- Appreciate the advantages and potential complications of intensive insulin therapy in the hospitalized patient
- List hospital-based safeguards available to prevent insulin-related hypoglycemia
- Understand the difference between efficacy and effectiveness with regards to results of clinical trials
Case & Commentary: Part 1
A 28-year-old man with insulin-dependent diabetes mellitus was admitted with hyperglycemia and an infected foot ulcer. Due to new hospital initiatives aimed at tighter glucose control, he was started on an insulin drip rather than subcutaneous insulin. The patient eventually required 8 units of regular insulin per hour to maintain a fingerstick glucose in the 180-220 mg/dL range.
Traditionally, only patients hospitalized for acute decompensation of diabetes were treated with intravenous (IV) insulin drips. More recently, intensive IV insulin therapy is being used in post-surgical patients for whom diabetes or hyperglycemia is a co-morbid, rather than a primary, condition. Emerging literature supports this new approach. In one study, the wound infection rate was reduced from 3% to 1% using a nurse-managed IV insulin protocol targeted to keep blood glucose less than 200 mg/dL for 72 hours after heart surgery.(1) In another, 1500 critically ill surgical patients were randomized to intensive (IV drip for blood glucose > 110 mg/dL, goal 80-110 mg/dL) and traditional (IV insulin reserved for blood glucose > 215 mg/dL) treatment of hyperglycemia.(2) The intensively treated group had about 50% lower ICU and hospital mortality rates, shorter hospital stays, and less morbidity. Although a few episodes of hypoglycemia were noted in the intensively managed group, there were no serious complications. These studies have generated a new consensus regarding the benefits of tighter glucose control in surgical ICU patients.
Given these results, and other studies demonstrating that hyperglycemia greater than 200 mg/dL is associated with more infections in patients undergoing "clean" surgery and in outpatients (and in vitro studies that show that ambient glucose over 200 mg/dL reduces macrophage migration and phagocytosis [3,4]), it is tempting to extrapolate that tight control will benefit medical, and not just surgical, inpatients. However, there are no data to support this extrapolation. Without demonstrated benefit in these broader populations, any efforts to improve glycemic control in such patients (including the one presented here) must place a premium on safety.
In non-critically ill hospitalized patients with hyperglycemia, subcutaneous insulin may be safer than an IV insulin drip. Physiologic insulin replacement requires a basal (delayed or extended absorption) form of insulin, especially for insulin-dependent (type 1) diabetes. This basal insulin is supplemented before meals with fast-acting insulin. New insulin analogues (such as insulin glargine [Lantus], insulin aspart [Novolog] and insulin lispro [Humalog]) meet both basal and mealtime insulin needs. For example, a once-daily dose of insulin glargine will meet the basal requirement and need not be held or adjusted for NPO status; whereas, doses of insulin aspart or insulin lispro will meet mealtime requirements and are easily adjusted for the patient's nutritional intake. Since the physiologic attractiveness of using subcutaneous insulin in the non-ICU patient now conflicts with the potentially generalizable benefits of IV insulin drips seen in the ICU, we believe that we are at equipoise, and thus the two methods should be compared in randomized trials.
Case & Commentary: Part 2
On hospital day 2, the patient was evaluated by an orthopedic surgeon, who decided the wound needed surgical debridement. The orthopedic resident notified the cross-covering medicine resident that the patient would be taken to the operating room (OR) the following day. In preparation for the operation, the orthopedic service made the patient "NPO after midnight." At 2 am, when the nurse came to measure the hourly fingerstick, she found the patient somnolent, tremulous, and diaphoretic. His fingerstick glucose was 20 mg/dL.
Hypoglycemia is the most common complication of any insulin therapy and is an extremely frequent adverse event in hospitals worldwide.(5) In the perioperative population, oral intake is highly variable; patients receiving insulin therapy are at particularly high risk for hypoglycemia in this period. Broader use of insulin drips outside the ICU requires that safeguards be put in place to prevent a potential increase in the frequency of hypoglycemia. The clinical trials cited used strict protocols (1,2); most hospitals do not have such guidelines in place (H.R.R., unpublished data, 2004). Protocols for insulin drips should be created, which clearly outline proper adjustment of the drip at specific glucose levels, alterations in rate and IV fluid type if the patient is NPO, and frequency of glucose monitoring.(6,7) Automated orders and preprinted order sheets have been effective in reducing chemotherapy-dosing errors and in ensuring appropriate therapy for myocardial infarction in the emergency department.(8,9) Such interventions could be instituted for insulin dosing. Where computerized physician order entry (CPOE) and computerized medication administration records are available, standard orders could be brought to the screen whenever a patient with diabetes is admitted or an insulin drip is ordered or begun.
