Improving Diabetes Care in the Hospital Using Guideline-Directed Orders

  1. Stephen F. Quevedo, MD, FACP,
  2. Elaine Sullivan, MS, RN, CDE,
  3. Randi Kington, MS, APRN, CDE and
  4. Warren Rogers, RPh
  1. Address correspondence and reprint requests to Stephen F. Quevedo, MD, Director, Joslin Diabetes Center Affiliate at Lawrence & Memorial Hospital, 50 Faire Harbour Pl., New London, CT 06320.


Improving glycemic control for inpatients with diabetes remains a formidable challenge. While the use of inadequate and nonphysiological insulin regimens is widespread, there are many potential barriers to improving physician and nursing hospital practices in this area. This article describes an integrated program that utilizes subcutaneous and intravenous insulin order forms incorporating guidelines that encourage more appropriate insulin therapy in the hospital. This approach has resulted in a significant decrease in one factor associated with suboptimal insulin use for inpatients: reliance on sliding-scale short-acting insulin alone for blood glucose control.

Much recent attention has focused on the results of studies relating improved chronic control of both type 1 and type 2 diabetes to more favorable clinical outcomes. However, a quieter but also important growing literature is documenting the effects of improved diabetes care and better glycemic control in the hospital. Specifically, decreased length of stay and readmission rates have been associated with the use of diabetes care teams,1,2 and recent reports have amplified the literature documenting a reduced risk of infectious complications with improved perioperative glycemic control.3,4 In addition, the DIGAMI study5 showed an immediate and sustained benefit from the use of an insulin-glucose infusion followed by multiple-dose subcutaneous insulin therapy in patients with diabetes and acute myocardial infarction.

Despite this information, inpatient diabetes care in the United States is often suboptimal, owing at least in part to the continued widespread use of nonphysiological sliding-scale regular insulin coverage,6 and possibly to underutilization of insulin infusions in perioperative patients and those who can have nothing by mouth (NPO). Frequent errors in the use of insulin infusions in the treatment of diabetic ketoacidosis (DKA) have also been reported.7

Given these issues, it would seem reasonable to encourage the use of treatment guidelines, ideally covering all aspects of inpatient insulin therapy, to reduce the frequency of errors and improve inpatient glycemic control. However, compliance with published clinical guidelines, when the recommendations are divorced from the acute care episode, is often found to be poor.811 Barriers to the use of disseminated clinical guidelines include provider perception that they are too complex, invasive, vague, or controversial. Not surprisingly, those requiring a change in provider practice routines are also less likely to be followed.10,11

We have developed a model that incorporates some aspects of clinical guidelines with a physician order sheet, or guideline-directed orders, to encourage more appropriate use of both intravenous and subcutaneous insulin in the hospital. The goal is to improve the quality of inpatient diabetes care by making appropriate use of insulin therapy as provider-friendly and seamless as possible.

General Measures

The insulin infusion formulation was standardized in the pharmacy to consist of 25 units of human regular insulin mixed with 250 cc of normal saline. Each bag has an attached instruction/dosing label (Fig. 1) to reduce the chance of dosing error.

Sample order forms were mailed to the medical staff and were discussed at medical staff meetings. Similarly, they were reviewed with pharmacy and nursing, and suggested improvements were then incorporated. Final forms were presented for approval to the hospital’s Joint Practice Committee, an interdisciplinary team consisting of nurses, physicians, pharmacists, and administrators.

The medical staff received inservice education at two medical staff and Department of Medicine/Family Practice meetings. A video of a physician education session and a related information packet were made available to all medical staff. Viewing the video was mandatory for emergency room physicians and house physicians who admit patients and cover patient units overnight.

Education was offered by an advanced practice nurse/certified diabetes educator at varied times on all shifts to accommodate as many nurses, dietitians, and pharmacists as possible. Unit-specific inservices were offered for the emergency room, operating room, intensive and coronary care units, obstetrics unit, and an off-site emergency facility. A session was videotaped, and copies were distributed to each patient care unit, along with a study guide developed with the assistance of a staff nurse working towards her bachelor’s degree in nursing. Follow-up reminders and education were provided as needed. Staff were encouraged to call the diabetes center staff to clarify any questions about the guidelines.

Insulin Infusion Orders

The use of an insulin and dextrose intravenous (IV) infusion has been shown to offer advantages, especially in the intraoperative and perioperative period,12,13 and has been recommended for NPO patients with type 1 or type 2 diabetes.14 To encourage the implementation of this recommendation, an insulin infusion order sheet was developed (Fig. 2). The sheet consists of several parts, including a review of general infusion use, frequency of bedside capillary glucose testing, stat IV insulin orders, insulin infusion rates adjusted to the most current glucose level, treatment of hypoglycemia, parameters for automatic physician communication, IV fluid and potassium orders, and laboratory orders. The order sheet contains wording reminding physicians to consider the most appropriate infusion rates of dextrose (5–10 g/h) and to add potassium if appropriate.

