Sufentanil

Study objectives: Neurologic disorders belong among the most serious complications of cardiac surgery. We tested the hypothesis that combinations of hemodynamic events from apparently normal cardiopulmonary bypass (CPB) procedures are related to the development of postoperative neurologic complications and affect the impact of common clinical risk factors.

Design: Retrospective study.

Setting: Cardiothoracic surgery department in a university hospital. Methods and patients: A multivariate statistical procedure (ie, cluster analysis) was applied to a data set of automatically recorded perfusions from 1,395 patients who had undergone coronary artery bypass grafting. One-way analysis of variance was used to select five parameters with the strongest significant correlation to postoperative neurologic complications for further cluster analysis. The dependencies in the clusters were tested against common clinical risk factors. To our knowledge, this is the first study of its kind.

Results: The following five parameters emerged for cluster analysis: mean arterial pressure (MAP); dispersion of MAP; dispersion of systemic vascular resistance; dispersion of arterial pulse pressure; and the maximum value of mixed venous saturation. Using these parameters, we found four clusters that were significantly different by CPB performance (first cluster, 389 patients; second cluster, 431 patients; third cluster; and fourth cluster, 229 patients). The frequency of postoperative neurologic complications was 0.3% in the first cluster and increased to 3.9% in the fourth cluster. Importantly, the impact of common clinical risk factors for postoperative neurologic complications was affected by the performance of the CPB procedure. For example, the frequency of neurologic complications among patients with cerebrovascular disease in their medical history was 22% in the fourth cluster, whereas it was zero in the second cluster.

Conclusions: This study shows that apparently normal CPB procedures affect the impact of common clinical risk factors on postoperative neurologic complications. Patients who underwent CPB procedures with large fluctuations in hemodynamic parameters particularly showed an increased risk for the development of postoperative neurologic complications.

Key words: CABG; cardiopulmonary bypass; postoperative neurologic complications; risk factors

Abbreviations: APpulse = arterial pulse pressure; BFI = blood flow index; CABG = coronary artery bypass grafting; CPB = cardiopulmonary bypass; MAP = mean arterial pressure; SV[O.sub.2] = mixed venous saturation; SVR = systemic vascular resistance

**********

CNS disorders after coronary artery bypass grafting (CABG) significantly increase perioperative mortality and hospitalization time, and can lead to a decrease in the patient's quality of life. The reported rate of major postoperative neurologic complications, such as stroke, varies from 0.8 to 6%. (1-4) Mechanisms causing these complications are not well understood. Previous studies have focused primarily on the separate identification of demographic and medical risk factors. A number of risk factors thus have been identified (eg, age, female sex, craniocervical and aortic atherosclerosis, previous cerebral stroke, hypertension, duration of the operation and cardiopulmonary bypass [CPB], use of the intra-aortic balloon pump, and postoperative cardiac arrhythmias). (2-7) As to the intraoperative factors, neurologic complications after cardiac operations have been attributed mainly to the effects of CPB. (8) Reported potential mechanisms are the microembolization and macroembolization of gas, particle matter, aim inadequate cerebral perfusion pressure or flow. However, studies on the impact of conventional, uncomplicated CPB are rare and usually are performed on small, selected groups of patients. Furthermore, the multifactorial nature of neurologic dysfunction alter CPB makes the results of such investigations controversial.

We undertook this retrospective study to determine which combination of preoperative and intraoperative risk factors, including the characteristics of CPB, might predispose CABG patients to neurologic complications following surgery. The use of a fully automated system of perfusion record-keeping provided continuous data and eliminated the possible bias that is inherent to manual record entry. Groups of patients with similar perfusion characteristics were separated by means of cluster analysis to study the interaction between the characteristics of CPB and common clinical risk factors. To our knowledge, this is the first study of its kind.

MATEIUALS AND METHODS

Patients

Retrospectively, 1,408 isolated CABG operations that had been carried out between May 1996 and January 1999 in the University Hospital Maastricht were reviewed. Patients with incomplete intraoperative or postoperative data were excluded from study.

