Efficacy and Safety of Heparinase I  versus Protamine in Patients Undergoing Coronary Artery Bypass Grafting with and without Cardiopulmonary Bypass 

This was a prospective randomized, double-blind, active-controlled, double-dummy noninferiority clinical trial of patients undergoing CABG surgery, with the study medication (heparinase I) or standard care (protamine) randomized in a 1:1 ratio, stratified by on- or off-pump CABG surgery. After institutional approval and written informed consent were obtained, 167 subjects received study drug at 46 centers (range, 1–26 patients/site) in the United States, Canada, and Germany from February through September 2003 ( appendix 2).

Eligible patients were aged 18–85 yr old and received heparin as part of a primary nonemergent on- or off-pump CABG surgery. Exclusion criteria were concomitant surgical procedures, hematologic disease including heparin-induced thrombocytopenia, preoperative platelet count less than 150,000/mm³, an international normalized ratio exceeding 1.5 within 48 h of surgery, severe liver dysfunction, serum creatinine greater than 2.0 mg/dl, ejection fraction of 20% or less, and allergy to heparin or protamine. Patients receiving preoperative heparin infusion were eligible; however, those receiving eptifibatide (Integrilin; Millennium Pharmaceuticals, Inc., Cambridge, MA and Schering Corporation, Kenilworth, NJ) within 12 h, clopidogrel (Plavix; Bristol-Myers Squibb/Sanofi Pharmaceuticals Partnership, New York, NY) or low-molecular-weight heparin within 3 days, abciximab (ReoPro; Centocor, Inc., Malvern, PA) or ticlopidine (Ticlid; Roche Laboratories, Inc., Nutley, NJ) within 7 days of surgery, or with another condition that compromised surgical hemostasis or patient safety were also excluded. Patients whose procedure changed from on- to off-pump after randomization remained in the trial if the decision to switch was unrelated to patient safety or bleeding. These patients (3% of total) were grouped by performed rather than scheduled procedure and were included for analysis. To minimize the need for readministration of heparin after study drug, intraoperative eligibility criteria included confirmation of clinical stability by the attending physicians immediately before heparin reversal (i.e. , no major intraoperative concerns).

Eligible patients were stratified by study site and procedure (on- or off-pump CABG) and randomized preoperatively (1:1) to receive heparinase I or protamine. Unstable patients not receiving study drug were followed up postoperatively for safety criteria only.

Management of preoperative medications, anesthesia agents, antifibrinolytic therapy (for on-pump procedures), and cardiopulmonary bypass (CPB) management was per local practice. Although intraoperative heparin administration was carefully documented, there were no dosing guidelines, stipulations regarding type (porcine vs.  bovine), or units of measure (International Units vs.  United States Pharmacopeia Units). Performance of an activated clotting time test was required, in duplicate, for all time points when heparin effect was assessed. Because alternate point-of-care coagulation testing (Hepcon, Thromboelastograph) posed no risk of study drug unblinding, these tests were also permitted.

A protocol was developed for heparin reversal. A total dose of up to 650 mg protamine or 35 μg/kg heparinase I was available. The protamine regimen represented a consensus from all study sites, and heparinase I dosing was based on previous safety, preclinical, and clinical data.17Protocol development involved a high-fidelity simulation laboratory23,24; a draft protocol was evaluated and refined using observations from CABG surgery scenarios resembling the typical complexity of heparin-reversal decisions, including mandatory and optional drug doses, activated clotting time values, clinical cues, and hemodynamic perturbations. For all doses, study drug was administered as a combination of bolus (heparinase I or saline placebo) and infusion (protamine or saline placebo) in a double-dummy fashion.

