J Adv Nurs. 2019;00:1–27. wileyonlinelibrary.com/journal/jan  | 1© 2019 John Wiley & Sons Ltd

Received:18July2018  |  Revised:25February2019  |  Accepted:2April2019DOI: 10.1111/jan.14078

R E V I E W P A P E R

Drawing blood from peripheral intravenous cannula compared with venepuncture: A systematic review and meta‐analysis

Linda L. Coventry RN, MS, PhD, Post‐Doctoral Research Fellow1,2  | Alycia M. Jacob BA, Research Assistant3  | Hugh T. Davies RN, MS, PhD, Lecturer3  | Laurita Stoneman RN, Clinical Support Nurse, Transfusion Trainer4 | Samantha Keogh RN, BSc(Hons), PhD, Professor of Nursing2,5  | Elisabeth R. Jacob RN, MEd., PhD, Associate Dean Nursing3

1CentreforNursingResearch,SirCharlesGairdnerHospital,SchoolofNursingandMidwifery,EdithCowanUniversity,Joondalup,Perth,Australia2AllianceforVascularAccessTeachingandResearchGroup(AVATAR),MenziesHealthInstituteQueensland,GriffithUniversity,Brisbane,Queensland,Australia3SchoolofNursingandMidwifery,EdithCowanUniversity,Joondalup,Perth,Australia4BairnsdaleRegionalHealthservice,Bairnsdale,Victoria,Australia5SchoolofNursing,InstituteofHealthandBiomedicalInnovation(IHBI),QueenslandUniversityofTechnology,Brisbane,Queensland,Australia

CorrespondenceLindaCoventry,SchoolofNursingandMidwifery,Building21,EdithCowanUniversity,270JoondalupDrive,Joondalup,Perth,6027,Australia.Email:l.coventry@ecu.edu.au

AbstractAims: To synthesize the evidence evaluating if blood samples are similar when obtainedfromperipheralintravenouscannulacomparedwithvenepuncture.Design: Asystematicreviewandmeta‐analysiswasundertaken.Data sources: Searcheswereconducted indatabases forEnglish languagestudiesbetweenJanuary2000–December2018.Review methods: ThesearchadheredtotheMeta‐analysisofObservationalStudiesinEpidemiologyguidelines.ThemethodologicalqualityofstudieswasassessedusingJoannaBriggscriticalappraisalinstruments.TheoverallqualityoftheevidencewasassessedusingtheGRADE.Results: Sixteenstudieswereidentified.Findingssuggesthaemolysisratesarehigherinbloodsampled fromperipheral intravenouscannula.However,haemolysis ratesmaybelowerifaperipheralintravenouscannulabloodsamplingprotocolisfollowed.Forequivalenceofblood test results,even thoughsome resultswereoutside thelaboratory,allowableerrorandwereoutsidetheBland–AltmanLevelofAgreement,noneofthesevalueswouldhaverequiredclinical intervention.Withregardtothecontaminationratesofbloodcultures,theresultswereequivocal.Conclusion: Furtherresearchisrequiredtoinformtheevidenceforbestpracticerec‐ommendations,including,ifaprotocolfordrawingbloodfromaperipheralcannulaisofbenefitforspecificpatientpopulationsandinothersettings.Impact: Venepuncture can provoke pain, anxiety and cause trauma to patients.Guidelinesrecommendbloodsamplesfromperipheralintravenouscannulabetakenonlyoninsertion.Anecdotalevidencesuggestsdrawingbloodfromexistingcannulasmaybeacommonpractice.Furtherresearchisrequiredtoresolvethisissue.

K E Y W O R D S

acutecare,adultnursing,diagnostictests,haemolysis,peripheralvenouscatheterization,phlebotomy,systematicreviewsandmeta‐analyses,venepuncture

2  |     COVENTRY ET al.

1  | INTRODUC TION

Patients admitted tohospital are frequently subjected tomultipleinvasive tests including venepuncture and peripheral intravenouscannula(PIVC)insertion.Patientsmayrequiremultiplebloodteststoassistindiagnosisandmanagementofmedicalconditionsandtheappropriatemethodofobtainingthebloodsamplecanbeatopicofdebate.Venepuncturecanprovokeanxiety,bepainfulanduncom‐fortable,causebruising,haematoma,infections,vasovagalreactionsandinrarecasesperipheralnervedamage(Buowari,2013;Tsukudaetal.,2016).Intheemergencydepartment(ED)itisacommonprac‐ticeforstafftotakethebloodsamplefromaPIVCwhenanewlineisplaced.Thisreducestheneedforanadditionalpainfulvenepunc‐ture.ItisestimatedthatoverabillionPIVCsworldwideareinsertedeachyear(Alexandrouetal.,2018).

1.1 | Background

Current Australian (Clinical Excellence Commission, 2013;GovernmentofWesternAustraliaDepartmentofHealth,2017;QueenslandGovernmentDepartmentofHealth,2015)andUKnational (RoyalCollegeofNursing,2016)guidelinesstatethatbloodsamplesmaybedrawn fromaPIVCdirectlyafter inser‐tion,butnotatothertimes.Twoguidelines(Gorskietal.,2016;GovernmentofWesternAustraliaDepartmentofHealth,2017)alsostateconsiderobtainingabloodsamplefromaPIVCinanemergency,whenthepatienthas limitedvascularaccess,or isatincreasedriskofbleeding,orreceivingthrombolytictherapy.Irrespectiveofcurrentguidelines,anecdotalevidencesuggeststhatwithdrawingbloodfromPIVCmaybeacommonpractice.Patientsmayoftenneedmultipleblood tests tomonitor theircondition. Examples include the patient with gastrointestinalbleedingmayneedrepeathaemoglobin;thepatientwithacutecoronarysyndromemayneedrepeattroponin;andthepatientrequiring glucose tolerance testing requires repeat blood glu‐cosetests.

