crosswww.510dd.com

RESEARCHOpenAccessInsufficientevidencefortheuseofaphysicalexaminationtodetectmaltreatmentinchildrenwithoutpriorsuspicion:asystematicreviewEvaMMHoytemavanKonijnenburg1,3*,ArianneHTeeuw1,TessaSieswerda-Hoogendoorn1,ArnoldGELeenders2andJohannaHvanderLee1AbstractBackground:Althoughitisoftenperformedinclinicalpractice,thediagnosticvalueofascreeningphysicalexaminationtodetectmaltreatmentinchildrenwithoutpriorsuspicionhasnotbeenreviewed.
Thisarticleaimstoevaluatethediagnosticvalueofacompletephysicalexaminationasascreeninginstrumenttodetectmaltreatmentinchildrenwithoutpriorsuspicion.
Methods:WesystematicallysearchedthedatabasesofMEDLINE,EMBASE,PsychINFO,CINAHL,andERIC,usingasensitivesearchstrategy.
Studiesthati)presentedmedicalfindingsofacompletephysicalexaminationforscreeningpurposesinchildren0–18years,ii)specificallyrecordedthepresenceorabsenceofsignsofchildmaltreatment,andiii)recordedchildmaltreatmentconfirmedbyareferencestandard,wereincluded.
Tworeviewersindependentlyperformedstudyselection,dataextraction,andqualityappraisalusingtheQUADAS-2tool.
Results:Thesearchyielded4,499titles,ofwhichthreestudiesmettheeligibilitycriteria.
Theprevalenceofconfirmedsignsofmaltreatmentduringscreeningphysicalexaminationvariedbetween0.
8%and13.
5%.
Thedesignsofthestudieswereinadequatetoassessthediagnosticaccuracyofascreeningphysicalexaminationforchildmaltreatment.
Conclusions:Becauseofthelackofinformativestudies,wecouldnotdrawconclusionsaboutthediagnosticvalueofascreeningphysicalexaminationinchildrenwithoutpriorsuspicionofchildmaltreatment.
Keywords:Childabuse,Diagnosis,PhysicalexaminationBackgroundChildmaltreatmentisaworldwideproblemwithmanyadverseconsequences,bothintheshortandlongterm[1-5].
Earlydetectionofchildmaltreatmentisextremelyimportantinordertointerveneandimprovethesitu-ation,andtopreventrecurrence,severemorbidity,orevendeath[6-9].
Thelargediscrepancybetweenthemuchhigherprevalenceofself-reportedmaltreatmentcomparedtotheprevalenceofmaltreatmentofwhichprofessionalsareaware,evenwhenusingidenticalcriteria,meansthatasubstantialamountremainsundetected[1,8-10].
Thecontributionofhospitalstothetotalnumberofchildmaltreatmentreportsisrelativelysmall.
Severalstudieshaveshownthatchildmaltreatmentisunder-detectedbyhospitalstaff[9,11,12].
Toimprovethede-tectionofchildmaltreatmentinhospitals,anumberofstrategies,suchaschecklistsandtrainingofpersonnel,havebeendeveloped[13,14].
Anotherstrategythatiswidelyusedinemergencydepartmentsandotherhealthcaresettingstodetectchildmaltreatmentistoperformascreeningphysicalexamination.
Thephysicalexaminationistargetedtowardsexposingsignsofchildmaltreatment,andissometimescalled'top-to-toe'inspection.
IntheNetherlands,41%ofDutchemergencydepartmentsusea*Correspondence:eva.
hoytemavankonijnenburg@amc.
nlEqualcontributors1DepartmentofPaediatrics,EmmaChildren'sHospital/AcademicMedicalCentre,UniversityofAmsterdam,Postbus22660,Amsterdam,1100DD,TheNetherlands3AcademicMedicalCentreAmsterdam,DepartmentofPaediatrics,UniversityofAmsterdam,RoomK01-138,Meibergdreef9,Amsterdam1105AZ,TheNetherlandsFulllistofauthorinformationisavailableattheendofthearticle2013HoytemavanKonijnenburgetal.
;licenseeBioMedCentralLtd.
ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.
org/licenses/by/2.
0),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
HoytemavanKonijnenburgetal.
SystematicReviews2013,2:109http://www.
systematicreviewsjournal.
com/content/2/1/109physicalexaminationasascreeningtool,mainlyinyoun-gerchildren[15].
Theexaminationcanalsobeusedaspartofabroaderscreeningtool,forexample,aspartofachecklist[13,16-19].
Inthesesettings,thephysicalexamin-ationisusedasascreeningtoolinallchildren(withoutpriorsuspicionofmaltreatment),andthusperformedregardlessofthecomplaintsofthechild.
Ascreeningphysicalexaminationisrelativelyeasy,inexpensive,andinprinciplewithoutadverseeffects.
Duringtheexamination,thechildisundressedcompletelyandspecificallyinspectedforanysignsofphysicalabuseandphysicalneglect(e.
g.
