Abstract
Aims. Electrocardiography (ECG) has a high specificity but unfortunately low sensitivity to detect anatomic left ventricular hypertrophy (LVH). In this study, ECG amplitude and products were examined as continuous variables together with blood pressure (BP) and body mass index (BMI) to find out a simple method to predict echocardiographic (ECHO)-LVH. An age- and gender-stratified population-based sample of men (n = 121) and women (n = 135) aged 35–64 years enriched with newly diagnosed untreated hypertensive men (n = 138) and women (n = 97) in the Turku area in south-western Finland was studied. Major findings. Cornell voltage (or Cornell product), systolic BP (SBP) and BMI were all independent determinants of ECHO-LVH and left ventricular mass (LVM) indexed by height (LVMI). According to multivariate regression analyses with Cornell voltage (Cornell product), BMI and BP as explanatory variables, the three determinants explained 46–48% (47–49%) of the variation in LVMI among men and 50–54% (52–57%) among women. Score tables were constructed to estimate the probability of LVH. The estimated probability of ECHO-LVH increased in men gradually from 0% to 81% (79%) along with increased Cornell voltage (Cornell product) tertiles and in women respectively from 0% to 95% (97%). Conclusion. The sensitivity of ECG to detect ECHO-LVH can be markedly enhanced by using ECG amplitudes and products as continuous variables. The risk tables using Cornell voltages or products, BMI and SBP enable an easy and effective way to estimate the probability of ECHO-LVH.
Introduction
Left ventricular hypertrophy (LVH) is associated with increased cardiovascular morbidity and mortality, and is traditionally detected by electrocardiography (ECG), radiological examinations and since the 1980s mainly echocardiography (ECHO) (Citation1). LVH detected by EGG or ECHO predicts mortality independently of each other, indicating that they may in part carry different prognostic information (Citation2–4). However, low sensitivity of the traditional ECG-LVH detection criteria to identify subjects with anatomic LVH has limited their usefulness in clinical practice (Citation5). Moreover, the ability of ECG to discover LVH from overweight subjects decreases along with increasing obesity and subcutaneous fat, particularly if ECG is recorded using only precordial leads (Citation6,Citation7).
Because of low sensitivity of different ECG criteria to detect anatomic LVH, it has been suggested that ECG should not be used as a surrogate for ECHO in detecting LVH in the general population (Citation8). On the contrary, decreases in ECG voltage and voltage product along with reductions in left ventricular mass (LVM) suggest that it could substitute ECHO in the follow-up of patients with diagnosed LVH (Citation5,Citation8).
Everyday clinical praxis and the majority of the published papers have used a “cut-off” strategy in ECG-LVH detection. Continuous ECG values are dichotomized as either LVH positive or negative. In this conversion, valuable information may be lost. Obesity, higher levels of sodium intake and elevated blood pressure (BP) are independent determinants of increased anatomic LVH (Citation9–12). Thus, examining continuous ECG together with simple patient-based LVH risk factors, like for example BP and obesity, might improve detection of ECHO-LVH.
The present paper introduces a method in which ECG voltages and voltage products are used as continuous variables together with gender, BP and body mass index (BMI) data to find out best possible ways to detect ECHO-LVH. Finally, we studied whether simple score tables with ECG, systolic BP (SBP) and BMI data could be created to assess the likelihood of ECHO-LVH.
Methods
From the Turku area in south-western Finland (200 000 inhabitants), age- and gender-stratified population-based sample of 252 individuals aged 35–64 years (called the population sample) and 275 newly diagnosed untreated hypertensive patients aged 35 and 54 years (called the hypertension sample) were recruited into the study. The inclusion criteria of the hypertensive population were a SBP or diastolic blood pressure (DBP), in the range of 180–220 mmHg or 100–120 mmHg, respectively, as measured in the primary healthcare. Patients with coronary artery disease, cerebrovascular disease, insulin-treated diabetes mellitus, haemodynamically significant valvular diseases or pregnancy were excluded from study. Subjects having left bundle branch block (LBBB), left anterior hemi block (LAHB), right bundle branch block (RBBB) or unsuccessful echocardiographic readings were further excluded. As a result, data was obtained from 481 Caucasian subjects, who make out the (). The study was conducted following the Second Declaration of Helsinki and was approved by ethical committee of the Social Insurance Institution of Finland. All subjects gave their informed consent.