Two mandatory and two optional doses of study drug were available as part of the heparin-reversal protocol (table 1). Dose 1 was a calculated mandatory initial dose, by body weight for heparinase I, and by cumulative intraoperative heparin dose for protamine. Dose 2 was an optional supplementary dose available 5–10 min after dose 1 as judged necessary by the clinical team. Dose 3 was a mandatory pump blood dose (for on-pump patients only) given after return of heparin-containing blood from the extracorporeal circuit; the protocol required blood return within 40 min of initiating heparin reversal. Dose 3 was not required if returned blood was washed by an autotransfusion device (e.g. , Hemonetics Cell Saver, Braintree, MA), and was optional if blood return was completed within 5 min of dose 2. Dose 4 was an optional postoperative dose for use within 12 h of intensive care unit (ICU) arrival. If residual heparin effect was suspected after available doses of study drug were administered, investigators could use open-label protamine, and the treatment was considered a failure in that subject. To minimize any potential effect of inexperience on performance of this complex protocol, education for all study monitors and coordinators, in addition to the standard investigator meeting, included participation in a full-day course including involvement in several simulations of the final protocol in an operating room setting.24Although guidelines to identify residual heparin effect requiring optional doses or open-label protamine were provided (e.g. , persistent increased activated clotting time, “wet” surgical field), these decisions were left to the judgment of the clinical team. As a safety consideration, education was provided about the need for immediate unblinding of drug assignment if heparin readministration was required within 1 h of study drug, because of the special heparin dosing requirements for patients receiving heparinase I. In addition, immediate unblinding for study drug was available during the study, and individualized heparin dosing instructions for urgent anticoagulation were provided for each patient.

The primary efficacy measurement was postoperative hemorrhage, defined as the cumulative chest tube drainage during the first 12 h after ICU admission.25,26Chest tube drainage was measured and recorded at hourly intervals, beginning with patient arrival in the ICU and continuing for 12 h, after which chest tube drainage was measured and recorded every 4 h until the 24th hour or until chest tubes were removed.

The secondary efficacy endpoint was hemodynamic stability, as determined by a composite of systolic arterial blood pressure decrease (> 30 mmHg) or pulmonary artery systolic pressure increase (> 40 mmHg with an increase of at least 10 mmHg from baseline) within 30-min of study drug dosing. Baseline hemodynamic parameters were the average values during the 5 min after CPB separation and before heparin reversal. Electronic data capture from an automated anesthesia record-keeping system was used to record minute-to-minute hemodynamic parameters. A freestanding laptop-based system (Saturn Information System; Draeger Medical Inc., Telford, PA) was provided if electronic data capture was not already available. Interventions to treat hemodynamic instability were also documented.

Tertiary measurements were used to derive a composite morbidity score reflecting the percentage of patients who died within 30 days of surgery or sustained at least one of the following severe adverse events known to be related to protamine reactions: congestive heart failure, catastrophic pulmonary hypertension or acute myocardial depression within 1 h of study drug, anaphylaxis, acute renal failure/dialysis, stroke/cerebral vascular accident, or encephalopathy/delirium.

Safety was assessed on the basis of adverse events and serious adverse events. Adverse events were recorded throughout the patients' hospitalization and summarized by severity and relation to study drug. Serious adverse events were recorded throughout the 30-day duration of the study. An independent Data Safety Monitoring Board, comprised of five individuals with expertise in the areas of cardiothoracic surgery, cardiovascular anesthesia, heparinization, and biostatistics, from around the United States reviewed safety analyses at regular scheduled intervals.

A statistical plan was developed before the start of the study and followed for this analysis. Unless otherwise noted, patients in the on-pump CABG stratum were analyzed separately from those in the off-pump CABG stratum.

The primary analysis included an as-treated data set and comprised all patients who received study drug. Preliminary analyses suggested that chest tube drainage was skewed toward higher values; therefore, statistical analysis was conducted using a natural logarithm scale. A regression model was developed for the ln (chest tube drainage) and included the treatment allocation and clinical center as predictors, as well as preidentified bleeding risk factors including age, sex, body weight, preoperative hemoglobin, total dose of heparin, and duration of CPB (in the on-pump CABG group only). From the characteristics of the log-normal distribution, the median chest tube drainage, adjusted for covariates, is the exponent of the adjusted mean on the natural logarithm scale. Medians were estimated separately by treatment arm, and the heparinase I minus protamine difference was derived. Bootstrap procedures were applied, and the empirical, one-sided, 95% confidence interval was derived.

