Earlier this week, I discussed the preliminary results of the POISE trial, the blockbuster that showed that perioperative beta blockers may cause more harm than good. I've asked my UCSF colleague Andy Auerbach, one of the nation's experts on this intervention, to help us understand these truly surprising results. Andy's comments follow:
"The POISE trial vividly demonstrates that giving beta blockers to surgical patients is not a simple intervention. That is, perioperative beta blockade (at a fairly high dose – 200 mg metoprolol/day ) produces a mix of risks (stroke, significant hypotension, bradycardia, plus that pesky all-cause mortality) that outweighs the benefit (largely, fewer non-fatal MIs).
In drilling a bit deeper into the POISE results, it appears the excess mortality was fueled by a far higher mortality rate in beta blocked patients who also had a non-cardiac complication (largely sepsis). Because non-cardiac post-op complications are more common than cardiac ones, it is easy to see how this double-whammy (beta blocker + sepsis) might produce POISE’s troubling results.
Interestingly, the concerns about beta blockers aren’t new: a trend toward higher stroke risk in beta blocked patients was also seen in a previous study, but the strokes were too infrequent to be analyzed with any accuracy. Other complications (such as bradycardia and hypotension) were also common in prior studies, but not clinically problematic. Taken together, since the excess risk of strokes wasn’t statistically significant, and the bradycardia and hypotensive episodes were self-limited and benign, these prior studies were taken as supporting the safety of beta blockers rather than as a harbinger of problems to come.
Where to go from here? At the least, it seems clear that beta blockers need to be used far less often, and when used we need to pay closer attention to patients (e.g., assess pain, infection, and hypovolemia before increasing beta blockers) after surgery. It also seems unsafe to be starting beta blockers right before surgery. Data from the Netherlands suggest that, in high risk patients, starting them at least 2 weeks beforehand and titrating them carefully is safe. Once we figure out how to operationalize this practice, this might be an effective strategy. For now, it is uncertain.
Did we endorse beta blockers as a practice and as a quality measure too early? Initial studies were relatively small, but they did show salutary effects on perioperative ischemia, MI, and death that closely paralleled beta-blockers’ known benefits seen in large RCTs from other settings (acute coronary syndromes, hypertension). So I agree with Bob – it wasn’t unreasonable to include them as a quality measure at the time. We were wrong, but at least we are in good company (can anybody say estrogen replacement therapy?).
It will be interesting to see how the field reacts to this change in evidence. Many experts have encouraged us to wait for the publication of a large RCT from the Netherlands (DECREASE V), expected in 2008, before drawing final conclusions about perioperative beta-blockers, but to my eye this seems unsafe given the POISE results.
Beta blockers are also the subject of public reporting initiatives (e.g., SCIP), but the SCIP-endorsed measure (maintaining beta blockers throughout the surgical period for patients on them already) is a reasonably safe standard and (dare I say) something we are doing as part of medication reconciliation. At the moment, I don’t think the SCIP beta blocker measure needs to be changed. This will bear close watching after POISE’s full results are published and DECREASE becomes available.
How is my practice today different than it was before last week? I’ll continue the beta blockers for my patient who was on them previously, I’m more likely to recommend starting beta blockers a couple of weeks before surgery in high risk patients, and I’m less likely to prescribe perioperative beta blockers in the intermediate risk pre-op patient (in whom I might now perform some testing to better define the true risk). I’m also less likely to use a high dose to start, or to titrate up the beta blocker dose for tachycardia unless I am certain that I’m not treating tachycardia due to hypovolemia, bleeding, or infection first. Finally, I’ll do all I can to prevent noncardiac postop complications.
And I’m more likely to look at quality measurement and practice guidelines with the slightly jaundiced eye of the clinical researcher, recognizing that, from time to time, even the best science will turn out to be completely wrong, or at least partly off base."