To the Editor.
—Lowenstein et al1 recently described 40 cases of new-onset atrial fibrillation admitted to Denver General Hospital from the emergency room. They concluded that alcohol caused or contributed to 35% of the cases. Although there does seem to be some theoretical support for their proposition, there are methodological problems in their observational report that make the results uninterpretable. No attempt was made to establish causality by comparing cases of atrial fibrillation with an appropriate control group. Also, there was no adjustment for potentially confounding variables.Causality was established by declaration. The authors considered alcohol to be the cause of atrial fibrillation if the patient was intoxicated and no other cause could be identified, and considered alcohol to be a contributing cause if the patient was intoxicated and other known causes of atrial fibrillation were identified. Alcohol intoxication was present in 14 of 40 cases; therefore, the authors