The secondary efficacy endpoint was hemodynamic stability as described above (see Efficacy and Assessment). The tertiary efficacy endpoint was a composite morbidity score including all cause mortality, congestive heart failure, catastrophic pulmonary hypertension, acute myocardial depression, anaphylaxis, acute renal failure, stroke/cerebral vascular accident, and encephalopathy/delirium. The proportions of patients meeting these criteria were compared in a stratified analysis using the Cochran-Mantel-Haenszel procedure.

The incidences of adverse events and serious adverse events were compared using the Fisher exact test. Several endpoints were examined, including all adverse events, hemodynamic changes, 12-h chest tube drainage, safety variables (including death), reoperation, vital signs and physical examination, and open-label protamine use after study drug administration. Composite variables representing the need for intervention to stabilize the patient hemodynamically because of hypotension or pulmonary hypertension within 30 min after study drug administration were similarly examined. Changes from baseline values in laboratory tests and vital signs were compared using a one-way analysis of variance test.

Based on a review of the literature and a broad background of clinical experience, a group of clinicians were invited to develop a noninferiority bleeding threshold; the group included several experienced cardiac anesthesiologists, surgeons, and established clinician–scientist investigators in the field. The consensus opinion was that a difference in 12-h chest tube drainage of less than or equal to 400 ml after on-pump CABG surgery was clinically acceptable. Using previously published clinical data for estimates of 12-h chest tube drainage differences between patients receiving protamine and heparinase I for on-pump procedures,17power and sample size calculations for 12-h chest tube drainage were performed separately for the on- and off-pump CABG strata. The 12-h chest tube drainage estimates for off-pump CABG patients were proportionately reduced 10% relative to the on-pump CABG stratum, based on published reports.27–34Bootstrap procedures were performed 1,000 times for each strata, with empirical power defined as the proportion of the 1,000 simulations in which the one-sided 95% confidence interval for the difference between medians was less than or equal to the noninferiority margin. The sample size was then successively changed until 90% power was achieved. To achieve the minimum 90% power, a study sample size of 300 (150 in each group) for on-pump CABG patients and 150 (75 in each group) for off-pump CABG patients was required. To be conservative, the sample size was set at 600 total patients: 400 on-pump and 200 off-pump CABG. All summaries and statistical analyses were generated using statistical software (SAS version 8.2; SAS Institute, Cary, NC).

The 167 subjects undergoing CABG surgery received study drug between February and September 2003, including 46 off-pump (protamine/heparinase I, 23/23) and 131 on-pump (protamine/heparinase I, 64/57) patients. All patients receiving study drug were included in the final analysis, which represents just over a quarter of the expected final data set because of the early discontinuation of the study, an action based on a recommendation from the study Data Safety Monitoring Board.

Overall, patients receiving heparinase I were older (65.6 vs.  62.1 yr; P = 0.04; table 2), more likely to have a history of congestive heart failure (11.3 vs.  3.5%; P = 0.05; table 2), had a higher baseline heart rate in the off-pump CABG group (68 ± 14 vs.  58 ± 10 beats/min; P = 0.008), and were less likely to have a history of peripheral vascular disease in the on-pump CABG group (4 vs.  14%; P = 0.04). No substantial differences were noted in the preoperative characteristics between those patients who underwent on- or off-pump CABG surgery. The median total intraoperative heparin dose was 40,000 units (46,000 units for on-pump CABG patients, 20,500 units for off-pump CABG patients). Overall, 84% patients had two to four bypass grafts, and an internal mammary artery graft was used in 99% of all procedures. The mean operative time was 4.7 h, with the duration of CPB in the on-pump CABG group averaging 84 min. A total of 14 patients received open-label protamine, and 2 patients were unblinded because they required readministration of heparin anticoagulation after receiving study drug; both patients were in the protamine group.

