|Slideshow | Pitfalls of the Current Bleeding Definitions|
|Sunil V. Rao, MD
Assistant Professor of Medicine
Duke University Medical Center
Durham VA Medical Center
Duke Clinical Research Institute
Washington, D.C.—Due to the variety of bleeding definitions used in clinical trials and the potential various outcomes on patients and clinical practice from different bleeding thresholds, Sunil V. Rao, MD, from Duke University Medical Center in Durham, N.C., argued that the time has come to arrive at a consensus definition of bleeding, at the Cardiovascular Research Technologies conference on Feb. 21.
There are significant differences in bleeding definitions across all the acute coronary syndromes trials, said Rao, who asked whether it was the patient, the trial or the definition that was the difference.
He compared the difference in major bleeding rates for PURSUIT (1.5 percent) versus CURE (3.7 percent) versus SYNERGY (9.1 percent) versus TRITON (2.4 percent), noting that the difference in definition significantly impacts the reported incidence of bleeding.
While Rao acknowledged that “statistical purists” discourage cross-trial comparisons, clinicians are forced to weigh all the available data—including the risks and benefits—for their daily practice.
He also noted that there seems to be a “clear association” between outcomes and bleeding. “A variety of different outcomes seems to be correlated—but not causal unless bleeding is severe—among even mild bleeding to subsequent mortality,” he explained. “This is true according to the definitions of bleeds in TIMI, CURE, ACUITY.”
There have not been many trials that have directly compared two different bleeding definitions and their correlation to subsequent outcomes. “Primarily that’s because a lot of trials use a variety of definitions to capture the same bleeding events” Rao said. He added that definitions have become more aligned.
In comparing the outcomes in the 15,858 patients enrolled in PURSUIT and PARAGON B trials, there was a stepwise increase in outcomes (30-day death/MI) as GUSTO bleeding definition worsened, Rao explained.
He referenced a trial by Mehran et al, presented at ESC.09, which compiled a pooled data set of all the bivalirudin trials. In these trials, TIMI major bleeds led to a greater incidence of death within one year, driven mainly by transfusion. In this study, large hematomas (5 cm or greater) did not seem to correlate with one-year mortality.
However, Rao said that many have questioned whether hematomas should be considered a major bleed. In fact, White et al in a 2009 American Heart Journal study determined that hematomas are better classified as minor bleeds. Along the same lines, recent trials have sought to assess the importance and impact on nuisance bleeds. The major caveat to nuisance bleeds, according to Rao, is regardless of how they are classified, they often lead to patients discontinuing their dual-antiplatelet therapy—which leads to an increased risk of stent thrombosis.
“As a result, bleeding seems to have downstream effects on how we treat patients,” he said.
Currently, bleeding definitions are constructed from various data elements, including clinical (intracranial hemorrhage and hematoma); laboratory (Hgb decrease); and consequences (hemodynamic compromise, transfusion and fatal). These contribute to the variety of severity classifications.
Bleeding definitions are designed to “assess the bleeding risk of a therapeutic strategy, but the “risk” can be dialed up or down depending on the definition,” said Rao. “As a result, the definitions should allow clinicians to weigh the risks and benefits of one therapy against another.”
According to Rao, the pitfalls of the current bleeding definitions are:
- Little overlap in data elements;
- Some “major” definitions are considered “minor” by other scales and vice versa; and
- Risk of “gaming” safety data.
Although many bleeding definitions appear to be associated with mortality, very few studies have directly compared two definitions in terms of prognosis.
Rao argued for a universal bleeding definition because “many definitions are associated with adverse outcomes. While the data elements are important, many of the definitions do not overlap.” Rao also said that “minor” bleeding appears important in terms of medication adherence, and some studies define minor bleeds as major bleeds.
He concluded that the time has come for a consensus definition, which he hoped would be established at