Educational inservices included actual case studies of infusion use, including appropriate initial insulin infusion ranges based on clinical estimates. These sessions also stressed the improved glucose control possible with more physiological insulin use and stressed that this should result in fewer calls from the nursing units in contrast to the frequent additional instructions often needed with the use of intermittent subcutaneous short-acting insulin.

A separate insulin/dextrose infusion order sheet was developed for intrapartum use on the labor and delivery unit (Fig. 3). Using a similar format and current recommendations for the peripartum period, insulin infusion rates and weight-adjusted dextrose infusion rates are given.15,16

A more concentrated infusion of 100 units regular insulin in 100 cc normal saline can be used in the intensive care units and emergency room for patients requiring significant fluid restriction or very high infusion rates.

Diabetic Ketoacidosis

There are many potential areas for error in the treatment of DKA.7 By request of the departments of medicine and family practice, the ketoacidosis insulin and IV fluid orders were included in a comprehensive two-page admitting order form (Fig. 4). Guidelines are provided for appropriate fluid resuscitation, potassium replacement, and the addition of dextrose to the IV fluid. Insulin infusion rates are based on a very-low-dose regimen, using the glucose reduction rate as the trigger for titration.17 This reduces the risk of lowering effective arterial volume before fluid equilibration occurs and avoids unnecessarily rapid potassium flux.

Safeguards include a reminder for the nurse to call the physician when the blood glucose falls below 250 mg/dl for the addition of dextrose to the IV fluid and to call if the glucose is not falling at an adequate rate. This identifies the occasional patient who may require higher therapeutic insulin levels. Proper drip discontinuation guidelines are also included.

Subcutaneous Insulin

The use of sliding-scale short-acting insulin coverage remains a mainstay in many United States hospitals, despite its retroactive nature and lack of any supporting literature. Significant problems identified with the use of such regimens include infrequent use of basal insulin along with the short-acting component, and not changing the insulin dosage schedule based on the previous day’s experience.6,18 Informal review of baseline practice at our institution revealed that approximately half of the sliding scales were ordered without basal insulin and that insulin doses were infrequently adjusted regardless of the level of glycemic control.

Given these difficulties, an anticipatory basal insulin program with the optional use of supplemental, usually pre-meal short-acting insulin, would appear to be a better approach.14 We developed a subcutaneous/supplemental insulin order form to encourage such an approach (Fig. 5). Nursing instructions include timing preprandial bedside capillary glucose monitoring so that there is enough lead time for regular insulin and to distinguish that from the immediate pre-meal use of lispro insulin. The routine order section allows physicians flexibility in ordering single or mixed pre-meal doses, along with intermediate- or long-acting insulin administration at bedtime, if desired. The option of using basal twice-daily NPH insulin for patients who are NPO or on continuous enteral alimentation is also available.

The section on supplemental insulin is placed beneath the routine orders, along with a reminder to add the dose of short-acting insulin to the routine dose in order to minimize improper use of such insulin as a stand-alone sliding scale. We have noted that the doses of supplemental short-acting insulin ordered are often inadequate; therefore, a suggested initial dose algorithm is included.14

It is also necessary to close the feedback loop and allow for changes, if needed, in the routine orders for the following day. This should incorporate the current day’s experience with supplemental insulin or episodes of hypoglycemia. The final section instructs the nurse to call the covering provider in the evening if there has been usage of supplemental insulin or an episode of hypoglycemia. This should avoid unnecessary calls during the day or night.

An order form for the use of U-500 insulin has also been developed, for use when this concentrated specialty insulin is needed.


Nursing resistance to insulin infusion use on general medical/surgical units was overcome with persistent case-by-case education. We found that the apprehension of nurses on the general units was greatly alleviated by speaking to their colleagues in labor and delivery, who had quickly embraced the use of insulin infusion in obstetrics. Staff nurse concerns were reduced further when the insulin infusion orders were compared with a heparin infusion protocol with which they were already very comfortable.

Intensive care nurses were already experienced in the use of insulin infusions to treat patients with DKA or hyperosmolar nonketotic syndrome, so their use with other patients in the unit was a logical extension for them.

Infusion use in the operating room was initially hampered by several issues. Some anesthesiologists, despite having a number of lengthy vascular and neurological cases, were not convinced of the importance of tight blood glucose control during the operative and immediate perioperative period. A review of the pertinent literature presented at one of their staff meetings addressed this issue.

