Peripheral Vascular Disease The effects of perioperative hh-blockade: Results of the Metoprolol after Vascular Surgery (MaVS) study, a randomized controlled trial Homer Yang, MD,a Karen Raymer, MD,b Ron Butler, MD,c Joel Parlow, MD,d and Robin Roberts, M Teche Ottawa, Hamilton, London, and Kingston, Ontario, Canada

Background Patients undergoing vascular surgery comprise the highest risk group for perioperative cardiac mortality and morbidity after noncardiac procedures. Many current guidelines recommend the use of h-blockers in all patients undergoing vascular surgery. We report a trial of the perioperative administration of metoprolol and its effects on the incidence of cardiac complications at 30 days and 6 months after vascular surgery. Methods Patients undergoing abdominal aortic surgery and infrainguinal or axillofemoral revascularizations were recruited to a double-blind randomized controlled trial of perioperative metoprolol versus placebo. Patients were randomized to receive study medication, starting 2 hours preoperatively until hospital discharge or maximum of 5 days postoperatively. Primary outcome were postoperative 30-day composite incidence of nonfatal myocardial infarction, unstable angina, new congestive heart failure, new atrial or ventricular dysrhythmia requiring treatment, or cardiac death. Results Patients were randomized to receive either metoprolol (n = 246) or placebo (n = 250). Primary outcome events at 30 days postoperative occurred in 25 (10.2%) versus 30 (12.0%) ( P = .57) in metoprolol and placebo groups, respectively (relative risk reduction 15.3%, 95% CI 38.3% to 48.2%). Observed effects at 6 months were not significantly different ( P = .81) (relative risk reduction 6.2%, 95% CI% 58.4% to 43.8%). Intraoperative bradycardia requiring treatment was more frequent in the metoprolol group (53/246 vs 19/250, P = .00001), as was intraoperative hypotension requiring treatment (114/246 vs 84/250, P = .0045). Conclusion Our results showed metoprolol was not effective in reducing the 30-day and 6-month postoperative cardiac event rates. Prophylactic use of perioperative h-blockers in all vascular patients is not indicated. (Am Heart J 2006;152:983290.)

Patients undergoing vascular surgery are thought to countershock, new or worsened CHF, and coronary comprise the highest risk group for perioperative cardiac insufficiency.2 Perioperative cardiac complication rates mortality and morbidity after noncardiac procedures.1,2 for infrainguinal revascularization and aortic surgery are Event rates have been reported as 13.2% for inhospital comparable.3-5 Although long-term event rates for post- all-cause mortality, myocardial infarction (MI), angina, operative all-cause mortality as well as for cardiac death, and congestive heart failure (CHF)3 to 21.3% for 6-day MI, unstable angina, and CHF were reported to have been postoperative cardiac death, MI, pulmonary edema, reduced by atenolol in one study,6 the study excluded ventricular tachycardia or fibrillation requiring inhospital deaths. When all events post randomization in this trial were included, the findings would become 7 From the aDepartment of Anesthesiology, University of Ottawa, Ottawa, Canada, insignificant. Another trial showed a reduction in 30-day bDepartment of Anesthesia, McMaster University, Hamilton, Ontario, Canada, cDepart- perioperative cardiac death and nonfatal MI rates in a ment of Anesthesia and Peri-operative Medicine, University of Western Ontario, London, d selected high-risk population, but it was not blinded, and Ontario, Canada, Department of Anesthesiology, Queen’s University, Kingston, 8 Ontario, Canada, and eDepartment of Clinical Epidemiology and Biostatistics, McMaster the study had 20 events in total. Thus, available data are University, Hamilton, Ontario, Canada. sparse and unreliable. Despite this, recent guidelines The study was funded by the Heart and Stroke Foundation of Canada (grant no. have recommended use of h-blockers routinely in NA3779). patients undergoing noncardiac surgery.9 Side effects Submitted April 6, 2006; accepted July 3, 2006. Reprint requests: Homer Yang, MD, Department of Anesthesiology, University of Ottawa, from h-blockers may occur, including prolonged hypo- B309, 1053, Carling Ave, Ottawa, Canada K1Y 4E9. tension or bradycardia, which require interventions. E-mail: [email protected] Therefore, a definite assessment of their benefits and 0002-8703/$ - see front matter n 2006, Published by Mosby, Inc. risks is needed. We report a trial to test the hypothesis doi:10.1016/j.ahj.2006.07.024 that, at 30 days and 6 months after vascular surgery, the American Heart Journal 984 Yang et al November 2006

