| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
From the Departments of *Anesthesiology and Critical Care Medicine, Endocrinology, and Cardiothoracic and Vascular Surgery, Onze-Lieve-Vrouw Hospital, Aalst, Belgium.
Address correspondence and reprint requests to L. Foubert, MD, PhD, Department of Anesthesiology and Critical Care Medicine, Onze-Lieve-Vrouw Hospital, Moorselbaan 164, 9300 Aalst, Belgium. Address e-mail to Luc.Foubert{at}olvz-aalst.be.
Abstract
BACKGROUND: Tight blood glucose control reduces mortality and morbidity in critically ill patients, but intraoperative glucose control during cardiac surgery is often difficult, and risks hypoglycemia. In this study, we evaluated the safety and efficacy of a nurse-driven insulin protocol (the Aalst Glycemia Insulin Protocol) for achieving a target glucose level of 80–110 mg/dL during cardiac surgery and in the intensive care unit (ICU).
METHODS: We included 483 nondiabetics and 168 diabetics scheduled for cardiac surgery with cardiopulmonary bypass. To anticipate rapid perioperative changes in insulin requirement and/or sensitivity during surgery, we developed a dynamic algorithm presented in tabular form, with rows representing blood glucose ranges and columns representing insulin dosages based on the patients’ insulin sensitivity. The algorithm adjusts insulin dosage based on blood glucose level and the projected insulin sensitivity (e.g., reduced sensitivity during cardiopulmonary bypass and normalizing sensitivity after surgery).
RESULTS: A total of 18,893 blood glucose measurements were made during and after surgery. During surgery, the mean glucose level in nondiabetic patients was within targeted levels except during (112 ± 17 mg/dL) and after rewarming (113 ± 19 mg/dL) on cardiopulmonary bypass. In diabetics, blood glucose was decreased from 121 ± 40 mg/dL at anesthesia induction to 112 ± 26 mg/dL at the end of surgery (P < 0.05), with 52.9% of patients achieving the target. In the ICU, the mean glucose level was within targeted range at all time points, except for diabetics upon ICU arrival (113 ± 24 mg/dL). Of all blood glucose measurements (operating room and ICU), 68.0% were within the target, with 0.12% of measurements in nondiabetics and 0.18% in diabetics below 60 mg/dL. Hypoglycemia < 50 mg/dL was avoided in all but four (0.6%) patients (40 mg/dL was the lowest observed value).
CONCLUSIONS: The Aalst Glycemia Insulin Protocol is effective for maintaining tight perioperative blood glucose control during cardiac surgery with minimal risk of hypoglycemia.
Glycemic control in cardiac surgical patients has become an integral part of standard care as a means for reducing infections and, possibly, for improving patient outcomes.1–4 In a landmark trial, Van den Berghe et al.5 demonstrated that tight glycemic control (glucose 80–110 mg/dL) decreased in-hospital mortality by 34% in a mixed medical-surgical intensive care unit (ICU) population that included cardiac surgery patients.5 These benefits were observed only in patients admitted to the ICU for >5 days. The benefits of strict glycemic control during cardiac surgery, however, remains controversial with one study suggesting higher mortality for patients receiving insulin to achieve similar targeted blood glucose levels.6,7 In contrast, intraoperative glycemic control to maintain blood glucose levels <200 mg/dL has been associated with decreased mortality and sternal wound complications in diabetics.8 Strict intraoperative glucose control was further found to significantly reduce cardiovascular, pulmonary, and renal morbidity in diabetic cardiac surgical patients,9 whereas hyperglycemia during cardiopulmonary bypass (CPB) was found to be an independent risk factor for mortality.10 Others suggest no benefit on neurological complications with an aggressive insulin protocol during cardiac surgery.11
Cardiac surgery induces counter regulatory hormone release and alterations in carbohydrate metabolism, such as enhanced hepatic gluconeogenesis, insulin resistance, and relative insulin deficiency.12,13 These responses to surgical stress often make maintaining euglycemia difficult during surgery. Furthermore, aggressive insulin administration during surgery risks hypoglycemia after surgery when surgical stress abates.14 Many protocols for insulin administration have been proposed (Table 1), but these protocols do not balance the differing insulin demands of the intraoperative and postoperative periods. Furthermore, many of the proposed insulin regimens were exclusively evaluated in diabetic or in nondiabetic patients.5,8,9,15–18 In addition, most protocols either accept a risk of hypoglycemia in attempting to achieve strict glycemic control or accept higher glucose levels than those associated with improved outcomes.8,9,14,17 Indeed, the incidence of serious hypoglycemic events (blood glucose <40 mg/dL) in these reports ranges from 3.4% to 7.7% of patients.5,9,15,17 Because of the static nature of many protocols that depend on the results of blood glucose testing, there is a delay in the adaptation of the insulin dosage, resulting in either hyperglycemia or severe hypoglycemia. Finally, in many studies, the protocols were executed only by study nurses and/or by study physicians not involved in the clinical care of the patients.5,7,8 Whether these strategies are applicable to the day-to-day care of surgical patients is thus unclear.
