Intensive Glucose Control in the ICU: An Expert Interview With James S. Krinsley, MD

Antonios Liolios, MD


April 12, 2004

Editor's Notes:
Recently published studies have indicated that intensive glycemic control in critically ill patients may have a significant effect on mortality in these patients. Some of the most current studies were discussed at the recent 33rd Critical Care Congress of the Society of Critical Care Medicine, which took place February 20-25, 2004, in Orlando, Florida, including an abstract by James S. Krinsley, MD,[1] who is Associate Clinical Professor of Medicine and Director of Critical Care at The Stamford Hospital in Stamford, Connecticut. His abstract was titled "Decreased Mortality of Critically Ill Patients With the Use of an Intensive Glycemic Management Protocol," and reported decreased mortality in a mixed intensive care unit (ICU) after implementing intensive insulin therapy. The study was conducted in a 305-bed community hospital with a 14-bed "mixed" ICU (medical, surgery, and cardiac patients). It was an "open" unit but highly data-conscious and protocol-driven. The house staff was medical and surgical. There were 800 consecutive admissions following institution of the protocol and 800 historical controls -- consecutive admissions preceding institution of the protocol. The goal was to maintain blood glucose at less than 140 mg/dL. Baseline hospital mortality rate was 20.9%. Treatment hospital mortality rate dropped to 14.8% (6.1% absolute reduction, 29.3% relative reduction, P = .002). New renal failure was seen in 12 patients in the baseline group and in 3 patients in the treatment group (P = .034). A total of 25.2% of the patients required blood transfusion in the baseline group and 20.5% in the treatment group (18.7% relative reduction; P = .035) The median length of stay (days) was 1.9 in the baseline group and 1.6 in the treatment group (P = .011). No difference was seen in the post-ICU length of stay. In this interview, Antonios Liolios, MD, Attending Staff Physician in the ICU at the University Hospital Saint Luc, Brussels, Belgium, discussed the issue of intensive insulin therapy with Dr. Krinsley.

Figure 1. James Krinsley, MD.

Figure 2. Antonios Liolios, MD.

Dr. Liolios: Dr. Krinsley, can you please tell us what prompted you to do your study?[1]

Dr. Krinsley: Van den Berghe's study[2] prompted me to examine my ICU's comprehensive database for the relationship between glucose values in the ICU and hospital mortality. These results have been published recently in the Mayo Clinic Proceedings.[3] There was a very strong relationship between increasing glucose values in the ICU and increasing hospital mortality. There is also an accompanying editorial.[4] Van den Berghe's study population was primarily postoperative cardiac surgery patients. There have been no similar studies published that describe the effect of an intensive glycemic management protocol in a medical or a mixed medical-surgical population of patients.

Dr. Liolios: What are your comments on the findings of the study by Finney and colleagues?

Dr. Krinsley: I believe that the study by Finney and coworkers[5] complements my Mayo Clinic Proceedings publication, suggesting that hyperglycemia was associated with increasing mortality of critically ill patients and that glycemic control was responsible for decreased mortality of these ill patients.

Dr. Liolios: Why did you use historical controls in your study?

Dr. Krinsley: The database allowed comprehensive and relatively expeditious analysis of the 800 consecutively admitted historical controls. I did not have the resources to perform a randomized, controlled clinical trial at my institution. My ICU is part of a university-affiliated community hospital. While we train medical and surgical residents, we do not have fellows and I do not have a research nurse or other staff directly assisting me in this work.

Dr. Liolios: How much do you think this limits the applicability of your findings?

Dr. Krinsley: Although the highest level of scientific evidence is a randomized, controlled, prospective trial, I believe that the findings of my study are generalizable and applicable to ICUs similar in configuration to mine, treating a mixed population of medical and surgical patients. This is, by the way, probably the most common configuration of ICUs in the 5000 or so hospitals in the United States. The two groups (baseline and treatment) were very well matched (age, gender, race, severity of illness reflected by Acute Physiology and Chronic Health Evaluation II scores, distribution of diagnoses, etc.). They represented consecutive admissions immediately preceding institution of the protocol compared with consecutive admissions following institution of the protocol; there were no other significant changes in clinical practice during these two periods.

Dr. Liolios: What was the duration of the study?

Dr. Krinsley: The 800 consecutive admissions comprising the control cohort and the 800 consecutive admissions comprising the study cohort each took approximately 11 months to accrue.

Dr. Liolios: Why did you choose 140 mg/dL as the cut-off level for applying intensive insulin therapy as opposed to the 80- to 110-mg/dL goal selected in the study by Van den Berghe and colleagues?[2]

Dr. Krinsley: The data published in my Mayo Clinic Proceedings article[3] indicated that the mean glucose level of hospital survivors was 138, while the mean glucose level of the nonsurvivors was 172. The use of 110 as a goal increased the risk of iatrogenic hypoglycemia. Van den Berghe's 2001 study[2] was not designed to test 110 vs 140; no other data in the literature supported such an extreme goal (ie, 110). Successfully implementing an intensive glucose management protocol is hard work. It requires extra effort by the nurses -- close monitoring and frequent interventions. It requires a change in culture; typical ICU practice before the last several years ignored even moderate hyperglycemia (eg, up to 200 to 225 mg/dL). Therefore, we had to achieve "buy-in" by the entire staff. Our nurses were comfortable with a goal of 140 and willing to do the work to achieve it. Our data strongly support their success.