perioperative administration of metoprolol reduces the including phenylephrine, ephedrine, nitroglycerine, and low- incidence of cardiac complications defined as cardiac dose dopamine were allowed. Surgical and anesthesia care as death, nonfatal MI, CHF, unstable angina, and dysrhyth- well as postoperative clinical decisions were made by attending mias requiring treatment. physicians and not dictated by protocol. Open-label h-blocker use was strongly discouraged except when deemed absolutely necessary by the attending physician. Methods Circumstances for open-label use were generally for rapid heart Study population rate control. Intraoperatively, esmolol, if deemed absolutely The study was conducted in 3 tertiary care centers in Canada: necessary, was allowed. In the postoperative period, if open- General Campus, Hamilton Health Sciences; Victoria Campus, label h-blocker was used, study medication was discontinued. London Health Sciences; and Kingston General Hospital be- Blinding was maintained. tween 1999 and 2002. After approval from the Research Ethics Boards, all patients undergoing vascular surgery were screened Follow-up and outcome measures for eligibility. Elective vascular surgical patients are evaluated by Patients were followed up daily by research nurses until internists, cardiologists, or anesthesiologists in preoperative hospital discharge. Troponin I or T levels and 12-lead clinics. Screening was also undertaken on the wards when electrocardiography were performed routinely on postopera- applicable. Inclusion criteria were American Society of Anes- tive days 1, 2, and 3. Additional measurements were performed thesiology class 3 or less and undergoing abdominal aortic at the discretion of the attending physician. Thirty-day and surgery and infrainguinal or axillofemoral revascularization. 6-month follow-ups were done by telephone for discharged Exclusion criteria were current or recent h-blocker use, current patients. All data with supporting documentation were col- amiodarone use, airflow obstruction requiring treatment, histo- lected by the participating centers and evaluated by the ry of CHF, history of atrioventricular block, previous adverse adjudication committee in a blinded fashion. drug reactions to h-blockers, and previous participation in the All outcome measures were defined before starting the study. MaVS study. The primary outcome was defined as a composite of cardiac complications at 30 days postoperative: cardiac death, nonfatal Study protocol MI, CHF, unstable angina, and dysrhythmia requiring treatment After informed consent by research nurses, study patients defined as atrial fibrillation or ventricular dysrhythmias. In the were randomized to receive either metoprolol or placebo. presence of N1 outcome, the first outcome was noted. An Randomization was constructed in blocks of 4 by the study independent adjudication committee reviewed all primary statistician (Roberts), and study medication preparations, by the composite outcomes. Cardiac death was defined as either the pharmacists. The patients, investigators, and all caretakers were ultimate cause of death traceable to an initiating cardiac blinded to the study randomization. Patients weighing z75 kg complication or death in which the cause was not clearly received metoprolol 100 mg or matching placebo; between 40 identifiable or was insufficient to account for the demise in a and 75 kg, metoprolol 50 mg or placebo; and V40 kg, patient who was not expected to succumb at the time of death. metoprolol 25 mg or placebo. Following the protocol of Nonfatal MI within 3 postoperative days was diagnosed if at 6 Mangano et al for timing, h-blocker therapy was commenced least one of the following was present: chemical evidence of MI preoperatively on the day of surgery and continued for the or new Q waves N0.04 s on 2 contiguous leads. Beyond 3 days, duration of the hospital stay. Oral study drug was administered nonfatal MI was determined by attending physicians with within 2 hours preoperatively. After surgery, oral or intravenous supporting documentation of hospital chart, troponins, and pre- (IV) study drug was administered within 2 hours. The IV study and postoperative electrocardiograms. The diagnosis of unsta- drug was either metoprolol 1 mg/mL or saline at 0.2 mL/kg (to a ble angina was made by the attending physicians when anginal maximum of 15 mL), diluted with 20 mL of saline, and given for symptoms necessitated a change in medications, coronary 15 minutes. Study medication was continued IV every 6 hours revascularization, or intensive care admission. Congestive heart or orally twice a day for 5 days or until hospital discharge, failure was diagnosed clinically with the requisite radiographic whichever occurred sooner. Intravenous study drug was evidence. Dysrhythmia requiring treatment was defined as one converted to oral as soon as oral intake was tolerated. of the following: ventricular fibrillation requiring counter Blood pressure and heart rate were monitored before each shock, ventricular tachycardia requiring counter shock or dose and 30 minutes after starting intravenous drug adminis- medication, or atrial fibrillation N15 minutes in duration tration. If the systolic blood pressure (BP) was b100 mm Hg or requiring counter shock or medication. Troponin-I was assayed if the heart rate was b50 beat/min while awake or b45 beat/ using Sanofi Access (Sanofi Diagnostics Pasteur, Madrid, Spain), min while asleep, the study drug was withheld. In perceived Abbott Axsym (Abbott Diagnostics, Chicago, IL), Dade Stratus II h-blocker toxicity, subsequent study drug was halved in dose (Dade Behringer, Buckinghamshire, United Kingdom), or intravenously or in frequency orally. In the event of a primary Beckman Coulter Access II (Fullerton, CA) as per each center’s study outcome, exacerbation of airflow obstruction, CHF, or laboratory preference. Troponin-T was assayed using the Roche advanced heart blocks, the study medication was discontinued. (Roche, Mannheim, Germany) kit. A significant rise was defined Blinding of randomization was maintained throughout clinical as a measurement exceeding the 99th percentile of a reference decisions on reducing or discontinuing the study medication. control group, with acceptable imprecision (coefficient of variation) at 99th percentile for each assay =10%.10 Clinical care Intraoperative hypotension or bradycardia was defined On the day of surgery, premedications were as per attending as systolic BP b90 mm Hg or 50 beat/min, respectively. anesthesiologists’ orders. Short-acting vasoactive medications Intraoperative hypotension or bradycardia requiring American Heart Journal Yang et al 985 Volume 152, Number 5

