• Nie Znaleziono Wyników

Methods of Evidence-Based Anatomy : a guide to conducting systematic reviews and meta-analysis of anatomical studies

N/A
N/A
Protected

Academic year: 2022

Share "Methods of Evidence-Based Anatomy : a guide to conducting systematic reviews and meta-analysis of anatomical studies"

Copied!
6
0
0

Pełen tekst

(1)

ContentslistsavailableatScienceDirect

Annals of Anatomy

j o ur na l h o me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / a a n a t

METHODS

Methods of Evidence-Based Anatomy: a guide to conducting systematic reviews and meta-analysis of anatomical studies

Brandon Michael Henry

a,b,∗,1

, Krzysztof A. Tomaszewski

a,b,1

, Jerzy A. Walocha

a,b

aInternationalEvidence-BasedAnatomyWorkingGroup,12KopernikaStreet,31-034Krakow,Poland

bDepartmentofAnatomy,JagiellonianUniversityMedicalCollege,12KopernikaStreet,31-034Krakow,Poland

a r t i c l e i n f o

Articlehistory:

Received13September2015

Receivedinrevisedform6December2015 Accepted15December2015

Keywords:

Evidence-basedanatomy Methodology

Meta-analysis Systematicreview

a b s t r a c t

Evidence-BasedAnatomy(EBA)istheconceptofapplyingevidence-basedprinciplesandresearchmeth- odstotheanatomicalsciences.Whilenarrativereviewsarecommonintheanatomicalsciences,true systematicreviews(SR)andmeta-analysis(MA)areonlybeginningtogrowinpopularity.Inorderto enhancethequalityoffutureEBAstudies,andensuretheclinicalreliabilityoftheirresults,auniform methodologyisneeded.Inthispaper,wepresentastep-by-stepmethodologicalguideforperforming SRsandMAsofanatomicalstudies.WeaddresstheEBA-specificchallengesineachstepoftheSRand MAprocess,anddiscussmethodsandstrategiestoovercomethesedifficulties.Furthermore,wediscuss indetailthestatisticalmethodsusedinMAofanatomicaldata,includingmulti-categoricalandsingle- categoricalpooledprevalenceestimates,aswellaspooledmeansofonegroup.Lastly,wediscussthe majorlimitationsofEBA,includingthelackofaproperqualityassessmenttoolforanatomicalstudies.

ThemethodsdescribedinthispaperpresentauniformroadmapforfutureEBAstudies.

©2016TheAuthors.PublishedbyElsevierGmbH.ThisisanopenaccessarticleundertheCCBY license(http://creativecommons.org/licenses/by/4.0/).

1. Introduction

While many of the basic medical sciences have seen great advancementsinthepastcentury,thebasicprinciplesandmeth- ods of anatomical research have remained largely unchanged.

Furthermore,asothermedicalscienceshaveprogressedintothe evidence-basedera,thecurrentfoundationofclinicalanatomyis still largelybased onfindings from singleepidemiological-type studies.Assuch,themajorityofcurrentanatomicalliteraturelacks thepowerallowingforconclusivefindingsonthepopulationas awhole.Thismayleaveclinicianspartiallyinthedarkonimpor- tantvariableanatomicalcharacteristics,potentiallyleadingtoan increasedriskofiatrogenicinjuryduringproceduresandmisinter- pretationoftheresultsfromdiagnosticstudies.

Ina2006studyanalyzingsurgicalerrorinmalpracticeclaims, 13%ofinjuriesduetosurgicalerrorswereattributedto“abnormal ordifficult”anatomy(Rogersetal.,2006).Wesuspectthismaybe inlargepartduetopooranatomical knowledgewithrespectto variationsamongpracticingclinicians.Thisknowledge gapmay beattributabletochangesin undergraduatemedicaleducation,

∗ Correspondingauthorat:DepartmentofAnatomy,JagiellonianUniversityMed- icalCollege,12KopernikaStreet,31-034Krakow,Poland.Tel.:+48795677090.

E-mailaddress:bmhenry55@gmail.com(K.A.Tomaszewski).

1 Theseauthorsequallycontributedtothiswork.

which have reduced the number of curriculum hoursand fac- ultymembersdevotedtogrossanatomy(Cottam,1999).However, irregardlessthecause,asvariantanatomyinpartcontributestosig- nificantnumberofmalpracticeclaims,thereisaneedtoimprove thebasisofanatomicalknowledgeamongpracticingphysiciansby enhancingthemethodsthroughwhichanatomicalinformationis synthesizedandpresentedtothemedicalcommunity.

Evidence-Based Anatomy (EBA), first proposed by Yammine (2014),isaconceptwhichaimstoapplytheevidence-basedprin- ciplesandtechniqueswhicharecommonlyusedinothermedical sciences tothefield ofanatomy. Similarly,as inother fieldsof evidence-basedmedicine(EBM),EBAreliesonsystematicreviews (SR)andmeta-analysis(MA)toprovideahigh-leveloverviewof primaryanatomicalresearch.Althoughnarrativeliteraturereviews arequitecommoninanatomical writing,trueSRswitha clear, detailed,andreproduciblemethodologyallowingforsynthesisof evidence-baseddataarefewandfarbetween.

