Section 1. Current Scientific Evidence on Glycemic Control and Targets: the Noncritically

Section 1. Current Scientific Evidence on Glycemic Control and Targets: The Noncritically Ill Hospitalized Patient

A 58-year-old obese woman with type 2 diabetes mellitus is admitted for treatment of an infected diabetes-related foot ulcer. There is a 20- x 10-cm area of cellulitis surrounding the ulcer, which has some purulent drainage and contains significant fibrinous material. The patient is started on intravenous (IV) antibiotics. The surgeons have requested that she be kept nothing by mouth (NPO) after midnight for surgical debridement in the morning. Her current weight is 100 kg (body mass index [BMI] 38), and her recent glycemic control can be summarized as having blood glucose values that are usually in the mid 200s mg/dL and a recent glycosylated hemoglobin (A1c) measurement of 11.2%. Her home medical regimen includes glipizide 10 mg twice daily, metformin 1000 mg twice daily, and exenatide 10 mg subcutaneous twice daily. Her blood glucose in the emergency department is 292 mg/dL. Which statement is MOST correct about the role of glycemic control in this patient’s care?
A.There is evidence from randomized-controlled trials supporting tight glycemic control in this type of patient. Optimal glucose target is below 110 mg/dL.
B.There is evidence from randomized-controlled trials supporting tight glycemic control in this type of patient. Optimal glucose target is below 180 mg/dL.
C.There are observational data demonstrating an association between poor outcomes and hyperglycemia in this type of patient. Optimal glucose target is uncertain.
D.There are no data linking glycemic control to outcomes in this type of patient. Optimal glucose target is uncertain.
Answer: C. There are many observational trials demonstrating an association between hyperglycemia and adverse clinical outcomes in hospital patients. The patient in this case requires active management of her blood glucose levels given her poor baseline blood glucose control and her current active medical and surgical issues. The optimal glycemic target is uncertain, although maintaining a blood glucose level below 180 mg/dL would be generally in accord with expert recommendations.

Current Scientific Evidence on Glycemic Control and Targets: The Noncritically Ill Hospitalized Patient

There is substantial evidence demonstrating an association between hyperglycemia in hospitalized patients and adverse outcomes.1-6 As an example, Umpierrez et al showed that 38% of patients admitted to their hospital suffered from hyperglycemia, and 33% of these patients had not been diagnosed with diabetes prior to the hospital admission. The patients with diabetes and hyperglycemia had a significantly higher in-hospital mortality than the patients with normal blood glucoses (3% vs 1.7%), but the patients with hyperglycemia that were not known to have diabetes had the highest mortality of all (16%).2 Other cohort studies have also shown associations between hyperglycemia and infections.3-5
Several randomized trials have evaluated the effects of targeting “euglycemia” in intensive care unit (ICU) patients. Unfortunately, there have been no comparable randomized studies in noncritically ill hospitalized patients. Without randomized trials, clinicians try to extrapolate from the conflicting results of the ICU trials, which may not apply to noncritically ill patients. Moreover, the overall management goals for the noncritically ill are unclear. Is the goal to only avoid severe illness related to hyperglycemia, such as diabetic ketoacidosis; to avoid moderate levels of hyperglycemia and its pathophysiologic implications; or to achieve true euglycemia in hopes of preventing all theoretical adverse consequences of hyperglycemia in a patient with active illness?
Currently, hospitalists are saddled with the responsibility of managing inpatient diabetes and hyperglycemia without good data to define specific glycemic targets or treatments. Although some argue that the lack of data on the subject means that we should not waste our time discussing it, the subject is ubiquitous and difficult to ignore. At the time of admission to the hospital, most patients with diabetes hand over the management of their blood glucose levels to the doctors and nurses. Given the lack of strong scientific data, clinicians often turn to expert opinion for a rational approach. Most of the recommendations in this module are based on the opinions of experts.1,7-9
In a consensus statement, the American Diabetes Association (ADA) and the American Association of Clinical Endocrinologists have recommended that fasting blood glucoses should be below 140 mg/dL, and random glucoses should be below 180 mg/dL when patients are in the hospital, as long as this can be safely achieved (Table 1).7 However, these values are given only as guidance and the authors note that glycemic control should be individualized. These recommendations are based solely on the opinion of experts, as opposed to scientific trial evidence. Critics call the goals arbitrary and lament that they are potentially difficult to achieve.