AdvantagesofwithdrawingbloodfromaPIVCincludeconve‐nienceofaccess,decreasedstaffworkload,lowcostandlesspainforthepatientduetoanadditionalvenepuncture.Disadvantagesmayincluderiskofhaemolysis,non‐equivalenceofthebloodtestresults, riskof infectionand risk to thepatencyof thecannula.Haemolysis,or redcellbreakdown,canpotentially lead to inac‐curate blood test results andmay require a secondblooddrawthat leadstodelay intreatment, increasedstaffworkload,addi‐tionalcostsandunnecessarypaintopatientsduetotherequire‐mentof repeatedblood tests. TheAmericanSocietyofClinicalPathologybenchmarkforbestpracticedefinethattheacceptablerateofsamplerejectionduetohaemolysisis2%orless(Loweetal.,2008;Phelan,Reineks,Schold,Kovach,&Venkatesh,2016).Estimatesofhaemolysisratesrangefromlessthan1‐36%(Phelanetal.,2016).

A recently published systematic review (McCaughey et al.,2017) explored differences in haemolysis rates; however, they

did not conduct meta‐analysis.We found no published system‐atic review that analysed the equivalence of blood test results.Asystematicreview(Snyderetal.,2012)examinedeffectivenessforreducingbloodculturecontaminationratesandsearchedtheliterature up to 2011, so an updatewas timely. Although blooddrawsviavenepunctureareconsideredastandardpractice,acriti‐calevaluationofthepotentialvalueofblooddrawsusingthePIVCtechniqueisrequired.Therefore,asystematicreviewincludingameta‐analysiswasconductedtogiveanevidence‐basedanswertotheresearchquestion.

2  | THE RE VIE W

2.1 | Aims

The aimof this reviewwas to synthesize the evidence evaluatingifhaemolysisrates,equivalenceofbloodresultsandcontaminationrates,betweenbloodsamplesobtainedfromPIVCarecomparablewithvenepuncture.Assuch,thisreviewquestionis:Arehaemoly‐sisrates,bloodtestresultsandcontaminationratescomparableforbloodsamplesobtainedbyPIVCandvenepunctureforpatients inacutehealthservices?

2.2 | Design

2.2.1 | Types of participants

Thisreviewincludedstudiesinvolvingadultsaged18yearsandoverwhowereadmitted inanacutecarehospital settingand requiredbloodsamplestobecollected.

2.2.2 | Types of interventions

TypesofinterventionswerestudiesthatinvestigatedtheeffectofdrawingbloodfromaPIVC.

2.2.3 | Comparator

Onlystudieswithvenepunctureasthecomparatorwereincluded.

2.2.4 | Outcome

Thisreview includedstudiesthat investigatedthefollowingout‐comes; haemolysis of blood samples, equivalenceof blood sam‐plesandcontaminationofbloodculturesamples. Itwasdecideda priori for equivalence of blood samples that only studies thatconductedBland–Altmanplotsandanalysedmeandifferencesinbloodtestresultswouldbeincluded(Bland&Altman,1986).Otheroutcomeswe considered but did not find any research onwereriskof:catheterocclusion,phlebitis,dislodgement,devicefailure,catheter‐relatedbloodstreaminfections,infiltration,blockageandcannulapatency.

     |  3COVENTRY ET al.

2.2.5 | Types of studies

This review considered published observational studies includingrandomized control trials, non‐randomized control trials, quasi‐ex‐perimentalstudies,beforeandafterstudies,prospectiveandretro‐spectivecohortstudiesandanalyticalcross‐sectionalstudies.Thisreviewalsoconsidereddescriptivestudydesignsforinclusion.

2.3 | Search methods

Thesearchstrategyadheredto theMeta‐analysisofObservationalStudies in Epidemiology study guidelines (Stroup et al., 2000) andwas undertaken using the databases CINAHL, Cochrane Library,MEDLINE, Scopus, ISI Web of Science and Joanna Briggs. Twosearches were conducted. The first search (January 2000–April2017)wasperformedusingacombinationofsearchterms,includingintravenouscatheterORintravenouscannulaORperipheralvenouscatheter OR peripheral venous cannula AND phlebotomy OR ve‐nepunctureORdirectvenouspuncture.Thesecondsearch(January2000–December2018)wasperformedtoupdatetheliteratureandincluded theoutcomemeasures in the search strategy. In addition,totheabovetermswealsoincludedriskfactors,infection,phlebitis,morbidity mortality, dwell time, device failure, device malfunction,occlusion,blockage,infiltration,extravasationanddislodgementwithassociatedBoolean logic. The search strategywas adapted for the

differentdatabasesandalltermsweresearchedwithMedicalSubjectHeadingsandaskey(text)words(Appendix1&2). Inaddition,thereferencesofretrievedarticleswerecheckedandotherarticlesthatcitedtheretrievedarticleswerecheckedusingcitationalertwiththeISIWebofKnowledge.Selectionofpapersforinclusioninthestudywasundertakenindependentlybytwomembersoftheresearchteam.

