Article Text

Download PDFPDF

Accuracy of ECG interpretation in primary care was limited for detecting atrial fibrillation

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

D A Fitzmaurice

Dr D A Fitzmaurice, University of Birmingham, Birmingham, UK; d.a.fitzmaurice{at}


What is the accuracy of different types of electrocardiography (ECG) and different interpreters for diagnosis of atrial fibrillation (AF) in primary care?



blinded comparison of 3 types of ECG interpreted by 3 different groups with 12-lead ECGs read by cardiology consultants (substudy of a randomised controlled trial [Screening for Atrial Fibrillation in the Elderly, SAFE]).


49 primary care practices in central England, UK.


2595 patients from the SAFE intervention group (⩾65 y) who were identified for an ECG because of an irregular pulse on opportunistic screening (n = 238) or were selected randomly to have a 12-lead ECG (n = 2357).

Description of tests:

1 of 3 ECG types was randomly selected for each patient (847 single chest-lead, 858 limb-lead, and 848 12-lead). Batches of 100 ECGs were sent for interpretation to 2 primary care practitioners (1 physician and 1 nurse) in each of a control and intervention practice. 2556 12-lead ECGs were interpreted using Biolog interpretative software.

Diagnostic standard:

all patients had a 12-lead ECG read independently by 2 consultant cardiologists. A third consultant arbitrated disagreements.


sensitivity and specificity.


Prevalence of AF was 8.4%. Diagnostic test characteristics were calculated for each type of ECG and for each form of interpretation (table). Different ECG types appeared equally accurate for detecting presence of AF (p = 0.52 matched for physicians and p = 0.08 for nurses); 12-lead ECGs read by physicians (p<0.001) but not nurses (p = 0.12) were better than other ECG types for detecting absence of AF. Accuracy of individual practitioners was variable.

Different ECG types and interpreters v 12-lead ECGs interpreted by consultants for detection of atrial fibrillation*


Accuracy of different types of ECGs interpreted by general practitioners, practice nurses, and software was limited for detection of atrial fibrillation in primary care.


Mant J, Fitzmaurice DA, Hobbs FDR, et al. Accuracy of diagnosing atrial fibrillation on electrocardiogram by primary care practitioners and interpretative diagnostic software: analysis of data from screening for atrial fibrillation in the elderly (SAFE) trial. BMJ 2007;335:380.

Clinical impact ratings: GP/FP/Primary care 6/7; IM/Ambulatory care 6/7; Cardiology 6/7


In this prospective screening study of elderly patients (⩾65 y), nested within a randomised controlled trial, Mant et al assessed the accuracy of general practitioners (GPs), practice nurses, and a computer software program for identifying AF based on ECGs, with blinding to clinical information. When ECGs were used for screening, both health professional groups correctly identified about 4 out of 5 elderly patients with AF. For a test where high sensitivity is most desirable, this false negative rate of nearly 20% is too high. Addition of the software program to GP interpretation of the ECG improved sensitivity to >90%. The authors concluded that if accurate identification of AF in the community is required, either through systematic or opportunistic screening, then ECGs should be read by appropriately trained people (eg, cardiologists). These findings have important implications for ongoing community-based screening interventions aimed at identifying AF in elderly people.

The findings of Mant et al should not be interpreted as meaning ECG testing by GPs in the community is not worthwhile. When considering a diagnosis of AF, important demographic (age and sex) and clinical factors (history of ischaemic heart disease, diabetes, and hypertension) affect the probability of AF, and ECG has an important confirmatory diagnostic role. The clinical examination—palpation of the patient’s radial pulse—has been shown to have a sensitivity of 94% but a lower specificity of 72%,1 highlighting the need for further diagnostic testing in using ECG to “rule in” a diagnosis of AF. In this clinical context, the use of ECG in diagnosing AF has an important confirmatory role.


View Abstract


  • Source of funding: Health Technology Assessment Programme.

Linked Articles