Hyperglycemia During Critical IllnessFrom the Department of Surgery, the Tufts–New England Medical Center, and the Tufts University School of Medicine, Boston, Massachusetts Correspondence: Stanley A. Nasraway, Jr, MD, FCCM, Department of Surgery, Tufts–New England Medical Center, Washington St, Box 4630, Boston, MA 02111. Electronic mail may be sent to Snasraway{at}tufts-nemc.org. Background: We sought to review the literature describing the benefits of tight glycemic control in critically ill patients, comparing outcome differences in subgroup populations. Methods: We searched PubMed for relevant literature on the topic of hyperglycemia and its management in the intensive care unit. Results: Overwhelming evidence in both surgical and medical patients conclusively demonstrates that hyperglycemia is a marker of severity of illness and is also an independent determinant of bad outcome, largely from infectious complications. Randomized trial evidence, in conjunction with historically controlled trials, supports the use of intensive insulin therapy and euglycemic control in critically ill patients, with nondiabetics possibly benefiting even more than diabetic patients. Euglycemia is best achieved, and hypoglycemia attenuated, through use of a protocolized approach. Further elaboration as to what threshold range defines euglycemia in patient subpopulations is needed and what pitfalls must be avoided in this practice. Development of continuous blood glucose monitoring has started and will someday be incorporated into routine practice in the same way that continuous electrocardiographic monitoring and pulse oximetry are standards of care in the intensive care unit. Conclusions: Hyperglycemia is a predictor of death and complications in critically ill patients. Early aggregated study results show that control of hyperglycemia improves outcomes. Well-designed studies involving thousands of patients have started to better elucidate the concomitant promoters of hyperglycemia and to better quantify the benefits from tight glycemic control. The catabolic host typically sustains stress hyperglycemia.1 The causes of stress hyperglycemia include disproportionate release into the circulation of counterregulatory hormones, including glucagon, epinephrine, cortisol, and growth hormone. These hormones promote gluconeogenesis and more than offset the increased secretion of insulin. Moreover, the release of proinflammatory cytokines during inflammation aggravates the hyperglycemic state; tumor necrosis factor, for example, promotes insulin resistance. Finally, the critical care environment further amplifies the hyperglycemic response through the use of corticosteroids, adrenergic agents, and high-dextrose nutrition support.
Hyperglycemia or relative insulin deficiency (or both) may represent a compensatory physiologic response; until recently, there have been few scientific reasons put forth for correcting the hyperglycemia often observed during critical illness. Nevertheless, despite very little exploratory investigation to precede it, van den Berghe and colleagues2 boldly embarked upon the Leuven study, a very large prospective, randomized, surgical intensive care unit (ICU), controlled trial testing the hypothesis that strict euglycemic control could increase survival and reduce morbidity in a critical care population. They studied 1548 surgical patients (62% post–cardiac surgery). The experimental group underwent intensive insulin infusions guided by frequent blood glucose measurements, with the aim of maintaining a blood glucose range between 80 and 110 mg/dL (4.4–6.1 mmol/L). The control group underwent usual care, for which measured blood glucose >215 mg/dL (1.94 mmol/L) triggered a continuous insulin infusion titrated to maintain the blood glucose concentration between 180 and 200 mg/dL (10–11.1 mmol/L). The euglycemic, intensive insulin group of patients (mean blood glucose 103 mg/dL [5.72 mmol/L]) sustained a 42% reduction in risk of death, which was attributed to a decrease in the incidence of sepsis and multiple-organ failure. Critically ill patients with a prolonged stay (ICU stay >5 days) were the primary beneficiaries of strict euglycemic control, with ICU mortality of 10.6% compared with 20.2% in the usual care group. Intensive insulin therapy also decreased morbidity, with associated decreases in bacteremia, the need for dialysis, transfusion, prolonged mechanical ventilation, and critical illness polyneuropathy. The major concern with intensive insulin in the critically ill would be hypoglycemia. Hypoglycemia was conservatively defined in the Leuven Study as blood glucose  40 mg/dL
(2.22 mmol/L); 5.1% of intensive insulin recipients sustained hypoglycemic
events vs just 0.8% of the usual care group. The lowest levels of
hypoglycemia were not described; nevertheless, no serious or irreversible
consequences of hypoglycemia were observed. These sensational observations by
van den Berghe and
associates2 are a
consequence of the use of intensive administered insulin, a widely available
and relatively inexpensive therapeutic modality. Nevertheless, conclusions
from this pioneering trial are tempered by study limitations, which include
the preponderance of cardiac surgery patients, the surprisingly low severity
of illness (median acute physiology and chronic health evaluation [APACHE] II
score 9), and the unblended nature by which the insulin was administered. Limitations aside, intensive insulin for strict euglycemic control, given its potential broad application across large numbers of heterogeneous patient groups, may very well represent the single greatest advance in critical care medicine since the discovery of fluid resuscitation and positive end expiratory pressure during the Vietnam War era >35 years ago.
