Z scores

The use of Z-scores in paediatric cardiology
Henry Chubb, John M Simpson
Department of Congenital Heart Disease, Evelina Children’s Hospital, Guy’s and St Thomas’ NHS Trust, London, UK

1. The use of Z-scores in paediatric cardiology

Henry Chubb, John M Simpson

PTS anne mern nen naiss va
nosrancr

Zocor area nena expressing te drétion Fa givenmonanenet fromthe sat orge ape population
‘mean By taking acount of roth or age, Zs oes ore an excl means af hat serial measurements in
[pnd cardologcal price Theycan be applied to echocandigraphi measurements, blood pressure and

Patent gro, and sy sit nl

el

Keywonts: Blood pressure, codi, echocardiography, pora caology, Zscore

WHAT ARE Z-SCORES?

Measurements arc an portan put of mal assesment
An the practice of paediatric cardiology. At the most
‘asic Level this fypeally includes measurement of the
llas weight, height and blood pressure at clinic visit.
Echocardiographic assessment 4 integral to patient
assessment ina majority of arena decisions ath
pect to surgery or cate intervention ae frequent
based on echocardographic findings. In adult practico,
Witrasound measurements are often reported with
respecto a single “normal range” but this approach i
{mmpossblein growing children becuse te norma! range
of measurement wil be impacted by patient growth
and / or patient age. Therefore, the interpretation of
‘Hnsemeastrements during chldhood presents unique
Challenge in determining whether a given measurement is
‘within the expected range In addition, a measurement
ttes from normality its necessary for the clinician
o gange the magnitude of such ation

An approach to the description of clinical and
‘echocardiographic variables is to describe the
measurement terms ofa Z score, The Z score describes
how many standard deviations above or below a sie
or age pee population mean à given measurement
lies, Tis approach has major attractions in paediatric

NE o
ae Saone

cardiology and fs increasingly being adopted. As an
«example, the left ventricle wil become larger in al
“re as they grow: However, fa patient vith chronic
aoric or mitral valve regurgitation is being followed.
(hrong serial assessment then clearly tie an anormal
and inappropriate dation ofthe left venice that mst
e excluded. These of scores faites the detection
pathologie creases sn Left ventricular dimensions,
over and above that expected due to normal growth y
Showing an increased Z score over tie.

Z-SCORES VERSUS CENTILES

Many paediatricians and paediatric cardiologists are
Familar with cents, particularly with regardto paient
ete and weight. The relationship between cents ad
“scores, fora normally distributed parameter shown
An Figure 1; thee of ether means of expresion asses
‘noma distrito of the dat.

DERIVATION OF Z-SCORES
AZ acone ts defmed as? 2
here the observed measurement, i the expected
‘measurement (population mean) and 95 the standard
evsation of the population. Thus Zscores above the
Population mean havea pasive value and those below
‘he population mean havea negative value. The Z score
value conveys the magnitude of deviation fromthe mean.
For example, where the mean ste ofthe orc valves
20mm, with a defined standard deviation of 3 mu the
‘score of am aortic valve th ans 14 mm

as

‘oso ep Dejo Spun Ena Ours Hoe Go Moras Hota NS Tu Lone U

ena names

2. ‘Therefore, in order to calculate a Zscore one must
define the mean for exch body size point and the
corresponding standard deviation, These values have
‘been derived in many separate Studies, of varying
sample sizes, (Table 1) and the word ‘llometry” has
been sed since 1936 to describe the “relationship
ten changes sn shape and overall ze". The
determination of the "best relationship ree upon.
‘nding the best mathematical fit forthe data, and it
has ong been recognised that the correlation between
size of body structure (eg, heart valve) and surrogate
marker of total body size (e. body surface area
‘or weight) s rarely a simple tinear one! In other
‘words, structures do not tend to obey the relationship

Far Te na of Zar an cates, auna
‘inal constant alas tnt rom e mean penny over
Hea antun rn mun) vt a Zar crias

Y =aX +D (were y isthe measured structure, xis the
marker of total body size 1 the sealing coefficient
and bisa constant

Instead, a much better ft is generally shown by more
‘complex polynomial equations (eg. yan sexed) or
by apower lawrdationship (mae, whereas theses
factor, and D the scaling exponent).

