Dr Poldermans was particularly critical of POISE [24, 25]. The debate could have continued for a long time but in 2011 Erasmus University Medical Center dismissed Dr Poldermans because of scientific misconduct [31], as several of the studies from his group were considered potentially unsound [32]. Yet, to this day, no paper has been formally retracted. Dr Poldermans rejected the case made against him [33]. More recent meta-analyses have excluded the papers from Dr Poldermans’ group as they were considered insecure [18, 34, 35]. The meta-analysis by Bouri and colleagues highlighted the increased risk of death where beta-blockade is initiated preoperatively and asked for guidelines to be urgently rewritten [34]. Shortly after the on-line publication of this meta-analysis a joint statement was published by ACC/AHA and ESC [35]. It emphasised that beta-blockers should only be given perioperatively after a careful evaluation of risks and benefits in individual patients.

Several recent meta-analyses have tried to identify groups of patients likely to benefit from perioperative initiation of beta-blockade, drawn from the available RCTs, often including the “insecure trials”. For example the benefits of beta-blockade were seen only in studies with high risk of bias [36]. Benefits were seen in studies with up-titration of beta-blockade [36], as the best control of heart rate provided the most benefits [37], but with the caveat of risk of side effects such as bradycardia and congestive heart failure [37]. However, another meta-analysis did not find an association between heart rate control and outcome [38]. Separating high risk patients from moderate- and low-risk patients shows that all-cause mortality is significantly increased in 10,367 low and medium risk patients, while there is a trend towards lower mortality in 1,096 high risk patients [18].

Despite all the controversy surrounding the introduction of a beta-blocker at the time of surgery, there was (and still is) unanimity on the need to continue the chronic administration of beta-blockers perioperatively based on observational studies, some very large. There is also evidence that stopping chronic beta-blocker therapy is harmful, causing a three-fold increase in the risk of adverse cardiovascular events [21–23].

How long before surgery should beta-blockade be started is still unclear. Various durations have been used, or proposed, ranging from the day of surgery as in POISE [6] and most previous RCTs, to 30 days or more in some studies [2, 39]. While previous guidelines recommended initiating beta-blockade, where indicated, 7 to 30 days before surgery [19, 20], in the absence of clear evidence to support a particular duration, the latest guidelines state that treatment should be started at least one day before surgery [40], or, when possible at least 1 week and up to 30 days before surgery, starting with a low dose [41]. This reflects a study using population-based administrative databases by Wijeysundera and colleagues [42]. They analysed data on a cohort of 48,103 patients. The majority (92 %) were on long-term beta-blockade (more than 31 days), 5.5 % had taken a beta-blocker for 8 to 30 days, and 2.3 % for 1 to 7 days. Compared with those treated for more than 31 days, patients in the 1 to 7 days group had a significantly increased 30-day mortality (OR 1.49; CI 1.03–2.6; p = 003). However there was no association with increased non-fatal myocardial infarction, stroke, and one-year mortality. This suggests that starting beta-blockade only one week before surgery may not be sufficient to reduce the risk of adverse events. Indeed, based on a meta-analysis of 8 RCTs, Dai and colleagues found that starting beta-blockade more than 7 days before surgery improved outcomes [43], but their analysis was not concerned with starting treatment more than 30 days before surgery.

Because of the risks uncovered by POISE, several authors have tried to consider the type of surgery and the level of cardiovascular risk that would identify patients likely to benefit from perioperative beta-blockade. In a retrospective cohort evaluation of over 136,000 patients, just over 55,000 were exposed to beta-blockers at the time of surgery, mostly patients undergoing vascular surgery. Exposure to beta-blockade was much higher in patients with 4 or more risk factors according to the revised cardiac risk index (RCRI [44]) [45]. In the propensity-matched cohort, exposure to beta-blockade was associated with a significantly reduced 30-day risk of death. The relative risk was much lower in the higher RCRI classes than in those with no or only one risk factor who showed no significant benefit. This confirmed the findings of a previous study by Lindenauer and colleagues [46]. In over 780,000 patients, 122,000 of whom received a beta-blocker, beta-blockade was protective in those with RCRI 2 and above, while it could cause harm in those with RCRI 0 or 1. Importantly and unexpectedly, the study by London and colleagues showed no benefits of beta-blockade in patients undergoing vascular surgery, yet, intuitively, such patients may have been expected to benefit most from beta-blockade [45]. This is important data as previous guidelines recommended initiating beta-blockade in patients undergoing vascular surgery [19]. Finally, a large Danish Nationwide cohort study with over 28,000 patients showed that perioperative beta-blockade benefits only patients with heart failure or recent myocardial infarction [47]. Thus analyses of observational studies to identify patients likely to benefit from beta-blockade rather than be put at risk are likely to continue.

