Relationship Between Hyperglycemia and Infection in Critically Ill Patients

Simona O. Butler, PharmD; Imad F. Btaiche, PharmD; Cesar Alaniz, PharmD


Pharmacotherapy. 2005;25(7):963-976. 

In This Article

Pathogenesis of Hyperglycemia-Associated Infection

A rise in the risk of postoperative infectious complications results from depressed immune function secondary to inhibition of IL-1 release from macrophages, impaired phagocytosis, and diminished production of oxygen radicals from neutrophils.[21,35] As a result, studies addressed the effect of transient acute hyperglycemia in a nondiabetic rat model on immune function after surgery[36] or splenectomy.[37] The authors of the first study evaluated the function of peripheral neutrophils, peritoneal macrophages, and alveolar macrophages in postoperative nondiabetic rats in two groups: normoglycemic (control) and hyperglycemic.[36] Their results showed significant change in immune function based on blood glucose concentration.

The second study focused on demonstrating decreased immunologic activity in rats after splenectomy and induction of bacteremia in the setting of hyperglycemia.[37] Survival times were markedly lower in the hyperglycemic group. Consequently, the authors concluded that short-term exposure to high glucose concentration has detrimental effects to the immune function of the studied cells and could be one of the primary factors contributing to increased susceptibility to infection in this population. Furthermore, the effect of blood glucose control with exogenous insulin was determined in vivo in a study of rabbits with severe injury.[27] Results revealed improvement in phagocytosis and oxidative killing with prevention of weight loss, hyponatremia, lactic acidosis, and excessive inflammation in the group maintained at normoglycemia (60-110 mg/dl). Similar findings came from an in vitro study that showed significant reduction in polymorphonuclear leukocyte phagocytic activity of normal cells in intermittent hyperglycemia compared with control.[38]

Hyperglycemia and Infection in Patients With Diabetes

Blood glucose control was originally investigated in the setting of diabetes because diabetic patients seemed to be predisposed to various infections (e.g., cystitis, cellulitis, postoperative wound infections[10]) and infectious complications as a result of having diabetes.[39] Although the exact pathogenesis was unknown, it was speculated that the increased risk of infection was related to decreased cellular immune function.[10] Later studies (both in animals and humans) demonstrated that the depressed ability to fight off infection was a result of impaired polymorphonuclear leukocyte function, chemotaxis, and phagocytosis.[31,39,40,41,42,43]

Studies in diabetic animals examined the phagocytic activity[44] and membrane fluidity[45] of certain immune cell types. Results showed impaired intracellular killing[44] and higher susceptibility to infection although decreased fluidity[45] in animals with poorly controlled diabetes compared with well-controlled ones. In vitro studies in diabetic patients further strengthened the evidence of cellular immune function impairment by demonstrating a significant reduction in polymorphonuclear leukocyte mobilization, chemotaxis, and bactericidal and phagocytic activity in poorly controlled[42,46] diabetic cells compared with well-controlled ones[42,46] and diabetic[47] cells in general compared with control.[47] Data derived from animal and in vitro trials led to further investigation of the link between hyperglycemia and infectious complications in patients with diabetes.

In 1991, one group of authors performed a retrospective chart review to compare operative mortality and complications (e.g., wound infections, myocardial infarction, arrhythmias, respiratory failure) in diabetic versus nondiabetic patients after coronary artery bypass graft (CABG).[7] Their analysis included 711 patients, 79% nondiabetic and 21% diabetic. The results did not show a difference between the two groups with respect to mortality but demonstrated a significantly increased morbidity in the diabetic group especially in the wound infection parameter.

Another group looked specifically at the rate of deep sternal wound infections in diabetic patients after cardiac surgery in a before and after retrospective chart review.[8] A strict blood glucose control protocol was implemented, with a target of blood glucose concentration less than 200 mg/dl. A total of 999 patients had surgery before the protocol was implemented, and 595 had surgery after. The results revealed a significant reduction in deep wound infection after protocol implementation, with elevated blood glucose level being directly related to infection rate. Later, these authors further demonstrated that the use of perioperative continuous intravenous infusion in this patient population significantly reduced morbidity from deep sternal wound infections and the costs associated with it.[11]

An increased rate of nosocomial infection in diabetic patients undergoing elective surgery was demonstrated in 1998. This was a noninterventional study with the objective to determine the relationship between perioperative glucose control (good blood glucose level was defined as < 220 mg/dl) and postoperative infection rate.[9] Analysis of 93 patients and their serum glucose concentrations on postoperative day 1 indicated that the nosocomial infection rate was 2.7 times higher in diabetic patients with a blood glucose level above 220 mg/dl and 5.7 times higher when urinary tract infections were eliminated compared with patients with better glucose control. Thus, these authors concluded that a blood glucose level above 220 mg/dl was a sensitive predictor (87.5%) of nosocomial infection.

