echocardiographic measurements normal...

Echocardiographic Measurements in Normal Subjects

Growth-related Changesthat Occur between Infancy and Early Adulthood

WALTER L. HENRY, M.D., JAMES WARE, PH.D., JULIUS M. GARDIN, M.D.,SEYMOUR I. HEPNER, M.D., JOYCE MCKAY, R.N., AND MICHAEL WEINER, M.D.

SUMMARY Echocardiographic measurements of the left ventric-ular dimensions and wall thicknesses at end diastole and end systole,aortic root and left atrial dimensions, mitral valve E-F slope, left ven-tricular ejection fraction, percent fractional shortening of the left ven-tricular internal dimension, estimated left ventricular mass andpercentage systolic thickening of the ventricular septum and left ven-tricular free wall were obtained in 105 normal subjects ranging fromone day to 23 years of age. Each parameter was found to follow alinear regression upon one of three functions of the body surface area.The internal dimensions of the left ventricle, the left atrium, and the

ECHOCARDIOGRAPHY IS BEING USED ROU-TINELY to evaluate infants, children, and young adultswith suspected heart disease.' In order to detect changes inthe structure and function of the heart produced by disease,it is important to determine accurately the effect of normalgrowth and development on echocardiographicmeasurements of chamber size, thickness, and function.Previous studies have reported growth-related changes inechocardiographic measurements as a function of either thebody surface area5' 6 or the weight of the individual.7 Duringa recent echocardiographic evaluation of chronically-transfused children with fl-thalassemia,' we obtained datafrom normal children for comparison with the data obtainedin the children with thalassemia. Because some of the data inour normal children differed with previously published data,we decided to expand our study of normal individuals anddetermine our own normal values for several echocardio-graphic parameters. The present paper contains the resultsof this study of normal infants, children, and young adults.Analysis of these normal data indicates that the dimensionsof various cardiac structures appear to follow a linearregression upon one of three functions (linear, square root,or cube root) of the body surface area.

Methods and Material

Study Population

Normal subjects ranged in age from one day to 23 years ofage. Fifty-two were male and 53 female. These subjects con-sisted of 93 Caucasians, 9 blacks, and 3 Orientals. Thirteenof the 105 subjects were newborn infants who ranged from16 to 118 hours of age (mean 67 hours) and from 2.5 to 4.0

From the Cardiology Branch and the Biometrics Research Branch,National Heart, Lung, and Blood Institute, National Institutes of Health,Bethesda, Maryland and Pediatric and Adult Cardiology Divisions,Georgetown University Medical Center, Washington, D.C., and PediatricHematology Division, New York University Medical Center, New York,New York.

Address for reprints: Walter L. Henry, M.D., Cardiology Branch, Building10, Room 7B-15, National Heart, Lung, and Blood Institute, Bethesda,Maryland 20014.

Received June 29, 1977; revision accepted September 23, 1977.

aortic root, and the mitral valve E-F slope varied in a linear relationto the cube root of the body surface area. Thickness of the ventric-ular septum and left ventricular free wall varied in a linear relation tothe square root of the body surface area. Estimated left ventricularmass varied linearly with the direct measurement of body surfacearea. Ejection fraction, percent fractional shortening of the left ven-tricle and percent systolic thickening of the ventricular septum andleft ventricular free wall were independent of body surface areadespite a marked increase in the size of the left ventricle during nor-mal growth and development.

kg (mean 3.4 kg) in weight. Questioning of the individual orthe parent failed to reveal any evidence of cardiac orsystemic disease symptoms and physical examination ofeach subject was normal. Twelve lead electrocardiogramswere available in 87 individuals and were normal in each in-stance. Height, weight and age were available for each sub-ject and cuff blood pressure measurements were obtainedand found to be normal in 70. Body surface area was es-timated using the Boothby and Sandiford modification ofthe DuBois nomogram. In every subject, the body weightwas within 25% of the desirable weight for height as definedby Metropolitan Life Insurance Company statistical tables.

Echocardiographic Techniques

Echocardiograms were obtained using either an Ekoline20A or a Hoffrel 201 ultrasound receiver interfaced with aHoneywell 1856 Line Scan Recorder. A 2.25 MHz, 1.25 cmdiameter Aerotech ultrasound transducer was used in theyoung adults and older children, a 3.5 MHz, 1.25 cmdiameter transducer was used in the younger children, and a5.0 MHz, 0.6 cm diameter transducer was employed tostudy the infants.The transverse dimensions of the left ventricle at end

diastole and at end systole were obtained with the ultra-sound beam passing through the left ventricle slightly belowthe tips of the mitral valve leaflets.' The end-diastolic andend-systolic transverse dimensions were taken as the max-imum and minimum transverse dimensions, respectively(fig. 1). Three consecutive measurements were made andaveraged. Measurements of septal and posterobasal freewall thicknesses in late diastole (ust prior to the thinning ofthe walls) and at end systole were obtained with the ultra-sound beam passing through the left ventricle at the tip ofthe mitral valve and also slightly caudal to this position." Aswitched-gain circuit was used to simplify measurement ofposterobasal free wall thickness." When the switched-gaincircuit was activated, it was usually possible to adjust thegain until two distinct parallel lines were seen at theepicardial-lung interface. Wall thickness measurementswere made to the center of the anterior line. When only one

