Subjective analysis of cross-sectional echocardiograms: reproducibility and sources of variability

Subjective analysis of left ventricular segmental wall motion is widely used to assess regional left ventricular function and the effects of intervention in patients with ischaemic heart disease. To determine its reproducibility we studied 20 consecutive men with proven ischaemic heart disease and 10 age-matched control subjects. Standard parasternal, apical and subcostal views were attempted in all subjects by two echocardiographers, and echoes were reported by two analysers, each on two occasions. The left ventricle was subdivided into 5 segments, and segmental wall motion scored as normal, hypokinetic, akinetic, dyskinetic or inadequately visualised. The different sources of variability were evaluated. The echocardiographers contributed little variability, the frequency and reproducibility of reporting being the same for echoes produced by each echocardiographer. The only exception was for subcostal views in which one echocardiographer produced significantly more reportable segments than the other (42% vs 27%; P less than 0.001). Within and between analyser variability was large. Replicate analysis of the same segment showed within analyser agreement of 92% for one analyser but only 80% for the other. Between analyser agreement was 89% for some segments but only 63% for others and was best in those segments where both analysers infrequently reported abnormal wall motion. Reporting bias is thus an important variable. The subjective assessment of regional left ventricular function thus has major limitations for general applications particularly when being used in longitudinal studies.