Perioperative Sympatholysis

Abstract

McSPI-Europe Research Group*.*See Appendix 1 for participating European centers and investigators, and Appendix 2 for other analysis investigators.Received from the Departments of Anesthesiology, Cardiology, Surgery, Medicine, and Epidemiology and Biostatistics of the member centers of the McSPI-Europe Research Group, the Ischemia Research and Education Foundation (IREF), the San Francisco Veterans Affairs Medical Center, and the University of California, San Francisco, and Stanford University, Palo Alto, California. Submitted for publication December 8, 1995. Accepted for publication November 1, 1996. Supported by grants from UCB Pharma, Belgium, and the Ischemia Research and Education Foundation.Dr. Mangano, as director of McSPI, assumes full responsibility for the integrity of the data and analyses.Address reprint requests to Dr. Mangano: c/o McSPI-Europe, 250 Executive Park Boulevard, Suite 3400, San Francisco, California 94134.Of the 60 million patients a year who undergo noncardiac surgery in the United States, Canada, Europe, Australia, New Zealand, and Japan, [1,2]approximately 30% have, or are at risk of, coronary artery disease (CAD). Of these, 3 million have a serious adverse perioperative cardiac outcome-myocardial infarction (MI), cardiac death, unstable angina, heart failure, or life-threatening dysrhythmias. [1]These complications add approximately $40 billion annually to worldwide surgical health care costs. [3]Mangano DT, and colleagues [4]demonstrated that the single most important predictor of such adverse cardiac outcome is the occurrence of myocardial ischemia soon after surgery (i.e., within 48 h). Occurring in 41% of patients undergoing elective noncardiac surgery, ischemia is associated with a ninefold increase in the odds of having in-hospital cardiac death, nonfatal MI, or unstable angina. In addition, the risk of long-term (up to 2 yr) adverse cardiac outcomes increases twofold in patients with postoperative ischemia alone, and 14- to 20-fold in patients with postoperative MI or unstable angina. [5]The entire perioperative period is known to be stressful, characterized by complex and rapidly changing physiologic responses that may be poorly tolerated by patients with compromised circulation or poor left ventricular function. During, and particularly after, anesthesia and surgery, hypertension, tachycardia, and increased sympathetic activity are common occurrences. [1,6]These hyperdynamic cardiovascular responses adversely affect the balance between myocardial oxygen supply and demand, predisposing the vulnerable myocardium to ischemia.Compounds with alpha-2 adrenoceptor agonist activity are reported to have sympatholytic, sedative, analgesic, and anxiolytic effects and attenuate the catecholamine response to perioperative stress. Therefore, alpha2-adrenoceptor agonists may prevent perioperative hyperdynamic changes, mitigating imbalances between myocardial oxygen supply and demand, and possibly reduce the incidence of myocardial ischemia. [7–9]Mivazerol (2-hydroxy-3 [1H-imidazole-4-methyl] benzamide hydrochloride) is a new alpha2-adrenoceptor agonist with a high affinity and a marked specificity for alpha2-adrenergic receptors. In animal studies, mivazerol blunts the surge in heart rate during emergence from halothane anesthesia, decreases basal norepinephrine concentrations, and reduces ischemia in animal models of coronary occlusion.* In clinical pharmacology studies, mivazerol decreases basal plasma levels of norepinephrine and, at higher doses, produces mild bradycardia, negative inotropic effects, and an early and transient increase in afterload.** In patients with stable angina undergoing a treadmill exercise tolerance test, mivazerol reduced ischemia and angina. [10]Most recently, in high-risk patients undergoing noncardiac surgery, a 24-patient dose-response trial demonstrated safety and tolerability.***Based on the above, we conducted a multicenter phase II trial to investigate whether mivazerol could reduce the incidence of, and treatment for, tachycardia and hypertension in patients with, or at risk for, CAD who were undergoing vascular surgery and general anesthesia. We also determined the effect of mivazerol on the incidence and severity of perioperative myocardial ischemia, adverse clinical events, and anesthetic and analgesic requirements.After institutional approval and informed consent were obtained, we enrolled 317 patients from 23 medical centers in 7 European countries (Table 8Appendix 1) between March and December 1993; analysis for the current study was completed in March 1996. The study was placebo-controlled, double-blind, and randomized with parallel groups using block randomization within each center to assign placebo or one of two doses of mivazerol to patients. For inclusion in the study, patients were required to be at least 21 yr old, to have CAD and normal renal function, and to be undergoing vascular surgery (excluding