WeaimedtoincludeallpublishedresearchstudiesthatwerewritteninEnglish.Studiespublishedbefore2000wereexcluded.Therationaleforthiswassuchthatthereviewreflectedthecontemporarypracticeinproductswithvascularaccessandphlebotomy.Theinvasivecomponentofmodern‐daycathetersaremuchmorepliableandsmoothcomparedwiththepolymericnatureofPIVCsbefore2000thatmayhavehadanimpactontheresults.Studieswereexcludediftheywereconductedinpaediatric(age<18years)settingsandiftherewasnodirectcomparisonbetweenbloodsamplesobtainedbyPIVCandvenepuncture.

2.4 | Search outcomes

The study selectionprocess resulted in855 studies being identifiedfromthesearchstrategy(Figure1).Basedoncomparingthetitleandabstractofthecitationagainsttheinclusioncriteria,16studieswereidentified as eligible (Barnard et al., 2016; Corbo, Fu, Silver, Atallah,& Bijur, 2007; Dietrich, 2014; Grant, 2003; Hambleton, Gomez, &BernabeuAndreu,2014;Himberger&Himberger,2001;Kelly&Klim,2013; Lowe et al., 2008; Munnix, Schellart, Gorissen, & Kleinveld,

F I G U R E 1  PRISMAflowchartofthestudyselectionandinclusionprocess

Records identified through database searching

(N = 851)

Scr

eeni

ngIn

clud

edE

ligib

ility

Iden

tific

atio

n Additional records identified through other sources

(N = 4)

Initial Literature Search(N = 855)

Possible relevant citations identifiedby literature search

(N = 716)

Studies retrieved for detailed examination

(N = 19)

Studies excluded after evaluation of title and

abstract(N = 697)

Records excluded as duplicates(N = 139)

Studies included in analysis

(N = 16)

Studies excluded

(N = 11)

Studies identified through reference

and citation checking

(N = 8)

4  |     COVENTRY ET al.

TAB

LE 1

 GRADEevidenceprofile–PeripheralIntravenousCannula(PIVC)comparedwithvenepuncturefordrawingblood

Cert

aint

y as

sess

men

tM

etho

d of

blo

od d

raw

Effe

ct

Cert

aint

yIm

port

ance

No

of

stud

ies

Stud

y de

sign

Risk

of b

ias

Inco

nsis

tenc

yIn

dire

ctne

ssIm

prec

isio

nO

ther

con

side

ratio

ns

Perip

hera

l In

trav

enou

s C

annu

la (P

IVC)

Vene

punc

ture

Rela

tive

(95%

CI)

Abs

olut

e (9

5% C

I)

HemolysisofbloodsamplesobtainedbyPIVCcomparedwithvenepuncture

12Observational

studies

Veryserious

aNotserious

Notserious

Notserious

Publicationbias

stronglysuspected

Verystrongassocia‐

tion

Allplausibleresidual

confoundingwould

reducethedemon‐

stratedeffect

b

5673/59032

(9.6%)

70/6091(1.1%)OR4.58

(3.61–

5.80)

39 m

ore

per

1,000

(from

29 m

ore

to52

more)

⨁⨁

◯◯

lo

wImportant

EquivalenceofbloodsamplesobtainedbyPIVCcomparedwithvenepuncture

5Observational

studies

Veryserious

cNotserious

Notserious

Notserious

Publicationbias

stronglysuspected

dTwostudiessummarizedtheresultsasoutsideofthe

laboratoryallowableerrorandoutsideBland–Altman

LOA,noneofthesevalueswouldhaverequiredclinical

intervention.Twostudiessummarizedtheresultsasall

parameterswerewithinthelaboratory'sacceptederror

orconsideredequivalentexceptforvenousbloodgases.

Onestudyfoundbloodsamplesforpotassium,bicarbo‐

nateandglucosewerenotclinicallyequivalent.

⨁◯

◯◯

ve

ry lo

wCritical

 (Continues)

     |  5COVENTRY ET al.

Cert

aint

y as

sess

men

tM

etho

d of

blo

od d

raw

Effe

ct

Cert

aint

yIm

port

ance

No

of

stud

ies

Stud

y de

sign

Risk

of b

ias

Inco

nsis

tenc

yIn

dire

ctne

ssIm

prec

isio

nO

ther

con

side

ratio

ns

Perip

hera

l In

trav

enou

s C

annu

la (P

IVC)

Vene

punc

ture

Rela

tive

(95%

CI)

Abs

olut

e (9

5% C

I)

ContaminationofbloodculturesobtainedbyPIVCcomparedwithvenepuncture

2Observational

studies

Serious

eVeryseriousfNotserious

Very

seriousg

Publicationbias

stronglysuspected

hOnestudyreportedbloodculturescouldbetakenac‐

curatelyfromaPIVCwhencomparedwithvenepunc‐

ture;incontrast,theotherstudyreportedtakingblood

culturesfromPIVCincreasestheriskofcontamination.

⨁◯

◯◯

ve

ry lo

wImportant

Abbreviations:CI:Confidenceinterval;OR:Oddsratio.

GRADEWorkingGroupgradesofevidence.

Highquality:Furtherresearchisveryunlikelytochangeourconfidenceintheestimateofeffect.