Table I. Characteristics of the study subjects.
Clinical BP was measured by an experienced nurse. It was recorded in a seated posture with a mercury sphygmomanometer, always between 08:00 and 10:00 h, according to the guidelines of the American Society of Hypertension (Citation13). A cuff with a bladder width of 15 cm was used. Subjects were requested to refrain from heavy exercise in the morning and to avoid cola drinks, coffee, tea and smoking for at least 1 h before the measurement. BP was averaged over duplicate measures made in four separate sessions within 3 weeks (Citation14).
Home BP was self-measured with a semiautomatic oscillometric device (Omron HEM 705C). The device meets the criteria for accuracy according to the revised protocol of the British Hypertension Society and the revised standards of the Association for the Advancement of Medical Instrumentation. A cuff with a bladder width of 13 cm was used for subjects with an arm circumference of ≤ 35 cm, and a cuff with a bladder width of 15 cm was used for subjects with an arm circumference of > 35 cm. Preparations for self-measured home BP were the same as for clinic BP. Seated BP was measured twice, approximately at a 2-min interval every morning between 06:00 and 09:00 h and every evening between 18:00 and 21:00 h on 7 consecutive days. Home BP was determined as the mean of 14 duplicate measures.
Ambulatory BP was recorded with an auscultatory device (Suntech, Accutracker II) according to the guidelines of the Berlin Consensus Document. A cuff with the same bladder width as that used in home measurements was used. Correct position of the microphone was controlled when the recorder was fitted by use of three or more auscultatory readings with a mercury column sphygmomanometer connected to the recorder's BP cuff. Ambulatory BP was recorded during daytime (06:00 to 23:00 h) at 15-min intervals and during night-time (23:00 to 06:00 h) at 30-min intervals, and the results were processed as outlined in detail elsewhere (Citation14).
Body weight was measured in light clothing without shoes with an accuracy of 0.1 kg and height of 1 cm. ECG was recorded with digital ECG recorder (Marquette MAC) with paper speed of 25 mm/s and sensitivity of 10 mm/mV after 10 min of supine rest for 5 min while subjects were resting in a supine position and breathing with a controlled frequency of 15/min (0.25 Hz). Cornell voltage (SV3 + RaVL), Cornell product [men: Cornell voltage × QRS duration, women: (Cornell voltage + 0.6 mV)× QRS duration], Sokolow–Lyon voltage (SV1 + RV5/6) and Sokolow–Lyon product [(SV1 + RV5/6) × QRS duration] indexes were automatically calculated using measurements from the limb and precoridal leads V1–V6 of a standard 12-lead ECG.
Two-dimensional controlled M-mode echocardiographic examinations were performed with the use of an Aloca SSt-860 colour Doppler ultrasound device and 3.5-MHz phased-array transducer. All ECHO studies were performed and measured by the same experienced physician (HK); measurements were performed according to the recommendations of the American Society of Echocardiography (ASE) (Citation15). The leading edge to leading edge convention was used. Left ventricular echograms were measured at or immediately below the tips of mitral leaflets and averaged over at least three heart cycles. LVM was calculated using method by Troy et al. (Citation16), as 1.05×[(interventricular septal thickness in diastole+ left ventricular internal dimension in diastole + posterior wall thickness in diastole) −left ventricular internal dimension in diastole]. Corrected LVM was calculated with the equation developed by Devereux and co-workers: 0.80×(ASE cube LVM)+ 0.6 (Citation17). LVM was indexed using body surface area (Citation18–20). ECHO-LVH was defined as LVM index (LVMI) > 143 g/m in men and > 102 g/m in women.
Statistical analyses
The data were analysed by SAS for Windows 9.1 program. The associations between the LVM indices and ECG-LVH parameters, and selected patient characteristics were studied by Pearson correlations. To assess the independent determinants of ECHO and LVMI, multivariate regression models were formed with LVH (and LVMI) as dependent variables, and age and statistically significant correlates (Cornell voltages and products, BP, BMI) as independent determinants. First, the interactions between the independent variables were studied. As the interactions were non-significant, models with only main effects were formed.