aortic surgery) that required general anesthesia for at least 1 h. Coronary artery disease was confirmed by the existence of any of the following conditions:(1) a history of either typical angina pectoris (as defined by the Canadian Heart Classification system [11]) or atypical angina with an ischemic (electrocardiographic, echocardiographic) response to exercise testing, or with scintigraphic evidence of myocardial perfusion defect;(2) a history of MI;(3) new Q wave on electrocardiogram typical of MI, with no history of MI; or (4) angiographic evidence of CAD. Patients were excluded from study if they:(1) had been taking alpha-methyl dopa, alpha2-agonists, or tricyclic antidepressants;(2) were in cardiogenic shock and had clinical signs of congestive heart failure or required chronic inotropic support for ventricular dysfunction;(3) had unstable angina or treated, uncontrolled hypertension;(4) had conduction defects that precluded electrocardiographic analysis of the ST segment;(5) were pregnant or of childbearing potential but not using reliable contraception; or (6) were American Society of Anesthesiologists physical status V.The primary measures used to judge the efficacy of mivazerol and placebo were hemodynamic instability (tachycardia, hypertension) and the need for cardiovascular medications to treat instability in the first 24 h after surgery, and secondarily for the intraoperative and 24- to 72-h postoperative periods. Other secondary measures consisted of the incidence and severity of myocardial ischemia and the need for medications to treat ischemia. Safety was assessed by the occurrence of adverse clinical events and hemodynamic abnormalities (bradycardia, hypotension), and by the use of cardiovascular medications to treat these hemodynamic abnormalities.The investigators gathered demographic and clinical data by taking a complete history and physical examination that included a neurologic examination and gathering information on previous and concurrent medications. Research data were obtained from measurement of hemodynamic variables and cardiac enzymes, and from ambulatory (Holter) and 12-lead electrocardiography. Preoperative cardiac medications were continued until the day of surgery. Premedication consisted of 5 mg diazepam given orally on the day of surgery.Anesthesia was induced by intravenous administration of sodium thiopental (as much as 5 mg/kg) and fentanyl (as much as 2 micro gram/kg). Anesthesia was maintained by continuous intravenous infusion of fentanyl (1 micro gram [center dot] kg-1[center dot] h-1) and isoflurane (end-tidal concentration of up to 2.0%). Vecuronium provided muscle relaxation. Systolic blood pressure (SBP) and heart rate (HR) were kept within 20% of preoperative baseline values by the use of cardiovascular drugs and prespecified changes in the anesthetic. The anesthetic concentration was first altered if necessary to treat hemodynamic abnormalities. After this, tachycardia with concomitant hypertension was to be treated with a beta blocker. Hypertension was to be treated with hydralazine or sodium nitroprusside. Bradycardia was to be treated with atropine or, when necessary, isoproterenol. Hypotension was treated with phenylephrine, methoxamine, or norepinephrine. Intravenous fluids were used, as was dopamine, when necessary. Prophylactic use of nitrates, calcium-channel blockers, or beta-adrenergic blocking drugs to prevent ischemia was specifically prohibited; nitroglycerin was used only to treat documented myocardial ischemia (as demonstrated by ST-segment changes on the clinical monitor).The “low-dose” group was given 2 micro gram/kg mivazerol for 20 min before induction of anesthesia, followed by 0.75 micro gram [center dot] kg-1[center dot] h-1intraoperatively and for as long as 72 h postoperatively. The “high-dose” group was given 4 micro gram/kg and then 1.5 micro gram [center dot] kg-1[center dot] h-1, following the same protocol. These doses were selected to produce plasma levels of mivazerol of approximately 1.0 ng/ml in the low-dose group and 2.0 ng/ml in the high-dose group during surgery and for as long as 72 h after surgery.Heart rate was to be kept between 40 and 100 beats/min, and SBP between 90 and 180 mmHg, during the 96-h postoperative period, using prespecified analgesic, sedative, and cardiac adjuvant therapies. Morphine sulfate was administered intravenously or by patient-controlled analgesia, and midazolam was given for anxiety.Routine clinical monitoring included continuous five-lead electrocardiography, measurement of radial artery pressure and arterial blood oxygen saturation, and assessment of inspired concentrations of isoflurane and oxygen. In parallel with clinical monitors, research monitors included three-channel, seven-lead Holter electrocardiography, continuous measurement of HR and blood pressure, 12-lead electrocardiography, and assays of cardiac enzymes. Clinical decisions were not controlled by study protocol, and clinicians