After cardiacsurgery, most patients have a short intensive care unit (ICU) and hospital stay; however, up to 10% of patients (1,2) need prolonged postoperative care, mainly because of organ dysfunction or multiple organ failure. This increases ICU and hospital resource use and health care costs (3). Although the causes of prolonged ICU and hospital stay are multifactorial, limited cardiovascular resources and inadequate hemodynamic response to postoperative surgical stress have recently been shown to be independent predictors of prolonged ICU stay (4,5). Increased levels of oxygen delivery and consumption have been associated with improved outcome (6) and the concept has been tested in a variety of clinical situations (7–17). Randomized studies in high-risk surgical or trauma patients, with intervention started perioperatively, have shown a marked decrease in mortality and morbidity rates (7,8,13,14). However, when heterogeneous, critically ill, and high-risk surgical patients have been treated after admission to the ICU, results have been conflicting—no difference in outcome between the groups (9,10,12,16) or even worse outcome in the treated patients (11). Perioperative volume optimization has been shown to improve outcome in cardiac surgery patients (18), whereas inadequate oxygen delivery immediately after cardiac surgery has been associated with prolonged ICU stay (19). In our previous study (20), increased oxygen extraction immediately after cardiac surgery was an independent predictor of prolonged ICU stay. Consequently, we hypothesized that a treatment targeting Svo2 >70% and blood lactate concentration ≤2 mmol/L immediately after cardiac surgery, would shorten the length of hospital stay and length of ICU stay, compared with the control group treated according to standard clinical care.
The study protocol was approved by our hospital institutional review board, and each patient gave written, informed consent. A total of 403 consecutive, elective cardiac surgical patients were prospectively enrolled in the study. Patients were randomly assigned either to the control group, in which standard care was given, or to the protocol group, in which besides standard care, the goal was to maintain Svo2 >70% and a serum lactate concentration ≤2.0 mmol/L from admission to the ICU and up to 8 h (Figure 1). During the operation, standard treatment was given to both groups. Patients were randomized the day before surgery by sealed envelope technique, and after randomization, the caregivers were aware of the randomization group. Nine patients were dropped from the study after the randomization but before the operation. Of these nine, eight patients were already enrolled in another study, including a Jehovah’s witness because the protocol algorithm required red blood cell infusion. One patient died during the operation and, therefore, was excluded from the study population. Altogether, 393 patients fulfilled the study protocol.
Anesthetic technique was standardized and based on continuous infusion of fentanyl and midazolam (30 μg/kg and 0.1 mg/kg, respectively) supplemented with isoflurane when necessary. Pancuronium and alcuronium were used for muscle relaxation. Extracorporeal circulation with systemic heparinization (3 mg/kg), pump flows from 2.0 to 2.4 L/m2 of body surface area, moderate systemic hypothermia (30°C) and intermittent cold crystalloid cardioplegia were used. Patients were weaned from cardiopulmonary bypass (CPB) after the rectal temperature had reached 36°C.
The standard clinical postoperative care after weaning from CPB included the following: 1) If cardiac index was <2.5 L · min−1 · m−2 despite volume expansion [pulmonary capillary wedge pressure 12–18 mm Hg], dobutamine infusion was started. 2) Mean arterial pressure was kept between 60 and 90 mm Hg by using either vasopressor (dopamine or norepinephrine) or vasodilator (sodium nitroprusside) infusion when appropriate. 3) Hemoglobin concentration was kept ≤100 g/L with packed red blood cells. 4) Patients were weaned from mechanical ventilation when rewarmed and hemodynamically stable.
Radial arterial and thermodilution pulmonary arterial catheters were routinely used. Cardiac output was measured in triplicate and the mean value was used for calculations. Oxygen delivery was calculated according to standard formula, multiplying the thermodilution cardiac output with arterial oxygen content (Cao2) and indexed to body surface area. Oxygen consumption was calculated by multiplying cardiac output with the arteriovenous oxygen content difference. Oxygen contents were derived as [1.39 × hemoglobin concentration × Sao2 or mixed venous oxygen saturation + dissolved oxygen]. Oxygen extraction rate was calculated dividing V̇o2 by Do2. Hemoglobin oxygen saturation was measured by using co-oximetry (IL 282; Instrumentation Laboratories, Lexington, MA) and oxygen tension was measured by clinical blood gas analyzer (Stat profile 4®; NOVA Biomedical, Waltham, MA). Arterial blood lactate levels were measured by an enzymatic reaction (YSI 2300; Yellow Springs Instrument Laboratory, Yellow Springs, OH).
