sign in sign up
<<
Home , Ventricular hypertrophy

Journal of Human Hypertension (2008) 22, 394–400 & 2008 Nature Publishing Group All rights reserved 0950-9240/08 $30.00 www.nature.com/jhh ORIGINAL ARTICLE The electrocardiogram is an unreliable method of identifying left ventricular in stable, treated patients

DSC Ang, LL Ti and AD Struthers Division of Medicine and Therapeutics, University of Dundee, Ninewells Hospital and Medical School, Dundee, UK

In (CAD), a potentially reversible sensitivity, specificity, positive predictive value and factor leading to cardiac death is left ventricular negative predictive value were calculated for each ECG hypertrophy (LVH). While the electrocardiogram (ECG) LVH criteria. The prevalence of echo LVH in the entire is a widely available way to diagnose LVH, its sensitivity CAD population was 43%. All the proposed ECG criteria and specificity has never been assessed in this were poor at identifying echo LVH. The Cornell product particular patient group where added ischaemic yielded the highest rate of change value but still missed changes on ECG might complicate things. Furthermore, up to 80% of the echo LVH cases. We conclude that in there are at least 11 different ECG criteria proposed to a group of stable, treated angina patients, ECG is an identify LVH. We sought to determine how many cases unreliable method of identifying LVH. As LVH is very of (echo) LVH would be missed if all common in this patient population, screening by means of these different ECG criteria were applied in a group of of echo might be indicated. This will enable intensified stable, treated angina patients. A total of 241 consecu- efforts to ensure LVH regression, which is associated tive patients with angiographically confirmed CAD with reduction in both cardiovascular morbidity and were prospectively recruited and 11 ECG criteria were mortality. assessed on each subject and compared with the Journal of Human Hypertension (2008) 22, 394–400; presence or absence of LVH on echo. Individual doi:10.1038/jhh.2008.18; published online 20 March 2008

Keywords: electrocardiogram; echocardiogram; left ventricular hypertrophy; sensitivity; specificity; coronary artery disease

