"Beta blockers should not be routinely used for perioperative treatment of patients undergoing noncardiac surgery unless patients are already taking them for clinically indicated reasons," say Bangalore and colleagues in their paper. "The ACC/AHA guidelines committee should soften their stance on perioperative beta blockade until definitive evidence shows clear benefit."

But not everyone agrees with these conclusions. Dr Lee Fleisher (University of Pennsylvania, Philadelphia), who is on the ACC/AHA committee that writes the guidelines for perioperative beta blockade use, told heartwire: "My greatest fear from all of this is that people will not use beta blockers at all." He added that a focused update of the guidelines is under way. "We have not ignored it; we have to look at the actual data. It's coming, that's all I'm allowed to say."

And Dr Martin London (University of California, San Francisco) added: "I would say that the opinion of the expert anesthesiologists I've interacted with . . . is that widespread use of beta-blocker protocols in relatively low-risk patients is going to distinctly decline. However, chronic use will continue to be maintained, and for sure, anesthesiologists who have liberally used beta blockers for many years at times of well-defined hemodynamic stress will and should continue to do so."

My greatest fear from all of this is that people will not use beta blockers at all.

In the meta-analysis, Bangalore et al included 33 trials with a total of 12 036 patients, including both the POISE trial—a landmark trial first reported last year in which there was a 33% increased risk of all-cause mortality and a 117% increased risk of stroke but a 30% reduction in nonfatal MI in the continued-release metoprolol arm—and the DECREASE study, which showed a decreased risk of all-cause mortality and nonfatal MI with a very different regimen of bisoprolol in a similar setting.

In this updated meta-analysis, beta blockers were not associated with any significant reduction in the risk of all-cause mortality, cardiovascular mortality, or heart failure but were associated with a 35% decrease in nonfatal MI. For the overall cohort, Bangalore et al estimate that treatment of 1000 patients with beta blockers resulted in 16 fewer nonfatal MIs in survivors, but at the expense of three nonfatal disabling strokes, 45 patients with clinically significant perioperative bradycardia, 59 with hypotension, and "potentially" increased mortality.

And as well as declaring that the totality of evidence "does not support the use of beta-blocker therapy for the prevention of perioperative clinical outcomes in patients having noncardiac surgery," Bangalore et al say that "use of perioperative beta blockade as a performance measure, when there is not robust evidence for improved outcome, is inappropriate."

But Fleisher told heartwire this is an "inaccurate" statement. "The only performance measure that the US Surgical Care Improvement Project ever endorsed was the continuation of beta blockers in patients already taking these agents," he noted, adding, "This recommendation has never been in question; no trial has demonstrated otherwise."

In a comment accompanying publication of the new meta-analysis [2], Drs Eric Boersma and Don Poldermans (Erasmus University, Rotterdam, the Netherlands) say it is skewed by the POISE results. London agrees, but also adds that the DECREASE studies—for which Poldermans is the lead investigator—also bias the meta-analysis in the opposite direction.

London commented to heartwire: "The meta-analysis by Bangalore et al is a major addition to the existing prior meta-analyses in several respects. First, it incorporates the data reported by the very large multinational POISE study. Second, it stratifies the studies into high and low risk for bias based on quality assessment of the trials, and third, it reports a series of interesting subanalyses on covariates of obvious clinical interest when deciding on the potential efficacy of beta blockers.

The meta-analysis by Bangalore et al is a major addition to the existing prior meta-analyses in several respects.

"Poldermans's DECREASE study is one that accounts for a substantial statistical impact in the high-bias-risk group, driving the majority of benefit with the least amount of risk. This contrasts with the lower-bias-risk studies, the bulk of which is accounted for by the POISE study, which reports a lesser degree of efficacy associated with significant risk. Both of these studies have come under substantial criticism for study design issues," he adds.

And Fleisher commented: "In this new meta-analysis, the most important thing that was brought up as a limitation was that the risks and benefits and the protocols used in the studies included are so different and that that is critical as part of the interpretation.

"What this meta-analysis did show, particularly as it was so heavily weighted toward POISE, is that standard protocols without titration do have significant risks," Fleisher added. "But it's important that people do not ignore tachycardia in the perioperative period, and beta blockers are very appropriate for treatment. If a patient is tachycardic, I would use a beta blocker, but I would be cautious about a protocol that didn't titrate."

Poldermans agrees: "Heart-rate control is mandatory," he told heartwire. "We use a different approach [from POISE]: start early before surgery and adjust the dose if necessary to prevent overtreatment during surgery, bradycardia, and eventually hypotension. This might be the reason why the POISE data—which were favorable for cardiac events—showed an increase in stroke."

What this meta-analysis did show . . . is that standard protocols without titration do have significant risks.

"The [beta-blocker] dose used in the DECREASE trials is only 10% of the maximum prescribed daily dose, in contrast to POISE, which goes up to 100% of the maximal prescribed daily dose," he added. "In these cases it is better to add beta blockers perioperatively than to start with too high a dose, because there is no way back," he added.

Poldermans also presented data on stroke risk in the DECREASE trials overall at the AHA meeting, which showed that his titration scheme, using a low dose of bisoprolol, "was safe, reduced cardiac events, and was not associated with an increased risk of stroke," he said.

Nevertheless, in their comment, Boersma and Poldermans state that this new meta-analysis is "methodologically sound" and concede that it "has emphasized that risk of stroke might be a serious issue." However, they point out that a mechanism to explain this stroke risk has not been revealed.

"To a large extent this is true," says London, "but there is tangible evidence . . . that decreasing hemoglobin postoperatively in the presence of beta blockade has serious adverse consequences. If confirmed, this will be a major stumbling block for perioperative beta blockade, as blood loss is part and parcel of the perioperative period, and it is often difficult to rapidly assess and treat such changes." There is also growing evidence for elevated risk for stroke in patients with certain genetic polymorphisms of the beta-adrenergic receptor, he notes.

The implications for immediate clinical management are quite complex. This is unlikely to be resolved by this analysis.

"Thus, I think that experts in this field need to now think much more expansively about this topic rather than simply rehashing heart rate and blood pressure from a variety of older studies," London adds. "The implications for immediate clinical management are quite complex. I know that the AHA group is working on updated guidelines, but I don't know how they will react and certainly would hope they have access to this meta-analysis but also consider these newer concepts that I've outlined above."

Poldermans said: "We call for people to report the hemodynamics of the patients who experienced adverse events, so the medical community can learn how to use this medication sensibly, to the benefit of the patients."

London concludes: "Despite the elegant statistical analyses performed we are still left in a polarized environment of clinicians —and experts—who have biases toward either of the two studies [POISE and DECREASE] and approaches. This is unlikely to be resolved by this analysis."

Senior author of the Lancet paper, Dr Franz Messerli (St Luke's-Roosevelt Hospital, New York), told heartwire: "We are not pretending that our analysis resolves all the issues surrounding perioperative beta blockade. Clearly, there remains uncertainty as to patient characteristics and dose and time of initiation of beta blockade. However, we strongly reject the unambiguous recommendations of the ACC/AHA guidelines. In view of our analysis, these guidelines violate the principle of primum non nocere."