Cardiovascular Anesthesiology

Cardiovascular and Thoracic Education

Hemostasis and Transfusion Medicine

Section Editor: Charles W. Houge, Jr.

Section Editor: Martin J. London

Section Editor: Jerrold H. Levy

Does Tight Heart Rate Control Improve Beta-Blocker Efficacy? An Updated Analysis of the Noncardiac Surgical Randomized Trials W. Scott Beattie, MD, PhD* Duminda N. Wijeysundera, MD† Keyvan Karkouti, MD, MSc† Stuart McCluskey, MD, PhD* Gordon Tait, PhD*

BACKGROUND: Recent meta-analyses assessing the efficacy of perioperative ␤-blockade trials have failed to show a reduction in postoperative morbidity and mortality. Tight control of heart rate (HR) has been suggested to improve these outcomes. Metaanalyses have not considered the influence of tight HR control on the efficacy of perioperative ␤-blockade. METHODS: Using previously published search strategies, we identified all randomized trials evaluating perioperative ␤-blockers after noncardiac surgery. This search yielded 10 trials with 2176 patients. We used the data from these studies to correlate measures of HR control with major postoperative outcomes, primarily in-hospital myocardial infarction (MI). Odds ratio (OR) and 95% confidence intervals (CI) were calculated, and metaregression was performed correlating measures of HR control with MI. RESULTS: The combined results of all studies did not show a significant cardioprotective effect of ␤-blockers, with considerable heterogeneity among the studies (OR ⫽ 0.76; 95% CI ⫽ 0.4 –1.4; P ⫽ 0.38 heterogeneity: I2 ⫽ 34%). However, grouping the trials on the basis of maximal HR showed that trials where the estimated maximal HR was ⬍100 bpm were associated with cardioprotection (OR ⫽ 0.23; 95% CI ⫽ 0.08 – 0.65; P ⫽ 0.005) whereas trials where the estimated maximal HR was ⬎100 bpm did not demonstrate cardioprotection (OR ⫽ 1.17; 95% CI ⫽ 0.79 –1.80; P ⫽ 0.43) with no heterogeneity. Moreover, metaregression of the HR response to ␤-blockade against the log OR of postoperative MI demonstrated a linear association between the effect of ␤-blockade on the mean, maximal, and variation in HR and the OR of an MI (r2 ⫽ 0.63; P ⬍ 0.001) where a larger effect of ␤-blockers on HR was associated with a decreased incidence of postoperative MI. Across all studies, ␤-blockade resulted in a reduction in postoperative HR (weighted mean difference: 8.6 bpm; 95% CI ⫽ ⫺9.6 to ⫺7.6; I2 ⫽ 85.3%) with considerable heterogeneity. This large heterogeneity in HR response to ␤-blockade was found to be related, in part, to the type of ␤-blocker, specifically, metoprolol, and the concomitant use of calcium channel blockers. Calcium channel blocker use and ␤-blockers other than metoprolol resulted in more effective control of HR. There was wide variability in the HR response to ␤-blockade. Twenty-five percent of patients receiving ␤-blockers had episodes when the HRs were more than 100 bpm, although 15% of placebo patients also had bradycardia, which would have required a dose reduction had they been administered ␤-blockers. Finally, this analysis found that perioperative ␤-blockade was associated with an increased incidence of bradycardia (OR ⫽ 3.49; 95% CI ⫽ 2.4–5.9) and congestive heart failure (OR ⫽ 1.68; 95% CI ⫽ 1.00–2.8). CONCLUSIONS: The trials that achieve the most effective control of HR are associated with a reduced incidence of postoperative MI, suggesting that effective control of HR is important for achieving cardioprotection. Second, this analysis demonstrates that administration of ␤-blockers does not reliably decrease HRs in all patients, and may be associated with increased side effects. Judicious use of combination therapy with other drugs may be necessary to achieve effective postoperative control of HR. (Anesth Analg 2008;106:1039 –48)

D

espite advancements in surgical and anesthetic techniques, perioperative mortality and morbidity remain a significant problem. Four times as many patients From the Departments of *Anesthesia, and †Health Policy Management, and Evaluation/Clinical Epidemiology, Toronto General Hospital and University of Toronto, Toronto, Ontario. Accepted for publication November 20, 2008. Address correspondence and reprint requests to W. Scott Beattie, MD, PhD, FRCPC, Department of Anesthesia, Toronto General Hospital and University of Toronto, EN 3-450 200 Elizabeth St., Toronto, Ontario M5G 2C4, Canada. Address e-mail to [email protected]. Copyright © 2008 International Anesthesia Research Society DOI: 10.1213/ane.0b013e318163f6a9

Vol. 106, No. 4, April 2008

die on the first day after noncardiac surgery compared with the daily expected mortality rate for an age and sex-adjusted general population,1 and cardiovascular events are the single most frequent cause of postoperative mortality, causing approximately one-third of deaths.2 Beta-blockade is thought to reduce postoperative cardiac morbidity. In 2006, the American College of Cardiology/American Heart Association (ACC/AHA) Task Force on Practice Guidelines issued a focused update on perioperative ␤-blockade.3 The committee issued Class l recommendations for patients having surgery who are receiving ␤-blockers to treat angina, 1039

