Journal of Human Hypertension (2008) 22, 460–467 & 2008 Nature Publishing Group All rights reserved 0950-9240/08 $30.00 www.nature.com/jhh REVIEW The prognostic value of the ECG in hypertension: where are we now?

DSC Ang and CC Lang Division of Medicine and Therapeutics, Ninewells Hospital and Medical School, University of Dundee, Dundee, UK

In hypertension, the presence of left ventricular hyper- review article, we discuss the individual strength and trophy (LVH) is associated with increased risk of both weaknesses of the commonly used ECG criteria in cardiovascular morbidity and mortality. To date, the diagnosing LVH. In addition, we present the latest data electrocardiogram (ECG) remains the cornerstone of on the prognostic significance of ECG LVH and the LVH diagnosis in clinical practice because it is uni- survival differences conferred in different genders. In versally available, technically easy to perform and view of the recent Losartan Intervention for Endpoint highly specific. In the most recent European Society of Reduction in Hypertension trial, the prognostic benefit Hypertension/European Society of Cardiology guide- of LVH regression will also be addressed. Finally, with lines for the treatment of arterial hypertension, the the wider availability of echocardiography, the role of Sokolow–Lyon voltage criterion was recommended as combining both modalities to improve risk stratification part of all routine assessment of subjects with hyperten- in hypertension is reviewed. sion. However, the use of the ECG in the diagnosis of Journal of Human Hypertension (2008) 22, 460–467; LVH is somewhat limited by its poor sensitivity. In this doi:10.1038/jhh.2008.24; published online 24 April 2008

Keywords: ; echocardiography; left ; prevalence; prognosis and regression

Introduction How good is the ECG in detecting true anatomic LVH? Left ventricular hypertrophy (LVH) diagnosed on electrocardiography (ECG) is an ominous prognostic One major limitation of the ECG is its low sensitivity sign that predicts a high rate of cardiovascular (CV) for detection of LVH.1 For example, the prevalence events. In hypertension, this condition is of parti- of ECG-LVH detected by the Framingham method cular importance because it helps guide risk strati- in the general population was only 2%, which fication. Various ECG criteria for diagnosing LVH compared with a prevalence of 20% when echocar- exist (Table 1). They range from the more widely diography was used.2,3 The sensitivity of the ECG is employed fixed voltage criteria like the Sokolow– also dependant on the population in which it is Lyon criterion (sum of SV1 þ RV5X3.5 mV or max applied. Although the sensitivity of several ECG RV5/6X2.6 mV), the McPhie criterion (the sum of criteria for LVH may be acceptable, albeit low (from the tallest R and deepest S in the precordial 58% for the Romhilt–Estes score4 to 41% for the leadsX4.5 mV) and the gender-specific Cornell Cornell voltage5), among patients with disparate voltage (sum of leads RaVL þ SV3X2.8 mV in men cardiac conditions, the sensitivity of the ECG andX2.0 mV in women) to the more complicated consistently decreases in the general population. Romhilt–Estes score of five points or more, which For example, the sensitivity of the Cornell voltage combines QRS voltage and voltage-independent criterion was only 10% in men and 22% in women signs (QRS duration, intrinsicoid deflection in V5 in the Framingham Heart Study2 and only 12% in or V6, left , left-axis deviation and men and 19% in women in the Progetto Ipertensione left ventricular (LV) strain). Figure 1 illustrates an Umbria Monitoraggio Ambulatoriale (PIUMA) example of how these commonly used ECG criteria study.6 In a longitudinal study involving 923 are calculated. untreated hypertensive participants, Schillaci et al.7 showed that the performance of the Cornell voltage was superior to the Sokolow–Lyon voltage Correspondence: Dr DSC Ang, Division of Medicine and Thera- criterion in predicting echo LVH. In addition, a peutics, Ninewells Hospital and Medical School, University of modified sex-specific partition value of the Cornell Dundee, Dundee DD1 9SY, UK. E-mail: [email protected] voltage (2.4 mV in men and 2.0 mV in women) Received 23 August 2007; revised 16 March 2008; accepted 21 further improved its predictive value. Subsequent March 2008; published online 24 April 2008 studies evaluated the inclusion of conduction Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 461 Table 1 Various ECG LVH criteria employed

