Aims: We sought to compare the predictive power and reproducibility between minimum and maximum left atrial (LA) volume for the development of first atrial fibrillation (AF)/flutter.
Methods and results: This prospective study included 574 adults, mean age 74+/-6 years, in sinus rhythm, and had no history or evidence of prior atrial arrhythmias. During a mean follow-up of 1.9+/-1.2 years, 30 (5.2%) developed first AF/flutter. The 3-year risk estimates of freedom from AF/flutter by tertiles of minimum and maximum LA volumes were, respectively, 97, 87, and 74% (P<0.0006) and 94, 85, and 78% (P=0.03). Minimum LA volume was incremental to clinical and other echocardiographic parameters of AF/flutter prediction [per tertile, hazard ratio (HR) 2.4], as was maximum LA volume (per tertile, HR 1.8) in a separate model. When both volumes were entered into the same model and adjusting for covariates, minimum but not maximum LA volume retained significance. However, in terms of interobserver reproducibility, maximum LA volume compared more favourably (mean difference 3.1+/-7.1 vs. 7.4+/-7.3 mL/m(2)).
Conclusion: Minimal LA volume was an independent predictor of first AF/flutter. Although it was marginally superior to maximal LA volume in terms of predictive ability, the interobserver variability was greater.