ThecentralprocessofanSRconsistsofidentifying,selecting, andappraisinghigh-qualitystudiesonafocusedreviewtopic,fol- lowedbytakingtheobtaineddata,analyzingit,andconstructing high-qualityresultswhichprovideevidence-basedinformationto clinicians (Uman, 2011). A SR may also contain an MA, which uses statistical techniquesto pool datafrom severalstudies in order to obtain single quantitativeeffect size estimate(Uman, 2011).ThroughEBA,wecanenhanceourunderstandingofclin- ical anatomy and provide more accurate evidence-based data http://dx.doi.org/10.1016/j.aanat.2015.12.002

0940-9602/©2016TheAuthors.PublishedbyElsevierGmbH.ThisisanopenaccessarticleundertheCCBYlicense(http://creativecommons.org/licenses/by/4.0/).

(2)

thatcanimproveeverydayclinicalpractice.Suchinformationcan alsobeincorporated intothe anatomical educationcurriculum, asevidence-basedsynthesisofanatomicaldatamaybeespecially beneficialforhighlightingthemostdangerousvariants forboth medicalstudentsandyoungsurgeons.InEBA,wecanprobeasso- ciationsbetweenanatomyandvariablessuchasraceorgenderin detail,andthusformconclusionsnotpossiblefromsinglestudies withsmall sample sizes. Furthermore,EBA allows usto recog- nizeareasofanatomythatstillrequireadditionalbasiclaboratory researchand can allowusto explore theconnections between anatomyandpathologyindepth.

While evidence-based methods are commonplace to clini- cal researchers, for many anatomists and other basic science researchersitisarelativelynewandunexploredconcept.Theaim ofthispaperistoprovideastep-by-stepguideforperformingSRs andMAs of anatomicalstudiesas wellastoprovidea uniform methodologyforconductionofEBA.Additionally,wesharetipsand advicesfromourownexperiencewithEBA(Henryetal.,2015a,b;

Ramakrishnanetal.,2015;Royetal.,2015a,b).Lastly,weaddress thelackofuniformstatisticalmethodscurrentlyusedinEBA,which canleadtomisinterpretationofdataandproductionofinaccurate clinicalinformation.

2. Beforeyoubegin

BeforestartinganEBAproject,wehighlysuggestauthorsfamil- iarizethemselveswiththebasictechniquesandproceduresofMA andSR, aswellas ofEBM.While manyresourcesareavailable, wehighly recommendauthors read theCochraneHandbookof SystematicReviewsofInterventions(TheCochraneCollaboration, 2011).Althoughitisfocusedoninterventionalstudies,itnonethe- less provides a very comprehensive source of information on

performingSRandMA.Additionally,werecommendreading“How to Reada Systematic Reviewand Meta-analysis and Apply the ResultstoPatientCare:Users’GuidestotheMedicalLiterature”

byMuradetal.(2014).

3. Step1:Determinethereviewtopicandsetobjectives

An overview of the EBA process is presented in Fig.1. The firststepinEBA,asinallsystematicreviews,isdeterminingthe topicofthereview.However,unlikeclinicalsystematic reviews whichoftencenteronafocusedquestion,reviewsinEBAcanbe eitherfocusedoronabroadertopic,encompassingtheentiretyof anatomicalcharacteristicsofaspecificstructure.Assuch,setting specificprimaryandsecondaryobjectivesforthereviewareessen- tialintheEBAprocess.Whensettingobjectives,werecommendthe authorsstronglyconsidertheclinicalrelevanceoftheanatomical characteristicsofthestudyandhowsynthesizingevidence-based datacouldimproveclinicalpractice.

Oncetheauthorshavedeterminedthetopicforreview,aclear, short,and descriptivetitlefor theprojectshouldbedeveloped, which in a straightforward manner identifies the study as an MAorSR.Furthermore,wewouldrecommendtheregistrationof thereviewonPROSPERO,aninternationalregisterofprospective systematicreviews(http://www.crd.york.ac.uk/PROSPERO/).This preventsduplicationofongoingstudies,andhelpsidentifypoten- tialreportingbiasinMAsandSRs,whichisabiasthatmayhave influencedtheidentificationofrelevantstudies(Sedgwick,2015).

ReportingbiascanoccurinMA/SRinseveralcircumstances,and includesformsofbiassuchas(Sedgwick,2015):

• Citation bias—thetendency forarticles morecommonly cited thanotherstobeidentifiedandincludedintheMA,

Fig.1. TheEBAprocess.SR—systematicreview;MA—meta-analysis;SoF—summaryoffindings.

(3)

• Locationbias—failuretoidentifyarticlesinlessaccessiblejour- nalsordatabases,

• Languagebias—inclusionofarticlesonlyinaccessiblelanguages,

• Publication bias—omission of unpublishedstudies, which are oftennot publisheddue tothe lackof significantresults; for example,ifastudyfoundnosignificantdifferencesinthepreva- lenceofanatomicvariationinadiseasedgroupversusahealthy controlgroup,

• Time-lagbias—rushedordelayedpublicationofastudy,which mayhaveinfluenceditsinclusionintheMA,

• Multiplepublicationbias—inclusionofmultiplearticlesreporting dataandresultsofthesamestudy.

