Current Diabetes Reports

, Volume 12, Issue 1, pp 101–107

Glycemic Targets and Approaches to Management of the Patient with Critical Illness


  • Dieter Mesotten
    • Laboratory and Department Intensive Care Medicine, Katholieke Universiteit LeuvenCatholic University of Leuven
    • Laboratory and Department Intensive Care Medicine, Katholieke Universiteit LeuvenCatholic University of Leuven
Hospital Management of Diabetes (M Korytkowski, Section Editor)

DOI: 10.1007/s11892-011-0241-8

Cite this article as:
Mesotten, D. & Van den Berghe, G. Curr Diab Rep (2012) 12: 101. doi:10.1007/s11892-011-0241-8


Hyperglycemia during critical illness is associated with adverse outcome. The proof-of-concept Leuven studies assessed causality, and revealed that targeting strict normoglycemia (80–110 mg/dL) with insulin improved outcome compared with tolerating hyperglycemia to the renal threshold (215 mg/dL). A large multicenter trial (NICE-SUGAR [Normoglycaemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation]) found an intermediate blood glucose target (140–180 mg/dL) safer than targeting normoglycemia. Differences in design and in execution of glycemic control at the bedside may have contributed to these results. In NICE-SUGAR (Normoglycaemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation), the blood-glucose target range in the control group was lower, there were problems to reach and maintain normoglycemia in the intervention group, and inaccurate handheld blood glucose meters and variable blood sampling sites were allowed. Inaccurate tools led to insulin-dosing errors with consequently (undetected) hypoglycemia and unacceptable blood glucose variability. Also, the studies were done superimposed upon different nutritional strategies. Thus, such differences do not allow simple, evidence-based recommendations for daily practice, but an intermediate blood glucose target may be preferable while awaiting better tools to facilitate safely reaching normoglycemia.


HyperglycemiaCritical illnessNormoglycemiaNICE-SUGARGlycemic targetsStress hyperglycemiaBlood glucose management

Clinical Trial Acronyms


Impact of Early Parenteral Nutrition Completing Enteral Nutrition in Adult Critically Ill Patients


Comparing the Effects of Two Glucose Control Regimens by Insulin in Intensive Care Unit Patients


Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation


Volume Substitution and Insulin Therapy in Severe Sepsis


Hyperglycemia is part of the natural stress response of humans. By physiologic reasoning, “stress hyperglycemia” was regarded as a beneficial, adaptive response to provide energy to those organs that predominantly rely on glucose as metabolic substrate, such as the brain and the blood cells [1]. Nevertheless, this viewpoint has always conflicted with data from observational studies that relate “stress hyperglycemia” with adverse outcome in critically ill patients [2] and after myocardial infarction [36] and cardiac surgery [7]. The association between blood glucose levels and mortality risk follows a J-curved relationship with the lowest risk in the “normal for age” zone. For adults this means fasted blood glucose between 90 and 126 mg/dL (5 and 7 mmol/L) (Fig. 1). In the hyperglycemic zone there is a quasi-linear relationship between blood glucose levels and mortality risk. In critically ill patients with established diabetes mellitus the J-shaped curve is significantly blunted in the hyperglycemic zone and the nadir is shifted to higher blood glucose levels.
Fig. 1

The statistical association in observational studies between blood glucose level and risk of death follows a J-shaped curve, with normal, fasting blood levels associated with lowest risk of death. In patients with diabetes mellitus this curve is blunted and shifted toward higher blood glucose level

Neither the physiologic nor the observational reasoning can actually prove whether stress hyperglycemia is an adaptive response or is contributing to adverse outcome. Causality can only be demonstrated in a randomized controlled trial (RCT) that targets and achieves different blood glucose levels while looking at mortality as an outcome measure.