Intraoperative Patient Management

The conduct of anesthesia and CPB was standardized by hospital practice. Anesthesia was induced with midazolam (0.06 to 0.08 mg/kg), sufentanil (1 to 1.5 [micro]g/kg), and pavulon 0.1 mg/kg), and was maintained with sufentanyl (0.5 to 1.25 [micro]g/kg/h) and/or propofol infusion (3 to 6 mg/kg/h). The bypass circuit consisted of a hollow fiber oxygenator (Univox-IC; Bentley/ Baxter; Irvine, CA; or Capiox SX: Terumo Corporation; Tokyo, Japan), an arterial line filter (Pall Autovent-SV; Pall Biomedical Ltd; Portsmouth, UK), and a roller pump (Stockert Instrumente GmbH; Munich, Germany). The circuit was primed with a mixture of colloidal intravenous infusion solution (Haemaccel; Beacon Pharmaceuticals; Tunbridge Wells, UK) [3.5%; 1,200 to 1,300 roll, mannitol (211% solution; 200 mL), human albumin (20%; 100 mL), NaHC[O.sub.3] (8.4%; 50 mL), potassium chloride (20 mmol), and heparin (50 mg/1,000 mL colloidal intravenous infusion solution). CPB was carried out by means of pulsatile flow with a flow range of approximately 2.4 L/min/[m.sup.2]. Moderate hypothermia (ie, temperature, 28[degrees]C to 32[degrees]C) was applied during the CPB. A modified St. Thomas solution was used to arrest the heart and to maintain it in an isoelectric state. After the release of the aortic clamp, a nitroglycerine infusion (0.5 [micro]g/kg/min) was started.

Data Collection

MI demographic and clinical data were stored ill using a software program (Summit Medical Systems: Minneapolis, MN). Postoperative neurologic complications were defined as follows: permanent or transient central neurologic deficit; mental disturbance marked by illusions, confusion, and cerebral excitement; and unresponsive coma for > 24 h. In all cases, patients with possible neurulogic complications were investigated by a neurologist. This as well as other definitions for clinical parameters were derived from the terms of the Society of Thoracic Surgeons National Cardiac Surgery Database (Table 1).

In 1996, the department of Extra-Corporeal Circulation introduced a computer program (homemade) for the continuous registration of perfusion parameters. Routinely, every 6 s, 21 parameters of CPB are read and stored in a data file. For the present study, the following parameters were extracted from the perfusion database: mean arterial pressure (MAP); arterial pulse pressure (APpulse); blood flow index (BFI); systemic vascular resistance (SVR); arterial blood temperature; rectal temperature; Pa[O.sub.2]; and mixed venous saturation (Sv[O.sub.2]). Also the duration of full extracorporeal circulatory support (ie, aortic cross-clamp period or flail bypass) was included in the analysis. Continuous perfusion parameters from each single CPB procedure were included in the study as their mean, maximum, and minimum values during each single procedure. As a degree of the scattering of data, we also included a variable that we called dispersion. Dispersion is a measure that points out how often and how far the values of the included parameters changed through the full period of CPB for each single procedure. Thus, continuous parameters in our study were presented by their four properties, as follows: mean; dispersion; maximum; and minimum.

Data Analysis

>From the initial 1,408 cases, 13 patients were excluded from further analysis because of incomplete or missing data from the CPB. None of these excluded patients had severe postoperative neurologic complications. Subsequently, a set of data from 1,395 cases was analyzed within a multidimensional space of 32 variables in order to identify data-dependent groups of perfusion patterns. One-way analysis of variance was used to compare the values of perfusion variables between patients who developed postoperative neurologic complications and those in whom such complications did not occur (Fig 1).

[FIGURE 1 OMITTED]

A cluster analysis was used to look for "subpopulations" of patients, with respect to a selection of perfusion variables that were clinically relevant for postoperative neurologic complications. The following two procedures for cluster analysis were used: a hierarchical cluster analysis; followed by a K-means cluster analysis. This was done to fully profit from the strength of both procedures. The K-means cluster analysis is the ideal analysis for large patient groups, however, the algorithm requires one to specify the number of clusters. We wanted these tests to be as objective as possible. Therefore, we first performed a hierarchical cluster analysis in which the decision making on the number of clusters to be used is automated. The drawback of the hierarchical cluster analysis is the fact that it is not specifically designed for large patient groups. Therefore, we randomly selected (SPSS software; SPSS Inc; Chicago, IL) a group of 387 case patients (28% of the total population) for the hierarchical cluster analysis. Thus, the steps for the total clustering process were as follows.