278

by guest on May 8, 2018

http://circ.ahajournals.org/D

ownloaded from

http://circ.ahajournals.org/

ECHO MEASUREMENTS IN NORMAL SUBJECTS/Henry et al.

tY:;:: :0:0:X;; A.

fMo.,44S , 0,I

FIGURE 1. M-mode echocardiogram obtainedfrom a normal sub-ject with the ultrasound beam passing through the body of the leftventricle slightly below the tips of the mitral leaflets. Left ventric-ular transverse dimensions in diastole (lvtd4) and systole (lvtd8) weremeasured from the maximum and minimum dimensions, respec-tively. Diastolic measurements of the thicknesses of the ventricularseptum (VS) and posterobasal left ventricular free wall (pw) wereobtained in late diastole just prior to the thinning of the walls.

line was seen, measurement was made to its center. In allsubjects studied, the measurements of wall thickness ob-tained at the level of the mitral valve tip were similar (i.e.,within one millimeter) to those obtained slightly caudal tothe valve tip. As with the other parameters in this study,three consecutive wall thickness measurements were madeand averaged.

In addition to measuring the internal dimensions and wallthicknesses of the left ventricle, several derived parametersof left ventricular size and function were computed. Leftventricular mass was estimated by the method of Troy etal.12 Ejection fraction was calculated using the cubedassumption to estimate left ventricular volume.'3 Percentfractional shortening of the ventricular internal dimensionwas aiso computed.14

Left atrial dimension and aortic root dimension were ob-tained by angling the ultrasound beam in a medial andcephalad direction from the mitral valve tip until both theaortic root and aortic valve signals were visualized." 15 Atthis point, signal intensity was reduced until the posteriorleft atrial wall was represented by a thin line. Alternatively,the posterior left atrial wall was identified using theswitched-gain circuit." After the left atrial wall had beenidentified, signal intensity was adjusted until both theanterior and posterior walls of the aorta were represented bytwo thin lines. Left atrial dimension and aortic root dimen-sion were measured from this reduced-gain portion of therecord (fig. 2). Measurement of left atrial dimension wasmade in late systole (or early diastole) at the point of max-

FIGURE 2. M-mode echocardiogram obtainedfrom a normalsub-ject with the ultrasound beam passing through aortic root and leftatrium at the level of the aortic valve leaflets. This record was ob-tained using the switched-gain circuit and with the transceiver gainreduced so that the signals from the walls of the aorta appear assingle lines. A ortic root dimension (A o) is measuredfrom the centerof the aortic root signals in late diastole at or slightly before the"'notch" that is seen on the aortic root echogram. Maximum atrialdimension (LA) is measured at end systole from the center of theposterior aortic wall signal to the left atrial wall represented by thesingle dark line obtained with the switched-gain circuit in operation.

imum left atrial dimension. Aortic root dimension wasmeasured in late diastole (i.e., at or just prior to the notchthat is seen on the aortic root signal in late diastole).Measurement was made from the center of the anterior aor-tic wall signal to the center of the posterior aortic wall signal(fig. 2).The closing velocity of the anterior leaflet of the mitral

valve in early diastole (E-F slope) was measured with the ul-trasound signal being reflected from the tip of the mitralleaflets. Three consecutive heart beats were measured fromthe portion of the record in which both anterior andposterior mitral leaflets were seen and the excursion ofanterior mitral leaflet was maximal. The slope between the Eand F point (and not F0)' was used in this study.

In addition to these measurements, we also measured car-diac dimensions using a variety of methods previouslydescribed or suggested. Left ventricular transverse dimen-sion and ventricular septal and posterobasal free wall thick-nesses were measured at end diastole at the peak of the Rwave and left ventricular transverse dimension measuredalone at the onset of the Q wave in 77 subjects in whom anelectrocardiogram had been recorded simultaneously withthe echocardiogram. Aortic root dimension in late diastolewas measured in all 105 patients in the undamped portion of

279



ownloaded from


VOL 57, No 2, FEBRUARY 1978

60r

z

z 500

('7

z

.I

' 30

1.1

J 2010

cWa0

AAj

AA

iL A

ft

AA-A

AA A

F, ..

b- P

o l0.5 1.0 1.5

BODY SURFACE AREA IN SQUARE METERS

A

A 50r-

z

o 40z

.