Leuven Landmark Studies

The first RCT on blood glucose management was the Leuven surgical intensive care unit (SICU) study from 2001. In this study a “strictly normal level for fasting blood glucose” (ie, 80–110 mg/dL [4.4–6.1 mmol/L]) was targeted in the intervention group, compared with the “usual care” of adult surgical ICU patients in the year 2000 [8]. “Usual care” in Leuven and in most centers worldwide [9] in those days was to tolerate hyperglycemia minimally up to 215 mg/dL (12 mmol/L), which is the renal threshold for glycosuria. In fact, the study compared the two extremes on the J-shaped curve, namely the nadir and the ultimate right side of the linear hyperglycemia line. Consequently, the mean blood glucose levels in the two randomization groups of the study were well separated and with virtually no overlap. The good separation of the blood glucose levels between the two study groups was helped by the rigorous execution of the tight glycemic control in the intervention group and by the stop of insulin administration in the control group when blood glucose levels fell below 180 mg/dL.

Blood glucose management involved frequent blood glucose measurements (interval 0.5–4 h) on whole arterial blood by an accurate blood gas analyzer. The continuous infusion of insulin was done exclusively via a central venous line with an accurate syringe pump. The insulin dose adaptations were based on a nurse-driven guideline, which stimulates intuitive and anticipating decision making. In line with the usual care in Leuven at that time, patients were kept in a nonfasting state at all times. Therefore, dextrose 20% was administered on the first day. When tolerated, enteral nutrition was started. If enteral nutrition was insufficient, early supplemental parenteral nutrition was given, resulting on average in 25 kcal/kg per day.

Maintaining strict normoglycemia by “intensive insulin therapy” lowered ICU mortality from 8.0% to 4.6% (absolute risk reduction [ARR] 3.4%) and in-hospital mortality from 10.9% to 7.2% (ARR 3.7%). It also reduced morbidity by preventing organ failure, reflected in a shorter duration of mechanical ventilation, a decreased incidence of acute kidney failure, severe infections, and critical illness polyneuropathy, and fewer blood transfusions.

The Leuven SICU study, which mainly included adult, surgical critically ill patients, was further expanded to the medical ICU (MICU) [10] and the pediatric ICU (PICU) [11]. In line with the landmark 2001 study, both trials showed that here too blood glucose control to normoglycemia decreased mortality, compared with a strategy tolerating hyperglycemia up to the renal threshold. In the PICU study normal fasting levels for age were targeted in the intervention group. For infants (<1 year) this meant 50–80 mg/dL (2.8–4.4 mmol/L) and for children between 1 and 16 years 70 and 100 mg/dL (3.9–5.6 mmol/L). In light of the low target blood glucose levels, which are close to the threshold for hypoglycemia (<40 mg/dL, 2.2 mmol/L), the same methodology for glucose management as in the Leuven SICU study was used. Only arterial blood was used for frequent blood glucose monitoring in the ABL blood gas analyzer. Diluted insulin (10 IU per 50 mL when the body weight is <15 kg) was continuously infused with an accurate syringe pump through a central venous line. Blood glucose management was done by the nursing staff utilizing a guideline instead of a strict protocol.

The Multicenter Confirmation Studies (VISEP, Glucontrol, and NICE-SUGAR)

The multicenter confirmation studies aimed to test whether tight glucose control would be generally applicable outside the well-controlled setting of the Leuven single-center studies. The VISEP (n = 537) multicenter trial was designed as a four-arm study to assess the efficacy of fluid resuscitation (10% pentastarch vs modified Ringer’s lactate) and of blood glucose control (intensive insulin therapy vs “usual care”) in patients with severe sepsis and septic shock [12]. The insulin arm of the study was stopped early after 488 patients had been included, because of the high incidence of hypoglycemia (12.1%) in the intervention group. The 90-day mortality was 39.7% in the “tight blood glucose control” versus 35.4% in the “usual care” group.

The Glucontrol RCT included 1,101 patients over 21 participating medical-surgical ICUs. However, the blood glucose target in the control group (140–180 mg/dL [7.8–10.0 mmol/L]) differed from the 180–200 mg/dL (10–11 mmol/L) target used in the Leuven studies [13]. This study was also untimely stopped because the target glycemic control was not reached and the incidence of hypoglycemia (9.8%) was deemed too high. Hospital mortality did not differ between the intervention (19.5%) and the control group (16.2%).

The NICE-SUGAR compared tight blood glucose control with the intermediate target of 140–180 mg/dL (8–10 mmol/L) in the control group [14••]. The study included 6,104 patients over 42 participating centers in Australia, New Zealand, Canada, and the United States. NICE-SUGAR revealed that tight blood glucose control increased 90-day mortality from 24.9% to 27.5% compared with the control group. Excess deaths were attributed to cardiovascular causes.