Moderatequality:Furtherresearchislikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandmaychangetheestimate.

Lowquality:Furtherresearchisverylikelytohaveanimportantimpactonourconfidenceintheestimateofeffectandislikelytochangetheestimate.

Verylowquality:Weareveryuncertainabouttheestimate.

a Therewasahighriskofbias:astheexposure(PIVC)couldhavebeenfromaPIVConinsertion,newlyinserted,oranexistingPIVC;theoutcome(haemolysis)wasmeasureddifferently–eithervisuallyor

automated;somestudiesdidnotcontrolforconfounding.

b TheFunnelplotisnotsymmetrical,suggestingthatpublicationbiasmaybeofconcern.

c Therewasahighriskofbiasas:theexposure(PIVC)couldhavebeenfromanewlyinsertedPIVC,oranexistingPIVC;theoutcome(equivalence)wasmeasureddifferentlyamongstudies.Somedefined

clinicalacceptableintervalothersusedBland–Altman95%or99%LevelofAgreement(LOA).SomestatedthateventhoughtheresultswereoutsidetheclinicalacceptableintervalortheBland–Altman

LOAthevalueswouldnothaverequiredclinicalintervention.

d WewereonlyabletocombinethreestudiesandthusunabletodoaFunnelPlot.TheGRADEHandbookrecommends,“Itisextremelydifficulttobeconfidentthatpublicationbiasisabsentandalmost

asdifficulttoplaceathresholdonwhentoratedownqualityofevidenceduetothestrongsuspicionofpublicationbias.ForthisreasonGRADEsuggestsratingdownqualityofevidenceforpublication

biasbyamaximumofonelevel”.

e Therewasahighriskofbiasas:theexposure(PIVC)couldbefromarecentlyinsertedPIVCoranexistingPIVC.

f Therewasseriousinconsistencyintheresultsofthetwostudies.

g Therewasseriousimprecisionconsideringthesmallnumberofstudiesandwideconfidenceintervals.

h WeconductedanarrativereviewandwereunabletodoaFunnelPlot.TheGRADEHandbookrecommends,“Itisextremelydifficulttobeconfidentthatpublicationbiasisabsentandalmostasdifficult

toplaceathresholdonwhentoratedownqualityofevidenceduetothestrongsuspicionofpublicationbias.ForthisreasonGRADEsuggestsratingdownqualityofevidenceforpublicationbiasbya

maximumofonelevel”.

TAB

LE 1

 (Continued)

6  |     COVENTRY ET al.

TAB

LE 2

 Resultsofqualityappraisal(MAStARI)

JBI c

ritic

al a

ppra

isal

che

cklis

t for

cro

ss‐s

ectio

nal s

tudi

es

Aut

hor

Crite

ria

clea

rly

defin

edSu

bjec

t and

set

‐tin

gs d

escr

ibed

Expo

sure

mea

sure

d in

va

lid a

nd re

liabl

e w

ayO

bjec

tive,

sta

nd‐

ard

crite

riaCo

nfou

ndin

g fa

c‐to

rs id

entif

ied

Stra

tegi

es fo

r con

‐fo

undi

ng fa

ctor

sO

utco

mes

mea

sure

d in

va

lid a

nd re

liabl

e w

ayA

ppro

pria

te s

ta‐

tistic

al a

naly

sis

Barnardetal.(2016)

Unclear

Yes

Unclear

Yes

Yes

Yes

Yes

Yes

Corboetal.(2007)

Yes

Yes

Yes

Yes

Unclear

Unclear

Yes

Yes

Dietrich(2014)

Unclear

Unclear

Yes

Yes

Yes

No

Yes

Yes

Grant(2003)

Yes

Yes

Yes

Yes

Yes

No

Yes

Yes

Loweetal.(2008)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Munnixetal.(2010)

Yes

Yes

Yes

Yes

Yes

Unclear

Yes

Unclear

Ongetal.(2008)

Yes

Unclear

Unclear

Yes

Yes

Yes

Unclear

Yes

Ortells‐Abuyeetal.

(2014)

Yes

Yes

Yes

Yes

Unclear

Unclear

Yes

Yes

Phelanetal.(2018)

Unclear

Yes

Unclear

Yes

Yes

Yes

Yes

Yes

Seemannand

Reinhardt(2000)

Yes

Yes

Yes

Yes

Yes

No

Yes

Yes

Wollowitzetal.

(2013)

Unclear

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Zlotowskietal.

(2001)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Hambletonetal.

(2014)

Yes

Yes

Yes

Yes

Unclear

Unclear

Yes

Yes

Himbergerand

Himberger(2001)

Yes

Yes

Yes

Yes

Unclear

Unclear

Yes

Yes

Selfetal.(2012)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

KellyandKlim(2013)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

     |  7COVENTRY ET al.

2010;Ong, Chan,& Lim, 2008;Ortells‐Abuye, Busquets‐Puigdevall,Díaz‐Bergara,Paguina‐Marcos,&Sánchez‐Pérez,2014;Phelanetal.,2018;Seemann&Reinhardt,2000;Selfetal.,2012;Wollowitz,Bijur,Esses,&Gallagher,2013;Zlotowski,Kupas,&Wood,2001).