, 2a 30

z

X 10

2.0

FIGURE 3. Plot ofend-diastolic left ventricular transverse dimen-sion in millimeters versus a linear function of body surface area insquare meters. In this and all subsequent figures, data from normalsubjects from one month to 23 years of age will be indicated bytriangles while data from newborn infants will be indicated bycircles with a dot in the center.

the record using three different boundaries of the aortic wall:a) outermost echo of anterior aortic wall to the outermostecho of the posterior aortic wall, b) innermost echo ofanterior aortic wall to innermost echo of posterior aorticwall, and c) outermost echo of anterior aortic wall to inner-most echo of posterior aortic wall. Aortic root dimension atend systole was also measured in every patient. Left atrialdimension was measured in this same undamped portion ofrecord from a) innermost echo of posterior aortic wall to leftatrial wall, and b) outermost echo of posterior aortic wall toleft atrial wall. In these measurements of left atrial dimen-sion, the left atrial wall was identified either by use of theswitched-gain circuit or from the damped portion of therecord.

Results

Statistical Analysis

The transverse dimension of the left ventricle at enddiastole (maximum distance between septum andposterobasal free wall) is plotted in figure 3 versus a linearfunction of body surface area. This plot demonstrates a cur-

vilinear relation of left ventricular end-diastolic dimensionand body surface area. Graphical analysis of this curve

_ ,"~~~~~~~~~~~~~~~~~~~~~~~~~~1,~~~~~~~~~~~~~,

.~~~~~~~~~~~~~~~~~-

,qr~~~~~~~-,I 11

)

iBSA) A1,

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 2.0 2.4 2.8BODY SURFACE AREA IN SQUARE METERS

FIGURE 4. Plot of end-diastolic left ventricular transverse dimen-sion in millimeters versus a cube root function ofbody surface area

in square meters. In this and all subsequent figures, the 95% predic-tion intervals for the data will be indicated by dashed lines.

suggested that it had the shape of a cube root function. Inorder to test this hypothesis, the logarithm of left ventric-ular end-diastolic dimension was plotted versus thelogarithm of body surface area. The slope of this regressionwas found to be 0.40. This value was between the slope of a

cube root function (0.33) and the slope of a square root func-tion (0.50). Therefore, left ventricular end-diastolic dimen-sion was analyzed by computing the residual sum ofsquares"6 for both the cube root and the square root of thebody surface areas. In this example, the cube root functionproduced the smallest residual sum of squares (table 1) and,therefore, left ventricular end-diastolic dimension was

plotted versus the cube root of body surface area in figure 4.A similar analysis was also performed for the other

echocardiographic parameters (table 1). The cube root func-tion of body surface area produced the better fit of the datafor left ventricular transverse dimension at end diastole, aor-

tic root and left atrial dimensions, and mitral valve E-Fslope. The left ventricular dimension at end systole had es-

sentially the same residual sum of squares for the square

root and cube root functions. Since the end-diastolic dimen-sion was better fit to the cube root function, we chose to use

this same function for the end-systolic dimension as well.Wall thicknesses at end diastole and end systole were betterfit to the square root function of the body surface area whileestimated left ventricular mass was better fit to the direct

TABLE 1. Statistical Results Used to Determine Type of Data AnalysisParameter Residual sum of squares X 103 -2 log Lt Males-females

BSA (BSA) 1/2 (BSA)1/3 Constant Proportional (%A)

LV dimension (diastole) 0.39 0.36 388 383 2.5 *LV dimension (systole) - 0.36 0.36 401 396 5.4 **Septal thickness (diastole) 0.38 0.41 188 182 1.5 NSSeptal thickness (systole) - 0.91 0.95 346 340 -3.4 NSFree wall thickness (diastole) - 0.30 0.31 163 163 1.5 NSFree wall thickness (systole) 0.99 1.00 350 345 -5.2 NSAortic root dimension 0.22 0.20 352 338 5.4 **Left atrial dimension - 0.48 0.47 415 410 5.7 **Mitral E-F slope 32.8 32.4 795 802 2.6 NSLV mass 31.4 45.8 - 797 757 3.6 *

t(-2) times the logarithm of the likelihood function.15*p = 0.05.**P = 0.01.NS = Not significant.

280 CIRCULATION



ownloaded from



TABLE 2. Regression Equations and Prediction LimitsParameter Regression equation 95% Prediction limits

LV dimension (diastole) = 45.2 [ (BSA) 1/3] - 6.6 10% (of mean value)LV dimension (systole) = 28.3 [ (BSA) 1/3] - 3.8 18%Septal thickness (diastole) = 6.4 [(BSA) 1/2] + 1.1 19%Septal thickness (systole) = 7.6 [(BSA) 1/2] + 2.1 21%Free wall thickness (diastole) 6.6 [(BSA) 1/2] + 0.9 16%Free wall thickness (systole) = 9.7 [(BSA) 1/2] + 1.5 20%Aortic root dimension 26.8 [(BSA) 1/31 - 6.1 16%Left atrial dimension = 28.8 [(BSA) 1/31 - 1.9 17%Mitral E-F slope = 131 [(BSA) 1/3] - 14 33%LV mass = 115 (BSA) - 11 32%

measurement of body surface area. As a result of theserelations, we have plotted the figures in this paper so that thevertical axis will display the echocardiographic parameteron an arithmetic scale. The horizontal axis will be arrangedaccording to the function of the body surface area thatproduced the better fit with the particular echocardio-graphic parameter (figs. 4-11).