The first step in the duster analysis was the selection of variables to identify a likely set of variables to "carve" the patient population into subgroups (Fig 1). A preliminary analysis revealed that patients with and without complications in the early postoperative period had undergone perfusions that differed from each other on more than a dozen parameters. For the cluster procedure, we selected the most representative parameters from groups of closely correlated parameters (see "Results" section). As an exception to this rule, we included the properties of the MAP in the cluster procedure. This exception was based on the hypothesis that the MAP during CPB is of high clinical significance. Thus, five parameters (ie, mean MAP, dispersion of MAP, dispersion of SVR, dispersion of APpulse, and maximum Sv[O.sub.2]) were selected for the duster analysis. From these parameters, standardized values (ie, Z-score variables) 'also were generated. This Z-score transformation was done to prevent the variables with greater variability to dominate the clustering.

The second step was a hierarchical cluster analysis on a random subset of cases (387 cases, as described above), which was performed with the Z-score variables generated before (Fig 1). During the clustering process, clusters were formed by grouping patients into bigger and bigger clusters until all patients were members of a single duster. The results of the hierarchical cluster analysis were summarized in a so-called agglomeration schedule, which was presented as a table. This agglomeration schedule showed us the step-by-step clustering process and identified the clusters being combined at each stage. The results of this agglomeration schedule in a later phase helped us to decide how many clusters should be included in our present study. The cluster method chosen was the between-groups linkage, and the dissimilarity measure we chose was the "squared Euclidean distance." The latter allowed us to specify the distance or similarity measure to be used in clustering.

In the third step, one of the variables in the agglomeration schedule, the coefficients, were used to select a four-cluster solution for further analysis. Then, a K-means cluster analysis based on the Z-scores of the five selected parameters was used to cluster all cases (9) (Fig 1).

In order to study the interactions of patient age, duration of CPB, and development of neurologic complications, these continuous numeric data were converted to four discrete categories using the data transpose option of the software package (ie, SPSS; SPSS Inc) [Table 1]. Analyses of the qualitative variables throughout the study" were made using the [chi square] test or extensions of the two-tailed Fisher exact test. One-way analysis of variance was used to compare the mean values of the quantitative variables in clusters. The multiple comparison Tukey honestly significant difference test or the Tamhane T2 test was used to find which means were different in the clusters. Data analyses were performed using a statistical software package (SPSS, version 9.01; SPSS/PC+; Chicago, IL).

RESULTS

Among the 1,395 patients enrolled in our final analysis, 27 (1.9%) had severe neurologic complications in the postoperative period. Demographic and clinical characteristics of all patients are shown in Table 2. A direct comparison of both patient groups points out that patients with postoperative neurologic complications were older, had previously experienced a cerebral stroke more often, had undergone more distal anastomoses, and had longer CPB and cross-clamp times. Postoperatively, the development of neurologic complications was associated with a number of other postoperative complications included in our study. Postoperatively, the majority of patients with a neurologic complication had this in combination with at least one other postoperative complication included in the analyses (22 of 27 patients; 81.5%). The most common complication to coincide with a neurologic complication was cardiac arrhythmia (21 of 27 patients; 77.8%).

CPB parameters in patients with or without postoperative neurologic complications are presented in Table 3. For the majority of parameters included in the present study, dispersion was significantly higher in patients with postoperative neurologic complications.