Finally, to make the risk score tables of ECHO-LVH, the independent determinants of LVH of the multivariate analyses were categorized into three categories and multivariate regression models with LVH as dependent variable were calculated. Subjects were categorized into three BMI groups [normal (< 25 kg/m2), overweight (25–30 kg/m2), obese (> 30 kg/m2)], three SBP groups [normal (< 120 mmHg), prehypertensive (120–139 mmHg) and hypertensive (> 140 mmHg)] and into three groups according to tertiles of Cornell voltages (men and women separately) and products (men and women combined). The interactions were again checked and found non-significant. Thus the effects of independent determinants were additive. The coefficients of multiple regression models were used to calculate the predicted prevalence of LVH for males and females.
Results
Characteristics of the study subjects are presented in . Of all men, 30% had normal weight (BMI < 25 kg/m2), 47% were overweight (BMI between 25 and 30.0 kg/m2) and 23% were obese (BMI > 30 kg/m2). In women, the respective figures were 39%, 39% and 22%. Of all men, 29% had a LVMI > 143 g/m and of all women 36% had a LVMI > 102 g/m, the thresholds of which were chosen as criteria for ECHO-LVH (Citation16). The prevalence of ECHO-LVH was 10% in men and 26% in women of the population sample, and 46% in men and 51% in women of the hypertensive sample. Cornell and Sokolow voltages and products were significantly higher among hypertensive men and women as compared with men and women of the population sample ().
The univariate correlations of patient characteristics with LVM and LVMI are presented in .
Table II. Correlations between left ventricular mass and selected patient characteristics.
All correlations were similar (p > 0.05) among men and women in the hypertensive and population samples, except for higher correlations (p < 0.05) between BP and LVM or LVMI among women in the population sample compared with women in the hypertensive sample. The results allowed combining the two populations in the further analyses to increase variations in the measured parameters. No associations were found between age and LVM or LVMI.
BMI, BP and Cornell voltages and products showed high correlations with LVM and LVMI (). SBP, regardless of the measuring methods (home BP, office BP or 24-h ambulatory BP) correlated slightly stronger with LVM and LVMI than did DBP. Sokolow–Lyon product showed weak though significant correlations with LVM and LVMI. Sokolow–Lyon voltage had no correlation with LVM or LVMI (). Accordingly, BMI, SBP, Cornell product and Cornell voltages were chosen as determinants for further analysing.
A linear regression was performed separately for men and women to identify independent determinants of ECHO-LVH and LVMI ( and Supplementary Table I to be found online at http://informahealthcare.com/doi/abs/10.3109/08037051.2013.803313). With ECHO-LVH or LVMI set as the dependent variable, higher BMI, higher SBP and higher Cornell voltages (or Cornell products) were independently associated with higher prevalence of ECHO-LVH () and LVMI (Supplementary Table I to be found online at http://informahealthcare.com/doi/abs/10.3109/08037051.2013.803313). With Cornell voltages/Cornell products, BMI and BP (office, home and 24-h ambulatory SBP), these three determinants explained 33–36% (34–37%) of the variance in the prevalence of ECHO-LVH in men and 37–41% (37–41%) in women (). Respectively these parameters explained 46–48% (47–49%) of the LVMI variance in men and 50–54% (52–57%) in women (Supplementary Table I to be found online at http://informahealthcare.com/doi/abs/10.3109/08037051.2013.803313).
Table III. Multivariate linear regression models for the prevalence of echocardiographic left ventricular hypertrophy (ECHO-LVH)a in men (n = 253) and women (n = 228).
and present the predicted prevalence of ECHO-LVH in men and women according to BMI, office SBP and Cornell voltage (Cornell product) categories. According to the score tables, the estimated prevalence of ECHO-LVH increased in men gradually from 0% to 81% (79%) along with increased categories of BMI, SBP and Cornell voltages (Cornell products) and in women respectively from 0% to 95% (97%).
Table IV. The predicted left ventricular (LVH) prevalencea by body mass index (BMI), office systolic blood pressure (SBP) and Cornell voltage categories.
Table V. The predicted left ventricular (LVH) prevalencea by body mass index (BMI), office systolic blood pressure (SBP) and Cornell product categories.