Graphical inspection of the data also indicated that thevariability (i.e., standard deviation) of many of the echocar-diographic measurements was less at the lower body surfaceareas than at the higher body surface areas. This suggestedthat the standard deviation of our data might be propor-tional to the mean value rather than being constant. In orderto determine if this was so, the data were analyzed using twodifferent assumptions: a) the standard deviation was con-stant at all average levels of the measurement and b) thestandard deviation was proportional to the mean value ofthe measurement. The appropriate assumption of variabilitywas determined for each parameter by identifying whichassumption produced a regression model giving best fit tothe data, i.e., which assumption minimized the quantity (-2times the logarithm of the likelihood function) (table 1).16Except for the mitral E-F slope all parameters were found tobe best fit by the assumption that the standard deviation wasproportional to the mean value of the measurement (leftventricular free wall thickness in late diastole was equallywell fit by either assumption). As a result of this analysis, theprediction intervalsi6 for 95% of normal values that weredetermined for each variable were computed using theproportional assumption. These prediction intervals areshown on each figure as dotted or dashed lines that diverge

50r

z0a 40z

E. 2xo230z

0Cc 20

uw 10

(BSA)'/-// ~I 1 1 1 1

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 2.0 2.4 2.8


FIGURE 5. Plot of end-systolic transverse dimension of the left

ventricle (millimeters) versus a cube root function of body surfacearea (square meters).

from each other as body surface area increases. In thederived parameters that are independent of body surfacearea, a constant 95% prediction interval was used.

Data Relations

In five of the 10 parameters indicated in table 1, theregression equations determined in the male subjects hadsimilar slopes but statistically greater (P < 0.05) y-intercepts than the female subjects. In three of the fourderived parameters of left ventricular function (ejection frac-tion, fractional shortening percentage and percent thicken-ing of left ventricular free wall), the female subjects hadstatistically greater (P < 0.05) values than the male subjects.However, since the absolute magnitude of the differenceswas small (i.e., < 6%), the data from males and femaleswere combined. Although the number of subjects is small,no differences were noted when blacks or Orientals werecompared with Caucasian subjects.

20 -

215Z

0)=co

LU

u5

10

5

AAA,&AsA A

. M A L A

eA1L .A& ..

... e a AgA v. AA4r.

-(BSA)14

IE I I I II, I I II,1, ,102 OA 0.6 0.8 1.0 12 1.4 1.6 1. 2.0 2.2 2A2.6


FIGURE 6. Plot of thickness of the ventricular septum(millimeters) versus a square root function of body surface area(square meters). Systolic thickness measurements are indicated bythe unfilled symbols and diastolic thickness measurements are in-dicated by the filled symbols. The dotted lines indicate the 95%prediction limits for the systolic measurements, while the dashedlines indicate the 95% prediction intervals for the diastolicmeasurements.

281



ownloaded from



A A

_* aA

AA

-A A&b66AAXb MA AA A.

A A^ A-' A

_ .--A A A --AAAA

AMAA A&A~ A4AM%

A. A A-A

e -0

TABLE 3. Correction Factors for Determining Normal ValuesObtained by Other Measurement Methods

Left ventricle (internal dimensions at end diastole)At peak of R wave -0.01**At onset of Q wave -0.02**

Left ventricle (wall thicknesses at end diastole)Septum at peak of R waveFree wall at peak of R wave

Aortic root (undamped record in late diastole)Outer wall to outer wallOuter wall to inner wallInner wall to inner wall

Aortic Root (undamped record in late systole)Outer wall to inner wall

Left atriumOuter aortic wall to atrial wallInner aortic wall to atrial wall

-0.03**- 0.08**

+0.12**-0.005*-0.14**

- 0.06**+0.05**

(BSA)"20.2 III0.4 0.6 0.8 1. 1.2 I1. 1. 2. 2. 2.4I26I0.2 OA4 0.6 OS 1.0 1.2 1A 1.6 1.8 ZO 2.2 Z4 6


FIGURE 7. Plot of thickness of the posterobasal left ventricularfree wall (millimeters) versus a square root function ofbody surfacearea (square meters). As in figure 6, systolic measurements are in-dicated by unfilled symbols and diastolic measurements by filledsymbols.

Most of the echocardiographic parameters were found tobe related to body surface area. These parameters are

plotted versus the appropriate root function of body surfacearea in figures 4-1 1. The regression equations relating eachparameter and body surface area as well as the 95% predic-tion limits are given in table 2.

