Electrocardiographic detection of left ventricular hypertrophy using echocardiographic determination of left ventricular mass as the reference standard Comparison of standard criteria, computer diagnosis and physician interpretation

Electrocardiographic findings of left ventricular hypertrophy were compared with echocardiographic left ventricular mass in 148 patients to assess performance of standard electrocardiographic criteria, the IBM Bonner program and physician interpretation. On echocardiography, 43% of the patients had left ventricular hypertrophy (left ventricular mass > 215 g). Sokolow-Lyon voltage (S in V1+ R in V5or V6) and Romhilt-Estes point score correlated modestly with left ventricular mass (r = 0.40, p < 0.001 and r = 0.55, p < 0.001, respectively). Sensitivity of Sokolow-Lyon voltage greater than 3.5 mV for left ventricular hypertrophy was only 22%, but specificity was 93%. Point score for probable left ventricular hypertrophy (≥ 4 points) had 48% sensitivity and 85% specificity, whereas definite hypertrophy (≥ 5 points) had 34% sensitivity and 98% specificity. Computer analysis resulted in 45% sensitivity and 83% specificity. Overall diagnostic accuracy of the IBM Bonner program (67%) was better than that of Sokolow-Lyon voltage (62%), but worse than the Romhilt-Estes point score (69% for ≥ 4 points or 70% for ≥ 5 points). Three cardiologists interpreted electrocardiograms independently and in a blinded fashion. Physician sensitivity was 56%, specificity 92% and accuracy 76%. Correlation with left ventricular hypertrophy was good (r = 0.70, p < 0.001).It is concluded that: 1) computer diagnosis of left ventricular hypertrophy by the IBM Bonner program is no more accurate than diagnosis by Sokolow-Lyon or Romhilt-Estes criteria, and 2) physician recognition of left ventricular hypertrophy is more accurate. This suggests that additional information about left ventricular hypertrophy is present in the electrocardiogram that is not detectable by standard criteria or the IBM computer program