,scars,bruises,caries,unkemptappearance).
Furthermore,abnormalphysicalandemo-tionaldevelopment,behaviourandparent–childinter-actioncanbeobserved.
Alloftheabovecouldleadtosuspicionsofchildmaltreatment.
Dependingontheageofthechild,thephysicalexaminationislikelytoshowdiffer-entfindingsaccordingtothechild'sphysicaldevelopmentandthemechanismofabuse(forexample,abusiveheadtraumaisusuallyseeninveryyoungchildren,presentingwithspecificfeatures)[20].
Thephysicalexaminationmightbemostrelevantinyoung,non-verbalchildren,whoareunabletotalkaboutmaltreatment.
Possibleun-desirableeffectsofascreeningphysicalexaminationmightoccurifanegativescreeningresultisfalselyreassuringforprofessionalsoriftheresultisafalsepositive.
Inaddition,itcouldbethatmaltreatingparentsarediscouragedfromvisitingahealthcaresettingiftheyknowthattheirchil-drenwillbephysicallyexaminedforpossiblemaltreat-ment.
Ascreeningphysicalexaminationwouldmostlyidentifyphysicalabuseandneglect,andcanneveridentifyallformsofchildmaltreatment.
Therefore,itisgenerallyusedincombinationwithotherscreeningstrategiesinordertoincreasethesensitivityofchildmaltreatmentdetection[21].
Toourknowledge,althoughmanychildmaltreatmentprotocolsinvarioushealthcaresettingsincludeascreeningphysicalexaminationandcliniciansrelyontheresults,thediagnosticvalueofascreeningphysicalexaminationtodetectmaltreatmentinchildrenwithoutpriorsuspicionhasnotyetbeenreviewed.
Twosystem-aticreviewsinvestigatedtheperformanceofvariousscreeningmethodsformaltreatmentinchildrenpresent-ingatemergencydepartments[14,22].
Ofall17studiesincludedinbothreviews,onlyonestudyinvestigatedacompletephysicalexaminationaspartofascreeningmethod,incombinationwithachecklistanddiscussionwithaphysician[23].
However,thediagnosticvalueofthisphysicalexaminationisunclearsinceresultswerenotreportedseparatelyfromtheotheraspectsofthescreeningmethod.
Evidencesuggestiveofabusewasfoundin10%[24]and63%[25]ofchildrenwhowerephysicallyexaminedbecauseof(suspected)maltreat-ment.
However,thephysicalexaminationprobablyyieldsdifferentresultswhenusedasascreeningmethodtodetectchildmaltreatmentinchildrenwithoutpriorsus-picion.
Althoughascreeningphysicalexaminationisoften(butnotalways)performedincombinationwithotherscreeningtools,itisimportanttoalsoassessitsaddeddiagnosticvalue.
Inpractice,cliniciansusethere-sultsofthescreeningphysicalexaminationtomakeariskassessment,andshouldthereforeknowitsdiagnos-ticvalue.
Ifphysiciansareunawareofthis,theymightover-orunder-detectchildmaltreatment,whichcanhaveseriousadverseconsequences.
Wethereforeperformedasystematicreviewtoevalu-atethediagnosticvalueofacompletephysicalexamin-ation,minimallyconsistingofavisualinspectionoftheentireskinandoralcavity,asascreeninginstrumentformaltreatmentinchildrenwithoutpriorsuspicioninvari-oushealthcaresettingscomparedtoa'compositerefer-encestandard'(acombinationofreferencestandards,consideredtobepositiveifatleastoneofthecomponentsispositive).
Unfortunately,nogoldstandardisavailableforchildmaltreatment;therefore,todeterminediagnostictestaccuracies,derivedstandardshavetobeusedasareference,i.
e.
,adiagnosisofmaltreatmentbyeitheri)acourt,ii)theChildProtectiveServices(CPS),iii)anexpertpanel,iv)aforensicphysician,orv)self-report.
MethodsSearchmethodsWesystematicallysearchedtheelectronicdatabasesofMEDLINE(throughPubMedandthroughOvid,from1947toAugust8,2013),EMBASE(1980toAugust8,2013),PsychINFO(1806toAugust8,2013),CINAHL(1982toAugust8,2013)andERIC(1965toAugust8,2013).
Themainsearchstrategyconsistedofthreecom-ponentscombinedby'AND',namely'physicalexamin-ation','child',and'abuse'.
Synonymsforthesetermswerecombinedwiththecorrespondingcomponentwith'OR'.
Furthermore,database-specificMeSHandthesaurustermsandtextwordswereadded.
Becauseweexpectedasmallnumberofeligiblearticles,weusedasensitivesearchstrat-egy.
SeeAdditionalfile1forfullsearchstrategies.
Thelistsofcitedandcitingreferencesofincludedarti-clesandofarticlesthatwereconsideredforinclusionatanearlystagewerehandsearchedforadditionalrelevantarticles.