Four parameters (ejection fraction, percent fractionalshortening of the left ventricle, percent thickening of theventricular septum, and percent thickening of the left ven-

tricular free wall), when plotted against body surface area,

had slopes that did not differ from zero. Thus, theseparameters can be considered to be independent of body sur-

face area. The 95% prediction interval limits for ejectionfraction are 64 and 83% (mean 74), while those for percentfractional shortening are 28 and 45% (mean 36). The 95%prediction interval limits for ventricular septal thickeningare 14 and 48% (mean 31) and for left ventricular free wallthickening are 29 and 69% (mean 49).

50F

z

z0

(10z 30

0

0 200cc

10

z 40

z

z 30

20

, 10

.~~~~~~~~~~~~~~~~

(BSA)P3I/ I,I II IIIIIIII I,,I, , ,, ,

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 2.0 2.4 2.8BODY SURFACE AREA IN SQUARE METERS

*P <0.05.**P <0.01.

The data obtained using other described or suggestedtechniques (e.g., left ventricular end-diastolic dimension atpeak of R wave) were also analyzed (table 3) and comparedto the data plotted in figures 4-1 1. The percentage differencebetween the data obtained using our method and that ob-tained using these alternate measurement methods was com-

puted. In each instance, the percentage difference was ap-

proximately the same in subjects with smaller body surfaceareas compared to subjects with larger body areas. As a

result, the numbers shown in table 3 can be used to calculatethe normal values at any given body surface area for each ofthese alternate measurement methods. For example, if aor-

tic root dimension was measured in late diastole in the un-

damped portion of the record from the outermost echo ofthe anterior aortic wall to the outermost echo of theposterior aortic wall, it is possible to derive the range of nor-mal data at a given body surface area in the followingmanner. First it is necessary to either go to the appropriatefigure (in this instance fig. 8) or use the regression equationand determine the range of normal values at the particularbody surface area. For a body surface of 0.5 m2, the upper

and lower 95% prediction limits are 17.4 and 13.0. The cor-

rection factor obtained from table 3 is + 0.12. Multiplyingthis correction factor by the 95% prediction limit values ob-

50 r

'SA At_ ~~~Y\A A d iSi;--

(BSA) 1/A} i I IIII0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 2.0 2.4 2.8


FIGURE 8. Plot of aortic root dimension (millimeters) versus a

cube root function of body surface area (square meters).FIGURE 9. Plot ofleft atrial dimension (millimeters) versus a cuberoot function of body surface area (square meters).

20r

z

,,, 15

z

J

w 10wrIL

>5

282 CIRCULATION



ownloaded from



250 -

z

0

0

uUEn200_2

Xj150

50E

tab AAA AAAA A

A~~~~4A 4A~A A

-BSA)h_

0.2 0.4 0.6 0.8 1.0 1. 14 1.6 2.0 2.4 2.8


FIGURE 10. Plot of early-diastolic closing velocity of the anteriorleaflet of the mitral valve (E-F slope) expressed in millimeters persecond versus a cube root function ofbody surface area expressed insquare meters.

tained from figure 8 or the regression equation and addingthe result to the previous prediction limits yields a range ofnormal values of 19.5 to 14.6 mm. In a similar manner, thenumbers given in table 3 can be utilized to calculate therange of normal values obtained when using any of the otheralternate measurement techniques.

Discussion

The results of this study of normal subjects indicate thatechocardiographic measurements of the internal dimen-sions, thicknesses and mass of the chambers of the heartfollow a linear regression upon one of three functions (directlinear, square root, cube root) of the body surface area. Theinternal dimensions of the left ventricle, the left atrium, andthe aortic root and the mitral E-F slope vary in a linear rela-tion to the cube root of the body surface. Stated anotherway, cubing these measurements produces a linear relationwith body surface area. The wall thicknesses of the left ven-tricle, however, vary in a linear relation to the square root ofthe body surface area while estimated left ventricular massvaries directly with the body surface area. Ejection fraction,fractional shortening percentage, and percent thickening of

300-

"20IDz 200In('

cc 150:5

~-100zu50

K

I -

Ksoo -A AA

AA-1 AA

A AA

A

tl:

I TW200i

the ventricular septum and left ventricular free wall are in-dependent of body surface area. These relations appear to beconsistent with the study of Lundstrom who found that theleft ventricular dimension at end diastole and the aortic rootand left atrial dimensions vary as a cube root function ofbody weight.7 These relations are also largely consistent withthe data of Epstein et al.5 Although not stated in theiroriginal paper, subsequent publication of the data graphs intwo textbooks"2 has included the notation that left ventric-ular end-systolic dimension, left ventricular septal thickness,aortic root dimension and left atrial dimension are rootfunctions of the body surface area. The specific root functionis not described. Also, the left ventricular end-diastolicdimension is indicated to vary as a linear function of thebody surface area while the left ventricular posterior wallthickness is described as a logarithmic function. Our data in-dicate that all of the dimensions of the heart (including theleft ventricular end-diastolic dimension and the left ventric-ular free wall thickness) appear to vary as root functions ofthe body surface area. As indicated above, we believe thatspecific root functions can be chosen that best fit theseechocardiographic parameters.The three methods used to assess the systolic function of

