Postoperative insulin resistance (PIR) is a characterizing feature of the catabolic response to surgical injury (1). The association with stress hyperglycemia is commonly observed in critical illness and leads to an increase of postoperative complications (2). Intensive insulin therapy to normalize glucose levels reduces morbidity and mortality in critically ill patients (3). Traditional overnight preoperative fasting acts as additional metabolic stress superimposed on surgical insults and other trauma (4). New concepts are designed to minimize stress reactions by improving nutritional status before operation (1,4). In animal studies, rodents that were fed before stress induction showed improved muscle and cardiac function, better immunologic performance, and most prominently, better survival rates with complete recovery after hemorrhage or endotoxemia when compared with fasting subjects (4–6). A carbohydrate (maltose and fructose)-rich, clear beverage (CHO) was developed for preoperative clinical use (1,4). CHO elicits an endogenous release of insulin, comparable to a small breakfast in halting the overnight fasting metabolic state, and can be taken up to 2 h before surgery (1). In ASA I-II patients, CHO significantly reduced preoperative discomfort, postoperative nausea and vomiting, loss of lean body mass and muscle strength (4,7), without adverse effects (8). Preoperative IV glucose administration and oral CHO caused significant reductions of PIR (9–12) and seemed to speed up the recovery measured by reduced length of hospital stay (LOS) (13).
Particularly in cardiopulmonary bypass (CPB)-guided cardiac surgery, the commonly associated systemic inflammatory response syndrome (SIRS) leads to marked antiinsulinergic metabolic disorders and is a major cause of PIR (14). Whether metabolic stress response in cardiac surgery patients is reduced by CHO has not been investigated. However, preoperative IV glucose treatment has been shown to benefit cardiac surgery patients; it has been associated with reduced postoperative impairment to cardiac muscle suggested by cardiac enzyme decrease, fewer complications such as serious arrhythmias, need for vasopressor and inotropic agents, and shorter durations of ventilatory support requirements and stays in the intensive care unit (ICU) (15–18).
Therefore, the primary outcome measure of this prospective, randomized, double-blind, controlled study was to investigate whether preoperative CHO in cardiac surgery patients attenuates PIR as indicated by lower insulin requirements. The secondary outcome measure was to investigate whether CHO improves preoperative discomfort without affecting gastric fluid volume (GFV). The tertiary outcome measure was morbidity as measured by organ dysfunction.
Study Design and Population
The study was designed as an internal pilot study with recalculation of the sample size after interim analysis (19). Along with two double-blind study arms, CHO versus placebo (flavored water), an additional open-labeled control arm (fasting from midnight) was included to monitor overall efficacy. The study began with 25 patients per group. Interim analysis showed cumulative insulin consumption of 43 ± 19 (sd) IU by control patients. With an estimated 25% reduction of insulin consumption in CHO compared with placebo patients (¼ times 43 IU), recalculation of required sample size specified a need for at least 33 more patients.
Adult patients (≥18 yr) including Type-2 diabetes patients (noninsulin-dependent) undergoing elective coronary artery bypass graft (CABG) or valve replacement surgeries and eligible for preoperative clear fluid intake according to current national recommendations (20) were considered for admission. Exclusion criteria included conditions likely to impair gastrointestinal motility or enhance gastroesophageal reflux (i.e., obstructions or carcinomas of the upper gastrointestinal tract, gastroesophageal hernias, gastroesophageal reflux disease, etc.), potentially difficult airway management, ASA physical status >IV, nonelective or emergent surgery, presence of infection, pregnancy, maltose or fructose intolerance, and/or refusal to participate in the study. Patients with Type-1 diabetes were excluded because of metabolic risks. The study was approved by the local ethics committee (No. 1919/269, 2003/04/17).
One hundred eighty-eight patients gave written informed consent for study inclusion. Stratification was performed regarding the presence or absence of Type-2 diabetes (noninsulin-dependent). Within each subgroup, patients were randomized, computer-based, in blocks of six. After randomization, 28 patients were excluded. Eighteen patients did not receive the beverage because of logistical reasons, one refused the beverage, one withdrew consent, one was diagnosed with an axial esophageal hernia, one patient underwent emergency surgery, another received insulin bolus therapy, and two were excluded due to previous inclusion in another study.