Is NICE-SUGAR Evidence against Blood Glucose Control?

With the publication of the NICE-SUGAR trial it become obvious that tight blood glucose control is not readily, generally applicable in daily clinical practice [15••, 16]. Many differences between the Leuven landmark studies and the NICE-SUGAR multicenter trial may have contributed to the opposing effects of tight blood glucose control in critically ill patients (Table 1).
Table 1

Key differences between the Leuven studies [8, 10] and NICE-SUGAR [14••] trial


Leuven SICU + MICU study


Patients (N)




2 × 1 center

42 centers

Patient sample (% of admissions)

95% (SICU)


60% (MICU)

Methodologic aspects

Control group target

180–215 mg/dL

144–180 mg/dL

10–11.9 mmol/L

8–10 mmol/L

Intervention group target

80–110 mg/dL

81–108 mg/dL

4.4–6.1 mmol/L

4.5–6.0 mmol/L

Blood sampling site

Predominantly arterial line


Glucose measurement device

Blood gas analyzer (SICU)

Not standardized, not recorded, all types allowed

1 blood glucose meter (MICU)

Insulin infusion

Continuously via central line by syringe pump

Continuous + bolus via all routes, no standardization of infusion systems

Nurse instructions

Guideline + intuitive decision making

Strict “if-then” algorithm

Nutritional strategy

Early parenteral nutrition

Late parenteral nutrition

Mean caloric intake in ICU

1100 kcal/day

880 kcal/day

Blood glucose control performance

Glycemic target reached



Overlap between study groups





x 6

x 13


Reduced organ failure and infections

No effect


Lowered by absolute 3%

Increased by absolute 3%

ICU intensive care unit; MICU medical intensive care unit; NICE-SUGAR Normoglycaemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation; SICU surgical intensive care unit

First, the NICE-SUGAR study was executed in the “flatter” part of the J-shaped glycemia-mortality risk curve (Fig. 2). The control group in the Leuven studies reflected the assumption that hyperglycemia is a beneficial adaptation. Thus, blood glucose levels were left alone unless they exceeded the renal threshold of 215 mg/dL (12 mmol/L). In the NICE-SUGAR trial the “usual care” was already affected by the Leuven studies and at the time of the study design tolerating glycemia up to 215 mg/dL (12 mmol/L) was deemed unethical [17].
Fig. 2

The NICE-SUGAR (Normoglycemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation) trial was not a true confirmation study of the Leuven surgical intensive care unit (SICU) study. Despite using similar blood glucose targets for the intervention group, the Leuven SICU and NICE-SUGAR differed in the targets for the control group. Whereas in the Leuven SICU study hyperglycemia up to the renal threshold was accepted, NICE-SUGAR, already affected by the results from the Leuven SICU study, used an intermediate blood glucose target for the control group

Second, the success in reaching and maintaining the preset target in the intervention group varied greatly between the Leuven studies and the NICE-SUGAR trial. Mean morning blood glucose levels in the Leuven studies were 105 ± 24 mg/dL (5.8 ± 1.3 mmol/L) in the intervention group versus 152 ± 32 mg/dL (8.4 ± 1.8 mmol/L) in the control group. In NICE-SUGAR the mean blood glucose level in the intervention group was already 118 ± 25 mg/dL (6.6 ± 1.3 mmol/L) versus 145 ± 26 mg/dL (8.1 ± 1.4 mmol/L) in the control group. In the Leuven studies, the mean blood glucose of 70% of the patients in the intervention group was in target [18], whereas this was less than 50% in NICE-SUGAR. This could be important, as a recent meta-analysis suggested that studies that actually managed to adequately achieve the blood glucose target showed a reduced mortality, whereas studies that did not succeed in reaching the target reported no benefit or even an increased mortality [16, 19]. Not as much expertise of the nursing team with blood glucose control and less controlled circumstances in the NICE-SUGAR trial contributed to the lower protocol compliance in the intervention group.