Finally, in 1999, another group performed a nonconcurrent, prospective, cohort, blinded study with the purpose of assessing the independent relationship between perioperative glycemic control and subsequent risk of infectious complications (e.g., pneumonia, urinary tract infection, wound infection) on postoperative day 2 or later in diabetic patients undergoing coronary artery surgery.[10] Their results echoed those of the previous studies and showed an increased infection rate in patients with a perioperative blood glucose level higher than 200 mg/dl. There was no relationship between peak perioperative blood glucose level and subsequent risk of infectious complications. Table 1 summarizes the studies relating hyperglycemia with infection and mortality in the surgical and critically ill patient population.

Hyperglycemia and Infection in Critically Ill Patients

As evidence accumulated in support of tight glycemic control in patients with diabetes, there still remained significant controversy regarding the pathophysiology and management of stress-induced hyperglycemia. Increased glucose turnover and insulin resistance were regarded as potentially appropriate responses during critical illness.[22] Also, although the association between hyperglycemia and infection had been well established,[31] debate continued regarding which condition precipitated the other.[51] However, data are now emerging that provide important information on the consequences and management of hyperglycemia in critically ill patients.[20]

In 2001, a group of authors published the results of a large, prospective, randomized, controlled trial that was the first to challenge the accepted clinical practice of loose glycemic control in critically ill patients.[20] During a 1-year period, 1548 patients with mechanical ventilation who were admitted to an intensive care unit (ICU) after surgery or trauma were randomly assigned to either intensive insulin or conventional therapy. The conventional therapy group was targeted to maintain a blood glucose level of 180-200 mg/dl with insulin therapy to begin at a blood glucose level higher than 215 mg/dl, with the insulin dose adjusted by the nursing staff. The intensive insulin therapy group was targeted to maintain a blood glucose level at 80-110 mg/dl with insulin therapy to begin at a blood glucose level higher than 110 mg/dl following a strict algorithm.

Hypoglycemia was defined as a blood glucose level less than 40 mg/dl. After discharge from the ICU, both groups were switched to subcutaneous insulin with a target blood glucose level of less than 200 mg/dl. The primary outcome was all-cause ICU mortality and secondary outcomes included in-hospital mortality, number of ICU and ventilation days, ICU stay longer than 14 days or readmission, use of renal replacement therapy and inotropic or vasopressor support, occurrence of infections, and critical illness polyneuropathy.

Results showed that use of intensive insulin therapy was associated with a 34% decrease in overall in-hospital mortality. The ICU mortality rate decreased from 8% to 4.6% (> 40% reduction), with the effect occurring primarily in patients in the ICU for more than 5 days (intensive therapy group 10.6% mortality vs 20.2% mortality in conventional therapy group). After adjustment for repeated interim analyses, the median unbiased estimate of mortality reduction was 32% (adjusted 95% confidence interval 2-%, p<0.04). No significant difference was noted in mortality between the two groups within the first 5 days in ICU. Furthermore, tight blood glucose control significantly decreased morbidity as demonstrated by a reduction in the duration of ventilatory support and ICU stay, reduced need for blood transfusions, 46% reduction in frequency of sepsis, decreased frequency of excessive inflammation, 44% reduction in polyneuropathy, and 41% reduction in acute renal failure. The overall conclusion was that use of exogenous insulin to maintain a blood glucose level below 110 mg/dl reduced morbidity and ICU and in-hospital mortality among surgical critically ill patients regardless of their history of diabetes.

This study demonstrated that tight glycemic control can prevent infections and associated complications in critically ill surgical patients. Sixty-one patients (7.8%) developed sepsis (bacteremia with systemic inflammatory response syndrome) in the conventional therapy group compared with 32 patients (4.2%) in the intensive insulin therapy group (p<0.003). In addition to the decrease in the frequency of septicemia, mortality from multiorgan failure with a proven septic focus decreased from 4.2% to 1.0%. Indeed, this subgroup represented the patient population that had the greatest reduction in death with intensive insulin therapy. Further, patients in the intensive insulin therapy group were less likely to receive prolonged antibiotics. One hundred thirty-four patients (17.1%) received antibiotics for more than 10 days in the conventional therapy group compared with 86 patients (11.2%) in the intensive insulin therapy group (p<0.001).