caring for the patients had no knowledge of any research data.Baseline HR and blood pressure were determined by averaging 3 readings obtained 5 min apart before induction of anesthesia. Intraoperatively and for as long as 96 h after surgery, HR, SBP, diastolic blood pressure, and mean arterial pressure were measured every 10 s and stored by portable patient monitor (McSPI/Ischemia Research and Education Foundation-(IREF) Patient Monitor System) until the arterial line was removed, and then an automated blood pressure cuff recorded pressures, which also were stored in the computer. Heart rate continued to be recorded every 10 s throughout all periods. The HR and blood pressure data were compressed into 1-min data samples and were then reviewed to eliminate artifacts (such as those associated with drawing of blood or flushing of the catheter). Median values for these data were calculated, and each value was evaluated for out-of-bounds conditions, which were annotated. The occurrence of episodes of hemodynamic instability was determined, and the frequency distribution and characteristics of these episodes were derived for the intraoperative and postoperative periods. (Table 9)For the intraoperative period, a hemodynamic episode was said to have occurred if:(1) HR increased at least 20% from baseline or was 40 beats/min or less;(2) SBP increased or decreased 20% or more from baseline; and (3) the episode lasted at least 5 min. For the postoperative period, episodes were said to have occurred if:(1) HR decreased to 40 beats/min or less, or increased to 100 beats/min or more;(2) SBP decreased to 90 mmHg or less, or increased to 180 mmHg or more. In addition, treatment for hemodynamic changes (tachycardia, bradycardia, hypertension, hypotension) were recorded and characterized. To evaluate hemodynamic rebound, the incidence of, and treatment for, either tachycardia or hypertension during the 12-h post-infusion period (> 72 h) was compared between the placebo and mivazerol groups.Two independent, blinded clinicians individually scanned the annotated hemodynamic files, located the out-of-bounds conditions, and reviewed the annotated data to determine whether the hemodynamic abnormalities were real or artifact. A computerized report that described all episodes identified by the two clinicians and any discrepancies between the two clinicians was generated after both primary readers completed their analyses. A third independent reviewer reviewed the generated report and validated all previous readings and resolved any discrepancies. To assess concordance of the hemodynamic analysis, a concordance plan was pre-specified, and 20% of the patient population was randomly selected and reanalyzed for hemodynamic events. All analysis individuals (IREF) were blinded to previous results, patient identification, clinical events, and other outcome events. The results of this analysis, which occurred after completion of the main analysis, were compared statistically as described earlier.Patients were monitored with a three-channel AM Holter electrocardiogram recorder (series 8500; Marquette Electronics, Milwaukee, WI), commencing at least 8 h before surgery and continuing up to the 96th postoperative hour. Bipolar leads (CC5, CM5, ML) were used. Each complete electrocardiographic recording was scanned using an electrocardiographic analysis system (Marquette Series Laser Holter SXP); all abnormal QRS complexes (i.e., ventricular ectopic beats, conduction abnormalities) were excluded, and a continuous three-lead ST-segment trend was generated, as described elsewhere. [4]ST deviation values were determined at 60 msec past the J point, in each of the three channels, unless that point fell within the T wave, in which case the measurement point of ST segment depression was shortened to a minimum of J + 40 msec. Electrocardiographic episodes of ischemia were defined as:(1) horizontal or downsloping ST segment depression from baseline of greater or equal to 1 mm lasting at least 1 min and separated from other episodes by greater or equal to 1 min;(2) ST segment evaluation from baseline greater or equal to 2 mm measured at the J point lasting at least 1 min and separated from other episodes by greater or equal to 1 min. The reversal of an ischemic episode was defined by the return of the ST segment to baseline for at least 1 min. Each episode was assessed for duration, magnitude, severity (area under the curve), as well as for “ischemic burden”(minutes of ischemia/hours monitored). Results of Holter scanner interpretation were recorded on data sheets by the first electrocardiographer. A second electrocardiographer reviewed these results. A third-level electrocardiographer also reviewed all the results and resolved all differences.The intraoperative results also were analyzed by dividing the period into the nonemergence period and the emergence from anesthesia period, with emergence being defined as the 30-min period before leaving the operating