Abstract
Acute perioperative beta-adrenergic blockade may be cardioprotective in the high-risk cardiac patient for major non-cardiac surgery. We have investigated the association between the heart rate achieved with perioperative beta-blockade and the incidence of perioperative cardiac complications.
We identified eight randomized studies (1931 patients) reporting acute perioperative beta-blockade and major perioperative cardiovascular outcomes after non-cardiac surgery. The mean heart rates within the first 72 h after operation were analysed. A meta-analysis of means was conducted using a random effects model. A bivariate correlation analysis was conducted using Spearman’s correlation coefficient to assess for an association between the mean postoperative heart rate and the 30 day cardiac outcomes.
Acute perioperative beta-blockade did not significantly reduce 30 day cardiac death [odds ratio (OR) 0.35, 95% confidence interval (CI) 0.08–1.52] or non-fatal myocardial infarction (OR 0.90, 95% CI 0.52–1.56) in the studies with adequate methodology. The mean (95% CI) heart rate was 73 (71–74) beats min−1 in the beta-blockade group, which was significantly lower than the placebo group (mean heart rate 82, P=0.0001). There was no correlation between heart rate and 30 day cardiac complications (P=0.848). The reduction in heart rate was associated with increased drug-associated adverse events (OR 2.53, 95% CI 2.05–3.13, P<0.0001). A major limitation of this analysis may be that postoperative heart rate was not a primary outcome in any of the studies identified and the mean postoperative heart rate achieved may be too high to realize optimal cardioprotection.
This meta-analysis cannot confirm that heart rate control with beta-adrenergic blockade is cardioprotective. A randomized controlled trial examining the effect of tight perioperative heart rate control with beta-adrenergic blockade on clinically important outcomes and adverse events is warranted.
Acute perioperative beta-adrenergic blockade (beta-blockade) may be cardioprotective in the perioperative period, although this is controversial.1 It has also been shown to be cost-effective in patients undergoing major non-cardiac surgery where the control major cardiovascular complication rate exceeds 10% within 30 days of surgery in the control group.2 However, acute beta-blockade is also associated with a significant increase in bradycardia and hypotension needing treatment.1 The American College of Cardiology/American Heart Association (ACC/AHA) guidelines for the cardiovascular evaluation of patients undergoing non-cardiac surgery recommend that, where appropriate, the dose of beta-blocker is titrated to a resting heart rate of 50–60 beats min−1 with the aim of maintaining an intraoperative and postoperative heart rate of <80 beats min−1.3 However, this recommendation is not evidence-based, as the single study quoted when making this recommendation4 was not blinded, selected a very high-risk subgroup of patients, had unclear randomization, and was stopped early.5
Methods
The aim of this study was to determine if there was an association between the reported mean heart rates within the first 72 h after surgery and the incidence of 30 day cardiac death and non-fatal myocardial infarction in both the beta-blocked and the placebo groups.
In order to evaluate the effect of acute perioperative beta-blockade on heart rate in patients likely to benefit from acute perioperative beta-blockade, we have analysed the heart rate data of all the perioperative beta-blockade studies reporting major cardiovascular complications, identified in five recent (2005–07) systematic reviews.1,5–8 These systematic reviews have, between them, searched at least 11 large databases up to October 2005. They identified at least 5000 potential studies, of which 29 studies reporting on acute perioperative beta-blockade for non-cardiac surgery were identified for inclusion in these reviews.1,5–8 From these, we identified the studies that reported major cardiovascular complications of either cardiac mortality or non-fatal myocardial infarction for inclusion in our analysis of heart rate and cardiovascular outcome.
The data abstracted from the identified trials included the number of patients randomized to beta-blockade or placebo/control group (intention to treat), the drug regimen including the duration of treatment, the mean heart rate and standard deviation in each group, and the time at which the heart rate was recorded. In studies which recorded heart rate on a number of occasions, we have recorded the heart rate in the early postoperative period (up to 72 h after operation), as it is during this period that prolonged myocardial ischaemia is most likely to result in myocardial infarction.9 Where the data required were not available in the individual papers, the authors were contacted in order to obtain the relevant missing data. Each randomized trial was evaluated for study quality and assigned a Jadad score (0–5) in order to measure the likelihood of bias.10 Low quality trials are considered to have a score of ≤2.