the left ventricle (ejection fraction, percent fractionalshortening, and percent wall thickening) are essentially in-dependent of body surface area. Despite an almost three-fold increase in the linear dimensions of the left ventricleduring growth (and hence an approximate 8-9 fold increasein its volume), the percentage of blood ejected, the percent-age the diameter shortens, and the percentage the wallsthicken each beat is constant. Thus, in terms of using theseparameters to assess left ventricular function, all have the at-tractive feature of being independent of the growth of theheart. However, ejection fraction may have an advantage inthat the scatter of the ejection fraction data (± 13%) is con-siderably less than that of the percent fractional shorteningdata (± 23%) or the wall thickening data (+ 48%).The actual numerical values of the normal data reported

in this study are in general agreement with previouslypublished data. This is true even in the extremes of body sur-face area in that the measurements previously published forinfants from several centers'7-"1 are in good agreement withour data in infants. Previously published normal data inadults' are similar to our data for individuals with largerbody surface areas. Although the data of Lundstrom are ex-pressed as a cube root function of body weight, estimation ofthe corresponding body surface area indicates that our dataand those of Lundstrom are in good agreement. In the threeparameters common to both studies, the mean values andthe variance of the data are similar.The shape of our data curves (and in most cases their

values as well) are also in agreement with the data of Epsteinet al.5 However, some discrepancies were noted. For exam-ple, our 95% prediction intervals are considerably wider thantheir 5th and 95th percentiles. Although not stated in theiroriginal manuscript, the 5th and 95th percentiles representthe confidence limits for the regression equation relating thedata and do not indicate the standard deviation of the data(personal communication). At the median value of body sur-face area, these confidence limits define the standard error ofthe estimate of the mean value. Hence, the 5th and 95th

0.5 1.0 1.5 2.0BODY SURFACE AREA IN SQUARE METERS

FIGURE 11. Plot of estimated left ventricular mass (grams) versusa linear function of body surface area (square meters).

0 - -

283



ownloaded from



percentiles indicated on the figures of Epstein et al. do notprovide a clinically useful indicator of the scatter of normaldata and cannot be used in their present form to identify in-dividuals with abnormal cardiac dimensions. Another dis-crepancy was noted in the values for left ventricular wallthickness (it was this discrepancy that led us to collect ourown normal data). In the data of Epstein et al., wallthickness values for individuals with body surface areas of1.7 to 1.8 m2 ranged from 6 to 7 mm.5 Adult normal valuesfor individuals of similar body surface area are in the 8 to 11mm range.' In order for the values of Epstein et al. to becompatible with the adult data, a significant increase in wallthickness would have to occur in early adulthood and atbody surface areas between 1.6 and 1.9 M2. After collectingand analyzing our normal data, we did not find such an in-crease but, rather, a smooth curve relating wall thicknessand body surface area.

Several factors could account for the difference betweenthe two studies in regard to wall thickness measurements.First, the wall thickness values were measured duringdifferent portions of the cardiac cycle. In our study, wallthickness measurements were made in late diastole prior tothe thinning of ventricular septum and posterobasal freewall. Wall thicknesses in the study of Epstein et al. weremeasured at the onset of the Q wave and therefore may haveresulted in measurement during thinning of the wall. In ad-dition, most of the data in our study have been collected inthe past few years using ultrasound equipment that has beenmodified to simplify measurement of wall thickness (i.e.,switched-gain circuit)." We believe that both technical andmethodological differences account for the discrepancynoted in wall thickness measurements.

Because of the obvious importance of how the echocardio-graphic measurements were made and because standards formeasurement have not yet been agreed upon, we decided touse a variety of described or suggested methods to measurethe echocardiographic parameters. Although the dataplotted in figures 4-11 were derived using the methods weprefer to use, the data given in table 3 can be used to derive(from our data graphs or regression equations) the corre-sponding normal values obtained using these other measure-ment methods. Examination of these correction factorsemphasizes that rather large differences in normal valuescan result when data obtained using different measurementmethods are compared. For example, aortic root dimensionsobtained in the undamped portion of the record from outerto outer aortic wall are approximately 25% greater thanvalues obtained from the inner to inner aortic wall.

Regardless of the method used, it appears useful to con-sider the data in relation to the various root functions of thebody surface area. Using this approach, linear regressionequations can be written which simplify computation whendetermining normal values for a specific body surface area.The specific relation to body surface area was determined byidentifying the function that produced the best fit to thedata. In some instances (e.g., left ventricular dimension atend diastole), the cube root function clearly produced abetter fit than the square root function (table 1). In other in-stances (e.g., left ventricular dimension at end systole), thechoice between the cube root and square root functions wasnot clear. Graphical analysis of the data revealed that eitherroot function produced an acceptable fit to the data.

Therefore, our decision to use the cube root function wassomewhat arbitrary. In these instances, we decided to usethe function that resulted in similar measurements (i.e., wallthickness) being analyzed by the same root function.

