Lsehpodcmesdmtc

Cr

0d

Relation of Electrocardiographic Criteria for Left AtrialEnlargement to Two-Dimensional Echocardiographic Left Atrial

Volume Measurements

Kwan S. Lee, MBBCha, Christopher P. Appleton, MDa,*, Steven J. Lester, MDa,Terrence J. Adam, MD, PhDb, R. Todd Hurst, MDa, Carlos A. Moreno, BSa,

and Gregory T. Altemose, MDa

Left atrial (LA) enlargement by 2-dimensional (2-D) echocardiography predicts adversecardiovascular outcomes. Electrocardiographic (ECG) criteria for LA enlargement arebased on M-mode echocardiographic LA diameter, which is inferior to 2-D–derived LAvolumes. This study compared established ECG criteria for LA enlargement with atrialvolume obtained by 2-D echocardiography to determine if traditional ECG criteria accu-rately represent LA chamber enlargement, therefore offering a low-cost screening tool. Atotal of 261 randomly selected patients who underwent electrocardiography and 2-Dechocardiography were enrolled. ECG parameters and electronically derived P-wave me-dians were analyzed with electronic calipers for maximal accuracy. LA volumes by 2-Dechocardiography were measured with Simpson’s method of discs, with enlargementdefined as 32 ml/m2. Sensitivity and specificity tables and receiver-operating characteristiccurves were constructed for each criterion. Univariate and multivariate analyses wereperformed for predictors of 2-D echocardiographic LA enlargement. LA enlargement waspresent in 43% of patients. ECG P-wave duration was the most sensitive for the detectionof LA enlargement (69%) but had low specificity (49%). Conversely, a biphasic P wave wasthe most specific (92%) but had low sensitivity (12%). The maximum area under thereceiver-operating characteristic curve for any criterion was 0.64, too low to be of clinicalutility. In conclusion, established ECG criteria for LA enlargement do not reliably reflectLA enlargement and lack sufficient predictive value to be useful clinically. These resultssuggest that P-wave abnormalities should be noted as nonspecific LA abnormalities, withthe term “LA enlargement” no longer used. © 2007 Elsevier Inc. All rights reserved. (Am

J Cardiol 2007;99:113–118)

M

TdeAt

tocpoehsml

ofcog

eft atrial (LA) enlargement measured using cardiac ultra-ound is associated with an increased risk for cardiovascularvents.1 Although cardiac ultrasound is noninvasive andarmless, a less costly technique such as electrocardiogra-hy for identifying LA enlargement would be desirable. Tour knowledge, all previous studies correlating electrocar-iographic (ECG) criteria for LA enlargement with echo-ardiographically defined LA enlargement have used M-ode echocardiographically derived LA diameters.2–14 LA

nlargement as defined by echocardiographic volume mea-urement is more accurate15 and a better predictor of car-iovascular outcomes.16 This study was designed to deter-ine the predictive value of established ECG criteria used

o define LA enlargement as assessed by echocardiographi-ally derived LA volumes.

Divisions of aCardiovascular Diseases and bInternal Medicine, Mayolinic Arizona, Scottsdale, Arizona. Manuscript received May 2, 2006;

evised manuscript received July 19, 2006 and accepted July 25, 2006.*Corresponding author: Tel: 480-301-8000; fax: 480-301-8081.

aE-mail address: cappleton@mayo.edu (C.P. Appleton).

002-9149/07/$ – see front matter © 2007 Elsevier Inc. All rights reserved.oi:10.1016/j.amjcard.2006.07.073

ethods

his was a retrospective cross-sectional study, which ran-omly selected adult patients who underwent transthoracicchocardiography for any indication at our institution fromugust 2001 to August 2003. The study was approved by

he Mayo Foundation Institutional Review Board.We excluded patients who had not authorized the use of

heir records, had no analyzable P waves on electrocardi-graphy, were not in sinus rhythm at the time of electro-ardiography or echocardiography, and had pacemakers. Allatients underwent electrocardiography �1 week after theirutpatient echocardiographic studies or �1 day after theirchocardiographic studies if hospitalized. Age, gender, race,eight, weight, vital signs, body mass index, and bodyurface area as well as history of hypertension, hyperlipide-ia, diabetes, ischemic heart disease, stroke, or atrial fibril-