Discussion
Our study shows that BP, BMI and Cornell voltages or products are independent determinants of LVMI and the prevalence of ECHO-LVH. Normal BMI, normal SBP and Cornell voltage or product in the lowest tertile virtually excluded LVH. Respectively, ECHO-LVH was present in almost every hypertensive overweight women and in 80% of hypertensive overweight men, who had Cornell voltage or product in the highest tertile. Cornell voltage and product have a LVH predictive power also when their values are below the generally used cut-off points of LVH. Results of the present study clearly suggest that the ECHO-LVH detection ability of Cornell voltage and product improves when they are used as continuous variables and interpreted together with other determinants of LVH. The presented score tables can be used to estimate the probability of LVH among middle-aged subjects. Either Cornell voltage or product can be used. It is also noteworthy that the method of BP measurement (office, self-measured home and ambulatory) is not very critical in the assessment of LVH when repeated measurements and appropriate techniques are used as was done in this study (Citation14).
The most studied ECG-based LVH criteria are Sokolow–Lyon voltage, Cornell voltage and product, and the 12-lead voltages (Citation8,Citation21,Citation22). They are easy to calculate but unfortunately quite insensitive when using generally accepted cut-off points of LVH. ECG-based LVH detection usually improves when independent ECG parameters (time voltages and QRS-duration) are combined (Citation21–23). Therefore, we decided to calculate both plain voltage and voltage–QRS products in the present study. We found that Cornell voltage and product were superior to Sokolow voltage and product in the detection of ECHO-LVH. This is not a surprise because the adoption of limb lead makes Cornell voltage less sensitive to pericardial fat (Citation23,Citation24). However, there were no clear differences between Cornell voltage and product in detecting ECHO-LVH. ECG-based LVH approach has been used also in large studies; for example, Cornel voltage and product were used to ascertain LVH in the LIFE study (Citation25). These observations clearly support the use of Cornell voltage or product in the detection of LVH among middle-aged subjects.
LVMI increases along with BP and obesity. Both factors are independent determinants of LVMI and ECHO-LVH (Citation7,Citation25). Quite recently Ang et al. (Citation7) published a clinical score to identify patients with LVH using simple patient-based parameters such as age, BMI, history of hypertension, previous myocardial infarction, high SBP and the presence of bundle branch block (BBB). The method was developed in patients with coronary artery disease and validated in subjects with peripheral arterial disease. In this patients group, conventional Cornell voltage, when used as dichotomized to LVH positive and negative, was inferior to BBB in detecting ECHO-LVH and was not included in the final score. To our knowledge, a score-table type approach by using common ECG dependent LVH determinants of ECHO-LVH continuously together with simple patient characteristics, such as BMI and BP, and developed in an apparently healthy population, has not been published. In that respect, the present paper introduces a new dimension into ECG-based LVH detection.
ECHO as a basic LVH measurement tool has some limitations, but we adopted the ECHO method because of its availability and cost-efficacy (Citation16,Citation24). LVM indexed by height (LVMI) was calculated and widely accepted Framingham thresholds levels for LVH were chosen (Citation19,Citation20). Because all ECHO examinations were performed and measured by a single experienced physician (HK), the variation in the LVH detection was minimized. The coefficient of variation for the measurement of LVM has shown to be good (5.8%) in our laboratory (Citation26).
ECG has been widely used in the detection of LVH nearly for a century (Citation27–29). Albeit specific, it suffers from low sensitivity, which may be overcome by redefinition of the ECG criteria or adopting table-type extensions like the one presented in our study (Citation22). In a recent paper, Casiglia et al. (Citation8) were critical against the use of ECG in detecting LVH in the general population. They have analysed the results from a survey of relatively aged patients (mean 65.8 years) whose average BP values were relatively high (SBP 160.5 and DBP 89.9 mmHg). ECHO detects myocytes plus fibrous matrix while ECG detects only electrically active vital myocytes. Results and conclusions by Casiglia and colleagues reflect aged hypertensive patients in whom vital myocytes are probably partly replaced by a fibrous matrix. In line with these observations, Tsiachris and colleagues (Citation30) showed that sensitivity of several ECG on-off LVH criteria to detect ECHO-LVH decreases along with increasing age (Citation30). Our study subjects were younger and did not suffer from CHD or haemodynamically significant valvular disease. This may explain the relatively high correlation we found between ECHO and ECG-LVH.
