Postgrad Med J: first published as 10.1136/pgmj.67.789.646 on 1 July 1991. Downloaded from Postgrad Med J (1991) 67, 646 - 648 D The Fellowship of Postgraduate Medicine, 1991

How reliable is the electrocardiogram in detecting left ventricular hypertrophy in hypertension? S.G. Vijan, Gillian Manning and M.W. Millar-Craig Department ofCardiology, Derbyshire Royal Infirmary, London Road, Derby DE] 2Q Y, UK

Summary: This paper assesses the sensitivity and specificity ofthe electrocardiogram in detecting left ventricular hypertrophy in 75 hypertensive patients. Each patient underwent a 12 lead electrocardiogram and echocardiogram. Left ventricular mass index, using echocardiogram, was calculated according to the Penn convention and left ventricular hypertrophy was assessed by standard electrocardiographic criteria. The electrocardiogram was found to be very specific but insensitive in the detection of left ventricular hypertrophy as compared with the echocardiogram. Other non-voltage dependent markers appeared to have similar reliability. We conclude that the electrocardiogram may be unreliable in the detection of left ventricular hypertrophy in hypertensive patients. Accurate assessment of left ventricular hypertrophy, in these patients should be by echocardiography.

Introduction Left ventricular hypertrophy imparts a substantial measurements of resting blood pressure and risk of morbidity and mortality.' Echocardio- anthropometric measurements were made. copyright. graphy has permitted the reliable, non-invasive Echocardiography was performed, using a Hew- estimation ofleft ventricular mass.2 However, most lett Packard Sonos 500 series echocardiograph, by physicians have relied on the electrocardiogram to the same trained technician. The echocardiogram detect the presence of left ventricular hypertrophy was performed in the left lateral position and all in hypertension. Several electrocardiographic crite- recordings were obtained during quiet respiration. ria have been suggested, including the Romhilt-Estes The echocardiogram was coded and independently

point score,3 Sokolow-Lyons voltage criteria,4 and analysed by two experienced investigators. Left http://pmj.bmj.com/ the presence of ST-T wave changes ('strain').5 We ventricular mass (LVMI) was calculated using the conducted a study to assess the sensitivity and Penn convention2 and corrected for body surface specificity ofthe electrocardiogram in detecting left area. A left ventricular mass index of > 115 g/m2 ventricular hypertrophy in patients with essential (mean ± 2 s.d. of 35 normal subjects in our labora- hypertension. tory) was considered to be indicative of left ventri- cular hypertrophy. The electrocardiogram was recorded on a Hew- Patients and methods lett Packard pagewriter and independently inter- on September 27, 2021 by guest. Protected preted by both investigators who were also Eighty-six patients with treated essential hyperten- unaware of the echocardiographic findings. Left sion between the ages of 30-65 were enrolled in a ventricular hypertrophy was assessed by two com- prospective study. The mean duration oftreatment monly used methods. The Romhilt-Estes point was 2 years (95% CI 1-3 years). Patients with score system3 (using a score based on amplitude previously known myocardial infarction or angina and duration of the QRS complex in limb and were excluded. All patients came off their drug precordial leads, a score of 4 or greater indicating therapy for a period of 2 weeks. Each patient left ventricular hypertrophy) and the Sokolow- underwent a 12 lead electrocardiogram and an Lyons score4 (using a sum of amplitude SV, and echocardiogram on the same occasion. In addition, RV5 or RV6 > 35 mm and RV5 or RV6 > 26 mm indicating left ventricular hypertrophy). Other non-voltage markers of left ventricular, i.e. ST-T wave changes, left atrial abnormalities (terminal negativity of P in lead VI) and left axis deviation Correspondence: S.G. Vijan, M.D., M.R.C.P. (axis <- 30) were also studied. In addition the Accepted: 8 February 1991 combination of voltage criteria of left ventricular Postgrad Med J: first published as 10.1136/pgmj.67.789.646 on 1 July 1991. Downloaded from LEFT VENTRICULAR HYPERTROPHY IN HYPERTENSION 647 hypertrophy (SVI and RV5 or RV6 > 35 mm) with Table I Sensitivity, specificity and accuracy of ECG ST-T wave inversion ('LVH with strain')5 was also criteria for detecting left ventricular hypertrophy in 75 studied. patients A standard Bayes analysis was used to calculate sensitivity, specificity and diagnostic accuracy for Criteria Sensitivity Specificity Accuracy electrocardiographic criteria of left ventricular Romhilt-Estes 20.4% 96% 46% hypertrophy present on the echocardiogram. Com- score (10/49) (25/26) parison of quantitative variables was performed Sokolow-Lyons 16% 100% 45% using standard least-square linear regression ana- score (8/49) (26/26) lysis. LVH with strain 15% 100% 44% (7/49) (26/26) ST-T wave 16.3% 70% 34% Results changes* (8/49) (18/26) Left atrial 48.2% 55% 50% abnormality** (24/49) (14/26) Of the 86 patients studied, 75 had diagnostic Left axis 35% 15% 28% quality echocardiograms for analysis. Hence, ana- deviation*** (17/49) (4/26) lysis was restricted for these 75 patients. The mean (s.d.) systolic blood pressure was 184 (14) mmHg * = isolated ST-T wave depression; ** = isolated ter- and mean (s.d.) diastolic blood pressure was 104 minal -ve P in VI; *** = left axis < - 30°. (10) mmHg. Of these, 49 (39 males, mean age 55 years, range 35-65) had echocardiographic left ventricular hypertrophy (mean 165 g/m2, range 130-390) and 26 (20 males, mean age 54 years, its reliability has also been confirmed angio- range 30-65) did not show left ventricular hyper- graphically.7 trophy (mean LVMI = 104 g/m2, range 89-114). The homogenous population we studied permit- There was no difference in blood pressure between ted the assessment of Romhilt-Estes point score