Name Criteria Cut-point for LVH

Sokolow–Lyon voltage SV1+RV5X3.5 mV or max RV5/6 X2.6 mV Cornell voltage RaVL+SV3 X2.8 mV (men) X2.0 mV (women) Modified Cornell voltage RaVL+SV3 X2.4 mV (men) X2.0 mV (women) Framingham adjusted Men: RaVL+SV3+0.0174  (age, 49)+0.191  (BMI, 26.5) X2.8 mV (men) Cornell voltage Women: RaVL+SV3+0.0387  (age, 50)+0.212  (BMI, 24.9) X2.0 mV (women) Minnesota code 3.1 RV5/V642.6 mV, RI/II/III/aVF 42.0 mV or RaVL 41.2 mV Lewis index (RI+SIII)À(RIII+SI) 41.7 mV Gubner and Ungerleider RI+SIII X 2.2 mV Sum of 12 leads Sum of max (R+S) amplitude in each of the 12 leads 417.9 mV McPhie criterion The sum of the tallest R and deepest S in the precordial leads 44.5 mV 12-lead product 12-lead sum voltage  QRS duration 17472 mm  ms Cornell product (RaVL+SV3)  QRS duration 2436 mm  ms Framingham score RI+SIII 42.5 mV, SV1/2+RV5/643.5 mV, SV1/2/342.5 mV+RV4/5/642.5 mV plus left ventricular strain pattern Perugia score Positivity of at least one of the following: (1) SV3+RaVL 4 2.4 mV (2) Left ventricular strain pattern (3) Romhilt–Estes point scoreX5 Romhilt–Estes point score (1) Any limb lead R or S X2.0 mV or X 5 points–definite LVH SV1/2X3.0 mV or RV5/6X3.0 mV (3 points) X 4 points–probable LVH (2) Left ventricular strain pattern (1 point) (3) Left atrial enlargement (3 points) (4) (5) QRS duration X 90 ms (1 point) (6) Intrinsicoid QRS deflection of X50 ms in V5/6 (1 point)

Abbreviations: BMI, body mass index; LVH, left ventricular hypertrophy. abnormalities with voltage criteria to improve the cohorts that have assessed the risk associated with sensitivity of the ECG. The Cornell voltage duration ECG-LVH and these have reinforced in general, the product (Cornell voltage  QRS duration) further prognostic value of ECG-LVH. The prevalence of enhances sensitivity of the ECG while maintaining ECG-LVH and the estimates of risk associated with high specificity, with a sensitivity of 51 versus 31% ECG-LVH vary however, quite considerably in these for Sokolow–Lyon voltage criterion when examined studies and the variability of these estimates is at a matched specificity of 95%.8 Despite this, the related to the different populations studied and the original Sokolow–Lyon voltage criterion9 is the still differences in the diagnostic criteria used to define the most commonly used ECG criterion because ECG-LVH and to define the clinical end points of more complex criteria and scoring systems are interest. Furthermore, in most of these studies, the difficult to apply in clinical practice. The Soko- number of individuals with ECG-LVH and the low–Lyon voltage criterion was successfully used number of outcome events were relatively small, together with the Cornell voltage duration product accounting for difficulties in obtaining precise in detecting patients with LVH in the Losartan estimates of risk, in spite of the large number of Intervention for Endpoint Reduction in Hyperten- subjects studied and the prolonged periods of sion (LIFE) study. On the basis of the LIFE study, the follow-up. It should be noted that there have only European Society of Hypertension guidelines re- been a few published reports on the risk associated commend the use of the Sokolow–Lyon voltage with ECG-LVH based solely on the Sokolow–Lyon criterion and the Cornell voltage duration product voltage criterion. The seminal data for the prognos- for the assessment of LVH.10 A summary of the tic value of the ECG (Sokolow–Lyon voltage criter- epidemiology studies employing various ECG LVH ion) came from the Framingham Heart Study more methods in identifying anatomic LVH is shown in than 35 years ago.16,17 In this longitudinal popula- Table 2. tion-based study, the risk of fatal and nonfatal CV morbid events increased from threefold to eightfold higher in middle-aged persons with ECG-LVH Voltage ECG LVH criteria as a prognostic compared to normal adults of similar age, and this indicator finding remained significant after adjustment for co- existent risk factors. It should be noted that these There have been a number of large population-based findings relate to ST-T abnormalities (LV strain epidemiological studies and studies in hypertensive defined as ST-J point depressionX0.1 mV þinverted

Journal of Human Hypertension Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 462

