Ethnic differences in electrocardiographic criteria for left ventricular hypertrophy: The LIFE study

Background: African Americans have greater precordial QRS voltages than whites, with concomitant higher prevalences of electrocardiographic (ECG) left ventricular hypertrophy (LVH) and lower specificity of ECG LVH criteria for the identification of anatomic hypertrophy. However, the high mortality associated with LVH in African American patients makes more accurate ECG detection of LVH in these patients a clinical priority. Methods: Electrocardiograms and echocardiograms were obtained at study baseline in 120 African American and 751 white hypertensive patients enrolled in the Losartan Intervention For Endpoint (LIFE) echocardiographic substudy. The ECG LVH was determined using Sokolow-Lyon, 12-lead sum, and Cornell voltage criteria. Echocardiographic LVH was defined by LV mass indexed to height^2.7 >46.7 g/m^2.7 in women and >49.1 g/m^2.7 in men. Results: After adjusting for ethnic differences in LV mass, body mass index, sex, and prevalence of diabetes, mean Sokolow-Lyon and 12-lead sum of voltage were significantly higher, but Cornell voltage was lower, in African Americans than in whites. As a consequence of these differences, when identical partition values were used in both ethnic groups, Sokolow-Lyon and 12-lead voltage criteria had lower specificity in African Americans than whites (44% v 69%, P = .007 and 44% v 59%, P = .10) but had greater sensitivity in African Americans (51% v 27%, P < .001 and 62% v 45%, P = .003). In contrast, Cornell voltage specificity was higher (78% v 62%, P = .09) but sensitivity was slightly lower (49% v 57%, P = 0.16) in African Americans. However, when overall test performance was compared using receiver operating curve analyses that were independent of partition value selection, ethnic differences in test performance disappeared, with no differences in accuracy of any of the ECG voltage criteria for the identification of LVH between African American and white hypertensive individuals. Conclusions: When standard, non-ethnicity-specific thresholds for the identification of LVH are used, Sokolow-Lyon and 12-lead voltage overestimate and Cornell voltage underestimates the presence and severity of LVH in African American relative to white individuals. However, these apparent ethnic differences in test performance disappear when ethnic differences in the distribution of ECG LVH criteria are taken into account. These findings demonstrate that ethnicity-specific ECG criteria can equalize detection of anatomic LVH in African American and white patients.