The landmark Leuven study in 2001 spawned a series of efforts to reproduce the initial findings and to elucidate the pivotal mechanisms behind these observations. The first important question that was asked was whether the improvements in morbidity and mortality sustained by critically ill patients were a consequence of the nature and amount of insulin, per se, or from the beneficial effects of tight glycemic control.3 Two separate investigations published in 2003 indicated that tight glycemic control was of paramount importance.4,5 In both of these studies, patients were stratified for insulin administration; when standardizing for the amount of insulin, patient survival increased with tighter glycemic control. van den Berghe et al4 reexamined their original database from the Leuven study, this time separating patients into 3 groups according to blood glucose range: (a) BS >150 mg/dL (8.3 mmol/L); (b) BS 110–150 mg/dL (6.1–8.3 mmol/L); (c) BS <110 mg/dL (6.1 mmol/L). In stepwise fashion, the lower the blood glucose, the lower the mortality; patients in the lowest range of blood glucose sustained the greatest decreases in mortality and complications. Finney et al5 reported that insulin administration was positively correlated with mortality, regardless of the prevailing blood glucose level, suggesting that glycemic control was the responsible factor among the cohort of 523 patients predominantly undergoing cardiac surgery. This study was extremely important in that these English investigators demonstrated that although critically ill patients may receive intensive insulin therapy with the aim of maintaining blood glucose below a predetermined threshold, the practical reality is that in a given patient, blood glucose concentration tends to fluctuate quite widely throughout the day. Moreover, patients commonly spend considerable time outside the target glucose range. Zimmerman et al6 recently reported the same observation (ie, that a protocol-driven intensive insulin regimen nevertheless resulted in blood glucose concentrations outside the desired range 39% of the time).
There are scant data examining the importance of insulin administration to a heterogeneous population of hyperglycemic patients receiving medical intensive care. Two studies have been published by Krinsley,7,8 an intensivist who directs a 14-bed medical-surgical intensive care unit at a community hospital. In a retrospective analysis of 1826 consecutive patients, of which most were nonsurgical, Krinsley7 observed a linear relationship between declining mean blood glucose and hospital survival. Notably, even within the normal range for blood glucose of 80–120 mg/dL (4.44–6.67 mmol/L), those patients within the lower band of 80–99 mg/dL (4.44–5.5 mmol/L) had a lower mortality (9.6%) than those patients within the higher normal range of 100–120 mg/dL (5.5–6.67 mmol/L; 12.2% mortality). Patients with a mean blood glucose >300 mg/dL (16.7 mmol/L), in contrast, sustained a much higher mortality of 42.5%. Hyperglycemia correlates with severity of illness; therefore, to adjust for this confounder, patients were grouped according to mild, moderate, and severity of illness using APACHE II scores. Krinsley7 found survivors sustained a lower mean blood glucose concentration than did nonsurvivors at any given severity of illness. These observations are important and consistent with earlier research but require caution in light of the retrospective nature of the data, in combination with the fact that daily blood glucose measurements were random and varied widely in number from patient to patient. The same author subsequently published a study of 800 consecutive patients who were subjected to an intensive glucose management protocol; this cohort of patients was compared with a historical control group whose blood glucose concentration was not managed by protocol.8 The protocol involved intensive monitoring and treatment to maintain plasma glucose values <140 mg/dL (7.8 mmol/L). Using a protocol, mean blood glucose decreased from 152 mg/dL (8.44 mmol/L) in the control group to 131 mg/dL (7.3 mmol/L) in the protocol group (p < .001). Hospital mortality in the protocol group decreased 29%. The incidence of acute renal failure and the need for blood transfusion both decreased substantially. ICU length of stay decreased 11%. These findings reinforced those of van den Berghe2,4 and Finney,5 demonstrating applicability to medical, as well as surgical, patients. Just this year van den Berghe et al9 completed a prospective, randomized controlled study of euglycemic control in patients in a medical ICU, again demonstrating a significant mortality decrease of 18% in those with a prolonged length of stay. For all patients, on an intent-to-treat analysis, morbidity but not mortality was decreased. Benefits included decreased duration of mechanical ventilation, decreased incidence of acute renal failure, and decreased ICU length of stay.