In 2003, the same group of authors addressed the issue of the relative impact of insulin dosage versus blood glucose concentration and their contributions to the demonstrated outcome benefits.[25,26] Results showed that control of hyperglycemia, rather than dosage of insulin, was associated with the beneficial outcomes of decreases in mortality, bacteremia, polyneuropathy, inflammation, and anemia, but not to the prevention of acute renal failure. The data also showed that there was no specific threshold below which no further risk reduction occurred. When patients were stratified according to glucose concentration, even moderate hyperglycemia (110-150 mg/dl) was associated with significantly increased risk of bacteremia and mortality. The authors speculated that the association of insulin dosage with prevention of renal failure may be due to either a renal protective effect of insulin or reduced insulin requirements secondary to either increased glucose removal in patients receiving continuous renal replacement therapy or reduced insulin clearance in renal failure.

In 2003, the results were published of a 6-month, prospective, single center, observational study evaluating glucose control and mortality risk in 523 adult patients in the ICU.[49] The goals of the study were to determine whether control of glucose metabolism or the degree of insulin administration was most likely to influence patient outcome and if there is a threshold glucose level above which there is an increased risk of death. Patients were predominantly male (72.8%) with a mean age of 64 years. Most subjects were cardiac surgery patients (85.1%), with a small subset of medical patients (11.7%).

Investigators defined six different bands of glycemic control and determined the amount of time each patient spent within each band. Blood glucose results were stratified based on ICU survival. Results showed that increased administration of insulin was positively and significantly associated with increased mortality. The authors concluded that glucose control rather than administration of exogenous insulin is the predominant factor in realizing decreased mortality. Furthermore, they observed a threshold blood glucose level between 145 and 180 mg/dl above which an increased risk of mortality exists. The authors suggest that a target blood glucose level of less than 145 mg/dl may be adequate and should be associated with less risk of hypoglycemia. The results of this study must be interpreted with caution because of its noninterventional design.

In 2003, results of a retrospective study were published in which the effect of hyperglycemia in a heterogeneous patient population was evaluated.[48] The study, which involved 1826 critically ill medical and surgical patients, found that increased mean serum glucose concentrations were associated with an increased mortality rate. Even modest increases in mean serum glucose concentration were associated with an increased rate of mortality. Due to its retrospective nature, this study did not allow for definitive conclusions regarding the effects of hyperglycemia. However, it provided compelling evidence for further studies into the role of hyperglycemia in causing an increased risk of mortality in a heterogeneous population.[52]

In a subsequent study, the same author evaluated the effect of a glucose management protocol in 800 patients compared with a historic control group of 800 patients.[50] The blood glucose goal was 140 mg/dl or less, and the study population included surgical and medical critically ill patients. Results showed that use of the glucose protocol led to significant decreases in hospital mortality (29.3%, p=0.002), new-onset renal failure (75%, p=0.03), patients requiring blood transfusions (18.7%, p=0.04), and ICU length of stay (10.8%, p=0.01). Despite its nonrandomized and historic control design, this study is important in showing mortality benefit in a heterogeneous critically ill population.

Unlike the previously mentioned 2001 study,[20] this study[50] showed no decrease in the frequency of infections. The author suggested that failure to decrease the frequency of infection may have been because the infection rate in the baseline period was already low owing to multiple interventional policies such as antibiotic-impregnated catheters, universal infection precautions, and procedures for prevention of ventilator-associated pneumonia (e.g., head of bed elevation). However, the level of blood glucose control needs to be considered. The mean blood glucose level in the latter study[52] was 130.7 mg/dl compared with the mean morning glucose level of 103 mg/dl in the former[20] study. Post hoc analysis of the former study[20] showed that although intermediate glucose control (110-150 mg/dl) was associated with decreased mortality, it appeared that the benefits of decreased morbidity, including prevention of infections, required stricter glycemic control.[26]

Although the latter study[50] did not achieve the same reduction in mortality as that of the former[20] (29% vs 34%), the benefits of an insulin infusion protocol outside of a clinical trial were demonstrated. In addition, episodes of hypoglycemia (blood glucose level < 40 mg/dl) in the treatment groups occurred with greater frequency in the former study[20] (5.2%) than in the latter study[50] (0.34%). However, neither study reported clinically significant adverse outcomes in patients who developed hypoglycemia.

Insulin Effect on Pathophysiology of Hyperglycemia-Related Cellular Derangements

Arguing for the benefit of exogenous insulin administration, one group of authors showed that intensive insulin therapy exerts a powerful antiinflammatory effect through modulation of C-reactive protein and appears to counter the effects of low levels of mannose-binding lectin (MBL).[53] C-reactive protein and MBL are acute-phase proteins synthesized by the liver. C-reactive protein is used as an inflammatory marker, which may be involved in the activation of leukocytes and the complement system; MBL is involved in innate immunity by initiating opsonization of microorganisms. Deficiency in MBL has been associated with increased risk of infection.

