Creating a Prediction Model to Identify Patients at High Risk of undetected Non-valvular Atrial Fibrillation

Background:
Atrial fibrillation (AF) is the most common arrhythmia in clinical practice and is a leading cause of morbidity and mortality. It is a major risk factor for thromboembolic disease and it is well known that even short episodes of intermittent AF convey an increased risk for stroke. These events can be avoided with anticoagulation treatment for patients in high risk (according to CHADS2 score).

Objective:
Creating a risk score within the general population for identifying individuals at high risk for diagnosis of non-valvular AF.

Methods and Results:
A retrospective cohort study, based on Clalit Health Services (CHS) data warehouse. The study population consists all adult members older than 40 years, with no previous diagnosis of AF.

Within the five years following the index date, we have compared the baseline characteristics of patients who developed AF with those who did not. We calculated the 5 years risk to develop AF for every individual in our cohort, creating a risk score to identify those at high risk.

Results:
The research population included 1,256,617 patients, with an AF incidence rate of 2.4%.

The best predictors to develop AF within 5 years were: Age, Gender, Congestive Heart Failure, Hypertension, anti-hypertensive treatment, BMI and Ischemic Heart Disease. The Logistic Regression Model we created utilizing those risk factors showed a 9.4% incidence of AF within the highest decile and a 14.3% within the highest percentile, a lift of 3.9 and 5.9 respectively, compared to the risk in the overall population.

Conclusion:
Our risk score model for detecting AF consists of classic and simple risk factors. We suggest that systematic screening for AF in asymptomatic patients, in accord with our risk score model, has the potential of helping to identify patients likely to benefit from more intense screening.