symptomatic arrhythmia, hypertension, or other ACC/ AHA Class l guideline indications. The Class l indications include patients undergoing vascular surgery at high cardiac risk due to findings of ischemia on preoperative testing.3 These recommendations are essentially based on the results of two trials,4,5 which the committee acknowledges by indicating that the recommendation was based on level C evidence. The recent publication in this journal of a pro-con debate on the subject,6,7 yet another meta-analysis8 and an editorial by the chairman of the guidelines committee,9 serves to highlight the controversy surrounding perioperative ␤-blockade. The two most recent systematic reviews of perioperative ␤-blockade8,10 suggest that there is insufficient evidence to recommend universal perioperative use of ␤-blockers. A moderate-to-high amount of heterogeneity complicates these meta-analyses*. Heterogeneity in meta-analysis refers to the variation in study outcomes among studies. High heterogeneity indicates that the difference between studies is more than one would expect by chance, and that the results of the individual studies are sufficiently different to make combining the outcomes inappropriate. High heterogeneity suggests that one or more of the studies in the group show distinctly different results from the rest of the group. If removal of one or more studies from the group eliminates the heterogeneity, the outcomes of the remaining studies represent a range of consistent results, which can be combined with confidence. As an example of dealing with heterogeneity, one metaanalysis8 eliminated the bisoprolol trial2 from the analysis. The result was an increase in the relative risk of the composite outcome [nonfatal myocardial infarction (MI), death] from 0.79 to 0.93 with no heterogeneity. In that analysis, elimination of the bisoprolol trial, as well as 2 other trials,11,12 was justified on the basis of breaches in methodology, in that there was inappropriate allocation concealment† and both the care and administration of the study drug (which allowed for titration) was not blinded, although the data were collected by blinded assessors. Recently reported observational data suggest that cardioprotective efficacy of perioperative ␤-blockers is improved by achieving tight heart rate (HR) control.13,14 We reasoned that if HR control is an important factor in postoperative cardioprotection then we *Heterogeneity in meta-analysis refers to the variation in study outcomes between studies. The I2 statistic that was used in this analysis describes the percentage variation that occurs as the result of this variation rather than by chance. Many reviews that have identified heterogeneity in the analysis have dealt with it by using a random effects model. This is not the optimal solution to a finding of excessive variability in outcomes between studies. It is more appropriate to search for confounding variables as we are attempting in this analysis. † Poor allocation concealment may allow the investigator to know which arm the next patient will be randomized to. Thus, it is possible to skew the randomization process and allow for bias.

1040

Rate Control Reduces Postoperative Myocardial Infarction

should be able to demonstrate the effect with chronotropic data extracted from the trials in these metaanalyses. Furthermore, since HR control has not been previously considered in these meta-analyses, it may account for the observed heterogeneity, which has complicated these analyses. In this communication, we re-examined these systematic reviews, attempting to determine the degree of HR control in each study and the factors which may have influenced cardioprotection and variability in the effectiveness of HR control.

METHODS The data used in this analysis include noncardiac surgical trials in the two most recent metaanalysis.8,10 Before this analysis, we conducted a final search (completed May 24, 2007) to determine whether any further trials had been published with the same electronic search criteria used in previous meta-analyses.8,10 We included trials reporting both cardiovascular events and details of HR responses to ␤-blockade. Our primary dependent variable was MI; secondary outcomes included all-cause mortality, hypotension requiring treatment, bradycardia requiring treatment, postoperative congestive heart failure (CHF), and stroke. The final analysis also evaluated dosing adjustments in response to hypotension and bradycardia. Odds ratio (OR) and 95% confidence intervals (CI) were calculated using REVMAN 4.2 (Cochrane collaboration). As postoperative myocardial ischemia and MI occur predominately in the first 24 h after surgery,5,15 we focused our analysis of HR responses on this period. In trials in which a series of postoperative HR were reported, we used the first 24-h postoperative HR. In three trials, we used the reported average for the first seven postoperative days. For our analysis, we defined the maximal HR as the upper range of HR and, when this range was not given, this was estimated as 99th percentile (3 standard deviations above the mean HR). Reduction in maximal HR is an important measure of perioperative ␤-blockade since periods of rapid HR increase myocardial oxygen demand and the chance of myocardial ischemia and tissue damage.16 In this analysis, we have assessed four measures of HR control with ␤-blockade: 1. The weighted mean difference (WMD‡) between mean HRs in the placebo and ␤-blocked groups and the 95% CI. 2. The mean HRs achieved in the ␤-blocked patients. 3. The maximal HRs recorded in the ␤-blocked patients. 4. The difference between the mean and maximal HRs in ␤-blocked patients (HR variability adjusted for the basal HR). ‡

The weighted mean difference uses an inverse of the variance to calculate the weighting.

ANESTHESIA & ANALGESIA

Table 1. Trials Included in This Analysis of Perioperative Beta Blockade

Year

Number

Surgery and population

Intervention

Jakobsen et al.22

Trial

1997

36

Inpatient elective thoracotomy for lung resection, patients without cardiovascular problems

Wallace et al.5

1998

200

Inpatient elective noncardiac surgery, patients with or at risk for CAD

Bayliff et al.17

1999

99

Oral metoprolol 100 mg or placebo 90 min before surgery and once daily thereafter until day 11 postsurgery or hospital discharge if sooner Atenolol 5 mg (for a HR ⱖ55 bpm and a SBP ⱖ100 mm Hg) twice 30 min before surgery, dosing postoperative daily Oral propranolol 10 mg every 6 h

Poldermans et al.4

2001

112

Raby et al.21

1999

26

Inpatient elective vascular surgery, ischemia demonstrated during preoperatively

Zaugg et al.12

1999

63

Inpatient elective major noncardiac surgery, patients with CHF were excluded

Urban et al.11

2000

120

Inpatient elective total knee arthroplasty, patients had known or probably CAD

POBBLE18

2005

103

Elective Vascular surgery

Oral bisoprolol 5 mg daily weekly before surgery, 10 mg daily if the HR was ⬎60 metoprolol infusions to keep the HR below 80 bpm, in NPO patients Esmolol infusion adjusted every hour to reduce the HR below a predetermined ischemic threshold Atenolol 5 mg (for a HR ⬎54 bpm and a SBP ⬎99 mm Hg) or atenolol 5 mg injections every 5 min during surgery to maintain a HR ⬍80 bpm and a MAP ⱕ20% of preoperative MAP Esmolol infusion titrated to keep the HR ⬍80 bpm metoprolol 25 mg BID with titration to keep the HR ⬍ 80 bpm Metoprolol

MaVS20

2006

496

Elective Vascular

Metoprolol

DIPOM19

2006

921

Elective surgery in patients with diabetes

Metoprolol XL

Inpatient major thoracic surgery

Inpatient elective abdominal aortic or infrainguinal arterial surgery, RDRI ⬎2 and DSE

Outcomes available for analysis CHF, MI, bronchospasm

Death, MI, CHF, hypotension, bradycardia, bronchospasm Death MI, cardiac arrest, CHF, hypotension, bradycardia, bronchospasm Death MI

MI

MI

MI

Death, MI, CVA, bradycardia, hypotension Death, MI, CHF, hypotension, bradycardia Death, MI CVA, CHF, hypotension, bradycardia

RCRI ⫽ revised cardiac risk index; DSE ⫽ dobutamine stress echocardiogram; CAD ⫽ coronary artery disease; CHF ⫽ congestive heart failure; HR ⫽ heart rate; SBP ⫽ systolic blood pressure; MAP ⫽ mean arterial pressure; MI ⫽ myocardial infarction; CVA ⫽ cerebrovascular accident.

