Methods
To assess differences in gastrointestinal blood loss, we measured quantitative fecal hemoglobin equivalents (HemoQuant, Mayo Medical Laboratory, Rochester, Minn) in 117 patients, mean age 71 years, 1 month after initiation of assigned therapy in the Stroke Prevention in Atrial Fibrillation III Study. Sixty-three of these patients who had characteristics for high risk of stroke were randomly assigned to conventional adjusted-dose warfarin therapy (international normalized ratio, 2.0 to 3.0) or low-dose combined therapy (warfarin [international normalized ratio, <1.5] plus 325 mg/d of entericcoated aspirin). The remaining 54 patients with low risk of stroke received 325 mg/d of enteric-coated aspirin.

Results
Among the 63 patients at high risk of stroke, abnormal values (>2 mg of hemoglobin per gram of stool) were detected in 11% and values greater than 4 mg of hemoglobin per gram of stool were found in 8%. Mean (±SD) values were more for those randomly assigned to receive combined therapy (1.7±3.3 mg of hemoglobin per gram of stool vs adjusted-dose warfarin therapy, 1.0± 1.9 mg/g; P=.003). The 54 nonrandomized patients with low risk of stroke receiving aspirin alone had a mean (±SD) Hemo-Quant value of 0.8±0.7 mg of hemoglobin per gram of stool 1 month after entry in the study.

Conclusions
Abnormal levels of fecal hemoglobin excretion were common in elderly patients with high risk of atrial fibrillation 1 month after randomization to prophylactic antithrombotic therapy. Combined warfarin and aspirin therapy was associated with greater fecal hemoglobin excretion than standard warfarin therapy, suggesting the potential for increased gastrointestinal hemorrhage.

(Arch Intern Med. 1996;156:658-660)