Furthermore,13keyauthorswereapproachedandaskediftheycouldrecommendanyrelevantstudiesinthisarea.
Finally,articlesknownbyanyoftheauthorswereaddedtothesearchresults.
Wesearchedforbothpublishedandunpublishedreports.
Therewasnolan-guagerestriction.
StudyselectionTheoptimalstudydesigntoanswerourresearchques-tionwouldbeacross-sectionaldiagnosticaccuracyHoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page2of8http://www.
systematicreviewsjournal.
com/content/2/1/109study,comparingaphysicalexaminationtoauniversalreferencestandardinallchildren.
However,becausewedidnotexpecttofindmanystudies,webroadenedoureligibilitycriteriatothefollowing:anyempiricalstudyincludingchildren0–18yearsvisitinganyhealthcaresetting,evaluatingacompletephysicalexamination(de-finedasminimallyconsistingofavisualinspectionoftheentireskinandoralcavity)specificallyperformedas(partof)ascreeningprocedureforchildmaltreatmentbyahealthcareprofessionalandprovidingsystematicdocumentationofthepresenceorabsenceofsignsindi-catingchildmaltreatmentincomparisontooneofthefollowingreferencestandards:i)acourt,ii)theCPS,iii)anexpertpanel,iv)aforensicphysician,orv)aself-report.
Sincetheaimofthisreviewwastoevaluatethephysicalexaminationasadiagnosticinstrumenttode-tectmaltreatmentinchildrenwithoutpriorsuspicion,wedidnotincludestudiesinchildrenwhowerephysic-allyexaminedbecauseof(orasuspicionof)childmal-treatment.
Furthermore,becausethisreviewaimstoevaluatethecomplete'top-to-toe'examination,studieswereexcludedifthephysicalexaminationwasonlyper-formedwiththeintentiontodetectsexualchildabuse.
Tworeviewers(EH,AT)selectedthestudiesinde-pendently,firstbasedontitles,thenonabstractsandkeywords,andfinallyonfulltexts.
Disagreementswerediscusseduntilconsensuswasreached.
DataextractionandassessmentAppraisalofthemethodologicalqualityanddataextrac-tionwereperformedbybothreviewers(EH,AT)inde-pendently.
Datafromincludedstudieswereextractedandthemethodologicalqualitywasassessedwithacom-binedformincludingdataextractionitemsandtheitemsoftheQUADAS-2toolforqualityassessment[26].
Thecombinedformwaspilotedindependentlyandadjustedbytworeviewersuntiltherewasconsensusonafinalversion.
Disagreementsindataextractionorqualityas-sessmentwerediscusseduntilconsensuswasreached.
Wherenecessary,athirdreviewer(TS)wasthefinaljudge.
Extracteddataincluded:i)characteristicsofthestudy(design,yearofpublication,typeofpublication,studycountry,fundingsource);ii)characteristicsofthestudypopulation(includingage,sexdistribution,previ-ousdiagnosisofchildmaltreatment);iii)characteristicsofthescreeningphysicalexamination(includingsetting);iv)characteristicsofthereferencestandard;v)character-isticsoftheoutcomemeasure(childmaltreatmentdiag-nosis,typeofmaltreatmentfindings);vi)anyreportedharmcausedbyascreeningphysicalexamination;vii)sensitivity(truepositives,proportionofmaltreatedchildrenwithapositivephysicalexamination),andviii)specificity(truenegatives,proportionofnon-maltreatedchildrenwithanegativephysicalexamination).
ResultsTheelectronicliteraturesearchprovided4,215titlesaftertheremovalofduplicates.
Furthermore,284stud-ies,retrievedbycitationsearch,personalknowledgeorcommunicationwithkeyauthors,wereaddedaftertheremovalofduplicates.
Ofallthese,762studieswerese-lectedbasedontitleandsubsequently147studieswereselectedbasedonabstractandkeywordsofwhichfulltextstudieswereread.
Applicationofinclusionandexclu-sioncriterialedtotheinclusionofthreestudies[27-29].
Figure1presentsthestudyselectionprocesswithreasonsforexclusioninaPRISMAflowdiagram[30].
ThecharacteristicsofthethreeincludedstudiesarepresentedinTable1.
Noneoftheincludedstudiesweredesignedtoevaluatethediagnosticaccuracyofthephys-icalexaminationtoidentifychildmaltreatment.
Twostudiesweredesignedtodeterminetheprevalenceandassociatedriskfactorsforchildmaltreatment,onestudyinacommunitysample[27]andonestudyinchildrenvisitingtheemergencydepartment[28].
Thethirdstudywasdesignedtoevaluateascreeningtoolforchildmal-treatment,includingacompletephysicalexamination,attheemergencydepartment[29].
ThereferencestandardswereachildmaltreatmentdiagnosisbytheCPS[29],anexpertpanel[28],andself-reportbythechild[27].
QualityofthestudiesThequalityofthethreestudiesthatwereincluded,ac-cordingtotheQUADAS-2tool[26],ispresentedinTable2.
