Temporal variation in the diagnosis of resolved atrial fibrillation and the influence of performance targets on clinical coding: cohort study

Research output: Contribution to journalArticlepeer-review

Colleges, School and Institutes


To investigate whether the introduction of performance targets for anticoagulation in atrial fibrillation (AF) was associated with a change in use of the ‘resolved atrial fibrillation’ code.
Retrospective cohort studies.
Data from The Health Improvement Network (THIN), a UK database of electronic patient records, from 2000 to 2016.
250,788 adult patients aged ≥18 years with a diagnosis of AF, including 14,757 with an incident diagnosis of ‘resolved AF’.
Main outcome measures
Annual and monthly incidence of ‘resolved AF’ from 2000 to 2016. Among patients with ‘resolved
AF’, for each year we calculated median duration of the preceding AF diagnosis and the proportion prescribed anticoagulants prior to ‘resolved AF’.
Incidence of ‘resolved AF’ increased from 5.7 to 26.3 per 1000 person-years between 2005 and the introduction of AF performance targets in 2006. Compared to the years prior to the introduction of the performance targets, incidence has remained higher in every year since their implementation.
Since 2007, monthly incidence has been highest between January and March. Between 2005 and
2006, median duration between AF and ‘resolved AF’ diagnoses increased from 276 days (9 months) to 1343 days (3 years 8 months). Among ‘resolved AF’ patients with CHA2DS2-VASc score ≥1, 81.9% (95%CI 81.1 to 82.6) had no current anticoagulant prescription, and 62.3% (95%CI 61.4 to 63.2) had no record of any anticoagulant prescription.
The introduction of AF performance targets was followed by a large increase in use of the ‘resolved AF’ code, particularly in the months immediately before practices make their anticoagulant performance target submissions. Although most AF patients are prescribed anticoagulants, few patients diagnosed with ‘resolved AF’ are prescribed anticoagulants and most have never been
prescribed them. Untreated patients are much more likely to be coded as having ‘resolved AF’.


Original languageEnglish
Article numbere030454
JournalBMJ open
Issue number11
Early online date21 Nov 2019
Publication statusE-pub ahead of print - 21 Nov 2019


  • anticoagulation, cardiology, primary care

ASJC Scopus subject areas