Figure 1

Patient screening.

treatment was as per the clinical judgment of the attending 50% relative risk reduction [RRR]) were likely.3,11 Our study anesthesiologist. population was considered highest risk as a surgical popula- Secondary outcomes included incidence of study drug tion, and as such, our sample size had 80% power to detect a discontinuation due to bronchospasm, advanced heart blocks, 45% RRR (2-tailed a = .05), assuming a control group event hypotension, or bradycardia; incidence of reoperation or rate of 20% at 30 days. The analysis was based on an intent-to- amputation; and the incidence of intraoperative hypotension treat principle. Baseline comparisons for categorical data were and bradycardia requiring treatment by the attending anes- achieved via a m2 test, and for continuous data, by either the thesiologists. Student t test or the Wilcoxon test in the presence of severe skewness. The primary analysis of efficacy compared the Statistical analysis between-group difference in outcome at 30 days via a m2 test The sample size of 500 was calculated based on the event (independent binomial proportions). The treatment effect was rates in the literature2 and the claim that large benefits (eg, represented as RRR with an associated 95% CI. The 6-month American Heart Journal 986 Yang et al November 2006

Table I. Preoperative data Table II. Distribution of risk scores (RCRI) by treatment

Placebo Metoprolol Risk score Placebo (n = 250) Metoprolol (n = 246) (n = 250) (n = 246) 1 159 (63.6%) 138 (56.1%) Age 65.9 F 10.04 66.4 F 10.0 2 63 (25.2%) 89 (36.2%) Sex (male/female) 184/66 193/53 3 25 (10.0%) 15 (6.1%) Height (cm) 170.3 F 9.0 171.3 F 8.7 z4 3 (1.2%) 4 (1.6%) Weight (kg) 78.2 F 15.9 78.3 F 14.7 BP v2, P = .036; Wilcoxon, P = .22. Systolic 138.7 F 18.2 139.8 F 17.0 Diastolic 77.3 F 9.5 77.3 F 9.6 F F Heart rate (beat/min) 75.8 10.3 75.8 10.8 Table III. Intraoperative data ASA classification I 0 (0.0%) 1 (0.4%) Placebo Metoprolol II 91 (36.4%) 92 (37.4%) (n = 250) (n = 246) III 159 (63.6%) 152 (61.8%) Unknown 0 (0.0%) 1 (0.4%) Types of anesthesia Qualifying surgery General 103 (41.2%) 113 (46.3%) Infrainguinal 106 (42.4%) 101 (41.4%) Regional 37 (14.8%) 31 (12.7%) Extraanatomical 38 (15.2%) 44 (17.8%) Combined 110 (44.0%) 100 (41.0%) Abdominal aortic 116 (46.4%) 111 (45.1%) Sedation 0 (0%) 0 (0%) Risk factors X-clamp duration (min) 55.0 F 19.1 (n = 135) 54.6 F 20.6 Prior MI 30 (12.0%) 37 (15.0%) (n = 135) Angina 25 (10.0%) 18 (7.3%) Surgery Time (h) 2.7 F 0.9 2.7 F 0.9 DM on treatment 37 (14.8%) 54 (22.0%) Estimated blood loss 350 (100-700) 400 (200-800) Permanent pacemaker 1 (0.4%) 0 (0.0%) (mL) (median IQR)4 12-lead ECG Postoperative BP (mm Hg) Rhythm other than 2 (0.8%) 7 (2.9%) Systolic 138.8 F 25.2 131.7 F 24.8 sinus or sinus + APB Diastolic 71.2 F 15.1 69.5 F 13.1 PVC N5/min 0 (0.0%) 1 (0.4%) Postoperative heart 79.1 F 15.7 69.4 F 12.9y Q waves 3 (1.2%) 1 (0.4%) rate (beat/min) ST