We then assessed the effect of HR on the heterogeneity described in previous meta-analyses using HR as both a dichotomous and a continuous variable. For the dichotomous analysis, we divided the trials equally on the basis of maximal HR with one group of trials having maximal HRs ⬍100 bpm and the other having Vol. 106, No. 4, April 2008

maximal HRs more than 100 bpm. Ischemic events have been shown to occur when HRs increase to 100 bpm.16 We also performed a sensitivity analysis where we eliminated the bisoprolol trial2 from the above analysis, since it was possible that this had exerted disproportionate weighting in the analysis. For the © 2008 International Anesthesia Research Society

1041

Table 2. Details of Heart Rate (HR) Data Extracted Study Poldermans et al.4 Bayliff et al.17 POBBLE18 MaVS20 DIPOM19 Zaugg et al.12 Jakobsen et al.22 Urban et al.11 Raby et al.21 Wallace et al.5

Data extracted from: (timing)

Postoperative HR (beta blocked)

sd or range

Highest HR (sd ⫻ 3)

Table 4 (7 d average) Text pg 184b Text pg 605 (7 d average) Table 3 PACU Table 6 (Day 3, which is the first postoperative day) Figure 2 (24 h) Table 4 (18 h) Text Table 2 Tables 4 and 6 (7 d average)

70 90 75 69.4 74

80a 140 16 12.9 12/120c

80 169 117 108 120

7 13

84 119 95 100 126

63 80 65 72 82

100 14

a

Poldermans HR data are presented as mean and range. Bayliff HR data are for patients with arrhythmia and neither timing nor baseline HR is reported. c DIPOM reports both SD and range. All other studies report SD. b

continuous analysis, we used metaregression§ to compare the measures of HR control against the OR for MI in each trial. Metaregression was performed using S-plus/R statistical software and the MiMa function㛳. A sensitivity analysis of this evaluation was also performed where the Bayliff et al.,17 and Urban et al. trials11 were eliminated as the HR data were based on collated details of arrhythmias and from the patient with a MI.

RESULTS Our search did not identify new noncardiac ␤-blockade publications since the publication of Wiesbauer et al.8 The present analysis is based on the 10 trials,4,5,11,12,17–22 which reported data on both HR and MI (Table 1). HR data, location of where the data can be found in the publication, and the methods used to calculate the peak HR are presented in Table 2. In three trials, the data are the average of the first 7 days after the operation; in four trials, the data are for the first postoperative day; and in one trial, the data are for the recovery room only. In one trial, the HR data were given only for patients with arrhythmias, and in one trial, the HR data were given only for the patient who sustained a postoperative MI. The combined results of all studies did not show a significant cardioprotective effect of ␤-blockers, with considerable heterogeneity among the studies. Figure 1 shows a Forrest plot of the effects of ␤-blockade on MI for all studies (OR ⫽ 0.76; 95% CI ⫽ 0.4 –1.4; P ⫽ 0.38; heterogeneity: I2 ⫽ 34%). Two analyses were then § Metaregression is a tool which can be used to measure the variability of a given response. By assessing the effects of a covariate (in this case the measure of HR control), on the outcome of interest (in this case the odds ratio of a myocardial infarction), we can measure the extent that this covariate explains the heterogeneity. The r2 has been used to measure the extent to which a covariate explains the heterogeneity. 㛳 Veichtbauer: MiMa: An S-plus/R function to fit meta-analytic mixed-random- and fixed effects models [computer and software manual] Retrieved from http://www.wvbauer.com, URL accessed Feb. 9, 2007.

1042

Rate Control Reduces Postoperative Myocardial Infarction

performed to determine whether variability among studies in the effective control of HR could account for this heterogeneity. In the first analysis, the trials were divided into two groups based on maximal reported postoperative HR. In trials where patients receiving ␤-blockers had a maximal HR of ⬍100 bpm, there was a significant reduction in the incidence of postoperative MI (OR ⫽ 0.23; 95% CI ⫽ 0.08 to 0.65; P ⫽ 0.005; heterogeneity: I2 ⫽ 0%), whereas in trials where patients had maximal HR more than 100 bpm, there was no reduction in MI (OR ⫽ 1.17; 95% CI ⫽ 0.79 to 1.80; P ⫽ 0.43; heterogeneity: I2 ⫽ 0%). A sensitivity analysis where the results of the Poldermans et al.’s4 trial was removed did not substantially change this result (OR ⫽ 0.28; 95% CI ⫽ 0.10 to 0.79; P ⫽ 0.02; heterogeneity: I2 ⫽ 0%). The elimination of heterogeneity achieved by dividing the groups according to maximal HR suggests that effective control of HR is an important confounding variable, which may be responsible for the heterogeneity of the combined results of all the studies. The relationship between effective control of HR and postoperative MI was further analyzed using meta regression of continuous measures of heat rate control (Figs. 2 and 3). There was a significant relationship between the log OR for MI and the WMD between mean HRs in the placebo and ␤-blocked groups (r2 ⫽ 0.53, Fig. 2a), the maximal HR in ␤-blocked patients (r2 ⫽ 0.59, Fig. 2b) and the difference between the mean and maximal HRs in ␤-blocked patients (r2 ⫽ 0.63). This relationship was not substantially changed in a sensitivity analysis in which the Bayliff et al. trial17 was removed (r2 ⫽ 0.62). In considering all 10 studies, ␤-blockade resulted in a decrease in mean HR with considerable heterogeneity (WMD ⫽ ⫺10.8; 95% CI ⫽ 13.6 to ⫺8.6; heterogeneity I2 ⫽ 66.1%), (Table 3). We explored four potential confounding variables, which might account for this heterogeneity: the time that the HRs was obtained, blinding of treating physicians, the ANESTHESIA & ANALGESIA