In addition to the 167 who received study drug, 34 patients (16.5%) were randomized but withdrawn before receiving study drug; the most common reasons were meeting exclusion criteria or investigator withdrawal. During the drug administration period, 47% of patients received an optional dose 2, and 18% received an optional postoperative dose 4. In the on-pump CABG group, 80 of 131 patients received dose 3; washing of heparin-containing blood before return to the patient accounted for the majority of omitted dose 3s.

The use of heparinase I was associated with an increased level of chest tube drainage compared with protamine; however, this difference did not exceed the predefined noninferiority margin (table 3). The adjusted heparinase I minus protamine difference in 12-h chest tube drainage was +181 ml (230 ml, one-sided 95% confidence interval) for on-pump CABG surgery and +310 ml (361 ml, one-sided 95% confidence interval) for off-pump CABG surgery.

There was no significant difference in the incidence of hemodynamic instability between the heparinase I and protamine groups (46.8% vs.  46.3%; P = 0.98; table 4). Similarly, there were no major differences in the rate of intervention to treat hemodynamic changes. No significant differences were noted between the heparinase I and protamine groups, adjusted for procedure type, for a composite morbidity score including death, major organ dysfunction, and rehospitalization (55.0% vs.  46.0%; P = 0.24; table 5).

Although efficacy analyses demonstrated noninferiority of heparinase I compared with protamine for heparin reversal, an inferior safety profile for heparinase I emerged when the study was approximately one quarter complete, prompting the Data Safety Monitoring Board to recommend discontinuation of the study. No patients were enrolled after the recommendation was issued; however, data collection was completed for all enrolled patients.

Adverse events and serious adverse events were common in both study groups; at least one adverse event or serious adverse event occurred in 93% and 99% of patients receiving protamine or heparinase I, respectively. Serious adverse events were more common in the heparinase I group compared with the protamine group (42.5 vs.  24.1%; P = 0.01). Although both the on- and off-pump strata demonstrated the same trend, the difference in serious adverse events between the protamine and heparinase I arms was more pronounced in the off-pump strata (off-pump: 13 vs.  43.5%; P = 0.02; on-pump: 28.1 vs.  42.1%; P = 0.11). No subset of complications grouped by body system was primarily responsible for this observation.

Several parameters reflected a pattern of increased bleeding and transfusion in the heparinase I group (table 6). Lowest hemoglobin values were lower in the heparinase I group at 24 h after surgery and before hospital discharge, and these patients were also more likely to have received a transfusion at these time points. In addition, the median duration of hospital stay was 1 day longer for patients in the heparinase I group than in the protamine group (6 vs.  5 days; P = 0.04). Although similar patterns of increased bleeding, transfusion, and extended hospital stay were evident in both strata, more findings reached statistical significance in the off-pump CABG group, despite the smaller numbers in this group.

One patient in the protamine group having on-pump CABG surgery died within 30 days of surgery; death was attributed to saphenous vein graft thrombosis and occurred in the operating room.

As a potential alternate to protamine that deactivates heparin using a different mechanism, the current noninferiority study was designed to demonstrate that heparinase I reverses heparin anticoagulation safely and without excess bleeding. We found that heparinase I was associated with an increased level of chest tube drainage compared with protamine, although the difference in bleeding did not exceed the predefined noninferiority threshold. This finding was made in an analysis adjusted to account for a priori  identified bleeding risk factors; however, the findings were almost identical in an unadjusted analysis. The secondary efficacy endpoint was hemodynamic instability related to heparin reversal. Curiously, despite previous indications of benefit with heparinase I,16,35we observed no difference in the frequency of hemodynamic instability between the two drugs in our study. Finally, the tertiary outcome was a composite morbidity index, reflecting the percentage of patients sustaining at least one major complication after their procedure. We found no difference between groups in this composite of all-cause mortality, rehospitalization, or major organ dysfunction. However, an overall inferior safety profile of heparinase I relative to protamine caused the study to be terminated on the advisement of the Data Safety Monitoring Board before complete enrollment. Although adverse events rates were similar, there were almost twice as many patients sustaining at least one serious adverse event in the heparinase I group (P = 0.01). Heparinase I patients also had lower hemoglobin values (P < 0.005) both at 24 h postoperatively and at hospital discharge, despite having greater transfusion rates (P < 0.006). The median duration of hospital stay was 1 day longer for patients in the heparinase I group than in the protamine group (6 vs.  5 days; P = 0.04). Therefore, our study does not support the use of heparinase I as a protamine replacement for elective CABG surgery.