Data from the different studies were collated in binary form with reference to drug therapy and outcome. To correct for differences in sample size of the various studies, the weighted odds ratio (OR) was calculated with a weighting proportional to the total size of the study. The results were analysed to calculate the OR, 95% confidence intervals (CI), and two-tailed P-values using RevMan version 4.2.8 software (The Nordic Cochrane Centre, Copenhagen, Denmark). Drug-associated adverse events were calculated using Epicalc 2000 Version 1.2 software (Brixton Health, UK).
A meta-analysis of means was conducted using a random effects model when determining the postoperative mean heart rate for the beta-blockade and the placebo groups. Statistical analysis was conducted with NCSS Statistical Software (NCSS, Utah, USA).
We analysed the reported mean postoperative heart rates of the beta-blockade and placebo patients in the included studies, and compared these data with the incidences of the complications of cardiac death and non-fatal myocardial infarction within 30 days post-operation. A bivariate correlation analysis was conducted using Spearman’s correlation coefficient to assess if there was an association between the mean postoperative heart rate and the combined 30 day cardiac outcome of cardiac death and non-fatal myocardial infarction. The Statistical Package for Social Sciences (SPSS) 13.0 for Windows (September 1, 2004) was used for this analysis.
We used the ‘test of heterogeneity’ and the I2-statistic to describe the proportion of variability due to heterogeneity between the studies. An I2-statistic of 25% was considered a cut-off between low and moderate heterogeneity between the studies.8
Results
Of the 29 studies of acute perioperative beta-blockade, only 10 reported major cardiovascular events.4,11–18 Two papers were excluded as the required heart rate data were unavailable from the manuscript and from the authors.15,19 Thus, data from eight studies (n=1931 patients, age range 40–89) were included in the analysis (Table 1). Three of these studies had a Jadad score of <3.4,13,14 Drug therapy in the studies varied, being atenolol (2), metoprolol (4), bisoprolol (1), and a postoperative infusion of esmolol in the remaining study. The duration of treatment ranged between 48 h and 30 days post-operation.
Randomized studies of acute perioperative beta-blockade included in the analysis of mean postoperative heart rate and associated cardiac outcome. HR, heart rate; SVT, supraventricular tachycardia; VT, ventricular tachycardia; MI, myocardial infarction; CHF, congestive heart failure
| First author, year | Target heart rate | Number of patients | Mean heart rates (sd) | Time of heart rate recording | Type of surgery | Outcome measured | Jadad score10 | ||
|---|---|---|---|---|---|---|---|---|---|
| Beta-blocker | Placebo | Beta-blocker | Placebo | ||||||
| Jakobsen, 199711 | No target | 18 | 17 | 75 (9) | 85 (17) | First 18 h postop | Lung resection | Bradycardia, hypotension, MI, SVT, VT | 3 |
| Wallace, 199812 | 55<HR<65 | 99 | 101 | 77.3 (11.4) | 87.4 (13.4) | First 48 h postop | Various non-cardiac surgery | Mortality, MI, ischaemia, cerebrovascular, VT, length of stay | 5 |
| Poldermans, 19994 | 50<HR<80 postoperatively | 59 | 53 | 71 (9.3) | 82 (9.7) | First 24 h postop | Major vascular surgery | Mortality, MI | 2 |
| Raby, 199913 | 60<HR<77 (20% below ischaemic threshold) | 15 | 11 | 72 (10) | 80 (13) | First 48 h postop | Major vascular | MI, ischaemia | 1 |
| Zaugg, 199914 | 55<HR<80 intraoperatively | 23 | 20 | 63 (14) | 82 (10) | First 24 h postop | Non-cardiac surgery | MI | 2 |
| Yang, 200616 | HR>50 awake | 247 | 250 | 69 (12.9) | 79 (15.7) | Postoperative | Vascular surgery | Mortality, MI, ischaemia, CHF, VT, AF | 4 |