|
We sought to evaluate whether blood glucose levels of both diabetics and nondiabetics can be controlled between 80 and 110 mg/dL by our regular nursing staff, with minimal risk for hypoglycemia. We modified an existing algorithm20 to anticipate the varying insulin needs that occur during and after the various stages of cardiac surgery with CPB (the Aalst Glycemia Insulin Protocol, ALGIP).
METHODS
After IRB approval and written informed consent from the patient, 651 consecutive patients (168 diabetics) scheduled for elective coronary artery bypass and/or cardiac valve surgery with the use of CPB were enrolled in this prospective, open label, single-center study between June 2005 and May 2006. Nondiabetics were defined as patients not previously treated for diabetes mellitus with a fasting glucose <125 mg/dL.21 Patients treated for diabetes mellitus, as well as patients not previously known as diabetics, but with a fasting glucose 
125 mg/dL, were considered diabetics, according to international guidelines.21 Cardiac medication was continued until the morning of surgery, except for angiotensin-converting enzyme inhibitors. Patients were premedicated with 1 mg lorazepam orally. Anesthesia was induced with a combination of propofol, diazepam, fentanyl, or remifentanil according to the preference of the attending anesthesiologist. Tracheal intubation was facilitated by 1 mg/kg of rocuronium bromide and mechanically controlled ventilation was adjusted to an end-tidal ETco2 of 35 mm Hg. Anesthesia was maintained with 0.8 to 1.5 MAC of sevoflurane and/or IV propofol targeting a plasma concentration of 1.5 to 3 µg/mL. IV cisatracurium was administered at a rate of 0.75 µg · kg–1 · min–1. All drugs were diluted in normal saline. Hypothermic (temperature 28°C), nonpulsatile CPB between 2.4 and 3.0 L · min–1 · m–2 with a membrane oxygenator was used. The CPB prime consisted of 1.8 L prime (0.9 L lactated Ringer’s solution, 0.9 L Gelofusine®, 50 mEq sodium bicarbonate, and 12.5 g mannitol). Alpha-stat blood gas management was used. Antegrade and/or retrograde cardioplegia (St. Thomas solution, 4–6°C) was used at the surgeon’s discretion. Separation from CPB was facilitated with IV inotropic and/or vasoactive drugs at the discretion of the attending anesthesiologist. The trigger for blood transfusion was any hemoglobin <8.5 g/dL. Postoperative care was standardized for all patients. Sedation was stopped 6 to 7 h after arrival in the ICU according to our institutional protocol.
The ALGIP Algorithm
Glycemic control using ALGIP was initiated after induction of anesthesia. This dynamic algorithm is presented in a tabular form, with rows representing blood glucose ranges and columns representing insulin dosages based on the patients’ insulin sensitivity (Fig. 1). Insulin (Actrapid®, Novo Nordisk, Brussels, Belgium) and 5% dextrose (1 mL · kg–1 · h–1) were given via separate lumens of a central IV catheter. During surgery, a nurse measured blood glucose levels from arterial blood using an on-site blood-gas analyzer (ABL715, Radiometer Medical, Copenhagen, Denmark) every 30 min and the insulin infusion rates were modified as follows. If glucose was between 85 and 110 mg/dL (green area), the insulin dose is adjusted within the same column. If glucose is above 110 mg/dL, insulin dosage is increased by moving one column to the right, thus considering the intrinsic insulin-sensitivity. If blood glucose level decreases by at least 1 range denoted in the rows, but is still above 110 mg/dL, the insulin dosage is decreased within the same column. At a blood glucose level between 70 and 84 mg/dL, insulin dosage is decreased by using the dose noted on the next leftward column. The lower limit of the green area is deliberately set at 85 mg/dL instead of 80 mg/dL. This allows for the insulin dosage to be decreased at a blood glucose of 84 mg/dL to reduce the risk of hypoglycemia. Based on our preliminary data (Fig. 2), insulin dosage is preemptively increased by moving three columns to the right (within the same row) at the moment rewarming is initiated on CPB. When the esophageal temperature is 36°C, insulin dosage is reduced to the dose 3 columns to the left.