the two groups. and Sokolow-Lyons criteria in a clearly defined copyright. Calculated sensitivity, specificity and accuracy clinical setting. In our population, both the electro- for the electrocardiographic criteria assessed are cardiographic criteria appeared to have a very low given in Table I. sensitivity due to a high incidence offalse negatives The Romhilt-Estes point score system was high- but both methods appeared to be very specific ly specific but had a low sensitivity and the (specificity> 90%). Therefore, in a population correlation with echocardiographic left ventricular with prevalence of left ventricular hypertrophy the mass index was poor (r = 0.28). The Sokolow- high false-negative rate renders both criteria of

Lyons criteria was also highly specific but was less limited value. Both scores were not affected by http://pmj.bmj.com/ sensitive and correlation with echocardiographic increasing body weight and had similar sensitivity left ventricular mass index was poor (r = 0.17). and specificity in men and women. Both scores had similar sensitivity and specificity in Non-voltage markers of left ventricular hyper- men and women. trophy, including ST-T strain, left atrial abnorma- The combination of voltage left ventricular lity, left-axis deviation, gave diagnostic accuracy hypertrophy with ST-T wave inversion was very similar to the Romhilt-Estes point score and the specific but sensitivity was poor. We also evaluated Sokolow-Lyons score. However, these markers, other non-voltage markers of left ventricular although more sensitive than the standard electro- on September 27, 2021 by guest. Protected hypertrophy. Isolated ST-T wave changes had a cardiographic criteria, were consistently less poor sensitivity and moderate specificity. Left specific (<70%) except for the combination of atrial abnormalities and isolated left axis deviation voltage left ventricular hypertrophy and ST-T had similar sensitivity and specificity. Further wave inversion which was very specific (100%). analysis of the data showed similar diagnostic Thus, the clinician, when assessing a hypertensive accuracy in overweight (body weight> 110% ideal patient, could use any of these criteria as a marker body weight) when compared with non-obese of hypertensive left ventricular hypertrophy. subjects. The results of our study differ from the sugges- tions of early reports8-10 where good correlation has been shown between electrocardiographic Discussion criteria and echocardiographic or post-mortem left ventricular hypertrophy. However, these studies Echocardiographic criteria for left ventricular were either population based or involved a hetero- hypertrophy have been shown to have excellent geneous group of patients and did not address the sensitivity, specificity and accuracy6 when com- specific question of the reliability of the electro- pared with postmortem left ventricular mass, and cardiogram in hypertensive subjects. Our study Postgrad Med J: first published as 10.1136/pgmj.67.789.646 on 1 July 1991. Downloaded from 648 S.G. VIJAN et al. suggests that, in hypertensive subjects, the reli- hypertensive patients, then this should be done by ability ofelectrocardiographic criteria is very poor echocardiography. Such a policy could lead to a and that it is less than that in the general popula- large increase in the workloads of echocardio- tion. Recent reports'-12 have suggested that echo- graphic departments throughout the country. cardiographic left ventricular hypertrophy predicts However, in view of recent reports'"-12 regarding complications of hypertension in men and women. the prognostic implications of echocardiographic Due to the low sensitivity and poor accuracy of left ventricular hypertrophy in hypertension and electrocardiographic diagnosis of left ventricular the beneficial effects ofanti-hypertensive therapy in hypertrophy, we conclude that if accurate assess- reducing echocardiographic left ventricular hyper- ment of left ventricular hypertrophy is required in trophy,'3 such a policy could be justified.