| Brady, 200517 | HR>50 | 53 | 44 | 74.8 (15.3) | 81.9 (13.9) | Up to 72 h after test dose | Vascular surgery | MI, cerebrovascular, ischaemia, VT, length of stay | 5 |
| Juul, 200618 | HR>55 | 462 | 459 | 74 (12) | 83 (14) | First 24 h postop | Orthopaedic and intra-abdominal surgery | Mortality, MI, ischaemia, heart failure | 5 |
| First author, year | Target heart rate | Number of patients | Mean heart rates (sd) | Time of heart rate recording | Type of surgery | Outcome measured | Jadad score10 | ||
|---|---|---|---|---|---|---|---|---|---|
| Beta-blocker | Placebo | Beta-blocker | Placebo | ||||||
| Jakobsen, 199711 | No target | 18 | 17 | 75 (9) | 85 (17) | First 18 h postop | Lung resection | Bradycardia, hypotension, MI, SVT, VT | 3 |
| Wallace, 199812 | 55<HR<65 | 99 | 101 | 77.3 (11.4) | 87.4 (13.4) | First 48 h postop | Various non-cardiac surgery | Mortality, MI, ischaemia, cerebrovascular, VT, length of stay | 5 |
| Poldermans, 19994 | 50<HR<80 postoperatively | 59 | 53 | 71 (9.3) | 82 (9.7) | First 24 h postop | Major vascular surgery | Mortality, MI | 2 |
| Raby, 199913 | 60<HR<77 (20% below ischaemic threshold) | 15 | 11 | 72 (10) | 80 (13) | First 48 h postop | Major vascular | MI, ischaemia | 1 |
| Zaugg, 199914 | 55<HR<80 intraoperatively | 23 | 20 | 63 (14) | 82 (10) | First 24 h postop | Non-cardiac surgery | MI | 2 |
| Yang, 200616 | HR>50 awake | 247 | 250 | 69 (12.9) | 79 (15.7) | Postoperative | Vascular surgery | Mortality, MI, ischaemia, CHF, VT, AF | 4 |
| Brady, 200517 | HR>50 | 53 | 44 | 74.8 (15.3) | 81.9 (13.9) | Up to 72 h after test dose | Vascular surgery | MI, cerebrovascular, ischaemia, VT, length of stay | 5 |
| Juul, 200618 | HR>55 | 462 | 459 | 74 (12) | 83 (14) | First 24 h postop | Orthopaedic and intra-abdominal surgery | Mortality, MI, ischaemia, heart failure | 5 |
Randomized studies of acute perioperative beta-blockade included in the analysis of mean postoperative heart rate and associated cardiac outcome. HR, heart rate; SVT, supraventricular tachycardia; VT, ventricular tachycardia; MI, myocardial infarction; CHF, congestive heart failure
| First author, year | Target heart rate | Number of patients | Mean heart rates (sd) | Time of heart rate recording | Type of surgery | Outcome measured | Jadad score10 | ||
|---|---|---|---|---|---|---|---|---|---|
| Beta-blocker | Placebo | Beta-blocker | Placebo | ||||||
| Jakobsen, 199711 | No target | 18 | 17 | 75 (9) | 85 (17) | First 18 h postop | Lung resection | Bradycardia, hypotension, MI, SVT, VT | 3 |
| Wallace, 199812 | 55<HR<65 | 99 | 101 | 77.3 (11.4) | 87.4 (13.4) | First 48 h postop | Various non-cardiac surgery | Mortality, MI, ischaemia, cerebrovascular, VT, length of stay | 5 |
| Poldermans, 19994 | 50<HR<80 postoperatively | 59 | 53 | 71 (9.3) | 82 (9.7) | First 24 h postop | Major vascular surgery | Mortality, MI | 2 |
| Raby, 199913 | 60<HR<77 (20% below ischaemic threshold) | 15 | 11 | 72 (10) | 80 (13) | First 48 h postop | Major vascular | MI, ischaemia | 1 |
| Zaugg, 199914 | 55<HR<80 intraoperatively | 23 | 20 | 63 (14) | 82 (10) | First 24 h postop | Non-cardiac surgery | MI | 2 |
| Yang, 200616 | HR>50 awake | 247 | 250 | 69 (12.9) | 79 (15.7) | Postoperative | Vascular surgery | Mortality, MI, ischaemia, CHF, VT, AF | 4 |
| Brady, 200517 | HR>50 | 53 | 44 | 74.8 (15.3) | 81.9 (13.9) | Up to 72 h after test dose | Vascular surgery | MI, cerebrovascular, ischaemia, VT, length of stay | 5 |
| Juul, 200618 | HR>55 | 462 | 459 | 74 (12) | 83 (14) | First 24 h postop | Orthopaedic and intra-abdominal surgery | Mortality, MI, ischaemia, heart failure | 5 |
| First author, year | Target heart rate | Number of patients | Mean heart rates (sd) | Time of heart rate recording | Type of surgery | Outcome measured | Jadad score10 | ||