|
|
Guidelines for managing hypoglycemia are provided at the bottom of the Figure. Adjustment of the insulin infusion rate and dextrose 30% is given IV in different doses beginning when blood glucose is between 60 and 69 mg/dL. We defined hypoglycemia as a blood glucose <60 mg/dL to remain above levels associated with adrenergic symptoms or cognitive dysfunction in nonanesthetized or sedated patients (glucose <55 mg/dL).22
To anticipate a decrease in insulin requirements with reduced surgical stress, upon arrival to the ICU, the insulin infusion rate is decreased to the dose depicted in one column to the left of that used at the end of surgery. Blood glucose levels were measured every 60 min using blood from a fingerprick, using a bedside plasma calibrated glucose analyzer (Glucotouch®, Lifescan, Beerse, Belgium). When four consecutive glucose levels were within the green area of Figure 1, glucose measurements were performed every 2 h. The ALGIP protocol was continued until enteral feeding was restarted and/or patients were discharged from the ICU. At that time, the IV insulin infusion was stopped and subcutaneous insulin started using an institutional protocol. Since most of our cardiac surgical patients are discharged from the ICU within 24 h after surgery, the study period was confined to the first 24 h.
The following outcome data were registered from the individual case report forms: blood glucose levels, patients and number of measurements within target, incidence of hypoglycemia, time to target after arrival in the ICU, insulin consumption as well as the coefficient of glucose variability, recently suggested to be an independent predictor of ICU and hospital mortality.23,24
Statistical Analysis
Patient characteristics, differences in mean blood glucose level and mean insulin consumption, incidence of hypoglycemia, coefficient of glucose variability, percentage of measurements within target, and the time to reach glycemia target were tested between nondiabetics and diabetics at all time points with a two-sample Student’s t-test, Wilcoxon two-sample test or a Fisher’s exact test as appropriate. P values <0.05 were considered statistically significant.
RESULTS
Demographic and clinical characteristics of the patients are listed in Table 2. Of the 168 diabetics, 29 were receiving insulin treatment, 81 were noninsulin-treated diabetics, and 58 patients were not previously known diabetics. Diabetics had higher Body Mass Index, HbA1C values, fasting blood glucose, and glucose on anesthesia induction.
|
A total of 18,893 blood glucose levels were measured: 5356 during surgery and 13,537 in. the ICU. In nondiabetics, the mean blood glucose during surgery was within the targeted range during all study time points, except during (112 ± 17 mg/dL) and after rewarming (113 ± 19 mg/dL) on CPB (Fig. 3). The highest glucose level during CPB was 246 mg/dL but 95% of the measurements were below 147 mg/dL. In the ICU, the mean blood glucose level in nondiabetics was within the targeted range at all time points. Overall, 93.0% of measurements were 
130 mg/dL both during and after surgery.
|
In diabetics, the mean blood glucose level at the end of surgery (112 ± 26 mg/dL) was lower compared with after anesthesia induction (121 ± 40 mg/dL; P < 0.05), but higher than in the nondiabetic group (105 ± 18 mg/dL; P < 0.001). The highest blood glucose level during CPB was 293 mg/dL, but 95% of the measurements were below 163 mg/dL. In diabetics, the mean blood glucose level was above the targeted range on arrival in the ICU (113 ± 24 mg/dL). After 1 h in the ICU, the mean blood glucose level of diabetics was not different from values from nondiabetics for the remaining 24 h (P = 0.17) (Fig. 3).
The distribution of patients with a blood glucose level above, within, or below the targeted range at any given time point during or after surgery is shown in Figures 4A and B. The overall success rate (operating room and ICU) of the ALGIP, expressed as the percentage of all blood glucose measurements within target, was 68% with 71.0% in nondiabetics and 59.5% in diabetics, respectively (P < 0.0001). Overall, 22.7% of measurements in nondiabetics and 24.9% in diabetics were between 110 and 130 mg/dL (Figs. 4A and B). The success rate during surgery was significantly higher in nondiabetics than in diabetics (67.7% vs 41.2%; P < 0.0001). Similarly, 22.1% of measurements in nondiabetics and 26.5% in diabetics were between 110 and 130 mg/dL (P < 0.0001). Upon arrival in the ICU, 72.0% of nondiabetics had a blood glucose level between 80 and 110 mg/dL compared with 54.2% in the diabetics group (P = 0.002) (Fig. 4B). During hospitalization in the ICU, the success rate of nondiabetics and diabetics was 72.4% and 66.2%, respectively (P < 0.001). For patients with a blood glucose level outside of the targeted range upon ICU arrival, a median of 3 h [2–14] in nondiabetics and 3 h [2–20] in diabetics was needed to reach target (P = 0.06).