Introduction valence would be around 30–35% in controls. Secondly, in a comparative study in CAD patients, Left ventricular hypertrophy (LVH) is associated Liao et al.4 showed that the independent relative with increased risk of cardiovascular morbidity, all 1 risk of cardiac death conferred by LVH was higher cause mortality and sudden cardiac death. This has (2.4) and accounted for more deaths than multi- been shown to be independent of conventional vessel coronary disease (relative risk ¼ 1.6) or LV cardiovascular risk factors. Recent evidence shows systolic dysfunction (relative risk ¼ 2.0). Therefore, that LVH is an important cause of cardiac death even it appears that LVH is a strong, if not the strongest, in patients with overt coronary artery disease (CAD). risk factor in CAD. In addition, recent studies have This is based on two main observations. Firstly, LVH also demonstrated that LVH regression is associated is more prevalent in this patient population than in with improved clinical outcome. In the hyperten- the general population. In a study carried out in a sive population, ECG and echo LVH regression has predominantly black population with established been associated with a significant reduction in both CAD, the prevalence of echocardiography (echo) 5,6 2 fatal and non-fatal cardiovascular events. In a LVH was as high as 50%. We recently found similar 7 3 recent review, Mancini et al. emphasized, ‘We now figures, while according to Framingham, the pre- have the ultimate documentation for both ECG LVH and echo LVH that reversal has an independent Correspondence: Dr DSC Ang, Division of Medicine and Thera- prognostic value, independent of therapy and blood peutics, University of Dundee, Clinical Pharmacology, Ninewells pressure’.7 This view was endorsed again by Hospital and Medical School, Dundee DD1 9SY, UK. Schillaci et al.8 E-mail: [email protected] Received 11 December 2007; revised 15 February 2008; accepted Nevertheless, detection of LVH in CAD remains a 16 February 2008; published online 20 March 2008 challenge. While echo is a reliable method for ECG LVH in stable angina patients DSC Ang et al 395 detecting anatomic LVH, it is less widely available Table 1 Demographic characteristics of the CAD population especially in the non US countries and obviously studied (n ¼ 241) requires technical expertize. Conversely, the 12-lead Variable Mean (s.d.) ECG is universally available, technically easy to perform and highly specific. The major limitation of Male subjects (%) 76 the ECG in other populations is its low sensitivity in Age (years) 64 (8) detecting LVH. Previous epidemiology studies have Body mass index (kg mÀ2) 29 (4) examined the sensitivity and specificity of various Office systolic BP (mm Hg) 136 (2) ECG criteria in picking up echo LVH in the general Office diastolic (mm Hg) 74 (10) 9,10 Haemoglobin (g per 100 ml) 13.6 (1.5) population and in hypertensives. Despite the Creatinine (umol lÀ1) 100 (21) importance of detecting LVH in CAD, the ability of Total Cholesterol (mmol lÀ1) 3.87 (0.76) different ECG criteria to detect echo LVH in this History of hypertension (%) 53 specific population has never been addressed. In History of type II diabetes (%) 14 History of hypercholesterolaemia (%) 51 addition, ischaemic related ST segment changes in Current smokers (%) 5 CAD might complicate ECG interpretation. Our Ex-smokers (%) 56 study aims to answer this question and in this Previous CABG (%) 33 analysis, we used a larger number of ECG criteria Previous percutanoues coronary intervention (%) 39 (11) than in previous studies. The main aim of our Previous (%) 25 study is to see how many cases of echo LVH would LVMI (BSA) be missed in a prospective cohort of 241 stable, Male subjects 127.3 (25.7) treated angina patients by applying each of these Female subjects 109.9 (21.8) many ECG criteria. Left ventricular ejection fraction (%) 58 (6) Treated with Aspirin 89% Statin 96% Methods b blockers 74% ACE inhibitors 56% A total of 351 white patients with the diagnosis of ARB 14% stable angina who had all undergone diagnostic Clopidogrel 15% coronary angiography (between 2002 and 2005) Calcium antagonist 27% Nicorandil 16% were consecutively recruited from the Nitrate 32% Department, Ninewells Hospital, Dundee. The in- Bendrofluazide 15% clusion criteria were a history of ischaemic chest Loop diuretic 14% pain and the presence of angiographically proven CAD (greater than 50% reduction in the cross Abbreviations: ARB, angiotensin receptor blocker; BP, blood pressure; BSA, body surface area; CAD, coronary artery disease; LVMI, left sectional diameter of a major coronary artery). ventricular mass index. Majority of them had been treated with stable antianginal medication for approximately 1 year Definition of a normal clinic blood pressure (Table 1 shows the breakdown of antianginal A ‘normal’ clinic BP was defined as 140/90 mm Hg treatment of the entire study population at the p as previously described in the British Hypertension time of study) and the majority had no ongoing Society (BHS) IV guidelines.12 In contrast, the symptoms of angina pectoris. The exclusion criteria ‘target’ BP for treating CAD was 130/80 mm Hg. include: p

(1) Recent hospital admission with -T 40.01 ng mlÀ1 in the last year. ECG assessment of LVH (2) Valvular, pericardial or congenital disease Simultaneous 12-lead electrocardiograms were ob- À1 (3) An impaired LV function (characterized by tained for analysis, calibrated to 20 mm mV and À1 regional wall motion abnormality or LV ejection 25 mm s . ECG analysis was performed manually fractiono45%) on ventriculography or echo. with an ECG image digitizer board, by a second researcher (IT) blinded to patient details and the echo results. Three QRS complexes’ per lead were All subjects attended a single clinic visit and analysed and the mean value used. We applied two underwent the following: history, past medical ECG criteria for LVH used in the Losartan Interven- history, clinic blood pressure (BP) measurement, tion for Endpoint Reduction (LIFE) study as well as , routine blood tests and trans- an additional nine previously reported ECG criteria thoracic echo. Ethical approval was obtained from for LVH, all of which are sufficiently straightforward the Tayside Committee of Medical Research Ethics to be used in routine clinical practice. The ECG and all participating subjects gave written informed criteria are as follows: consent. Other details of this study are already published, especially data on the relationship (1) Sokolow–Lyon voltage, SV1 þ (RV5 or RV6) between BP and LVH.11 X38 mm