Review: Perioperative beta-blockers Jan 2004 Comparison: 02 HEART RATE CONTROL Outcome: 01 CARDIAC EVENT (30 days) Study or sub-category

BETA-BLOCKER n/ N

CONTROL n/N

01 HEART RATE UNDER 100 ( in all patients receiving BB) 0/59 Poldermans (BIS) (4) 0/15 Raby (esmolol inf) (21) 0/43 Zaugg (Atenolol) (20) 1/60 Urban (esmolol inf) (16) 3/53 POBBLE (Metop) (17) 230 Subtotal (95% CI) Total events: 4 (BETA-BLOCKER), 21 (CONTROL) Test for heterogeneity: Chi² = 3.47, df = 4 (P = 0.48), I² = 0% Test for overall effect: Z = 2.80 (P = 0.005)

OR (random) 9 5% C I

Weight %

9/53 1/11 3/20 3/60 5/44 188

02 HEART RATES ABOVE 100 ( in some of the patients receiving BB) 1/15 0/15 Jakobsen (Metop) (22) 3/99 2/101 Wallace (Atenolol) (5) 1/50 0/49 Bayliff (Prop) (15) 28/462 21/459 DIPOM (Metop) (18) 19/246 21/250 MaVS (Metop) (19) 872 874 Subtotal (95% CI) Total events: 52 (BETA-BLOCKER), 44 (CONTROL) Test for heterogeneity: Chi² = 1.57, df = 4 (P = 0.81), I² = 0% Test for overall effect: Z = 0.82 (P = 0.41)

0.01

1

0.1

Favours treatment

10

Yea r

4.15 3.24 3.80 6.13 11.87 29.19

0.04 0.23 0.06 0.32 0.47 0.23

[0.00, [0.01, [0.00, [0.03, [0.11, [0.08,

0.69] 6.09] 1.17] 3.19] 2.08] 0.65]

1999 1999 1999 2000 2005

3.26 8.95 3.37 28.42 26.81 70.81

3.21 1.55 3.00 1.35 0.91 1.19

[0.12, [0.25, [0.12, [0.75, [0.48, [0.79,

85.20] 9.46] 75.44] 2.41] 1.74] 1.80]

1997 1998 1999 2004 2004

100.00

1102 1062 Total (95% CI) Total events: 56 (BETA-BLOCKER), 65 (CONTROL) Test for heterogeneity: Chi² = 13.69, df = 9 (P = 0.13), I² = 34.3% Test for overall effect: Z = 0.88 (P = 0.38)

OR (random) 9 5% C I

0.76 [0.41, 1.41]

100

Favours control

Figure 1. Forrest plot of 10 trials assessing ␤-blockers and postoperative myocardial infarction (MI). The trials are grouped according to the highest heart rate (HR) recorded in the trials. Trials in Group 1 had the maximal recorded HR ⬍100 bpm, whereas trials in Group 2 had maximal recorded HR more than 100 bpm. There is a significant reduction in postoperative MI in trials in Group 1. The 95% confidence intervals of the odds ratios for the two groups do not overlap. type of ␤-blocker, and the concomitant use of a calcium channel blocker (Table 4). Dividing the trials into two groups according to each of these variables resulted in the elimination of the heterogeneity in one group, but heterogeneity remained in the other group, indicating that these variables did not fully account for the heterogeneity of the trials as a whole. Although ␤-blockers did not reliably control increases in HR (many patients had maximal HR exceeding 100 bpm), 31% of patients receiving ␤-blockers required a dose reduction due to bradycardia (OR ⫽ 2.21; 95% CI ⫽ 1.75 to 2.8; heterogeneity: I2 ⫽ 0%). However, more than 17% of placebo patients also had a postoperative HR that fulfilled the trial definition of bradycardia and would have required a dose reduction had they been in the active drug group. The effects of ␤-blockers on the remainder of the important cardiovascular events are tabulated. All cause mortality was reported in six trials including 1927 patients with an incidence of 1.7% (OR ⫽ 0.75; 95% CI ⫽ 0.27 to 2.07). Beta-blockade was associated with an increased incidence of treated bradycardia (OR ⫽ 3.49; 95% CI ⫽ 2.4 to 5.9; heterogeneity: I2 ⫽ 0%) and CHF (OR ⫽ 1.68; 95% CI ⫽ 1.0 to 2.8; heterogeneity: I2 ⫽ 0%). We are unable to assess Vol. 106, No. 4, April 2008

whether HR control had an influence on the occurrence of CHF or stroke. These events were not reported in the Poldermans et al.,4 Zaugg et al.,12 Urban et al.,11 or Raby et al.21 trials. The POBBLE study18 reported the incidence of stroke but there were no data on CHF.

DISCUSSION We have demonstrated that the perioperative ␤-blockers trials are complicated by significant heterogeneity in both the HR response to ␤-blockade and on MI. This analysis also shows that chronotropic response to ␤-blockade is highly variable, as expressed in both the difference in means between the placebo and active arms of the study and the variability in the occurrence of tachycardia in the treated group. This variation in HR response to ␤-blockade may explain much of the heterogeneous response that ␤-blockers have on perioperative MI. The metaregression shows that approximately 60% of the variation seen in the cardioprotective effect of ␤-blockade is due to the HR response; with better control of HR being associated with better outcomes. Thus, the results of this analysis are consistent with the hypothesis that tight HR control may lead to fewer postoperative MIs. © 2008 International Anesthesia Research Society

1043

a

1

r = 0.73 r2 = 0.53

0.5

LOG OR

0 -20

-15

-10

-5

0

5

-0.5

Figure 2. (a) Metaregression analysis comparing the lower bound 95% confidence interval (CI) of the weighted mean difference for each of the studies with postoperative heart rate (HR) reported. The lower CI represents the minimal difference in HR between the active and placebo arms of each study. (b) Metaregression analysis of the mean HR (open square) and the maximal HR (closed square) in each study against the log OR of a myocardial infarction for each study. The divergent lines indicate increasing variability and higher maximal HRs are associated reduced cardioprotection. The differences between mean and maximal HR are plotted in Figure 3.