Sokolow-Lyon voltage SV1 + RV5 1.5mV + 2.4mV = 3.9mV

Cornell voltage RAVL + SV3 1.2mV + 1.0mV = 2.2mV

Cornell product (RAVL + SV3) QRS duration 22mm × 130ms = 2860m m ms

12 lead sum 12 lead sum 15.2 mV

12 lead product 12 lead sum of QRS amplitude 152mm × 130ms =19760m m ms

McPhie criterion RV4 + SV1 2.7mV + 1.5mV = 4.2mV

Gubner-U voltage RI + SIII 1.3mV + 1.1mV = 2.4mV

Lewis voltage (RI + SIII) – (RIII + SI) (1.3mV + 1.1mV) – 0mV = 2.4 mV

Figure 1 An example of calculating various electrocardiogram (ECG) left ventricular hypertrophy (LVH) criteria of a 67-year-old man with a long-standing history of hypertension.

Table 2 Summary of various epidemiology studies employing different ECG LVH methods in identifying anatomic LVH

Study Criteria Prevalence (%) Sensitivity/ specificity (%)

Framingham Heart Study2 Framingham score 2.9 in men 6.9/98.8 (general population, n ¼ 524) 1.5 in women PIUMA7 (hypertensive population, n ¼ 923) Modified sex-specific Cornell voltage 22/95

LIFE11 (treated hypertensive, n ¼ 8854) Sokolow–Lyon or Cornell voltage duration 11 product +/À strain HOPE12 (high-risk population, n ¼ 9541) Sokolow–Lyon +/À strain 8.3 Molloy et al.8 (post-mortem series, n ¼ 220) Cornell voltage duration product 51/95 Copenhagen Heart Study13 Minnesota code 3.1 19.3 in men (general population, n ¼ 11 634) 5.6 in women Okin et al.14 (mild hypertension, n ¼ 389) Cornell voltage duration product 37/96 12-lead product 50/96 PIUMA15 (hypertensive population, Cornell/strain (C/S) index 16.3 33.6/91 n ¼ 2190) PIUMA6 (hypertensive population, Perugia score 17.8 n ¼ 1717)

Abbreviations: HOPE, Heart Outcomes Prevention Evaluation; LIFE, Losartan Intervention for Endpoint Reduction in Hypertension; PIUMA, Progetto Ipertensione Umbria Monitoraggio Ambulatoriale.

Journal of Human Hypertension Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 463 in leads I, aVL or V5/V6) in addition to the shown that patients who were diagnosed with LVH Sokolow–Lyon voltage criterion defined as ‘definite at baseline and went on to develop increasing ECG LVH’ by the Framingham investigators. More recent voltage or re-polarization abnormalities had an supportive data came from the Heart Outcomes increased risk of subsequent CV disease events.19 Prevention Evaluation (HOPE) study involving high- Nevertheless, it is important to note that previous risk individuals with established CV disease.12 In evidence has not consistently demonstrated the this landmark trial, LVH defined solely by the increased risk of CV mortality associated with Sokolow–Lyon voltage criterion was found to be an voltage-only LVH criteria. In the Copenhagen City independent predictor of all cause deaths, CV death Heart Study (a population-based study), men young- and .12 The association with risk of heart er than 55 years of age with voltage-only LVH failure was particularly pronounced (a twofold (employing the Minnesota code) had the same risk increase), considering the fact that all patients had of developing future and normal LV ejection fraction at study entry. A link as those with a normal ECG.13 between diastolic dysfunction and ECG LVH was This observation was supported by Hsieh et al.20 postulated. In the PIUMA study, the prognostic role who compared 17 different ECG methods of identi- of the Sokolow–Lyon voltage criterion was less fying LVH. They demonstrated that a composite of convincing. When compared with various other ECG criteria (Framingham, Perugia and Romhilt– ECG LVH criteria, the Sokolow–Lyon voltage criter- Estes point scores) was more strongly predictive of ion yielded the lowest hazard ratio (HR 1.19) in CV mortality when compared to voltage-only LVH predicting fatal CV morbid events among hyperten- criteria.20 In addition, the Framingham investigators sive subjects. This was in contrast with other more found that the excess CV risk associated with predictive methods like the Cornell voltage (HR voltage-only LVH was eliminated when adjusted 1.34) and Romhilt–Estes score (HR 2.63).6 In this for hypertension.21 They postulated that the voltage- study the Perugia score carried the highest popula- only LVH in hypertensive individuals reflects tion-attributable risk for CV morbidity and mortality chiefly the severity and duration of the associated compared with classic methods for detection of LVH hypertension and therefore signifies ventricular (Figure 2). muscle hypertrophy without myocardial damage. In addition to employing conventional cut-points to define LVH, the application of the voltage criteria as a continuous variable has also been evaluated as a Left ventricular strain pattern as a continuous prognostic indicator. In the LIFE study, prognostic indicator every 5 mm increase in the Sokolow–Lyon voltage was associated with a significant increase in both Although the evidence regarding the association CV mortality and morbidity. This also applied to between voltage-only LVH criteria and adverse increasing Cornell voltage duration product. Conse- clinical outcomes is less robust, there is good quently, less severe ECG LVH using the above evidence to consistently show a strong relationship methods translated to a reduced likelihood of between the presence of both voltage and re- subsequent events.18 It is fair to say that the polarization abnormalities and CV mortality. In a prognostic value of an increase in voltage is not recent longitudinal study evaluating 19 434 male new as the Framingham Heart Study had previously patients with a mean follow-up of 7±4 years, the presence of LV strain pattern in addition to voltage criteria yielded an HR of 3.9 after adjustment for 100 6 conventional cardiac risk factors.20 In Framingham, 90 the presence of a strain pattern on ECG was the most 5 important predictor of future CV disease across all 80 age and sex.21 In LIFE, ECG strain was a significant Fatal + Nonfatal 4 predictor of CV death, non-fatal myocardial infarc- 70 Cardiovascular Events tion or stroke after adjustment for standard CV risk P<0.0001 3 60 LVS absent factors, baseline blood pressure (BP) and severity of * 22 LVH present 2 ECG LVH in the setting of aggressive BP lowering 50 (Figure 3). ECG strain also identified hypertensive * Romhilt-Estes score > =5, or 40 typical left ventricular strain pattern, or 1 patients in LIFE who were at increased risk of RaVL – SV3 > 2.4 mV (men) developing chronic heart failure (CHF) and dying as or > 2.0 mV (women) 30 0 (per 100 patient-years) Rate of Events 11