Hemodynamics and oxygen transport data were recorded 15 min after weaning from CPB, at admission to the ICU, and 2, 6, and 8 h after ICU admission. The data for organ functions were recorded on the first postoperative morning and on the day the patient was discharged from the hospital. Dysfunction of organ systems was defined as follows: (a) central nervous system: hemiplegia, stroke, or Glasgow coma scale score <10 in the absence of sedation; (b) circulatory: need for vasoactive medication to treat hypotension (dopamine or norepinephrine) or decreased cardiac output (dopamine, dobutamine or epinephrine), or intraaortic balloon counterpulsation, (c) respiratory : need for mechanical or assisted ventilation, (d) renal : urine output <750 mL/24 h or increase of serum creatinine concentration >150 μmol/L from preoperatively normal levels, (e) hepatic : serum alanine aminotransferase activity >40 IU/L and serum bilirubin concentration >40 μmol/L, (f) gastrointestinal : macroscopic bleeding or paralytic ileus, (g) hematological : leukocyte count <3.5 × 109/L and platelet count <80 × 109/L.
In the protocol group, additional targets of hemodynamic management were to maintain Svo2 >70% and arterial blood lactate concentration ≤2 mmol/L from admission to the ICU to 8 h thereafter (Figure 1). If the goals could not be achieved at a certain time point despite volume substitution, dobutamine infusion up to 15 μg · kg−1 · min−1 was started to increase cardiac index. The effect of the treatment on Svo2 and lactate was assessed 2 h later. In patients who achieved the goals, therapy was maintained and adjusted on the following data recording time points, when necessary: 1) In those who had not achieved the goals, therapy was continued with incremental volume expansion and increasing doses of dobutamine. 2) The same protocol algorithm was used at the following time points, except at 8 h after admission to the ICU, the end point of the protocol treatment, in which patients who had Svo2 >70% and arterial blood lactate ≤2.0 mmol/L were deemed as responders. 3) All others were deemed as nonresponders, even if they had achieved the targets at earlier time points and had no inotropes. 4) After the end of data recording, protocol treatment was gradually withdrawn.
Because of the small, expected mortality rate in cardiac surgical patients, the length of hospital and ICU stay were chosen as primary outcome variables. Postoperative morbidity was estimated by the number of organ dysfunctions and by the use of ICU and hospital resources. Secondary end points of the study were hospital, 6-mo, and 12-mo mortality.
Sample size was estimated from our previous observational study (20). A sample size of 400 patients was estimated to have at least 90% power at α = 0.05 to detect 3 days difference in the length of hospital stay, given an sd of 8 days, and to allow for a 10% dropout rate. The primary analysis was performed on an intention-to-treat basis. The χ2 test or Fisher’s exact test was applied to categorical variables, and Student’s t-tests or Mann-Whitney U-test was used for independent continuous variables when appropriate. Analysis of variance for repeated measures was used for repeated continuous variables to determine differences between the two groups over the study period. One-way analysis of variance with Bonferroni correction was used to locate the differences between the groups during the study protocol. Kaplan-Meier analysis with log-rank statistics was used to test the differences in the use of hospital and ICU resources between the groups. Statistical significance was assumed at P < 0.05. Data are presented as median (range) or mean ± sd when appropriate.
Group demographics were comparable (Table 1). There were 700 ICU care days. Intensive care longer than 4 days was needed in 3.3% (13 of 393) of patients. These patients used 38% (259/700) of the cardiac surgery ICU resources and 11.8% (353/2990) of the total postoperative hospital care days during the study period. The overall hospital and 28-day mortality rate after randomization was 2.0% (8 of 393). The mortality rate at 6 mo was 2.5% (10 of 393) and at 1 yr was 3.3% (13 of 393).