abnormalities 40 (16.1%) 36% (14.6%) LVH 1 (0.4%) 2 (0.8%) 4Data on anesthesia type missing for 2 patients. LBBB 1 (0.4%) 6 (2.4%) yP b .0001. Others CVA, RIND, or TIA 30 (12.0%) 26 (10.6%) Renal insufficiency or 7 (2.8%) 3 (1.2%) N participation in MaVS, 51 (2.5%); previous adverse drug creatinine 1.8 mg/dLy h History of cigarette smoking 225 (90.0%) 216 (87.8%) reaction to -blockers, 24 (1.2%); current amiodarone Concomitant medications use, 11 (0.5%); and others, 33 (1.6%); for a total of 1262. Nitrates (long-acting and daily) 9 (3.6%) 9 (3.7%) Of 762 eligible patients, 262 refused consent and 4 were Calcium-channel blockers 42 (16.8%) 46 (18.7%) not randomized for other reasons. As a result, 496 were ACE inhibitors 71 (28.4%) 71 (28.9%) included and randomized to receive placebo (n = 250) ASA, American Society of Anesthesiologists; DM, diabetes mellitus; ECG, and metoprolol (n = 246), respectively. electrocardiogram; APB, atrial premature beat; PVC, premature ventricular Preoperatively, groups were well balanced (Table I). contractions; LVH, left ventricular hypertrophy; LBBB, left bundle-branch block; CVA, cerebrovascular accident; RIND, reversible ischemic neurologic deficits; TIA, Risk groups by Revised Cardiac Risk Index (RCRI) are transient ischemic attack; ACE, angiotensin-converting enzyme. presented in Table II. Type of anesthesia, aortic cross- 4Mean F SD. yCreat N160 Amol/L. clamp time, surgery time, and estimated intraoperative blood loss are not different between groups (Table III). All study patients received the allocated study drug outcomes for the 2 treatment groups have been displayed as preoperatively (Table IV). Postoperative heart rate at survival curves estimated via the Kaplan-Meier technique. F F 12 post-anesthesia care unit was 79.1 15.7 and 69.4 Survival curves were compared with a log-rank test and the 12.9 beat/min in the placebo and metoprolol groups , treatment effect (and CI) computed via a Cox model. All P respectively, P b .0001. Completion of study protocol values are 2-sided with P b .05 being considered significant. was 77.6% and 75.2% in the placebo and treatment groups, respectively. Discontinuation of the study Results protocol in the placebo and treatment groups was due We screened 2847 patients (Figure 1), of whom 2024 to z1 of the following: primary outcome event (30 and met the inclusion criteria. Patients were excluded for z1 25, respectively); patient/family/physician preference of the following: current h-blocker use, 638 (31.5%); (27 and 14, respectively); and open-label h-blockers (24 airflow obstruction, 283 (14.0%); history of CHF, 149 and 14, respectively). In placebo and treatment groups, (7.4%); atrioventricular heart block, 99 (5.0%); previous 239 (95.6%) and 231 (93.9%) received z1 doses of American Heart Journal Yang et al 987 Volume 152, Number 5

Table IV. Study drug protocol compliance Figure 2

Placebo Metoprolol (n = 250) (n = 246)