Probability of Event-Free Survival, % Probability of Event-Free a result of CHF. Numerous mechanisms have been 0 100 200 300 400 500 LVH LVH postulated to account for the strong association absent present Time to Event, weeks between re-polarization abnormalities and increased CV mortality. One possible mechanism is the Figure 2 Event-free survival curves for total cardiovascular (CV) development of myocardial fibrosis. Evidence sug- events in the study cohort grouped by the presence or absence of gests that myocardial fibrosis may impair micro- left ventricular hypertrophy (LVH) according to the Perugia score. 23 The rate of major CV morbid events was significantly higher in the vascular coronary reserve, leading to ischaemia. subset with LVH (with permission from Verdechia et al.6). Offering support to this observation is a study which

Journal of Human Hypertension Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 464 25 ECG LVH and prognosis: does gender Strain + matter? * Strain - 20 21.0 It is important to note that the majority of available ECG LVH criteria do not employ gender-specific cut- 15 * offs. This is contrary to the definition of echocardio- 14.6 graphic LVH that frequently incorporates different LV mass cut-offs for men and women. The Cornell 10 11.2 * * 10.2 voltage criterion, which has been shown in some 9.1 * 8.1 8.3 studies to be superior to the Sokolow–Lyon voltage

5-Year Event Rate (%) 7 5 5.8 criterion is the only method that employs sex- 4.3 3.9 specific cut-offs. In a recent study, the sensitivity and specificity of four ECG criteria were evaluated 0 29 Composite Endpoint CV Mortality Myocardial Infarction Stroke All-Cause Mortality against LV mass measured by cardiac MRI. At a Figure 3 Five-year event rates according to the presence or matched specificity of 95%, the Sokolow–Lyon absence of strain on the baseline electrocardiogram (*Po0.001 voltage criterion yielded the highest sensitivity in versus patients without strain) (with permission from Okin women (26.2%), the Cornell voltage criterion had et al.22). the highest sensitivity in men (26.2%) and the Cornell product criterion had a relatively high sensitivity in both men and women (25 and 23.8%). demonstrated that patients with LV strain experi- Few studies have specifically addressed the differ- enced more episodes of exercise-induced ECG ential prognostic value of LVH in men and women. abnormalities and more reversible thallium perfu- Recent literature suggests that increased LV mass is a sion abnormalities.24 In addition to myocardial stronger risk factor in women than in men. This is ischaemia, LV systolic dysfunction characterized particularly true of echocardiographic (echo) LVH. by lower stress-corrected midwall shortening was Liao et al.30 described an unfavourable prognosis in also more prevalent in patients with ECG strain terms of total death and cardiac-related deaths in pattern.25 On the contrary, the relationship between black women with echo LVH compared to black men strain pattern and LV diastolic abnormalities is with echo LVH. Data from the Glasgow Blood weak. In a recent subgroup of the LIFE study, Pressure Clinic31 also support this observation of a combination of diastolic parameters (E/A ratio, survival disadvantage conferred by LVH in women. E wave deceleration time, atrial filling fraction and In the presence of ECG LVH, women had a substan- isovolumic relaxation time) and different LV filling tially higher risk of dying from CV causes. The patterns (normal filling pattern, abnormal relaxa- Framingham Heart Study highlighted an increased tion, pseudonormal filling pattern and restrictive risk of stroke and CVD mortality among women with pattern) assessed by means of Doppler echocardio- ECG LVH.32 In a recent prospective study involving graphy did not differ significantly between patients 6668 hypertensive patients with a mean follow-up with versus those without strain pattern.26 duration of 11.2 years, the Sokolow–Lyon voltage criterion predicted stroke mortality in women but not in men.33 The difference in survival between genders conferred by ECG LVH has no clear underlying Prolonged QTc and QRS duration predicts mechanism but a higher prevalence of the concentric prognosis geometric abnormality in women than men is a possible explanation. The latter carries the