Screening for atrial fibrillation: waiting for randomised trial data

We agree with Berge that Wilson and Jungner would not approve of screening for atrial fibrillation.1 The key problems are that we do not know whether screen detected atrial fibrillation carries the same risk of stroke as disease that is clinically detected and that the treatment, anticoagulation, has adverse effects.

In their original 1986 report for the World Health Organization,2 Wilson and Jungner specifically state that “in enthusiastically attacking disease at an early stage, the Hippocratic principle, previously mentioned, of ‘primum non nocere’ should not be neglected.” This has become modified into the more realistic principle that screening programmes should “do more good than harm at reasonable cost,”3 as we now recognise that all screening does some harm.

Thus, unless we have robust evidence that finding and treating atrial fibrillation in an asymptomatic population reduces stroke rates or mortality without relying on lead time and volunteer bias—and can do so without conferring significant morbidity from anticoagulation—we can’t recommend the introduction of a national screening programme. Such evidence can only be obtained in randomised trials, where randomisation takes place at the point of invitation to screening. As outlined in Berge’s letter, studies are under way and will be given full attention by the UK National Screening Committee when they report.