2.5 | Quality appraisal

Studies selected for retrieval were assessed by two independentreviewers formethodological validity prior to inclusion in the re‐view.Weused thestandardizedJoannaBriggs Institute (JBI)criti‐cal appraisal instrument from the JBI Meta‐Analysis of StatisticsAssessment andReview Instrument (JBIMAStARI). Any disagree‐mentsthatarosebetweenthereviewerswereresolvedthroughdis‐cussion.Elevenstudieswereexcluded(Appendix3).

2.6 | Data abstraction

Datawere extracted from the included studies by two reviewerstocheckaccuracy.Thedataextractedincludeddetailsaboutstudyyear,studycountry,studyaim,studysetting,studydesign,interven‐tionsandcomparators.Datawereextractedseparatelyforstudiesinvestigatinghaemolysis,accuracyofbloodresultsandcontamina‐tionofbloodcultures.Dataincludedsampletype,samplesize,meth‐ods,resultsandauthorrecommendations.

2.7 | Synthesis

Meta‐analysis was conducted for studies examining haemolysis.Forrest plots were produced to display the effect measures ofeach study that were expressed as prevalence, odds ratio (OR)with95%confidenceintervals(CIs).TheORistheratiooftheoddsofhaemolysisoccurring inabloodsampleobtainedfromaPIVCcomparedwiththeoddsofhaemolysisoccurringinabloodsampleobtainedbyvenepuncture.Aratioofoneimpliesthehaemolysisof ablood sample is equally likely if obtainedbybothPIVCandvenepuncture, a ratio of greater than one implies haemolysis ismore likely inabloodsampleobtained fromPIVCanda ratioofless than one implies haemolysis is less likely if blood sample isobtainedbyPIVC.

Meta‐analysis was also conducted for three studies (Corboet al., 2007; Hambleton et al., 2014; Zlotowski et al., 2001) ex‐amining equivalence of blood results. We attempted to contactthe authors for raw data and were unsuccessful for two studies(Himberger&Himberger,2001;Ortells‐Abuyeetal.,2014).Foronestudy(Hambletonetal.,2014)weusedRevMancalculator(ReviewManager(RevMan),2014)toinputthestandarddeviationandcon‐ductstatisticalmeta‐analysis.Effectsizeswereexpressedaspooledmeandifferencesandtheir95%CI.Resultswerepooledusingfixedeffectsmodels.Heterogeneitymeasuresthevariabilityamongthecombinedstudiesandthechi‐squaretestandthe I2 statisticwereusedtoassessheterogeneity.Thepooledresultwasconsideredhet‐erogeneousifthe I2statisticwas>40%andthepvaluewas<0.05(Higgins&Green,2011).

Forsomestudiesassessingequivalenceofbloodresultsandcon‐taminationofbloodcultures,meta‐analysiscouldnotbeperformed,andthefindingshave,therefore,beenpresentedinanarrativeform.Tablesaredisplayedtoaidindatapresentationwhereverappropriate.

Publication bias may occur when studies with non‐significantfindingsarenotsubmittedbytheinvestigatororarerejectedbytheeditorsofthejournal(Gordis,2009).When10ormorestudieswerecombined,publicationbiaswasassessedusingfunnelplotsandin‐terpretedbyvisualinspection(Higgins&Green,2011).

TheoverallqualityoftheevidencewasassessedusingtheGradingof Recommendations Assessment, Development and Evaluation(GRADE)assessment(Guyattetal.,2008).AGRADEassessmentin‐cludesassessmentofriskofbias,inconsistencyofresults,indirectnessofevidence,imprecisionofresults,thelikelihoodofpublicationbias,themagnitudeoftheeffectandtheeffectofplausibleresidualcon‐founding.Theoverallqualityofthebodyoftheevidenceisthengradedashigh,moderate,loworverylow.Twoindependentreviewers(LCandHD)performedtheGRADEassessments,differenceswerediscussedandconsensusagreed(Table1).Anarrativesummaryofequivalenceofbloodresultsandcontaminationofbloodcultureswasconducted.

3  | RESULTS

3.1 | Characteristics of included studies

The16studieswerecriticallyappraised(Table2)formethodologicalqualityusingtheJBIcriticalappraisaltools.Theoverallmethodolog‐icalqualityoftheincludedstudieswasgenerallypoor.Differencesamongthestudies included ifbloodsampleswereobtainedon in‐sertion,fromanewlyinserted,oranexistingPIVC.Theoutcomeofhaemolysiscouldhavebeenmeasuredeitherbyvisualinspectionorbyautomatedspectrometry.Confoundingfactorswerenotalwaysidentified and strategies to account for confounding factorswerenotalwaysincluded.

Theaimsofthestudiescanbesummarizedasfirstlyto:examineblood sample haemolysis rates between blood samples drawn viavenepuncturecomparedwithPIVC(Barnardetal.,2016;Corboetal.,2007;Dietrich,2014;Grant,2003;Loweetal.,2008;Munnixetal.,2010;Ongetal.,2008;Ortells‐Abuyeetal.,2014;Phelanetal.,2018;Seemann&Reinhardt,2000;Wollowitzetal.,2013;Zlotowskietal.,2001).Secondly,toexamineequivalenceofbloodtestresultsbetweenbloodsamplesdrawnviaPIVCcomparedwithvenepunc‐ture (Corbo et al., 2007; Hambleton et al., 2014; Himberger &Himberger,2001;Ortells‐Abuyeetal.,2014;Zlotowskietal.,2001).Thirdly, to examine blood culture contamination between bloodsamplesdrawnviavenepuncturecomparedwithPIVC(Kelly&Klim,2013;Selfetal.,2012).