Finally, 160 patients (85%) were included for data analysis assigned to one of the treatment groups: the two double-blind drinking groups receiving either the carbohydrate drink (CHO, n = 56) or flavored water (placebo, n = 60), or fasting after midnight (control, n = 44). The external hospital pharmacy was responsible for blinding the beverages according to the randomization list exclusively posted by the Institute of Medical Statistics. The placebo drink (water with acesulfame-K, 0.64 g/100 mL citrate, 0 kcal/100 mL, 107 mOsm/kg, pH 5.0) was formulated to taste and appear identical to the CHO drink (iso-osmolar, 12.5% carbohydrates, 50 kcal/100 mL, 290 mOsm/kg, pH 5.0). All beverages were stored in uniform bottles labeled “CHO/placebo” along with the identification number of the corresponding patient and transported to the cardiac surgery ward. Throughout the duration of the study, patients, nursing staff distributing the beverages, all other health care providers, the study physician, and the data collectors were blinded to the contents of the bottles. On the evening before surgery, CHO- and placebo-patients consumed 800 mL of the assigned beverage. After midnight, patients received nothing by mouth except for another single, morning dose of 400 mL of the appropriate drink 2 h before induction of general anesthesia (GA). Patients were clinically monitored beginning with anesthetic procedures, perioperatively, and for 24 h after admission to the ICU.
All anesthetic procedures and CPB management were performed according to institutional standards (21). Patients were premedicated with flunitrazepam (1–2 mg PO). GA was induced IV with midazolam (0.04–0.08 mg/kg), fentanyl (1–4 μg/kg), and etomidate (0.2–0.3 mg/kg). Pancuronium (0.1 mg/kg) was given for muscle relaxation and GA was maintained with fentanyl (0.5–1.0 mg/h), isoflurane (0.5–1.5 vol %), and hourly midazolam boluses. Membrane oxygenators Quadrox (Jostra, Hirlingen, Germany) and a centrifugal pump (Rotaflow, Jostra, Hirlingen, Germany) were used to conduct normothermic, nonpulsatile CPB. The pump was primed with methylprednisolone (1 g), hydroxyethyl starch (500 mL HES sterile 6%), and mannitol (250 mL). All patients received 50,000 KIU/kg of aprotinin at priming. Cardiac arrest was induced and maintained by intermittent anterograde administration of warm-blood cardioplegia solution enriched with potassium. Pump flow substituted cardiac output for hemodynamic measurements during CPB. Before CPB, heparin 400 IU/kg was given to maintain an activated clotting time (ACT) of at least 410 s measured by Hemochron® Jr. ACT+ (ITC, Edison, NJ). Heparin was fully reversed with protamine after discontinuation of CPB to achieve an ACT of 100–130 s. According to standardized CPB-weaning protocol, anesthesiologists aimed to maintain normovolemia and mean arterial blood pressure of 60–70 mm Hg supported by a baseline infusion of dopamine (1.5 μg·kg−1·min−1) and nitroglycerin (0.1–0.5 μg·kg−1·min−1). Cardiac insufficiency was assumed in cases of increasing dopamine requirements when no relevant surgical cause was identified, and was confirmed clinically by the presence of a dilated and weakly contracting heart. Transesophageal echocardiography was performed at the discretion of the anesthesiologist to confirm the diagnosis. Reduced afterload was assumed when targeted mean arterial blood pressure was not reached despite normal biventricular contractility and regular filling of the beating heart. In cases of cardiac insufficiency and/or reduced afterload, inotropic, and vasopressor support was administered. Inotropic treatment was prospectively defined as dopamine ≥5 μg·kg−1·min−1 and epinephrine, and enoximone per se, and vasopressor treatment as dopamine >10 μg·kg−1·min−1, and norepinephrine per se.