Data from the Multi-Ethic Study of Atherosclerosis (MESA) showed that the sensitivity and overall performance of on-off ECG-LVH criteria also varies by ethnicity (Citation28). African Americans showed the highest sensitivity and overall performance, Caucasians the lowest with Chinese and Latin populations in-between. The performance of 14 different ECG-LVH criteria showed considerable variation between the four ethic groups.
Limitations
The study sample was a bit modest and BP of the study patients perhaps too well controlled to represent fully an average BP patient. The score tables are going to be checked using a larger material in the near future. Magnetic resonance imaging (MRI) is a reproducible and accurate method to rule out LVH (Citation31). With MRI, we might have yielded perhaps more accurate LVH detection, but MRI was not a realistic alternative for our study. ECG criteria may contain some ethnic variations. The reliability of our score table approach should also be ascertained in different age groups of subjects and in different populations.
Conclusions
Cornell voltage and product are easy to measure, readily available and the methods are quite inexpensive. Their sensitivity for detecting LVH can be markedly enhanced by using ECG amplitudes or products as continuous variables and interpreting them together with other patient-based determinants of ECHO-LVH such as BMI and SBP. The risk tables presented in this paper enable an easy and effective way to detect subjects with ECHO-LVH. The score table approach of the present study need to be examined also in larger study populations, different ethnic groups and in elderly subjects.
Supplementary Table I
Download PDF (82.9 KB)Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.
References
- Popp RL. Echocardiography. N Engl J Med. 1990;323: 165–172.
- Devereux BB. Is the electrocardiogram still useful for detection of left ventricular hypertrophy?Circulation. 1990; 81:1144–1146.
- Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP. Prognsotic implications of echocardiographically determined left ventricular mass in the Farmingham Heart Study. N Eng J Med. 1990;322:1561–1566.
- Sundström J, Lind L, Ärnlöv J, Zethelius B, Andrén B, Lithell HO. Echocardiographic and electrocardiographic diagnoses of left ventricular hypertrophy predict mortality independently of each other in a population of elderly men. Circulation. 2001;103:2346–2351.
- Okin PM, Devereux RB, Liu JE, Oikarinen L, Jern S, Kjeldsen SE, et al. For the LIFE study investigators: Regression of electrocardiographic left ventricular hypertrophy predicts regression of echocardiographic left ventricular mass: The LIFE study. J Hum Hypertens. 2004;18: 403–409.
- Tochikubo O, Miyajima E, Shigemasa T, Ishii M. Relation between body fat-corrected ecg voltage and ambulatory blood pressure in patients with eseential hypertension. Hypertension. 1999;33:1159–1163.
- Ang DSC, Fahey TP, Wright GA, Struthers AD. Development and validation of a clinical score to identify echocardiographic left ventricular hypertrophy in patients with Cardiovascular disease. Am J Hypertens. 2008;21: 1011–1017.
- Casiglia E, Schiavon L. Tikhonoff V, Bascielli A, Martini B, Mazza A, et al. Electrocardiographic criteria of left ventricular hypertrophy in general population. Eur J Epidemiol. 2008;23:261–271.
- Vottonen P, Husso M, Sipola P, Vanninen R, Peuhkurinen K, Magga J. Electrocardiographic left ventricular hypertrophy has low diagnostic accuracy in middle-aged subjects. Blood Press. 2007;16:328–334.
- Kupari M, Koskinen P, Virolainen J. Correlates of left ventricular mass in a population sample aged 36 to 37 years. Focus on lifestyles and salt intake. Circulation. 1994;89: 1041–1050.
- Jula A, Karanko H. Effects on left ventricular hypertrophy of long-term nonpharmacological treatment with sodium restriction in mild-to-moderate essential hypertension. Circulation. 1994;89:1023–1031.
- Vaidya A, Bentley-Lewis R, Jeunemaitre X, Adler GK, Williams JS. Dietary sodium alters the prevalence of electrocardiogram determined left ventricular hypertrophy is hypertension. Am J Hypertens. 2009;22:669–673.