Other studies besides that of van den Berghe and Finney5 have confirmed the utility of intensive insulin in patients undergoing cardiac surgery. Furnary et al10 conducted a retrospective analysis of intermittent insulin vs continuous insulin in 3354 diabetic patients undergoing coronary artery bypass grafting. More rigorous insulin administration via continuous infusion was associated with a 57% reduction in the likelihood of death, both from cardiac and noncardiac causes. The same investigator had previously shown that tighter glycemic control resulted in a decrease in sternal wound infections.11 Investigators at Boston University recently concluded a more ambitious study12; using a prospective randomized controlled design, diabetic patients undergoing cardiac surgery received standard therapy aimed at maintaining a blood glucose <250 mg/dL (13.9 mmol/L) or a more intensive regimen aiming for a range of 125–200 mg/dL (6.94–11.1 mmol/L). Insulin in the protocol group was initiated at the onset of surgery and continued for the first postoperative day in the ICU. Mean blood glucose was decreased from 260 mg/dL in the standard care group to 138 mg/dL (7.7 mmol/L) in the intensive therapy group. Accordingly, the protocol group of cardiac surgery patients sustained a decreased incidence of atrial fibrillation, decreased sternal wound infections, a decrease in ICU length of stay, and an increase in 2-year survival rate.12 Generally speaking, patients with cardiovascular disease seem to sustain the greatest benefits, with or without diabetes.13 Patients who undergo noncardiac surgery may also require intensive insulin therapy for euglycemia control. After head injury or generalized trauma, hyperglycemia is independently and linearly related to death and infection.14–17 Investigators at Vanderbilt17 attempted to test the hypothesis that tight glycemic control by a nurse-driven intensive insulin protocol would improve outcome from trauma. This study reported no change in the incidence of ventilator-associated pneumonia, surgical-site infection, or mortality between the 2 groups. Unfortunately, in their prospective, historically controlled study, they did not achieve enough separation in mean blood glucose between the study groups (control mean blood glucose 130 mg/dL [7.25 mmol/L], protocol mean blood glucose 124.5 mg/dL [6.9 mmol/L]; p = .168) and were not able to demonstrate any improvements in outcome with intensive insulin therapy.17 However, even this study showed that for blood glucose values in excess of 150 mg/dL (8.33 mmol/L), for even a single measurement, mortality increased 2.2 times despite stratification for severity of illness. A universal concern with intensive insulin therapy has been that of the magnitude and consequences of hypoglycemia. A number of studies have conservatively defined hypoglycemia as a value of <40–60 mg/dL [2.2–3.3 mmol/L].2,16–18 Using this standard, the incidence of hypoglycemia using a protocol-driven continuous insulin regimen is reportedly in the range from 4.0% to 6.9%. In no trial has there been any reported observation of a hypoglycemic event that was severe or irreversible to the patient. Risk factors for hypoglycemia are primarily renal failure, requiring renal replacement therapy, which prolongs the duration of action of insulin; other factors include sepsis and the need for vasoactive medications, corticosteroids, and any decrease in dextrose or nutrition administration without a concomitant decrease in the insulin infusion.18,19
Specifically why tight glycemic control leads to substantially better outcomes in critically ill patients is not known.20 Under normal conditions, insulin signaling promotes protein synthesis, glycogen synthesis, cellular glucose uptake, the preservation of cellular integrity, and facilitates cellular proliferation with an antiapoptosis effect. There are 3 major proposed mechanisms by which hyperglycemia may be toxic to the critically ill host. The first is that of cellular glucose overload. Cellular glucose overload promotes intracellular oxygen free radical formation; this, in turn, suppresses oxidative phosphorylation and increases cellular apoptosis. Recent ultrastructural evidence demonstrates mitochondrial preservation in patients receiving intensive insulin and contrasts strikingly with those patients who remained hyperglycemic.21 Hyperglycemia is also typically associated with an impaired host immune response, characterized by impairments in macrophage function and migration. Hyperglycemic patients are more susceptible to nosocomial infection, as seen in the Leuven study. Insulin has other beneficial effects on the host, ameliorating dyslipidemia, exerting anabolic effects on skeletal muscle and healing, and retarding apoptosis.20 Insulin has other interesting properties, including inducing vasodilation by promoting nitric oxide and offsetting the effects of vasoconstrictors.22 Conceivably, in this way, insulin administration may increase tissue blood flow.