Although no peer-reviewed studies have reported comparisons of protamine and heparinase I in cardiac surgery patients, some data exists on this subject. Heres et al.  35reported safety data in a dose-escalation study describing heparin reversal in 49 patients receiving between 5 and 20 μg/kg heparinase I; however, these treatments were not compared with protamine therapy. Two previous double-blind, double-dummy studies compared heparinase I with protamine17; heparinase I was administered in smaller and differently timed doses compared with those used in the current study. The first involved 176 patients undergoing CABG surgery and found a single 7-μg/kg postbypass bolus dose of heparinase I for heparin reversal to be inferior to 1 mg protamine per 100 units heparin (infused over 10 min) with regard to bleeding. This study also demonstrated insufficient heparinase I dosing. The second trial, evaluating 10 and 15 μg/kg heparinase I with optional additional doses, was terminated early for business reasons. Analysis of 94 enrolled patients was not sufficiently powered to demonstrate statistically significant findings. However, substantial improvements in bleeding were evident as compared with the previous trial, with the 15-μg/kg dose performing better than the 10-μg/kg dose. Hemodynamic data from both studies indicated better hemodynamic stability in the heparinase I groups, but the dosing regimen provided only immediate post-CPB heparin reversal activity, without addressing the possibility of delayed heparin-reversal needs (e.g. , return of pump blood, ICU reheparinization). Based on the findings of these two trials, the current study was designed to include a further significant increase in the dose of available heparinase I (up to 35 μg/kg) and match heparinase I dose timing to include all postbypass heparin reversal needs. However, despite modifications to address these concerns, our study regimen did not achieve efficacy without increased postoperative transfusion and lower hemoglobin concentrations. Given the short half-life of heparinase I, it is possible that our findings could be explained by residual heparin that was not degraded; however, we believe that the increased heparinase I dosing and the protocol requirement for return of pump blood (and contained heparin) within 40 min of initiation of heparin reversal assured adequate intraoperative heparinase I concentrations. In addition, we included an optional ICU dose to address the possibility of postoperative reheparinization. Although no differences in coagulation test results were evident between the two groups (table 6), samples drawn for a substudy specifically to evaluate residual anti–Factor IIa and Xa activity were unfortunately not processed because of excessive cost. However, we believe that the observed excess transfusion and anemia in our study with heparinase I treatment is primarily due to the previously reported property of this agent to incompletely reverse anti–Factor Xa activity during heparin degradation17and not related to residual uncorrected heparin activity.

Examination of the hourly chest tube drainage data (fig. 1) highlights the differing patterns of bleeding between the two agents. Notably, heparinase I was associated with more chest tube effluent during the first 6 postoperative hours. This effect is particularly obvious in the off-pump CABG group, possibly because of the lack of antifibrinolytic therapy in this population, and is consistent with the known pattern of residual anti–Factor Xa anticoagulant activity after heparin degradation with heparinase I.17A key aspect of the design of any noninferiority study is the predefined tolerable threshold for no difference. A panel of independent clinicians was gathered to develop the noninferiority bleeding threshold, including several experienced cardiac anesthesiologists, surgeons, and established clinician–scientist investigators in the field. Among the evidence they considered was data from two large outcomes databases with postoperative bleeding and transfusion records. A consensus was developed that less than or equal to 400 ml excess bleeding over the first 12 h after CABG surgery would be a tolerable difference in chest tube drainage if an alternate heparin-reversal agent had other potential advantages. This consensus was also accepted by regulatory authorities, institutional review boards, and study investigators and was used to perform power calculations to determine the necessary study size. Despite the smaller-than-anticipated number of study subjects enrolled in our study, this noninferiority bleeding threshold was easily achieved in both on- and off-pump groups. However, despite the consensus of numerous individuals regarding the reasonableness of the a priori  determined noninferiority boundary, successful avoidance of the boundary did not assure patient safety, and no benefits were apparent.