Outcome Measures and Data Collection
Insulin requirement (primary outcome measure) was deliberately chosen as a surrogate marker to estimate PIR, assuming accurate maintenance of equivalent glucose levels among the three study groups. A similar rationale measuring the ratio of fasting-glucose/ fasting-insulin levels is used to determine the degree of insulin resistance in diabetic patients (22). We hypothesized that CHO would require significantly less insulin to control blood glucose levels compared with placebo and control. Patients' serum glucose levels were monitored and documented together with the corresponding insulin dosages. Serum glucose was measured hourly by arterial blood gas analysis (Analysator ABLTM 700, Radiometer, Copenhagen, Denmark). Blood gas monitors were checked for accuracy three times per day. Adjustments of insulin dose were based on a continuous insulin infusion therapy protocol and adjusted according to hourly blood glucose measurements (Appendix). Because of previous experience with hypoglycemia while attempting to maintain blood glucose levels between 4.4–6.1 mmol/L, we chose a wider range of 4.4–10 mmol/L.
GFV and preoperative discomfort (secondary outcome measure) as well as drink-related complications, such as aspiration, regurgitation and metabolic disorders, were determined and recorded. In a subgroup of 96 patients (CHO n = 33, 59%; placebo n = 38, 63%; control n = 25, 57%), GFV was estimated according to passive gastric reflux over a single-lumen gastric tube inserted shortly after the induction of GA. At the end of operation, the collected volume was recorded. Patients rated five subjective discomfort variables of hunger, thirst, mouth dryness, nausea, and anxiety using visual analog scales (VAS) on the ward before premedication and transport to the operating suites. VAS ratings have been used to measure corresponding variables within similar preoperative settings (7,8,23). The systems used were horizontal, ungraded scales bounded by vertical lines from 0 to 100 mm, signifying the minimal and maximal extreme values of measured variables. Because no pilot study had ensured that drink flavor did not cause bias, the CHO- and placebo-groups were asked to evaluate the taste of their drinks.
To determine morbidity (tertiary outcome measure), variables of organ dysfunction were obtained: clinical appearance, vital signs, and administered medications were documented at least hourly intraoperatively and postoperatively in the ICU, and subsequently every 4 h. Further, the acute physiology and chronic health evaluation score (24), sepsis-related organ failure assessments (25), and simplified therapeutic intervention scoring systems (26) were calculated for the first 24 h. A SIRS was diagnosed according to Society of Critical Care Medicine criteria (27). LOS and mortality were recorded. Infection complications were evaluated to the time of hospital discharge according to definitions of the Centers for Disease Control (28). Postoperative pain, known to affect glucose metabolism (29), was estimated using a 10-cm VAS, where a score >3 is estimated as the threshold for clinically relevant pain in surgical patients (30). Piritramide in boluses of 3–5 mg was administered.
Categorical data were described using medians and 25%–75% quartiles or minimum-maximum (min-max), and continuous data using means and standard deviation (sd). Potential differences among the study groups were tested with adequate statistical methods depending on the scaling and distribution of observed variables, with the t-test for symmetrically distributed continuous data or the Kruskal-Wallis and Mann-Whitney U-tests, respectively, for categorical data. Proportions (frequencies) were analyzed using Fisher's exact test. For repeated measurements of blood gas analysis variables blood glucose and blood lactate as well as the hourly insulin infusion rate, a nonparametric repeated measures analysis in a factorial design was applied (31). A P value of <0.05 was considered significant. Data processing was performed with SPSS 11.5 and SAS 8.02 for Windows.
Demographic, clinical, and preoperative laboratory characteristics did not differ among groups (Table 1). The majority of patients underwent CABG procedures under CPB (Table 2). Almost all patients required exogenous insulin (Table 3). There was no significant difference in glucose levels (Fig. 1A) or insulin requirements (Fig. 1B). This was also independent of preexisting Type-2 diabetes (data not shown).
There was no difference in median (min-max) GFV between the CHO, placebo, and control groups (Fig. 2A). There were no cases of apparent or suspected pulmonary aspiration or other drinking-related complications. VAS ratings did not differ with respect to hunger, nausea, anxiety, and/or dryness of mouth (Table 1). There was a significant decrease in thirst ratings between CHO and control groups [7 (0–75) mm vs 30 (0–90) mm, P < 0.01) and a tendency between placebo and control [8 (0–76) vs 30 (0–90), P = 0.06] (Fig. 2B). CHO and placebo groups did not differ significantly in thirst. There was no significant difference between CHO and placebo groups in flavor ratings (Table 1).