|
In the nondiabetic patients, the incidence of intraoperative hypoglycemia (glucose <60 mg/dL) was 1 of 4016 measurements (0.02%). In the ICU, hypoglycemia occurred in 16 of 9913 measurements (0.16%), of which 87.5% were between 50 and 59 mg/dL. The lowest glucose level in nondiabetics during surgery and in the ICU was 52 and 40 mg/dL, respectively. In nondiabetics, there were no patients with more than 1 hypoglycemic event during surgery as well as in the ICU. In diabetic patients, 1 of 1340 (0.07%) measurements during surgery and 8 of 3624 measurements (0.22%) in the ICU were below 60 mg/dL. The lowest measured blood glucose level during surgery and in the ICU was 56 mg/dL and 40 mg/dL, respectively. There were no diabetics with more than 1 hypoglycemic event both during surgery and in ICU.
To demonstrate the stability in time of mean blood glucose level, the coefficient of variation was calculated for each patient, according to Egi et al.23 Intraoperatively, the coefficient of blood glucose variation was 15.4% ± 3.2% in nondiabetics and 16.9% ± 5.9% in diabetics (P = 0.01). Postoperatively, however, there was no difference in blood glucose variability between diabetics and nondiabetics (P = 0.32). Logarithmic regression at all time points shows that the coefficient of blood glucose variability of both groups was converging (Fig. 5).
|
The mean insulin consumption in diabetics was higher than in nondiabetics, both during surgery (4.5 ± 2.3 vs 3.2 ± 1.5 IU/h; P < 0.0001) as in the ICU (2.8 ± 1.4 vs 2.3 ± 1.2 IU/h; P < 0.001). In the ICU, insulin-treated diabetics needed more insulin than noninsulin-treated diabetics (4.1 ± 1.6 vs 2.3 ± 1.1 IU/h; P < 0.0001). The use of inotropic drugs during weaning from bypass or in the ICU was comparable between groups (Table 2). Intraoperatively, there were no protocol violations. In the ICU, major protocol violations occurred in 0.7% of all measurements, of which 91.0% occurred during the first 6 wk after the introduction of the algorithm.
DISCUSSION
Our results show that the use of the ALGIP during and after cardiac surgery with CPB was effective in maintaining blood glucose levels within the targeted ranges of 80 to 110 mg/dL in 71.0% of nondiabetics and in 59.5% of diabetics. The success rates of the ALGIP during surgery and in the ICU were significantly higher in nondiabetics than in diabetics, 67.7% vs 41.2% during surgery (P < 0.0001), and 72.4% and 66.2% in the ICU (P < 0.001), respectively. Overall, 88.9% of blood glucose measurements of nondiabetics and 78.9% of diabetics were between 80 and 130 mg/dL. Hypoglycemia occurred in 0.12% of measurements in nondiabetics and in 0.18% of measurements in diabetics. Hypoglycemia <50 mg/dL was avoided in all but four patients (0.6%), with 40 mg/dL as the lowest observed value.
Complex metabolic derangements occur during cardiac surgery particularly with the use of hypothermic CPB.25–27 Insulin resistance results from counter-regulatory mechanisms invoked by the stress of surgery and hypothermia including increased plasma levels of glucagon, catecholamines, growth hormone, and cortisol.25,26 This leads to decreased insulin-stimulated skeletal muscle glycogen synthesis, which can mostly be attributed to decreased glucose transport (Glut 4) receptor sensitivity.27 Our preliminary data showed that anticipation of changes in insulin sensitivity during key portions of surgery may be more effective for achieving intraoperative normoglycemia than other protocols in which insulin dose is adjusted based on glucose levels only.5,8,9,14,15,19 After rewarming on CPB, the nondiabetics treated with ALGIP had a significantly lower mean glucose level than patients without the proactive approach, 113 ± 19 mg/dL vs 141 ± 18 mg/dL, respectively (P < 0.0001) (Fig. 2). Acutely increasing insulin dosage at rewarming on CPB and then decreasing insulin dosage when the temperature reaches 36°C counters the effect of transient insulin resistance during rewarming. These modifications enabled us to achieve normoglycemia (80–110 mg/dL) in 67.4% of all patients at arrival in the ICU. Using ALGIP, patients who arrive in the ICU with a glucose level above 110 mg/dL were within the targeted range after a median of 3 h [2–20], which is shorter than 13.6 h reported in a previous investigation.9 Most probably, this difference can be attributed to the fact that our patients were already well controlled during surgery. Weaning from mechanical ventilation and subsequent tracheal extubation in the ICU appears to induce a stress response, temporarily increasing mean glucose (Fig. 3) and decreasing the percentage of patients within the target blood glucose range (Figs. 4A and B). However, within 2 h after tracheal extubation, mean glucose levels were restored to the pre-extubation values (Fig. 3).