-1

-1.5

1

r = 0.77 r2 = 0.59

r = 0.63 r2 = 0.39

0.5

0 LOG OR

50

75

100

125

150

b

175

-0.5

-1

-1.5 HR

The rationale for the cardioprotective effect of ␤-blockers involves a reduction in myocardial oxygen demand related primarily to the reduction in HR from the blockade of adrenergic activity. Tachycardia is common in the postoperative period in response to many factors—stress, hypovolemia, and hypoxia to name a few—and ␤-blockade reduces both the resting mean and the maximal HR in response to exercise and hypoxia.23 The relationships that we have shown between cardioprotection and the effective attenuation of postoperative tachycardia with ␤-blockade are consistent with this rationale for cardioprotection. However, this analysis also shows that tight HR control with ␤-blockers may be difficult to achieve in all patients and may be associated with increased side 1044

Rate Control Reduces Postoperative Myocardial Infarction

effects, including bradycardia and CHF. Several factors may be responsible for the variability that we observed in HR. First, four trials were not blinded to treating physicians,4,11,21,24 as the protocols were designed to achieve a target HR. These trials were associated lower maximal HRs and the greatest cardioprotection (Fig. 1). Recent trials with blinding and fixed upper dosage regimens were associated with more frequent tachycardia and failed to show decreases in MI. The type of drug may also be associated with the variability in the HR response. The use of metoprolol was associated with less difference between the active and placebo arms than trials using other ␤-blockers (Table 4). Part of the HR variability could also have been due to the use of calcium channel ANESTHESIA & ANALGESIA

Figure 3. Metaregression analysis con-

r = 0.79 r2 = 0.63 p < 0.001

1.5

Log Odds Ratio for Myocardial Infarction

ducted using the Viechtbauer method comparing the difference between maximal and mean HR and the log OR for myocardial infarction in each study. Data are from the first recorded postoperative HR. Each trial is specified by name of the first author. The size of the circle represents the weighting of each trial in the regression. There is a significant correlation between the difference between maximal and mean HR and the log odds ratio for MI. The r2 ⫽ 0.63 is highly statistically significant. A separate sensitivity analysis was conducted where the data from the Bayliff et al. and Urban et al. trials were excluded is still highly significant (r2 ⫽ 0.57 P ⬍ 0.01) and the slope of the relationship is unchanged.

JAKOBSEN BAYLIFF

0.5

MANGANO

5

15

MaVS 35

25

45

DIPOM55

65

-0.5

POBBLE

URBAN -1.5

RABY

-2.5

ZAUGG POLDERMANS -3.5

Difference between Maximum and Mean Heart Rate

Table 3. Effects of Dosing on Heart Rate (HR) Response Difference in postoperative HR WMD (95% CI) 22

Jakobsen et al. n ⫽ 36 Wallace et al.5 n ⫽ 200 Bayliff a et al.17 n ⫽ 100 Poldermans et al.4 n ⫽ 112 Raby et al.21n ⫽ 26 Zaugg et al.12 n ⫽ 63 Urbanb et al.11 n ⫽ 120 POBBLE18 n ⫽ 103 MaVS20 n ⫽ 496 DIPOM19 n ⫽ 921 Pooled results

Per protocol dose reductions Drug

Placebo

⫺7 (⫺13.4 to ⫺0.5)

NA

NA

⫺12 (⫺14.6 to ⫺9.3)

25 (25%)

13 (13%)

⫺16 (⫺27.2 to ⫺4.2)

NA

⫺11 (⫺13.2 to ⫺8.8)

Treated bradycardia Placebo

Drug

Placebo

NA

NA

NA

2 (2%)

0

22 (22%)

34 (34%)

NA

25 (51%)

4 (8%)

NA

NA

NA

NA

NA

NA

23 (40%)

18 (41%)

⫺8 (⫺12.7 to ⫺3.3) ⫺20 (⫺24.8 to ⫺15.2)

NA NA

NA NA

NA 2 (10%)

NA 0

NA 9 (20%)

NA 9 (45%)

NA

NA

NA

NA

NA

NA

NA

1 (2.2)

30 (57%)

6 (14%)

NA

NA

28 (12%) 102 (22%)

53 (22%) NA

19 (17%) NA

46 (19%) 70 (15%)

42 (17%) 81 (18%)

⫺8 (⫺14.8 to ⫺1.2)

4 (7.5%)

⫺10 (⫺12.6 to ⫺7.4) ⫺5 (⫺6.6 to ⫺3.4)

55 (22%) 168 (36%)

WMD ⫺8.6 (⫺9.6 to ⫺7.6) I2 ⫽ 85.3%

OR 2.23 (1.7 to 2.8) I2 ⫽ 0%

Drug

Use of calcium channel blockers

OR 3.49 (2.4 to 5.9) I2 ⫽ 0%

OR 0.85 (0.68 to 1.08) I2 ⫽ 39.8%

a

Data from Bayliff et al. documents arrhythmia HR (used to calculate the peak heart rates only). Data from Urban et al. text only for the beta blocked patient who sustained a myocardial infarction. I2 Measure of heterogeneity in the meta-analysis where 0% is associated with no heterogeneity. NA ⫽ data were not available; WMD ⫽ weighted mean difference. b

blockers. Trials in which these drugs were used in more than 25% of patients were associated with a greater difference between active and placebo groups (Table 4). The trials with the best outcomes also had the highest use of these drugs. Consistent with this observation is the report that the use of the nondihydropridine calcium channel blockers has been reported to also improve noncardiac surgical outcomes.25 Tight HR control, while desirable, may not always be achievable. Several trials, some not included in this Vol. 106, No. 4, April 2008

analysis due to their design, consistently show that a certain proportion of patients receiving ␤-blockers are unable to achieve a low postoperative target HR. Recently, the DECREASE II study13 failed to achieve its prescribed degree of HR control in 35% of the study population, despite a rigorous protocol of ␤-blockade. Harwood et al.,26 using IV esmolol with a feedback algorithm to increase dosage in the event of inadequate HR control, was unable to keep HR below 100 bpm in 30% of patients after vascular surgery. Wallace et al. reported that 35% of patients receiving atenolol © 2008 International Anesthesia Research Society