Meta‐analysiswasconductedforthestudiesexamininghaemo‐lysis.Forthestudiesassessingequivalence,meta‐analysiswascon‐ductedforthreestudies(Corboetal.,2007;Hambletonetal.,2014;Zlotowskietal.,2001).Datacouldnotbeaggregatedfortwostud‐ies (Himberger &Himberger, 2001;Ortells‐Abuye et al., 2014) ofequivalenceandthestudiesexaminingbloodculturecontamination.

8  |     COVENTRY ET al.

TA B L E 3  Summaryofcharacteristicsofincludedstudies

AuthorCountry Setting Data Collection Sample type Sample size Methods

Barnardetal.(2016) UK

UniversityteachinghospitalEmergencydepartment

Prospective ConveniencesampleCollectedover3months

Bloodsamples(N=844)Bloodsample:‐ Venepuncture(N=257)

‐ PIVC(N=587)

Corboetal.(2007) USA

Urbantertiaryhospi‐talLevel1TraumaCenterAdultemergencydepartment.

ProspectiveObservationalCase–Control

ConveniencesampleCollectedover2monthsInclusion‐ ExistingPIVCsalinelock

Patients(N=81)Usablepatientsamples(N=73)

ConcurrentsamplesExisting PIVC‐ Infusionshalted2minpriortotourniquet

‐ Tourniquetproximaltointravenousline

‐ Alcoholwipe‐ 5mldiscard‐ Vacutainerusedtoaspiratebloodsample

Venepuncture‐ 21Gbutterflyneedle‐ Vacutaineradaptor

Dietrich(2014) USA

188‐bedlevelIIITraumaCentreEmergencyDepartment

ProspectiveObservational

ConveniencesampleCollectedover4‐monthperiod

Bloodsamples(N=8,944)Bloodsample:‐ OnPIVCinsertion(N=3,803)

‐ Venepuncture(N=3,301)

‐ ExistingIVcatheter(N=1,840)

Grant(2003) USA

Metropolitanteach‐inghospitalEmergencyDepartment

ProspectiveObservational

ConveniencesampleCollectedover19days

Bloodsamples(N=454)Bloodsample:‐ OnPIVCinsertion(N=255)

‐ Venepuncture(N=117)

‐ ExistingIVcatheter(N=82)

Loweetal.(2008) USA

450‐bedLevelIItraumacentreCommunityteach‐inghospitalEmergencyDepartment

ProspectiveObservational

Non‐consecutivesampleCollectedover55days

Bloodsamples(N=853)Bloodsample:‐ OnPIVCinsertion(N=498)

‐ Venepuncture(N=355)

Munnixetal.(2010) Netherlands

HospitalEmergencydepart‐ment&Outpatientclinic

ProspectiveObservational

ConveniencesampleCollectedover3months

EDPatients(N=100)OutPatients(N=50)BloodSamples(N=600)Bloodsampledrawn:‐ OnPIVCinsertion(N=400)

‐ Venepuncture(N=200)

ConsecutivepatientspecimensFourconsecutivesampleswerecollectedforeverypatient

Ongetal.(2008) Singapore

HospitalEmergencyDepartment

ProspectiveObservational

Conveniencesample Bloodsamples(N=227)Bloodsampledrawn:‐ PIVC(N=168)‐ Venepuncture(N=59)

(Continues)

     |  9COVENTRY ET al.

AuthorCountry Setting Data Collection Sample type Sample size Methods

Ortells‐Abuyeetal.(2014)Spain

Reference100‐bedhospitalInpatientwardandShortStayUnit

Cross‐sectionalstudySimplecrossoverdesign

Collectedover8monthsInclusion‐WithaPIVCExclusion‐ PIVCcollectiontime>20s

‐ Difficultvenoclysis‐ Arterio‐venousfistula‐ Languagedifficulties‐ Criticalcondition‐ Alteredstateofconsciousness

Patients(N=272) ConcurrentsamplesRandomizedcollectionsequence

Existing PIVC‐ IVfluidstoppedfor15s

‐ Aspiratedanddis‐carded4mlofblood

‐ Removedbloodsample‐FlushedPIVCwith4mlofsaline

Venepuncture‐ Oppositearm‐ 21‐gaugeneedle‐ 10mlsyringe

Phelanetal.(2018) USA

UrbantertiarycarehospitalEmergencydepartment

RetrospectiveObservational

AllED‐obtainedsamplesinwhichpotassiumanalysiswascompletedCollectedover12months

Bloodsamples(54,531)Bloodsample:‐ PIVC(47,266)‐ Venepuncture(615)

SeemannandReinhardt(2000)USA

Medium‐sizedcomprehensivehealthcarefacilityInpatientmedical

ward

ProspectiveObservationalCase–Control

ConveniencesampleInclusion‐ Nocoagulopathiesorsepsis

Bloodsamples(N=34)Bloodsample:‐ ExistingPIVC(N=17)‐ Venepuncture(N=17)

Wollowitzetal.(2013)USA

UrbanacademictertiaryhospitalAdultemergencydepartment.