Outcome by Treatment
The median hospital stay was 6 days (range 4–64) in the protocol group and 7 days (range 2–93) in the control group (Mann-Whitney U-test, P < 0.05). In the protocol group, discharge of patients from the hospital was significantly faster at 14 days and at 28 days than patients in the control group (Kaplan-Meier, P < 0.05 and P = 0.08, respectively). Discharge from the ICU was similar between the groups (Kaplan-Meier, P = 0.8). Patients in the protocol group used 48.1% and patients in the control Group 51.9% of the total of 2990 hospital care days required postoperatively for the study population. In the protocol group, there were one renal and one hepatic dysfunctions, whereas in the control group, patients had four central nervous system, five hepatic, three renal, and one respiratory dysfunctions at hospital discharge. One patient in the control group had three organ dysfunctions (Table 2). Mortality rate was similar between the groups at 28 days, at 6 mo and 1 yr after the operation (Table 2).
Oxygen Transport and Hemodynamics by Treatment
Because the study protocol Do2I, V̇o2I, O2ER, and Svo2 were different between the groups (Table 3). Accordingly, each oxygen transport (Table 3) and hemodynamic (Table 4) variable included in the analysis changed during the first 8 h in the ICU. During the study protocol, cardiac index and stroke volume index were higher, whereas central venous pressure was lower in the protocol group compared with the control group (Table 4).
Fluid Replacement and Use of Inotropes
In the protocol group, patients received more crystalloid (2271 ± 1523 vs 1970 ± 1219 mL, P < 0.05) and colloid substitution (922 ± 431 vs 802 ± 408 mL, P < 0.01) than patients in the control group during the first postoperative night. The use of blood products was similar. In the protocol group, the use of inotropic drugs was more frequent and the need for vasopressors less frequent than in the control group. There were no differences in the use of vasodilators (Figure 2).
Outcome by Achievement of the Targets in the Study Groups
The hemodynamic targets of the protocol group (Svo2 >70% and arterial blood lactate ≤ 2 mmol/L) were achieved in 70 (35.1%) patients in the protocol group and in 66 (33.5%) patients in the control group by means of volume substitution only and in 42 (21.4%) patients in the protocol group and in 17 (8.6%) patients in the control group with inotropes in addition to volume substitution (Figure 3). In the protocol group, there were 84 (42.9%) patients and in the control group there were 114 (57.9%) patients who did not achieve the targets at 8 h after arrival at the ICU. Of the nonresponders 38% (32 of 84) in the protocol group and 55.3% (63 of 114) in the control group had volume substitution only and 62% (52 of 84) in the protocol group and 44.7% (51 of 114) in the control group in addition had inotropes.
Post hoc analysis in the protocol group showed that postoperative morbidity decreased in patients who achieved the targets as reflected in significantly faster patient discharge from the hospital and from the ICU (Kaplan-Meier, P < 0.001 for both) than in patients who did not achieve the targets. The median hospital stay was 5 days (range 5–11 days) in patients who achieved the targets and 7 days (range 4–64 days) in patients who did not achieve the targets (Table 5). The median ICU stay was 1 day (range 1–4 days) in patients who achieved the targets and 1 day (range 1–52 days) in patients who did not achieve the targets. Patients who achieved the targets had fewer organ dysfunctions on the first postoperative morning in the ICU (n = 19 vs 37, P < 0.01) and at hospital discharge (n = 0 vs 2, P = 0.1); a smaller number of patients had a prolonged ICU stay (n = 0 vs 5, P < 0.05) while tending to have fewer readmissions to the ICU (n = 0 vs 3, P = 0.08). Mortality rate was decreased at 6 mo (n = 0 vs 2, P < 0.05) and at 1 yr (n = 0 vs 4, P < 0.05) after randomization in patients who achieved the targets. Patients who achieved the targets had less colloids (853 ± 423 mL vs 1014 ± 428 mL, P < 0.01) and red blood cell substitution (3.6 ± 2.3 vs 6.6 ± 7.9 units, P < 0.001) and, bleeding from the chest tubes was less (997 ± 578 vs 1342 ± 903 mL, P < 0.01) than in patients who did not achieve the targets. Patients who achieved the targets were younger (58.6 ± 8.3 vs 61.9 ± 7.4 yr, P < 0.01), had higher preoperative ejection fraction (65.9 ± 14.1 vs 60.6 ± 16.1%, P < 0.05), shorter aortic occlusion time (111 ± 37 vs 122 ± 40 min, P < 0.05), shorter perfusion time (128 ± 42 vs 142 ± 46 min, P < 0.05), shorter operation time (232 ± 57 vs 256 ± 78 min, P < 0.05) and fewer postoperative reoperations for bleeding or tamponade (4.5 vs 13.1%, P < 0.05) than patients who did not achieve the targets in the protocol group.