Pain management did not differ among groups. Four hours after ICU admission, 68% of CHO, 68% of placebo, and 75.0% of control patients scored >3 cm (P = 0.69). Sixteen hours later, these scores decreased to 25.0%, 28.3%, and 36.4% (P = 0.47), respectively. The corresponding piritramide administration was also not different (7.9 ± 5.3, 6.4 ± 2.5, 5.2 ± 2.7 mg, P = 0.28; and 5.3 ± 3.7, 4.5 ± 1.8, 4.3 ± 1.7 mg, respectively, P = 0.95).
After initiation of CPB weaning until the end of operation, CHO patients required significantly less inotropes than did placebo and control patients (Fig. 3). There was no difference in pre-CPB and postoperative inotropic requirements (Fig. 3, Table 2).
There was no significant difference in severity of illness scores or incidence of SIRS (Table 2). The incidence of typical postoperative complications and durations of ICU/hospital LOS also did not differ (Table 2). Two patients died during hospital admission: one control, of acute left ventricular failure in the ICU, and one placebo patient of bowel cancer on the ward.
As insulin requirements did not differ among our study groups, it seems that CHO administration before elective cardiac surgery does not affect PIR in ASA III-IV patients. This contrasts with findings from previous investigations performed with ASA I-II patients (11,12), which have identified significant reductions in PIR using a hyperinsulinemic-euglycemic clamp technique. We assumed that clinically relevant reductions of PIR should be identifiable by decreased insulin requirements to maintain comparable blood glucose levels. It is possible that benefits to glucose metabolism were suppressed by SIRS, which occurred in most of our patients. The incidence of CPB-associated SIRS of 66%–93% of patients, as reported elsewhere (32), was confirmed by our data and is even more distinct in patients undergoing cardiac compared to noncardiac surgery (14). Other variables of metabolic stress known to influence PIR, i.e., duration of surgery (11), use of vasoactive medication (14), and pain (29), did not differ among the groups and should not have interfered with the potential therapeutic effects of preoperative CHO administration.
With a median value of 0 mL, GFV did not differ significantly among groups. This confirms findings of other randomized, controlled trials (mainly in ASA physical status I-II patients) that have reported reliable gastric emptying of limited amounts of clear fluids (1,20). Hausel et al. (8,23) showed that even relatively large quantities (400 mL) of clear liquids did not affect the GFV or cause any adverse effects when administered 2 h before anesthetic induction. In contrast to the present study, former investigations have used blind aspiration and a single marker dilution technique to estimate GFV. Regardless of the methods, however, a median GFV of approximately 20 mL was obtained (8). The lower median value of 0 mL in our study is likely due to the less exact measurement method, and does not prove that the risk of aspiration remains unaffected. However, because of the extremely low incidence of 1.4–6.0 aspirations per 10,000 patients undergoing GA, it is difficult to conduct proper trials regarding such risks (1,20). In the end, pulmonary aspiration has a good prognosis and is more likely a consequence of insufficient airway protection and inadequate anesthetic depth rather than with patients' fasting state (1,20).
Subjective VAS-ratings revealed a reduction of thirst in beverage-drinking patients when compared with that in fasting patients (CHO versus control, P = 0.004 and placebo versus control, P = 0.06). CHO and placebo did not differ in this respect. Hausel et al. (8), using VAS for a larger sample size of ASA I-II patients undergoing abdominal surgery (n = 252), also found no difference in thirst after the morning drink (CHO and placebo). However, in that study, patients given CHO reported significantly decreased hunger and anxiety. Henriksen et al. (7) showed contrasting results in a smaller study (n = 48) comparing CHO administration with fasting in patients before elective bowel resection. In that study, patients showed no difference even in thirst (7). It is possible, therefore, that the sample size of the current study (n = 160) was insufficient to show potential effects on the other subjective measures of patient discomfort. Regardless, current data show that thirst, suggested as the major determining factor of preoperative discomfort (33), is effectively reduced by clear fluid intake.