Noneofthestudiescontainedaflowdiagram,thus,aflowdiagramwashand-drawnandreviewedforeachstudy.
Thefirststudy,thatofAfifietal.
,isacross-sectionalstudywithasamplesizeof555subjects[27].
Thestudyaimedtoidentifytheprevalenceandunderlyingriskfac-torsofchildmaltreatmentinschool-agedchildreninruralEgypt.
Thisistheonlystudyperformedinacom-munitysample.
Thestrengthsofthestudyareitslargesamplesize,therandomselectionofparticipantsandtheparticipationofalleligiblesubjects.
Animportantlimitationisthatsubjectswereconsiderednon-abusediftherewerenosignsofphysicalabuseonexamination,evenwhenabusewasself-reported.
Ontheotherhand,ifsubjectsdeniedabuse,theywereconsiderednon-abusedevenwhenpositivesignsofabusewerepresentduringexamination[31,32].
Thisleadtoreferencestand-ardrelatedbias,andprobablytofalsenegativetestresults.
Twostudieswereperformedinanemergencydepart-ment[28,29].
ThestudybyPalazzietal.
isacross-sectionalstudywithasamplesizeof10,175subjects[28].
ThestudyaimedtoidentifytheprevalenceandassociatedriskfactorsofsuspectedchildmaltreatmentinpaediatricemergencydepartmentsinItaly.
ThestrengthsofthestudyaretheverylargesamplesizeandHoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page3of8http://www.
systematicreviewsjournal.
com/content/2/1/109themulticentredesign.
Animportantlimitationisthattheresultsofdifferentpartsofthephysicalexaminationarepresentedseparately(skinlesions,orallesions,etc.
)andthecumulativeprevalenceofanypositivesignofchildmaltreatmentisunknown,butmostlikelyhigherthanthereportedprevalenceofskinlesionsonly(report-ingbias).
Furthermore,theresultsofthephysicalexam-inationwereusedtoestablishthereferencestandard(incorporationbias).
Thethirdstudy,thatbyRosenbergetal.
,isaprospect-ivestudywithafollow-upofoneyear,including476subjects[29].
Thestudyaimedtoprospectivelyevaluateabriefscreeningassessmentforchildmaltreatmentatanemergencydepartment.
Thestrengthsofthestudyarethelargesamplesize,therandomenrolmentofchil-dren(althoughtherandomizationprocessisnotde-scribed),andtheindependenceofthereferencestandardfortheresultsofthephysicalexamination.
Limitationsarethatthereferencestandardwasappliedat1-yearfollow-upandwasconsideredpositiveifmaltreatmenthadeveroccurred.
Therefore,itispossiblethatmaltreat-mentwasconfirmed,evenifthishappenedaftertheemergencydepartmentvisit.
Otherimportantlimita-tions,probablyleadingtounderestimationoftheprevalenceofsignsofchildmaltreatmentduringphys-icalexamination,arethatchildrenwereexcludedfromthestudyiftherewasasuspicionofchildmaltreatmentbeforeorduringthevisittotheemergencydepartment,andthatthenumberofchildrenwithanypositivesignofchildmaltreatmentisnotreported.
Finally,thisstudywaspublishedin1982,atwhichtimesomeviewsonchildmaltreatmentweredifferentfromtoday(suchasconsideringabaldocciputasignofmaltreatment),pos-siblyleadingtoinformationbias.
ResultsofthephysicalexaminationTable3showstheprevalenceofanysignsofchildmaltreat-mentfounduponscreeningphysicalexamination(uncon-firmed)andsignsofmaltreatmentfoundinchildrenwhowereindeedmaltreatedasconfirmedbyareferencestand-ard(confirmed)asreportedinthethreeincludedstudies.
Theprevalenceofunconfirmedsignsofmaltreatmentrangedbetween7.
8%and14.
6%ofthechildrenexamined.
Theprevalenceofsignsofchildmaltreatmentconfirmedbyareferencestandardrangedbetween0.
8%and13.
5%.
Duetothestudydesigns,itwasimpossibletousesensitiv-ityandspecificityofthestudiestodeterminethediagnosticaccuracyofthescreeningphysicalexaminationtodetectFigure1PRISMAflowdiagram.
HoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page4of8http://www.
systematicreviewsjournal.
com/content/2/1/109Table1CharacteristicsofincludedstudiesFirstauthorYearofpublicationCountryTypeofstudySettingStudyaimSamplesizeAgeSexdistribution(%)male/femaleIndextestReferencestandardAfifi[27]2003EgyptCross-sectionalPreparatoryandsecondaryschoolstudentsfromaruralcommunity,selectedbyrandomclustersamplingToidentifytheprevalenceandunderlyingriskfactorsofchildmaltreatment55512–18years;meanage15.
6±1.