Also, post hoc analysis of the control group showed less postoperative morbidity in patients who achieved the targets. They had earlier discharge from the hospital and the ICU compared with patients who did not achieve the targets (Kaplan-Meier, P < 0.05 for both). Accordingly, the median hospital stay was 6 days (range 2–36) vs 7 days (range 3–93) (Table 5) and the median ICU stay was 1 day (range 1–25) vs 1 day (range 1–39). Patients who achieved the targets had fewer organ dysfunctions on the first postoperative morning in the ICU (n = 9 vs 27, P < 0.05) and at hospital discharge (n = 7 vs 14, P < 0.05). There was no difference in mortality rate. Also, patients who achieved the targets had less red blood cell substitution (3.8 ± 2.8 vs 5.6 ± 8.1 units, P = 0.06) despite similar bleeding from the chest tubes (P = 0.12) and less postoperative reoperations for bleeding or tamponade (2.4% vs 10.5%, P < 0.05) than patients who did not achieve the targets. Additionally, patients who achieved the targets had less combined surgery (n = 16 vs 43, P < 0.01) than patients who did not achieve the targets.
Oxygen Delivery and Outcome in the Whole Study Population
To clarify the role of oxygen delivery on outcome a posteriori, we compared patients who achieved the targets (n = 195) to those who did not achieve the targets (n = 198) in the whole study population (353 ± 83 and 329 ± 81 mL · min−1 · m−2 after weaning from bypass, 418 ± 87 and 377 ± 76 mL · min−1 · m−2 after arrival at the ICU, 488 ± 110 and 419 ± 95 76 mL · min−1 · m−2 at 2 h in the ICU, 508 ± 105 and 434 ± 96 76 mL · min−1 · m−2 at 6 h in the ICU, 533 ± 96 and 441 ± 88 mL · min−1 · m−2 after 8 h in the ICU, respectively). Patients who achieved the targets had faster discharge from the hospital and from the ICU (Kaplan-Meier, P < 0.001 for both). The median hospital stay was 5 days (range 2–36) in patients who achieved the targets and 7 days (range 3–93) in patients who did not achieve the targets (P < 0.001). The median ICU stay was 1 day (range 1–25) in patients who achieved the targets and 1 day (range 1–52) in patients who did not achieve the targets (P < 0.001). Prolonged ICU stay was less (P = 0.05) in patients who achieved the targets. Mortality rate was similar at 28 days and at 6 mo but less at 1 yr after randomization (P = 0.05) in patients who achieved the targets. Patients who achieved the targets had better preoperative ejection fraction (P < 0.01) and less diabetes (P < 0.05) than patients who did not achieve the targets. Patients who achieved the targets had fewer valvular or combined valvular and coronary artery bypass procedures (P < 0.01), shorter aortic occlusion time (P < 0.05), perfusion time (P < 0.01), and operation time (P < 0.05). Reoperations for bleeding or tamponade (P < 0.01), myocardial infarctions (P < 0.05), and arrhythmias (P = 0.05) were less common in patients who achieved the targets.
The concept based on increasing oxygen delivery and consumption to supranormal values perioperatively has been associated with improved outcome in high-risk surgical patients (7,8,17), whereas in high-risk peripheral vascular surgical patients preoperative hemodynamic optimization did not improve outcome (16). In critically ill patients, the use of supranormal oxygen transport values as the target of therapy has shown conflicting results (9,10,11,12). A recent meta-analysis of prospective randomized studies (21) concluded that these studies show a trend toward decreased mortality rates in the treatment groups. The concept has not been tested in cardiac surgical patients known to be at risk of increased postoperative morbidity and mortality, although inadequate oxygen transport balance has been associated with increased use of ICU resources postoperatively (19,20). Although perioperative volume optimization improves outcome (18), this is the first randomized, controlled trial in cardiac surgery targeting to normal perioperative oxygen transport, as judged by normal Svo2 and arterial lactate.
