Glycemic Control in the Critically Ill

Daniel Runde and Jarone Lee

BACKGROUND

Glycemic control is one of the most controversial topics in critical care. A large body of evidence demonstrates a clear association between elevated blood glucose levels and increased morbidity and mortality.¹^–⁴ Persistent hyperglycemia also correlates with poor outcomes in all subtypes of critically ill patients: postoperative, myocardial infarction, ischemic and hemorrhagic stroke, neurologic trauma, and sepsis.⁵^–¹⁰ Pathophysiologic changes caused by hyperglycemia affect a wide variety of biologic responses, from immune function to wound healing. Despite extensive research, however, it is yet to be determined whether hyperglycemia is a marker of disease severity; whether it directly causes poor outcomes; and whether interventions to control and regulate blood glucose levels result in improved outcomes in critically ill patients. The controversy surrounding glycemic control is driven in large part by the fact that, in contrast to hyperglycemia, even transient hypoglycemic episodes, which commonly occur in the setting of strict blood glucose control, can have potentially disastrous consequences for the critically ill patient.¹¹

PATHOPHYSIOLOGY OF HYPERGLYCEMIC STATES

Stress hyperglycemia is common in the acutely ill and in both diabetic and nondiabetic patients. Hyperglycemia occurs as a response to a range of events that result in physiologic stress, including, but not limited to, trauma, hemorrhage, hypoxia, myocardial ischemia, and infection.¹²^–¹⁶ This stress response is mediated by a complex interaction of hormones and proinflammatory cytokines, including a dramatic increase in the production of cortisol, epinephrine, and norepinephrine, as well as tumor necrosis factor-alpha and interleukins 1 and 6. The hypothalamic–pituitary–adrenal axis plays a key role, as does the sympathoadrenal system.¹⁷^–¹⁹ Metabolically, these changes produce increases in gluconeogenesis, glycogenolysis, and insulin resistance.²^,²⁰

Short-term hyperglycemia may be an adaptive response to stress, resulting in improved glucose delivery to tissue that is poorly perfused at the microvascular level.²¹ Macrophages, which play a key role in immune response, rely on glucose as a means of energy production, and adequate glucose delivery is necessary to ensure optimal function.²²^,²³ Furthermore, laboratory, animal, and human studies have demonstrated that hyperglycemia may initially limit ischemic myocardial injury.²⁴^,²⁵

In contrast, chronic hyperglycemia appears to be associated with a variety of negative effects at the cellular level. In vitro models have found that hyperglycemia inhibits glucose-6-phosphate dehydrogenase activity, which in turn decreases oxygen radical production by neutrophils.²⁶^,²⁷ It has been theorized that this could result in impaired bactericidal activity and immune function.²⁸ Chronic hyperglycemia is also associated with increased myocardial cell death in the setting of cardiac ischemia.²⁹

HYPERGLYCEMIA AND CLINICAL OUTCOMES

There is ample evidence that hyperglycemia is associated with poor clinical outcomes in a wide variety of patients who present to the emergency department. Among these are patients with acute coronary syndrome, neurologic injuries, and sepsis.

Acute Coronary Syndrome

In patients with acute myocardial infarction (AMI), several studies suggest that elevated serum glucose on admission is associated with increased risk of reinfarction, the development of congestive heart failure, the incidence of future cardiac events, and increased mortality.³⁰^,³¹ A related study tracked long-term outcomes in patients with AMI (either known diabetics or those with elevated serum glucose at the time of their event) and demonstrated a significant decrease in mortality in the group with more strict blood sugar regulation.³²

Neurologic Injury

In several studies of patients presenting with ischemic strokes, hyperglycemia on admission was independently associated with worse long-term neurologic outcomes and increased mortality.⁷^–¹⁰^,³³^,³⁴ Hyperglycemia is also associated with an increased rate of hemorrhagic transformation in patients receiving thrombolytic therapy.³⁵ Again, these findings were independent of the patient’s diabetic status at the time of the event.

