Observer variability in M-mode echocardiographic diagnosis of mitral valve prolapse: Effect of tracing quality and depth of prolapse

To study observer variability in diagnosing mitral valve prolapse (MVP) from M-mode echocardiographic tracings, 6 experienced echocardiographers contributed 10 M-mode tracing each: three with MVP, three normal tracings, three equivocal tracings, and one other from any of these groups. The 60 tracings were randomized and read blindly twice by each echocardiographer. First they judged MVP to be present, absent, or 'equivocal' (three-choice mode). If equivocal, each observer further chose whether it was closer to MVP or normal (two-choice mode). A significant amount of inter- and intraobserver variability was noted. Intraobserver variability was smaller than interobserver variability. There was a direct relationship between the depth of mitral prolapse behind the C-D line and the likelihood of MVP being diagnosed, with satisfactory observer agreement (91%) on normality at 0-mm depth of prolapse and on MVP at ≥3 mm, and more observer variability at depths of prolapse between 0.5 and 2.5 mm. The eight tracings with a 'depth of mitral prolapse' of 3 mm or more below the C-D line had 98% observer agreement on the presence of MVP. The 14 tracings judged excellent in quality had significantly less observer variability than the other tracings, using Shannon's index of diversity. It is concluded that reader variation in M-mode echocardiographic diagnosis of MVP is larger than desirable in a technique considered to be the 'gold standard' for this diagnosis. Interobserver variability is larger than intraobserver variability. Excellent-quality studies have less reader variability than other studies. A depth of mitral prolapse of 3 mm or more results in excellent interobserver agreement on the presence of MVP, but adoption of this as a diagnostic criterion may result in underdiagnosis of mild instances of MVP as a heritable entity.