|---|---|---|---|---|---|---|---|---|---|
| Beta-blocker | Placebo | Beta-blocker | Placebo | ||||||
| Jakobsen, 199711 | No target | 18 | 17 | 75 (9) | 85 (17) | First 18 h postop | Lung resection | Bradycardia, hypotension, MI, SVT, VT | 3 |
| Wallace, 199812 | 55<HR<65 | 99 | 101 | 77.3 (11.4) | 87.4 (13.4) | First 48 h postop | Various non-cardiac surgery | Mortality, MI, ischaemia, cerebrovascular, VT, length of stay | 5 |
| Poldermans, 19994 | 50<HR<80 postoperatively | 59 | 53 | 71 (9.3) | 82 (9.7) | First 24 h postop | Major vascular surgery | Mortality, MI | 2 |
| Raby, 199913 | 60<HR<77 (20% below ischaemic threshold) | 15 | 11 | 72 (10) | 80 (13) | First 48 h postop | Major vascular | MI, ischaemia | 1 |
| Zaugg, 199914 | 55<HR<80 intraoperatively | 23 | 20 | 63 (14) | 82 (10) | First 24 h postop | Non-cardiac surgery | MI | 2 |
| Yang, 200616 | HR>50 awake | 247 | 250 | 69 (12.9) | 79 (15.7) | Postoperative | Vascular surgery | Mortality, MI, ischaemia, CHF, VT, AF | 4 |
| Brady, 200517 | HR>50 | 53 | 44 | 74.8 (15.3) | 81.9 (13.9) | Up to 72 h after test dose | Vascular surgery | MI, cerebrovascular, ischaemia, VT, length of stay | 5 |
| Juul, 200618 | HR>55 | 462 | 459 | 74 (12) | 83 (14) | First 24 h postop | Orthopaedic and intra-abdominal surgery | Mortality, MI, ischaemia, heart failure | 5 |
The clinical outcome data of all eight studies included for 30 day cardiac mortality (Figs 1 and 2) and for 30 day non-fatal myocardial infarction (Figs 3 and 4) appear to favour treatment. However, when the studies with inadequate methodology are removed, neither 30 day cardiac mortality nor 30 day non-fatal myocardial infarction shows significant difference between the beta-blockade and the placebo groups (Figs 2 and 4, respectively). Despite adequate methodology, the study by Juul and colleagues18 is not included in the meta-analysis of 30 day cardiac mortality as these data are unavailable. The all-cause mortality at 30-days in this study was 19 of 462 patients randomized to beta-blockade and 14 of 459 patients randomized to placebo (A. Juul, personal communication).
Thirty day cardiac mortality.
Thirty day cardiac mortality in studies with adequate methodology.
Thirty day non-fatal myocardial infarction.
Thirty day non-fatal myocardial infarction in studies with adequate methodology.
Three of the eight studies reported bradycardia and hypotension needing treatment.16–18 The beta-blockade patients had a significantly increased rate of bradycardia and hypotension needing treatment (53% and 42%, respectively) with an OR and 95% CI of 2.53 and (2.05–3.13), P<0.0001.
The time at which the postoperative heart rate value was recorded varied between that in the post-anaesthesia care unit16 and a mean value over 18–72 h after operation. The weighted mean heart rate recorded for patients receiving acute beta-blockade was 10 (95% CI 8–11) beats min−1 lower than the control group (73 compared with 82 beats min−1) (Fig. 5).
Forest plot of mean heart rate difference in beta-blocked vs placebo groups in randomized controlled perioperative studies.
A Spearman’s correlation comparing heart rate and 30 day cardiovascular complications (a composite of cardiac death or non-fatal myocardial infarction) was conducted for the beta-blockade group, the placebo group, and for both groups compared. There was no significant association between mean postoperative heart rate and 30 day cardiac complications (P=0.848, P=0.478, and P=0.264) for beta-blockade, placebo, and combined beta-blocker and placebo groups, respectively (Fig. 6).
Scatter plot of mean postoperative heart rates and combined 30 day cardiac mortality and non-fatal myocardial infarction.