Dr. Liolios: Is intensive insulin control an integral part of your ICU care now?

Dr. Krinsley: Our protocol has been in place since February 2003. It applies to all patients, regardless of presence or absence of known diabetes.

Dr. Liolios: The Van den Berghe study[2] focused on mechanically ventilated patients. Do you think that patients who do not have respiratory failure will still benefit as much from intensive insulin therapy?

Dr. Krinsley: Our data suggest that critically ill patients without respiratory failure and the need for mechanical ventilation also benefit from intensive glycemic management. Subgroup analysis of my study suggested an "across-the-board" benefit.

Dr. Liolios: How often do you see hypoglycemia in your ICU? Did you have any major complications?

Dr. Krinsley: The percentage of glucose values less than 40 mg/dL was .35% in the control period and .34% in the treatment period. There were no adverse effects in either period attributable to hypoglycemia.

Dr. Liolios: How much has the intensive insulin approach added to your nursing staff's workload?

Dr. Krinsley: Our protocol is locally designed. We consider it a "starting point" for glycemic control of our patients. Many patients require modifications of the protocol (ie, either more or less insulin than is indicated by the protocol). Glucose levels and insulin administration are charted on a unified flowsheet. The protocol is nurse-driven. Nurses administer insulin based on the protocol parameters; if there is a variance, a physician (medical or surgical house staff) is required to cosign the protocol flowsheet. The increased monitoring (frequent fingersticks) and increased insulin treatment, including the frequent use of continuous insulin infusions (required if the glucose value exceeds 200 mg/dL on 2 successive occasions) has definitely increased the workload of the nurses and the nursing assistants (who perform many of the fingerstick checks).

Dr. Liolios: You mentioned fingersticks. In the original Van den Berghe paper,[2] arterial blood was used for blood glucose measurement. Aren't you concerned that fingerstick blood glucose determination may not be as reliable in an edematous, vasoconstricted, and poorly perfused critically ill patient?

Dr. Krinsley: I agree that arterial blood provides a more accurate measure of glucose levels than fingersticks in certain situations. However, routine arterial line insertion is not a policy that many ICUs would be willing to adopt, due to the associated risk and cost. The use of fingerstick measurements contributed strongly to the "achievable" and "generalizable" nature of my study. This was work that was done in a real-life ICU without research fellows or research nurses.

Dr. Liolios: What was the extra cost? What protocol do you use?

Dr. Krinsley: The protocol is very inexpensive. We have not hired additional staff to perform the protocol. We have purchased a few extra bedside glucose monitors. The use of insulin has increased, but the cost of this drug is minimal. An interesting contrast is the use of drotrecogin alfa (activated). This drug was shown to yield a 6% absolute and 20% relative reduction in mortality of very sick patients with sepsis; it costs at least $10,000 per use in our institution. We achieved a 6.2% absolute and 29.3% relative reduction in hospital mortality in our study at a tiny fraction of this cost.

Dr. Liolios: Did you encounter any unexpected problems while trying to apply the intensive insulin therapy?

Dr. Krinsley: No.

Dr. Liolios: What would you advise the staff of an ICU that has decided to implement this protocol?

Dr. Krinsley: This question has forced me to think about why my ICU has achieved success with this initiative, while many ICUs are still talking about "doing glucose." The necessary components for effective promulgation of this ambitious and difficult program include:

  • A culture of multidisciplinary cooperation and good communication between nurses and physicians;

  • A strong leadership bond between the nursing director and the medical director of the unit, with a shared vision for the unit;

  • A culture that accepts standardization of care using best available medical evidence. Our unit attempts to "protocolize" all routine aspects of care. Adding the glucose protocol to the numerous other protocols in place in our unit was a natural outgrowth of the data that we generated documenting a strong relationship between increasing glucose values during ICU admission and hospital mortality;

  • The choice of achievable goals. We had initially discussed using Van den Berghe's goals, but that met with tremendous resistance and reluctance among the nurses. Without their total buy-in, the extra effort imposed by the protocol would not have been expended;

  • An ICU data management system that allows regular feedback of progress and outcomes. The nurses in my unit are barraged on a nearly continual basis with a blizzard of multicolor charts and graphs documenting our protocol's progress with glycemic control and the associated improvement in mortality and morbidity. This validates their work and ensures ongoing participation in the initiative.

Dr. Liolios: Could you share your protocol with us?

Dr. Krinsley: Sure. We consider our protocol a starting point, or road map, for glycemic management of our patients. Many patients require variances to the protocol. The protocol is nurse-driven; medical and surgical house staff are required to countersign any variances that are required. All patients admitted to the unit are treated with the protocol.

Dr. Liolios: Based on your work and the study by Van den Berghe, would you recommend intensive insulin therapy for all ICU patients?

Dr. Krinsley: Yes. I believe that the results we achieved in our institution are generalizable and achievable. Intensive glycemic management of critically ill patients will become the standard of care.


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.