highest In addition to the conventional voltage LVH criteria 34 and strain pattern, recent studies have also con- risk and eccentric hypertrophy an intermediate risk. firmed the role of QRS and QT prolongation in the risk stratification of hypertensive patients. In the ECG LVH and Echo LVH: is there a role PIUMA study, Schillaci et al.27 studied 2110 in combining both? hypertensive patients over a mean duration of 5.3 years. This study found a significant but weak With the wider availability of echocardiography, association between corrected QT interval (QTc) there may be a role of combining ECG and echo LVH and the Cornell voltage (r ¼ 0.06, P ¼ 0.006). In in evaluating prognosis. In the second Strong Heart addition, the presence of prolonged QTc (defined Study, 2193 American Indians underwent both ECG asX450 ms in women andX440 ms in men) was and echocardiography and were followed up over an associated with a twofold increase in risk of average of 3 years. The presence of both echo LVH coronary events and CV deaths, independent of and ECG ST depression was associated with a 6.3- traditional cardiac risk factors and ECG LVH. In the fold increased of CV death in this group of patients35 LIFE study, prolonged QRS duration further risk (Figure 4). In addition to this, Sundstrom et al.36 stratified patients with established ECG LVH. In this showed that the prognostic value of ECG LVH study, increased QRS duration was an independent (employing the Cornell product) to some extent is predictor of both CV and all cause mortality.28 independent of echocardiographic LV mass index in

Journal of Human Hypertension Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 465 40 Log rank=86.1 STD+/LVH+ P<0.001 (n=52) 30

20

STD-/LVH+ or STD+/LVH- 10 (n=238) Cardiovascular Mortality (%) Cardiovascular STD-/LVH- (n=1903) 0 0 12 24 36 48 Follow-Up (Months)

70 Log rank=100.4 STD+/LVH+ (n=52) 60 P<0.001

50

40

STD-/LVH+ 30 or STD+/LVH- (n=238) 20 All-Cause Mortality (%) 10 STD-/LVH- (n=1903) 0 012243648 Follow-Up (Months) Figure 4 Kaplan–Meier plots of cumulative cardiovascular mortality (a) and all cause mortality (b) according to combined ST depression (STD) and echocardiographic left ventricular hypertrophy (LVH) criteria. ST depression defined asX0.5 mV in any lead (excluding lead aVR). Echocardiographic LVH (LVH þ ) is defined as LV mass4104 g mÀ2 in women and4116 g mÀ2 in men (STDÀ/LVHÀ represents both negative; STDÀ/LVH þ or STD þ /LVHÀ represents either positive; STD þ /LVH þ represents both positive) (with permission from Okin et al.35). predicting total mortality in a population-based to whether regression or reversal of ECG LVH is sample of 475 men. The author concluded that echo associated with improved prognosis. Various obser- LVH and ECG LVH carried different prognostic vational studies suggest that LVH reversal is bene- information.36 The combination of ECG and echo ficial beyond BP reduction and treatment by LVH also predicted future risks of ischaemic stroke. demonstrating that CV events occur in higher Kohsaka et al.37 demonstrated that LVH diagnosed proportion of individuals in whom LVH progresses based on the Cornell voltage criteria was associated than regresses.19,38 These findings suggest that the with an increased risk of ischaemic stroke after level of LV mass during antihypertensive treatment adjustment for other stroke risk factors and echo reflects independent information regarding disease LVH. The author suggested that ECG LVH may progression or its control. Confirmation of this detect electrical abnormalities induced by the hypothesis comes from two landmark trials; the hypertrophy process that are not detectable by HOPE and LIFE trials. In LIFE, lower Cornell echocardiography, and this information may aid in product and Sokolow–Lyon voltage criterion during stroke risk prediction. antihypertensive treatment were associated with lower risk of CV mortality, myocardial infarctions or strokes.18 These findings were independent ECG LVH regression and prognosis of BP reduction or treatment modalities. In HOPE, treatment with the ACE inhibitor, ramipril in a high- While the evidence is clear that ECG LVH is linked risk cohort was associated with significant regres- to adverse clinical outcomes, the question arises as sion or prevention of ECG LVH (defined using the

