To the Editor.
—Brown and colleagues1 study of antibiotics and bleeding is astonishing in its methodology and conclusions. The authors make much of the fact that a total of 1493 patients were laboriously followed. Nevertheless, their claims regarding the results of the use of moxalactam are made on the basis of 18 patient histories with a total of two occurrences of decreased hemoglobin concentration and a single patient with increased prothrombin/ partial thromboplastin time. Will readers who get beyond the headlinelike title consider this clinically or statistically relevant?More importantly, Brown and colleagues claim to have controlled for patient "conditions" (nutritional status, etc) and "treatments" (use of anticoagulants, vitamin K) that "could have been associated with bleeding." How can conditions and treatments be meaningfully controlled for without knowledge of why and how patients were treated? Specifically, does not the need for and/or the performance of surgery influence various