Although the noninferiority threshold for heparinase I–related bleeding was achieved, the termination of the study before complete enrollment reduced the total number of study participants and diminished the power of our study to identify more subtle differences. It is possible that a hemodynamic advantage of heparinase I was missed in our study. In addition, data collection for the hemodynamic analysis at trial termination was incomplete, with 54% and 68% of computerized data available in the protamine and heparinase I groups, respectively. However, the incidence of systemic hypotension and/or pulmonary hypertension was almost identical between groups (table 4). Our enrollment criteria selected for a low-risk study population, but these subjects were similar to others in whom differences in hemodynamic stability have previously been observed.17One possible explanation is that slowly developing hemodynamic instability may be masked by corrective action; however, no striking differences in hemodynamic drug interventions were evident. It is also possible that hemodynamic reactions were muted by the slow rate of infusion of protamine; rapid administration predicts greater hemodynamic responses.36Although patterns of protamine administration vary widely among clinicians, we believe it is not typical to complete a primary dose of protamine (dose 1) more rapidly than the 10 min in our study design. For similar reasons, it is also possible that a difference in the composite morbidity index, the percentage of patients with at least one major postoperative complication, was missed in our study. However, these data were captured for almost 100% of study participants as part of required study safety data. Data collection for other adverse event/serious adverse event and morbidity reporting was almost complete, exceeding 95% except for the incidence of dialysis in the protamine group (87%, 76/87).

Almost twice as many heparinase I patients had at least one serious adverse event. Closer examination of serious adverse events reveals a pattern of morbidity; serious adverse events related to bleeding (cardiac tamponade, pericardial effusion, pleural effusion, hemothorax, postprocedure hemorrhage, mediastinal hemorrhage, and hemorrhage) occurred more frequently in the heparinase I group (P = 0.01 by log-linear model; table 7). In addition, serious adverse events specifically or likely related to infectious complications (catheter site discharge, cellulitis, wound infection, staphylococcal infection, rectal abscess, sepsis, and multiorgan failure) were more common in the heparinase I group (P = 0.05 by log-linear model; table 7). Blood transfusions are associated with significant immunomodulatory effects that have been linked to increased postoperative infections,37–41and transfusion and reoperation are strong independent predictors of extended hospital stay.38,42The median hospital stay was 1 day longer in the heparinase I group. It is plausible that increased transfusion, infectious complications, and extended hospital stay in the heparinase I group are all consequences of the increased bleeding in this group. Bleeding and infectious-related serious adverse events represented 44% and 21% of total for heparinase I and protamine groups, respectively, and accounted for much of the excess in serious adverse event reporting in the heparinase I group. Although patterns of increased bleeding, transfusion, and extended hospital stay were evident in both strata, these findings were more statistically significant in the off-pump CABG group, despite the smaller numbers. Notably, off-pump patients did not receive antifibrinolytic therapy and, therefore, may have been predisposed to complications from residual anticoagulation; in this trial, we observed more bleeding and greater transfusion requirements in the off- than the on-pump surgery group.