The separation of patients into four clusters is shown in Figure 2. Final cluster centers and related parameters are shown in Table 4. Clusters differed by all studied parameters, except for the maximum value of the arterial blood temperature. The first cluster united patients with a relatively "warm," stable lull bypass period, as shown by the low dispersion values for all parameters. On the contrary, the fourth cluster united CPB procedures with high alterations, especially of the hemodynamic parameters. Table 4 shows that CPB procedures in the fourth cluster had the highest dispersion of MAP (mean [[+ or -] SD] MAP dispersion, 13.7 [+ or -] 3.2 mm Hg). In the fourth cluster, the MAP dining CPB varied from 38 [+ or -] 8 to 97 [+ or -] 12 mm Hg. In other words, the MAP during this type of CPB procedure varied by almost 60 mm Hg. The same tendency was found for the BFI. The mean dispersion of blood flow during CPB in cluster 4 was 0.2 [+ or -] 0.1 L/min/[m.sup.2] (p < 0.001 compared to any other cluster). The mean value of the BFI in CPB procedures from the cluster 4 was significantly lower than those in CPB procedures from cluster 1 (p < 0.001), but they did not differ from those in CPB procedures included in the clusters 2 and 3. The lower BFI in these clusters could be related to a lower rectal temperature during CPB. A relatively high MAP in combination with low values of blood flow arbitrated that clusters 3 and 4 had the highest mean values of SVB during full bypass (1,440 [+ or -] 192 and 1,455 [+ or -] 266 dyne*s*[cm.sup.-5], respectively). The fourth cluster had the highest values of SVR with an average maximum value up to 3,093 [+ or -] 1,860 dyne*s*[cm.sup.-5] (p < 0.001 compared to any other cluster).

[FIGURE 2 OMITTED]

The clusters were also different in terms of demographic, clinical, and intraoperative characteristics, as shown in Table 5. Table 5 shows that the fourth cluster contains the oldest patients, the highest percentage of ferule patients, and patients with cerebrovascular disease in their medical histories. The cluster means of CPB duration, cross-clamp time, and number of distal anastomoses were statistically different among clusters (p < 0.001). Post hoc testing showed that the mean CPB duration in the second and third clusters and in the third and fourth clusters were equal (Table 5). Also, although a significant difference was found for the aortic cross-clamping times between clusters, the aortic cross clamp times in clusters 3 and 4 were identical. The mean number of distal anastomoses was significantly lower in the first duster compared to all other clusters (p < 0.05) and was identical for clusters 2 to 4.

The frequency of postoperative neurologic complications was the lowest in the first cluster (0.3%) and increased up to 3.9% in the fourth cluster (p < 0.01) [Table 5]. From the 389 patients in the first cluster, only 1 patient (0.3%) had a postoperative neurologic disorder, which was experienced in combination with cardiac arrhythmias and renal failure. In the first cluster, 12 patients (3.1%) had a cerebral incident in their medical history. Remarkably, none of them developed neurologic complications in the postoperative period. The frequency of postoperative neurologic complications in the second cluster was 2.3% (10 patients), and the occurrence of this was associated with the occurrence of postoperative cardiac arrhythmias (p < 0.001). We did not find any association between the development of neurologic complications and any other demographic or clinical characteristic in this cluster. In the second cluster, 35% of all patients with cerebrovascular disease in their medical history were included. None of these patients experienced a postoperative neurologic complication. Also, none of the 16 patients with carotid disease who were included in this cluster experienced a postoperative neurologic complication. In the third cluster, the frequency of postoperative neurologic complications was 2.0%. Cluster 3 contained 22 patients with a medical history of cerebral disease. Two of these patients (9.1%) developed neurologic complications postoperatively, but this association was not signit3cant. However, the occurrence of postoperative neurologic complications in patients of the third cluster was associated with carotid disease and postoperative renal failure (p < 0.01 and p < 0.0.5, respectively). In the fourth cluster, the development of postoperative neurologic complications was associated with the occurrence of a perioperative myocardial infarction and postoperative cardiac arrhythmias (p < 0.01 and p < 0.001, respectively). From nine patients in the fourth cluster who experienced postoperative neurologic complications, eight patients (88.9%) had a combination of several postoperative complications. The occurrence of postoperative neurologic complications in this group of patients was particularly associated with hypertension and the presence of cerebrovascular disease in their medical histories (p < 0.05 and p < 0.001, respectively). Despite the differences between the clusters with respect to gender, age, and CPB duration (Table 5), we did not find a statistically significant association between these factors and postoperative neurologic complications in any of the clusters.