room. The period definitions were arbitrary, but were specified before unblinding.Twelve-lead electrocardiograms were obtained at the time of patient screening, on arrival at the intensive care unit, daily on postoperative days 1 to 5, on postoperative day 7, and at hospital discharge. Creatine kinase myocardial band (CK-MB) isoenzyme levels were obtained preoperatively; on arrival at the intensive care unit; and at 4, 8, 12, 16, 20, 24, 32, 40, 48, 60, 72, 84, and 96 h after surgery. The occurrence of perioperative MI was assessed for the period from induction of anesthesia to hospital discharge. Myocardial infarction was diagnosed if any of the following occurred:(1) new Q wave on a postoperative 12-lead electrocardiogram, as determined by application of the Minnesota Code criteria [12](1–1 to 1–3) and analysis panel validation; or (2) an elevation in CK-MB levels to greater or equal to 100 ng/ml at any time after surgery, or to greater or equal to 70 ng/ml within 12 h after surgery; or (3) diagnosis of acute MI made during autopsy.Two blinded investigators applied the Minnesota Code criteria, as modified by Chaitman et al., [13]to code the 12-lead electrocardiograms. If disagreements arose between the two investigators regarding coding or diagnosis, a third investigator reviewed the set of electrocardiograms, and, along with the original two investigators, a final diagnosis was achieved by consensus.Determinations of CK-MB levels were performed centrally by the Bioanalytical Research Corporation (BARC, Gent, Belgium) using an immunoenzymetric assay (Hybritech Tandem-E CK-MB). Diagnosis of MI by autopsy was made by the pathologist at the participating center.Blood samples for clinical chemistry and for determination of plasma mivazerol concentrations were performed at predetermined times.Adverse events and serious adverse events were ascertained by site investigators and reported on Case Report Forms using the system organ classification of the World Health Organization. [14]All research data (Holter electrocardiogram, hemodynamic data, 12-lead electrocardiographic data) were analyzed at the coordinating center (IREF, San Francisco, CA) in a blinded fashion to ensure that uniform criteria for data analysis were used. Because block randomization was performed at each of the 23 centers, the analyses presented herein include an adjustment for any effect associated with individual centers.For the analysis of the incidence of hemodynamic or ischemic episodes, when the outcome variable was binary and the explanatory variable was treatment, the two-by-three contingency Table analysiswas controlled for center effect using the Cochran-Mantel-Haenszel general association chi-square statistic. For the high-dose versus placebo comparison, the same technique was performed using data for the high-dose and placebo groups. These analyses were performed using PROC FREQ of the Statistical Analysis System (SAS, SAS Institute, Cary, NC). For a continuous response variable, a general linear model was used and included center and treatment-by-center effects, to derive the adjustment treatment effect.The treatment-by-center effect was included in all models, regardless of whether the effect was statistically significant (at the 5% level), because the sample size per center was not sufficiently large to assess adequately whether there was treatment-by-center effect, and because randomization was carried out at each center. In addition, in most of the models, the center effect was significant, suggesting a high level of heterogeneity in the least-squares estimated means of the response variable across the 23 centers. PROC GLM (General Linear Model) was used to fit these models and to obtain the adjusted estimated treatment effect. The comparison between high-dose mivazerol and placebo was carried out using data from the high-dose and placebo groups, and the same technique as described was used.The secondary efficacy variables included myocardial ischemia, anesthetic and analgesic requirements, and adverse clinical outcomes. For these endpoints, incidence was compared using either chi-square or Fisher's exact test. Analysis of variance or the Kruskal-Wallis test was used to analyze characteristics of hemodynamic and ischemic abnormalities. For analysis of area under the CK-MB curve and the maximum CK-MB, the data window was taken to be 4–96 h after surgery, a period that encompassed 14 measurements of CK-MB. Values for area under the CK-MB curves and maximum CK-MB were compared across treatment groups using analysis of variance techniques. Missing data were considered unevaluable and therefore excluded from analysis. However, because the values were missing, at random, across study groups, the statistical inferences were still able to be generalized for the whole sample size.The statistical methods for the concordance analysis of the primary outcome variable, hemodynamic stability, includes the incidence, frequency, and severity of