β-Blockade before Noncardiac Surgery: The Devil Is in the Details

In this issue of Anesthesiology, McKenzie et al.1 address the question: “Is β-blocker initiation within 60 days (but more than 1 day) of major abdominal surgery associated with an increase in postoperative stroke?” They did so by using the Truven database, which provided a large U.S.-based dataset of mostly non-Medicare patients. Their question builds on more than 20 yr of controversy and changing opinion about the role of β-blockers initiated preoperatively and continued throughout the perioperative period.Of the more than 100 million adults who undergo inpatient, noncardiac surgery globally each year,2 2% develop major cardiac complications and 13% develop prognostically important myocardial injury, both of which have been shown to negatively impact patients’ short- and long-term outcomes.3 Based on the belief that postoperative cardiac events were caused by the sympathetic response to surgery, a small, single-center randomized controlled trial initiated 5 mg oral bisoprolol in high-risk patients 1 week before vascular surgery and found that the primary study endpoint of cardiac death or nonfatal myocardial infarction occurred in 3.4% of patients in the bisoprolol group and 34% of patients in the standard-care group.4 This finding was corroborated in a large, retrospective cohort study evaluating 782,969 patients undergoing major noncardiac surgery that observed a reduction in 30-day mortality in patients who had received β-blockers in the 2 days preoperatively, with greater benefit conferred to those at greater risk.5 In response, the 2007 American College of Cardiology, and American Heart Association Task Force perioperative guidelines issued varying levels of recommendation (according to type of surgery) that preoperative β-blockade be used to prevent adverse postoperative cardiovascular events.6 However, in 2008, the 8,351 patient PeriOperative ISchemic Evaluation (POISE) trial found that, although preoperative initiation of a β-blocker 2 to 4 h before noncardiac surgery resulted in a 16% decrease in the risk of nonfatal myocardial infarction, it also more than doubled the risk of stroke, such that patients who received metoprolol had a 33% increase in the risk of death.7 The increase in stroke and death with metoprolol was presumed to be because patients in that arm had a 55% increase in the incidence of clinically important hypotension, which was defined as a systolic blood pressure less than 90 mmHg that required fluid resuscitation, intra-aortic balloon pump, an inotropic agent, or study drug discontinuation.Since the clear signal of overall harm identified in POISE, multiple observational studies examining perioperative β-blockade have suggested that the true effect of β-blockers during the perioperative period is probably a lot more complicated,8 with recent guideline recommendations reflecting this complexity.9In contrast to POISE, McKenzie et al. used retrospective administrative data to examine the outcomes of 240,981 patients undergoing major abdominal surgery managed in routine care and did not identify an association between preoperative β-blocker initiation and the risk of postoperative stroke or major cardiovascular events. Taken together, the results of these two studies serve to highlight the questions facing perioperative clinicians that randomized controlled trials have not yet answered, as well as the relative roles of large randomized controlled trials and population-based registry research in answering them.Large, well-conducted randomized controlled trials generate results that allow clinicians to draw conclusions about causal relationships between studied interventions and outcomes. When a trial is large and well executed, the process of randomization ensures that the only difference between treatment arms is their exposure to the intervention of interest. This may not be the case for smaller or poorly executed randomized controlled trials that do not adequately ensure allocation concealment and blinding or fail to use transparent and well-documented study protocols and procedures. When well-executed, large randomized controlled trials have excellent internal validity, i.e., the extent to which a study establishes a trustworthy cause-and-effect relationship between an intervention and an outcome. In interpreting the results of a randomized controlled trial, one must consider the specifics of the research question being posed, the patient to whom they will be applied, and the quality with which the randomized controlled trial has been executed. Well-executed randomized controlled trials do an excellent job at answering the questions they are intended to answer. However, the questions answered by a randomized controlled trial are not always generalizable to real-life clinical practices and patient cohorts. The extent to which the question posed by a randomized controlled trial deviates from routine practice limits its external validity, i.e., the ability to generalize the