Without seeing the specific insulin drip order in this case, it is difficult to know how it might have been improved to avoid the adverse event. However, carefully developed decision support, to assist in adjusting insulin drips, and standardization of glucose monitoring in patients on IV insulin drips might have helped prevent this error. In an even more robust system of decision support, both orders (NPO status and insulin therapy) would have been entered into a CPOE system and adjustments suggested automatically. A pop-up flag could have alerted the surgeon to ensure the insulin drip was appropriate for the situation, or that the primary care physician be notified. Finally, one can even envision a "forcing function," in which IV insulin is automatically discontinued, or a dextrose drip started, when an NPO order is written (with the physician immediately notified), further decreasing the chance of a mistake harming a patient.(10)
Results obtained in clinical trials often achieve rates of adverse events lower than those found in routine clinical practice. There are several reasons for these differences. First, patients in day-to-day practice are rarely monitored as closely as those in trials. Secondly, clinicians may administer therapies to patients who would have been excluded from trials.(11). This broadening of inclusion criteria can lead to a lesser benefit in actual practice, an issue known as the difference between the "efficacy" vs. "effectiveness" of an intervention.(11-14) We believe that it will be important for the safety literature to distinguish between patient harm in a clinical trial vs. "practical safety"—that which can be expected in real world practice.
In this case, the orthopedic surgeon wrote the NPO order, and did not notice the insulin drip. Allowing consulting services to write orders on patients may lead to poor communication, which can then result in errors and poor outcomes (15): through delays in diagnosis, unnecessary or duplicative diagnostic tests, delays in needed treatment, or unnecessary treatment. In the absence of "smart" information systems, it may be helpful to provide a procedural forcing function. For example, a hospital policy might require that the primary physician approve all new orders (obviously, such a policy will raise important staffing issues). The widely adopted hospitalist model (16) may reduce the potential for error that results from multiple physicians writing orders, as hospitalist services can be designated to write all orders on inpatients.
Case & Commentary: Part 3
The patient was treated with 1 amp of D50 and his insulin drip was held. He recovered completely from this event.
Increasingly, quality improvement efforts focus on decreasing the extent of "underuse" of effective therapies. However, even as we focus on ensuring that such therapies are uniformly used, we must remember that harm can result if no protocols or systems are in place to help monitor for and prevent their common side effects. Quality improvement efforts, exemplified by those of the Institute for Healthcare Improvement in its "Breakthrough Collaboratives" (17), suggest that a single, central outcome should be monitored to help make quality improvement projects practical and do-able. Some refer to this as "moving the dot," derived from the popular run chart illustrating changes over time in a particular parameter.(18)
Assuming that "the dot" was the level of glucose control, this case sounds a cautionary note regarding such an approach. When our quality and safety improvement projects focus on correcting underuse of particular medications or therapies, we must carefully consider adverse events that may be increased by promoting a particular treatment, and implement safeguards to avoid them. Like physicians, quality improvement professionals will serve patients best if we focus on the Hippocratic mantra of primum non nocere: first, do no harm.
- Insulin, both intravenous and subcutaneous, can be a safe and effective means of controlling blood glucose levels in hospitalized patients when appropriate standard protocols are followed.
- Recent data support the practice of tighter control (through the use of IV insulin drips) in the surgical ICU. The extent to which such data can be generalized to medical and surgical patients on the general medical-surgery ward is unknown.
- Protocols for tighter glucose control must incorporate adequate monitoring frequency to avoid hypoglycemia.
- Hospitals should implement hospital-wide standard insulin drip protocols that have been tested and shown to be safe.
- Various strategies, including standard order sets, decision support, and computerized forcing functions, can be used to help prevent adverse events.
- Safety in clinical trials differs from safety in practice.
- Quality improvement efforts that focus on increasing use of medications and "tight control" of physiologic parameters must incorporate efforts to monitor and prevent the most common adverse events of over-treatment of those parameters.
