A 65-year-old obese (100 kg) man with type 2 diabetes, hypertension, and a pack-a-day smoking habit is admitted with moderately severe bilobar pneumonia
by Greg Maynard MD, MSc
A 65-year-old obese (100 kg) man with type 2 diabetes, hypertension, and a pack-a-day smoking habit is admitted with moderately severe bilobar pneumonia. His condition is manifest by fever, cough, chills, leukocytosis, and a modest oxygen requirement. You order oxygen, intravenous (IV) fluids, diet, and appropriate antibiotics while continuing the history and chart review. The patient uses metformin and glyburide, and his home glucose readings are generally in the 160 to 180 mg/dL range. An HbA1c level performed three months ago was 9.8, leading to an increased dose of glyburide. As you finish the history, the nurse reports a glucose reading of 198 mg/dL. What is the target blood glucose for noncritical care adult inpatients?
Diabetes mellitus is an epidemic in the United States. At least 9.3% of adults older than 20 (more than 20 million people) have diabetes. Approximately 30% are unaware they have diabetes.1 Concurrent with the increasing prevalence of diabetes in the U.S. from 1980 through 2003, the number of hospital discharges with diabetes as any listed diagnosis more than doubled between 1980 and 2003. These trends are expected to accelerate.2 Studies suggest 26% of inpatients have diabetes and 12% have pre-diabetes, previously undiagnosed diabetes, or stress hyperglycemia.3
A full review of the evidence is beyond the scope of this article. What follows is a sampling of the most representative or influential critical care studies.
Fluid and electrolyte balance, left ventricular (LV) function, leukocyte action, wound healing, endothelial function, and immunoglobulin function are all impaired with hyperglycemia.
A prothrombotic state and enhanced platelet aggregation have been demonstrated with even mild elevations of blood glucose.
The mechanisms are multifactorial and complex and involve metabolic derangements leading to oxidative stress, release of free fatty acids, and counter-regulatory hormones.4-6
A strong and consistent association with hyperglycemia and adverse outcomes is seen in a wide variety of critical care and peri-operative settings. Trauma survival, stroke survival and function, and the incidence of post-operative infections are all adversely affected by hyperglycemia.7-10 Acute myocardial infarction (MI) mortality, acute MI infarct size, and LV dysfunction are also consistently adversely affected in these studies.11-13
This association is typically present in hyperglycemic patients whether they have a diagnosis of diabetes or not, and the association is often even stronger in those lacking a pre-existing diagnosis. Dysfunction typically is detectable at only modest elevations of blood glucose and becomes more marked in a dose response relationship.
The Portland Diabetic Project is a prospective, non-randomized, observational study of 5,510 consecutive diabetic cardiac surgery patients.14-15 The three-day blood glucose average (3-BG) has been progressively reduced for the population through the use of continuous insulin infusion (CII).
The last reported glycemic target is less than 130 mg/dL, and the current glycemic target is less than 110 mg/dL. Both CII for three days and a favorable 3-BG were independently associated with improved mortality, deep sternal-wound infection rates, and length of stay. Mortality and deep sternal-wound infection rates for diabetic patients with well-controlled glucose levels are equal to patients without diabetes.
Another study compared 800 mixed medical-surgical ICU patients with tight glycemic control (mean BG 130.7 mg/dL) to historical controls with a mean glucose of 152.3 mg/dL. The insulin infusion group had associated significant reductions in mortality and median length of ICU stay.16
In the first Diabetes and Insulin-Glucose study (DIGAMI 1), patients with acute MI received IV insulin therapy for 24 hours, followed by multiple daily injections for three months or longer. The insulin group had lower glucose values and a 29% reduction in mortality at one year and 28% reduction at 3.4 years compared with the control group.17-18
In the most influential study to date, van den Berghe, et al., randomized 1,548 surgical intensive-care unit (ICU) patients to either intensive (IT) or conventional (CT) insulin therapy.19 The glycemic target in the IT arm was 80 to 110 mg/dL (mean glucose attained was 103 mg/dL), while the CT arm had a mean glucose level of 153 mg/dL. The IT group enjoyed substantial reductions in both ICU and total in-hospital mortality, as well as reductions in blood stream infections, acute renal failure, transfusions and the duration of mechanical ventilation (p<0.01 for all).
While a similar study in a medical ICU did not achieve statistical significance in the overall intention-to-treat analysis for mortality, it did demonstrate reductions in mortality in patients with at least three days of ICU treatment and significant reductions in morbidity.20
A meta-analysis of these two studies demonstrated a relative risk reduction in mortality (23.6 to 20.4%) and morbidity in all patients treated with intensive insulin therapy.21
A separate meta-analysis of 35 clinical trials evaluating the effect of intensive insulin infusion therapy on mortality in critically ill inpatients revealed a 15% reduction in short-term mortality.22
There are no randomized controlled trials establishing the optimal glycemic target for noncritical care inpatients. There are a number of observational and pilot studies that reinforce the studies performed in critical care settings.