DISCUSSION

Neurologic complications after coronary artery bypass surgery substantially increase mortality, put a strain on health-care resources, and reduce the clinical effectiveness of the procedure. (8,10,11) Customarily, these complications have been attributed mainly to the effects of CPB. (8) Nowadays, complications associated with the use of CPB have gained even more attention due to the rediscovering and growing interest in "off-pump" CABC. The reported incidence of neurologic complications after CABC with CPB varies from 0.8 to .3.2% in retrospective studies (1,12-14) and from 1.5 to 5.2% in prospective studies. (2,15,16) The frequency of severe postoperative neurologic disorders in the present study (1.9%) is in line with these previous findings.

Previous studies identified individual factors related to postoperative neurologic complications. Among them were age, gender, arterial hypertension, diabetes mellitus, carotid stenosis, previous cerebrovascular disease, preoperative and postoperative arrhythmias, requirement of using an intra-aortic balloon pump, duration of CPB and aortic cross-clamping, and the number of anastomoses performed. (9,14,17-22) However, in most of these studies these factors were examined individually, which led to conflicting reports in the literature. (15) For example, some studies, (8,18) have demonstrated that advanced age and previous cerebrovascular disease are powerful predictors of postoperative stroke, whereas others, (3,4,23) have not found such an association. The results from our direct comparison of groups of patients with and without postoperative neurologic complications are consistent with those of the studies in which age and a history of cerebral stroke increased the probability of postoperative neurologic complications. Also, in the present study, the group of patients with postoperative neurologic complications had more distal anastomoses performed, and had longer CPB and cross-clamp times. However, other common risk factors were not significantly associated with the incidence of postoperative neurologic complications in the present. Gender, hypertension, carotid disease, and diabetes mellitus all were not directly associated with these complications. Our findings, however, support reports in which postoperative neurologic events were linked to the development of cardiac arrhythinias, (24) myocardial infarction, (25) or renal failure. (25,26)

CPB is still an important, intrinsic part of coronary artery bypass surgery. Yet, data on the impact of apparently normal CPB procedures on the development neurologic complications are rare and usually are from studies performed in small, selected groups of patients. Furthermore, these studies were mainly focused on a single characteristic of CPB (eg, temperature, (27) pulsatility of flow, (28) or acid-base management (29)). The multifactorial nature of nenrologic complications after CPB makes the results of such investigations controversial. Determining "tree" predictors of postoperative neurologic complications is furthermore hampered by the small number of outcome events in data collection (16,30) and by the low incidence of major ueurologic complications. (1-4) This and the multifactorial etiology of postoperative neurologic complications (31) do not favor the use of conventional statistical methods. The cluster analysis used in the present study was developed in the 1960s and has been used in studies of pattern recognition, artificial intelligence and neural network development, social sciences, and epidemiology. (9) Until now, however, this method rarely has been applied in clinical studies. (31-34) In the present study, the separation of large groups of similar CPB characteristics by means of cluster analysis enabled us to overcome the above-mentioned limitations. Despite the small number of patients with postoperative neurologic complications, the use of cluster analysis enabled us to study the interaction between risk factors for postoperative neurologic complications and CPB. Most of the known risk factors appeared not to be the ultimate, independent predictors that predispose CABG patients to postoperative neurologic complications, yet their impact was shown to be altered by the performance of the CPB procedure.

In the present study, patients in the fourth cluster had the most frequent fluctuations of hemodynamic parameters, which resulted in an almost 10-fold increase of the risk for major postoperative neurologic complications compared to patients in the first cluster. Furthermore, in patients with cerebrovascular disease in their medical history who were included in the fourth cluster, the incidence of postoperative neurologic complications reached 22.2%. Oil the contrary, in the second cluster, in which 35% of all patients with cerebrovascular disease in their medical history were merged, none of the patients experienced a postoperative neurologic complication. Also, only in the fourth cluster was the development of major postoperative neurologic complications associated with a history of hypertension and/or cerebrovascular disease. Surprisingly, the occurrence of postoperative neurologic complications was associated with carotid disease only in patients in cluster 3.