Completed study protocol 194 (77.6%) 185 (75.2%) Discontinued before 56 (22.4%) 61 (24.8%) completion4 Atrioventricular block 2 (0.8%) 3 (1.2%) Bronchospasm 1 (0.4%) 4 (1.6%) Primary outcome event 30 (12.0%) 25 (10.2%) Open-label h-blockers 24 (9.6%) 14 (5.7%) Patient/family/ 27 (10.8%) 14 (5.7%) physician preference Patient death 3 (1.2%) 0 (0.0%) Other 11 (4.4%) 13 (5.3%) Preoperative dose taken 250 (100%) 246 (100%) Any postoperative IV drug 222 (88.8%) 191 (77.6%) Duration IV drug (d) 0.92 (0.8-1.05)y 0.92 (0.62-1.0) Any postoperative 216 (86.4%) 213 (86.6%) oral drug Cumulative risk primary outcomes at 6 months. Duration oral drug (d) 4.0 (1.5-5.0)y 3.5 (1.4-4.8) Any postoperative drug 239 (95.6%) 231 (93.9%)

4More than 1 reason can be specified. yMedian and IQR. metoprolol in the vascular population is smaller than previously reported and is not significant. Our cardiac death and nonfatal MI rate in the placebo group was 14-16 study drug postoperatively. Median length of stay after 8.8% and was comparable with other studies. The the first study drug administration was 6 (5-8) days primary composite event rate of 12% in our control (interquartile range [IQR]) for both groups. Observed group, although lower than expected, was comparable 2,3 effects at 6 months are shown in Figure 2 and are not with previous studies. In the study of Poldermans 8 significantly different. et al, a 91% relative risk reduction for cardiac death and Observed effects in primary composite outcomes at nonfatal MI was reported. However, that study was 30 days are shown in Table V: 30 patients (12.0%) in the unblinded and open to potential biases. The event rate placebo group and 25 (10.2%) in the metoprolol group in the standard treatment group was 34%, significantly 17 2 experienced z1 primary outcomes. The observed higher than the reported rates of 18% and 21% in relative risk reduction (RRR) was 15.3% ( P = .57), with most published studies. Our study was double-blinded 95% CI of 38.3% to 48.2%. Table VI shows event rates and larger in sample size, overcoming many of the and treatment effects by RCRI. limitations of the previous study. 8 Intraoperatively, incidences of hypotension (40.8% In the previous study, the patients commenced placebo vs 53.7% metoprolol), hypotension requiring bisoprolol therapy on average of 37 days preoperatively treatment (33.6% placebo vs 46.3% metoprolol), brady- and continued it for 30 days postoperatively. Our dosage cardia (10.4% placebo vs 34.6% metoprolol), and regimen started 2 hours before surgery and continued bradycardia requiring treatment (7.6% placebo vs 21.5% for 5 days after surgery and was comparable with the 6 metoprolol) were significantly different (Table VII). All methodology of Mangano et al. In most Canadian h b episodes requiring treatment were deemed manageable, centers, instituting -blockade at least 1 week preop- Q8 although vasopressors such as phenylephrine or eratively is not practical, and the results from a ephedrine were required. Atrioventricular block oc- protocol following this regimen would not be general- curred in 2 (0.8%) and 3 (1.2%) of the placebo and izable to our clinical setting. Differences in results 8 metoprolol groups, respectively (Table IV). One atrio- between the study of Poldermans et al and MaVS could ventricular block occurred during the intravenous be ascribed to this choice; however, the presence of h administration period, whereas the rest occurred during -blockade in the active treatment group in our study oral administrations. was confirmed by the differences in postoperative heart rates and in the intraoperative bradycardia requiring treatment between the study groups. Discussion In our study, open-label h-blocker use was allowed As a population, patients undergoing vascular surgery when absolutely necessary because denying h-blocker are considered to have one of the highest risks.13,14 Our use when clinically indicated would have been unethi- results show that the RRR achieved with perioperative cal. Furthermore, the use of a h-blocker when deemed American Heart Journal 988 Yang et al November 2006

Table V. Outcomes at 30 days postoperative Table VII. Intraoperative side effects

Placebo Metoprolol Placebo Metoprolol (n = 250) (n = 246) P (n = 250) (n = 246) P

Primary outcome4y 30 (12.0%) 25 (10.2%) .57 Intraoperative hypotension Cardiac death 1 (0.4%) 0 (0.0%) 1.00 Treatment 102 (40.8%) 131 (53.7%) .0069 MI 21 (8.4%) 19 (7.7%) .87 required 84 (33.6%) 114 (46.3%) .0045 CHF 3 (1.2%) 5 (2.0%) .50 Intraoperative bradycardia Unstable angina 0 (0.0%) 1 (0.4%) 1.00 Treatment 26 (10.4%) 84 (34.6%) b.000005 Dysrhythmia 10 (4.1%) 7 (2.9%) .62 required 19 (7.6%) 53 (21.5%) .00001 Noncardiac death 3 (1.2%) 0 (0.0%) Secondary outcomes Reoperation 78 Postoperative CVA 45 New or worsened 57 Table VIII. Comparison of treatment effects renal insufficiencyz Any rehospitalization 28 24 Study Event rates RRR (95% CI)