The main finding of our study was that protocolized care aiming to normal oxygen transport and normal Svo2 and lactate during immediate postoperative period (up to eight hours) after cardiac surgery, can improve outcome. In the protocol group, this was reflected in a one-day shorter length of hospital stay and reduced number of organ dysfunctions at hospital discharge. Although the difference in length of hospital stay was small, the impact on resource use may be substantial. Assuming the volume of 1000 cardiac operations per year, the care according to the protocol would lead to a reduction of 1000 care days in the cardiothoracic surgical ward. Organ dysfunctions detected in the control group are those typically related to peri- and postoperative hypoperfusion and causing increased use of hospital resources. These findings are in agreement with other studies of surgical patients in which decreased morbidity and shorter length of ICU and hospital stay have been reported in the protocol groups (7,8,17). The fact that there were no differences in mortality rate between the study groups may be because of overall low mortality rate in cardiac surgical patients and because we did not design our study to find differences in mortality rates. Faster hospital discharge and decreased morbidity in the protocol group suggests that optimizing cardiovascular function, either with volume loading or with the use of inotropes, in addition to volume substitution, has significantly contributed to improved outcome.
An additional important finding was that the achievement of the hemodynamic targets in the protocol group was difficult, even with the addition of inotropes. Approximately one half of the patients (57%) in the protocol group achieved the targets of Svo2 and lactate. Difficulty achieving the targets has been common also in other randomized studies (8,9,10–12). In the study by Gattinoni et al. (12), 66.7% of the patients in the Svo2 group and 44.9% of the patients in the cardiac index group achieved the targets. Similarly, 73% of the patients in the study by Tuchschmidt et al. (9) and 66% of the patients in the study by Yu et al. (10) were able to reach the targets. In the study by Hayes et al. (11), only 30% of the patients achieved the targets, whereas in the study by Ziegler et al. (16), all patients in the treatment group achieved the targets. Thus, the proportion of patients who achieve the targets varies according to the targets used. When V̇o2I is used as the target in addition to Do2I, CI, or Svo2 the achievement of the targets has been least frequent. In our study, the low achievement of the targets in the protocol group may partly be attributable to the protocol. For example, some patients who achieved the targets with volume substitution only at six hours were nonresponders at eight hours after arrival at the ICU (end point of the protocol). As in the study by Gattinoni et al. (12), patients who achieved the targets in the protocol group were younger and had better cardiovascular function reflected as a better preoperative ejection fraction. Shorter operative times may reflect more favorable operative conditions caused by a better underlying disease state in patients who achieved the targets. Also, reoperations for excessive bleeding or tamponade in the postoperative period may have jeopardized adequate oxygen delivery when cardiac function was already compromised. Improved outcome in patients who achieved the targets in the protocol group suggests better cardiovascular reserves and thus, capability to adequately respond to increasing surgical stress and hemodynamic crisis, either spontaneously with volume substitution or with the help of inotropes.
In the whole study population, achievement of hemodynamic targets was also associated with improved outcome. Patients who did not achieve the targets had more co-morbidity and limited cardiovascular reserves. These patients were operated on more often for valvular or combined procedures and thus, had longer operative times, indicating more complex surgery or advanced disease state. Arrhythmias, myocardial infarction, and resternotomies also complicated the postoperative course more frequently in patients who did not achieve the targets.
One limitation in our study design was that had we had enrolled patients for only coronary revascularization, the patient population would have been more homogeneous. Therefore, aposteriori, we ran the statistics for patients who had only a revascularization procedure and found that the results were similar with our protocol (Kaplan-Meier for the hospital stay, P < 0.05). Although the study was randomized, it was not blinded, and the data of the oxygen transport measurements of both groups were open to those taking care of patients during the study period. This may have caused some bias in the intensity of treatment in the control group despite the fact that targeted criteria for the groups were different. In addition, the study protocol period—eight hours after admission to the ICU—may have been too short to allow hemodynamic and metabolic stabilization to occur in all of patients because of excessive bleeding or immediate reoperations, although the rate of reoperations was equal in the study groups. The high proportion of nonresponders with volume substitution only in the protocol group may also be a source of bias.
In conclusion, the results show that therapy, targeting Svo2 >70% and lactate concentration ≤2 mmol/L immediately after cardiac surgery, improves outcome as shown in decreased use of hospital resources. Difference in hospital stay between groups was small, but assuming 1000 cardiac surgical operations per year, this could reduce the need for beds in the cardiothoracic surgical ward. The use of hospital resources is an adequate end point to reflect outcome both from the economical and quality of care point of view. We used the hospital and ICU length of stay as outcome measures, because prolonged stay in the ICU after cardiac surgery, although rare, has a major impact on postoperative use of intensive care and hospital facilities (5). In addition, achievement of the therapeutic targets in cardiac surgical patients was difficult, despite vigorous fluid resuscitation and use of inotropes.