Inthestepsbelow,weprovidetipsonavoidingreportingbias, aswellasotherformsofbias,inanatomicalMA/SRs.

4. Step2:Choosinginclusionandexclusioncriteria

ThenextstepintheEBAprocessistodeterminetheeligibil- itycriteriaforarticleinclusion,theMA/SR.Choosingtheinclusion andexclusioncriteriaforthereviewrequirescarefulevaluationto maintaintheintegrityoftheSR/MAandreducetheriskofbiasin thestudy.ThefirstandsimplestinclusionfactorforanySRorMAis asfollows:Doesthestudyaddresstheobjective(s)ofthereview?

Next,itisimportanttoconsiderwhichpatientdemographics(age, sex,race,healthstatus,etc.)areappropriateforthestudy.Authors shouldpayspecialattentiontopatientdemographics,especially diseasestatus,asitmayincreasetheriskofbiasbyinclusioninthe analysis.Forexample,patientswithmigraineheadacheshavebeen demonstratedtobeatanincreasedriskforvariationsintheCircle ofWillisascomparedtohealthyindividuals(Henryetal.,2015a).

Therefore,inanMAorSRonthevariationsinthecerebralcircu- lation,datafromstudiesthatassessedmigrainepatientsshould notbeincludedinfull,asthatwouldintroduceselectionbias(bias introducedbyinappropriatecollectionandinclusionofdatathus alteringthestatisticaloutcome)intothereview.Nevertheless,such studiesarenotcompletelyuselesstotheproject,astheymaystill includehealthycontrolgroupswhosedatacanbeincludedinthe SR/MA.

Anotherimportantfactortoconsiderinsettinginclusioncrite- riaisthemodalityusedbytheoriginalstudies.Whiledissection andimagingmodalitiesmakeupthemajorityofanatomicalstud- ies,othermethodssuchaselectrophysiologicalorintraoperative studiesalsocontainvaluableanatomicaldata.Whenjudgingwhich modalities to include, it is important to consider whether the modalityitselfmayintroduceselectionbiasintothereview.Occa- sionally,certainspecificcriteriacanbesetforamodalitytoreduce bias,insteadof completelyrejectingthemodality.For example, oldermethodsofimagingmaynothavebeenveryaccurateforthe assessmentoftheanatomicalstructureinquestion,andassuch, couldaffectthereportedprevalencerateorthereportedcharac- teristicsofthestructure.

Itisalsoimportanttoconsiderstudydesignwhenassessingthe modality.Forexample,inourrecentSRandMAontheanatomical variationsofthemediannerve(Henryetal.,2015b),wedecidedto excluderetrospectiveintraoperativestudies.Thesestudiesrelied uponareviewofmedicalrecordsfromcarpaltunnelreleasesur- geriestoreportprevalencedataonmediannervevariations,thus introducingahighriskofinterviewerbias(alsoknownasrecorder bias).Interviewerbiasoccurswhentherearesystematicdifferences intherecordingand/orinterpretationofdatainastudy.Thus,in studieswhichrelyuponpatientchartreviewforanatomicaldata, werecommendcarefulevaluationtodeterminewhetherinclusion ofsuchstudiesisappropriateintheSR/MA.

Case reports, conference abstracts, and letters to the editor shouldalwaysbeexcluded.Whiledatafromunpublishedstudies cantechnicallybeused,werecommendedcautionandcarefuleval- uationofallnon-peerreviewedstudies.Publishedstudiesreporting missing,unclear,orincompleteresultsshouldalwaysbeexcluded ifanattempttocontacttheauthorsoftheoriginalstudyforclari- ficationisnotpossible,unsuccessful,ordoesnotresolvetheissue.

Furthermore,wehighlyrecommendthecarefulevaluationofall studiesthatdonotclearlydefine(bytextorfigures)thedescriptive anatomyusedinthestudy,astheymayintroducebias.Inaccor- dancewiththeabove,werecommendclearanatomicaldefinitions bestatedintheinclusioncriteriaoftheMA/SR.

Forexample,inourMAand SRonthemediannerveandits thenarmotorbranch(TMB) (Henry etal.,2015b), ourinclusion criteriawereasfollows:

(1)thestudyreportedextractableprevalencedatarelatedtoLanz’s classificationtype1,2,3,or4ordataonsideofbranchingof theTMB,

(2)thestudyhadclearlydefineddescriptionsofTMBvariations, and

(3)itwasacadavericoraprospectiveintraoperativestudy.

5. Step3:Buildandexecuteasearchstrategy

Duetothecharacteristicsofanatomicalstudies,aswellasthe multitudeofdifferentstudytypeswhichcontainanatomicaldata, developinganeffectivesearchstrategyforEBAisparticularlytricky comparedtomanyotherfieldsofEBM.Abroadsearchstrategy shouldbeusedtominimizelocationbias.Authorsshouldbegin withdevelopingacomprehensivelistofkeywordsforthesearch.

Duringthekeywordbuildingprocess,authorsshouldreviewthe NationalLibraryofMedicine,Medical SubjectHeadlines(MeSH, http://www.ncbi.nlm.nih.gov/mesh).Anatomicalstructuresoften haveseveralnamesoreponymsthatrequireinclusion.