Similarly, patients with traumatic brain injury with elevated blood glucose on admission experience worse neurologic outcomes and increased mortality, both in the short and in the long term. In one study of severely brain-injured patients, the degree of hyperglycemia was inversely proportional to Glasgow Coma Scale score and to favorable outcome.⁹

Sepsis

In patients with sepsis, even moderate hyperglycemia is associated with increased rates of complication, length of stay, and unfavorable clinical outcomes.³^,¹⁶^,¹⁹^,³⁶^–³⁸

PATHOPHYSIOLOGY OF HYPOGLYCEMIC STATES

In contrast to hyperglycemia, whose negative effects occur over hours to days, even transient hypoglycemia can result in profound morbidity and potential mortality in the critically ill patient. The brain, in particular, is dependent on a near-continuous supply of glucose, and any interruption or decrease in glucose delivery can result in impaired judgment, confusion, seizures, coma, and even death. While there are no definitive cutoffs for the degree or duration of hypoglycemia required to produce permanent neurologic damage, the correlation between the two is clear.³⁹^,⁴⁰ The heart is similarly sensitive to hypoglycemia. One of the initial adaptive responses to hypoglycemia is an increase in heart rate, myocardial contractility, and stroke volume, with the end result being a dramatic increase in cardiac workload over a short period of time. This increase, while of little concern to a healthy patient, can produce demand ischemia in critically ill patients with coronary artery disease. In addition, hypoglycemia can result in cardiac conduction abnormalities, specifically a prolonged QT interval, and an increase in myocyte repolarization time. These changes are associated with an increased risk of arrhythmias, including atrial fibrillation and ventricular tachycardias.⁴¹

HYPOGLYCEMIA AND CLINICAL OUTCOMES

There is a convincing body of evidence, from both observational studies and randomized control trials, that even a moderate degree of hypoglycemia is associated with worse clinical outcomes and increased mortality. These findings are seen in both medical and surgical patients, and they appear to be independent of patients’ underlying pathology or reason for admission.⁴²^–⁴⁵ In a recent retrospective, case–control study examining the effects of mild hypoglycemia on medical ICU patients, even a single episode of mild hypoglycemia independently predicted increased mortality (OR 2.98).⁴⁶

CONTROVERSY REGARDING GLYCEMIC CONTROL IN THE CRITICALLY ILL

Given the observed relationship between hyperglycemia and poor clinical outcomes, the notion that strict control of a patient’s glycemic state would result in improved outcomes has been actively investigated for more than 20 years. Early studies were often limited to patients with diabetes undergoing the same treatment or procedure (e.g., therapy for acute MI or cardiac surgery). The DIGAMI study included patients with hyperglycemia, regardless of diabetic status, who presented with acute MI.³² This study demonstrated that patients who were randomized to receive intravenous insulin infusions during their inpatient stay and subsequently received 3 months of subcutaneous outpatient insulin therapy had dramatically improved survival at 1 year follow-up (7.5% ARR, NNT = 13 for survival). It should be noted, however, that there were no significant differences in mortality during the in-hospital period, or at 3-month follow-up, and it remains unclear the extent to which inpatient glycemic control affected outcomes. Though it did not involve glycemic targets, the CREATE-ECLA trial enrolled 20,000 subjects in a multicenter investigation of the effect of insulin and glucose infusion on patients with acute MI; it found no changes in mortality, cardiac arrest, recurrent MI, or cardiogenic shock.⁴⁷ A similar trial randomized nearly 2,500 patients undergoing cardiac surgery to receive either continuous insulin infusion or intermittent subcutaneous insulin injections for glycemic control. While the authors did not find any mortality benefit in the intervention group, they did note a small, but statistically significant decrease in deep sternal wound infections in the intervention group (0.8 vs. 2.0%).⁴⁸

In 2001, the Lueven intensive insulin therapy trial enrolled 1,548 critically ill surgical ICU patients and reported a 42% relative reduction in mortality (3.4% ARR, NNT = 30 for survival) and a 46% reduction in septicemia (3.6% ARR, NNT = 28 for preventing bloodstream infection) for patients targeted to tight glycemic control (70 to 110 mg/dL).⁴⁹ Following the study’s publication, strict glycemic control was rapidly adopted as the standard of care in many ICUs worldwide. In 2006, the same research group enrolled 1,200 subjects and examined the effects of tight glycemic control in critically patients in the medical ICU setting.⁵⁰ Unlike their previous trial, this investigation did not find any overall difference in mortality between the intervention and control groups, nor did it find any difference in the rate of bloodstream infections. In a subgroup analysis, the authors noted that for patients with ICU stays <3 days, there was decreased mortality in the intervention group. Unfortunately, this benefit was offset by an increase in deaths among intervention patients with ICU stays longer than 3 days.