The question of the best beta-blocker for perioperative protection is debated. In the study by London and colleagues atenolol was associated with better outcomes than metoprolol (the drug used in POISE) [45]. This confirms data from Wallace and colleagues [48] who showed greater protection by atenolol than metoprolol in a cohort study with nearly 39,000 Veterans Association Hospitals patients. Of the beta-blockers used in RCTs, only metoprolol is associated with increased all-cause mortality in 10,027 patients [18]. Atenolol appears to reduce mortality and myocardial infarction when compared to metoprolol [43]. This is not surprising since POISE used metoprolol, found an increased mortality, and the study represents the vast majority of the patients in any meta-analysis. Another single-centre cohort study with over 44,000 [49] showed that the risk of perioperative stroke was lower in patients receiving bisoprolol than atenolol or metoprolol. However, in most studies patients were receiving long-term treatment with beta-blockers, thus the conclusions may not apply to the initiation of beta-blockade prior to surgery.

The recent meta-analysis by Guay and Ochroch [50] was based on 27 RCTs including those by Mangano et al. [1], Poldermans et al. [2], and Dunkelgrun et al. [39] even though these studies have been criticised for their methodology [1], or concerns about scientific integrity [2, 39]. The authors concluded that addition of a beta-blocker did not increase the risk of 30-day mortality, yet the data showed 167 deaths in the beta-blocked patients and 146 in the placebo group (158 deaths and, respectively, 126 deaths if the criticised studies are excluded). These differences are not statistically significant but demonstrate an adverse trend. The authors found a significantly reduced mortality at one year in those given a beta-blocker (14 deaths vs 28 in the placebo group) but most of deaths (7 vs 16) were in the study by Mangano et al. [1]. The latter was criticised because it allowed patients on long-term beta-blocker therapy to be randomised to placebo, thus generating a group of patients with beta-blocker withdrawal, a known risk factor. Moreover all-cause mortality during the administration of atenolol (or placebo) was not included in the final analysis (4 in the atenolol and 2 in the placebo groups). Thus the data was intrinsically biased in favour of beta-blockade. If this study is excluded then the difference between 7 and 12 deaths in the remaining 3 studies is not significant. Unsurprisingly the meta-analysis confirmed the increased risk of stroke and the reduction of the risk of non-fatal myocardial infarction. Based on three studies, the latest published in 1983, and a total of 97 patients, the authors found no effect of stopping beta-blockers preoperatively but add that the small numbers “preclude drawing any conclusions” [50]. In view of advances in the management of patients with coronary artery disease over the last 25 years a meta-analysis based on three studies [51–53] published more than 30 years ago is unlikely to be relevant to current practice.

The latest meta-analysis by Neng and colleagues [43] shows that in eight RCTs including 11,180 patients, perioperative atenolol was associated with lower mortality. Similarly starting treatment more than one week before surgery was beneficial. Unfortunately this meta-analysis includes one of the insecure studies from Dr Poldermans’ group [39].

As the body of evidence from observational studies is quite large, it is worth mentioning the systematic review and meta-analysis of cohort studies by Wan and colleagues [54]. Eight cohort studies were included with a total of 470,000 participants of whom 180,000 were in the beta-blocker group. Perioperative beta-blockers were not associated with significantly reduced risks of mortality (RR 0.88; CI 0.75–1.04), postoperative myocardial infarction (RR 1.30; CI 0.76–2.33), and postoperative stroke (RR 1.17; CI 0.53–2.57). With such large number of patients the absence of cardiac protection, and the known risks seen in RCTs, indicates that a thorough risk/benefit analysis is needed before initiating beta-blockade preoperatively.

As the controversy surrounding POISE was settling new guidelines were prepared and they deserve to be discussed in detail.

In August 2014, the American College of Cardiology, the American Heart Association, and the American Society of Anesthesiologists (ACC/AHA) published their new guidelines on the management of patients with heart disease undergoing non-cardiac surgery [40]. At the same time the European Society of Cardiology and the European Society of Anesthesiology (ESC/ESA) published theirs [41]. Both comment extensively on perioperative beta-blockade, with some differences in their approach (Table 7.2).

Both sets of guidelines recommend maintaining long-term beta-blocker therapy during the perioperative period (Class I recommendation, Evidence level B). The ACC/AHA guideline adds that the management of beta-blockade after surgery should be guided by clinical circumstances independent of when the agent was started (IIa B). The ESC/ESA guideline recommends initiating beta-blockers in patients scheduled for high-risk surgery who have at least two risk factors or are ASA 3, and in those with known ischaemic heart disease or myocardial ischaemia (Class IIb B). The ACC/AHA considers it reasonable to begin beta-blockers before surgery in patient with 3 or more RCRI risk factors (IIb B), and in patients with high-risk myocardial ischaemia in preoperative tests (IIb C). Thus there are no longer Class I indications for starting beta-blocker therapy before non-cardiac surgery.

If there are compelling long-term indications for beta-blocker therapy, it may seem legitimate to institute treatment before elective surgery. However, the ACC/AHA guideline considers that, in the absence of other RCRI risk factors, initiating beta-blockers to reduce the perioperative risk is of uncertain benefit (IIb B). Moreover, at variance with previous guidelines, vascular surgery is not mentioned as a particular reason for initiating beta-blockade.