The results of our direct comparison of groups with and without postoperative neurologic complications are consistent with those of studies in which age and duration of CPB and clamp-time are significant predictors of postoperative neurologic complications. (15) However, the impact of these factors disappeared when we analyzed them using cluster analysis. However, what we found in the cluster analysis, as well as in a direct comparison of groups with and without neurologic complications, was a strong association of postoperative neurologic complications with other postoperative complications. The development of severe postoperative neurologic events in all but the first clusters was statistically related to at least one coexistent complication. However, the interrelationship of these events makes the identification of the relative contribution of each single variable difficult. Therefore, further studies are necessary to elucidate the relationship of these complications.

In conclusion, this study shows that apparently normal CPB procedures affect the impact of common clinical risk factors on postoperative neurologic complications. Particularly, patients who underwent CPB procedures with large fluctuations in hemodynamic parameters showed an increased risk for the development of postoperative neurologic complications.

REFERENCES

(1) Carrascal Y, Guerrero AL, Maroto LC, et al. Neurological complications after cardiopulmonary bypass: an update. Eur Neurol 1999; 41:128-134

(2) Bull DA, Neumayer LA, Hunter GC, et al. Risk factors tot stroke in patients undergoing coronary artery bypass grafting. Cardiovasc Surg 1993; 1:182-185

(3) Cernaianu AC, Vassilidze TV, Flum DR, et al. Predictors of stroke after cardiac surgery. J Card Surg 1995; 10:334-339

(4) Craver JM, Puskas JD, Weiniraub WW, et al. 601 octogenarians undergoing cardiac surgery: outcome and comparison with younger age groups. Ann Thorac Surg 1999; 67:1104-1110

(5) Herlitz J, Brandrup-Wognsen G, Haglid M, et al. Mortality and morbidity during a period of 2 years after coronary artery bypass surgery in patients with and without a history of hypertension. J Hypertens 1996; 14:309-314

(6) Gardner TJ, Horneffer PJ, Manolio TA, et al. Major stroke after coronary artery bypass surgery: changing magnitude of the problem. J Vasc Surg 1986; 3:684-687

(7) Alexander KP, Anstrom KJ, Muhlbaier LH, et al. Outcomes of cardiac surgery in patients > or - 80 years: results from the National Cardiovascular Network. J Am Coll Cardiol 2000; 35:731-738

(8) Murkin JM, Etiolegy and incidence of brain dysfunction after cardiac surgery. J Cardiothorac Vasc Anesth 1999; 13:12-17

(9) Kaufman L, Rousseeuw PJ. Finding groups in data: an introduction to cluster analysis. New York, NY: John Wiley & Sons, 1990

(10) Taylor KM. Central nervous system effects of cardiopulmonary bypass. Ann Thorac Surg 1998; 66:S20-S24

(11) Utley JR. Techniques for avoiding neurologic injury during adult cardiac surgery. J Cardiothorac Vasc Anesth 1996; 10:38-43

(12) Pompilio G, Lotto AA, Agrifoglio M, et al. Nonembolic predictors of stroke risk in coronary artery bypass patients. World J Surg 1999; 23:657-663

(13) Kolh PH, Torchiana DF, Buckley MJ, Atheroembolization in cardiac surgery: the need for preoperative diagnosis. J Cardiovasc Surg (Torino) 1999; 40:77-81

(14) Ahlgren E, Aren C. Cerebral complications after coronary artery bypass and heart valve surgery: risk factors and onset of symptoms. J Cardiothorac Vasc Anesth 1998; 12: 270-273

(15) McKhann GM. Goldsborough MA, Borowicz LM Jr, et al. Predictors of stroke risk in coronary artery bypass patients. Ann Thorac Surg 1997; 63:516 521

(16) Newman MF, Wolman R, Kanchuger M, et al. Multicenter preoperative stroke risk index for patients undergoing coronary artery bypass graft surgery: Multicenter Study of Perioperative Ischemia (McSPI) Research Group. Circulation 1996:94:1174-1180

(17) Hogue CW Jr, Murphy SF, Schechtman KB, et al. Risk factors for early or delayed stroke after cardiac surgery. Circulation 1999:100:642-647