Journal of Human Hypertension Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 466 Sokolow–Lyon voltage criterion).39 This was asso- LVH in hypertension mainly because it is easy to use, ciated with a significant reduction in CV events in highly specific and widely available. the group whereby LVH regression occurred. There- fore, based on these findings it is now evident that reversal of ECG LVH has an independent prognostic References value, independent of therapy and BP. 1 Devereux RB, Koren MJ, de Simone G, Okin PM, Kligfield P. Methods for detection of left ventricular hypertrophy: application to hypertensive heart dis- The implications of ECG LVH in clinical ease. Eur Heart J 1993; 14(Suppl D): 8–15. practice 2 Levy D, Labib SB, Anderson KM, Christiansen JC, Kannel WB, Castelli WP. Determinants of sensitivity On the basis of current evidence, the detection of and specificity of electrocardiographic criteria for LVH in hypertension is even more important than left ventricular hypertrophy. Circulation 1990; 81: before because its reversal is associated with a 815–820. favourable prognosis. However, the question arises 3 Levy D, Anderson KM, Savage DD, Kannel WB, as to how we should detect this condition in the vast Christiansen JC, Castelli WP. Echocardiographically number of hypertensive patients based on the detected left ventricular hypertrophy: prevalence and available resources. In clinical practice, the ECG is risk factors. The Framingham Heart Study. Ann Intern usually the first line of investigation to assess for the Med 1988; 108: 7–13. presence of LVH in hypertension. The presence of 4 Romhilt DW, Estes Jr EH. A point-score system for the ECG LVH is highly specific of anatomic LVH and ECG diagnosis of left ventricular hypertrophy. Am Heart J 1968; 75: 752–758. hence this subgroup of patients may not need 5 Casale PN, Devereux RB, Kligfield P, Eisenberg RR, further prognostic evaluation. However, in the group Miller DH, Chaudhary BS et al. Electrocardiographic of hypertensive patients without ECG LVH, should detection of left ventricular hypertrophy: development physicians be routinely referring all these patients and prospective validation of improved criteria. JAm for an echocardiogram? Due to the enhanced Coll Cardiol 1985; 6: 572–580. sensitivity of the echocardiogram in detecting LVH, 6 Verdecchia P, Schillaci G, Borgioni C, Ciucci A, the answer is likely to be ‘yes’ in an ideal world. In Gattobigio R, Zampi I et al. Prognostic value of a new addition, the echocardiogram has the advantage of electrocardiographic method for diagnosis of left providing a quick and accurate assessment of the ventricular hypertrophy in essential hypertension. valves and LV systolic/diastolic function. However, J Am Coll Cardiol 1998; 31: 383–390. 7 Schillaci G, Verdecchia P, Borgioni C, Ciucci A, advocating an echocardiogram for the assessment of Guerrieri M, Zampi I et al. Improved electrocardio- LVH in all hypertensive individuals would have graphic diagnosis of left ventricular hypertrophy. enormous cost implications on the National Health Am J Cardiol 1994; 74: 714–719. Service. This imaging modality is more time 8 Molloy TJ, Okin PM, Devereux RB, Kligfield P. consuming and requires considerable skills to per- Electrocardiographic detection of left ventricular form. Therefore, for the foreseeable future, the hypertrophy by the simple QRS voltage-duration combination of the patient’s risk factor profile, product. J Am Coll Cardiol 1992; 20: 1180–1186. ECG findings and 24 h ambulatory BP monitoring 9 Sokolow M, Lyon TP. The ventricular complex in left will be used to guide physicians or general practi- ventricular hypertrophy as obtained by unipolar pre- tioners as to which hypertensive patient merits an cordial and limb leads. 1949. Ann Noninvasive Electrocardiol 2001; 6: 343–368. echocardiogram for the evaluation of LVH. 10 European Society of Hypertension-European Society of Cardiology Guidelines Committee. European Society of Hypertension-European Society of Cardiology Conclusion guidelines for the management of arterial hyperten- In conclusion, the role of the ECG as a prognostic tool sion. J Hypertens 2003; 21: 1011–1053. 11 Okin PM, Devereux RB, Nieminen MS, Jern S, in hypertension has been well established and newer Oikarinen L, Viitasalo M et al. Electrocardiographic evidence continues to strengthen its clinical utility in strain pattern and prediction of new-onset congestive this respect. In addition, the favourable prognosis heart failure in hypertensive patients: the Losartan associated with ECG LVH regression should further Intervention for Endpoint Reduction in Hypertension prompt clinicians to actively screen for this condi- (LIFE) study. Circulation 2006; 113: 67–73. tion and implement aggressive secondary prevention 12 Lonn E, Mathew J, Pogue J, Johnstone D, Danisa K, therapy when present. With the wider availability of Bosch J et al. Relationship of electrocardiographic left echocardiography, both these noninvasive modalities ventricular hypertrophy to mortality and cardiovascu- can be used synergistically to yield prognostic lar morbidity in high-risk patients. Eur J Cardiovasc information far beyond the conventional CV risk Prev Rehabil 2003; 10: 420–428. 13 Larsen CT, Dahlin J, Blackburn H, Scharling H, factors in hypertension. Both ECG LVH and echocar- Appleyard M, Sigurd B et al. Prevalence and prognosis diographic LVH reflect different pathological pro- of electrocardiographic left ventricular hypertrophy, cesses and thereby influence prognosis in different ST segment depression and negative T-wave; the ways. However, for the foreseeable future, the ECG Copenhagen City Heart Study. Eur Heart J 2002; 23: will remain as an important tool for the diagnosis of 315–324.