Once a best fit relationship 4s defined. a standard
eviation and mean can be derived at each body
Size point. It is important to note that the standard
ewtation may vary vith body size, a property termica.
ter tot.

WHICH PATIENT FACTORS SHOULD BE
CONSIDERED?

‘The question of which patent factors shouldbe seo,
reflec overall body sie has been addressed in recent
Published guidelines” For most measurements the
recommendation has been to calculate Zacores vt
respecto patient body ace area rather than height
xr Weight alone! Many formulas have been used to
late body surface area including those of Boyd,
Dubois and Dubai and Haycock and shouldbe noted
tat there ss considerable discrepancy ın the values
derived by exch forma, particularly a low body size
Recent published gridance has recommended the use of
‘he Haycock formula, but there 1 a strong argument
that for vata comparison thesame BSA formula should
ed sin the original reference shady wach eas wed
to compute the scores.

‘Table 1: Key references with respect to important paediatric and fetal echocardlographic data

uo Year Numberofsubjects Parameters measured
Poner al zus m AS
meta zu ma Honra
nat
Dadowney ra 1998 ss Hours andate, ora, puma anales
ci
Datei agora am 1009 coronary ania
Mebane mas 2007 a Corona ari
sn
Sir era zus “= Coronary aris
‘Sores arabs zus Foul banat nennen
Etre £a ne inn e er
Em sar Zo Tite dpi eases Im Ma pued
epee
Komseneng ora Pr ise ;
owt ropes ie ood reas (Cc aig

3. However, it should not be assumed that body surface
area is always the favoured approach. For left
‘ventricular mass, for example, patient height has been
Preferred" Other factors, such as patient sex and
ace. may also e important for crta mensuremen
Echocardiographic measurements do not simply
relate to metsuentent of the size of heart stitutes.
Normal values of funcional data such as Blood pool
Doppler and tissue Doppler data also change across
the paediatric age range, and are heal Mec
by patient age" and heart rate.

APPLICATION OF Z-SCORES

In order to optimise the relevance of the computed
“score sn the cal station, the methodctogy used.
forthe messurement should match, as far as posible,
‘that ofthe orginal investigators.

In order for body surface area to be computed,
conventionally both height and weight are required.
and our normal practice has been to record both
‘whenever possible, However, fom experience, weight
ds recorded more commonly than height, particularly
An smaller babies and children. Weight any equations
for the calculation of BSA, such as BSA = 0.1023
(right), are convenient ot sk mssappropriaing
eZ score dat. n asin, for some cchocankographic
‘measurements there may be varaılty as o wheter a
measurement is madein systole or diastole or whether.
Leading edge to leading edge versus internal dimension
method has been sed

One of the most useful applications of Zscores 4 in
‘tacking allometrie growth over time. The same Zscore
algosti sho be wed each time. anda note made of
‘which reference data has been employed. Z cores may
vary agit between different authors (Figure 2).
fret 7 core references are wed in he erroneo
ele that these are interchangeable then an apparent
‘change in 2-score may be produce in the absence ofa
ue variation. In practice, an stitution should decide
‘which reference data 15 going to be used for which
measurements and this should be weed consistently
‘desig patent fll.

An example where cores can be very informative i

An following progressive dation of the aortic root in a
patient wth Marfan's syndrome, The diameter at the
mes of Valsalva val increase asthe ld grows, but
at times of rapid total body growth can be dict
to detec disproportionate growth ofa single structure,
Table 2 strates such a insel scemano where child
starts to grow rapidly a the age of 10 years The aortic
toot grows even faster, and the significant increase i
{he Z score ofthe sims of Valalva demonstrates this
succinctly and clearly, alerting the inician.

LIMITATIONS OF 2-SCORES

scores have major advantages fr the presentation of
paediatric echocardiographic and physiological data
However, scores remain an imperfect approxsnaten,
scussion. Theis, and most important, factor to bear
Anis that the mean and standard deviation at each
‘ody size point are only estimation and may vary widely
between investigators (igure 2)

Secondly, to have statistical confidence in both the
mean and the more extreme cores an extremely large
Sample size is rele, particulary in the asset
of patents across a wide range of sizes. The change in
‘ariance across body size (heteroscedasticty) means hat
{tis important to include sficent munbers of patents
At extremes of body size. The inappropriate averaging of
tance wonddtend1o undereimaleZs0oresvals for
te smallest children, and overestimate Zscore valves
for the anger sua.