1. Higgins TL, Yared J-P, Ryan T. Immediate postoperative care of cardiac surgical patients. J Cardiothorac Vasc Anesth 1996; 10:643–58.
2. Michalopoulos A, Tzelepis G, Pavlides G, et al. Determinants of duration of ICU stay after coronary artery bypass graft surgery. Br J Anaesth 1996; 77:208–12.
3. Mythen MG, Webb AR. Intra-operative gut mucosal hypoperfusion is associated with increased post-operative complications and cost. Intensive Care Med 1994; 20:99–104.
4. Christakis GT, Fremes SE, Naylor CD, et al. Impact of preoperative risk and perioperative morbidity on ICU stay following coronary bypass surgery. Cardiovasc Surg 1996; 4:29–35.
5. Ryan TA, Rady MY, Bashour A, et al. Predictors of outcome in cardiac surgical patients with prolonged intensive care stay. Chest 1997; 112:1035–42.
6. Shoemaker WC, Montgomery ES, Kaplan E, Elwyn DH. Physiologic patterns in surviving and nonsurviving patients: use of sequential cardiorespiratory variables in defining criteria for therapeutic goals and early warning of death. Arch Surg 1973; 106:630–6.
7. Shoemaker WC, Appel PL, Kram HB, et al. Prospective trial of supranormal values of survivors as therapeutic goals in high-risk surgical patients. Chest 1988; 94:1176–86.
8. Boyd O, Grounds RM, Bennett ED. A randomized clinical trial of the effect of deliberate perioperative increase of oxygen delivery on mortality in high-risk surgical patients. JAMA 1993; 270:2699–707.
9. Tuchschmidt J, Fried J, Astiz M, Rackow E. Elevation of cardiac output and oxygen delivery improves outcome in septic shock. Chest 1992; 102:216–20.
10. Yu M, Levy MM, Smith P, et al. Effect of maximizing oxygen delivery on morbidity and mortality rates in critically ill patients: a prospective, randomized, controlled study. Crit Care Med 1993; 21:830–8.
11. Hayes MA, Timmins AC, Yau EHS, et al. Elevation of systemic oxygen delivery in the treatment of critically ill patients. N Engl J Med 1994; 330:1717–22.
12. Gattinoni L, Brazzi L, Pelosi P, et al. A trial of goal-oriented hemodynamic therapy in critically ill patients. N Engl J Med 1995; 333:1025–32.
13. Berlauk JF, Abrams JH, Gilmour IJ, et al. Preoperative optimization of cardiovascular hemodynamics improves outcome in peripheral vascular surgery. Ann Surg 1991; 214:289–97.
14. Bishop MH, Shoemaker WC, Appel PL, et al. Prospective randomized trial of survivor values of cardiac index, oxygen delivery, and oxygen consumption as resuscitation endpoints in severe trauma. J Trauma 1995; 38:780–7.
15. Durham RM, Neunaber K, Mazuski JE, et al. The use of oxygen consumption and delivery as endpoints for resuscitation in critically ill patients. J Trauma 1996; 41:32–40.
16. Ziegler DW, Wright JG, Choban PS, Flancbaum L. A prospective randomized trial of preoperative “optimization” of cardiac function in patients undergoing elective peripheral vascular surgery. Surgery 1997; 122:584–92.
17. Wilson J, Woods I, Fawcett J, et al. Reducing the risk of major elective surgery: randomised controlled trial of preoperative optimization of oxygen delivery. BMJ 1999; 318:1099–103.
18. Mythen MG, Webb AR, Perioperative plasma volume expansion reduces the incidence of gut mucosal hypoperfusion during cardiac surgery. Arch Surg 1995; 4:423–9.
19. Routsi C, Vincent J-L, Bakker J, et al. Relation between oxygen consumption and oxygen delivery in patients after cardiac surgery. Anesth Analg 1993; 77:1104–10.
20. Pölönen P, Hippeläinen M, Takala R, et al. Relationship between intra- and postoperative oxygen transport and prolonged intensive care after cardiac surgery: a prospective study. Acta Anesth Scand 1997; 41:810–17.
21. Heyland DK, Cook DJ, King D, et al. Maximizing oxygen delivery in critically ill patients: a methodologic appraisal of the evidence. Crit Care Med 1996; 24:517–24.