Randomised trial of two approaches to screening for atrial fibrillation in UK general practice
Morgan S., Mant D.
Background: Atrial fibrillation is a common and treatable cause of stroke that often remains unrecognised. Screening has been suggested but there is very little evidence concerning the uptake of screening in the elderly population at risk, nor of the optimal method of screening in a general practice setting. Aim: To compare the uptake and effectiveness of two methods of screening for atrial fibrillation in general practice - systematic nurse-led screening and prompted opportunistic case finding. Design of study: Randomised controlled trial. Setting: Patients aged 65 to 100 years (n = 3001) from four general practices within the MRC general practice framework. Method: Each of the four study practices were selected from one quartile, after ranking all framework practices according to the small area standardised mortality ratio of the geographical area served. Patients were randomised either to nurse-led screening or to prompted opportunistic case finding. The proportion of patients assessed and the proportion found to have atrial fibrillation were compared. The sensitivity and specificity of clinical assessment of pulse are also reported. Results: Substantially more patients had their pulse assessed through systematic screening by invitation (1099/1499 [73%] ) than through opportunistic case finding (439/1502 [29%], difference = 44%, 95% confidence interval [CI] = 41% to 47%). Atrial fibrillation was detected in 67 (4.5%) and 19 (1.3%) patients respectively (difference = 3.2%, 95% CI = 2.0 to 4.4). Invitation to nurse-led screening achieved significantly higher assessment rates than case finding in all practices; however, the proportion of patients assessed in the case-finding arm varied markedly between practices (range = 8% to 52%). The number needed to screen to identify one additional patient with atrial fibrillation was 31 (95% CI = 23 to 50). The proportion of screened patients with atrial fibrillation receiving anticoagulation treatment was 25%, although in the majority (53/65 [82%]) atrial fibrillation had been previously recorded somewhere on their medical record. If the nurse used any irregularity of the pulse as the screening criterion, the sensitivity of screening was 91% and the specificity was 74%; sensitivity fell to 54% but specificity increased to 98% if the criterion used was continuous irregularity. Conclusions: Nurse-led screening for atrial fibrillation in UK general practice is both feasible and effective and will identify a substantial number of patients who could benefit from antithrombotic therapy. Although the majority of patients detected at first screening could be identified by careful scrutiny of medical records, review of record summaries was insufficient in the practices involved in this study and screening may be a more cost-effective option.