Test characteristics of electrocardiography for detection of left ventricular hypertrophy in asymptomatic emergency department patients with hypertension.

OBJECTIVES

The objective was to evaluate the diagnostic test characteristics of three validated electrocardiographic (ECG) criteria for the diagnosis of left ventricular hypertrophy (LVH) in undifferentiated, asymptomatic emergency department (ED) patients with hypertension (HTN).

METHODS

This was a prospective cohort study of ED patients with asymptomatic HTN at a single tertiary care facility. Patients 35 years of age or older with systolic blood pressure (sBP) ≥ 140 mm Hg or diastolic blood pressure (dBP) ≥ 90 mm Hg on two separate readings (at least 1 hour apart) were eligible for inclusion. At enrollment, ECGs were obtained for all patients. Presence of LVH on ECG was defined using Cornell voltage, Cornell product, and Minnesota Code 3.1/3.2 criteria. Echocardiography was then performed, with LVH defined by the presence of one or more of the following validated criteria: interventricular septal or posterior wall thickness ≥ 1.3 cm, LV mass ≥ 225 g (male) or ≥ 163 g (female), or LV mass indexed to height raised to the power of 2.7 ≥ 48 g/m(2.7) (male) or ≥ 45 g/m(2.7) (female). Descriptive statistics and diagnostic characteristics (i.e., sensitivity and specificity) with corresponding 95% confidence intervals (CIs) for each of the three ECG criteria were derived for both the composite and the individual echocardiographic determinants of LVH. Logistic regression was also used to model LVH before and after subsequent inclusion of clinically relevant variables.

RESULTS

A total of 161 patients (93.8% African American; mean [±SD] age = 49.8 [±8.3] years) were enrolled, and LVH was present in 89 patients (55.2%, 95% CI = 47.6% to 62.8%). On ECG analysis, mean Cornell voltage (21.5 mV vs. 28.7 mV; difference = -7.2 mV, 95% CI = -3.8 to -10.7 mV) and Cornell product (1868.4 msec × mV vs. 2616.4 msec × mV; difference = -748.0 msec × mV, 95% CI = -401.2 to -1094.8 msec × mV) were significantly lower among those without LVH on echocardiography. Subjects without LVH on echocardiography were less likely to meet Cornell voltage (30.5% vs. 48.3%; difference = -17.8%, 95% CI = -2.5% to -31.7%) or Cornell product (26.4% vs. 49.4%; difference = -23.0%, 95% CI = -8.0% to -36.5%) criteria for LVH. The diagnosis of LVH by Minnesota Code was less common (18.1% vs. 25.8%; difference = -7.7%, 95% CI = -20.1% to 5.3%) with no difference by group. Sensitivity and specificity were as follows: for the Cornell voltage, sensitivity 25.4% (95% CI = 15.3% to 37.9%), specificity 50.0% (95% CI = 67.6% to 93.2%); for the Cornell product, sensitivity 25.4% (95% CI = 15.3% to 37.9%), specificity 75.0% (95% CI = 19.4% to 99.4%); and for the Minnesota code, sensitivity 26.9% (95% CI = 16.6% to 39.7%), specificity 75.0% (95% CI = 19.4% to 99.4%). On logistic regression, the c-statistics for Cornell voltage and Cornell product were equivalent (0.67), with only marginal improvement after the addition of body mass index (BMI; 0.69 and 0.70, respectively), B-type natriuretic peptide (BNP; 0.68 and 0.69, respectively), or both (0.71 and 0.72, respectively) to the models.

CONCLUSIONS

In this cohort of predominately African American ED patients with asymptomatic HTN, sensitivity and specificity of standard ECG criteria were relatively poor for the diagnosis of LVH on echocardiography. Thus, ECG is of limited use for LVH risk stratification in asymptomatic ED patients with elevated blood pressure, with additional clinical information only modestly strengthening its predictive value.