Cohn et al1 concluded that ongoing diverticular bleeding after 4-unit blood transfusion requires subtotal colectomy without localization by angiography. Their conclusion was based on record review, including 4 successful subtotal colectomies, and on review of the literature. By the same devices, we reached the same conclusion in 1972.2
In 20 years of testing this hypothesis, I have learned that record and literature reviews are deceptive because they miss nondiverticular bleeds that, while actively bleeding, would have followed an algorithm to an inappropriate and failed operation. In a subsequent report,3 I excluded without details 22 patients who underwent transfusion for lower gastrointestinal tract bleeding, were studied, discharged with diagnoses of diverticulosis, and later proved to have bled from other sources. Ten cases exceeded the 4-unit threshold and might have had useless colectomies: 2 for missed varices, 2 for ulcers, 2 for atriovenous malformations with coagulopathy, 3 for colonic polyps, and 1 for ulcerative colitis. More impressive than these unreported numbers is the memory of patients undergoing blind subtotal colectomies in which surgeons missed bleeders that rebled; several of these patients died of complications of ineffective operations.