Journal of Human Hypertension ECG LVH in stable angina patients DSC Ang et al 396 (2) Gender-specific Cornell voltage, SV3 þ RaVL constructed. We also separately analysed the 11 420 mm in women, 428 mm in men proposed ECG criteria for LVH in the hypertensive (3) Gubner–Ungerleider voltage, RI þ SIII subgroup of the CAD population. X25 mm (4) Lewis voltage, (RI þ SIII)À(SI þ RIII) X17 mm (5) RaVL 411 mm Results (6) Cornell product, Cornell voltage  QRS dura- tion 42440 in men, with addition of six to the Of the 351 consecutive patients recruited, 29 Cornell voltage in women patients were excluded due to the presence of (7) Sum of 12-lead voltage, 12-lead QRS sum impaired LV systolic function or valvular heart 4175 mm disease on echo. Of the remaining 322 patients, an (8) 12-lead QRS product, 12-lead QRS sum  QRS LVM was obtainable in 267 patients (83%). Those in duration 417 472 whom LVM could not be assessed did not differ significantly in any parameters from those in whom (9) RV5 or RV6 426 mm an adequate M-mode measurement was obtained. A (10) Largest R þ largest S in V1–V6 445 mm further 26 patients were excluded from the remain- (11) RV6/RV5 41 ing 267 patients due to the presence of morphology on ECG (defined as QRS Echocardiography duration 4150 ms). This yielded 241 patients Transthoracic echo was performed by one trained whereby the various ECG LVH criteria were com- operator (D Ang) using an Acuson (Sequia 512) pared with the presence/absence of echo LVH. imaging system with a 3V2C transducer who is Table 1 demonstrates the demographics of the blinded to the ECG results. The scan was performed patient population. The average duration between with the patient lying in the left lateral position at patients undergoing diagnostic coronary angiogra- approximately 45%. The intraobserver correlation phy and clinic attendance for electrocardiography/ for LVMI was r ¼ 0.95 (mean difference 1.0 g mÀ2, echo was 19 months. The reason for this delay was s.d. 8 g mÀ2). A Bland Altman plot (n ¼ 30) of two that we wanted stable patients whereby antianginals separate LVMI readings carried out on the same that might potentially alter LVM were stable for patient on two different occasions demonstrated a around 1 year. The CAD population consisted of close level of agreement between both readings. predominantly male patients with a mean age of 64 years; 53% of the patients had a history of hypertension. About 70% of the patients had Left ventricular hypertrophy assessment coronary artery revascularization (CABG/PCI) in Patients were studied with two-dimensional guided the past and majority were on optimal antianginals M-mode echo in standard views. All measurements and secondary prevention therapy during clinic were made according to the American Society of attendance. Table 2 presents the prevalence of echo Echocardiography recommendation at end diastole, LVH according to the two criteria used and the sex taken as the onset of QRS complex. The leading- distribution. Regardless of the method of indexation edge to leading-edge convention was used to used, the prevalence of LVH in this patient popula- measure interventricular septal thickness, LV inter- tion was worryingly high (46% when indexed to nal diameter and LV posterior wall thickness. Measurements were made over at least three Table 2 Prevalence of echocardiographic LVH separate cardiac cycles and the average was taken. Total group Hypertensive LV mass (LVM) was calculated according to the (n ¼ 241) subgroup formula of Devereux et al.13 0.80 (American Society (n ¼ 128) of Echocardiography LVM) þ 0.6 and indexed to body surface area to give an LVM index (LVMI). LVH LVM indexed to height2.7 was defined as LVMI greater than 110 g mÀ2 in Overall prevalence of LVH 73 83 female and greater than 134 g mÀ2 in male subjects.14 (%) 2.7 LVH prevalence in female 71 78 LVM was also indexed to height and LVH was subjects (%) À2.7 defined as LVMI 447 g m in female and LVH prevalence in male 74 85 450 g mÀ2.7 in male subjects.15 LVMI was not subjects (%) calculated in cases in which either poor image LVM indexed to body surface area quality or inadequate image alignment prevented Overall prevalence of LVH 46 55 accurate M-mode measurements from being made. (%) LVH prevalence in female 55 65 subjects (%) Statistics LVH prevalence in male 43 51 subjects (%) We calculated the sensitivity, specificity, positive predictive value and negative predictive value for Abbreviations: LVH, left ventricular hypertrophy; LVM, left ventri- each criterion. Receiver–operator curves were also cular mass.