As in 2009, guidelines recommend commencing beta-blockers, where indicated, far enough in advance of surgery to assess safety and tolerability and preferably more than one day before surgery (ACC/AHA 2014, IIb B). Without including this as a formal recommendation, the ESC/ESA guideline states that, ideally, treatment, where indicated, should be started between 30 and 2 days before surgery; in addition the target heart rate should be 60–70 bpm and systolic arterial blood pressure more than 100 mmHg. By contrast the ACC/AHA guideline does not indicate targets for titration because of insufficient evidence.

The ESC/ESA guideline recommends, when oral beta-blockade is initiated, using atenolol or bisoprolol as the first choice (IIb B). By contrast, the ACC/AHA guideline makes no specific recommendation.

Both sets of guidelines include class III recommendations, highlighting the risk of harm: beta-blockers should not be started the day of surgery (ACC/AHA, III B). High-dose without titration is not recommended (ESC/ESA III B), and initiation of beta-blockers before low risk surgery is not recommended (ESC/ESA III B).

What the new guidelines highlight again is that continuing long-term beta-blockers is important as there is evidence, in the perioperative and in the non-operative setting, that abrupt withdrawal carries significant risks [21, 22].

By comparison with the 2009 recommendations, it is in the ESC/ESA guideline that the changes relevant to beta-blockers are the most striking. This is not surprising as in 2009 the ESC/ESA guideline there was much emphasis on the results of the DECREASE studies [2, 39] from Dr Poldermans’ group, leading the panel to promote perioperative beta-blockade despite the risks highlighted by POISE.

In the light of recent evidence, the new guidelines recommend that only patients at substantial cardiac risk undergoing intermediate or high-risk surgery should be considered for initiation of perioperative beta-blockade. If indicated it makes sense to start treatment well ahead or surgery to assess its safety and adjust the dose as a function of the patient’s response. While the ESC/ESA guideline indicates targets for heart rate and blood pressure, these are not based on strong evidence. Indeed it has been shown that a low target for heart rate is associated with cardiac protection but carries the risk of excessive bradycardia, congestive heart failure [37] and hypotension. A target blood pressure higher than 100 mmHg before the next dose of beta-blocker is recommended, yet it is not evidence-based. It was the limit in POISE and it can be argued that the limit was too low. In high cardiac risk patients 100 mmHg systolic blood pressure, as one size fits-all, is difficult to justify. Even today many patients present with arterial hypertension that is not fully controlled to less that 140 mmHg systolic and less that 80 mmHg diastolic blood pressure. The latest JNC (JNC8)[55] proposes 150 mmHg as the goal for patients above the age of 65 in daily clinical practice. In such patients is it legitimate to give another dose of a beta-blocker when their systolic blood pressure is only 100 mmHg? One could easily argue that it is not legitimate and that the target blood pressure should take into consideration the patient’s everyday blood pressure. This is why the ACC/AHA guideline does not recommend particular targets for heart rate and blood pressure, but draws attention to the need to prevent hypotension.

As the indications for using beta-blockers to protect the heart against perioperative cardiovascular complications of non-cardiac surgery are now very limited, it may be appropriate to consider briefly what is known in cardiac surgery, based on a recent Cochrane Collaboration review of 53 randomised controlled trials [18]. The meta-analysis showed that beta-blockade had no influence on all-cause mortality, acute myocardial infarction, and cerebrovascular accidents, based on studies including between 1700 and 3983 patients. For other adverse events such as myocardial ischaemia, bradycardia, hypotension and congestive heart failure beta-blockade did not influence these outcomes, but the number of patients included in the available studies was low (between 166 and 660). Therefore firm conclusions cannot be drawn. On the positive side, beta-blockade reduced the risk of ventricular and supra-ventricular arrhythmias and reduced the length of hospital stay in studies with between 2292 and 6420 patients. Thus, the role of beta-blockade in cardiac surgery is relatively limited but at least there is no evidence of increased all-cause mortality.

Alpha₂-adrenoceptor agonists reduce post-ganglionic noradrenaline output, thereby reducing the catecholamine output surge during the perioperative period. Thus they could be expected to protect the heart. Over the years several studies have shown benefits of this class of drugs. Studies used clonidine, dexmetedomidine, or mivazerol. The largest trial was the European mivazerol trial. It randomized 1,897 patients with ischaemic heart disease who underwent intermediate- or high-risk surgery. It is only in the subpopulation of 904 patients undergoing vascular surgery that mivazerol significantly reduced the risk of postoperative death or myocardial infarction [58]. As protection was seen only in patients undergoing vascular surgery, mivazerol was not commercialized.

A meta-analysis comprising 23 trials enrolling a total of 3,395 patients who were randomised to receive clonidine, dexmedetomidine, mivazerol or a placebo showed overall both mortality [RR 0.64 (CI 0.42-0.99)] and myocardial ischaemia [RR 0.76 (CI 0.63–0.91)] to be significantly reduced. Mortality and myocardial infarction were reduced in vascular surgical patients, and mortality in cardiac surgery patients [59]. In 2004 a relatively small RCT with190 patients showed clonidine to reduce perioperative myocardial ischaemia, and to reduce long-term mortality over the two years following surgery [60]. In view of the relatively limited evidence, based mostly on small trials, with the exception of the mivazerol trial, a large randomized controlled trial was needed.