ProspectiveObservationalCross‐Sectional

ConveniencesampleCollectedover40days

Bloodsamples(N=4,513)Bloodsample:‐ ExistingPIVC(N=3,727

‐ Venepunctureusingabutterflyneedle(N=786)

Existing PIVC‐ClosedIVcathetersys‐tem‐dual‐port,attachedtoaBDvacutainerleurlockand8.5mlBDvacutainertube

Venepuncture‐ Butterflyneedlecollectionset(pushbuttonwith21‐or23‐gaugebutterflyneedles)

Zlotowskietal.(2001)USA

TertiaryteachinghospitalEmergencyDepartment

ProspectiveObservationalCase–Control

Inclusion‐ Healthyvolunteers.

Samplesize(N=32)Bloodsamples(N=96)

Newly inserted PIVC‐ PIVCinsertedintoupperextremity

‐ 200mlbolusofNS administeredover10min

‐ 2minwaittime‐ Tourniquetapplied‐ 18‐gaugeneedleattachedtoa20mlsyringeaspirated12mlofblood

‐ Asecondaspirateof12mlwassimilarlyaspirated

Venepuncture‐ 21‐gaugebutterfly

needle‐ Vacutainer

TA B L E 3   (Continued)

(Continues)

10  |     COVENTRY ET al.

AuthorCountry Setting Data Collection Sample type Sample size Methods

Hambletonetal.(2014)Spain

UniversityhospitalEmergencydepartment

ProspectiveObservationalCase–Control

ConsecutiveenrolmentCollectedover7monthsExclusionPatientswith‐ anaemia‐ vasculardisease‐ coagulopathy‐ receivinganticoagulation

‐ immunocompromised‐ difficultvenousaccess

Pairedbloodsamples(N=259)

ConcurrentsamplesExisting Double lumen

PIVC‐ Infusionshalted2min‐ Flushedbothlumenswith1mlsaline

‐ 2minlateratourniquetwasapplied

‐ Alcoholwipe‐ 2mldiscarded‐ Vacutainerwasusedtoaspiratebloodsample

Venepuncture‐ Oppositearm‐ 21‐gaugebutterfly

needle

HimbergerandHimberger(2001)USA

MilitaryteachinghospitalRegionalLevel1TraumaCentreEmergencydepartment

ProspectiveObservationalCase–Control

ConveniencesampleCollectedover10monthsInclusion‐ Adults‐ EnglishSpeaking‐ ReceivingIVhydration

‐ NoThrombophlebitis‐ Haemodynamicallystable

‐ SBP>90mmHg‐ Capableofconsent.

Patients(N=64)Bloodsamples(N=559)

ConcurrentsamplesExisting PIVC‐ IVpaused30s‐ Tourniquetapplied‐ 30secondspause‐ 5mldiscarded‐ IVtubenotdiscon‐nectedfromhub

‐ 10mlsyringewithan18‐gaugeneedleaspi‐ratedthebloodsample

‐ 10mlsalineflushafterVenepuncture‐ Oppositearm‐ 20‐gaugeneedle

KellyandKlim(2013)Australia

CommunityTeachinghospitalEmergencydepartment

ProspectiveObservational

Collectedover7‐monthperiod

Inclusion‐ Requiredabloodculture

‐ PIVCrecentlyplaced(<1hr)

Exclusion‐ PIVCplacedbyparamedic

Samplesize(N=472) Hospitalpolicyonsterility,skincleans‐ingandbloodculturebottlepreparationwasfollowed.

Selfetal.(2012) USA

TeachingHospitalAdultemergencydepartment.

Matchedhistori‐calcohort

Collectedover12‐monthperiod

Samplesize(N=505)matchedcultures

Existing PIVC‐ Skinantisepsiswith2%chlorhexidine/70%isopropylalcoholpriortoPIVCplacement

‐ Antisepsisofthecatheterwith70%isopropyl

‐ DrawingbloodthroughthePIVC

Venepuncture‐ Skinantisepsiswith2%chlorhexidine/70%ispropylalcohol

‐ Withdrawbloodfromthevein

TA B L E 3   (Continued)

     |  11COVENTRY ET al.

Therefore, a narrative review is presented, asmeta‐analysis couldnotbeperformed.

StudieswereconductedintheUSA(Corboetal.,2007;Dietrich,2014;Grant,2003;Himberger&Himberger,2001;Loweetal.,2008;Phelanetal.,2018;Seemann&Reinhardt,2000;Selfetal.,2012;Wollowitzetal.,2013),Europe(Barnardetal.,2016;Hambletonetal.,2014;Munnixetal.,2010;Ortells‐Abuyeetal.,2014),Australia(Kelly&Klim,2013)andSingapore(Ongetal.,2008).Mostofthestudieswereprospective(Barnardetal.,2016;Corboetal.,2007;Dietrich,2014;Grant,2003;Hambletonetal.,2014;Himberger&Himberger, 2001;Kelly&Klim, 2013; Lowe et al., 2008;Munnix etal.,2010;Ongetal.,2008;Ortells‐Abuyeetal.,2014;Seemann&Reinhardt,2000;Wollowitzetal.,2013;Zlotowskietal.,2001)andretrospectiveinnature(Phelanetal.,2018;Selfetal.,2012).Manystudiesusedthesamegroupofpatients,thatis,onegroupofpa‐tientshadbloodsamplesfrombothPIVCandvenepuncture(Corboetal.,2007;Hambletonetal.,2014;Himberger&Himberger,2001;Ortells‐Abuyeetal.,2014;Seemann&Reinhardt,2000;Selfetal.,2012;Zlotowskietal.,2001).Otherstudiesusedseparategroupsofpatientsforbloodsamples,thatis,onegroupofpatientsbloodwassampled fromaPIVCandaseparategroupofpatientshadbloodsampled by venepuncture (Barnard et al., 2016; Dietrich, 2014;Grant,2003;Kelly&Klim,2013;Loweetal.,2008;Munnixetal.,2010;Ongetal.,2008;Phelanetal.,2018;Wollowitzetal.,2013).