5years63/37Generalphysicalexaminationbyphysician,specificallyincludingsignsofpreviousorrecentphysicalabuseSelf-reportofthechildincombinationwithpositivesignsuponphysicalexaminationPalazzi[28]2005ItalyCross-sectionalAllchildren0–14yearspresentingin19emergencydepartmentsToidentifytheprevalenceandassociatedriskfactorsofsuspectedchildmaltreatment10,1750–14years;meanage4.
8±3.
9years57/43Completephysicalexaminationwheneverpossible,especiallyinyoungerchildrenSix-pointsuspicionindexforchildmaltreatmentattributedbyanexpertpanelofalocalchildhealthteamincollaborationwithresearchassistants,basedonroutineassessmentsRosenberg[29]1982USAProspective,1-yearfollow-upArandomlyenrolledsampleofchildren0–2yearsvisitinganemergencydepartmentToprospectivelyevaluateabriefscreeningassessmentforchildmaltreatment4760–2years55/45Caregiverundresseschild,assessmentbynurseforbeingunkempt,havingabaldocciput,andthepresenceofphysicalbruises,burnsorbitesRegisteredasmaltreatedattheCPSi(theDepartmentofSocialServices)at1-yearfollow-upiCPS=ChildProtectiveServices.
HoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page5of8http://www.
systematicreviewsjournal.
com/content/2/1/109childmaltreatment.
SeeAdditionalfile2forthe2x2con-tingencytablesoftheresultsoftheincludedstudies.
DiscussionMainfindingsThisreviewdidnotestablishthediagnosticaccuracyofacompletephysicalexaminationasascreeninginstru-mentformaltreatmentinchildrenwithoutpriorsuspi-cioninhealthcaresettings.
Nostudiesprovidinganadequateestimationofsensitivityandspecificityofthescreeningphysicalexaminationforchildmaltreatmentcouldbeidentified.
Threestudieswereincluded[27-29].
Inthesestudies,theprevalenceofconfirmedsignsofmaltreatmentuponascreeningphysicalexaminationrangedbetween0.
8%and13.
5%.
Theriskofbiasofthereferencestandardwasconsideredhighforallthreestudies.
Thereferencestandardwasnotindependentoftheresultsofthescreeningphysicalexaminationintwoofthethreereviewedstudies[27,28].
Intwostudies,re-sultsofvariousaspectsofthephysicalexaminationwereTable2QualityassessmentofreviewedstudieswithQUADAS-2tool[26]PatientselectionIndextestReferencestandardFlowandtimingReviewercommentsRiskofbiasConcernsaboutapplicabilityRiskofbiasConcernsaboutapplicabilityRiskofbiasConcernsaboutapplicabilityRiskofbiasAffifi(2003)[27]LowLowLowLowHighHighLowSensitivityandspecificitycannotbecalculatedduetoreferencestandardrelatedbias(referencestandardisincorrectduetotheuseofself-reportincombinationwithsignsuponphysicalexamination,whichislikelytounderestimatetrueprevalence)Palazzi(2005)[28]HighLowLowUnclearHighLowHighSensitivityandspecificitycannotbecalculated,duetoi)reportingbias(cumulativeprevalenceofatleast1positivefindinguponphysicalexaminationnotbeingreported)andii)incorporationbias(resultsofphysicalexaminationareusedinestablishingthereferencestandard)Rosenberg(1982)[29]LowLowLowUnclearHighLowHighSensitivityandspecificitycannotbecalculated,duetoi)differenttimingofapplicationofthereferencestandard,ii)reportingbias(cumulativeprevalenceofatleast1positivefindinguponphysicalexaminationnotbeingreported)andiii)informationbias(duetoadifferentdefinitionofphysicalsignsofmaltreatmentusedatthetime)Table3SummaryofresultsofincludedstudiesAuthor(year)Childrenwithunconfirmedsignsofmaltreatmentuponphysicalexamination/childrenexaminedChildrenwithsignsofmaltreatmentuponphysicalexaminationconfirmedbyreferencestandard/childrenexaminedAfifi(2003)[27]81/555(14.
6%)75/555(13.
5%)(burns30,bruises20,scars19,scratches10,bitemarks2)Palazzi(2005)[28]Skinlesions:1,177/9,510(12.
4%)Skinlesions:75/9,510(0.
8%)Orallesions:123/9,137(1.
3%)Orallesions:8/9,137(0.
09%)Presentorpastburns,fracturesandheadtraumaarepresentedseparatelyintheoriginalarticle.
However,itisunclearwhetherthisisassessedduringphysicalexaminationand,therefore,theseresultsarenotincludedinthisreview.
ThenumberofchildrenwithatleastonefindinguponphysicalexaminationisunknownRosenberg(1982)[29]Unkempt:37/473(7.
8%)Unkempt:7/473(1.
5%)Bruises,burns,humanbites:18/473(3.
8%)Bruises,burns,humanbites:5/473(1.
1%)Baldocciput*:14/474(3%)Baldocciput*:0/474(0%)Thenumberofchildrenwithatleastonefindinguponphysicalexaminationisunknown*Thisisnolongerconsideredasignofchildabuse.
HoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page6of8http://www.
systematicreviewsjournal.
com/content/2/1/109presentedseparately,andcumulativenumbersofchil-drenwithatleastonepositivefindinguponphysicalexaminationwereunclear[28,29].
StrengthsandlimitationsofthisreviewThestrengthsofthisreviewarethesystematicapproachandtheextensiveliteraturesearch.
ThreestudieswereincludedandsystematicallyassessedformethodologicalqualityusingthethoroughlydevelopedQUADAS-2tool.
Noneofthesestudiesinvestigatedanypotentialharmcausedbyaphysicalexamination.
Itwasnotpossibletodetermineinwhichsubgroups(forexampleagegroups)ascreeningphysicalexaminationwasmoreorlessac-curate.
Unfortunately,becausethesearchdidnotiden-tifystudiesthatprovidedsensitivityandspecificity,wecouldnotdrawconclusionsaboutthediagnosticvalueofascreeningphysicalexaminationinchildrenwithoutpriorsuspicionofchildmaltreatment.
RecommendationsAlthoughitiswidelyusedinclinicalpractice,thereisinsufficientevidenceforaphysicalexaminationasascreeninginstrumenttoimprovedetectionofmaltreat-mentinchildrenwithoutpriorsuspicion.
Currently,whenusingascreeningphysicalexaminationtodetectchildmaltreatmentinpractice,cliniciansshouldbeawarethatitsdiagnosticaccuracyisunclearandchildmaltreatmentcanbebothover-andunder-detected.
Anegativeresultdoesnotexcludeabuse,becausei)notallabuseleavesinjuries,ii)evenseriousinjuriescausedbyphysicalabuse(suchasfractures)canbepresentwithoutanysignsuponphysicalexamination,andiii)priorinjur-iesofabusemayalreadyhavedisappeared[16,31,32].
Ontheotherhand,somefindingsuponphysicalexamin-ationcanmimicphysicalabusewhilebeingofanon-abusivenature(forexampleaMongolianspot)[33,34].
Finally,althoughitispossiblethatascreeningphysicalexaminationcouldidentifyemotionalorsexualmaltreat-mentinrarecases(forexamplebecauseofadisclosureduringtheexaminationorbecauseabnormaldevelop-ment,behaviour,orparent–childinteractionsobserved),theexaminationisaimedtowardsthedetectionofphys-icalabuseorneglect,andotherformsofchildmaltreat-mentcouldbeoverlooked.
Todeterminethediagnosticaccuracyofaphysicalexaminationasascreeninginstrumenttodetectmal-treatmentinchildrenwithoutpriorsuspicion,wewouldrecommendastudywithaprotocolizedsystematicphys-icalexaminationandreferencestandardforalarge,un-selectedgroupofchildren,atdifferentlevelsofriskformaltreatment,andindifferentsettings.
Althoughweacknowledgethatitischallengingtofindanoptimalref-erencestandard,itcouldbeathoroughcasereviewbyanexpertpanelincombinationwithchild-,parent-,informant-,and(ifthereisinvolvement)CPS-reports.
Ifparentsandchildrenareinterviewedoraskedtofilloutaquestionnaireinarespectfulway,theresearchwouldnotbeamajorburdenforthem,anditwouldnotbeun-ethicaltoalsoapplythereferencestandardtochildrenwithanegativescreeningresult.
Ideally,todeterminethesensitivity,specificity,andpositivepredictivevalueofthescreeningphysicalexamination,allchildrenshouldundergothephysicalexaminationandtherefer-encestandard,regardlessoftheresultsofthephysicalexamination.
However,inpractice,thismightnotbefeasiblegiventhelargenumberofchildrenthatwouldrequirethe(time-consuming)referencetest.
Tosolvethisissue,thereferencestandardcouldbeperformedinallchildrenwithapositivephysicalexaminationandinarandomsampleofchildrenwithanegativephysicalexamination,asiscurrentlybeingdoneinastudyfocusedontheemergencydepartment[17].
ConclusionsBecauseofthelackofinformativestudies,wecouldnotdrawconclusionsaboutthediagnosticvalueofascreeningphysicalexaminationinchildrenwithoutpriorsuspicionofchildmaltreatment.
AdditionalfilesAdditionalfile1:Fullsearchstrategies.
Additionalfile2:2x2contingencytablesofincludedstudies.
AbbreviationCPS:Childprotectiveservices.
CompetinginterestsTheauthorsdeclarethattheyhavenocompetinginterests.
Authors'contributionsEHandATcarriedoutdataextraction,qualityappraisalanddataanalysis,participatedinstudydesign,anddraftedthemanuscript.
ALdesignedthesearchstrategiesandrevisedthemanuscript.
TSparticipatedinstudydesignandassistedindataextractionandqualityappraisal,andrevisedthemanuscript.
JLparticipatedinstudydesign,advisedonallmethodologicalissues,andhelpedtodraftandrevisethemanuscript.
