Current issues of ACP Journal Club are published in Annals of Internal Medicine

Clinical Prediction Guide

A 5-point scoring system predicted bleeding in patients treated with warfarin

ACP J Club. 1999 Jan-Feb;130:25. doi:10.7326/ACPJC-1999-130-1-025

Source Citation

Beyth RJ, Quinn LM, Landefeld S. Prospective evaluation of an index for predicting the risk of major bleeding in outpatients treated with warfarin. Am J Med. 1998 Aug;105:91-9.



How accurate and useful is a 5-point index (Outpatient Bleeding Risk Index [OBRI]) for estimating the risk for major bleeding in patients who are treated with warfarin?


2 cohort studies—1 for derivation and 1 for validation of the index results.


University hospitals in Boston, Massachusetts, USA (derivation), and Cleveland, Ohio, USA (validation).


Data from 556 patients (mean age 61 y, 53% women, 93% white) were used for derivation of the index and from 264 patients (mean age 60 y, 53% women, 70% white) for validation. All patients started to receive warfarin at the time of hospital discharge and were monitored by their primary physician. Duration of follow-up was ≥ 2 years, and 98% of patients were accounted for.

Description of prediction guide

The OBRI was developed to divide patients into 3 risk groups (low, intermediate, or high) for bleeding. 1 point was given for each of the following variables: age ≥ 65 years, history of stroke, history of gastrointestinal bleeding, or a specific comorbid condition (recent myocardial infarction, serum creatinine level > 1.5 mg/dL, hematocrit < 0.3, or diabetes mellitus). Low-risk patients had a score of 0, intermediate-risk patients had a score of 1 or 2, and high-risk patients had a score of 3 or 4.

Main outcome measure

Major bleeding (≥ 2 units of blood loss in ≤ 7 d or life-threatening bleeding).

Main results

Major bleeding occurred in 12% of patients in the derivation cohort (cumulative incidence 7% at 1 mo, 11% at 12 mo, and 22% at 48 mo) and 8% of patients in the validation cohort (cumulative incidence 2% at 1 mo, 8% at 12 mo, and 12% at 48 mo). The validation cohort had 80 patients in the low-risk group, 166 in the intermediate-risk group, and 18 in the high-risk group. The risk for major bleeding for patients in the low-, intermediate-, and high-risk groups was, respectively, 3%, 8%, and 30% at 12 months and 3%, 12%, and 53% at 48 months. Physicians estimated the probability of major bleeding no better than that estimated by chance. 11 of the 18 patients who were classified as high risk and who had major bleeding had an elevated international normalized ratio (INR) (mean for the 11 patients 12.2).


The Outpatient Bleeding Risk Index classified patients who had started receiving warfarin into low-, intermediate-, or high-risk groups for major bleeding. The index performed better than primary care physicians. A substantial proportion of bleeding was accompanied by excessive anticoagulation.

Sources of funding: In part, National Institutes of Health; Claude D. Pepper Older Americans Independent Center; American Federation for Aging Research.

For correspondence: Dr. R.J. Beyth, Division of General Internal Medicine and Health Care Research, Case Western Reserve University, School of Medicine, WG29B, 10900 Euclid Avenue, Cleveland, OH 44106-4961, USA. FAX 216-368-0737.


In this well-done, validated study by Beyth and colleagues, patients who received anti-coagulation agents and who were determined to be at high risk by the OBRI clearly had more bleeding episodes than did patients at intermediate or low risk. Although 2 parameters of the index (history of stroke and gastrointestinal bleeding) make intuitive sense, the others (older age and comorbid conditions) are less clear. Patients bleed for other reasons aside from old age or heart disease. This assertion is supported by the facts that high-risk patients had bleeding episodes while their warfarin dose was too high, while they were taking nonsteroidal anti-inflammatory drugs (NSAIDs), and while they were having invasive procedures without appropriate correction of coagulopathy before the procedure. Therefore, the OBRI seems to predict therapeutic misadventure as much as an underlying propensity to bleed. Further support for this interpretation might come from knowledge of how many low- and intermediate-risk patients were given NSAIDs or had an INR outside of the desirable therapeutic range (data were not available in the paper).

Consecutive patients were studied, and follow-up was by chart review. A protocol with ongoing follow-up was not part of the study, and management was left to individual physicians. This might explain why the rates of bleeding with anticoagulation are lower in most recent randomized controlled trials, many of which use low-intensity anticoagulation with close attention to the INR.

Consider proceeding in the following manner when anticoagulation is being considered for a high-risk patient. First, reassess whether the benefits of anticoagulation outweigh the potential harms; second, treat potential bleeding lesions (e.g., peptic ulcers); third, use prophylactic therapy for lesions that cannot be definitively treated; fourth, avoid medications known to affect the prothrombin time or other clotting factors; and fifth, carefully monitor the INR, regardless of the predictions of the OBRI.

Walter Peterson, MD
Veterans Affairs Medical CenterDallas, Texas, USA