There were no glucose meters in the operating room, and a recovery room nurse would have to be called if a capillary glucose was needed.

The practical barrier of who would obtain the glucose readings in the operating room was overcome by meeting with the anesthesiology staff and the operating room administrator. Circulating nurse staff members were trained in the use of two glucose meters dedicated for operating room use. We are currently working with the department to add a section to the anesthesia record for the recording of blood glucose values and insulin administration. Insulin infusions are still underutilized in the operating room, but their use is steadily increasing, in no small part because of requests from the surgeons.

Other problems encountered in the institution of the entire program involved insulin infusion use in medical and surgical patients. Physicians, particularly those who had missed one or both of the relevant staff meetings, initially were unsure of what infusion rate (i.e., how many units/h) to begin with. Given this feedback, the infusion order form was amended to include a guideline for determining an approximate starting point for dosing in several common clinical situations. More recently, based on feedback from the departments of medicine and surgery, sample completed insulin infusion order forms were distributed to the medical staff. Two samples were included—one with an initial insulin infusion rate of 1 unit/h and one with an initial rate of 1.5 units/h, which cover the majority of routine medical and surgical situations.

Additional difficulties with the insulin infusions centered on adjustments to the initial orders when one of two situations presented. First, if the initial insulin infusion ranges were excessive or inadequate, nurses and physicians had to realize that the entire infusion order form would have to be rewritten to better cover the patient’s actual needs. For example, if a patient’s actual insulin requirements were 4 units/h, and the initial infusion orders only went up to a maximum of 3 units/h, the blood glucose level would remain unacceptable. Although, with increasing experience, physicians and nurses became better at realizing this, and a revised order form was adopted with a selectable initial infusion rate and incremental rate adjustments (Fig. 2).

Second, adjusting infusion rates after the initial adjustments were made was sometimes confusing with the first versions of the form, especially for nursing staff who had no previous experience with the orders. This was clarified by changing the order sheet to increase or decrease infusion rates in a predetermined fashion if the glucose was not in the target range (Fig. 2). With this enhancement, nurses more easily comprehend the orders, particularly if they have not worked with these forms previously.

A less critical concern was excessive blood glucose testing for patients who were on infusions for many days or even weeks. This resulted in increased cost, nursing time, and patient discomfort. As a patient’s situation stabilized, often associated with a fairly consistent rate of parenteral or enteral alimentation, we learned to decrease the frequency of glucose testing to every 3 h or less. Changing to basal coverage with subcutaneous NPH insulin administered every 12 h is also an option at this point.

Voluntary use of the subcutaneous/supplemental insulin order form has been adopted by most of the medical staff. Consequently, there has been a significant reduction in the use of sliding-scale insulin coverage without the use of basal insulin. The percentage of insulin orders using sliding-scale only insulin for patients in the hospital more than 1 day decreased from 42 to 29% during corresponding 3-month periods in 1999 and 2000 (P < 0.01, n = 252 and 256, respectively). These data include the patients of all physicians, not only those who used the new subcutaneous order forms, indicating a probable spillover effect from the education efforts.


The use of guideline-directed orders can facilitate more physiological use of insulin in the hospital setting. The effectiveness of these orders improved and resistance to change was overcome after ongoing education and protocol revisions based on feedback from physicians, nurses, and pharmacists.

Figure 1.

Insulin drip pharmacy label.

Figure 2.

Standard insulin infusion (0.1 unit/cc) order sheet (not for ketoacidosis or hyperosmolar states).

Figure 3.

Intrapartum insulin/dextrose infusion order sheet (for labor and delivery use only).

Figure 4.

Diabetic ketoacidosis admitting/insulin orders (for ER/ICU use, patients 16 y.o. and up: NOT for pediatric use).

Figure 5.

Subcutaneous/supplemental insulin orders.


The authors would like to acknowledge the enthusiastic assistance of Kristen Fiore, RN, who served as a liaison with the nursing staff and helped create the self-study guide.


  • Stephen F. Quevedo, MD, FACP, is director; Elaine Sullivan, MS, RN, CDE, is center manager; and Randi Kington, MS, APRN, CDE, is an adult nurse practitioner at the Joslin Diabetes Center Affiliate at Lawrence & Memorial Hospital. Warren Rogers, RPh, is director of pharmacy at Lawrence & Memorial Hospital in New London, Conn.

  • Note of disclosure: Ms. Kington has received honoraria for speaking engagements from Eli Lilly & Company and is a stock shareholder in Eli Lilly and Pfizer, Inc. Both companies manufacture insulin products for the treatment of diabetes.


| Table of Contents