(18) Redmond JM, Greene PS, Goldsborough MA, et al. Neurologic injury in cardiac surgical patients with a history of stroke. Ann Thorae Surg 1996; 61:42-47

(19) Dashe JF, Pessin MS, Murphy RE, et al. Carotid occlusive disease and stroke risk in coronary artery bypass graft surgery. Neurology 1997; 49:678-686

(20) Herlitz J, Brandrup-Wognsen G, Karlson BW, et al. Mortality, risk indicators, mode and place of death and symptoms of angina pectoris in the five years after coronary artery bypass grafting in patients with and without a history of hypertension. Blood Press 1999; 8:200-206

(21) Rolfson DB, McElhaney JE, Rockwood K, et al. Incidence and risk factors for delirium and other adverse outcomes in older adults after coronary artery bypass graft surgery. Can J Cardiol 1999; 15:771-776

(22) Morricone L, Ranucci M, Denti S, et al. Diabetes and complications after cardiac surgery: comparison with a nondiabetic population. Acta Diabetol 1999; 36:77-84

(23) Beall AC Jr, Jones JW, Guinn GA, et at. Cardiopulmonary bypass in patients with previously completed stroke. Ann Thorac Surg 1993; 55:1383-1384

(24) Ahnassi GH, Schowalter T, Nicolosi AC, et al. Atrial fibrillation after cardiac surgery: a major morbid event? Ann Surg 1997; 226:501-511

(25) Saimanen EI. Perioperative stroke in coronary artery bypass surgery: analysis of risk 5aetors. Seand Cardiovasc J 2000; 34:41-45

(26) Anderson RJ, O'Brien M, MaWhinney S, et al. Renal failure predisposes patients to adverse outcome after coronary artery bypass surgery: VA Cooperative Study #5. Kidney Int 1999; 55:1057-1062

(27) Insler SR, O'Connor MS, Leventhal MJ, et al. Association between postoperative hypothermia and adverse outcome after coronary artery bypass surgery. Ann Thorac Surg 2000; 70:175-181

(28) Chow G, Roberts IG, Harris D, et al. Stockert miler pump generated pulsatile flow: cerebral metabolic changes in adult cardiopulmonary bypass. Perfusion 1997; 12:113-119

(29) Stephan H, Weyland A, Kazmaier S, et al. Acid-base management during hypothermic cardiopulmonary bypass does not affect cerebral metabolism but does affect blood flow and neurological outcome. Br J Anaesth 1992; 69:51-57

(30) Kleinbaum DG, Kupper LL, Muller KE, et al. Applied regression analysis and other multivariable methods. 3rd ed. Pacific Grove, CA: Brooks/Cole Publishing Company, 1998

(31) Walzer T, Herrmann M, Wallesch CW. Neuropsychological disorders after coronary bypass surgery. J Neurol Neurosurg Psychiatry 1997; 62:644-648

(32) Knyshov HV, Palets BL, Nastenko Ie A, et al. The heart pumping function and the systemic regulation of blood circulation in groups of heart surgery patients. Fiziol Zh 1996: 42:3-18

(33) Marshall G, Henderson WG, Mofitz TE, et al. Statistical methods and strategies for working with large data bases. Med Care 1995; 33:OS35-OS42

(34) Marshsall G, Grover FL, Henderson WG, et al. Assessment of predictive models for binary outcomes: an empirical approach using operative death from cardiac surgery. Stat Med 1994: 13:1501-1511

* From the Departments of Extra-Corporeal Circulation (Drs. Ganushchak, Visser, and de Jong) and Cardiothoracic Surgery (Drs. Fransen and Maessen), University Hospital Maastricht, Maastricht, the Netherlands.

Manuscript received February 26, 2003; revision accepted May 16, 2003.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail: permissions@chestnet.org).

Correspondence to: Youri M. Ganushchak. PhD, Department of Extra-Corporeal Circulation, University Hospital Maastricht, P. Debyelaan 25, 6202 AZ Maastricht, the Netherlands; e-mail: yga@scpc.azm.nl

COPYRIGHT 2004 American College of Chest Physicians
COPYRIGHT 2004 Gale Group