Journal of Human Hypertension ECG LVH in stable angina patients DSC Ang et al 397 body surface area and 73% when indexed to patients with echo LVH will be undiagnosed if we height2.7). were to solely employ ECG as a diagnostic tool. Tables 3 and 4 demonstrate the sensitivity, Thirdly, the prevalence of echo LVH is high in our specificity, positive predictive value and negative study population and (46% when indexed to body predictive value and the area under the receiver– surface area and 73% when indexed to height2.7) and operator curve for the various ECG criteria in reiterates the potential importance of screening for detecting echo LVH. As expected, all ECG criteria this condition and finding novel ways to regress showed excellent specificity but very poor sensitiv- LVH especially in ‘normotensive LVH’. ity, resulting in poor rate of change values. Tables 5 ECG LVH has long been regarded as a harbinger of and 6 illustrate the same data in the hypertensive cardiovascular morbidity and mortality. Its associa- CAD subgroup. The same pattern emerges, whereby tion with adverse prognosis has been consistently all the 11 ECG criteria missed detecting majority of shown in both the general population and the the echo LVH cases. hypertensive cohort.16,17 However, despite its high specificity in detecting LVH, its clinical utility has Discussion been limited by its low sensitivity.9 Various ECG criteria for diagnosing LVH exist. Up to date, the There are several important aspects to our study. Sokolow–Lyon criterion remains the most widely Firstly, the value of various ECG criteria in detecting used criterion as it is easy to employ. In the most echo LVH has never been specifically assessed in a recent ESH–ESC guidelines, the Sokolow–Lyon CAD population. This is despite evidence suggest- criterion was recommended as part of all routine ing that angina patients have a higher prevalence of assessment of subjects with hypertension.18 In LIFE, echo LVH than the general population. Secondly, we the Sokolow–Lyon criterion and the Cornell product did a more comprehensive assessment of different were both used as entry criteria for the study.5 In our ECG criteria than previously published. Our find- study, the Cornell product consistently outper- ings suggest that a large proportion of angina formed the Sokolow–Lyon criteria in identifying

Table 3 Ability of ECG criteria to detect echo LVH (by height2.7) in the whole group (n ¼ 241)

Criterion Sensitivity Specificity Positive Negative predictive Area under the curve (%) (%) predictive value (%) value (%) (95% confidence interval)

Sokolow–Lyon 8.4 90 81.8 17.1 0.499 (0.417–0.582) Cornell voltage 6.8 87.6 60 25.8 0.473 (0.389–0.556) Gubner 5.1 95.4 75 27.1 0.502 (0.420–0.585) Lewis 11.4 86.2 69 26.4 0.488 (0.405–0.571) RaVL 411 6.8 93.8 75 27.1 0.503 (0.421–0.585) Cornell Product 10.2 96.9 90 28.5 0.536 (0.456–0.615) 12-lead QRS sum 2.8 100 100 27.5 0.514 (0.433–0.595) 12-lead QRS product 1.7 100 100 27.3 0.509 (0.427–0.590)

