Andersons pediatric cardiology 291

besymptomaticshortlyafterbirth,theyalsomaynotpresentuntilafterthe
patientisdischargedfromthehospital.Thedelayindiagnosismayincrease
mortality.Inresponse,manyofthestatesintheUnitedStateshaveimplemented
pulseoximetryscreeningforcriticalcongenitalheartdisease.Thisisperformed
atleast24hoursafterbirthorasdelayedaspossibleifearlydischargeis
planned.Thescreeningusespulseoximetrytomeasureoxygensaturationinthe
rightupper(preductal)andonelowerextremity(postductal).Thescreeningis
consideredpositiveifeitheroxygensaturationislessthan90%orifboththe
preductalandpostductalsaturationsarelessthan95%onthreerepeated
occasionsseparatedby1hour,orifthereisadifferenceof3%betweenthe
upperandlowerextremitysaturation.30Ifthescreeningispositive,an
echocardiogramistobeobtained.Implementingpulseoximetryscreeninghas
beenshowntohaveresultedinasignificantdecreaseinmortalityrelatedto
criticalcongenitalheartdisease.31
Alarge-scalestudyusingauniversalscreeningstrategyforascertainmenthas
yettobeperformed.InvestigatorsfromJapanperformedamorefocused
screeningofnewbornswithechocardiographyaimedatdeterminingthe
prevalenceandnaturalhistoryofmuscularventricularseptaldefects,alesion
knowntoinflateestimatesofprevalence.32Theyscreened1028newborns
withoutevidentchromosomalabnormalitiesandwhohadbeendeliveredatterm
gestation,andidentified21newbornswithmuscularventricularseptaldefects,3
withcoarctationoftheaorta,and1withadouble-outletrightventricle.This
gaveaprevalenceof24.3casesper1000livebirths.The95%confidencelimits
werewide,rangingfrom15.8to35.7,indicatingpoorreliability.Only60%of
theventricularseptaldefectswereassociatedwithacardiacmurmur,andfollowupshowedthat75%hadclosedspontaneouslybytheageof12months,withthe
majorityclosingbefore6months.Thishasimplicationsforstrategiesof
ascertainmentthatrelyonclinicalpresentationforidentification.Ifdiagnostic
confirmationisdelayed,manydefectswillhaveclosedspontaneouslyand
escapeverification.
Becausemanystudiesexaminedtheprevalenceofcongenitalheartdisease
fromdifferentpopulationsanderas,systematicreviewsmightyieldimportant

insightsandperhapsmorereliableandgeneralizableestimates.Twoteamsof
reviewershaveexaminedmultiplestudiesofprevalencewithanaimof
comparing,andpotentiallypooling,theestimates.Theinvestigatorsfromthe
Baltimore-WashingtonInfantStudycomparedtheprevalenceatlivebirthnoted
fromtheirinitial2yearsoftheirstudy,1981and1982,tothatofeight


previouslypublishedandwidelycitedreports.33Theprevalencefromthe
Baltimore-WashingtonInfantStudyforthatperiodwas3.7casesforeach1000
livebirths,withallcasesbeingconfirmedbyautopsy,surgery,cardiac
catheterization,orechocardiography.TheNewEnglandRegionalInfantCardiac
Programverifiedcasesbyautopsy,surgery,orcardiaccatheterization,witha
prevalenceof2.03.34Whencasesverifiedonlybyechocardiographywere
excludedfromtheestimateofprevalencemadefromtheBaltimore-Washington
InfantStudy,whichloweredtheestimateto2.38,theresultsofthetwostudies
weremoresimilar.Theremainingsevenstudiesincludedcasesforwhicha
clinicaldiagnosiswithoutverificationhadbeenmade,varyingfromthree-tenths
tohalfofincludedcases,withestimatesrangingfrom5.51to8.56.26,35–41If
casesthatwerediagnosedbyclinicalmeansonlywereexcludedfromthese
estimates,theprevalencethenrangedfrom3.75to4.30,whichiscomparableto
theestimatereportedbytheinvestigatorsconductingtheBaltimore-Washington
InfantStudy.
Amorerecentreviewprovidedananalysisofalargernumberofstudies,with
anaimatdeterminingthesourcesofvariabilityinestimatesofprevalencenoted
betweenstudies.42Thisanalysiswasbasedonreviewof62studiesreportedafter
1955.Widediscrepancieswerefoundinreportedestimatesofprevalence,
rangingfrom4to50per1000livebirths(Fig.13.2).Estimateswerestableover
time,atbetweenapproximately4and8per1000livebirths,until1985.The
adventofthewidespreaduseofechocardiographyatthispointresultedin
estimatesbecomingmorediverse,andthereportedestimatesincreasedand

begantoexceed10per1000livebirths(Fig.13.3).Theincreasesinthe
prevalenceestimateslikelyreflectedanoverallincreaseinascertainedcases,but
thiswasmoremarkedfortrivialormildlesions,particularlyventricularseptal
defects,withtheestimatedprevalenceoflesionsproducingcyanoticdisease
remainingrelativelystable(Fig.13.4).Themedianprevalenceofcongenital
heartdiseasewas7.7,withaninterquartilerangefrom6.0to10.6,excluding
nonstenoticaorticvalveswithtwoleaflets,silentarterialducts,andisolated
partiallyanomalouspulmonaryvenousconnections.Themedianprevalenceof
bifoliateaorticvalveswas9.2,withaninterquartilerangefrom5.3to13.8.
Inclusionofcaseswithsuchbifoliateaorticvalveswouldgreatlyinflate
estimatesofprevalence.Themedianprevalenceofcongenitallymalformed
heartsproducingcyanosiswas1.08,withaninterquartilerangeof1.27to1.53.
InTable13.9,theprevalenceofspecificlesionsper1millionlivebirthsis
shown.


FIG.13.2 Histogramoftheprevalenceofcongenitalcardiacdefectsper
1000livebirthasnotedin62reports.(FromHoffmanJI,KaplanS.The
incidenceofcongenitalheartdisease.JAmCollCardiol.2002;39:1890–
1900.)

FIG.13.3 Changesinthereportedprevalenceofcongenitalheartdisease
frommultiplereportsovertime,andtheinfluenceoftheproportionof
ascertainedcasesthatwereventricularseptaldefects.Eachcircle
representsthevaluederivedfromeachreport.Thehorizontallinesare
drawnarbitrarilyataprevalenceof10per1000livebirthsand40%ofall
congenitalheartdisease,andtheverticallinesaredrawnarbitrarilyat1985
and40%,respectively.Anincreasingproportionofthehighprevalence
estimatesarebeyond1985andforseriesthathavemorethan40%
ventricularseptaldefects.CHD,Congenitalheartdisease;VSD,ventricular

septaldefect.(FromHoffmanJI,KaplanS.Theincidenceofcongenital