Allauthorsreadandapprovedthefinalmanuscript.
AcknowledgementsThisstudywasfundedby'StichtingKinderpostzegels',anindependentchildren'scharityintheNetherlandswiththeslogan"Forchildrenbychildren",andbytheCityofAmsterdam.
Authordetails1DepartmentofPaediatrics,EmmaChildren'sHospital/AcademicMedicalCentre,UniversityofAmsterdam,Postbus22660,Amsterdam,1100DD,TheNetherlands.
2MedicalLibrary,AcademicMedicalCentre,UniversityofAmsterdam,Postbus22660,Amsterdam1100DD,TheNetherlands.
3AcademicMedicalCentreAmsterdam,DepartmentofPaediatrics,UniversityofAmsterdam,RoomK01-138,Meibergdreef9,Amsterdam1105AZ,TheNetherlands.
Received:20August2013Accepted:28November2013Published:6December2013HoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page7of8http://www.
systematicreviewsjournal.
com/content/2/1/109References1.
GilbertR,WidomCS,BrowneK,FergussonD,WebbE,JansonS:Burdenandconsequencesofchildmaltreatmentinhigh-incomecountries.
Lancet2009,373:68–81.
2.
FelittiVJ,AndaRF,NordenbergD,WilliamsonDF,SpitzAM,EdwardsV,KossMP,MarksJS:Relationshipofchildhoodabuseandhouseholddysfunctiontomanyoftheleadingcausesofdeathinadults.
TheAdverseChildhoodExperiences(ACE)study.
AmJPrevMed1998,14:245–258.
3.
AndaRF,FelittiVJ,BremnerJD,WalkerJD,WhitfieldC,PerryBD,DubeSR,GilesWH:Theenduringeffectsofabuseandrelatedadverseexperiencesinchildhood.
Aconvergenceofevidencefromneurobiologyandepidemiology.
EurArchPsychiatryClinNeurosci2006,256:174–186.
4.
LupienSJ,McEwenBS,GunnarMR,HeimC:Effectsofstressthroughoutthelifespanonthebrain,behaviourandcognition.
NatRevNeurosci2009,10:434–445.
5.
NormanRE,ByambaaM,DeR,ButchartA,ScottJ:Thelong-termhealthconsequencesofchildphysicalabuse,emotionalabuse,andneglect:asystematicreviewandmeta-analysis.
PLoSMed2012,9(11):e1001349.
6.
EthierLS,LemelinJ-P,LacharitéC:Alongitudinalstudyoftheeffectsofchronicmaltreatmentonchildren'sbehavioralandemotionalproblems.
ChildAbuseNegl2004,28:1265–1278.
7.
SeifertD,KrohnJ,LarsonM,LambeA,PüschelK,KurthH:Violenceagainstchildren:furtherevidencesuggestingarelationshipbetweenburns,scalds,andtheadditionalinjuries.
IntJLegalMed2010,124:49–54.
8.
GilbertR,KempA,ThoburnJ,SidebothamP,RadfordL,GlaserD,MacmillanHL:Recognisingandrespondingtochildmaltreatment.
Lancet2009,373:167–180.
9.
KingWK,KieselEL,SimonHK:Childabusefatalities:arewemissingopportunitiesforinterventionPediatrEmergCare2006,22:211–214.
10.
AlinkL,vanIjzendoornR,Bakermans-KranenburgM,PannebakkerF,VogelsT,EuserS:DeTweedeNationalePrevalentiestudieMishandelingvanKinderenenJeugdigen(NPM-2010).
Report.
LeidenAttachmentResearchProgramandTNO.
Leiden:CasimirPublishers;2011.
11.
RavichandiranN,SchuhS,BejukM,Al-HarthyN,ShouldiceM,AuH,BoutisK:Delayedidentificationofpediatricabuse-relatedfractures.
Pediatrics2010,125:60–66.
12.
OralR,BlumKL,JohnsonC:Fracturesinyoungchildren:arephysiciansintheemergencydepartmentandorthopedicclinicsadequatelyscreeningforpossibleabusePediatrEmergCare2003,19:148–153.
13.
TeeuwAH,DerkxBHF,KosterWA,vanRijnRR:Educationalpaper:Detectionofchildabuseandneglectattheemergencyroom.
EurJPediatr2012,171:877–885.
14.
WoodmanJ,LeckyF,HodesD,PittM,TaylorB,GilbertR:Screeninginjuredchildrenforphysicalabuseorneglectinemergencydepartments:asystematicreview.
ChildCareHealthDev2010,36:153–164.
15.
HoytemavanKonijnenburgEMMH,TeeuwAH,ZwaardSA,vanderLeeJH,vanRijnRR:Screeningmethodstodetectchildmaltreatment:highvariabilityinDutchemergencydepartments.
EmergMedJ2013[Aheadofprint].
16.