Journal of Human Hypertension Prognostic value of the ECG in hypertension: where are we now? DSC Ang and CC Lang 467 14 Okin PM, Roman MJ, Devereux RB, Kligfield P. ventricular in hypertension. Arch Intern Electrocardiographic identification of increased left Med 2006; 166: 909–913. ventricular mass by simple voltage-duration products. 28 Oikarinen L, Nieminen MS, Viitasalo M, Toivonen L, J Am Coll Cardiol 1995; 25: 417–423. Jern S, Dahlof B et al. QRS duration and QT interval 15 Verdecchia P, Angeli F, Reboldi G, Carluccio E, predict mortality in hypertensive patients with left Benemio G, Gattobigio R et al. Improved cardiovascu- ventricular hypertrophy: the Losartan Intervention for lar risk stratification by a simple ECG index in Endpoint Reduction in Hypertension Study. Hyperten- hypertension. Am J Hypertens 2003; 16: 646–652. sion 2004; 43: 1029–1034. 16 Kannel WB, Gordon T, Offutt D. Left ventricular 29 Alfakih K, Walters K, Jones T, Ridgway J, Hall AS, hypertrophy by electrocardiogram. Prevalence, inci- Sivananthan M. New gender-specific partition values dence, and mortality in the Framingham study. Ann for ECG criteria of left ventricular hypertrophy: Intern Med 1969; 71: 89–105. recalibration against cardiac MRI. Hypertension 2004; 17 Kannel WB, Gordon T, Castelli WP, Margolis JR. 44: 175–179. Electrocardiographic left ventricular hypertrophy and 30 Liao Y, Cooper RS, McGee DL, Mensah GA, Ghali JK. risk of coronary heart disease. The Framingham study. The relative effects of left ventricular hypertrophy, Ann Intern Med 1970; 72: 813–822. coronary artery disease, and ventricular dysfunction 18 Okin PM, Devereux RB, Jern S, Kjeldsen SE, Julius S, on survival among black adults. JAMA 1995; 273: Nieminen MS et al. Regression of electrocardiographic 1592–1597. left ventricular hypertrophy during antihypertensive 31 Dunn FG, McLenachan J, Isles CG, Brown I, Dargie HJ, treatment and the prediction of major cardiovascular Lever AF et al. Left ventricular hypertrophy and events. JAMA 2004; 292: 2343–2349. mortality in hypertension: an analysis of data from 19 Levy D, Salomon M, D’Agostino RB, Belanger AJ, the Glasgow blood pressure clinic. J Hypertens 1990; 8: Kannel WB. Prognostic implications of baseline elec- 775–782. trocardiographic features and their serial changes in 32 Kannel WB. Left ventricular hypertrophy as a risk subjects with left ventricular hypertrophy. Circulation factor: the Framingham experience. J Hypertens Suppl 1994; 90: 1786–1793. 1991; 9: S3–S8; discussion S8–9. 20 Hsieh BP, Pham MX, Froelicher VF. Prognostic value of 33 Antikainen RL, Grodzicki T, Palmer AJ, Beevers DG, electrocardiographic criteria for left ventricular hyper- Webster J, Bulpitt CJ. Left ventricular hypertrophy trophy. Am Heart J 2005; 150: 161–167. determined by Sokolow–Lyon criteria: a different 21 Kannel WB. Prevalence and natural history of electro- predictor in women than in men? J Hum Hypertens cardiographic left ventricular hypertrophy. Am J Med 2006; 20: 451–459. 