ation were recorded.Measured variables were the presence, type, and degree

f native valvular heart disease, the left ventricular ejectionraction, and LA volumes and ECG characteristics. Echo-ardiographic data were derived from the measurementsbtained during patients’ clinical studies. Echocardio-raphic studies were interpreted and vetted for quality and

ccuracy by experienced staff echocardiographers. Two-

www.AJConline.org

dasmvMea

vblo

psewa0WitRttFrLIitPi

(tbSt

’s meth

Fdl

114 The American Journal of Cardiology (www.AJConline.org)

imensional (2-D) LA volumes were assessed from thepical 4-chamber view at end-expiration apnea using Simp-on’s method of discs,17 as shown in Figure 1. LA maxi-um volume was obtained on the frame preceding mitral

alve opening at end-systole from the 4-chamber view.inimum volume was measured in a similar fashion at the

nd of atrial contraction. Both were indexed to body surface

Figure 1. Maximum and minimum LA volume measured by Simpson

igure 2. Electronic caliper measurement of negative terminal P-waveuration of electronically derived V1 median, magnified (12�) using off-ine analysis software.

rea. LA enlargement was defined as a maximal indexed t

olume of 32 ml/m2, a value larger than defined as abnormaly recent American Society of Echocardiography guide-ines18 but well validated as predicting adverse clinicalutcomes.19–22

Electrocardiograms were analyzed retrospectively forreviously validated P-wave parameters in relation to LAize.2,3,5 Electrocardiograms were retrieved from a centrallectronic database, and analysis of P-wave configurationas performed by a single blinded interpreter using off-line

nalysis software (Muse Information System version05A.32, GE Marquette Medical Systems, Inc., Milwaukee,isconsin). Lead waveforms were measured from electron-

cally derived medians of individual leads as described inhe manual Physician’s Guide to Marquette Electronicsesting ECG Analysis. Electronic calipers with a resolution

o a millisecond were used during measurements from elec-ronically magnified (12�) median waveforms, as shown inigure 2. Three measurements were performed for eacheading, and an average was obtained. P-wave criteria forA enlargement included P-wave duration in lead I, II, or

II �110 ms2,5,7,8; P-wave notching in lead I, II, or III, withnterpeak distance �40 ms (P mitrale)2,5,7; or area sub-ended by the terminal negative component of a biphasic

wave in precordial lead V1 �40 ms · mm (Morrisndex).2,3,5,7,9,11,23

Statistical analysis was performed with JMP version 5.12SAS Institute Inc., Cary, North Carolina). Descriptive sta-istics were obtained and are expressed as mean � SDecause most of the data approximated normal distributions.ensitivity and specificity tables with corresponding posi-

ive and negative predictive value tables were then created

od of discs from the apical 4-chamber view at end-expiration apnea.

o evaluate the performance of the ECG criteria in detecting

LeTtsqaetpmwTtdattcwisa

R

IsTtiec

mEWvihwbsa

cTP0fdct

momspe

TB

V

AWCBBHWH

S

D

L

L

L

L

L

TSad

E

P

B

N

CCCCC

115Methods/ECG Criteria for LA Enlargement

A enlargement. The performance of ECG criteria for LAnlargement was tabulated individually and in combination.he traditional thresholds defining positive test results were

hen tested by receiver-operating characteristic curve con-truction to identify optimal thresholds. To determine if auantitative association existed between the degree ofbnormality in the ECG criteria and the degree of 2-Dchocardiographic LA enlargement, Pearson’s correla-ion coefficients were calculated. Univariate analysis waserformed comparing LA enlargement with clinical, de-ographic, and ECG characteristics. Logistic regressionas used as the nonparametric univariate statistical test.he univariate statistical analysis used logistic regression

o fit the dependent LA enlargement variable against theemographic, clinical, and ECG variables. Multivariatenalysis was then performed to determine if addition ofhe ECG criteria improved the predictive ability of de-ermining LA enlargement above knowledge of basiclinical parameters. Results are expressed as odds ratiosith 95% confidence intervals. Backward stepwise elim-

nation was used for multivariate analysis, followed byelective forward regression to identify significant vari-

able 1aseline characteristics of study population

ariable All Patients(n � 261)