Only the 30 day non-fatal myocardial infarction (Fig. 3) had an I2-statistic of >25%, but with exclusion of trials with a Jadad score of <3, the remaining trials do not have significant heterogeneity (Fig. 4). With respect to the mean postoperative heart rates, a radial plot was generated for all the included studies. A single study was out of the limits for homogeneity.14
Discussion
The cardioprotective effects of beta-blockade are multifactorial, of which some are heart rate mediated.20 A reduction in heart rate improves myocardial oxygen balance and decreases mechanical stress on vulnerable plaques and the associated inflammation leading to plaque rupture.21 However, the heart rate theoretically necessary to realize these cardioprotective benefits is unknown. Clinical studies and physiological principles suggest that a heart rate <75 beats min−1 is desirable, as this would maximize diastolic time,22 and in patients with coronary artery disease it should be below the ischaemic threshold and the resting heart rate associated with plaque rupture during daily living.21,23
However, despite the 95% CI of the mean postoperative heart rate in the beta-blockade group in this study being <75 beats min−1, this analysis has not shown a correlation between heart rate and 30 day postoperative cardiac complications (Figs 2 and 4).
There may be a number of possible explanations for this. First, the combined sample size of all the studies identified is underpowered to show a 25% reduction in the combined 30 day outcome of cardiac death and non-fatal myocardial infarction. The optimal sample size based on the event rate in the placebo group for the combined endpoint is nearly 2900 patients per group for a power of 80% and a two-sided α of 0.05, to show a 25% reduction in the combined endpoint. Secondly, a maximum postoperative heart rate was only defined in the protocol of three studies.4,13,14 All three studies had a maximum heart rate target >75 beats min−1. This inability to control the maximum postoperative heart rate may have adversely affected postoperative cardiac outcome in the beta-blocked group.
Thirdly, a major limitation of this analysis is that it is not possible to relate a specific heart rate to a specific perioperative cardiac outcome, but rather we are limited to group analysis. It is possible that the beta-blockade patients, who sustained perioperative cardiac events, may have had a higher heart rate than those patients who had no perioperative cardiac events.24,25 If this was the case, then a lower target heart rate than that presented here may be appropriate. Another major limitation of this analysis is that postoperative heart rate was not a primary outcome variable in these studies. Five of the eight studies analysed did not specifically target a decrease in postoperative heart rate, and the three studies that did target a decrease in postoperative heart rate, had an upper limit which may have been too high to produce cardiac protection.4,13,14 Indeed, there is evidence to suggest that an even more aggressive reduction of heart rate (than the suggested 75 beats min−1) may further improve cardiac outcome. This has recently been shown in an observational cohort study of 272 vascular patients, where a multivariate analysis showed that increasing beta-blockade dose (per 10% increase) and increasing postoperative heart rate (per 10 beats min−1 increase) were independent predictors of postoperative mortality [P=0.008, hazard ratio (HR) 0.86, 95% CI 0.76–0.97] and (P<0.001, HR 1.45, 95% CI 1.16–1.67), respectively.25
A prospective study of preoperative cardiac testing in intermediate-risk vascular surgery patients administered beta-blockade to all patients in the study, with a preoperative and postoperative heart rate target of 60–65 beats min−1. Patients who had a preoperative heart rate of <65 beats min−1 had significantly less 30 day cardiac death and non-fatal myocardial infarction (P=0.003, OR 0.24, 95% CI 0.09–0.66).24 The incidence of the primary endpoint increased by a factor of 1.5, for every 5 beats min−1 increase above 65.24 However, this heart rate target was not achieved in 16.5% of the patients.24
These studies suggest that an exponential relationship may exist between an increase in perioperative heart rate and cardiac complications. Our analysis, however, could not assess this relationship, as some of the included studies had groups with no reported cardiac complications,11,13,20 which does not allow for exponential curve estimation. Exclusion of these studies may increase bias, as one study had no cardiac complications in the control group.11
We have shown that acute perioperative beta-blockade increases the incidence of profound bradycardia and hypotension, which would probably increase in studies which have tighter heart rate control. The impact of these adverse events on outcome would have to be assessed.
Finally, the inability of this analysis to show any cardioprotective protection of perioperative beta-blockade may be complicated by the inclusion of patients at lower cardiac risk. We have previously suggested that a higher cardiac event rate may be required to show benefit associated with perioperative beta-blockade.26
Our analysis of the relationship between acute beta-blockade, mean postoperative heart rate, and cardiac outcome at 30 days cannot confirm the value of tight postoperative heart rate control. For this reason, a well-conducted, randomized controlled trial examining the effect of tight perioperative heart rate control with beta-blockers on clinically important outcomes and adverse events is warranted.
Acknowledgements
We thank the following authors for access to their data and clarification of their studies; Drs Bayliff, Poldermans, Raby, Urban, Wallace, Yang, Zaugg and Drs Powell and Sydes on behalf of the POBBLE Trial Management Group.