In summary, there are abundant studies demonstrating an association between hyperglycemia and adverse outcomes in hospital patients, and there are practical reasons for controlling blood glucose levels in hospitalized patients. However, there is not strong evidence supporting any specific glucose target for noncritically ill hospitalized patients. Expert recommendations offer some guidance.

Section 2. Controlling Blood Glucose Levels in the Hospital: Which Medications Are Best

A 58-year-old obese woman with type 2 diabetes mellitus is admitted for treatment of an infected diabetes-related foot ulcer. There is a 20- x 10-cm area of cellulitis surrounding the ulcer, which has some purulent drainage and contains significant fibrinous material. The patient is started on IV antibiotics. The surgeons have requested that she be kept NPO after midnight for surgical debridement in the morning. Her current weight is 100 kg (BMI 38), and her recent glycemic control can be summarized as having blood glucose values that are usually in the mid 200s mg/dL and a recent A1c measurement of 11.2%. Her home medical regimen includes glipizide 10 mg twice daily, metformin 1000 mg twice daily, and exenatide 10 mg subcutaneous twice daily. Her blood glucose in the emergency department is 292 mg/dL. Which of the following treatment modalities would be MOST consistent with expert recommendations in this situation?
A.Continue her outpatient medications for diabetes, but add a dose of glargine insulin before bedtime and an insulin sliding scale.
B.Discontinue her diabetes medications and initiate sliding scale insulin to estimate her insulin needs.
C.Discontinue her diabetes medications and initiate scheduled glargine, scheduled short-acting insulin given with meals, and an insulin sliding scale.
D.Increase the dose of her home diabetes medications.

Answer: C. The patient in the case should be given scheduled basal and nutritional insulin.

Controlling Blood Glucose Levels in the Hospital: Which Medications Are Best

Medications to treat hyperglycemia in the hospitalized patient should have the following characteristics:

Act rapidly
Allow for rapid titration (ie, increases and decrease in doses should have immediate effects)
Have safety monitoring (ie, glucose testing)

Insulin is the medication that most often meets these criteria, and it is the medication on which the remainder of this module will focus. I believe that most noncritically ill patients who demonstrate blood glucose levels that are consistently outside of the target range should be treated with subcutaneous insulin. Most other diabetes medications have shortcomings when used in patients that are ill enough to be hospitalized.1

Table 2 summarizes the issues that should be considered when using noninsulin diabetes medications in the hospital, and illustrates why insulin is regarded as the medication of choice for controlling blood glucose in ill patients.8There are situations, such as a patient nearing discharge, where the use of oral medications (or other noninsulin medications) is appropriate.

Although insulin is the drug of choice for managing insulin in the hospitalized patient, it is important to recognize that insulin has historically been associated with high rates of adverse events, especially hypoglycemia, when used in the hospital. Therefore, many experts believe that insulin use in hospitals should be standardized to assure patient safety and reduce insulin-related errors.

In the multiple choice question for this section, the answer is relatively straightforward because this patient should be treated with insulin regardless of her acute hospitalization based on her recent blood glucose levels and her A1c level. Continuation of her oral medications would, at best, maintain her poor control. The addition of other noninsulin agents in this case would not have a rapid effect on her glycemic control and would be unlikely to achieve adequate control. Because the patient will be NPO after midnight, her sulfonylurea should be held to reduce the likelihood of hypoglycemia in the fasting state. Metformin should be held in the perioperative setting where there is an increased likelihood of hemodynamic changes or changes in renal function which could increase the risk of metformin-induced lactic acidosis.

