Testa and coworkers raise a number of questions related to our article1 on suspected increased mortality in patients with atrial fibrillation treated with digitalis. The only method for settling the question would be to perform a randomised controlled clinical trial on digitalis versus placebo in atrial fibrillation, with mortality as the primary outcome. For obvious reasons such a trial will probably not be done. The sample size would need to be large, long-term follow-up would be required, and …