In light of the potentially dramatic improvements in outcome that can be achieved with intensive insulin therapy, management of hyperglycemia should be a top priority in the care of critically ill patients. The conventional approach using insulin sliding scales has been shown to be both unsafe and ineffective.23 Repetitive doses of subcutaneous insulin may have a cumulative effect, leading to prolonged hypoglycemia.24,25 In addition, hypermetabolic, critically ill patients often have rapidly changing insulin requirements as a result of changes in clinical status, provision of nutrition, or medication regimen, which makes it difficult to achieve glycemic control using subcutaneous insulin. Alternatively, continuous IV insulin infusions are safe, effective, and timelier. Hypoglycemia that occurs during IV insulin therapy is generally short lived and more easily corrected in comparison to hypoglycemia resulting from subcutaneous insulin administration. Using an intensive insulin protocol requires frequent, often hourly, blood glucose monitoring; this practice is inherently labor intensive, placing significant demands on the bedside nurse. Since publication of the Leuven study in 2001, institutions have busied themselves with the task of creating and implementing blood glucose management protocols using an intensive insulin infusion. Protocolizing euglycemic management is a means of insuring standardized care, reducing variability, and increasing the likelihood of hitting the target blood glucose range in the earliest time possible. The benefits23,26 of the protocolized approach are:
The resulting gains come not only from best practice but also from the constancy of practice that leads to a decrease in errors, improved effectiveness, and the reduction in uncontrolled variables. As an example, the surgical ICU at Tufts–New England Medical Center has been engaged in intensive insulin therapy since 1998. When first initiated, the primary goal was simply to standardize insulin administration with an arbitrarily chosen blood glucose upper threshold of 180 mg/dL (10 mmol/L). After publication of the Leuven study, our protocol was revised in an interative fashion, with a euglycemic goal for blood glucose <135 mg/dL (7.5 mmol/L). This effort is shown in Figure 1, where it can be seen that, in 1997, average blood glucose in a busy, academic, liver transplant and general surgery ICU was approximately 230 mg/dL (12.8 mmol/L). The average blood glucose declined to <130 mg/dL (7.2 mmol/L) in 2002; by coincidence, ICU mortality decreased by approximately 30% over this time period (Figure 2) compared with 1997.
Looking to the future, research will need to ferret out implications of intensive insulin administration to critically ill patients. At the time of this writing, there are 2 extremely large prospective randomized, controlled, multicenter trials in progress; (1) NICE, Normoglycemia in Intensive Care, an intended study of 4000 patients in New Zealand and Australia; and (2) SUGAR, Survival Using Glucose Algorithm Regulation, an intended study of 5000 Canadian patients.27 Many questions remain to be answered. First of all, what is the proper blood glucose threshold that must be maintained? It may well be that different types of patients require different thresholds of blood glucose to achieve a benefit. Furthermore, what is the most effective strategic approach in sustaining low steady-state blood glucose and minimizing the daily fluctuations observed by Finney et al5 and Zimmerman et al?6 How can intensive insulin therapy be provided in the least laborious, nurse-intensive fashion? In our surgical ICU at Tufts–New England Medical Center, we currently have a glucometer at every bedside, so that nurses do not need to "compete" over a scarce supply of blood glucose measuring devices. There will have to be improvements in bedside blood glucose monitoring, including rapid input of demographics from scanning patient identification bracelets or information chips. These improvements may include wireless transmission of data to laboratory computer servers. The holy grail of device technology undoubtedly will be the ability to continuously monitor patient blood glucose concentrations, perhaps using fiberoptic or infrared technology.28 Continuous monitoring will provide multiple benefits: first, it will permit smoother, timelier adjustments in insulin infusions to more quickly achieve the blood glucose endpoint; and second, it will provide early warning to caregivers about incipient hypoglycemia. This latter concern has been a flashpoint of debate in intensive care units that are balancing strict euglycemia against safety concerns for the patient.8,17 Continuous monitoring will be pivotal not only in measuring absolute blood glucose values, but perhaps more importantly, in signaling emerging trends over time. Received for publication October 20, 2005. Accepted for publication February 2, 2006.
Journal of Parenteral and Enteral Nutrition, Vol. 30, No. 3,
254-258 (2006) This article has been cited by other articles:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Top
The Leuven Study
Are the Benefits From...