Haya R. Rubin, MD, PhD Professor of Medicine Director, Quality of Care Research The Johns Hopkins University
Vera T. Fajtova, MD Assistant Professor of Medicine, Harvard Medical School Medical Director, Diabetes Management Service, Brigham and Women's Hospital
Faculty Disclosure: Drs. Rubin and Fajtova have declared that neither they, nor any immediate member of their family, have a financial arrangement or other relationship with the manufacturers of any commercial products discussed in this continuing medical education activity. In addition, their commentary does not include information regarding investigational or off-label use of pharmaceutical products or medical devices.
1. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous intravenous insulin infusion reduces the incidence of deep sternal wound infection in diabetic patients after cardiac surgical procedures. Ann Thorac Surg. 1999;67:352-62.[ go to PubMed ]
2. Van den Berghe G, Wouters P, Weekers F, et al. Intensive insulin therapy in the critically ill patients. N Engl J Med. 2001;345:1359-67.[ go to PubMed ]
3. Pomposelli JJ, Baxter JK 3rd, Babineau TJ, et al. Early postoperative glucose control predicts nosocomial infection rate in diabetic patients. JPEN J Parenter Enteral Nutr. 1998;22:77-81.[ go to PubMed ]
4. Rayfield EJ, Ault MJ, Keusch GT, Brothers MJ, Nechemias C, Smith H. Infection and diabetes: the case for glucose control. Am J Med. 1982. 72:439-50.[ go to PubMed ]
5. Runciman WB, Roughead EE, Semple SJ, Adams RJ. Adverse drug events and medication errors in Australia. Int J Qual Health Care. 2003;15 Suppl 1:i49-59.[ go to PubMed ]
6. Surgical services. The Portland protocol for continuous intravenous insulin infusion in post operative diabetic cardiac surgery patients. The Starr Wood Web site. Available at: [ go to related site ]. Accessed April 20, 2004.
7. Managing diabetes in hospitalized patients. Brigham and Women's Hospital. Division of Pharmacoepidemiology and Pharmacoeconomics. Medication Education Program. Available at: [ 5).pdf" target="_blank">go to related site ]. Accessed April 20, 2004.
8. Lillis K. Automated dosing. Computerized physician order entry reduces risk of medication and dosing errors in neonatal ICU. Health Manag Technol. 2003;24:36-7.[ go to PubMed ]
9. Bradley EH, Holmboe ES, Wang Y, et al. What are hospitals doing to increase beta-blocker use? Jt Comm J Qual Saf. 2003;29:409-15.[ go to PubMed ]
10. Farbstein K, Clough J. Improving medication safety across a multihospital system. Jt Comm J Qual Improv. 2001;27:123-37.[ go to PubMed ]
11. Katzan IL, Furlan AJ, Lloyd LE, et al. Use of tissue-type plasminogen activator for acute ischemic stroke: the Cleveland area experience. JAMA. 2000;283:1151-8.[ go to PubMed ]
12. Mohr JP. Thrombolytic therapy for ischemic stroke: from clinical trials to clinical practice. JAMA. 2000;283:1189-91.[ go to PubMed ]
13. Cebul RD, Snow RJ, Pine R, Hertzer NR, Norris DG. Indications, outcomes, and provider volumes for carotid endarterectomy. JAMA. 1998;279:1282-7.[ go to PubMed ]
14. Wennberg DE, Lucas FL, Birkmeyer JD, Bredenberg CE, Fisher ES. Variation in carotid endarterectomy mortality in the Medicare population: trial hospitals, volume, and patient characteristics. JAMA. 1998;279:1278-81.[ go to PubMed ]
15. Pollack MM, Koch MA; NIH-District of Columbia Neonatal Network. Association of outcomes with organizational characteristics of neonatal intensive care units. Crit Care Med. 2003;31:1620-9.[ go to PubMed ]
16. Wachter RM. An introduction to the hospitalist model. Ann Int Med. 1999;130:338-42.[ go to PubMed ]
17. Collaboratives. The Institute for Healthcare Improvement Web site. Available at: [ go to related site ]. Accessed April 20, 2004.
18. Improvement Tracker. QualityHealthCare.org Web site. Available at: [ go to related site ]. Accessed April 20, 2004.