results of that randomized controlled trial to routine care. The POISE trial randomly assigned β-blocker–naive patients to receive 100 mg oral metoprolol 2 to 4 h in advance of surgery, a higher-than-normal initial dose timed more closely to surgery than usual. Although POISE left no doubt that initiating 100 mg oral metoprolol in β-blocker–naive patients immediately before noncardiac surgery causes an increase in postoperative stroke and mortality, it is unclear that these results can be extrapolated to lower or titrated doses of β-blocker, different timing of β-blocker initiation, or even different β-blockers. In other words, POISE had high internal validity and limited external validity. Nonetheless, POISE was a landmark trial that brought the serious issue of hypotension during the perioperative period to the attention of clinical researchers and provided a cautionary tale warning of the risks of widespread adoption of practices based on limited evidence.In contrast, large, registry-based population studies retrospectively examine routinely collected data and, rather than randomization, use advanced statistical techniques to control for known and measured confounders and achieve balanced risk between treatment arms. Because the data analyzed in registry-based studies have been obtained as part of routine care, the questions posed and results obtained usually have high external validity, although this depends on the extent to which the registry that is sampled represents the general population. However, because registry-based observational studies are limited to measured confounders available in healthcare records, the statistical techniques used to achieve a balance of risk between treatment arms are unable to control for unmeasured or unknown confounders. As such, registry-based designs have limited internal validity, which decreases the ability to make causal inferences. The strength of registry-based observational studies is that, because they reflect patterns of health and disease in routine clinical care, and often include very large numbers of patients, they can be used to explore hypotheses that can be tested in subsequent randomized controlled trials. The work of McKenzie et al. tells us that the initiation of β-blockers in usual care at more than 1 and less than 60 days before major abdominal surgery is likely neither associated with an increased risk of stroke nor a decreased risk of postoperative cardiac events, although its relationship with clinically important hypotension remains unclear. However, to confidently assert the absence of a causal relationship requires a large, multicenter randomized controlled trial, and substantial investment of time, financial resources, and expertise. As such, a randomized controlled trial addressing this specific question may or may not ever be done.Where then does this leave clinicians regarding which noncardiac surgery patients to give β-blocker medications to, how much β-blocker to give, and how to time β-blocker administration in relation to surgery? McKenzie et al. studied a multicenter sample that included mostly young, non-Medicare patients; only about 15% were more than 65 yr old. Within this sample, 22% of adults undergoing major abdominal surgery were chronically prescribed β-blockers and an additional 2% had β-blockers initiated within 60 days of surgery, presumably for the purposes of primary or secondary prevention of cardiovascular disease. Given the national, geographically diverse, and multicenter nature of their sample, rigorous attempts to control for confounders, and lack of a strong signal of harm, initiation of β-blockers within 60 days of major abdominal surgery may be safe in similar nonelderly populations. However, given the absence of a strong signal of benefit, it remains unclear whether β-blockers should be initiated before major abdominal surgery in patients with cardiovascular risk factors as a perioperative risk mitigation strategy. β-Blockers, when used during the perioperative period, are neither poison nor panacea. In deciding how to approach their use, perioperative clinicians are left to make patient-centered decisions that consider potential benefits and harms in light of the body of evidence and patient in front of them. To ensure that we provide safe and well-informed care, anesthesiologists and other perioperative practitioners need to strive to create a culture of evidence-based medicine that supports the conduct of robust clinical trials. As a profession, we need to learn to appraise the quality of evidence being generated, avoid blindly adopting practice based on a smattering of proof, and demand well-executed trials that address the questions essential to inform our day-to-day practice.Dr. Spence is supported by a Clinician Scientist Award from the Canadian Anesthesia Research Foundation and an IARS Mentored Research Award.The authors are not supported by, nor maintain any financial interest in, any commercial activity that may be associated with the topic of this article.