1045

Table 4. Sources of Heterogeneity (Heart Rate (HR) Response to Beta Blockade) Variable Timing Blinding Beta blockers Calcium channel blockers

1st Variable

WMD (95 % CI) I2

24 ha21 ⫺10.1 (⫺11.5 to ⫺8.5) 85% No blinding of Rx4,11,12,21 ⫺12.0 (⫺13.0 to ⫺10.4) 93% Metoprolol18–20,22 ⫺9.2 (⫺10.5 to ⫺7.8) 0% More than 25% of ⫺12.0 (⫺13.5 to ⫺10.5) 83% population on CCB4,5,12

2nd Variable

WMD (95% CI) I2

7 d average4,5,18 ⫺10.8 (⫺12.4 to ⫺9.3) 0% Rx blinded5,17,18–20,22 ⫺9.2 (⫺10.6 to ⫺7.9) 73% All others4,5,17,11,12,21 ⫺13.2 (⫺17.8 to ⫺8.6) 73% All others11,17–22 ⫺9.1 (⫺10.4 to ⫺7.8) 0%

NB the smaller the negative WMD the smaller difference between the active and placebo arms of the study. WMD is the weighted mean difference between beta blocker and placebo HR as extracted in Table 2. I2 is the degree of heterogeneity. a Raby et al. using a esmolol infusion reported 48 h, not included in this analysis. Rx ⫽ therapy; CCB ⫽ calcium channel blockers.

had peak HR more than 100 bpm.5 Approximately 16% of patients with acute coronary syndrome are homozygous for the AA subunit of the ␤-receptor.27 This genetic predisposition is associated with an ineffectiveness of ␤-blockade and an increased mortality after an acute coronary event. The variability in dose response to ␤-blockers demonstrated in this analysis is important, since it shows that patients may be unable to achieve a target HR when administered a fixed dose of ␤-blocker. This was recognized in all trials that had a dose adjustment algorithm. More importantly, it suggests that not all patients will be able to achieve effective control of maximal HR with ␤-blockers alone. The trials where a large proportion of patients were also receiving calcium channel blockers had better control of postoperative HRs. Although postoperative HR control appears to be cardioprotective, this finding must be tempered by the findings that ␤-blockers may be associated with increased CHF. Of concern is that these outcomes may be exacerbated with increasing doses of ␤-blockers to achieve a target HR. Unfortunately, this analysis is unable to adequately address this issue, as studies that attained effective HR control did not report CHF as an outcome. Obviously, the safety of postoperative ␤-blockers is an important area for future investigation. The ongoing POISE trial28 uses IV metoprolol and an extended-release oral formulation. Although the protocol has explicit instructions for dose reduction for bradycardia, there are no instructions for dose modification when HR is excessive. For reasons outlined above, this may be an important design flaw of double-blind trials of perioperative ␤-blockade.10 It will be incumbent upon the investigators of the POISE trial to conduct a secondary analysis to assess the association of tight HR control with cardioprotective efficacy. This study has numerous limitations. As with all meta-analyses, the results are only as good as the information derived from the included trials. The trials in this analysis comprise the best evidence on the perioperative use of ␤-blockers. Although all trials report MI as an outcome, the reporting of other cardiovascular outcomes was variable. This is an important limitation, since we found that ␤-blockers may increase the incidence of CHF. It is possible that increasing the dose of a ␤-blocker to achieve a target HR may increase the potential for 1046

Rate Control Reduces Postoperative Myocardial Infarction

CHF due to the well recognized negative inotropic effects of this class of drug. A second limitation is that the reporting of HRs in these trials was inconsistent. The HR measurements were taken at different postoperative times, and spanning different epochs, which may have occurred at different times of the day. We have tried to address this limitation in several ways. In our assessment, the heterogeneity of HR compared with the HR response at 24 h was compared to those averaged for 7 days and found no difference in the WMD (Table 4). In addition, three trials report numerous HR at different epochs. Inspection of these data shows that in these trials the postoperative HRs are consistent across all time frames as has been reported with other investigations of postoperative MI.15 It is also well recognized that there is a circadian variability in HRs and even in the HR response to stress.34 The extent to which this type of variability played a role in our analysis is unknown. The HRs in the Bayliff et al.17 trial were from episodes of arrhythmia, and in the Urban et al.11 trial, the data were only recorded for patients who sustained a MI. We conducted a sensitivity analysis and the results of the metaregression improved when we excluded data from these trials. Finally, the division of trials into two equal groups on the basis of maximal HR above and below 100 bpm was arbitrary. However, this division resulted in two groups of trials that were free of heterogeneity, and is supported by the observation that ischemic episodes are associated with HR increases to 100 bpm.16 This report has found that the variation in effective HR control among studies is a possible explanation for much of the heterogeneity reported in previous metaanalyses. The strength and the primary utility of this analysis is the generation of hypotheses to guide future research. This analysis supports the hypothesis that tight HR control is associated with better cardiovascular outcomes after noncardiac surgery, but before tight HR control can be adopted as a treatment goal, several issues must be addressed. First, many patients cannot achieve a target HR with a simple dosing regimen as designed in the double-blind, randomized trial. If the titration of medication is required to achieve adequate control of HR, then the doubleblind trial may not be a suitable study design for this ANESTHESIA & ANALGESIA