RV5 or RV6 426 3.4 98.5 85.7 27.4 0.509 (0.428–0.591) Maximum R+S in V1–V6 4.5 100 100 27.89 0.523 (0.442–0.603) RV6/RV5 41 9.1 90.8 72.7 26.9 0.499 (0.417–0.582)

Abbreviations: ECG, electrocardiogram; echo, echocardiography; LVH, left ventricular hypertrophy.

Table 4 Ability of ECG criteria to detect echo LVH (by BSA) in the whole group (n ¼ 241)

Criterion Sensitivity Specificity Positive predictive Negative predictive Area under the curve (%) (%) value (%) value (%) (95% confidence interval)

Sokolow–Lyon 9.9 91.5 50 54.3 0.507 (0.434–0.581) Cornell Voltage 9.9 93.1 55 54.8 0.515 (0.442–0.588) Gubner 5.4 95.4 50 54.1 0.504 (0.431–0.577) Lewis 12.6 88.5 48.3 54.2 0.505 (0.432–0.579) RaVL 411 8.1 94.6 56.3 54.7 0.514 (0.440–0.587) Cornell Product 14.4 96.9 80 57 0.557 (0.483–0.630) 12-lead QRS sum 4.5 100 100 55.1 0.523 (0.449–0.596) 12-lead QRS product 2.7 100 100 54.6 0.514 (0.440–0.587)

RV5 or RV6 426 4.5 98.5 71.4 54.7 0.515 (0.441–0.588) Maximum R+S in V1–V6 6.3 99.2 87.5 55.4 0.528 (0.454–0.601) RV6/RV5 41 9.9 91.5 50 54.3 0.507 (0.434–0.581)

Abbreviations: BSA, body surface area; ECG, electrocardiogram; echo, echocardiography; LVH, left ventricular hypertrophy.

Journal of Human Hypertension ECG LVH in stable angina patients DSC Ang et al 398 Table 5 Ability of ECG criteria to detect echo LVH (by height2.7) in the hypertensive subgroup (n ¼ 128)

Criterion Sensitivity Specificity Positive predictive Negative predictive Area under the curve (%) (%) value (%) value (%) (95% confidence interval)

Sokolow–Lyon 8.5 90.9 81.8 17.1 0.497 (0.364–0.630) Cornell Voltage 10.4 81.8 73.3 15.9 0.461 (0.324–0.598) Gubner 5.7 86.4 66.7 16 0.460 (0.322–0.598) Lewis 14.2 77.3 75 15.7 0.457 (0.320–0.594) RaVL 411 7.5 86.4 72.7 16.2 0.470 (0.333–0.606) Cornell Product 13.2 100 100 19.3 0.566 (0.445–0.688) 12-lead QRS sum 2.8 100 100 19.3 0.514 (0.384–0.645) 12-lead QRS product 1.9 100 100 17.5 0.509 (0.378–0.641)

RV5 or RV6 426 1 95.5 50 16.9 0.491 (0.357–0.626) Maximum R+S in V1–V6 4.8 100 100 18.2 0.533 (0.406–0.660) RV6/RV5 41 8.7 81.8 69.2 15.9 0.452 (0.313–0.590)

Abbreviations: ECG, electrocardiogram; echo, echocardiography; LVH, left ventricular hypertrophy.