In addition to simplifying computation, plotting thesedata versus the various root functions of the body surfacearea has illuminated several interesting relations. One suchrelation pertains to cardiac dimensions in newborn infants.When left ventricular transverse dimension at end diastole isplotted versus the direct linear function of body surface area(fig. 3), the newborn infant data appear consistent with theoverall shape of the curve. However, plotting these data as afunction of the cube root of the body surface area (fig. 4) in-dicates that the data from the newborn infants cluster nearor slightly below the lower 95% prediction limit. In fact, 8 of13 infants had values less than or equal to the value of thelower 95% prediction limit. This is a much greater numberof data points on or below the prediction limit than would beexpected (P < 0.01). A similar clustering of values equal toor less than the lower prediction limit was found for the leftventricular free wall thicknesses (at both late diastole andend systole), left atrial dimension, ejection fraction, and per-cent left ventricular free wall thickening (P < 0.05 in eachinstance). Thus our data appear to indicate that the internaldimensions, wall thickness, and function of the chambers ofthe left heart are reduced when compared to themathematical relations with body surface area that existfrom one month to 23 years of age. It is tempting tospeculate that these reductions are due to the fact that theleft ventricle and left atrium are partially by-passed duringintrauterine life and thus do not develop the expected rela-tion to body surface area until after birth. This speculation iscompatible with direct physiologic measurements in ex-

perimental animals.22 23 Further studies will be necessary toconfirm these observations and to characterize the timecourse of normalization of this apparent decrease in thedimensions and function of the left heart chambers duringthe first month of life.

AcknowledgmentThe authors acknowledge the superb technical assistance of Ms. Cora Burn

and Mrs. Estelle Cohen in obtaining the echocardiograms. The assistance ofDr. Stephen E. Epstein in reviewing and of Mrs. Exa Murray in typing themanuscript is greatly appreciated. Also Dr. Arthur Nienhuis, whocollaborated in a previous study of thalassemic patients, was greatly responsi-ble for the initial interest in collecting these normal data. The authors alsogreatly appreciate the assistance of Ms. Yee Wong in helping to analyze thedata.

References1. Feigenbaum H: Echocardiography, ed 2. Philadelphia, Lea and Febiger,

19762. Goldberg SJ, Allen HD, Sahn DJ: Pediatric and Adolescent Echocar-

diography: A Handbook. Chicago, Yearbook Medical Publishers, 19753. Gramiak R, Waag R: Cardiac Ultrasound. St. Louis, The C. V. Mosby

Company, 19754. Williams RG, Tucker CR: Echocardiographic Diagnosis of Congenital

Heart Disease. Boston, Little, Brown and Company, 19775. Epstein ML, Goldberg SJ, Allen HD, Konecke L, Wood J: Great vessel,

cardiac chamber, and wall growth patterns in normal children. Circula-tion 51: 1124, 1975

6. Allen HD, Goldberg SJ, Sahn DJ, Schy N, Wojcik R: A quantitativeechocardiographic study of champion childhood swimmers. Circulation55: 142, 1977

7. Lundstrom M: Clinical applications of echocardiography in infants andchildren without heart disease. Acta Paediat Scand 63: 23, 1974

8. Henry WL, Nienhuis AW, Weiner M, Miller DR, Canale VC, Piomelli

284 CIRCULATION



ownloaded from


GROWTH PATTERNS BY ECHO/Roge et al.

S: Echocardiographic abnormalities in patients with transfusion-dependent anemia and secondary myocardial iron deposition. Am J Med,in press

9. Boothby W, Sandiford RB: Nomographic charts for the calculation of

the metabolic rate by the gasometer method. Boston Med Surg J 185:337, 1921

10. Henry WL, Clark CE, Epstein SE: Asymmetric septal hypertrophy(ASH): Echocardiographic identification of the pathognomonic anatomicabnormality of IHSS. Circulation 47: 225, 1973

11. Griffith JM, Henry WL: Switched-gain: A technique for simplifying ul-

trasonic measurement of cardiac wall thickness. I.E.E.E. Trans BiomedEngr 22: 337, 1975

12. Troy BL, Pombo J, Rackley CE: Measurement of left ventricular wall

thickness and mass by echocardiography. Circulation 45: 602, 197213. Popp RL, Harrison DC: Ultrasonic cardiac echography for determining

stroke volume and valvular regurgitation. Circulation 41: 493, 197014. McDonald IG, Feigenbaum H, Chang 5: Analysis of left ventricular wall

motion by reflected ultrasound: Application to assessment of myocardialfunction. Circulation 46: 14, 1972

15. Brown OR, Harrison DC, Popp RL: An improved method forechographic detection of left atrial enlargement. Circulation 50: 58, 1974