When protamine is contraindicated (e.g. , catastrophic protamine reactions, protamine allergy), no satisfactory approach is currently available. Ralley and De Varennes43reported the successful use of heparinase I for cardiac surgery in a patient with protamine allergy. Heparinase I has not been evaluated as an alternative when a positive response to a protamine test dose raises concern that continued dosing may precipitate a full-blown reaction; current options are limited to tolerating the potential adverse consequences of residual heparin effect or continued protamine administration.44Other therapeutic options that eliminate protamine include avoidance of heparin–protamine altogether (ancrod,45,46danaparoid sodium,47recombinant hirudin,48,49bivalirudin50–52) or heparin without protamine (reduced heparin dosing with heparin-coated CPB circuits,53extracorporeal heparin adsorption54). No comparisons exist among these approaches, although when compared with standard heparin–protamine therapy, as with the current study, they have been associated with excess bleeding and transfusion.45,47,49Clearly, further trials are required to identify new strategies that avoid or attenuate the adverse events associated with protamine.

In summary, heparinase I for heparin reversal was associated with increased chest tube drainage after elective CABG surgery as compared with protamine, but the difference in bleeding did not exceed the predefined noninferiority threshold for 12-h cumulative mediastinal chest tube drainage. However, heparinase I demonstrated an inferior safety profile compared with protamine. Patients receiving heparinase I sustained more postoperative bleeding complications and had increased transfusion requirements and anemia, possibly because of residual anti–Factor Xa effects. Although all patients demonstrated the same trend, bleeding differences were less pronounced in the on-pump strata, possibly because of the use of antifibrinolytics in this group. The increased risk of infection and delayed hospital discharge observed with heparinase I may be secondary to increased bleeding. Postdrug hemodynamic stability and an index of major postoperative morbidity and mortality were not significantly different between the study groups. Heparinase I is not suitable as a replacement for protamine for nonemergent CABG surgery.

The authors thank Fiona Clements, M.D., Jeff Taekman, M.D., Melanie Wright, Ph.D., and Gene Hobbs, C.R.T. (Duke University Human Simulation and Patient Safety Center, Department of Anesthesiology, Duke University Medical Center, Durham, North Carolina), and Gaetano Paone, M.D. (Assistant Professor, Section of Cardiac Surgery, University of Michigan Medical Center, Crittenton Hospital Medical Center, Rochester, Michigan), for their assistance in developing and teaching the study protocol; and Elizabeth Schramm (Duke Clinical Research Institute, Duke University) and Cheryl Stetson, A.A.S. (Staff Assistant, Department of Anesthesiology, Duke University Medical Center), for assistance in manuscript preparation.

Codirectors:  Mark F. Newman, M.D., Department of Anesthesiology, Duke University Medical Center, Durham, North Carolina; Lee Fleischer, M.D., Department of Anesthesiology, University of Pennsylvania, Philadelphia, Pennsylvania. Study Principal Investigator:  Mark Stafford-Smith, M.D., F.R.C.P.C., Department of Anesthesiology, Duke University Medical Center. Quality Assurance:  Michael Cuffe, M.D., Department of General Medicine, Duke University Medical Center. Study Coordination:  Linda Barber, M.S.N., Duke Clinical Research Institute, Duke University; Nancy Newark, B.S.N., Duke Clinical Research Institute, Duke University; Tammy Reece, M.S., Duke Clinical Research Institute, Duke University. Human Simulation Laboratory:  Jeffrey Taekman, M.D., Department of Anesthesiology, Duke University Medical Center; Melanie Wright, Ph.D., Department of Anesthesiology, Duke University Medical Center; Gene Hobbs, C.R.T., Department of Anesthesiology, Duke University Medical Center; Fiona Clements, M.D., Department of Anesthesiology, Duke University Medical Center. Analysis Group:  James Rochon, Ph.D., Biostatistics and Bioinformatics, Duke University; Jasmine Mathias, M.S., Duke Clinical Research Institute, Duke University; Christina Sigmon, M.A., Duke Clinical Research Institute, Duke University.

The Global Perioperative Research Organization is an academic research organization representing a strategic collaboration between the International Anesthesia Research Society and the Duke Clinical Research Institute. These investigators are listed in alphabetical order with their medical centers.