References 1. Kannel, W.G., Gordon, T., Castelli, W.P. & Margolis, J.R. 8. Benneth, D.H. & Evans, D.W. Correlation of left ventricular Electrocardiographic left ventricular hypertrophy and risk of mass determined by echo with vectorcardiographic and coronary heart disease: The Framingham study. Ann Intern electrocardiographic voltage measurements. Br Heart J 1974, Med 1970, 72: 813-822. 36: 981. 2. Devereux, R.B. & Reichek, N. Echocardiographic determina- 9. Antran, E.M., Green, L.H. & Grossman, W. Physiologic tion of left ventricular mass in man: anatomic validation of determinants of the electrocardiographic diagnosis of left the method. Circulation 1977, 55: 613-618. ventricular hypertrophy. Circulation 1979, 60: 386. 3. Romhilt, D.W., Bone, K.E., Norris, R.J. et al. A critical 10. Horton, J.D., Sherber, H.S. & Lakata, E.G. Distance correc- appraisal of the electrocardiographic criteria for the diag- tion for precordial electrocardiographic voltage in estimating nosis of left ventricular hypertrophy. Circulation 1969, XL: left ventricular mass. Circulation 1977, 55: 509-512. 185-195. 11. Levy, D., Garisson, R.J., Savage, D.D., Kannel, W.B. & 4. Sokolow, M. & Lyons, T.P. The ventricular complex in left Castelli, W.P. Prognostic implications ofechocardiographic- ventricular hypertrophy as obtained by unipolar precordial ally determined left ventricular mass in the Framingham and limb leads. Am Heart J 1949, 37: 161. study. N Engi J Med 1990, 332: 1561-1566. 5. Hamer, J., Shinebourne, E. & Fleming, J. Significance of 12. Koren, M.J., Devereux, R.B., Pappas, T.W. et al. Echo-

electrocardiographic changes in hypertension. B MedJ 1969, cardiographic left ventricular mass predicts complications of copyright. 1: 79-82. hypertension in both men and women (abst). Am J Hyperten- 6. Reichek, N. & Devereux, R.B. Left ventricular hypertrophy: sion 1988, 1, 12A. relationship of anatomic, echocardiographic and electrocar- 13. Schulman, S.P., Weiss, J.L., Becker, L.C. et al. The effects of diographic findings. Circulation 1981, 63: 1391-1398. antihypertensive therapy on left ventricular mass in elderly 7. Nixon, J.V., Anderson, R.J. & Cohen, M.C. Alteration in left patients. N Engi J Med 1990, 322: 1350-1356. ventricular mass and performance in patients treated effect- ively for thyrotoxicosis. Am J Med 1979, 67: 268. http://pmj.bmj.com/ on September 27, 2021 by guest. Protected