16. Kendall MG, Stuart A: The Advanced Theory of Statistics, vol II. NewYork, Hafner Publishing Company, 1961

17. Winsberg F: Echocardiography of the fetal and newborn heart. InvestRadiol 7: 152, 1972

18. Meyer RA, Kaplan S: Echocardiography in the diagnosis of hypoplasiaof the left or right ventricle in the neonate. Circulation 46: 55, 1972

19. Solinger R, Elbl F, Minhas K: Echocardiography in the normal neonate.Circulation 47: 108, 1973

20. Hagan AD, Deely WJ, Sahn DJ, Friedman WF: Echocardiographiccriteria for normal newborn infants. Circulation 48: 1221, 1973

21. Godman MJ, Tham P, Kidd BSL: Echocardiography in the evaluation ofthe cyanotic newborn infant. Br Heart J 36: 154, 1974

22. Friedman WF: The intrinsic physiologic properties of the developingheart. Prog Cardiovasc Dis 15: 87, 1972

23. Kirkpatrick SE, Covell JW, Friedman WF: A new technique for the con-

tinuous assessment of fetal and neonatal cardiac performance. Am JObstet Gynecol 116: 963, 1973

Cardiac Structure Growth PatternDetermined by Echocardiography

CLAUDE L. L. ROGEI, M.D., NORMAN H. SILVERMAN, M.D.,

PATRICIA A. HART, M.A., R.D.M.S., AND ROSE M. RAY, PH.D.

SUMMARY Using M-mode echocardiography, we measureddimensions of the ventricular walls and cavities, great vessels, and leftatrium and atrioventricular valve excursions on 93 infants andchildren without heart disease. The data were analyzed by relatingeach dimension in mm to body surface area in m2 and the 90%

M-MODE ECHOCARDIOGRAPHY is useful inevaluating the child with heart disease. It allows theanatomic relationships of chambers, great vessels, andvalves to be assessed, and allows wall thickness, cavitydimensions, great vessel diameters, and atrioventricularvalve excursions to be measured. These measurements, whencompared to normal data, can be used quantitatively tomake judgments about normality. Several studies alreadyprovide such normal values for neonates" 2 and adults,3'but there is only one study providing "normal" values ofechocardiographic dimensions in the growing child withrespect to body 'surface area;' this study has thereforebecome the standard.3 6, The range of suggested normallimits in that study is narrow, and many of the childrenwithout significant heart disease whom we examined hadechocardiographic measurements outside of these previouslyestablished limits. We therefore reviewed the echocardio-graphic records of 93 children and adolescents without heartdisease examined at the University of California during thepast two years in order to re-evaluate the limits of nor-

mality.

tolerance limits for the data were calculated. The tolerance lines ofthe data were wider than previously recorded. At birth and maturitythey were similar to the range defined as normal by studies inTieonates and adults. We suggest that the tolerance lines of these nor-

mal data may be used for quantitative echocardiography in childhood.

Method

The echocardiograms of 93 children and adolescents, age

one day to 18 years, were used in this study. These subjectswere thought by two pediatric cardiologists to be free of anysignificant heart disease. Most were outpatients referred forevaluation of a heart murmur which was found to be inno-cent on clinical, electrocardiographic, and radiologicgrounds. The others, especially newborns, were inpatients inwhom cardiac evaluation and follow-up failed to detect any

heart disease.The patients were examined in the supine position with

slight left shoulder recumbency. The tracings were obtainedwith commercial M-mode echocardiographs and strip chartrecorders. Appropriate transducers (2.25, 3.5, 5, and 7.5MHz) were used to define cardiac structures. The echocar-diograms were obtained from standard precordialpositions.6' 8, 9

Right ventricular anterior wall thickness (RVAWD), leftventricular posterior wall thickness (LVPWD), right and leftventricular cavity diastolic dimensions (RVDD andLVEDD), interventricular septal thickness (SEPT D), andmitral valve excursion (MVDE) were measured at the levelof the posterior mitral leaflet at end-diastole, defined by thepeak of the R wave on the ECG (fig. I a and b). Left ventric-ular end-systolic dimension was measured at the peak up-ward motion of the posterior left ventricular endocardium.The right ventricular anterior wall thickness was obtained by

From the Cardiovascular Research Institute and the Department ofPediatrics, University of California, San Francisco, California.

Address for reprints: Norman H. Silverman, M.D., 1403-HSE, Universityof California, San Francisco, California 94143.

Received June 27, 1977; revision accepted September 16, 1977.

285



ownloaded from


W L Henry, J Ware, J M Gardin, S I Hepner, J McKay and M Weinerbetween infancy and early adulthood.

Echocardiographic measurements in normal subjects. Growth-related changes that occur

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 1978 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation doi: 10.1161/01.CIR.57.2.278

1978;57:278-285Circulation.

http://circ.ahajournals.org/content/57/2/278the World Wide Web at:

The online version of this article, along with updated information and services, is located on

http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer information about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. FurtherEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculationpublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:



ownloaded from

http://circ.ahajournals.org/content/57/2/278

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circ.ahajournals.org//subscriptions/


echocardiographic measurements normal...

Documents