hemodynamic abnormality (tachycardia, bradycardia, hypertension, or hypotension) during each of the study periods. Twenty percent (60 patients) were selected randomly from the 300-patient population, and hemodynamics were reanalyzed using the same methods as for the original analysis, with all patient identifiers blinded to the investigators. Concordance between the original analysis results and the reanalysis results was first determined for the incidence of each of four types of hemodynamic abnormality during the primary outcome period, as well as all other time periods. The concordance was evaluated using the Kappa statistic, [15]with an acceptance threshold prespecified to equal 0.60, which is considered to indicate substantial agreement. [16]The second part of the analysis assesses the concordance of the frequency and severity of hemodynamic episodes during each of the study periods. Frequency was defined as the number of hemodynamic episodes (tachycardia, bradycardia, hypertension, or hypotension) per patient for each period. Hemodynamic severity parameters were defined as:(1) the mean of all episodes' durations for each patient; and (2) the mean of all episodes' area under the curve for each patient. The concordance for frequency and severity was evaluated by the method described by Dunn, [17]which uses the traditional method in which the coefficient of correlation serves as the reliability coefficient; for the purposes of this analysis, a threshold greater or equal to 0.6 was prespecified before analysis, and is considered good agreement.All patient case report forms were entered (double-data entry) into a PC/MS-DOS SAS database by the sponsor. The data were backed up daily onto a Vax computer. A copy of the database that consisted of all the information from the case report forms was sent every month to the coordinating and analysis center in San Francisco in a PC/SAS 6.04 format. The technology data that included the Holter and 12-lead electrocardiographic data were double-data entered into a PC/MS-DOS database at the analysis center in San Francisco (IREF)-where error-checking and data validation were done individually by research investigators using SAS error-checking programs. The continuous hemodynamic data were sent by the centers to IREF and were converted to SAS formats. The laboratory data were sent from BARC (Gent, Belgium) to IREF on a DOS diskette, in a flat ASCII file, and then were converted to PC/SAS 6.04 format. All data management tasks were carried out using the PC/SAS 6.04 system. The final database was then converted to UNIX SAS 6.09, and all data analyses were performed using this system.Of the 317 patients enrolled in the study, 17 were not assigned to a study group, because of ineligibility (7), patient refusal (3), or technical reasons (7). Of the remaining 300 patients, 103 were given placebo; 99, low-dose mivazerol; and 98, high-dose mivazerol. Table 1shows that demographic data and preoperative cardiac medications were similar for the three groups. Generally, patients were older and hypertensive and were taking a number of cardiovascular medications. Exposure to the study drug did not differ (P = 0.082) for the three groups (placebo, 75.3 h; low-dose, 75.3 h; high-dose, 75.7 h, all medians).(Figure 1and Figure 2) show the median HR and SBP for various time periods. For all time periods, HR was significantly lower with high-dose mivazerol than with placebo. Systolic blood pressure was significantly lower in the high-dose group at 8, 16, and 24 h after leaving the operating room.(Table 2) shows the incidence of hemodynamic changes, and Table 3shows the therapeutic interventions to treat those changes. Before administration of mivazerol, the incidence of tachycardia tended to be higher in the high-dose group. However, during drug administration (i.e., during the intraoperative, early postoperative, and late postoperative periods), the incidence of tachycardia was significantly lower in the high-dose group (Table 2). For the low-dose group, the incidence of tachycardia during the intraoperative period was 38% versus 51% for placebo (P = 0.064), and was significantly reduced during the late postoperative period (54% vs. 70%, P = 0.031). The incidence of hypertension was significantly decreased with low-dose or high-dose mivazerol for the intraoperative period only (Table 2). For all time periods, the incidence of either tachycardia or hypertension was significantly less with mivazerol (at either dose level) than with placebo. Specifically, the incidences for the three groups during the various time periods were as follows: for the intraoperative period, 65%(high dose) and 64%(low dose), versus 80%(placebo)(P = 0.015); for the early postoperative period, 55% and 70%, versus 69%(P = 0.029); and for the late postoperative period, 63% and 66%, versus 82%(P = 0.011). An effect also was observed after discontinuation of the infusion of mivazerol: 85% and 85%, versus 96%(P = 0.015).The need