Inordertobeascomprehensiveaspossible,authorsshouldtake thetimetosearchseveralelectronicdatabasesduringthesearch process,usingasearchstrategythatistailoredindividuallytoeach.

Atminimum,wewouldsuggestsearchingPubMed,Embase,Web ofScience,andScienceDirect.Duetothelonghistoryofanatomi- calstudies,werecommendnotsettingadatelimit.Additionally, in order to obtain globally applicable data, we do not suggest applyinganylanguagerestrictionstothesearch.Whenbuilding asearchstrategy,thekeyistobalancesensitivity(identifyinga highproportionofrelevantarticles)withspecificity(retrievinga lowproportionofirrelevantarticles;Uman,2011).Wewouldurge cautionwhenusingrestrictivetermsorfilters(suchas“humans”) duringdatabasesearching.Fromourexperience,wehavenoticed thetendencytoloseviablearticleswhensuchtermsorfiltersare applied.Moreover,wewouldstronglyadviseauthorstoseekadvice whenbuildingasearchstrategybyconsultingalibrarianoracol- leaguewhohasextensiveexperiencewithSRandMA.Authorsmay alsoconsultandcooperatewiththeInternationalEvidence-Based AnatomyWorkingGroup(iEBA-WG,www.eba.cm.uj.edu.pl/),the coordinatorsofwhicharetheauthorsofthisguide.TheiEBA-WGis alwaysopentoassisting,helping,andcollaboratingwithallthose interestedinEBA.

Inadditiontodatabasesearching,werecommendahandsearch ofthemajoranatomicaljournals(e.g.,AnnalsofAnatomy,Jour- nalofAnatomy,AnatomicalRecord,ClinicalAnatomy,Surgicaland RadiologicAnatomy,AnatomicalScienceInternational,FoliaMor- phologica,etc.),aswellastherelevantclinicaljournalsrelatedto theanatomicalstructureofthestudy.Lastly,authorsshouldsearch thereferencesofallincludedstudiesforadditionalarticleseligible forinclusioninthemeta-analysis.Thisisparticularlyimportant

(4)

inEBA,asmanyoftheolderorforeignstudiesmaybedifficultto identifyduringdatabasesearches.

6. Step4:Studyselection

Identifying articles eligible for inclusion should begin with screeningand excludingarticlesbytitle andabstract.Afterthis initialscreening,full-textarticlesshouldbeusedtoassesstheeli- gibilityofarticlesforinclusioninthemeta-analysis.

Assessmentofstudyeligibilityforinclusioninthemeta-analysis shouldbepreferablyperformedindependently bytwoseparate reviewers.Intheeventofdisagreementbetweentworeviewers duringtheeligibilityprocess,wesuggestthatitshouldberesolved byformingaconsensusamongtheentirereviewteam,meanwhile consultingwiththeauthoroftheoriginalstudywhennecessary andpossible.

Translatorsshouldbeusedwheneveraccessible,limitingexclu- sion of studieswith valuable data in languages not spoken by theauthorsandthusreducinglanguagebias.Studieswithunclear anatomicaldescriptionsandincompleteormissingresultsrequire carefulevaluation.Thedecisiontoexcludeastudyduetoanyofthe aboveshouldonlybemadeafterattemptingtoseekclarification fromtheauthor(s)oftheoriginalstudy,wheneverpossible.

Duringthestudyselectionprocess,thereviewteamshouldbe vigilantofanyduplicatepublicationsofthesamestudydata.From ourexperienceinEBA,wehavenoticedtwocommontypesofdupli- catepublications.Thefirstismultiplepublicationofthesamestudy datainmultiplelanguages.Thesecondtype,slightlymoredifficult todetect,isthatsometimefollowingthepublicationofastudy,a secondstudybythesameauthor(s)onthesametopic,withalarger samplesize,ispublished.Whenthisoccurs,itisoftendifficultto determineifthesecondstudywasinfactacompletelynewsam- pleoranexpansionoftheoriginalsample.Thissituationrequires carefulevaluationofthestudiestominimizeriskofmultiplepubli- cationbias,andthereviewteamshouldmakeeveryeffortpossible tocontacttheauthorsoftheoriginalstudyforclarification.Ifthe reviewteamstronglysuspectsorconfirmssampleexpansionof apreviousstudy,werecommendexcludingtheolderstudy,and usingonlythemostrecentstudy,withthelargestsample.

7. Step5:Dataextraction

Dataextractionshouldpreferablybeperformedindependently bytwodifferentreviewers.Intheeventthatonlyonereviewerper- formsdataextraction,wehighlyrecommendthatitischeckedfor errorsbyasecondorevenathirdreviewerwheneverpossible.For easeofuse,werecommendthatdatabeextractedintotablesusing softwaresuchasMicrosoftExcel.Theextracteddatashouldalways include:author,year,country,studydesign,population,modality, andsamplesize/numberofstructuresassessed.Asageneralrule, werecommendtheextractionofasmuchrawanatomicaldataas possible.Wheneverpossible,subgroupdata[suchasanatomical datawithrespecttogender,laterality,side(leftvs.right),age,etc.]

shouldbeextracted.