Preoperative administration of CHO seemed to reduce intraoperative requirements for inotropic drugs after initiation of weaning CPB. A former study by Quiros and Ware (5) and recently by van Hoorn et al. (6) have investigated the cardiovascular effects of prestress nutrition versus starvation in rats within hemorrhage-induced hypotension and ischemia/reperfusion models, respectively. Both studies showed consistent, significantly improved cardiac function indicated by higher cardiac output and stroke volume and slower heart rates in fed versus fasted animals (5,6). In other clinical trials, preoperative IV carbohydrate administration before cardiac surgery has also led to markedly improved cardiac performance (15–18); given alone (16), or in combination with IV lipids (17) or insulin and potassium (15,18), carbohydrate was found to reduce incidence of cardiac insufficiency and other complications (i.e., fibrillation or need for vasopressors). In particular, Lazar et al. (18) reported reduced inotropic scores postoperatively after perioperative administration of IV glucose-insulin-potassium to patients undergoing urgent CABG surgery. However, the need for inotropic support, defined as dopamine ≥2 μg·kg−1·min−1 compared with our definition of ≥5 μg·kg−1·min−1, did not differ significantly in their study.
There are limitations of our study: First, GFVs were measured by passive gastric reflux, and not the gold standard. This may have led to an underestimation of GFV. Also, logistical problems and confounding factors such as intraoperative transesophageal echocardiography restricted the sample size of patients undergoing GFV measurements. Second, nosocomial pneumonias were not diagnosed according to the recently published new guidelines of the American Thoracic Society (34). Finally, a possible CHO-associated effect on cardiac performance can only be indirectly suggested by the reduced inotropic requirements. Also, cardiac insufficiency was diagnosed only according to clinical variables. Even so, more than half of the CHO patients (55%) needed inotropic support during CPB weaning, and there was no difference in cardiovascular drug requirements at any other time.
In conclusion, CHO administration before elective cardiac surgery does not appear to influence PIR. The intake of clear fluids reduced preoperative thirst, and may be recommended as routine procedure for ASA physical status III-IV patients. As GFVs were not increased, and other adverse events or metabolic disorders did not occur, oral CHO administration can be considered safe for cardiac surgery patients, including noninsulin-dependent Type-2 diabetes patients. The role of preoperative CHO administration with respect to a clinically relevant reduction in postoperative impairment of myocardial function remains a question for further investigation.
The authors thank Sarah Hamilton Halberg, MD, CFPC (Frankfurt, Germany) and Anand Rughani (McGill University, Faculty of Medicine, Montreal, Canada) for the diligent revision of the manuscript, as well as Gerd Kalb, Dipl-Ing. (Institute of Medical Statistics and Biometry; Campus Charité Mitte, CHARITÉ, University Medicine Berlin, Germany) for the detailed statistical advice and for principal data analysis.
APPENDIX: THERAPY PROTOCOL FOR INSULIN TITRATION (FIG. A1)
Concentration and mode of administration: Insulin is only given by continuous IV infusion through a central venous line using a syringe-driven pump. The standard concentration is 40 IU insulin in 40 mL NaCl (0.9%). Prepared solutions are not stable after 24 h.
Measurement of Blood Glucose Levels
Whole blood glucose levels are measured in undiluted arterial blood.
The primary aim of blood glucose level is a scope of 4.4–6.1 mmol/L (80–110 mg/dL), and secondary aim is 4.4–10 mmol/L (80–180 mg/dL).
Starting Insulin Infusion and Initial Stabilization of Blood Glucose Level
When blood glucose level is ≥145 mg/dL (≥8.1 mmol/L), insulin infusion is initiated at a starting dose of 2 IU/h. If blood glucose level on which insulin is started is ≥181 mg/dL (≥10.1 mmol/L), the starting dose of insulin is set at 4 IU/h. If the glucose level is ≥217 mg/dL (≥12.1 mmol/L), the dose amounts 6 IU/h.