4 N Primary outcomes from all sources; may have 1 outcome. Poldermans 18/53 (34.0%) 2/59 (3.4%) 90.6% yEstimated treatment effect primary outcome odds ratio 0.83, 95% CI 0.45-1.51. 8 N A z et al (28 d) (59.0%-97.6%) zPostop creatinine 1.8 mg/dL (160 mol/L) and 20% increase from preoper- 4 ative value. Mangano 12/101 (11.9%) 0/100 (0.0%) 95.9% et al6 (6 m) (35.7%-99.9%) POBBLE (30 d) 15/44 (34.1%) 17/53 (32.1%) 5.9% (66.0% to 46.7%) MaVS 30/250 (12.0%) 25/246 (10.2%) 15.3% (38.3% to 48.2%) Table VI. Event rates and treatment effect by RCRI Pooledy 27.0% (4.7% to 49.0%) Event rates Treatment effect 40.5 added to all frequencies. Risk Odds ratio yP (pooled RRR) = .087; P (homogeneity) = .0048. score Placebo Metoprolol (95% CI) P

1 19/159 (11.9%) 10/138 (7.2%) 0.576 (0.258-1.284) .24 2 7/63 (11.1%) 8/89 (9.0%) 0.790 (0.271-2.304) .78 serious morbidity, this finding, in conjunction with a 3 3/25 (12.0%) 6/15 (40.0%) 4.889 (0.999-23.930) .057 lack of clear clinical benefit, highlights the need to z 4 1/3 (33.3%) 1/4 (25.0%) 0.667 (0.025-18.060) 1.00 balance the potential cardioprotective effect of pro- h h Summary, OR = 0.864 (95% CI 0.491-1.518), P = .61; homogeneity, P = .14. phylactic -blockade against the incidence of -blocker– related side effects. Our study did not show a treatment effect at 6 months. This is in contrast to another previous report,6 but that necessary by the attending physician reflects current study did not include all events after randomization. One standard of practice; hence, our research question is possible explanation for the difference in results is that, more akin to a comparison between a prophylactic in the previous study, 8% of the patients assigned to the h-blocker group and a bstandard treatmentQ group. control group had been on h-blockers preoperatively Where an open-label h-blocker was used, a short-acting and, therefore, were subjected to an abrupt h-blocker h-blocker such as esmolol was recommended. If a long- withdrawal in the perioperative period. Another possi- acting h-blocker was used, the study medication was bility is that study’s long-term survival data did not discontinued, although blinding and data follow-up were include all patients who had been randomized. There continued. were 4 patients in the atenolol group and 2 patients in Our results show that the incidence of bradycardia the placebo group who died during the first 7 days and hypotension requiring treatment was significantly postoperatively while in hospital (the period of treat- higher in the metoprolol group confirming a pharma- ment administration). Only subjects who survived to cologic effect. The incidence of bronchospasm and discharge were included in the published analysis. atrioventricular blocks were low and similar in both Including all events after randomization, that is, intent- groups. The rate of drug discontinuation was low and to-treat analysis, the actual 2-year mortality was 13 in the similar in both groups. A recent study showed that there atenolol group and 23 in the placebo group ( P = .1). may be an increase in all-cause mortality with the Although promising, the difference is not significant, administration of h-blockers in RCRI 1 or 2 patients.18 and the associated 95% CI was wide (20.2% to 0.9% Although we have no evidence that increased side risk difference for atenolol); thus, this trial does not effects in our treatment group resulted in increased provide reliable results. A similar criticism could be American Heart Journal Yang et al 989 Volume 152, Number 5

