/L (80–139 mg/dL) as compared with CONV BG control, 7.8–11.1 mmol/L (140–200 mg/dL) in critically ill children (defined by the pediatric panel as ≥ 42 wk adjusted gestational age) (strong recommendation, moderate certainty evidence).
Rationale
INT targets were associated with increased frequency of severe hypoglycemia (< 2.2 mmol/L [40 mg/dL]), shorter ICU LOS, but no effect on mortality or neurocognitive outcomes among mixed ICU and postcardiac surgery patients (54–62). The high risk of severe hypoglycemia outweighs the trivial clinical benefits of INT glucose control among critically ill children. The impact of hypoglycemia on cognitive development is a special consideration in children. While RCT data were prioritized for this guideline, observational data suggest poorer cognitive performance among children with moderate or severe hypoglycemia events, lending additional importance to hypoglycemia avoidance (54,57,63,64). Like the adult population, the panel recommends prospective randomized clinical trials using individualized targets based on preexisting glycemic control to inform future practice changes.
Question:
In critically ill adults and children on insulin infusion therapy, should a protocol that includes explicit decision support tools be used compared with conventional protocols for the management of hyperglycemia?
Recommendations
We suggest use of a protocol that includes explicit decision support tools (tools) over a protocol with no such tools in critically ill adults receiving IV insulin infusions for the management of hyperglycemia (conditional recommendation, moderate certainty evidence).
We suggest use of explicit decision support tools over no such tools in critically ill pediatric patients receiving IV insulin infusions for the management of hyperglycemia (conditional recommendation; very low certainty evidence).
Rationale
We defined those elements of explicit clinical decision support tools that were critical components of acceptable protocols, preferably with computerized support and interoperability of the tool with the electronic health record. While patient outcomes were prioritized for this guideline, the panel acknowledges that insulin titration protocols add to bedside caregiver cognitive burden and workload and could be minimized with a well-designed explicit decision support tool that directs treatment (65,66). Protocols incorporating these tools were associated with reduced frequency of moderate hypoglycemia less than 3.3 mmol/L (60 mg/dL) and greater proportion of BG values within the target range (45,50,67–76). There were no effects on other critical outcomes such as hospital mortality or ICU LOS (moderate certainty), ICU mortality or quality of life at 90 days (low certainty). The TGC-Fast trial of INT vs. a glucose target of 10–11.9 mmol/L (180–215 mg/dL) used a computer algorithm integrated into the electronic health record with alerts to guide insulin dosing and monitoring intervals of 1–4 hours (45). With these components, a low rate of hypoglycemia was reported in this multicenter trial of adults in both INT and higher target groups. While most other studies evaluated adult protocols it was determined that the processes of glycemic management are comparable between adults and children, leading to comparable statements and endorsement of the need for high-quality interventional trials in both age groups.
CONCLUSIONS
Guidelines are limited by the quality of published data in RCTs and additional research on various aspects of glycemic control is needed. Key guideline statements are summarized in this executive summary but there is significant additional detail in the full document regarding hyperglycemic triggers, route of insulin administration, frequency of glucose monitoring, and monitoring devices (3). Clinicians should also examine the complete explanation of rationale and evidence to recommendation discussions to gain insight into strengths and weaknesses of existing data when considering how to incorporate guidelines into clinical practice.