Whenanatomicaldefinitionsvarybetweenstudies,carefuleval- uationofeach articleshouldbeperformedbytheentirereview team.Duringtheextractionprocess,inaccuraciesinthereported resultsoftheindividualstudiesareinevitable.Whentheydooccur, werecommendthatauthorsoftheoriginalstudiesbecontacted, andfurtherdatasoughtafter,ifnecessary.

Additionalrawstudydataobtaineddirectlyfromtheauthorsof thepublishedstudyshouldbecarefullyevaluatedforanyincon- sistencies beforeinclusion inthe meta-analysis.However, such additionaldatamaybehelpfulinseveralcircumstances,andnot onlyincasesofdatadiscrepancies.Forexample,astudymayhave

reportedthatnosignificantdifferenceswerefoundinthepreva- lenceofavariablewithrespecttogenderorside(leftvs.right), butdidnotreporttheactualproportionvalue.Insuchasituation, accesstotherawstudydatasetwouldallowfortheextractionand analysisofadditionalanatomicaldata.Thus,wewouldencourage reviewerstoseekoutadditionalrawdatafromthestudyauthors whenitmaypotentiallybebothusefulandavailable.

Oneissuethatrequiresspecificattentionduringextractionfor anEBAmanuscriptishowtherateforananatomicalcharacteristic ispresentedwithrespecttothesample.Forexample,ifastructure ispresentbilaterallyinhumans,aswiththemediannerve,itmay bepresentedinoriginalstudiesasarateperlimborperperson.It isimportanttoclarifyintheinclusioncriteriawhichratetheywill extract,andhow (ifappropriateandpossible) theywillconvert oneratetotheotherforthepurposesoftheanalysis.Iftheneedfor conversionarises,wehighlyrecommendcontactingtheauthorsof theoriginalstudiestoinquireaboutadditionaldata.

Lastly,whenextractingforprevalencerates,ifrareanomalies arereportedin thestudy,wesuggesttwowaystopreservethe statisticalintegrityofthedata.Optiononeistosubtractthenumber ofpatientswithrareanomaliesfromthetotalsamplesize,andthus notincludethem intheanalysis.Theexcludedanomaliesalong withdetailedexplanationsandrationaleforexclusionshouldbe clarifiedintheMA/SR.Option twoistocreate arareanomalies categorytoincorporatetheanomalies,thuscreatingaprevalence rateforthiscategory.Thechoiceshouldbemadeaftercarefully evaluatingthedataandconsideringtheclinicalimplications.When adecisionisreached,therationaleforthedecisionshouldbefully explainedinthemanuscript.

8. Step6:Statisticalanalysis

WhenperforminganMAofanatomicaldata,itisimportantto considerhowtoanalyzethedatainordertopresentresultswhich canbeeasilyunderstoodandappliedtotheclinicalsetting.Inan SRthatdoesnotincludeanMA,wehighlyrecommendtheuseof SummaryofFindings(SoF)tablestopresentdatainaclear,com- prehensibleformat(Langendametal.,2013).

First,duetomultiplestatisticalissues,wehighlyrecommend authorsavoidsimplepoolingofrawdata,a relativelypoorpro- cedureascomparedtoanMA(BravataandOlkin,2001).Simple pooling refers to the process of analyzing all the data as if it wasobtainedfromthesamesample,withoutproperweighting.

Meta-analysis allows for weightingof individualstudies before incorporationintothefinalanalysis,whichavoidsseveralofthe issues withsimple datapooling (Bravata andOlkin,2001).Fur- thermore,MAallowsforassessmentofheterogeneity(anyform ofvariability)betweenstudiesandtheuseofvariousstatistical modelstoimprovetheaccuracyofapooledeffectmeasureesti- mate(thestatisticalresultofthemeta-analysis).Werecommend consultationwithabiostatisticianwhohasexperienceinMAfor expertadviceonproperlyanalyzingdata.

InMA,effectsizesaregivenwiththeir95%confidenceinter- vals(CIs)and canbepresentedinboth quantitativeformatand graphicalrepresentation(forestplots;Uman,2011).Inanatomical MA,whichcontainsmanyvariablesand/orsubgroups,werecom- mendtopresentdatainhighlydetailedtablesforthepurposeof improvingreadability(Royetal.,2015a).

Theresultsofanatomicalstudiesgenerallyconsistoftwomain types of data—proportions (e.g., prevalence rates of variations, duplication,etc.)andmeans(e.g.,meannervelength,meanves- seldiameter,meannumber ofnervefibers, etc.).Depending on thetypeofdata,apropereffectmeasurecanbechosen.Fordata ofproportions,theeffectmeasureusedispooledprevalenceesti- mate(PPE),whichprovidesanestimateoftheprevalenceofan

(5)

anatomicalvariableinachosenpopulation.For PPE,werecom- mendtheuseofthefreesoftwareMetaXLversion2.0byEpiGear InternationalPtyLtd.(Wilston,Queensland,Australia).MetaXLis apluginforMicrosoftExcel whichallows forthecalculationof PPEforbothsinglecategory(i.e.,simpleprevalence)ormultiple categoryMA.ThemultiplecategoryMAisespeciallyusefulforcal- culatingtheprevalenceofdifferenttypes ofvariationsatonce;

forexample,thePPEofeachtypeofvariationinaclassification.