Table 6 Ability of ECG criteria to detect echo LVH (by BSA) in the hypertensive subgroup (n ¼ 128)

Criterion Sensitivity Specificity Positive predictive Negative predictive Area under the curve (%) (%) value (%) value (%) (95% confidence interval)

Sokolow–Lyon 5.8 91.2 44.4 44.4 0.498 (0.397–0.599) Cornell Voltage 13 91.2 64.3 86.7 0.527 (0.426–0.627) Gubner 5.8 91.2 44.4 44.4 0.484 (0.383–0.586) Lewis 14.4 82.4 50 44.3 0.483 (0.381–0.584) RaVL 411 8.7 91.2 54.5 45.2 0.498 (0.397–0.599) Cornell Product 17.4 98.2 92.3 49.6 0.583 (0.485–0.681) 12-lead QRS sum 1.4 100 100 45.6 0.521 (0.421–0.622) 12-lead QRS product 2.8 100 100 45.2 0.514 (0.413–0.615)

RV5 or RV6 426 2.8 96.5 50 44.4 0.497 (0.395–0.598) Maximum R+S in V1–V6 8.5 98.2 85.7 46.3 0.533 (0.433–0.634) RV6/RV5 41 9.9 89.5 53.9 44.3 0.497 (0.396–0.598)

Abbreviations: BSA, body surface area; ECG, electrocardiogram; echo, echocardiography; LVH, left ventricular hypertrophy.

echo LVH regardless of the method of indexation 2.14, 2.4 and 2.1 after adjustment for other factors used or the presence/absence of hypertension. In including hypertension and the number of coronary fact, the Cornell product yielded the highest rate of vessels affected. Secondly, there is now strong change value compared with the other 10 ECG evidence to suggest that regression of LVM is criteria. Okin et al.19 showed that the simple product associated with improved prognosis in hypertensive of Cornell voltage  QRS duration significantly im- patients. Therefore, it is conceivable that similar proved the identification of LVH. Despite this, benefit would occur in CAD patients and future employing the Cornell product would still result in studies should specifically address this issue. In a a substantial under detection of LVH in our study substudy of LIFE, lower LVM during antihyperten- population (missing out 80% of the LVH cases). In sive treatment was associated with lower rate of the Framingham Heart Study, factors that were clinical end points independent of BP lowering known to reduce the sensitivity of fixed voltage effects.21 In a separate study, the lack of decrease or thresholds include obesity and smoking.9 Not the increase in LVM during antihypertensive treat- surprisingly, in our CAD population with a mean ment was associated with increased risk of cardio- body mass index of about 30 and a large proportion vascular events.22 Thirdly, a hypothetical analysis of patients with a smoking history, majority of the suggests that identifying and regressing LVH in ECG criteria fared poorly in terms of sensitivity. In normotensive patients with angina should be worth- addition, the LIFE investigators also showed that while and cost-effective.23 obese patients tended to have lower Sokolow–Lyon It is worth noting that the mean BP of our voltages than non-obese patients but had higher study population (136/74 mm Hg) is still not within Cornell product.20 the target recommendation of the most recent The importance of detecting LVH in CAD patients Joint British Society (JBS) 2 guidelines (o130/ is based on several reasons. Firstly, three previous 80 mm Hg).24 In fact, EUROASPIRE shows that many studies in CAD patients have demonstrated that the patients with CAD in the real world fail to achieve relative risk of cardiac death conferred by LVH was target BP and this is sometimes called ‘therapeutic