Exenatide has not been well studied in the inpatient setting and carries the risk of nausea, gastroparesis, and pancreatitis. A once-weekly formulation (pending US Food and Drug Administration approval), with its long duration of action (up to 4–6 week), will pose a management challenge to the inpatient provider. Providers will have to prescribe appropriate amounts of insulin, likely at lower doses than most formulas would recommend, to account for the glucagon-like peptide-1 medication. Therefore, this patient’s home regimen should be held, and she should be treated with insulin.

Answer B (initiate a short-acting insulin sliding scale) is what many physicians would do in this case. An insulin sliding scale used in isolation is designed to provide insulin to patients who need insulin only on an intermittent basis. Sliding scale insulin should not be used alone in this scenario because the patient is demonstrating consistent hyperglycemia. When using a sliding scale, patients receive insulin only after they experience hyperglycemia (ie, the insulin is given reactively). Because the patient in the case is demonstrating consistent hyperglycemia, it is preferable to give insulin in an anticipatory way (ie, preemptively) to maintain the patient in a controlled metabolic state. In addition, sliding scales, when used alone, almost never deliver a well-calculated dose of insulin for the particular circumstances.

An insulin sliding scale may be appropriately used alone for the occasional patient who does not have a defined need for insulin therapy. As an example, sliding scale insulin may be prescribed for the patient who is being initiated on high-dose corticosteroids, with normal blood glucose levels so far. In that case, it is not certain that the patient will need insulin, and the sliding scale is used as a safety mechanism. For patients with a defined need for insulin (ie, those with a diagnosis of diabetes requiring oral medications or insulin or those exhibiting hyperglycemia), insulin should be given in an anticipatory and physiologic manner. A prospective, multicenter, randomized trial comparing the efficacy and safety of a basal-bolus regimen with that of sliding scale insulin in patients with type 2 diabetes demonstrated that basal-bolus insulin improved glycemic control without and increased rate of hypoglycemia.10 In the RABBIT-2 Trial, patients with type 2 diabetes randomized to basal-bolus insulin achieved a blood glucose target of less than 140 mg/dL in 66% of patients versus 38% of patients on sliding scale. Rates of hypoglycemia, defined as a glucose lower than 60 mg/dL were equal in each group.

In summary, patients who demonstrate consistent hyperglycemia in the hospital should be treated with insulin. Other agents are acceptable in some clinical circumstances, but insulin is fast acting, titratable, and almost never contraindicated in hospitalized patients. Although there are a few circumstances where an insulin sliding scale might be used alone, it should be the exception and not the rule. Patients with consistent hyperglycemia need anticipatory and physiologic insulin.

Section 3. Step 1: Computing an Appropriate Dose of Insulin to Manage Hyperglycemia in the Hospital

A 58-year-old obese woman with type 2 diabetes mellitus was admitted for treatment of an infected diabetes-related foot ulcer. There is a 20- x 10-cm area of cellulitis surrounding the ulcer that was debrided yesterday and looks much better. Her current weight is 100 kg (BMI 38), and her recent last A1c measurement was 11.2%. The surgical team appropriately stopped her oral diabetes medications, but only placed her on a sliding scale insulin regimen. She is now eating normally and her blood glucose before lunch is 293 mg/dL. How much total insulin would you estimate that this patient would require in a 24-hour period when eating 3 meals?
A.20 units or less
B.Approximately 30 units
C.Approximately 40 units
D.50 units or more

Answer: D. A weight-based calculation would suggest that the patient will require 50 units of insulin or more in a day when receiving adequate nutrition.

Step 1: Computing an Appropriate Dose of Insulin to Manage Hyperglycemia in the Hospital

There are many barriers to good glycemic control in the hospital. One of the biggest barriers is deciding how much insulin to prescribe. Physicians who are uncomfortable estimating a patient’s insulin needs are more likely to avoid the use of insulin even when it is appropriate and to underdose the insulin when it is prescribed.