MetaXLimplementsadoublearcsinetransformationwithaback- transformation toreport thePPE (Barendregt etal., 2013).The doublearcsinetransformationstabilizesthevarianceinamulti- categoricalprevalenceMAbymakingvariancedependentonlyon thepopulationsize(Barendregtetal.,2013).Thistypeoftransfor- mationhasbeenshowntobepreferentialtologittransformations inmulti-categoricalprevalenceMA(Barendregtetal.,2013).The PPEshouldalwaysbepresentedwithits95%CI.

Forcalculating apooledmean inanMAforanatomical data reportedasmeans,werecommendtheuseoftheComprehensive MetaAnalysissoftwarebyBiostatInc.(Englewood,NJ,USA).This softwareallowsforeasyMAofmeansinonegroup.Itrequiresthe inputofsamplesize,mean,andthestandarddeviationfromeach studytocalculateapooledmeanvaluewitha95%CI.

InbothtypesofMA,heterogeneityamongthestudiesshould alwaysbeassessed,preferentiallyusingboththeChi-squaredtest and I-squaredstatistic.For theChi-squaredtest, thep-value of Cochran’sQshouldbereported,withavalueof<0.10considered toindicate statisticallysignificant heterogeneitybetween stud- ies(TheCochraneCollaboration,2011).TheI-squaredstatistic,an overallmeasureofheterogeneity,shouldbereportedwithits95%

CI.TheI-squaredstatisticshouldbeinterpretedasfollows:0%to 40%mightnotbeimportant;30%to60%mayrepresentmoderate heterogeneity;50%to90%mayrepresentsubstantialheterogene- ity;and75%to100%mayrepresentconsiderableheterogeneity, inaccordance withtheguidelinesin thecurrentversion ofthe CochraneHandbookofSystematicReviewsofInterventions(The CochraneCollaboration,2011).

Fromourownexperience,wefoundthattheheterogeneityin anatomicalMAsisalmostalwayshigh.Assuch,duetointrinsichet- erogeneityinanatomicalstudies,andtomaximizethevalidityof theresults,werecommendonlytheuseofarandom-effectsmodel inanatomical MA.Unlike a fixed-effects modelwhich assumes thatthedifferencebetweentheresultsofstudiesareduesolely tochance,arandomeffectsmodelassumesthattheeffectsbeing estimated(i.e.,themeasurementsoftheanatomicalvariables)in thedifferentstudiesarenotidentical(TheCochraneCollaboration, 2011).Theimplementationofrandom-effectsmeta-analysisisas simpleasselectingtherandom-effectsresultstabinComprehen- siveMetaAnalysisorchoosingtousetherandom-effectsformula inMetaXL.

The sources of heterogeneity in the MA should always be explored.Toprobethem,subgroupanalysisandsensitivityanalysis shouldbeperformed.Subgroupanalysisbygeographicaldistribu- tionofthestudiesandbythemodalityofthestudy(e.g.,dissection vs.imagingstudies)shouldalmostalwaysbeperformed.Othersub- groupanalyses,suchasbygender,age,laterality,andside(leftvs.

right)shouldbeperformedwheneverdataareavailable.Forcom- paringsubgroups,werecommendthatthesamemeasuresusedfor themainanalysis(PPEorpooledmean)beusedforthesubgroup analysis.Thus,werecommendtheuseofCIstoassessforstatisti- callysignificantdifferencesbetweentwoormoresubgroups.Ifthe CIsbetweentwosubgroupsoverlap,thedifferencescanbecon- sideredinsignificant,whileifthereisnooverlapbetweenCIs,we canconsiderthedifferencesbetweenthegroupstobestatistically significant.Inourexperience,becauseofthehighheterogeneity inanatomicalMA,itiscommontohavewideconfidenceintervals, thusmakingitdifficulttodetectstatisticallysignificantdifferences.

However,theuseofeffectsizeestimateCIsforcomparingsub- groupdifferencesallowsthereaderstointerpretforthemselves clinicallysignificantdifferences,regardlessofthewidthoftheCIs orstatisticalsignificance.

Althoughtheeffectmeasuresofoddsratio(OR;themeasure ofassociationbetweentwovariables)ormeandifference(MD;the absolutedifferenceinmeansbetweentwogroups)canalsobeused tocomparestatisticaldifferencesbetweensubgroups,werecom- mendauthorsavoidtheseforthegeneralreportingofthedata,as theyareoftenconsidereddifficulttointerpretbycliniciansandlack theeaseofinstantclinicalapplicationthatpooledmeanorpooled prevalenceestimatesprovide(TheCochraneCollaboration,2011).

ThisdoesnotmeanORandMDdonothaveaplaceinEBA.Werec- ommendtheirusewhenstudyingaspecificassociation;forexam- ple,betweenavariationandapathology,suchasalinkbetween migraineandCircleofWillisvariations(Henryetal.,2015a).