Third, tight blood glucose control by intensive insulin therapy requires accurate glucose measurement technology. In the Leuven SICU and PICU studies only a blood gas analyzer was used, whereas in the Leuven MICU also the Hemocue© (HemoCue® AB, Ängelholm, Sweden) handheld blood glucose meter was endorsed. In NICE-SUGAR, however, a variety of blood glucose meters was allowed and not documented [20]. The accuracy of these handheld blood glucose meters in critically ill patients has recently been questioned [21]. Many drugs, depending on the enzymatic reaction, may interfere with the blood glucose measurement (eg, ascorbic acid, acetaminophen, icodextrin) [22]. The main culprit in the accuracy of the blood glucose meters during critical illness appears to be anemia [23], which is widely tolerated in critically ill patients due to restrictive transfusion policies [24, 25]. Anemia mainly results in overestimation of glycemia, leading to overtreatment with insulin and finally inducing hypoglycemia [26]. A mathematical correction for anemia may prevent these measurement and treatment errors [27•]. The use of capillary, instead of arterial, blood for blood glucose measurements further amplifies the inaccuracy of blood glucose meters. Kanji et al. [28] elegantly showed that when using arterial blood on a blood gas analyzer for blood glucose measurements only 1% of the data fell outside the 20% allowed error zone. Contrary, when using a handheld blood glucose meter measuring capillary blood, the proportion of data points outside the allowed error zone rose to 27%. The rise in mortality rate in patients on intensive insulin therapy in the NICE-SUGAR may thus be related to the use of variable blood sampling sites and inaccurate blood glucose meters unsuitable for use in critically ill patients [29]. Simulation modeling for tight glycemic control has revealed that large dosing errors occur often when a total error of 20% is accepted for glucose meters [30]. Conversely, when the total error is kept below 10% dosing errors are virtually avoided. Moreover, titrating insulin infusions on results from such inaccurate blood glucose meters induces undetected severe hypoglycemia and unacceptably high blood glucose variability [31]. Repeated and large (undetected) fluctuations in blood glucose with hypoglycemia alternating with hyperglycemia in ill patients may also be worse than tolerating constant moderate hyperglycemia [32, 33].

Fourth, two thirds of the NICE-SUGAR patients were medical critically ill patients. Tight blood glucose control may be less effective in this patient population, in line with the results of the Leuven MICU study. Because tight blood glucose control basically prevents the development of critical illness-related complications, it predictably works best when there is no delay between onset of hyperglycemia and the start of glycemic control, such as in SICU patients. In contrast, when ICU patients already suffered from chronic illness and organ failure prior to ICU admission, and hyperglycemia was present for a longer time, the optimal window of opportunity to prevent hyperglycemia-induced complications may have been missed.

Finally, the feeding strategies differed between the studies. In NICE-SUGAR, feeding relied almost exclusively on the enteral route which evoked substantial caloric deficit during the stay in ICU, whereas in Leuven early parenteral nutrition was used to supplement insufficient enteral feeding, a strategy that prevents caloric deficit during ICU stay. The more pronounced hyperglycemia with the use of parenteral nutrition may have played a role in the beneficial impact of blood glucose control in the Leuven studies [34]. The EPaNIC RCT has recently shown that late supplementation of insufficient enteral nutritional intake does not affect lethality as long as strict normoglycemia is maintained, but accelerates recovery, compared with early administration of parenteral nutrition (Leuven studies) [35••]. Tolerating important nutritional deficit lowered the incidence of infections and cholestasis, enhanced weaning from mechanical ventilation and from renal replacement therapy, and shortened ICU and hospital stay. Early administration of nutrition may suppress autophagy, which is an essential cellular housekeeping system responsible for the clearance of damaged cell organelles, proteins, and microorganisms [36]. Hyperglycemia induces cellular damage, most specifically to the mitochondria and to functional proteins, which should be cleared by autophagy. This coping mechanism is suppressed by the administration of parenteral nutrition [37, 38]. Thus, it could be inferred that blood glucose control could perhaps be loosened to a certain extent in the absence of the early administration of parenteral nutrition, since preserved autophagy, in case of late administration of parenteral nutrition, may suffice to clear the cell damage caused by moderate hyperglycemia. This could to some extent reconcile the different outcomes of tight blood glucose control in NICE-SUGAR (late parenteral nutrition) and the Leuven studies (early parenteral nutrition). However, it remains to be investigated in a large enough clinical study that, in a setting of late initiation of parenteral nutrition and while using accurate tools for blood glucose control, tolerating moderate hyperglycemia is not inferior to maintaining normoglycemia in ICU patients.