LindbergDM,ShapiroRA,LaskeyAL,PallinDJ,BloodEA,BergerRP:ExSTRAInvestigators:Prevalenceofabusiveinjuriesinsiblingsandhouseholdcontactsofphysicallyabusedchildren.
Pediatrics2012,130:193–201.
17.
SittigJS,UiterwaalCSPM,MoonsKGM,NieuwenhuisEES,vandePutteEM:Childabuseinventoryatemergencyrooms:CHAIN-ERrationaleanddesign.
BMCPediatr2011,11:91.
18.
LouwersECFM,KorfageIJ,AffourtitMJ,ScheeweDJH,vandeMerweMH,Vooijs-MoulaertA-FSR,vandenElzenAPM,JongejanMHTM,RuigeM,ManaBHAN,LoomanCWN,BosschaartAN,TeeuwAH,MollHA,deKoningHJ:Effectsofsystematicscreeninganddetectionofchildabuseinemergencydepartments.
Pediatrics2012,130:457–464.
19.
BengerJR,PearceV:Simpleinterventiontoimprovedetectionofchildabuseinemergencydepartments.
BMJ2002,324:780.
20.
MaguireSA,KempAM,LumbRC,FarewellDM:Estimatingtheprobabilityofabusiveheadtrauma:apooledanalysis.
Pediatrics2011,128:550–564.
21.
BossuytPM:Comparativeaccuracy:assessingnewtestsagainstexistingdiagnosticpathways.
BMJ2006,332:1089–1092.
22.
LouwersECFM,AffourtitMJ,MollHA,deKoningHJ,KorfageIJ:Screeningforchildabuseatemergencydepartments:asystematicreview.
ArchDisChild2010,95:214–218.
23.
PlessIB,SibaldAD,SmithMA,RussellMD:Areappraisalofthefrequencyofchildabuseseeninpediatricemergencyrooms.
ChildAbuseNegl1987,11:193–200.
24.
KirkCB,Lucas-HeraldA,MokJ:Childprotectionmedicalassessments:whydowedothemArchDisChild2010,95:336–340.
25.
RayS,CostolliV,TanM:Child-protectionmedicalassessments:theneedforauniformservicemodel.
ArchDisChild2010,95:1070.
26.
WhitingPF,RutjesAWS,WestwoodME,MallettS,DeeksJJ,ReitsmaJB,LeeflangMMG,SterneJAC,BossuytPMM:QUADAS-2Group:QUADAS-2:arevisedtoolforthequalityassessmentofdiagnosticaccuracystudies.
AnnInternMed2011,155:529–536.
27.
AfifiZEM,El-LawindiMI,AhmedSA,BasilyWW:AdolescentabuseinacommunitysampleinBeniSuef,Egypt:prevalenceandriskfactors.
EastMediterrHealthJ2003,9:1003–1018.
28.
PalazziS,deGirolamoG,LiveraniT:IChilMa(ItalianChildMaltreatmentstudygroup):observationalstudyofsuspectedmaltreatmentinItalianpaediatricemergencydepartments.
ArchDisChild2005,90:406–410.
29.
RosenbergNM,MeyersS,ShackletonN:Predictionofchildabuseinanambulatorysetting.
Pediatrics1982,70:879–882.
30.
MoherD,LiberatiA,TetzlaffJ,AltmanDG:PRISMAGroup:preferredreportingitemsforsystematicreviewsandmeta-analyses:thePRISMAstatement.
BMJ2009,339:b2535.
31.
ValvanoTJ,BinnsHJ,FlahertyEG,LeonhardtDE:DoesbruisinghelpdeterminewhichfracturesarecausedbyabuseChildMaltreat2009,14:376–381.
32.
RubinDM,ChristianCW,BilaniukLT,ZazycznyKA,DurbinDR:Occultheadinjuryinhigh-riskabusedchildren.
Pediatrics2003,111:1382–1386.
33.
OranjeA,BiloRA:Skinsignsinchildabuseanddifferentialdiagnosis.
MinervaPediatr2011,63:319–325.
34.
LabbeJ,CaouetteG:Recentskininjuriesinnormalchildren.
Pediatrics2001,108:271–276.
doi:10.
1186/2046-4053-2-109Citethisarticleas:HoytemavanKonijnenburgetal.
:Insufficientevidencefortheuseofaphysicalexaminationtodetectmaltreatmentinchildrenwithoutpriorsuspicion:asystematicreview.
SystematicReviews20132:109.
SubmityournextmanuscripttoBioMedCentralandtakefulladvantageof:ConvenientonlinesubmissionThoroughpeerreviewNospaceconstraintsorcolorgurechargesImmediatepublicationonacceptanceInclusioninPubMed,CAS,ScopusandGoogleScholarResearchwhichisfreelyavailableforredistributionSubmityourmanuscriptatwww.
biomedcentral.
com/submitHoytemavanKonijnenburgetal.
SystematicReviews2013,2:109Page8of8http://www.
systematicreviewsjournal.
com/content/2/1/109