Lastly,asensitivityanalysisshouldalmostalwaysbeperformed tohelpexplorethesourcesofheterogeneityintheMA.Asensitiv- ityanalysisassesseswhetherthefindingsoftheanalysisarerobust todecisionsmadeintheprocessofperformingthemeta-analysis (TheCochraneCollaboration,2011).Itisperformedbyrepeating theanalysis,aftersubstitutinganalternativedecision(e.g.,chang- ingtheminimumsamplesize,excludingaparticularstudy),and lookingforsignificantdifferencesbetweentheobtainedresultsand theprimaryanalysis.(TheCochraneCollaboration,2011).Ingen- eral,werecommendperformingasensitivityanalysisbylimiting inclusiontostudieswithasamplesizeof≥100.However,theinclu- sionsamplesizeshouldbedeterminedbasedontherangeofdata availableintheMA.Additionally,authorscantryaleave-one-out analysis,performedbyremovingonestudyata time,and then repeatingtheanalysistoprobeifasinglestudysignificantlydrove theresultsoftheprimaryanalysis.

9. Step7:Writethemanuscript

Whenpreparingthemanuscript,wehighly recommendthat authorsalwaysfollowspecificguidelinesforthereportingofSR andMA.WesuggesttheuseofPreferredReportingItemsforSys- tematicReviewsandMeta-Analyses(PRISMA)guidelines(Moher etal.,2009)ortheMeta-AnalysisOfObservationalStudiesinEpi- demiology(MOOSE)guidelines(Stroupetal.,2000).Itshouldbe notedthatitisbecomingmorecommonforjournalstohavespe- cificguidelinesforsubmissionofSRandMA,andmayspecifically requirePRISMAorMOOSE.Assuch,wehighlysuggestthatwhen preparingthemanuscript,theauthorschecktheguidelinesofthe journalforwhichtheyplantosubmitthemanuscript.

Inadditiontotherequirementsabove,werecommendafew EBA-specific additional items for the introduction and discus- siontoensurequalityofthemanuscript.Theintroductionofthe manuscriptshouldcontaina detailed,butconcisereviewofthe relevant anatomy, embryology, and clinical significance of the structure and its variations. Imagesof therelevant anatomical structure and its variations should be also included whenever possible.Additionally, theintroduction should alsoinclude the anatomicalandclinicalrationaleforthereview,aswellastheaims andobjectivesofthereview.Werecommendthatthediscussion sectionbeaclinicallyfocusedreview oftheresults,andshould addressthepracticalimplicationsofthedataandtheneedfor,or thedirectionoffurtherresearch.

10. LimitationsofEBA

AfewlimitationsofEBAshouldbementioned,especiallywith respecttoMA. Oneof themajor limitations of EBAis thelack ofanobjectivestudyqualityandriskofbiasassessmenttoolfor

(6)

anatomicalstudies.Generally,inEBM,studieswithpoormethod- ologicalqualityorathighriskofintroducingbiasintotheMA/SR areexcluded.FromourownexperienceinEBA,wehavefoundthat manyanatomicalstudiesraiseseriousquestionswithrespectto methodologicalqualityandthereportingofresults.Similarfind- ingsofpoorreportinginanatomicalstudieshavebeenreportedby others(Yammine,2014).Tolimitsuchbiasasmuchaspossible, westronglyrecommendcontactingtheauthorsoforiginalstudies wheneverencounteringanyinconsistencies.

Wearecurrentlyengagedinthedevelopmentofatooltoassess themethodologicalqualityandriskofbiasinanatomicalstudies.

Inaddition,wearedevelopingachecklisttoimprovethereporting ofmethodsandresultsinoriginalanatomicalstudies.

Lastly,assessmentofpublicationbiasisastandardpartofMAin EBM.However,thecurrenttoolsforassessmentofpublicationbias aredesignedprimarilyforMAofinterventionalstudies.Thus,for MAofmulti-categoricalPPEandonegrouppooledmean,thelack ofaproperassessmentofpublicationbiasremainsalimitation.

11. Conclusions

Evidence-basedanatomy representsthenext frontierin the advancementofclinicalanatomy,anditsexplorationwillimprove clinicalpracticeacrosstheexpanseofmedicalandsurgicalspecial- ties.Throughthemethodsandtechniquesdescribedinthispaper, auniformstrategyforSRandMAinanatomyispresented,which will improve the quality and clinical applicability of evidence derivedfromEBA.

Acknowledgments

KrzysztofA. Tomaszewskiwassupportedby theFoundation forPolishScience(FNP).Thepublicationofthismanuscriptwas supportedbytheFacultyofMedicine,JagiellonianUniversityMed- icalCollegeKNOW(LeadingNationalResearchCentre2012–2017) funds.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound, intheonlineversion,athttp://dx.doi.org/10.1016/j.aanat.2015.12.

002.

References

Barendregt,J.J.,Doi,S.A.,Lee,Y.Y.,Norman,R.E.,Vos,T.,2013.Meta-analysisof prevalence.J.Epidemiol.CommunityHealth67,974–978.

Bravata,D.M.,Olkin,I.,2001.Simplepoolingversuscombininginmeta-analysis.

Eval.HealthProf.24,218–230.

Cottam,W.W.,1999. Adequacyofmedicalschoolgrossanatomyeducation as perceivedbycertainpostgraduateresidencyprogramsandanatomycourse directors.Clin.Anat.12,55–65.

