Arrhythmias & electrophysiology
Risk calculator for incident atrial fibrillation across a range of prediction horizons
Authors' conclusion
The FIND-AF models demonstrate high discrimination and calibration across short- and long-term prediction horizons in 2 million individuals.
Study context
- Population
- Individuals aged ≥30 years without known atrial fibrillation (AF), UK primary care EHR linked to secondary care, 1998–2018
- Participants
- 2,081,139 individuals
- Comparison
- FIND-AF models vs CHA2DS2-VASc and C2HEST scores
- Outcome
- Discrimination (AUROC) and calibration of models predicting incident AF
- Follow-up
- 6 months, 1, 2, 5 and 10 years
Methodological detailsretrospective cohort, EHR data (UK CPRD-GOLD), 80% derivation / 20% validation split, bootstrap validation (200 samples)
- Retrospective cohort
- EHR data (UK CPRD-GOLD)
- 80% derivation / 20% validation split
- Bootstrap validation (200 samples)
Performance
As reported in the abstract
Discrimination (AUROC, 95% CI) and calibration slope of FIND-AF models for incident AF
- 6 months: AUROC (95% CI); calibration slope
- 0.803 (0.789-0.821); 0.952
- 1 year: AUROC (95% CI); calibration slope
- 0.807 (0.794-0.819); 0.962
- 2 years: AUROC (95% CI); calibration slope
- 0.815 (0.807-0.823); 0.973
- 5 years: AUROC (95% CI); calibration slope
- 0.807 (0.803-0.812); 1.000
- 10 years: AUROC (95% CI); calibration slope
- 0.780 (0.777-0.784); 1.010
Further findings
Incident AF (whole cohort)
- Within 6 months
- 7,386 (0.4%)
- In 1 year
- 15,349 (0.7%)
- In 5 years
- 38,487 (1.8%)
- By 10 years
- 79,997 (3.8%)
Predictors
- Valvular heart disease and HF strongest predictors; association of hypertension with AF increased at longer horizons
- Optimal models: age, sex, ethnicity and 8 comorbidities
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Risk calculator for incident atrial fibrillation across a range of prediction horizons. Am Heart J. Published 6 March 2024. doi:10.
Summarised from the published abstract; every number is checked against it