BackgroundAlthough structural heart disease is often present in patients with nonrheumatic atrial fibrillation, the echocardiographic precursors of atrial fibrillation have not been reported previously. In this elderly, population-based cohort, our objective was to examine prospectively the echocardiographic predictors of nonrheumatic atrial fibrillation.Methods and ResultsSubjects in the Framingham Heart Study were routinely evaluated with M-mode echocardiography; 1924 subjects, ranging in age from 59 to 90 years, comprised the population at risk. Cox proportional hazards modeling was used to analyze the association of selected echocardiographic features with atrial fibrillation risk after adjustment for age, sex, hypertension, coronary heart disease, congestive heart failure, diabetes, and valvular heart disease. During a mean follow-up interval of 7.2 years, 154 subjects (8.0%) developed atrial fibrillation. Multivariable stepwise analysis identified left atrial size (hazard ratio [HR] per 5-mm increment, 1.39; 95% confidence interval [CI], 1.14 to 1.68), left ventricular fractional shortening (HR per 5% decrement, 1.34; 95% CI, 1.08 to 1.66), and sum of septal and left ventricular posterior wall thicknesses (HR per 4-mm increment, 1.28; 95% CI, 1.03 to 1.60) as independent echocardiographic predictors of atrial fibrillation. For each of the echocardiographic predictors, risk increased progressively over successive quartiles. Moreover, risk increased markedly when highest-risk-quartile measurements for these features were present in combination; the cumulative 8-year age-adjusted atrial fibrillation rates were 7.3% and 17.0%, respectively, when one and two or more highest-risk-quartile features were present, compared with 3.7% when none was present.ConclusionsIn this elderly, population-based sample, left atrial enlargement, increased left ventricular wall thickness, and reduced left ventricular fractional shortening were predictive of risk for nonrheumatic atrial fibrillation. These echocardiographic precursors offer prognostic information beyond that provided by traditional clinical atrial fibrillation risk factors.