Use of the signal-averaged QRS duration for diagnosing left ventricular hypertrophy in hypertensive patients

Abstract

We analysed the diagnostic performance of the signal-averaged QRS duration for the detection of left ventricular hypertrophy in 100 consecutive outpatients (62 men and 38 women; mean age, 49.8 ± 11.8 years) with essential hypertension and compared the results with some of the currently employed electrocardiographic criteria. Forty-eight healthy subjects (24 men and 24 women; mean age, 46.4 ± 12.1) with normal physical, electrocardiographic and echocardiographic findings served as a control group to derive normal reference values for signal-averaged QRS duration. Twenty-six (26%) hypertensives (22 men and 4 women) had left ventricular hypertrophy echocardiographically defined as a left ventricular mass ≥ 261 g in men and ≥ 172 g in women or left ventricular mass index ≥ 125 g/m 2 in men and ≥ 112 g/m 2 in women. The signal-averaged QRS duration was different in patients with than in those without left ventricular hypertrophy (102.1 ± 10.8 vs. 95.8 ± 8.4 ms; P < 0.01). Also, in the group with left ventricular hypertrophy QRS duration was longer, although not significantly different, in men than in women (103.5 ± 10.7 vs. 94.2 ± 8.8 ms; P n.s.). The correlation between the signal-averaged QRS duration and left ventricular mass was weak but statistically significant in men ( r = 0.34; P < 0.05) in women ( r = 0.30; P < 0.05) and in men and women together ( r = 0.42; P < 0.01). Partition values of filtered QRS duration ≥ 114 ms in men and ≥ 107 ms in women were used to diagnose left ventricular hypertrophy as these values were above the upper limits in our control men and women when 95% confidence intervals were calculated. These criteria were insensitive (12%) but highly specific (99%) for left ventricular hypertrophy. The use of a single threshold value of filtered QRS duration ≥ 111 ms in both sexes combined improved sensitivity modestly (15%) while maintaining a good specificity (95%). Also, we tested the following standard electrocardiographic criteria: the Sokolow-Lyon index, the Romhilt-Estes point score ≥ 4 points and ≥ 5 points, the Cornell voltage criteria, the sum of QRS voltages in all 12 leads > 175 mm, and the QRS duration > 90 ms. Sensitivities ranged from 4% to 58% and specificities from 74% to 99%. To compare the performance for left ventricular hypertrophy of the electrocardiographic criteria we modified, when appropriate, partitions value for test positivity to reach a clinically relevant 95–100% range in specificity. The use of McNemar's test did not show significant differences among sensitivities. Using the same procedure, the 12% sensitivity of the sex-specific signal-averaged QRS duration criteria and the 15% sensitivity of the single QRS duration criterion for both sexes did not significantly differ from each other or from sensitivities of standard electrocardiographic criteria. Therefore, the signal-averaged QRS duration appears to have limitations comparable with standard electrocardiographic criteria in the diagnosis of left ventricular hypertrophy.