When Evidence Goes Bad

Posted by The Evidence Blog on August 13, 2013

There’s been much discussion about how to improve the clinical guideline process, especially when authors may have a financial conflict of interest. But what if patients were inadvertently harmed by a clinical guideline because evidence supporting the guideline was based on unethical research?

That’s the dilemma faced now by many European cardiac surgeons as they attempt to protect the heart during noncardiac-related surgery by prescribing beta-blockers just prior to the procedure. This practice is endorsed by the 2009 European Society of Cardiology (ESC) guidelines (and somewhat less aggressively by the American College of Cardiology Foundation/American Heart Association [ACCF/AHA] guidelines), so what’s the problem? Well, apparently the research on which the guidelines are based were discredited almost 2 years ago, and the results were made public shortly thereafter. But until now, there’s been no real effort to change the guidelines.

The ESC based its 2009 guidelines on evidence from both the Dutch DECREASE trials, which appeared to strongly support perioperative beta-blockade and the POISE trial, which raised concerns that beta-blockers might lead to an increased mortality. In an analysis of all the data, the ESC committee found a neutral effect on mortality but a strong benefit due to significant reductions in nonfatal myocardial infarction (MI) and stroke in the patient who took the beta blockers—due primarily to the results from the DECREASE trials.

And from here, things get interesting. In 2011, the principal investigator of the DECREASE trials was found guilty of several breaches of scientific integrity related to all but one of the individual DECREASE trials. These included protocol gaps, bias, fabrication of data, and submitting studies for publication knowing that the data could not be trusted. And in a remarkable irony, this same researcher had also been the chairman of the ESC committee that wrote the original guideline back in 2009.
Fast forward to 2013. The ESC guidelines and the AHA guidelines are still in place, despite the fact that the basis for the ESC guidelines has been publicly denounced as fraudulent in January 2o12.

A group of UK researchers is now trying to move things forward. A new meta-analysis published in Heart reexamined the original data without the original DECREASE trials. The findings now look significantly different from the original analysis on which the guidelines were based. In a combined population that excluded the DECREASE trials, the patients randomized to beta-blockers had a 27% increase in the risk of death (relative risk [RR], 1.27; confidence interval [CI], 1.01-1.60; P=0.04). In contrast, 2 randomized placebo-controlled DECREASE studies excluded from this new analysis showed a 58% reduction in mortality associated with beta-blockers (although this was not statistically significant). The new Heart analysis did find some benefits associated with beta-blockade, including a statistically significant reduction in nonfatal MI. However, it also found a significant increase in hypotension and stroke. The authors of the new meta-analysis believe that these beneficial effects are clearly overshadowed by the most important finding of an increased mortality risk. While there is insufficient data to estimate the number of patients that may have been harmed in the UK, the authors approximate that up to 10,000 patients may have died as a result of adherence to this guideline.

The authors conclude that “Patient safety being paramount, guidelines for perioperative beta-blocker initiation should be retracted without further delay. Routine initiation of β-blockers for this indication should not be recommended, except in the context of RCTs that should be designed carefully, conducted honestly and reported truthfully.” They also suggest that future guidelines be accompanied by a commitment from all those named as authors to retract them immediately if the advice given is later revealed to be harmful.

The ESC Practice Guidelines Committee is currently revising the guidelines, with a new version due in early 2014. Good for them, but what took so long?

Topics: Hayes Blog

Sign up to receive updates from our blog

Our latest articles

New Call-to-action