LA Size

�32 ml/m2

(n � 113)�32 ml/m2

(n � 148)

ge (yrs) 67 � 15 71 � 11 64 � 16omen 53% 51% 55%

aucasian 93% 98% 93%ody surface area (m2) 1.9 � 0.3 1.9 � 0.3 1.9 � 0.3ody mass index (kg/m2) 28 � 6 28 � 6 28 � 7eight (cm) 169 � 10 169 � 11 169 � 10eight (kg) 80 � 21 80 � 20 81 � 22istory ofhypertension 55% 62% 49%hyperlipidemia 44% 48% 41%ischemic heart disease 34% 51% 20%diabetes 19% 23% 16%mitral valve disease 5% 11% 0%atrial fibrillation 13% 19% 8%stroke 12% 13% 10%obesity 23% 22% 24%ystolic blood pressure

(mm Hg)130 � 21 133 � 22 129 � 19

iastolic blood pressure(mm Hg)

71 � 13 69 � 14 72 � 12

eft ventricular ejectionfraction (%)

60 � 13 55 � 16 63 � 11

A maximum volume,nonindexed (ml)

60 � 24 80 � 21 44 � 13

A maximum volume,indexed (ml/m2)

32 � 13 43 � 11 23 � 6

A minimal volume,nonindexed (ml)

32 � 19 47 � 19 22 � 10

A minimal volume,indexed (ml/m2)

17 � 10 25 � 11 11 � 5

Data are expressed as mean � SD or percentage.

bles. h

esults

n total, 261 patients were enrolled into the study by randomelection. The characteristics of the study group are listed inable 1. Of our study population, 106 (41%) were inpa-

ients. LA enlargement by 2-D echocardiographic maximalndexed volume was present in 43%. Patients with LAnlargement had a uniformly higher incidence of cardiovas-ular co-morbidities and risk factors.

The 3 ECG criteria used to detect LA volume enlarge-ent performed poorly. Sensitivities and specificities of theCG criteria alone or in combination are listed in Table 2.hen used as individual tests, the highest sensitivity for LA

olume enlargement was with a P-wave duration �110 msn limb lead I, II, or III, but this had low specificity. Theighest specificity for LA enlargement was for a bifid Pave separated by �40 ms (P mitrale) in lead I, II, or III,ut this had low sensitivity. When �1 criterion was present,ensitivity decreased, whereas specificity increased, withoutchange in predictive value.P-wave criteria were tested with the construction of re-

eiver-operating characteristic curves, as shown in Figure 3.he maximum area under the curve was greatest for a bifidwave in lead I, II, or III of 40 ms (P mitrale), at 0.64 (SE

.09, 95% confidence interval 0.47 to 0.79). All 3 thresholdsor ECG criteria closely approximated the accuracy pointsetected by the receiver-operating characteristic curves,onfirming that these were appropriate for maximal predic-ive power.

Scatterplots of ECG criteria were constructed against LAaximal indexed volume, as shown in Figure 4. The degree

f abnormality in all ECG criteria was associated with LAaximal indexed volume (p �0.0001) but with low Pear-

on’s correlation coefficients. Significant univariate clinicalredictors for LA enlargement were age, the left ventricularjection fraction, ischemic heart disease, left-sided valvular

able 2ensitivity and specificity of individual electrocardiographic parametersnd combinations of parameters in predicting left atrial enlargementefined by indexed volume �32 ml/m2

CG Criterion Detection of LA Enlargement (�32 ml/m2)

Sensitivity Specificity PositivePredictive

Value

NegativePredictive

Value

duration �110 ms inlead I, II, or III(criterion 1)

69% 49% 51% 68%

iphasic P wave �40ms in lead I,II, or III (criterion 2)

12% 92% 52% 58%

egative terminal Pforce in lead V1

�40 ms · mm(criterion 3)

46% 64% 49% 61%

riteria 1 and 2 4% 96% 46% 57%riteria 2 and 3 1% 99% 50% 57%riteria 1 and 3 20% 91% 64% 60%riteria 1, 2, and 3 5% 98% 67% 58%riteria 1 or 2 or 3 77% 33% 47% 65%

eart disease, history of atrial fibrillation, or history of

hEIls

Fcgsens � sensitivity; spec � specificity

F(

116 The American Journal of Cardiology (www.AJConline.org)

ypertension. The only significant univariate traditionalCG criteria of the 3 was a P-wave duration in lead I, II, or