Edwin Avery, M.D., Instructor in Anesthesia, Department of Anesthesia and Critical Care Medicine, Massachusetts General Hospital, Boston, Massachusetts; Richard Baillot, M.D., Quebec Heart Institute, Laval Hospital, Sainte Foy, Quebec, Canada; Denis Bouchard, M.D., Department of Surgery, Montreal Heart Institute, Montreal, Quebec, Canada; Steven W. Boyce, M.D., Cardiac Surgery, Washington Hospital Center, Washington, D.C.; Luis A. Dibos, M.D., Department of Surgery, Union Memorial Hospital, Baltimore, Maryland; H. Muhammad Ghotbi, M.D., Department of Anesthesiology, Kaiser Foundation Hospital, San Francisco, California; N. Martin Giesecke, M.D., Clinical Assistant Professor, Department of Cardiovascular Anesthesiology, Texas Heart Institute, Houston, Texas; Richard Hall, M.D., Pharmacology/Critical Care/Anesthesiology, Capital Health Center for Clinical Research, Halifax, Nova Scotia, Canada; John Harlan, M.D., Cardio-Thoracic Surgeons, Medical Center East, Birmingham, Alabama; Andreas Hoeft, M.D., Chairman, Department of Anesthesiology and Intensive Care Medicine, Universitatsklinkiken Bonn, Bonn, Germany; Marc Kanchuger, M.D., Associate Professor, Department of Anesthesiology, New York University Medical Center, New York, New York; Charles T. Klodell, M.D., Assistant Professor, Department of Surgery, University of Florida Health Science Center, Gainesville, Florida; Irving L. Kron, M.D., Professor and Chief, Department of Surgery, University of Virginia Medical Center, Charlottesville, Virginia; Edward A. Lefrak, M.D., Chief, Cardiac Surgery, Inova Fairfax Hospital, Falls Church, Virginia; Andrew Maitland, M.D., Cardiac Surgery, Foothills Medical Centre, Calgary, Alberta, Canada; David Marsh, M.D., Associate Director, San Diego Cardiac Center, San Diego, California; Nancy Nussmeier, M.D., Clinical Associate Professor, Cardiovascular Anesthesiology, Texas Heart Institute, Houston, Texas; Anjum G. Qazi, M.D., Cardiac Surgery, Washington Adventist Hospital, Takoma Park, Maryland; Scott T. Reeves, M.D., Professor, Department of Anesthesiology and Perioperative Medicine, Medical University of South Carolina Hospital, Charleston, South Carolina; Mohan R. Sarabu, M.D., Cardiac Surgery, Westchester Medical Center, Valhalla, New York; Linda Shore-Lesserson, M.D., Associate Professor, Department of Anesthesiology, Mount Sinai School of Medicine, New York, New York; Norman Silverman, M.D., Chief, Cardiothoracic Surgery, Henry Ford Hospital, Detroit, Michigan; Victor E. Tedesco, M.D., Touro Infirmary, New Orleans, Louisiana; Serge G. Tyler, M.D., Department of Anesthesia, John Stroger Hospital, Chicago, Illinois; Felipe Urdaneta, M.D., Assistant Professor, Department of Anesthesiology, V.A. Medical Center, Gainesville, Florida; Hugo Van Aken, M.D., Department of Anesthesiology and Intensive Care Medicine, University Hospital of Muenster, Muenster, Germany; Michael T. Weaver, M.D., Pendleton Memorial Methodist Hospital, New Orleans, Louisiana; Thorsten Wahlers, M.D., Klinikdirektor Prof., Klinikum der Friedrich Schiller, Universitat Jena, Jena, Germany; Ian J. Welsby, M.B.B.S., Assistant Professor, Department of Anesthesiology, Duke University Medical Center, Durham, North Carolina; Michael Zenz, M.D., Department of Anesthesiology, Intensive Care Medicine and Pain Therapy, University Hospital Bergmannsheil, Bochum, Germany.