Risk of Hypoglycemia

The Leuven studies already showed that the main drawback of tight blood glucose control is the increased incidence of hypoglycemia (SICU Leuven: 5%; MICU Leuven: 18%; and PICU Leuven: 25%). The early termination of the VISEP and Glucontrol studies brought the issue of iatrogenic hypoglycemia to the forefront of the debate on tight blood glucose control. Nevertheless, hypoglycemia has always been a major clinical concern, whether in outpatient diabetes management or during stress hyperglycemia management in hospital settings. Profound and prolonged hypoglycemia can have grave consequences and may even result in death. Clearly, severe hypoglycemia should be avoided as much as possible. Nevertheless, it is still undecided whether short-lasting, iatrogenic hypoglycemia is actively contributing to adverse outcome.

Observational data, as evidenced in the J-shaped curve, point to an association between hypoglycemia and death. This may still have two explanations. Either hypoglycemia is a marker of severity of critical illness or hypoglycemia may actually cause the increased risk of death. It is well known that particular patient populations are vulnerable to hypoglycemia [39, 40]. These include patients with liver failure, acute kidney injury requiring renal replacement therapy, diabetes mellitus, and septic shock. The most important risk factor for hypoglycemia is insulin therapy. As was recently shown by Egi et al. [41], hypoglycemia is still independently associated with mortality risk after correction of insulin therapy. When correcting for patient risk factors and duration of exposure to insulin (ICU stay), studies found that hypoglycemia could still be associated with increased mortality risk (Leuven MICU study: OR 2.4 [P < 0.001]) or it would disappear as an independent factor for mortality (Leuven SICU study: OR 1.9 [P = 0.17]; and Leuven PICU study: OR 2.1 [P = 0.32]). Nested-case control studies, also correcting for on-admission risk factors and duration of ICU stay, could not demonstrate a causal link between hypoglycemia and mortality or increased serum levels of brain injury markers [42, 43]. The latter study provided evidence that brain injury markers were already raised before the hypoglycemic event occurred, illustrating that baseline severity of illness is the chief contributor to the risk of hypoglycemia and elevated brain injury markers. Nonetheless, the lack of counterregulatory responses on hypoglycemia and the long-term neurocognitive effects of these hypoglycemic episodes have to be further explored [44].

Conclusions: Implications for Daily Practice

The NICE-SUGAR trials has demonstrated that tight blood glucose control is not yet ready to be broadly implemented in every ICU, as experience and accurate tools are a prerequisite for safety. The Leuven studies have demonstrated that targeting normoglycemia is better than tolerating hyperglycemia up to 215 mg/dL. From a practical point of view, targeting intermediate levels is safer than normal (for age) fasting blood glucose levels. Thus, blood glucose levels ought to be normalized as much as safely possible. Targeting blood glucose around 145 mg/dL (8 mmol/L) seems preferable. It is advisable to gradually tighten the glycemic management under diligent monitoring of the incidence of hypoglycemia. Irrespective of the chosen target for blood glucose, four conditions should always be met:
  1. 1)

    Frequent blood glucose measurements with one blood glucose meter methodology. Blood glucose measurements on on-site blood gas analyzers are currently the preferred devices. However, the use of a single, handheld blood glucose meter with an acceptable error range and using arterial blood may be an alternative. ICU nurses and physicians should know the limitations of these blood glucose meters to minimize measurement errors that could be harmful for the patients.

  2. 2)

    Capillary blood samples are unreliable in the ICU and should never be used.

  3. 3)

    Continuous intravenous insulin administration through a deep venous catheter using a dedicated lumen and accurate syringe pumps avoids undetectable fluctuations in insulin administration.

  4. 4)

    Thorough training of the ICU health care providers (ie, physicians and nurses) in the execution of the complex intervention blood glucose control improves patient safety.



No potential conflicts of interest relevant to this article were reported.

Copyright information

© Springer Science+Business Media, LLC 2011