In a retrospective review of almost 1,900 general medical-surgical admissions, Umpierrez, et al., reported an 18-fold increase in mortality in hyperglycemic patients without prior history of diabetes and a 2.5-fold increase in mortality in patients with known diabetes compared to controls. These associations persisted with adjustment for severity of illness.23
A variety of observational and pilot studies associate hyperglycemia with poor outcomes in community acquired pneumonia, renal transplant, and the durability of remission in acute lymphocytic leukemia.24-25
Spurred by the emerging controlled trial evidence, the American Association of Clinical Endocrinologists (AACE) convened a consensus conference involving nine organizations, including SHM. Recommendations for the management of inpatient hyperglycemia included stringent glycemic targets for critical care and noncritical care areas.26 The American Diabetes Association (ADA) produced an excellent technical review on inpatient diabetes that provided the basis for ADA Clinical Practice Guideline glycemic targets.27 The glycemic targets recommended are shown in Table 1 (above).
Before accepting these recommended glycemic targets, review the shortcomings of the literature supporting them and consider institutional and individual patient factors that might modify the glycemic target.
Insulin has beneficial vascular and anti-inflammatory effects in its own right, making it difficult in some studies to distinguish the benefit of glucose lowering from the benefit of the insulin used to attain improved control.
The majority of studies supporting inpatient glycemic targets are observational or non-randomized. Some use admission blood glucose concentrations as the sole measure of glucose control.
While most of these studies used valid methods to control for severity of illness and co-morbidities, these methods are not perfect. In some cases, hyperglycemia may have been a marker of a more stressed and sick patient rather than an independent source of adverse outcome.
The dramatic results from the first van den Berghe study have proved difficult to replicate, in part because other investigators have had difficulty achieving stringent glycemic targets safely. Two international multicenter studies recently stopped enrollment due to excess rates of hypoglycemia, but the studies have not yet been published in final form.28-29
Finally, it bears repeating that the proposed glucose targets for noncritically ill patients are based on essentially no clinical trial data in that population. In part, the glycemic targets reflect the evidence derived from landmark outpatient randomized trials.30-31 In the outpatient setting, insulin requirements and nutritional intake are far more reliable than the inpatient setting, where the iatrogenic induction of excessive hypoglycemia is a valid concern.
Rapidly fluctuating nutritional status, changing insulin requirements, varied levels of expertise, and hand-offs between geographic locations and providers are all common in the inpatient setting.
Aggressively pursuing glycemic targets without having systems and safeguards in place could lead to net harm.
The AACE recently identified these barriers and recommended a multidisciplinary team approach, reliable metrics, and a standardized method for insulin protocols, orders, and hypoglycemia prevention and treatment techniques.32
Both the AACE and SHM have produced toolkits to assist institutions to safely achieve improved glycemic control and care.33-34 The SHM Glycemic Control Task Force recently summarized key concepts to emphasize in formulating protocols and order sets in the noncritical care setting (see Table 2, left).
Stringent glycemic targets recommended by the ADA and the AACE may be appropriately moderated in centers that do not yet have the systems in place to achieve those goals safely.
Your glycemic target need not be identical to the ADA and AACE glycemic targets but should be similar to them. Examples of glycemic targets for noncritically ill inpatients are shown in Table 3 (see p. 48).
The glycemic target should be actionable, in that some institutionally endorsed action should result when a patient’s glycemic target is consistently not met.
Your patient has an active infection, a glucose of 198 mg/dL and an elevated HbA1c. You hold the oral agents and start a basal bolus insulin regimen.
You generate an estimate for a safe insulin total daily dose of 60 units (100 kg x 0.6 units per kg for an obese, type 2 diabetes patient), and administer half as basal insulin, with the remaining 30 units distributed as rapid acting insulin in three divided doses. Your orders include routine glucose monitoring, and you plan to adjust the insulin daily as needed to adhere to the institutional glycemic target for noncritically ill patients of 90 to 150 mg/dL. TH
Dr. Maynard is the division chief for hospital medicine at the University of California, San Diego. He is the leader of SHM’s Glycemic Control Task Force and a leader of the VTE Prevention Collaborative.
The Hospitalist newsmagazine reports on issues and trends in hospital medicine. The Hospitalist reaches more than 25,000 hospitalists, physician assistants, nurse practitioners, residents, and medical administrators interested in the practice and business of hospital medicine.