Henry,B.M.,Roy,J.,Ramakrishnan,P.K.,Vikse,J.,Tomaszewski,K.A.,Walocha,J.A., 2015a.Associationofmigraineheadacheswithanatomicalvariationsofthe CircleofWillis:evidencefromameta-analysis.Neurol.Neurochir.Pol.49, 272–277.

Henry,B.M.,Zwinczewska,H.,Roy,J.,Vikse,J.,Ramakrishnan,P.K.,Walocha,J.A., Tomaszewski,K.A., 2015b.Theprevalence ofanatomicalvariationsofthe mediannerveinthecarpaltunnel:asystematicreviewandmeta-analysis.PLoS One10,e0136477.

Langendam,M.W.,Akl,E.A.,Dahm,P.,Glasziou,P.,Guyatt,G.,Schünemann,H.J., 2013.AssessingandpresentingsummariesofevidenceinCochranereviews.

Syst.Rev.2,81.

Moher,D.,Liberati,A.,Tetzlaff,J.,Altman,D.G.,2009.Preferredreportingitemsfor systematicreviewsandmeta-analyses:thePRISMAstatement.PLoSMed.6, e1000097.

Murad,M.H.,Montori,V.M.,Ioannidis,J.P.A.,Jaeschke,R.,Devereaux,P.J.,Prasad,K., Neumann,I.,Carrasco-Labra,A.,Agoritsas,T.,Hatala,R.,Meade,M.O.,Wyer,P., Cook,D.J.,Guyatt,G.,2014.Howtoreadasystematicreviewandmeta-analysis andapplytheresultstopatientcare.J.Am.Med.Assoc.312,171.

Ramakrishnan,P.K.,Henry,B.M.,Vikse,J.,Roy,J.,Saganiak,K.,Mizia,E.,Tomaszewski, K.A.,2015.Anatomicalvariationsoftheformationandcourseofthesural nerve:asystematicreviewandmeta-analysis.Ann.Anat.—Anat.Anzeiger202, 36–44.

Rogers,S.O.,Gawande,A.A.,Kwaan,M.,Puopolo,A.L.,Yoon,C.,Brennan,T.A.,Stud- dert,D.M.,2006.Analysisofsurgicalerrorsinclosedmalpracticeclaimsat4 liabilityinsurers.Surgery140,25–33.

Roy, J.,Henry, B.M., P ˛ekala, P.A., Vikse, J.,Ramakrishnan, P.K.,Walocha, J.A., Tomaszewski,K.A.,2015a.Theprevalenceandanatomicalcharacteristicsofthe accessoryheadoftheflexorpollicislongusmuscle:ameta-analysis.PeerJ3, e1255.

Roy,J.,Henry,B.M.,P ˛ekala,P.A.,Vikse,J.,Saganiak,K.,Walocha,J.A.,Tomaszewski, K.A.,2015b.Medianandulnarnerveanastomosesintheupperlimb:ameta- analysis.MuscleNerve.

Sedgwick,P., 2015. Meta-analysis:testing forreporting bias.Br. Med.J. 350, g7857.

Stroup,D.F.,Berlin,J.A.,Morton,S.C.,Olkin,I.,Williamson,G.D.,Rennie,D.,Moher,D., Becker,B.J.,Sipe,T.A.,Thacker,S.B.,2000.Meta-analysisofobservationalstudies inepidemiology:aproposalforreporting.Meta-analysisofobservationalstudies inepidemiology(MOOSE)group.J.Am.Med.Assoc.283,2008–2012.

TheCochraneCollaboration,2011.CochraneHandbookforSystematicReviewsof Interventions,Version5.1.0[WWWDocument].

Uman,L.S.,2011.Systematicreviewsandmeta-analyses.J.Can.Acad.ChildAdolesc.

Psychiatry20,57–59.

Yammine,K.,2014.Evidence-basedanatomy.Clin.Anat.27,847–852.

Cytaty

Powiązane dokumenty

Wskazują one, iż w dokładnym oszacowaniu prawdopodobieństwa złośliwości zmiany należy oprzeć się na doświadczeniu klinicznym bądź też użyć jednego ze zweryfikowanych modeli

The purpose of the research undertaken was to analyse the knowledge and skills of nursing staff related to the use of scientific evidence in everyday professional

Ocena związku poszczególnych pozycji podskali IV („Częstości wykorzystywania poszczególnych elementów Evidence-based Practice w codziennej pracy klinicznej”) z ogólnym

Wyniki badań własnych wyodrębniły bardzo podobne czynniki ograniczające wykorzystanie EBN w praktyce polskich pielęgniarek, są to: niezrozumia- łość analiz

Celem pracy była analiza polskiego piśmiennictwa naukowego podejmującego problematykę zastosowania paradygmatu Evidence-based Medicine w badaniach naukowych oraz zastosowania

An analysis of highly cited trials published in the three journals with the highest impact factor (The New England Journal of Medicine, The Lancet, Journal of the American

Address for correspondence: Ioannis Tomos, Second Pulmonary Medicine Department, Medical School, National and Kapodistrian University of Athens, ATTIKON University Hospital, 1

Na przestrzeni lat pojawiło się kilka skal opisujących nasilenie zmian skór- nych w polu napromienianym rekomendowanych albo przez takie stowarzy- szenia, jak EORTC, RTOG, WHO,