Methods : A large international study was undertaken to compare 9 ECG and 6 vectorcardiographic (VCG) computer programs using 1220 clinically validated cases. The same ECGs were also read by 9 cardiologists, 8 of whom analysed the ECG and 5 the VCG. The following diagnostic groups were included : normals (N=382) ; left (N=183), right (N=55) and biventricular (N=53) hypertrophy ; anterior (N=170), inferior (N=273) and combined (N=73) myocardial infarction ; combined infarction and hypertrophy (N=31).

All individual program and cardiologist results were compared with the “truth”, based on ECG independent evidence, while all programs were also compared with the combined interpretations of the cardiologists.

Results : The median specificity of the ECG programs was 91.3% (range, 86.3% to 97.1%) against 80.9% (range, 71.4% to 86.6%) for the VCG programs (P<0.001). Corresponding values for the cardiologists were 96.0% (range, 92.7% to 97.6%) for the ECG and 80.6% (range, 73.8% to 97.2%) for the VCG. The median sensitivities for left and right ventricular hypertrophy, and for anterior and inferior myocardial infarction were 55.7%, 32.7%, 74.1% and 65.1%, respectively, for the computer programs versus 63.4%, 48.5%, 82.9% and 73.3%, respectively (P<0.02 for all four), with equal or lower rates of false positives for the cardiologists. However, programs with the best performance reached almost equal levels as the best cardiologists.

Programs which used statistical methodology for diagnostic classification generally had a higher diagnostic accuracy than the deterministic programs when compared with the “truth”, but the study design may have favored this outcome.

Combined results of programs and of cardiologists were in almost all cases more accurate than those of the individual programs and cardiologists. The combined program and the combined cardiologists agreed in 87.5% of the cases, which is almost as high as the highest intra-observer reproducibility of the cardiologists (median : 82.4% ; range, 73.6% to 90.8%).

Conclusion : The present study shows that some ECG computer programs perform almost as well as the best cardiologists in classifying 7 main diagnostic entities. However, the study also shows that other programs could be improved considerably.