Journal of Human Hypertension ECG LVH in stable angina patients DSC Ang et al 399 inertia’.25 The value of detecting LVH in high risk Conflict of interest population is also important because studies have shown that the knowledge of yet another risk factor None. i.e. LVH would prompt physicians towards institut- ing more aggressive secondary prevention treat- ment.26 Studies are now being conducted, looking References at further ways to regress LVH over and above ‘AII 1 Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli withdrawal therapy’. This could mean the achieve- WP. Prognostic implications of echocardiographically ment of lower than conventional BP targets in those determined left ventricular mass in the Framingham with LVH (for example, systolic BP of 120 mm Hg) or Heart Study. N Engl J Med 1990; 322: 1561–1566. even an individualized BP target level that ensures 2 Ghali JK, Liao Y, Simmons B, Castaner A, Cao G, full LVH regression in that individual. In the 4E Cooper RS. The prognostic role of left ventricular study, the addition of an aldosterone blocker to hypertrophy in patients with or without coronary ACE-inhibitor achieved greater reduction in LVM. A artery disease. Ann Intern Med 1992; 117: 831–836. third possibility, which has been shown already to 3 Ang DSC, Pringle SD, Struthers AD. The cardiovascu- regress LVMin normotensive diabetic patients is lar risk factor, left ventricular hypertrophy, is highly 27 prevalent in stable, treated angina pectoris. Am J trientine—a copper chelating agent. Hypertens 2007; 20(10): 1029–1035. In summary, our study demonstrates that using 4 Liao Y, Cooper RS, McGee DL, Mensah GA, Ghali JK. any of ECG criteria proposed to identify LVH in The relative effects of left ventricular hypertrophy, CAD patients is unreliable and would result in a coronary artery disease, and ventricular dysfunction substantial under detection of this condition. The on survival among black adults. JAMA 1995; 273: best performing ECG criteria in our study was the 1592–1597. Cornell product but this method still missed 5 Dahlof B, Devereux RB, Kjeldsen SE, Julius S, Beevers approximately 80% of the LVH cases. This implies G, de Faire U et al. Cardiovascular morbidity and that echo screening for LVH in all CAD patients mortality in the Losartan Intervention For Endpoint might be necessary as LVH regression is associated Reduction in hypertension study (LIFE): a randomised trial against atenolol. Lancet 2002; 359: 995–1003. with reduction of both cardiovascular morbidity and 6 Verdecchia P, Schillaci G, Borgioni C, Ciucci A, mortality. Future studies should address the cost Gattobigio R, Zampi I et al. Prognostic significance of effectiveness of screening for fully regressing echo serial changes in left ventricular mass in essential LVH in CAD patients, especially in the non US hypertension. Circulation 1998; 97: 48–54. countries, whereby the availability of echo is limited 7 Mancini GB, Dahlof B, Diez J. Surrogate markers for and not all CAD patients undergo echo. : structural markers. Circulation 2004; 109: IV22–IV30. 8 Schillaci G, Pirro M, Mannarino E. Left ventricular hypertrophy reversal and prevention of diabetes: two birds with one stone? Hypertension 2007; 50: 851–853. What is known about this topic: 9 Levy D, Labib SB, Anderson KM, Christiansen JC, K In coronary artery disease, LVH is an important risk factor Kannel WB, Castelli WP. Determinants of sensitivity that leads to increased cardiac death. and specificity of electrocardiographic criteria for K The ECG remains a widely used tool in identifying LVH left ventricular hypertrophy. Circulation 1990; 81: because it is easily available. 815–820. K However, the sensitivity and specificity of various ECG LVH 10 Devereux RB, Koren MJ, de Simone G, Okin PM, criteria have never been assessed in this patient population Kligfield P. Methods for detection of left ventricular whereby added ischaemic changes might complicate interpretation. hypertrophy: application to hypertensive heart dis- ease. Eur Heart J 1993; 14(Suppl D): 8–15. What this study adds: 11 Ang DS, Pringle SD, Struthers AD. The cardiovascular K We analysed 11 different ECG LVH criteria in 241 stable, risk factor, left ventricular hypertrophy, is highly treated angina patients and correlated them with prevalent in stable, treated angina pectoris. Am J echocardiographic LVH. Hypertens 2007; 20: 1029–1035. K The prevalence of echo LVH in the entire CAD population 12 Williams B, Poulter NR, Brown MJ, Davis M, McInnes was 43%. All the proposed ECG criteria were poor at GT, Potter JF et al. Guidelines for management of identifying echo LVH. The Cornell product yielded the hypertension: report of the fourth working party of the highest ROC value but still missed up to 80% of the echo LVH cases. British Hypertension Society, 2004-BHS IV. J Hum K Majority of ECG LVH criteria are poor at identifying LVH in Hypertens 2004; 18: 139–185. CAD. As this condition is very common in this patient 13 Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, population, screening by means of echocardiography might Campo E, Sachs I et al. Echocardiographic assessment be indicated of left ventricular hypertrophy: comparison to ne- cropsy findings. Am J Cardiol 1986; 57: 450–458. 14 Devereux RB, Lutas EM, Casale PN, Kligfield P, Eisenberg RR, Hammond IW et al. Standardization of Acknowledgements M-mode echocardiographic left ventricular anatomic measurements. J Am Coll Cardiol 1984; 4: 1222–1230. We thank the British Heart Foundation for gener- 15 de Simone G, Devereux RB, Daniels SR, Koren MJ, ously funding this project. Meyer RA, Laragh JH. Effect of growth on variability of