for treatment of tachycardia was significantly lower with high-dose mivazerol for the early and late postoperative periods (Table 3). The need for treatment of hypertension was similar in the high-dose, placebo, and low-dose groups (Table 3). The low-dose group did not differ from the placebo group regarding treatment for tachycardia or hypertension, except for the intraoperative period (34% vs. 46%, P = 0.028). After discontinuation of mivazerol, treatment for tachycardia or hypertension did not differ between groups.Episodes of bradycardia occurred in both mivazerol groups during all time periods of drug administration and even after discontinuation of the drug; the incidence was significantly higher than that with placebo (Table 2). However, treatment for bradycardia did not differ significantly across groups during any period. The groups did not differ in the incidence of hypotension, except during the intra-operative period, at which time the low-dose group had a higher incidence than the placebo group (87% vs. 77%, P = 0.028). There was no difference in the incidence of either bradycardia or hypotension for any of the postoperative periods. However, for the intraoperative period, the low-dose group had a higher incidence of either bradycardia or hypotension (high-dose, 78%; low-dose, 88%; placebo, 77%; P = 0.048). Treatment for hypotension did not differ significantly between the groups during or after drug administration (Table 3).(Table 4) shows the incidence of changes in the ST segment on Holter electrocardiogram. This incidence was significantly less (50% less) for the high-dose group than for the placebo group during the intraoperative period. Post hoc analysis, derived by subdividing the intraoperative period into the emergence-from-anesthesia period (i.e., the 30 min before transport of the patient from the operating room) and the nonemergence period, revealed that the incidence was 67% less in the emergence period. For other periods, these differences were not significant (Table 4).(Table 5) shows the characteristics of ST segment changes. The high-dose group had a significantly shorter duration of ischemia and area under the curve, and less ischemic burden for the late postoperative period.Regarding treatment of ischemia, the need for nitrates, calcium-channel blockers, or beta-blockers was significantly lower in the high-dose group for the late postoperative period, being 46% less than with placebo (high-dose, 21%; low-dose, 35%; placebo, 39%; P = 0.027). Regarding individual cardiac medications, the use of beta-blockers was significantly lower in the high-dose group in the late postoperative period (9% vs. 22% for placebo; P = 0.049).One hundred seventy-three patients (high dose, 55; low dose, 55; placebo, 63; P = 0.675) had an adverse event, as defined by the World Health Organization System Organ Classification. The incidence of serious adverse events also was not significantly different across study groups: 12%, 9%, and 13% for the high-dose, low-dose, and placebo groups, respectively. Five patients died during the study-four in the high-dose group (two from stroke after carotid surgery, one from pulmonary embolism, and one from MI and cardiogenic shock) and one in the placebo group (MI complicated by stroke and cardiogenic shock). None of these deaths was reported by the clinician (who had no knowledge of study group assignment) as probably or definitely being associated with drug therapy. Using a post hoc analysis, there was no difference in the incidence of hypertension between study groups during the acute drug infusion period before anesthetic induction (high dose, 28%; low dose, 18%; placebo, 21%; P = 0.244), nor was there any difference in the incidence of hypertension before drug infusion (high dose, 31%; low dose, 22%; placebo, 24%; P = 0.376).Myocardial infarction occurred in 6 of 103 patients given placebo, in 1 of 99 patients given low-dose mivazerol, and in 2 of 98 patients given high-dose mivazerol. The relatively small sample size limits the ability to interpret these results. The number of patients in the high-dose, low-dose, and placebo groups who had other adverse cardiac outcomes were, respectively, as follows: 0, 0, and 1 patient had unstable angina; 3, 1, and 3 patients had congestive heart failure; and 0, 0, and 1 patient had dysrhythmia. Five, six, and four patients had cerebrovascular accidents; and 1, 0, and 1 patient had cardiac death.Anesthetic requirement was similar for all groups. For the high-dose, low-dose, and placebo groups, the mean total dose of fentanyl (micro gram) required to induce anesthesia was, respectively, 129 +/- 29, 136 +/- 35, and 136 +/- 32. The mean induction dose of thiopental (mg) was 326 +/- 71, 313 +/- 93, and 330 +/- 83; and the mean end-tidal concentration of isoflurane required (vol%) was 0.46 +/- 0.19, 0.46 +/- 0.16, and 0.48 +/- 0.15. All patients required morphine sulfate equally during the first 48 h after surgery. Midazolam requireme