made of the current study. Although our observed risk selected by surgical procedure. The question of h- reduction was firmly nonsignificant, the upper end of blockers reducing cardiac complications in patients the 95% CI was 48.2%, leaving scope for clinically selected by medical comorbidities is being addressed by important effects. MaVS was designed to have 80% the POISE trial, a multicenter randomized controlled trial power for a 50% relative risk reduction, a lower than by Yang and Devereaux.22 anticipated event rate caused the breliablyQ detectable Our study setting was in vascular surgery, considered risk reduction to be about 60%. bhigh-risk surgeryQ and included as a risk factor in the A comparison of treatment effects (RRR) is shown in RCRI. Because vascular patients are considered the Table VIII. Variability in treatment effect between these highest risk group among surgical populations, the 4 studies is larger than would be expected by chance approach of applying h-blockade broadly to other (ie, variation, as indicated by the significant test of homo- lower risk) surgical populations may be of even less geneity. There are several factors, which may make the benefit in efficacy. Intraoperative bradycardia and studies different: the study of Mangano et al6 did not hypotension were more frequent in the treatment include inhospital mortality as an intent-to-treat study; group. Such side effects, occurring in the monitored the study of Poldermans et al8 was not blinded and was patient under anesthesia, would be acceptable if conducted on a medically preselected population; and evidence for improved outcomes is definite. Given the both POBBLE19 and MaVS were exclusively on vascular current level of evidence, there is a need for caution patients. MaVS remains the largest study to focus only on based on surgical populations in using h-blockers higher-risk vascular surgery: 496 patients undergoing perioperatively.7,23 abdominal aortic and axillofemoral bypasses in addition In summary, our study did not show a clinically useful to infrainguinal vascular surgery. metoprolol effect in reducing the cardiac event rate in The study DIPOM20 has likewise reported that meto- vascular patients at 30 days postoperative or at 6 prolol treatment did not result in any difference in months. Our results call into question the current primary outcome (time to composite outcome of all- recommendations to use h-blockers routinely in these cause mortality, acute MI, unstable angina, or CHF). That higher-risk patients. study randomized 921 diabetic patients; only 7% under- went vascular surgery and 40% underwent low-risk surgery. A significant increase in bradycardia and We thank Ms L Costantani for her help in coordi- hypotension in the treatment group, 32% versus 17%, nating the study. We also thank Drs C Cina, R Kolesar, was also noted. D Crosby, and Ms Mary-Helen Blackall RN from the Should patients undergoing vascular surgery receive Hamilton General Campus, Hamilton Health Sciences; routine perioperative h-blockade? In the publication of Dr D Mark from Kingston General Hospital; Drs K Lee et al14 on RCRI, patients who were using h-blockers Harris, L Patrick and Ms Marge Lovell RN from London at the time of admission had a similar rate of cardiac Health Sciences; and K Wilson-Yang, PhD, for clarifi- complications as patients who were not. The study of cation of the troponin assays and editing. We would Mangano et al6 was in patients with or at risk for also like to acknowledge the comments and sugges- coronary artery disease, but vascular patients constitut- tions provided by Dr S Yusuf. ed only about 40% of the study population. Others advocate perioperative h-blockers for patients with 1 or 9,21 2 RCRI factors. By virtue of undergoing vascular References surgery, all of our patients had at least 1 RCRI factor. 1. Goldman L, Caldera DL, Nussbaum SR, et al. Multifactorial index of Without referral bias and after 496 patients, our study cardiac risk in noncardiac surgical procedures. N Engl J Med observed a difference of only 15% in RRR and was not 1977;297:845 -50. statistically significant. Based on these results, we 2. Detsky AS, Abrams HB, McLaughton JR, et al. Predicting cardiac believe that a widespread prophylaxis of all vascular complications in patients undergoing non-cardiac surgery. J Gen patients is not indicated. Intern Med 1986;1:211 -9. The enrolment of only h-blocker–naive patients may 3. Bode Jr RH, Lewis KP, Zarich SW, et al. Cardiac outcome after be criticized for potentially excluding high-risk patients. peripheral vascular surgery. Comparison of general and regional MaVS was designed to show if a prophylactic h-blocker anesthesia. Anesthesiology 1996;84:3 - 13. protocol should be instituted in vascular patients. 4. Krupski WC, Layug EL, Reilly LM, et al. Comparison of cardiac Patients already on h-blockers would not be subjected to morbidity rates between aortic and infrainguinal operations: two- year follow-up. J Vasc Surg 1993;18:609 - 17. withdrawal. Similarly, patients with contraindications to h 5. L’Italien GJ, Cambria RP, Cutler BS, et al. Comparative early and -blockers would not be given such therapy. In clinical late cardiac morbidity among patients requiring different vascular h practice, only patients not on -blockers would be surgery procedures. J Vasc Surg 1995;21:935 -44. subjected to a prophylaxis protocol. We feel our study 6. Mangano DT, Layug EL, Wallace A, et al. Effect of atenolol on has answered the question of prophylaxis in patients mortality and cardiovascular morbidity after noncardiac surgery. American Heart Journal 990 Yang et al November 2006