The Society of Hospital Medicine’s Glycemic Control Task Force provides guidelines for choosing a starting dose of insulin for hospitalized patients with hyperglycemia.8The first step is to estimate a patient’s total daily dose of insulin (TDD). The TDD is the total amount of insulin that a patient requires for metabolic control over the course of an entire day when receiving nutrition. That dose includes the patient’s basal insulin needs (long-acting insulin such as glargine or neutral protamine Hagedorn [NPH]) and his or her nutritional needs (short-acting insulin such as aspart that is given with meals).

There are 2 methods for estimating TDD. Use the first method for patients who are on insulin at home. The home insulin dose can provide a starting point for the hospital insulin dose (even if the type of insulin will be modified). Of course, when using this approach, several other variables must be considered, including the patient’s glycemic control at home (if control is poor at home, the dose may need to be increased), the seriousness of the illness requiring hospitalization (the stress response to illness typically results in an increase in insulin resistance), and changes in nutritional intake. The second method uses the patient’s weight to calculate a starting dose of insulin. Both methods are outlined in Figure 1.8

The patient weights 100 kg and has a BMI consistent with obesity. Therefore, her TDD would be estimated to be at least 0.5 units/kg
TDD = 100 kg x 0.5 = 50 units of insulin

This means that this patient would be expected to require at least 50 units of insulin over the course of a day, when taking in normal levels of nutrition.

The estimates provided in Figure 1 are conservative, and many patients will require higher doses of insulin to achieve metabolic control. However, most patients will not require less insulin (ie, very few patients will suffer severe hypoglycemia at the suggested doses). Still, it is important to remember that these calculations are just estimates, and the initial estimate of the TDD is less important than the subsequent adjustments made to the dosing which are based on the patient’s response.

The fear of hypoglycemia often prompts clinicians to underdose insulin in the initiation phase. Perhaps the most common example of this is the use of sliding scale insulin to “see how much insulin the patient needs.” Although this approach is reasonable in a patient with borderline or normal blood glucose values, patients who are already demonstrating consistent hyperglycemia should be treated with scheduled (anticipatory) doses of insulin to achieve glycemic control. In this population, sliding scale insulin will almost always undertreat the patient and will result in an unnecessary delay in arriving at the appropriate TDD.

In summary, the first step toward glycemic control in the hospital is to calculate the insulin TDD. Method 1 is to calculate TDD based on the patient’s home insulin doses and make adjustments based on the factors described previously. Method 2 is to calculate TDD using a weight-based formula which takes into account the patient’s likely insulin sensitivity. This is a key step in formulating an anticipatory insulin program.

Section 4. Steps 2 and 3: Formulating an Anticipatory, Physiologic Insulin Regimen with Adjustments as Needed

A 100-kg 58-year-old obese woman with type 2 diabetes mellitus was admitted for treatment of an infected diabetes-related foot ulcer. There is a 20- x 10-cm area of cellulitis surrounding the ulcer that was debrided yesterday and looks much better. She was previously on oral medications to treat her diabetes. You have estimated that her insulin TDD is 50 units (100 kg x 0.5 = 50). She is now eating normally and her blood glucose before lunch is 293 mg/dL. What is the most physiologic way of delivering 50 units of insulin to the patient?
A.Give the entire amount as 3 doses of rapid-acting insulin (lispro) divided evenly before each meal.
B.Give the entire amount as intermediate-acting (NPH) or long-acting insulin (ie, detemir or glargine) in 1 or 2 injections per day, along with an insulin sliding scale.
C.Give half of the total daily insulin dose (25 units) as a long-acting insulin, and divide the other half to be given with meals.
D.Give insulin by sliding scale, escalating the dose until the target dose is reached.
Answer: C. The patient in the case has an estimated TDD of 50 units of insulin. Half of this could be given as glargine, 25 units subcutaneously daily. The remainder could be provided with meals, such as 8 units of a rapid-acting insulin analogue subcutaneously 3 times daily with meals.