ECG identification of left ventricular hypertrophy. Relationship of test performance to body habitus

Obesity is associated with the presence of left ventricular hypertrophy (LVH) and, conversely, with decreased sensitivity of the electrocardiogram (ECG) for LVH due to attenuating effects on QRS amplitudes. Although the Framingham-adjusted Cornell voltage, incorporating age, sex, and body mass index (BMI), was developed to correct for the effects of obesity on the accuracy of the ECG, the impact of body habitus on ECG detection of LVH for newer, more accurate ECG criteria based on the time-voltage area under the QRS complex has not been determined. The authors examined the test accuracy of the Sokolow-Lyon voltage, Cornell voltage, Cornell product (product of QRS duration and Cornell voltage), Framingham-adjusted Cornell voltage, and time-voltage area of the horizontal plane vector QRS for the detection of echocardiographic LVH in relation to body habitus in 250 patients. Normal-weight or overweight status was based on sex-specific population-based BMI partitions. Using partitions with a matched specificity of 98% in the overall population without LVH, the sensitivity of standard ECG criteria varied according to body habitus. Sensitivity of the Framingham-adjusted Cornell voltage was less in normal-weight than in overweight patients (49 vs 59%, P = .0004); there were also trends toward lower sensitivity in normal-weight patients for the Cornell voltage (40 vs 65%, P = .10) and the Cornell product (43 vs 65%, P = NS), but sensitivity of the Sokolow-Lyon voltage was lower in obese than in nonobese patients (18 vs 50%, P = .025). In contrast, the horizontal plane vector area had similar sensitivity in obese and normal-weight patients (76 vs 74%, P = NS). Specificity varied with body habitus only for the Framingham-adjusted Cornell voltage: 100% in normal-weight vs 95% in overweight patients (P < .05). Thus, accuracy of the Framingham-adjusted Cornell voltage and Sokolow-Lyon voltage varies significantly with body habitus. In contrast, accuracy of the Cornell voltage and the Cornell product appears less dependent on BMI, and the time-voltage area of the QRS minimizes the effects of obesity on the accuracy of the ECG for LVH.