Measurement After One Hour
Blood Glucose Level Increases
If the next blood glucose level is more than or equal to the previous, compare with the aspired blood glucose scope.
- If the blood glucose level is between 4.4 and 6.1 mmol/L, keep the insulin infusion rate.
- If the scope exceeded, change the rate while adding IUs to the previous rate:
If blood glucose level is
- >6.1 mmol/L (110 mg/dL), add 0.5 IU/h
- >7.0 mmol/L (126 mg/dL), add 1 IU/h
- >8.0 mmol/L (144 mg/dL), add 2 IU/h
- >10.0 mmol/L (180 mg/dL), add 4 IU/h
- >12.0 mmol/L (216 mg/dL), add 6 IU/h
Blood Glucose Level Decreases
If the next blood glucose level is <the previous, compare the aspired blood glucose scope.
- If blood glucose level is <4.4 mmol/L (79.2 mg/dL), stop infusion and check again blood glucose level after 30 min. Is blood glucose level <3.5 mmol/L (63.0 mg/dL), stop infusion, give a 10 g bolus of glucose and check blood glucose level every 30 min until blood glucose level is >4.4 mmol/L.
- If blood glucose is >8.0 mmol/L, leave insulin dose unaltered, except if there was an intense decrease of glucose level (>50%) or a decrease of glucose level >1 mmol/L, while glucose level is <10 mmol/L (in the latter case halve the rate).
- If blood glucose level is between 5.3–8.0 mmol/L 1 h after previous measurement, reduce insulin rate to 1.5 IU/h, except if the current rate is lower (in this case leave the insulin rate unaltered).
- If blood glucose level is <5.3 mmol/L, reduce the insulin rate: if current rate is ≥2 IU/h reduce to 1 IU/h, if it is <2 IU/h reduce to 0.5 IU/h.
Measurement After Two Hours
It is the same scheme as after the first measurement, except the blood glucose level is again <5.3 mmol/L, in this case stop infusion.
The insulin therapy starts again, if the blood glucose reaches a level of >6.1 mmol/L.
The following hourly measurements involve the same scheme, shown in previous sections.
Maximal insulin dose is arbitrarily set at 20 IU/h.
1. Ljungqvist O, Soreide E. Preoperative fasting. Br J Surg 2003;90:400–6.
2. Mizock BA. Blood glucose management during critical illness. Rev Endocr Metab Disord 2003;4:187–94.
3. 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.
4. Diks J, van Hoorn DEC, Nijveldt RJ, et al. Preoperative fasting: an outdated concept? J Parenter Enteral Nutr 2005;29:298–304.
5. Quiros G, Ware J. Cardio-vascuclar and metabolic alterations caused by hemorrhage in fed and starved rats. Acta Physiol Scand 1983;117:397–403.
6. van Hoorn EC, Middelaar-Voskuilen MC, van Limpt CJ, et al. Preoperative supplementation with a carbohydrate mixture decreases organ dysfunction-associated risk factors. Clin Nutr 2005;24:114–23.
7. Henriksen MG, Hessov I, Dela F, et al. Effects of preoperative oral carbohydrates and peptides on postoperative endocrine response, mobilization, nutrition and muscle function in abdominal surgery. Acta Anaesthesiol Scand 2003;47:191–9.
8. Hausel J, Nygren J, Lagerkranser M, et al. A carbohydrate-rich drink reduces preoperative discomfort in elective surgery patients. Anesth Analg 2001;93:1344–50.
9. Ljungqvist O, Thorell A, Gutniak M, et al. Glucose infusion instead of preoperative fasting reduces postoperative insulin resistance. J Am Coll Surg 1994;178:329–36.
10. Nygren J, Thorell A, Soop M, et al. Perioperative insulin and glucose infusion maintains normal insulin sensitivity after surgery. Am J Physiol 1998;275:E140–8.
11. Nygren J, Soop M, Thorell A, et al. Preoperative oral carbohydrate administration reduces postoperative insulin resistance. Clin Nutr 1998;17:65–71.