Thardly, the use of Z-scores may amply errors in
measurements, A degree of intra and interobserver
Variability in measurements 4s unavoidable, Dut
‘scores can ampli only mansa! changes sn absolute
measurement (Figure 2b), particularly at negative
Zscores The exact mature of any amplification effet
‘depends upon the algoitim used to derive the Zscore,
And some do not ext such a phenomenon.

nally, 4 45 portant thatthe wer does not accept
published data mentally If there isa wide standard
‘eviation for example in smaller patients, then a vale
fof zero can appear toe within 2 standard devsations of
‘Hemean, For instance, a complete absence of movement

“Table 2: Increasing Zscore over time ofthe sinuses of Valsalv, suggesting pathological aortic root
dilation
Age (ears) Helght(om) Weight (ig) Simuses of Vosaiva mm) Zacors Detroit)

4. Manz
nd gore Vale

‘on tissue Doppler at some postions, in the under one
year age group, may yield a score of > 2.8

HOW TO CALCULATE Z-SCORES IN
PRACTICE

For the dima tisimpractcal otra throng mde
paper reference sources ina ci ting, Us portant
‘Hugh that theres theataty to alte scores when,
required and there are some commercially available
‘echocardiography archive packages that faiitate this
‘Thon convenient, these integrated packages are often
imate toa single set of reference data and have not
‘exter to the incorporation of functional data such
as oo pool o issue Doppler: Fetal ochocanographic
tas avale on some arche soins ado others.

Some institutions do not have image archiving systems
installed and co web based calculators have evalved.
Probably the best known of these eat wwveparametre.
‘com. here the ser can input Ihe relevant data for online

‘leuaton of scores tse on published reference data
‘nar on unit has recently released the "Cari 7° APD
‘sich canbe installed on an Phone pad platform and
{oes not depend onan Internet connection This pens
ser configuration of the preferred reference source
for multiple fetal and paediatric echocardiggraphic
measurements. The user has the option to change the
body surface aren formada, Dt the deft postion i=
to we that of the original reference. Diagrams deta
the exact measurement methodology used by each
paper, in order to facia valid comparzons with the
reference data. Functional dat, such as pulsed wave
and tissue Doppler with derived functional indices,
fom alo be entered, Supplementary information such.
As scores for growth and blood pressure, and normal
‘ectrocardiographic data can also be analysed. A
screenshot ofthe App is shown in Figure 3.

CONCLUSIONS
‘These of scones represents a ph ay of presenting

5. Z-Scores

patient spect information in paediatric cardiclogy
Users need to have an avareness of how such scores
are calodated and their limitations, Purpose «designed
Software 1s making access to Z-score information far
‘easier than twas possible the past

DECLARATIONS

THC and JMS are the joint authors of Caro Z an App
‘develope for Phone and availabe for purchase. Profits
And are donated to the Evelina Children's Hospital
Appeal

REFERENCES

1. Haley JS, Tester G Ternanclogy of else Grow
Nana 16361377801.

2. Medahon T. Size and shape in biology

3, Nolan TG, rad AD King ME, ya AE Desa
da siteindependent vaable for sel of cuite
‘Gmnensions 1 4 normal peditie population. Eur J
Folclor 2001080

Science

4. Stuysmans T, Colan SD, Theoretical and empirical
‘ovation of canvas encre relationship in
ln) Appl Pass 2005398415

5. Lopez Coban SD Froumelt PC, Eng GI, Kendal K,
methode during the performance of à pediatric
Echocardiogram: À report from the Pediatric
Nasen Wing Group of te American Soi
‘of Echocardiography Pedic and Congenital Heart
Boe Gach hse Pant AN
method for measunng body suface avez: À height

‘Sel form validate fat, dln, nd des

Tear 197895626

Roman SU, de Dista O, er al Lal eme mac

and body ste in normotensive ddr and ads

eri: An Gl Coral 1992220125160,

8. Khoury PR Misnefes M Daniels SR, Kall TR Age
Specie reference tervals for mdexca let vente
Bain dren l'An Soc Echocdog 2009322709.

