Fihn and colleagues [1] have further fueled the controversy about how much, if at all, increasing age contributes to the risk for hemorrhage in patients who are treated with oral anticoagulant agents. It is difficult to reconcile the nearly contradictory conclusions of their recent study that "age did not appear to be an important determinant of risk for bleeding" but that "the single exception was life-threatening and fatal complications in patients 80 years of age or older (relative risk, 4.6)." The authors report that octogenarians have an astonishingly high absolute annual risk of 3.38% for life-threatening and fatal hemorrhage.

A methodologic flaw of considerable concern in this report is the departure from the traditional designation of hemorrhagic episodes as minor, major, or fatal [2]. The substitution of "minor," "serious," and "life-threatening or fatal" categories for severity appears arbitrary and has no established precedent in the literature. It is also unclear whether these definitions of hemorrhagic episodes were established before data extraction or collection or were derived after the complete data set had been obtained. Finally, the small number of life-threatening events and the admittedly difficult task of determining the severity of complications in the retrospective aspect of the trial further challenge the validity of the conclusions. It is apparent from Table 2 of the paper that no compelling evidence for an increased rate or severity of bleeding would be evident for octogenarians if "life-threatening" complications had been grouped with "serious" bleeding events instead of "fatal" bleeding events.

The consequence of arbitrarily redefining categories of bleeding is an overemphasis of the risk for hemorrhagic episodes in octogenarians. This distraction overshadows the primary findings of the study: that intensity of anticoagulation, not increasing age, is the major determinant of anticoagulant-related bleeding. The potential ramifications of this study are far reaching when one considers that 10% of patients older than 80 years of age are candidates for warfarin for the prevention of stroke in atrial fibrillation [3]. Clinicians who treat such patients should not withhold anticoagulant agents because of a heightened concern about "life-threatening" bleeding until additional prospective studies have confirmed or refuted these results.