Consistency of echocardiographic ejection fraction: variation and 'drift' by interpreter and practice site

Visual analysis of ejection fraction by echocardiography (echo) is commonly performed, but little data exists regarding consistency among interpreters. We postulate that an active echo group evolves a standard to which interpreters cluster, with variation or 'drift' occurring at satellite sites distant from the main practice centre. Assuming that echo ejection fraction follows a normal distribution, average ejection fraction for high volume readers should be similar.

From May 1997 to June 1999, 8187 echoes were read by 16 (average 512) readers of a single practice at several sites. Composite and individual ejection fraction (visual estimate) means and standard deviations were compared. Composite ejection fraction was 54.17+/-11.35. Mean ejection fraction varied significantly overall among readers (P<0.0001) with 37 comparisons between individual readers being significantly different (P<0.05). However, analysis suggested that these individual differences involved only five readers, four of whom practised at satellite sites. If these five readers were excluded, no significant individual differences were seen (n=6202). If only the four satellite readers were excluded, just four significant individual differences were seen (n=6915).

(1) Within a busy single practice site, mean echo ejection fraction is very similar, suggesting a self-evolved practice standard that is consistently adhered to. (2) 'Drift' or variability is seen at smaller satellite sites, perhaps due to less exposure to the echo interpretations of others. (3) This type of analysis may provide a useful quality assurance tool for echo laboratories.