II �110 ms (p � 0.003). Multivariate stepwise backwardogistic regression analysis showed that the addition of all

igure 4. Scatterplots of ECG criteria versus LA maximal indexed volumein milliliters per square meter). CI � confidence interval.

igure 3. Receiver-operating characteristic curves for traditional ECGriteria in detecting LA enlargement (32 ml/m2) identifying thresholds ofreatest accuracy. AUC � area under the curve. CI � confidence interval;

ignificant ECG parameters and criteria did not add incre-

mi

D

WomAbuMrnclbinpcvevcbdooAbvtb

mspnshwlf

espttPwace

lsdawnt

asiiRdamppdvrvo2tur

mfi

AcePz

TMb

C

AC

L

H

L

117Methods/ECG Criteria for LA Enlargement

ental value beyond that provided by patient demograph-cs, as listed in Table 3.

iscussion

hether electrocardiograms are interpreted by cardiologistsr automated proprietary algorithms, the term “LA enlarge-ent” continues to be used in their clinical interpretation.ll previous studies supporting this ECG diagnosis on theasis of echocardiographically defined LA enlargementsed a single length measurement of the left atrium by-mode echocardiography. Two-dimensional echocardiog-

aphy has been shown to be much more accurate in diag-osing LA enlargement15 and a more robust marker ofardiovascular events than M-mode diameter,16 yet the re-ation of P-wave abnormalities to 2-D LA volume has noteen studied. The purpose of this study was to readdress thessue of diagnosing LA enlargement by ECG criteria usingew technology (electronically derived median ECG com-lexes and using electronic ECG calipers) and a more ac-urate standard for LA enlargement (2-D echocardiographicolumes). Our study shows that current ECG criteria for LAnlargement correlate poorly with 2-D echocardiographicolume, with a predictive value that is too low to be clini-ally useful in individual cases. In addition, when analyzedy multivariate analysis, the ECG criteria did not add pre-ictive value for diagnosing LA enlargement above thatbtained by simple demographics, despite criteria thresh-lds that are appropriate for maximal predictive capability.s listed in Table 2, the specificity of individual or com-ined P-wave variables for identifying LA enlargement wasery high. However, in each case, the sensitivity was so lowhat the predictive values remained in a range that would note clinically useful in our real-world clinical population.

Compared with previous studies using M-mode measure-ents for defining LA enlargement, our results indicate

ignificantly lower specificity for the Morris index and arolonged P-wave duration.2,5,7,9,11 Our findings regardingotched P-wave criteria (P mitrale) and LA enlargement areimilar to previous results showing very low sensitivity withigh specificity.2,5 Other features shown to be associatedith “ECG LA enlargement” have been LA pressure over-

oad,13,24 atrial hypertrophy,25 or a combination of these

able 3ultivariate predictors of left atrial enlargement (�32 ml/m2) by

ackward nominal regression*

haracteristic OddsRatio

95% ConfidenceInterval

WaldChi-Square

p Value

ge (per 10 yrs) 1.3 1.1–1.7 7.2 0.007oronary arterydisease

2.9 1.5–5.6 10.8 0.001

eft ventricularejection fraction(per 10%)

0.8 0.7–1.0 4.0 0.05

istory of atrialfibrillation

2.6 1.2–6.0 5.3 0.02

eft-sided valvulardisease

3.8 1.2–12.5 4.8 0.03

* Regression with these variables produced an r 2 value of 0.16.

actors.13

The poor performance of ECG criteria in detecting LAnlargement has been previously suggested12 and is noturprising in light of the increased knowledge from electro-hysiologic studies over the past 20 years. The hypothesishat LA enlargement would lead to prolonged atrial conduc-ion time did not take into account that prolongation of the

wave can be a manifestation of intra-atrial block,10,26

hich is common in the elderly and not necessarily associ-ted with LA enlargement. In addition, after reaching theompensatory limits of LA hypertrophy, the atrium maynlarge and fail with a decrease in voltage.

Signal averaging with median complexes is an estab-ished technique for artifact filtering in computer analy-is,27,28 which has the advantage of excellent signal noiseiminishment. Given the resolution of electronic calipersnd the 12-fold magnification in this study, our methodould be expected to be more accurate than using a mag-ifying glass with 5� magnification, as was customary inhe original studies.