1983; 75: 4–11. 34 Lang C, Lip G. Complications of hypertension. Cardi- 22 Okin PM, Devereux RB, Nieminen MS, Jern S, ology. (Mosby: London) 2001: 489–500. Oikarinen L, Viitasalo M et al. Electrocardiographic 35 Okin PM, Roman MJ, Lee ET, Galloway JM, Howard strain pattern and prediction of cardiovascular mor- BV, Devereux RB. Combined echocardiographic left bidity and mortality in hypertensive patients. Hyper- ventricular hypertrophy and electrocardiographic ST tension 2004; 44: 48–54. depression improve prediction of mortality in Amer- 23 Houghton JL, Frank MJ, Carr AA, von Dohlen TW, ican Indians: the Strong Heart Study. Hypertension Prisant LM. Relations among impaired coronary flow 2004; 43: 769–774. reserve, left ventricular hypertrophy and thallium 36 Sundstrom J, Lind L, Arnlov J, Zethelius B, Andren B, perfusion defects in hypertensive patients without Lithell HO. Echocardiographic and electrocardio- obstructive coronary artery disease. J Am Coll Cardiol graphic diagnoses of left ventricular hypertrophy 1990; 15: 43–51. predict mortality independently of each other in a 24 Pringle SD, Macfarlane PW, McKillop JH, Lorimer AR, population of elderly men. Circulation 2001; 103: Dunn FG. Pathophysiologic assessment of left ventri- 2346–2351. cular hypertrophy and strain in asymptomatic patients 37 Kohsaka S, Sciacca RR, Sugioka K, Sacco RL, Homma with essential hypertension. J Am Coll Cardiol 1989; S, Di Tullio MR. Additional impact of electrocardio- 13: 1377–1381. graphic over echocardiographic diagnosis of left 25 de Simone G, Devereux RB, Roman MJ, Ganau A, Saba ventricular hypertrophy for predicting the risk of PS, Alderman MH et al. Assessment of left ventricular ischemic stroke. Am Heart J 2005; 149: 181–186. function by the midwall fractional shortening/end- 38 Verdecchia P, Schillaci G, Borgioni C, Ciucci A, systolic stress relation in human hypertension. JAm Gattobigio R, Zampi I et al. Prognostic significance of Coll Cardiol 1994; 23: 1444–1451. serial changes in left ventricular mass in essential 26 Palmieri V, Okin PM, Bella JN, Wachtell K, Oikarinen hypertension. Circulation 1998; 97: 48–54. L, Gerdts E et al. Electrocardiographic strain pattern 39 Mathew J, Sleight P, Lonn E, Johnstone D, Pogue J, and left ventricular diastolic function in hypertensive Yi Q et al. Reduction of cardiovascular risk by patients with left ventricular hypertrophy: the LIFE regression of electrocardiographic markers of left study. J Hypertens 2006; 24: 2079–2084. ventricular hypertrophy by the angiotensin-converting 27 Schillaci G, Pirro M, Ronti T, Gemelli F, Pucci G, enzyme inhibitor ramipril. Circulation 2001; 104: Innocente S et al. Prognostic impact of prolonged 1615–1621.

Journal of Human Hypertension