We appreciate Dr. Fiore's comments but believe that he has misinterpreted our data and conclusions. Although he focuses on "an astonishingly high absolute annual risk" for life-threatening and fatal hemorrhage in patients 80 years of age and older, he seems to have overlooked our strong rejoinder that this result must be "interpreted with great caution" because the numbers of patients and events were relatively small. Even though our study sample was larger than those in almost all other published studies, the confidence limits around the risk estimate were wide, ranging from 1.3 to 15.6. Excessive concern for this finding should not divert attention from the fact that we did not find any consistent, age-related trend in risk for bleeding.

We strongly disagree with the contention that our method for classifying complications of bleeding was flawed or manipulated in any manner after the study was initiated. We used the same system in many earlier reports and explained its rationale [1-4]. The major difference between our method and the traditional designations cited by Dr. Fiore is that we have added a fourth category for less-severe complications of bleeding that are typically deemed trivial and not counted by other investigators. Such events occur in as many as 60% of patients, have a clear negative influence on patients' well-being, and should not be dismissed as trivial [5]. The other three categories we use correspond closely to the traditional designations of minor, major, and fatal, with the exception that life-threatening events are restricted to the more-severe complications. In this report, we elected to combine life-threatening and fatal events because of the small number of events in each category and because it was sensible from a clinical standpoint. We believe that combining serious with life-threatening events, as Dr. Fiore suggests, would have misrepresented the data.

Despite his objections, Dr. Fiore's conclusion is the same as ours: Elderly patients who are maintained at a level of intensity of anticoagulation similar to that of younger patients do not seem to be at a greater overall risk for hemorrhagic complications. On the basis of our data, however, the possibility that very elderly persons may be at risk for more severe manifestations of bleeding must be recognized. Moreover, we concur wholeheartedly with Dr. Fiore that elderly candidates for warfarin therapy should not be denied this clearly beneficial therapy on the basis of age alone and that additional prospective studies on this topic are mandated.