research. The reasons for an inadequate HR response remain unclear but are likely multifactorial. First and importantly, ␤-blockers, specifically metoprolol, may not exhibit a class effect and may be a poor choice for achieving postoperative cardioprotection, as has been alluded to in other reports.24,29 Metoprolol differs from atenolol in that it has different solubility and central effects which are manifested as a different effect on the parasympathetic system.30 Second, it is likely that patient characteristics and phenotype are important concerning the response to ␤-blockers. This analysis provides some evidence that suggests combination therapy with a calcium channel blockers may be more effective at achieving tight control. Does combination therapy, with tight HR control, confer outcome benefit? Other drugs such as ␣-2 adrenergic agonists31 or thoracic epidural anesthesia/analgesia,32 which have negative chronotropic characteristics, have been shown to reduce postoperative MI as single drugs, and are worthy of evaluation as well. Finally, the safety issues must be investigated, as our analysis provides evidence that ␤-blockers may increase the incidence of CHF and bradycardia. Although this analysis suggests that HR control may be an important goal in treating patients at risk of a postoperative cardiac event, the best method of achieving this goal remains to be determined. REFERENCES 1. Lie SA, Englesaeter LB, Havelin LI, Furnes O, Vollset SE. Early postoperative mortality after 67,548 total hip replacements: causes of death and thromboprophylaxis in 68 hospitals in Norway from 1987 to 1999. Acta Orthop Scand 2002;73:392–9 2. Boersma E, Kertai MD, Schouten O, Bax JJ, Noordzij P, Steyerberg EW, Schinkel AF, van Santen M, Simoons ML, Thomson IR, Klein J, van Urk H, Poldermans D. Perioperative cardiovascular mortality in noncardiac surgery: validation of the lee cardiac risk index. Am J Med 2005;118:1134 – 41 3. American College of Cardiology, American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2002 Guidelines on Perioperative Cardiovascular Evaluation for Noncardiac Surgery), American Society of Echocardiography, American Society of Nuclear Cardiology, Heart Rhythm Society, Society of Cardiovascular Anesthesiologists, Society for Cardiovascular Angiography and Interventions, Society for Vascular Med and Biology, Fleisher LA, Beckman JA, Brown KA, Calkins H, Chaikof E, Fleischmann KE, Freeman WK, Froehlich JB, Kasper EK, Kersten JR, Riegel B, Robb JF, Smith SC Jr, Jacobs AK, Adams CD, Anderson JL, Antman EM, Faxon DP, Fuster V, Halperin JL, Hiratzka LF, Hunt SA, Lytle BW, Nishimura R, Page RL, Riegel B. ACC/AHA 2006 guideline update on perioperative cardiovascular evaluation for noncardiac surgery: focused update on perioperative beta-blocker therapy: a report of the American college of Cardiology/American heart association task force on practice guidelines (writing committee to update the 2002 guidelines on perioperative cardiovascular evaluation for noncardiac surgery) developed in collaboration with the American society of echocardiography, American society of nuclear cardiology, heart rhythm society, society of cardiovascular anesthesiologists, society for cardiovascular angiography and interventions, and society for vascular medicine and biology. J Am Coll Cardiol 2006;47:2343–55 4. Poldermans D, Boersma E, Bax JJ, Thomson IR, Paelinck B, van de Ven LL, Scheffer MG, Trocino G, Vigna C, Baars HF, van Urk H, Roelandt JR; Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echocardiography Study Group: bisoprolol reduces cardiac death and myocardial infarction in high-risk patients as long as 2 years after successful major vascular surgery. Eur Heart J 2001;22:1353– 8 Vol. 106, No. 4, April 2008