The investigators conducted a post hoc analysis of the subset of patients (> 5 days in the ICU) in whom the mortality benefit of intensive insulin therapy occurred.[53] Serum C-reactive protein and MBL were measured on admission, days 5 and 15, and the last day of ICU stay. Baseline serum C-reactive protein and MBL concentrations were similar between the two treatment groups. Serum C-reactive protein concentrations were significantly suppressed at all time points in the intensive insulin therapy group and, after day 5, were independently associated with decreased mortality (p<0.0001) and decreased frequency of acute renal failure (p<0.02) by multivariate analysis. Also, conventional insulin therapy and a high baseline C-reactive protein significantly and independently increased the risk for prolonged inflammation and prolonged antibiotic therapy. For all patients, nonsurvivors tended to have lower baseline MBL concentrations compared with that of survivors.

In addition, a subanalysis of 243 conventionally treated patients showed that baseline MBL concentrations were significantly lower in nonsurvivors compared with survivors (387 vs 897 B5g/ml, p=0.04).[53] During ICU stay, all patients had significant increases in MBL concentrations. Although intensive insulin therapy blunted increases in MBL concentrations compared with conventional insulin therapy (p<0.02), multivariate analysis showed that insulin therapy, not baseline MBL concentrations nor suppressed MBL concentrations over time, determined patient outcome. The authors suggested that the antiinflammatory activity associated with intensive insulin therapy partly explains the improvement in morbidity and mortality and that the possible adverse consequences of low baseline MBL levels are overcome. Further, results showed that the suppression of C-reactive protein was present among uninfected survivors, thus suggesting a direct antiinflammatory effect in the intensive insulin therapy group. However, the relative contributions of glycemic control versus insulin exposure are unknown.

The antiinflammatory effect of insulin has been suggested also through the suppression of proinflammatory cytokine and superoxide radical production and signaling.[26,27,28] Furthermore, postulated positive effects of insulin in critical illness include its role in the functional improvement of the insulin-sensitive organs through direct anabolic effect in acute illness, thus promoting tissue repair and preventing transfusions, dialysis, and critical illness polyneuropathy[21,24,26,29]; enhanced energy production and delivery to ischemic tissues; decreased free radical production and enhanced nitric acid production; direct anabolic effects on muscles; and maintenance of macrophage and neutrophil function.[28]

Hyperglycemia and Infection in Patients Receiving Parenteral Nutrition

Although adequate nutrition is essential to maintain a competent immune system,[54] overfeeding and inappropriate use of parenteral nutrition leads to serious complications.[55] Compared with enterally fed patients, patients who receive parenteral nutrition have a higher frequency of infectious complications, which include catheter-related sepsis, intraabdominal abscess, wound infections, and pneumonia.[56,57] The increase in infection rate is likely the result of several factors, including catheter-related infections,[58] reduced intestinal integrity and immunity during bowel rest leading to bacterial translocation,[59] and the higher frequency of hyperglycemia with parenteral nutrition.[60] Patients receiving parenteral nutrition commonly have a higher frequency of hyperglycemia and more severe cases than those of enterally fed patients, and this usually necessitates higher insulin doses to achieve normoglycemia.[20,56]

The Veteran Affairs Cooperative Study that evaluated the role of perioperative parenteral nutrition found that the benefits of this type of nutrition are mostly in severely malnourished patient.[60] A higher frequency of postoperative infections such as pneumonia and wound infections occurred in mildly malnourished patients compared with the severely malnourished patients who received parenteral nutrition. Although parenteral nutrition was beneficial in malnourished patients in reducing noninfectious complications, there was no change in infectious complications. A much higher frequency of severe hyperglycemia (serum glucose concentrations > 300 mg/dl) occurred in patients receiving parenteral nutrition (20%) compared with the control group (1%). Infection was 2.2 times more common in surgical patients who received parenteral nutrition. It was presumed that severe hyperglycemia in patients who received parenteral nutrition may have contributed to increased infectious complications.[61]

In a meta-analysis of enteral feeding versus parenteral nutrition studies in perioperative patients, 61% greater infectious complications were found in the parenteral nutrition group compared with those who received enteral nutrition.[62] Serum glucose concentrations were significantly higher in the parenteral nutrition group at all points of the analysis compared with the enteral nutrition group. Although these two publications[60,62] pointed at a possible correlation between hyperglycemia and infection in patients receiving parenteral nutrition, the relationship between the timing of hyperglycemia and infection and the threshold of serum glucose concentration above which there is a higher infectious risk were not determined. Because hyperglycemia may represent a response to sepsis rather than necessarily being a cause of septic complications, the cause-effect relationship between hyperglycemia and infection in critically ill patients regardless of the route of nutrition support has been a debatable subject.[63]