- American Society of Hypertension. Public policy position paper: Recommendations for routine blood pressure measurement by indirect cuff sphygmomanometry. Am J Hypertens. 1992;5:207–209.
- Jula A, Puukka P, Karanko H. Multiple clinic and home blood pressure measurements versus ambulatory blood pressure monitoring. Hypertension. 1999;34:261–266.
- Sheldon GS, Edward DF. Limited echocardiography for hypertensive left ventricular hypertrophy. Hypertension. 1997;29:560–563.
- Troy BL, Pombo J, Rackley CE. Measurement of left ventricular wall thickness and mass by echocardiography. Circulation. 1972, 45:602–611.
- Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, Campo E, Sachs I, Reichek N. Echocardiographic assessment of left ventricular hypertrophy: Comparison to necropsy findings. Am J Cardiol. 1986;57:450–458.
- Dubois D, Dubois EF. A formula to estimate the approximate surface area if height and weight be known. Arch Intern Med. 1916, 17:863–871.
- Devereux RB, Lutas EM, Casale PN, Kligfield P, Eisenberg RR, Hammond IW, et al. Standardization of M-mode echocardiographic left ventricular anatomic measurements. J Am Coll Cardiol. 1984, 4:1222–1230.
- Krumholz HM, Larson M, Levy D. Prognosis of left ventricular geometric patterns in the Farmingham Heart Study. JACC. 1995;25:879–884.
- Okin PM, Roman MJ, Decereux RB, Kligfield P. Electrocardiographic indetification of increased left ventricular mass by simple voltagea-duration products. J Am Coll Cardiol. 1995;25:417–423.
- Molloy TJ, Okin PM , Devereux RB, Klighfield P. Electrocardiographic detection of left ventricular hypertrophy by the simple QRS voltage-duration product. J Am Coll Cardiol. 1992;20:1180–1186.
- Oikarinen L, Karvonen M, Viitasalo M Takala P, Kaartinen M, Rossinen J, et al. Electrocardiographic assessment of left ventricular hypertropht with time-voltage QRS and QRST-wave areas. J Hum Hypertens. 2004;18:33–40.
- Dahlöf B, Devereux R, Kjeldsen S, Stevo J, Beevers G, de Faire U et al. Cardiovascular morbidity and mortality in the Losartan Intervention for Endpoint reduction in hypertension study (LIFE): A randomised trial against atenolol. Lancet. 2002;359:995–1003.
- Abächerli R, Zhou L. Schmid JJ, Kobza R, Niggli B, Frey F, Erne P.Correlation of relationship assessment between left ventricular hypertrophy voltage criteria and body mass index in 41,806 Swiss conscripts. Ann Noninvasive Electrocardiol. 2009;14:381–388.
- Jula A, Rönnemaa T, Tikkanen I, Karanko H. Responses of atrial natriuretic factor to long-term sodium restriction in mild to moderate hypertension. J Intern Med. 1992;213:521–529.
- Liebson PR, Grandits GA, Primeas R et al. Echocardiographic correlates of left ventricular structure among 855 mildly hypertensive men and women in the Treatment of Mild Hypertension Study (TOMHS). Circulation. 1993;87:476–486.
- Jain A, Tandri H, Dalai D, Silliman EZ, Prineas RJ, Folsom AR, et al. Diagnostic and prognostic utility of electrocardiography for left ventricular hypertophy defined by magnetic resonance imaging in relation to ethnicity: The Multi Ethnic Study of Atherosclerosis (MESA). Am Heart J. 2010;159:652–658.
- Estes EH, Jackson KP. The electrocardiogram in left ventricular hypertrophy: Past and future. J Electrocardiol. 2009;42:589–592.
- Tsiachris D, Chrysohoou C, Oikonomou E, Lazaros G, Dimitriadis K, Maragiannis D, et al. Distinct role of electrocardiographic criteria in echocardiographic diagnosis of left ventricular hypertrophy according to age, in the general population: The Ikaria Study. J Hypertens. 2011;29: 1624–1632.
- Alfakih K, Walters K, Jones T, Ridgway J, Hall S, Sivananthan M. New gender-specific partition values for ECG criteria of left ventricular hypertrophy: Recalibration against cardiac MRI. Hypertension. 2004;44:175–179.