5. Wallace A, Layug B, Tateo I, Li J, Hollenberg M, Browner W, Miller D, Mangano DT. Prophylactic atenolol reduces postoperative myocardial ischemia. McSPI research group. Anesthesiology 1998;88:7–17 6. London MJ. Con: beta-blockers are indicated for all adults at increased risk undergoing noncardiac surgery. Anesth Analg 2007;104:11– 4 7. Schouten O, Bax JJ, Dunkelgrun M, Feringa HH, Poldermans D. Pro: beta-blockers are indicated for patients at risk for cardiac complications undergoing noncardiac surgery. Anesth Analg 2007;104:8 –10 8. Wiesbauer F, Schlager O, Domanovits H, Wildner B, Maurer G, Muellner M, Blessberger H, Schillinger M. Perioperative betablockers for preventing surgery-related mortality and morbidity: a systematic review and meta-analysis. Anesth Analg 2007;104:27– 41 9. Fleisher LA. Perioperative beta-blockade: how best to translate evidence into practice. Anesth Analg 2007;104:1–3 10. Devereaux PJ, Beattie WS, Choi PT, Badner NH, Guyatt GH, Villar JC, Cina CS, Leslie K, Jacka MJ, Montori VM, Bhandari M, Avezum A, Cavalcanti AB, Giles JW, Schricker T, Yang H, Jakobsen CJ, Yusuf S. How strong is the evidence for the use of perioperative beta-blockers in noncardiac surgery? systematic review and meta-analysis of randomised controlled trials. BMJ 2005;331:313–21 11. Urban MK, Markowitz SM, Gordon MA, Urquhart BL, Kligfield P. Postoperative prophylactic administration of beta-adrenergic blockers in patients at risk for myocardial ischemia. Anesth Analg 2000;90:1257– 61 12. Zaugg M, Tagliente T, Lucchinetti E, Jacobs E, Krol M, Bodian C, Reich DL, Silverstein JH. Beneficial effects from beta-adrenergic blockade in elderly patients undergoing noncardiac surgery. Anesthesiology 1999;91:1674 – 86 13. Poldermans D, Bax JJ, Schouten O, Neskovic AN, Paelinck B, Rocci G, van Dortmont L, Durazzo AE, van de Ven LL, van Sambeek MR, Kertai MD, Boersma E; Dutch Echocardiographic Cardiac Risk Evaluation Applying Stress Echo Study Group: should major vascular surgery be delayed because of preoperative cardiac testing in intermediate-risk patients receiving betablocker therapy with tight heart rate control? J Am Coll Cardiol 2006;48:964 –9 14. Feringa HH, Bax JJ, Boersma E, Kertai MD, Meij SH, Galal W, Schouten O, Thomson IR, Klootwijk P, van Sambeek MR, Klein J, Poldermans D. High-dose beta-blockers and tight heart rate control reduce myocardial ischemia and troponin T release in vascular surgery patients. Circulation 2006;114:I344 –9 15. Badner NH, Knill RL, Brown JE, Novick TV, Gelb AW. Myocardial infarction after noncardiac surgery. Anesthesiology 1998;88:572– 8 16. Kop WJ, Verdino RJ, Gottdiener JS, O’Leary ST, Bairey Merz CN, Krantz DS. Changes in heart rate and heart rate variability before ambulatory ischemic events (1). J Am Coll Cardiol 2001;38:742–9 17. Bayliff CD, Massel DR, Inculet RI, Malthaner RA, Quinton SD, Powell FS, Kennedy RS. Propranolol for the prevention of postoperative arrhythmias in general thoracic surgery. Ann Thorac Surg 1999;67:182– 6 18. Brady AR, Gibbs JS, Greenhalgh RM, Powell JT, Sydes MR. POBBLE trial investigators: perioperative beta-blockade (POBBLE) for patients undergoing infrarenal vascular surgery: results of a randomized double-blind controlled trial. J Vasc Surg 2005;41:602–9 19. Juul AB, Wetterslev J, Gluud C, Kofoed-Enevoldsen A, Jensen G, Callesen T, Norgaard P, Fruergaard K, Bestle M, Vedelsdal R, Miran A, Jacobsen J, Roed J, Mortensen MB, Jorgensen L, Jorgensen J, Rovsing ML, Petersen PL, Pott F, Haas M, Albret R, Nielsen LL, Johansson G, Stjernholm P, Molgaard Y, Foss NB, Elkjaer J, Dehlie B, Boysen K, Zaric D, Munksgaard A, Madsen JB, Oberg B, Khanykin B, Blemmer T, Yndgaard S, Perko G, Wang LP, Winkel P, Hilden J, Jensen P, Salas N; DIPOM Trial Group. Effect of perioperative beta-blockade in patients with diabetes undergoing major noncardiac surgery: randomised placebo controlled, blinded multicentre trial. BMJ 2006;332:1482 20. Yang H, Raymer K, Butler R, Parlow J, Roberts R. The effects of perioperative beta-blockade: results of the metoprolol after vascular surgery (MaVS) study, a randomized controlled trial. Am Heart J 2006;152:983–90 © 2008 International Anesthesia Research Society

1047

21. Raby KE, Brull SJ, Timimi F, Akhtar S, Rosenbaum S, Naimi C, Whittemore AD. The effect of heart rate control on myocardial ischemia among high-risk patients after vascular surgery. Anesth Analg 1999;88:477– 82 22. Jakobsen CJ, Bille S, Ahlburg P, Rybro L, Hjortholm K, Andresen EB. Perioperative metoprolol reduces the frequency of atrial fibrillation after thoracotomy for lung resection. J Cardiothorac Vasc Anesth 1997;11:746 –51 23. Hopkins SR, Bogaard HJ, Niizeki K, Yamaya Y, Ziegler MG, Wagner PD. Beta-adrenergic or parasympathetic inhibition, heart rate and cardiac output during normoxic and acute hypoxic exercise in humans. J Physiol 2003;550:605–16 24. Rinfret S, Abrahamowicz M, Tu J, Humphries K, Eisenberg MJ, Richard H, Austin PC, Pilote L. A population-based analysis of the class effect of beta-blockers after myocardial infarction. Am Heart J 2007;153:224 –30 25. Wijeysundera DN, Beattie WS. Calcium channel blockers for reducing cardiac morbidity after noncardiac surgery: a metaanalysis. Anesth Analg 2003;97:634 – 41 26. Harwood TN, Butterworth J, Prielipp RC, Royster RL, Hansen K, Plonk G, Dean R. The safety and effectiveness of esmolol in the perioperative period in patients undergoing abdominal aortic surgery. J Cardiothorac Vasc Anesth 1999;13:555– 61 27. Lanfear DE, Jones PG, Marsh S, Cresci S, McLeod HL, Spertus JA. Beta2-adrenergic receptor genotype and survival among patients receiving beta-blocker therapy after an acute coronary syndrome. JAMA 2005;294:1526 –33

1048

Rate Control Reduces Postoperative Myocardial Infarction

28. POISE Trial Investigators, Devereaux PJ, Yang H, Guyatt GH, Leslie K, Villar JC, Monteri VM, Choi P, Giles JW, Yusuf S. Rationale, design, and organization of the PeriOperative ISchemic evaluation (POISE) trial: a randomized controlled trial of metoprolol versus placebo in patients undergoing noncardiac surgery. Am Heart J 2006;152:223–30 29. Redelmeier D, Scales D, Kopp A. Beta blockers for elective surgery in elderly patients: population based, retrospective cohort study. BMJ 2005;331:932 30. Kardos A, Long V, Bryant J, Singh J, Sleight P, Casadei B. Lipophilic versus hydrophilic beta(1) blockers and the cardiac sympatho-vagal balance during stress and daily activity in patients after acute myocardial infarction. Heart 1998;79:153– 60 31. Wijeysundera DN, Naik JS, Beattie WS. Alpha-2 adrenergic agonists to prevent perioperative cardiovascular complications: a meta-analysis. Am J Med 2003;114:742–52 32. Beattie WS, Badner NH, Choi P. Epidural analgesia reduces postoperative myocardial infarction: a meta-analysis. Anesth Analg 2001;93:853–58 34. Dufour Doiron M, Prud’homme D, Boulay P. Time-of-day variation in cardiovascular response to maximal exercise testing in coronary heart disease patients taking a beta-blocker. Appl Physiol Nutr Metab 2007;32:664 –9

ANESTHESIA & ANALGESIA