Journal of Human Hypertension ECG LVH in stable angina patients DSC Ang et al 400 left ventricular mass: assessment of allometric signals in treatment and the prediction of major cardiovascular adults and children and their capacity to predict cardio- events. JAMA 2004; 292: 2343–2349. vascular risk. J Am Coll Cardiol 1995; 25: 1056–1062. 22 Muiesan ML, Salvetti M, Rizzoni D, Castellano M, 16 Kannel WB, Gordon T, Offutt D. Left ventricular Donato F, Agabiti-Rosei E. Association of change in left hypertrophy by electrocardiogram. Prevalence, inci- ventricular mass with prognosis during long-term dence, and mortality in the Framingham study. Ann antihypertensive treatment. J Hypertens 1995; 13: Intern Med 1969; 71: 89–105. 1091–1095. 17 Hsieh BP, Pham MX, Froelicher VF. Prognostic value of 23 Witham MD, Davies JI, Dawson A, Davey PG, Struthers electrocardiographic criteria for left ventricular hyper- AD. Hypothetical economic analysis of screening for trophy. Am Heart J 2005; 150: 161–167. left ventricular hypertrophy in high-risk normotensive 18 European Society of Hypertension-European Society of populations. QJM 2004; 97: 87–93. Cardiology Guidelines Committee. 2003 European 24 Hall RJ, Lefroy D, Simpson I, Mills P. JBS 2: Joint Society of Hypertension-European Society of Cardiol- British Societies’ guidelines on prevention of cardio- ogy guidelines for the management of arterial hyper- vascular disease in clinical practice. Heart 2005; tension. J Hypertens 2003; 21: 1011–1053. 91(Suppl 5): v1–52. 19 Okin PM, Roman MJ, Devereux RB, Kligfield P. 25 Okonofua EC, Simpson KN, Jesri A, Rehman SU, Electrocardiographic identification of increased left Durkalski VL, Egan BM. Therapeutic inertia is an ventricular mass by simple voltage-duration products. impediment to achieving the healthy people 2010 J Am Coll Cardiol 1995; 25: 417–423. blood pressure control goals. Hypertension 2006; 47: 20 Okin PM, Jern S, Devereux RB, Kjeldsen SE, Dahlof B. 345–351. Effect of obesity on electrocardiographic left ventricu- 26 Hall LM, Jung RT, Leese GP. Controlled trial of effect of lar hypertrophy in hypertensive patients: the losartan documented cardiovascular risk scores on prescribing. intervention for endpoint (LIFE) reduction in hyper- BMJ 2003; 326: 251–252. tension study. Hypertension 2000; 35: 13–18. 27 Cooper GJ, Phillips AR, Choong SY, Leonard BL, 21 Okin PM, Devereux RB, Jern S, Kjeldsen SE, Julius S, Crossman DJ, Brunton DH et al. Regeneration of the Nieminen MS et al. Regression of electrocardiographic heart in diabetes by selective copper chelation. left ventricular hypertrophy during antihypertensive Diabetes 2004; 53: 2501–2508.

Journal of Human Hypertension