Hypoglycemia is a major concern with aggressive insulin administration in anesthetized and sedated patients who cannot report symptoms of low blood glucose.28 The incidence of hypoglycemia in this study is lower compared with other investigations (Table 1). None of our patients experienced hypoglycemia <40 mg/dL and only 4 of 651 patients (0.61%) had a blood glucose <50 mg/dL, when compared with 1.2% to 7.1% below 40 mg/dL, and 7.7% below 50 mg/dL in other studies.5,9,15,17,19
Egi et al.23 showed in a large multicenter cohort that the benefit of intensive insulin therapy is not only related to a reduction in the average glucose level, but also to a reduction in the variability of glucose levels perioperatively. They described blood glucose variability as an independent predictor of ICU and in-hospital mortality. Using the ALGIP, we found that the coefficient of blood glucose variation in the ICU is lower than that in the survivors-group described by Egi et al.23 In some patients, especially in diabetics, the time needed to reduce fluctuations in blood glucose is long, nearly the duration of surgery. However, it is not clear whether decreasing glucose variability during surgery improves patient outcomes. Of note, Gandhi et al.7 could not demonstrate a beneficial effect of intraoperative glycemic control on the incidence of mortality, sternal infections, cardiac arrhythmias, prolonged ventilation, and renal failure in patients undergoing on-pump cardiac surgery. Although the mean blood glucose levels in their treatment group were significantly lower than in their control group, the mean blood glucose level of their treatment group did not reach the preset target of 80–100 mg/dL during surgery as well as in the ICU.7 These data, together with the fact that 15% of patients in the control treatment group received insulin during surgery, may mask a potential benefit of strict intraoperative glycemic control.
Several limitations of ALGIP must be considered. First, the insulin dosage column can only be decreased one column at a time when the glucose level is below 85 mg/dL. Perhaps, adjusting the lower glucose limit from 85 to 90 mg/dL and narrowing the target range to 90–110 mg/dL could further reduce the number of hypoglycemic events. Second, although 88% of all diabetic patients had blood glucose levels <130 mg/dL during surgery, additional modifications of the algorithm are necessary to further improve the effectiveness of the protocol. Our results during the ICU admission, however (75.4% glucose levels 110 mg/dL, 3 h after arrival) support the efficacy of the ALGIP algorithm. Finally, that the protocol can be implemented with clinical nurses, as opposed to study personnel, suggests that our result might be replicated in other cardiac surgical centers.
ACKNOWLEDGMENTS
We thank the anesthetic nurses, perfusionists and intensive care nurses for their dedication and cooperation during the implementation of this protocol.
Footnotes
Accepted for publication February 8, 2008.
Financial support for this study was provided solely by intramural departmental sources.
The authors have no conflict of interest to report.
Data have been presented in part at the 21st annual meeting of the European Association of Cardiothoracic Anaesthesiologists 2006 in Venice, Italy, and at the 59th annual meeting of the American Society of Anesthesiologists 2006 in Chicago, USA.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  J. Blaha, P. Kopecky, M. Matias, R. Hovorka, J. Kunstyr, T. Kotulak, M. Lips, D. Rubes, M. Stritesky, J. Lindner, et al. Comparison of Three Protocols for Tight Glycemic Control in Cardiac Surgery Patients Diabetes Care, May 1, 2009; 32(5): 757 - 761. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
) and diabetics (
) during surgery and in intensive care unit (ICU). Mean preCPB = mean values before cardiopulmonary bypass; CPB1 = first blood glucose level on cardiopulmonary bypass; perCPB = mean blood glucose levels during hypothermic cardiopulmonary bypass; preSTOP = last blood glucose level before separating from bypass (at normothermia); postCPB = first blood glucose level after separation from cardiopulmonary bypass; preICU = last blood glucose level before admission in ICU; ICU0 = blood glucose level at arrival in the ICU; ICU1,2,3, . . . = blood glucose level after 1,2,3, . . . h after arrival in ICU.