12. Soop M, Nygren J, Myrenfors P, et al. Preoperative oral carbohydrate treatment attenuates immediate postoperative insulin resistance. Am J Physiol Endocrinol Metab 2001;280:E576–83.
13. Ljungqvist O, Nygren J, Thorell A, et al. Preoperative nutrition—elective surgery in the fed or the overnight fasted state. Clin Nutr 2001;20:167–71.
14. Jakob SM, Ensinger H, Takala J. Metabolic changes after cardiac surgery. Curr Opin Clin Nutr Metab Care 2001;4:149–55.
15. Oldfield GS, Commerford PJ, Opie LH. Effects of preoperative glucose-insulin-postassium on myocardial glycogen levels and on complications of mitral valve replacement. J Thorac Cardiovasc Surg 1986;91:874–8.
16. Berggren H, Ekroth R, Hjalmarson A, et al. Enhanced myocarial protection from preoperative carbohydrate infusion in addition to maintained beta-blockade. J Cardiovasc Surg 1985;26:454–6.
17. Lolley DM, Myers WO, Ray JF, et al. Clinical experience with preoperative myocardial nutrition management. J Cardiovasc Surg 1985;26:236–43.
18. Lazar HJ, Philippides G, Fitzgerald C, et al. Glucose-insulin- potassium solutions enhance recovery after urgent coronary artery bypass grafting. J Thorac Cardiovasc Surg 1997;113:354–62.
19. Kieser M, Friede T. Simple procedures for blinded sample size adjustment that do not affect the type I error rate. Stat Med 2003;22:3571–81.
20. Spies CD, Breuer JP, Gust R, et al. [Preoperative fasting. An update]. Anaesthesist 2003;52:1039–45 (in German).
21. Kox WJ, Spies D. Check-up anästhesiologie. 2nd ed. Heidelberg: Springer Medizin Verlag, 2005.
22. Gungor N, Saad R, Janosky J, Arslanian S. Validation of surrogate estimates of insulin sensitivity and insulin secretion in children and adolescents. J Pediatr 2004;144:47–55.
23. Hausel J, Nygren J, Thorell A, et al. Randomized clincial trial of the effect of oral preoperative carbohydrates on postoperartive nausea and vomiting after laparoskopic cholecystektomy. Br J Surg 2005;92:415–21.
24. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:818–29.
25. Vincent JL, Moreno R, Takala J, et al; on behalf of the working group on sepsis-related problems of the European Society of Intensive Care Medicine. The SOFA (sepsis-related organ failure assessment) score to describe organ dysfunction/failure. Intensive Care Med 1996;22:707–10.
26. Miranda DR, de Rijk A, Schaufeli W. Simplified Therapeutic Intervention Scoring System: the TISS-28 items—results from a multicenter study. Crit Care Med 1996;24:64–73.
27. Bone RC, Balk RA, Cerra FB, et al; The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/ Society of Critical Care Medicine. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest 1992;101:1644–55.
28. Garner JS, Jarvis WR, Emori TG, et al. CDC definitions for nosocomial infections, 1988. Am J Infect Control 1988;16:128–40.
29. Greisen J, Juhl CB, Grofte T, et al. Acute pain induces insulin resistance in humans. Anesthesiology 2001;95:578–84.
30. Geissler B, Neugebauer E, Angster R, Witte Dagger J. Quality management during postoperative pain therapy. Chirurg 2004;75:687–93.
31. Brunner E, Domhof S, Langer F. Nonparametric analysis of longitudinal data in factorial experiments. New York: Wiley, 2002.
32. Pittet D, Rangel-Frausto S, Li N, et al. Systemic inflammatory response syndrome, sepsis, severe sepsis and septic shock: incidence, morbidities and outcome in surgical ICU patients. Intensive Care Med 1994;21:302–9.
33. Madsen M, Brosnan J, Nagy VT. Perioperative thirst: a patient perspective. J Perianesth Nurs 1998;13:225–8.
34. American Thoracic Society Documents. Guidelines for the management of adults with hospital-acquired, ventilatory-associated, and healthcare-associated pneumonia. Am J Respir Crit Care Med 2005;171:388–416.