Multicenter Study of Perioperative Ischemia Research Group. N Engl 15. Ashton CM, Petersen NJ, Wray NP, et al. The incidence of J Med 1996;335:1713 - 20. perioperative myocardial infarction in men undergoing noncardiac 7. Devereaux PJ, Yusuf S, Yang H, et al. Are the recommendations surgery. Ann Intern Med 1993;118:504 -10. to use perioperative b-blocker therapy in patients undergoing 16. Landesberg G, Mosseri M, Wolf Y, et al. Perioperative myocardial noncardiac surgery based on reliable evidence? Can Med Assoc J ischemia and infarction: identification by continuous 12-lead 2004;171:245 -7. electrocardiogram with online ST-segment monitoring. Anesthesi- 8. Poldermans D, Boersma E, Bax JJ, et al. The effect of bisoprolol on ology 2002;96:264 - 70. perioperative mortality and myocardial infarction in high-risk 17. Mangano DT, Browner WS, Hollenberg M, et al. Association of patients undergoing vascular surgery. Dutch Echocardiographic perioperative myocardial ischemia with cardiac morbidity and Cardiac Risk Evaluation Applying Stress Echocardiography Study mortality in men undergoing noncardiac surgery. N Engl J Med Group. N Engl J Med 1999;341:1789 - 94. 1990;323:1781 -8. 9. Grayburn PA, Hillis LD. Cardiac events in patients undergoing 18. Lindenauer PK, Pekow P, Wang K, et al. Perioperative beta-blocker noncardiac surgery: shifting the paradigm from noninvasive risk therapy and mortality after major noncardiac surgery. N Engl J Med stratification to therapy. Ann Intern Med 2003;138:506 - 11. 2005;353:349 - 61. 10. Alpert JS, Thygesen K, Antman E, et al. Myocardial infarction 19. Brady AR, Gibbs JS, Greenhalgh RM, et al. Perioperative beta- redefined—a consensus document of The Joint European Society of blockade (POBBLE) for patients undergoing infrarenal vascular Cardiology/American College of Cardiology Committee for the surgery: results of a randomized double-blind controlled trial. J Vasc redefinition of myocardial infarction. J Am Coll Cardiol Surg 2005;41:602 -9. 2000;36:959 - 69. 20. Juul AB, Wetterslev J, Gluud C, et al. Effect of perioperative beta 11. Baron JF, Mundler O, Bertrand M, et al. Dipyridamole-thallium blockade in patients with diabetes undergoing major non-cardiac scintigraphy and gated radionuclide angiography to assess cardiac surgery: randomised placebo controlled, blinded multicentre trial. risk before abdominal aortic surgery. N Engl J Med BMJ 2006;332:1482. 1994;330:663 -9. 21. Boersma E, Poldermans D, Bax JJ, Steyerberg EW, Thomson IR, 12. Hosmer DW, Lemeshow S. Applied survival analysis: regression Banga JD, et al. Predictors of cardiac events after major vascular modeling of time to event data. New York, NY: Wiley, 1990. surgery: role of clinical characteristics, dobutamine echocardiogra- 13. Eagle KA, Berger PB, Calkins H, et al. ACC/AHA guideline update phy, and beta-blocker therapy. JAMA 2001 Apr 11;285(14): for perioperative cardiovascular evaluation for noncardiac sur- 1865-73 2001;285:1865-73. gery—executive summary a report of the American College of 22. The POISE Trial Investigators. Rationale, design, and organization Cardiology/American Heart Association Task Force on Practice of the PeriOperative Ischemic Evaluation (POISE) Trial: a Guidelines (Committee to Update the 1996 Guidelines on Peri- randomized controlled trial of metoprolol versus placebo in operative Cardiovascular Evaluation for Noncardiac Surgery). patients undergoing noncardiac surgery. Am Heart J Circulation 2002;105:1257 - 67. 2006;152:223 - 30. 14. Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and 23. Devereaux PJ, Leslie K, Yang H. The effect of peri-operative beta- prospective validation of a simple index for prediction of cardiac blockers on patients undergoing noncardiac surgery—is the answer risk of major noncardiac surgery. Circulation 1999;100:1043 -9. in? Can J Anaesth 2004;51:749 -55.