A second difference from the original M-mode studies isdefinition of LA enlargement that is indexed for body

urface area. Correction for body surface area is a standard-zed attempt to normalize previously identified differencesn LA size with respect to gender, height, and weight.25,29

ecent studies that correlated LA volume and adverse car-iovascular events were all indexed to body surface areand used either the same definition for LA enlargement (32l/m2) or the alternative (28 ml/m2).1 Although this study

resents data based on the first definition, when we com-ared the sensitivities and specificities using the alternativeefinition, there was minimal difference. We did not studyolumes derived by magnetic resonance imaging, the cur-ent gold standard for LA volume assessment. LA 2-Dolumes are approximately 15% smaller, probably becausef the geometric simplification of the atrium.30 Because thevolumes have similar slopes except for the underestima-

ion by 2-D echocardiography, a repeat of the current studysing magnetic resonance imaging LA volume would likelyesult in similar results.

In conclusion, established ECG criteria for LA enlarge-ent do not reliably reflect LA enlargement and lack suf-cient predictive value to be useful clinically.

cknowledgment: We thank Paul Elko from GE Health-are for his assistance in providing research regarding thelectronic measurement of electrocardiograms and Elianeurchase from the Library Department, Mayo Clinic Ari-ona, Scottsdale, Arizona.

1. Abhayaratna WP, Seward JB, Appleton CP, Douglas PS, Oh JK, TajikAJ, Tsang TS. Left atrial size: physiologic determinants and clinicalapplications. J Am Coll Cardiol 2006;47:2357–2363.

2. Hazen MS, Marwick TH, Underwood DA. Diagnostic accuracy of theresting electrocardiogram in detection and estimation of left atrialenlargement: an echocardiographic correlation in 551 patients. AmHeart J 1991;122:823–828.

3. Aronow WS, Schwartz KS, Koenigsberg M. Prevalence of enlargedleft atrial dimension by echocardiography and its correlation with atrialfibrillation and an abnormal P terminal force in lead V1 of the elec-trocardiogram in 588 elderly persons. Am J Cardiol 1987;59:1003–

1004.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

118 The American Journal of Cardiology (www.AJConline.org)

4. Jin L, Weisse AB, Hernandez F, Jordan T. Significance of electrocar-diographic isolated abnormal terminal P-wave force (left atrial abnor-mality). An echocardiographic and clinical correlation. Arch InternMed 1988;148:1545–1549.

5. Munuswamy K, Alpert MA, Martin RH, Whiting RB, Mechlin NJ.Sensitivity and specificity of commonly used electrocardiographiccriteria for left atrial enlargement determined by M-mode echocardi-ography. Am J Cardiol 1984;53:829–832.

6. Perosio AM, Suarez LD, Torino A, Llera JJ, Ballester A, RoisinblitJM. Reassessment of electrovectorcardiographic signs of left atrialenlargement. Clin Cardiol 1982;5:640–646.

7. Termini BA, Lee YC. Echocardiographic and electrocardiographiccriteria for diagnosing left atrial enlargement. South Med J 1975;68:161–165.

8. Chirife R, Feitosa GS, Frankl WS. Electrocardiographic detection ofleft atrial enlargement. Correlation of P wave with left atrial dimensionby echocardiography. Br Heart J 1975;37:1281–1285.

9. Ikram H, Drysdale P, Bones PJ, Chan W. The non-invasive recognitionof left atrial enlargement: comparison of electro- and echocardio-graphic measurements. Postgrad Med J 1977;53:356–359.

0. Josephson ME, Kastor JA, Morganroth J. Electrocardiographic leftatrial enlargement. Electrophysiologic, echocardiographic and hemo-dynamic correlates. Am J Cardiol 1977;39:967–971.

1. Waggoner AD, Adyanthaya AV, Quinones MA, Alexander JK. Leftatrial enlargement. Echocardiographic assessment of electrocardio-graphic criteria. Circulation 1976;54:553–557.

2. van Dam I, Roelandt J, Robles de Medina EO. Left atrial enlargement:an electrocardiographic misnomer? An electrocardiographic-echocar-diographic study. Eur Heart J 1986;7:115–117.

3. Di Bianco R, Gottdiener JS, Fletcher RD, Pipberger HV. Left atrialoverload: a hemodynamic, echocardiographic, electrocardiographicand vectorcardiographic study. Am Heart J 1979;98:478–489.

