Management of Inpatient Hyperglycemia

Lowell R. Schmeltz, MD


Lab Med. 2011;42(7):427-434. 

In This Article

Insulin: The Standard of Care

Insulin is the preferred agent for glycemic control in hospitalized patients. The pharmacodynamics of insulin allow it to be adaptable to the changing physiology of the sick patient, is easily titrated, and has no dosage threshold. Furthermore, insulin has a rapid onset of action, minimal side effects except for hypoglycemia, and has minimal drug-drug interactions. The ideal insulin protocol will help reach the glucose target range timely, effectively treat all degrees of hyperglycemia, minimize glycemic variation and the risk of hypoglycemia, and is easy for nurses to carry out in a timely fashion.[43]

The Society of Hospital Medicine (SHM) has created a workbook to guide hospitals in creating safe and effective glycemic control plans. Included in the workbook appendix are multiple successful inpatient insulin protocols. The common themes in these protocols include the use of regular insulin for continuous insulin infusion and a basal/bolus SC insulin regimen including a long-acting insulin analog (insulin glargine or detemir) and prandial and correctional doses of a rapid-acting insulin (insulin aspart, glulisine, or lispro).[44–47]

Continuous infusion of regular insulin is suggested for critically ill ICU patients, pre- and postoperative patients, peripartum women with hyperglycemia, severe hyperglycemia with metabolic decompensation (diabetic ketoacidosis and hyperosmolar non-ketotic states), and any patient in whom tight glycemic control is clinically indicated. Paper-based and computer-based insulin infusion algorithms are available to help clinicians achieve optimal glycemic control.[44,48,49]

Conversion from IV to SC insulin commonly occurs when the critical illness resolves when the patient is extubated, off vasopressors, and ready to begin eating, or is at a stable tubefeed rate. When the patient is being converted from an IV insulin drip, the drip rate is used as a guide to determine total daily insulin requirements. The insulin drip rate over the preceding 6 hours is averaged to obtain a stable hourly rate. The average hourly rate is multiplied by 24 hours to calculate the total daily dose (TDD) of insulin required. The basal insulin dose ordered is 60%-80% of the TDD, and the prandial insulin dose for each meal is 10% of the TDD. The prandial insulin dose is adjusted accordingly as the patient's appetite improves. The proportion of insulin given for prandial dosing is substantially less because these patients are generally consuming only a clear liquid diet initially with a reduced caloric content. The insulin infusion should be continued for 4 hours after the first injection of basal insulin is given (if insulin glargine or detemir; 2 hours if NPH insulin). Practical necessities sometimes outweigh physiologic reasoning in that the conversion from IV to SC insulin often coincides with the transfer of the patient out of the ICU rapidly. Basic insulin may be given, and the insulin infusion is simply stopped without an overlap period. In this scenario, a conversion dose of a rapid-acting insulin (10% of TDD) can be given simultaneously with the basal insulin, and the insulin infusion can be discontinued without an overlap period.[50]

For all non-critically ill patients, a basal/bolus insulin regimen is the preferred method of glycemic control. Basal insulin suppresses hepatic gluconeogenesis between meals and overnight. During illness, basal insulin requirements rise with any physical stress, including surgery, infection, infarction, or fever. For patients who are eating, a scheduled mealtime insulin dose with a rapid-acting insulin analog helps prevent the glucose from rising from carbohydrate intake. Whether eating or not, when blood sugars are outside the glycemic target range, a correctional dose of rapid-acting insulin should be administered.

Hospitalized patients have unpredictable eating and diagnostic testing schedules and thus are more susceptible to an insulin-food dyssynchrony; hypoglycemia occurs if insulin peaks before the patient has eaten or consumed enough carbohydrates; hyperglycemia results if the insulin peak is insufficient to meet glucose intake or metabolic stress needs. Rapid-acting insulin analogs can be given immediately before or up to 20 minutes following food consumption and thus are more flexible and less likely to cause hypoglycemia compared to regular insulin.[51]

Insulin analogs are preferred for basal, mealtime, and correction doses instead of human insulins (regular and NPH). Insulin analogs have a more predictable absorption and action profile in addition to less pharmacokinetic fluctuation in patients with renal insufficiency. High doses of human insulins have not only a greater peak effect than lower doses but also result in a longer duration of action. Overlap of insulin doses (known as insulin stacking) increases the risk of a hypoglycemic event. The duration of action of rapid-acting insulin analogs are predictable at low and high doses, thereby decreasing the risk of insulin stacking. Insulin analogs have a more consistent pharmacokinetic and pharmacodynamic profile (less inter- and intra-individual variability),[52,53] so it is easier to predict the effect the dose will have on an individual's BG concentration. Table 2 describes the types of insulin recommended for hospital use.[54,55]

Regular insulin should be avoided for SC postprandial BG correction and should not be used as monotherapy in a sliding-scale regimen. Numerous studies over the past 50 years show that sliding-scale insulin (SSI) alone is not effective for inpatient glycemic control (Figure 1) and more recently has been associated with increased inpatient mortality.[56] Sliding-scale insulin regimens do not allow for basal or mealtime insulin requirements and grossly underestimate total daily insulin requirements. Furthermore, SSI regimens respond to hyperglycemia after it has happened, rather than preventing it, and the sliding scale depends on the inaccurate assumption that insulin sensitivity is uniform among all patients.

Figure 1.

Example of a patient on a sliding-scale-only insulin regimen for the first few days of hospitalization with high mean glucose with large glycemic variability. When switched to a basal/bolus insulin regimen, mean glucose gradually decreased, and glycemic variability was significantly less

In a randomized, prospective study, the use of a basal/bolus regimen vs SSI regimen in diabetic patients naïve to insulin (on oral agents only) was superior in achieving glycemic control with almost no hypoglycemia (RABBIT-2).[57] In this study and the subsequent RABBIT surgery study,[58] the TDD was calculated by multiplying the patient weight in kilograms by either 0.4 or 0.5 units/kg and ordering 50% of TDD as basal insulin and the other 50% of TDD as mealtime insulin divided equally into 3 mealtime doses. Glucose targets were reached in the majority of patients in the basal bolus intervention group with minimal hypoglycemia.

Premixed insulins are generally not recommended for use in the hospital setting, as there is an increased risk of hypoglycemia in patients with variable oral intake. Some hospitals have eliminated premix insulins via therapeutic interchange so patients do not get mealtime insulin if at a procedure or NPO, therefore reducing the risk of hypoglycemia. The TDD can be calculated by adding up the total daily home dose of premixed insulin. This allows basal and pre-meal insulins to be ordered separately; thereby, mealtime insulin can be administered only when the patient is eating. Subsequently, at the time of discharge, these patients can be converted back to their premixed insulin if appropriate.

All hyperglycemic patients should have their hemoglobin A1c (HbA1C) checked on admission to help differentiate between pre-hospital or acute onset of hyperglycemia. The Joint Commission on Accreditation of Healthcare Organizations (JCAHO) recommends a A1C during the hospital stay if one is not documented in the past 60 days to identify previously unrecognized diabetes or to help guide optimization of the outpatient diabetes regimen if the A1C is elevated. Recently, an international expert committee recommended an A1C ≥ 6.5% indicates a diagnosis of diabetes.[59]


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.