Intensive Glucose Control for Critically Ill Patients

No benefit and increases hypoglycemia

Benefits in NNT

No significant benefits

Harms in NNT

12 for severe hypoglycemia
8.3% higher risk of severe hypoglycemia
View As:


Fu Y, Sun Y, Zhang J, et al. Intensive glucose control for critically ill patients: an updated meta-analysis. Endocr Connect 2018;7:1288–98.

Study Population: 17,582 adults hospitalized in critical care settings and enrolled in 27 trials

Efficacy Endpoints

3- to 6-month mortality, short-term mortality, sepsis, new dialysis

Harm Endpoints

Severe hypoglycemia


In the past decade, emergency department (ED) to intensive care unit (ICU) admissions increased by 79% to 2.2 million admissions annually, reflecting the increasing role of emergency medicine physicians in providing care for critically ill patients.1 Optimal glucose control in critical care patients has been a topic of contention for decades. In 2001 a single-center trial of mechanically ventilated surgical patients found intensive glucose control (maintaining glucose at 80–110 mg/dL) reduced mortality compared to conventional control (180–200 mg/dL only if glucose exceeded 215).2 Subsequent studies provided conflicting data, and in 2009, the multicenter NICE-SUGAR trial, the largest trial yet, demonstrated increased mortality with intensive glucose control.3 Current American Diabetes Association (ADA) guidelines, reflecting concern about harms associated with intensive glucose control, recommend conventional glucose control with a target glucose range of 140 to 180 mg/dL for critically ill patients who experience persistent hyperglycemia.4

The meta-analysis summarized here provides an updated review of intensive glucose control effects on critical care patients.5 A total of 27 randomized trials enrolling 17,582 patients compared intensive with conventional glucose control in adult medical, surgical, and mixed critical care settings. Most had similar glucose targets. The primary outcomes were 3- to 6-month and short-term mortality (mainly within 28 days). Secondary outcomes were severe hypoglycemia (defined as serum glucose < 40 mg/dL: associated with increased mortality in multiple studies),6, 7, 8 sepsis, and need for dialysis.

There was no significant difference found in any primary outcome, and among secondary outcomes, only severe hypoglycemia in the intensive group was more common (relative risk = 4.9, 95% CI = 3.2– 7.5, NNH = 12). Notably, there was no significant difference found in any outcome between patients in medical, surgical, or mixed ICUs.


This meta-analysis is limited in several ways. There was variation, in glucose targets, type of insulin, dose and mode of administration, duration of follow-up, and concomitant therapy. Additionally, not all trials reported on all outcomes of interest, and patient-level data are not available, limiting secondary research.

The quality of evidence included in this meta-analysis is high. For most outcomes, despite the clinical heterogeneity noted above, there was little statistical heterogeneity. The only outcome with significant heterogeneity was severe hypoglycemia (I2 = 76.1%, p < 0.001), suggesting that clinical variation between studies affected this outcome.

Despite these flaws three additional, slightly less recent reviews have pooled these data as well with similar results despite differing numbers of trials, subjects, and point estimates. This consistency across author groups and approaches is reassuring.9, 10, 11

This meta-analysis also fails to address some ongoing research that has identified subgroups of patients who may stand to benefit from intensive glucose control. For example, two recent studies from the surgical ICU setting have found that among nondiabetic patients who had undergone major cardiothoracic surgery, intensive glucose control reduced morbidity.12, 13 No similar benefit was found for patients with a prior diagnosis of diabetes. Despite these interesting findings and ongoing research, conventional glucose control currently remains the standard of care in hospitalized patients.14

In summary, there was no benefit found with intensive glucose control in critical care patients but there was increased incidence of severe hypoglycemia. With no benefits and increased harms, the most appropriate color rating for intensive glucose control is black (harms > benefits). Current ADA guidelines, citing the findings of prior meta-analyses, recommend conventional glucose control with targeted blood glucose of 140 to 180 mg/dL in critically ill patients who experience persistent hyperglycemia.4

The original manuscript was published in Academic Emergency Medicine as part of the partnership between and AEM.


John Conway; Benjamin Friedman, MD
Supervising Editor: Shahriar Zehtabchi, MD


May 3, 2019