4. Bartall H, Desser KB, Benchimol A, Massey BJ. Echocardiographicleft atrial enlargement. Comparison of vectorcardiogram and electro-cardiogram for detection. J Electrocardiol 1978;11:355–359.

5. Lester SJ, Ryan EW, Schiller NB, Foster E. Best method in clinicalpractice and in research studies to determine left atrial size. Am JCardiol 1999;84:829–832.

6. Tsang TS, Abhayaratna WP, Barnes ME, Miyasaka Y, Gersh BJ,Bailey KR, Cha SS, Seward JB. Prediction of cardiovascular outcomeswith left atrial size: is volume superior to area or diameter? J Am CollCardiol 2006;47:1018–1023.

7. Wang Y, Gutman JM, Heilbron D, Wahr D, Schiller NB. Atrialvolume in a normal adult population by two-dimensional echocardi-ography. Chest 1984;86:595–601.

8. Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pel-likka PA, Picard MH, Roman MJ, Seward JB, Shanewise JS, et al.

Recommendations for chamber quantification: a report from the Amer-

ican Society of Echocardiography’s Guidelines and StandardsCommittee and the Chamber Quantification Writing Group, developedin conjunction with the European Association of Echocardiography, abranch of the European Society of Echocardiography. J Am SocEchocardiogr 2005;18:1440–1463.

9. Tsang TS, Barnes ME, Gersh BJ, Takemoto Y, Rosales AG, BaileyKR, Seward JB. Prediction of risk for first age-related cardiovascularevents in an elderly population: the incremental value of echocardi-ography. J Am Coll Cardiol 2003;42:1199–1205.

0. Barnes ME, Miyasaka Y, Seward JB, Gersh BJ, Rosales AG, BaileyKR, Petty GW, Wiebers DO, Tsang TS. Left atrial volume in theprediction of first ischemic stroke in an elderly cohort without atrialfibrillation. Mayo Clin Proc 2004;79:1008–1014.

1. Tsang TS, Barnes ME, Gersh BJ, Bailey KR, Seward JB. Left atrialvolume as a morphophysiologic expression of left ventricular diastolicdysfunction and relation to cardiovascular risk burden. Am J Cardiol2002;90:1284–1289.

2. Moller JE, Hillis GS, Oh JK, Seward JB, Reeder GS, Wright RS, ParkSW, Bailey KR, Pellikka PA. Left atrial volume: a powerful predictorof survival after acute myocardial infarction. Circulation 2003;107:2207–2212.

3. Morris JJ Jr, Estes EH Jr, Whalen RE, Thompson HK Jr, McIntoshHD. P-wave analysis in valvular heart disease. Circulation 1964;29:242–252.

4. Heikkila J, Hugenholtz PG, Tabakin BS. Prediction of left heart fillingpressure and its sequential change in acute myocardial infarction fromthe terminal force of the P wave. Br Heart J 1973;35:142–151.

5. Romhilt DW, Scott RC. Left atrial involvement in acute pulmonaryedema. Am Heart J 1972;83:328–331.

6. Spodick DH. Unappreciated prevalence of interatrial block and asso-ciated consequences: a poorly perceived pandemic. Mayo Clin Proc2004;79:668–670.

7. Willems JL, Zywietz C, Arnaud P, van Bemmel JH, Degani R, Mac-farlane PW. Influence of noise on wave boundary recognition by ECGmeasurement programs. Recommendations for preprocessing. ComputBiomed Res 1987;20:543–562.

8. Willems JL, Arnaud P, van Bemmel JH, Bourdillon PJ, Degani R,Denis B, Graham I, Harms FM, Macfarlane PW, Mazzocca G, et al. Areference data base for multilead electrocardiographic computer mea-surement programs. J Am Coll Cardiol 1987;10:1313–1321.

9. Pritchett AM, Jacobsen SJ, Mahoney DW, Rodeheffer RJ, Bailey KR,Redfield MM. Left atrial volume as an index of left atrial size: apopulation-based study. J Am Coll Cardiol 2003;41:1036–1043.

0. Rodevan O, Bjornerheim R, Ljosland M, Maehle J, Smith HJ, Ihlen H.Left atrial volumes assessed by three- and two-dimensional echocar-diography compared to MRI estimates. Intl J Card Imaging 1999;15:

397–410.