There have been multiple subsequent attempts to reproduce the results of the original 2001 Leuven trial. Ensuing investigations failed to demonstrate any mortality benefit for patients receiving tight glycemic control. The majority of these trials did, however, again demonstrate intensive insulin therapy to be associated with an increase in hypoglycemic events, with at least one trial stopped early because of an increase in mortality in the intervention group.⁵¹^–⁵⁹ High-quality systematic reviews have likewise failed to demonstrate a benefit for tight glycemic control in more defined populations, such as perioperative patients with diabetes or in patients following ischemic stroke. These reviews also reinforced the finding of increased hypoglycemic events in intervention groups.³⁸^,⁶⁰^–⁶² As a result of these studies, there has been a trend away from tight glycemic control in the treatment of the critically ill.²⁴^,⁶³^,⁶⁴

Follow-up studies investigating the effect of tight glycemic control in critically ill patients culminated in the 2009 NICE-SUGAR study, a landmark multicenter trial that randomized over 6,000 medical and surgical ICU subjects expected to require ICU care for 3 or more days to either intensive (goal 80 to 110 mg/dL) or conventional (goal <180 mg/dL) glycemic control.⁴⁵ In stark contrast to the results of the Lueven trial and in agreement with the smaller studies discussed above, all-cause mortality at 90 days (the primary endpoint) was actually higher in the intensive glucose control group (27.5% vs. 24.9%, NNH = 38 for death). The rate of severe hypoglycemic episodes was also found to be dramatically increased in the intervention group (6.9% vs. 0.5%, NNH = 15).⁶⁵ Importantly, there was no difference in outcomes for medical or surgical patients and no difference in secondary outcomes (ICU/hospital days, days requiring mechanical ventilation, need for renal replacement therapy). As a result of this study, aggressive glucose control in the ICU to goal <110 mg/dL is definitively no longer recommended. Although an ideal target serum glucose remains unclear and prolonged hyperglycemia remains, in general, undesirable, the risks associated with hypoglycemia have led most centers to target 140 to 180 mg/dL for patients in the critically care setting.

CONCLUSION

The majority of data on glycemic control in the critically ill are derived from studies performed in an ICU setting. At this time, there are little data on glycemic control among critically ill patients in the ED.⁶⁶^,⁶⁷ Extrapolating current evidence from the ICU-based studies, targeting a blood glucose target of 140 to 180 mg/dL is recommended.

Many emergency departments, however, lack the staffing, protocols, and resources to safely ensure even this level of glycemic control. Data clearly suggest hypoglycemic events to be of greater danger to the patient than modest increases in blood glucose, and insulin continues to be a top five “high-risk” medications, with one in three fatal medical errors being linked to insulin therapy.⁶⁸ As such, in emergency departments in which resources are limited, targeting a more liberal blood glucose target of 160 to 220 mg/dL may be appropriate.

In summary, while an abundance of data demonstrates a clear association between hyperglycemia and worsened outcomes for critically ill patients of all types (sepsis, acute MI, ischemic and hemorrhagic stroke, postsurgical), it is unclear whether this association is causal. It remains a distinct possibility that hyperglycemia is in fact an independent marker of disease severity—similar to lactate in severe sepsis—rather than a cause of increased morbidity and mortality.⁶⁹ Causality notwithstanding, it appears that hyperglycemia is associated with worsened outcomes when present on the scale of hours to days. In contrast, even very brief episodes of hypoglycemia may be catastrophic for a critically ill patient. Despite the development of well-defined protocols, the introduction of continuous serum glucose monitoring devices, and highly trained and attentive ICU staff, modern medicine is still unable to adequately anticipate or avoid hypoglycemic events in patients targeted to tight—or even moderate—glycemic control. Acknowledging these limitations, as well as the near-zero tolerance for harm from iatrogenic hypoglycemia, helps justify the current trend away from tight glycemic control despite the known association between hyperglycemia and poor clinical outcomes.

LITERATURE TABLE