9. Foster BJ Mackie AS, Mtsnefes M. Alt H, Mauer 8,
(Clan SD. A novel method of expresnglef ventricular

10. iden BW, Mahon CL Cohen RR Ws, ikleteya
Kovach Je pac of cardiac prt on Doppler
Bee ging Velos Asay callo alee. J
Ka Soe Rainer 200487212 21
Shale and dastle tne intervals measured from
Doppler tse inaging Normal values and Z.core
table and efecto age heart ate, and boy suas
rea j Am or heard 2008211301 70.

12. Petersen MD, DW, Skeens ME Hoes RA. Regression
{Sr large cobort of hall tanto, caros, and
‘iolescents Au echocariogrape study ] Aa Soe
Eshocanlog 2008521922 34

13. 2thermon NV, Khoury PR KünballRT Two «menciona
chocardiographie valve measurement du Realty
‘Stren: Genderspeae differences Pediatr Cardiol

14. Daubeney PE, Bickstone IL Weintrub RG, las 2,
Scanlon I, Webber SA Relata ofthe dimension of
Se suce to body sr! tm echocardloprephie
Shady la normal adas nd children, Goal Yong

hoff CM, Wenzel A, Stole 6,
Beiancor NL Wppermans Ce a Normal aes ef
mode hoc plie messwenets o moxe ta
2000 healthy mine and enden ental Europe

16. Warren AE Boy ML, "Cannell Dodds Dilatation of
{he ascending acta puc patients vn nep
arte valve eee rate of progression and ck
tore Hee 2000502 196: $00.

6. 17. Gautier N, Detaint D, Fermanian C, Aegerter P,
Delorme 6; Amon F et al Nomograme for orte
carga. Au] Cadel 2010105:88804.

Dale E, Dada N. New Equations and a Catal
Appralal of Coronany Artery 2 Scores in Healy
lares] Am Soc Echocaiogs 2011520074,
NicCidle BW, L JS, Musel LA, Cola SD, At AML
Taha M er al Coronary artery moles ı
Gilden with Kana disease: ak facıare rom
‘uals oral named Creation
nera.

‘vert, Ang, Fier NL Sable C oromany artery
{core regresen equations and cleats derived
fom lage heterogenemns population of cilren.
ego echocndiogapy. J Am Soc Echo io
POSE

EN
Schneider € MeCrndle B, Cara JS Hornberger,
Man KE Daubeney PE Development of scones
for fn cardiac menden from spy.
Ührasound Obstet Greed 2005-2000 405.

».

20.

a

Pau L Melander M, Seale À Mat Ro NL
Ho AL Locos othe fetal sxc ias ad dt

“Quick Response Cade” ink fr fll text articles

Amat to ases ach hypopla, Urasound Obstet

‘cca 2007200283,

Kocsteberger N, Ravekes WY, Everett AD, Steger HP,

eal Gamocheg Ae al Right sense actin

dba calco an adden Reference het

be wcuspd amd plane syste exec (TABS m.

o healthy patents ud calculan À? cor vale]

am Soc Echotandiog 2009227159

24. National High Blood Pressure Education Program
Woxlang Group on High Hood Pressures Crea
and Adcessens. The fourth sept u the diagnose,
‘valuation, and treatment of high blood pressure i
ren and adelwecute Pda 200817402 Suppl
ka 76

25, Bley BJ, Bss GL. Etting the sue aca of the
Iba Boy Stat Med 199631511525 32.

2.

on en is wil GIE Sesa I Terz
a co, io Ge OS 1508

Source of Sppart Crta ret Ha 5

‘Schon el Cato to up apa row ond oo
acs Pos tnino o und marta ro racer

Eu snaestobe in Curr et ees

‘Th jalisco hs au

eee ett a joa website what ig ón et Each are nis st pa

has a Quick Response Code. Usng any mail a aha ho dv vi camera and GPR ar tom sure, an can nach

tothe totor atari
mom ya) and pito camara to OF odo pits
arco. Oe can abe ses eshte or lp ith web camara o
amy torte apes

aril ent puma! web te. Start aOR cn rad sar (

tte appear
inthe eral wi utomatay tke yo he HTL fl wett at
y Sar uncanay Se yu con 27 rt.