Most studies were conducted in an emergency department(Barnardetal.,2016;Corboetal.,2007;Dietrich,2014;Grant,2003;Hambletonetal.,2014;Himberger&Himberger,2001;Kelly&Klim,2013;Loweetal.,2008;Munnixetal.,2010;Ongetal.,2008;Phelanetal.,2018;Selfetal.,2012;Wollowitzetal.,2013;Zlotowskietal.,2001).Onestudywasconductedinaninpatientwardandshortstayunit (Ortells‐Abuye et al., 2014) and one study in amedicalward(Seemann&Reinhardt,2000).

Conveniencesampling(Barnardetal.,2016;Corboetal.,2007;Dietrich,2014;Grant,2003;Himberger&Himberger,2001;Kelly&

Klim,2013;Loweetal.,2008;Munnixetal.,2010;Ongetal.,2008;Phelanetal.,2018;Seemann&Reinhardt,2000;Selfetal.,2012;Wollowitzetal.,2013;Zlotowskietal.,2001)wascommonwiththreestudiesusingconsecutivesampling(Hambletonetal.,2014;Munnixetal.,2010;Ortells‐Abuyeetal.,2014).Samplesizesvariedsignifi‐cantlywith thenumberof patients beingbetween17 and54,531anddatacollectionperiodsvaryingbetween19daysand12months.A fewstudiesexcludedpatientswhowereunstableorwithmulti‐plecomorbidities(Hambletonetal.,2014;Himberger&Himberger,2001;Ortells‐Abuyeetal.,2014;Seemann&Reinhardt,2000)andonestudyonlyincludedhealthyvolunteers(Zlotowskietal.,2001).

Many studies clearly articulated protocols for collecting bloodsamples (Corbo et al., 2007;Hambleton et al., 2014;Himberger&Himberger, 2001; Kelly & Klim, 2013; Ortells‐Abuye et al., 2014;Seemann&Reinhardt,2000;Selfetal.,2012;Wollowitzetal.,2013;Zlotowski et al., 2001) and others did not. Most studies sampledbloodfromexistingPIVCs(Corboetal.,2007;Hambletonetal.,2014;Himberger&Himberger,2001;Ortells‐Abuyeetal.,2014;Seemann&Reinhardt,2000;Selfetal.,2012;Wollowitzetal.,2013);andafewstudiessampledbloodonPIVCinsertion(Loweetal.,2008;Munnixet al., 2010).Two studies (Dietrich, 2014; Grant, 2003) comparedbloodsampledfrombothexistingPIVCsandonPIVCinsertion;andtwostudies(Kelly&Klim,2013;Zlotowskietal.,2001)statedbloodwas sampled from newly inserted PIVC (Table 3).The results havebeenpresentedaccordingtostudiesinvestigatinghaemolysis,equiv‐alenceofbloodresultsandcontaminationofbloodcultures.

3.2 | Haemolysis

TheratesofhaemolysisfrombloodsamplesobtainedbetweenPIVCandvenepuncturewasreportedin10studies(Figure2).Meta‐analy‐sisfoundthattheoddsratioofhaemolysiswere4.58(CI,3.61–5.80)timesmorelikelyinbloodsamplesobtainedviaPIVCcomparedwithvenepuncture. There was evidence of both clinical and statistical

F I G U R E 2  ForestplotofstudiesusingORincomparinghaemolysisinbloodsamplestakenviaPIVCcomparedwithvenepuncture.CI,confidenceinterval;M‐H,Mantel‐Haenszel;OR,oddsratio[Colourfigurecanbeviewedatwileyonlinelibrary.com]

CI, confidence interval; M-H, Mantel-Haenszel; OR, odds ratio

Total (95%Cl)Total eventsHeterogenity: Chi2 = 33.96, df = 10 (p = 0.0002); I2 = 71%

567359032 6091 100.0% 4.58 [3.61, 5.80]

0.01Favours [PIVC] Favours [venepuncture]

0.1 1 10 100

70

Test for overall effect: z = 12.57 (p < 0.00001)

Study or SubgroupPIVC

Events EventsTotal Total WeightOdds Ratio

M-H, Fixed,95% ClOdds Ratio

M-H, Fixed,95% ClVenepuncture

Barnard 2016Corbo 2007Dietrich 2014Grant 2003Lowe 2008Munnix 2011Ong 2008Ortells-Abuye 2012Phelan 2018Seeman & Reinhart 2000Wollowitz 2013Zlotowski 2001

840

555828264110

48214

5442

58781

5643337470400168272

4726617

372764

70311040

330

210

25781

3301117354200

59272615

17786

32

7.6%

3.4%1.1%1.0%0.6%4.1%0.4%

53.6%0.3%

27.1%0.6%

5.96 [2.72, 13.09]Not estimable

10.82 [3.38, 34.61]24.11 [3.30, 176.17]22.36 [3.03, 165.16]28.38 [1.72, 4.68.06]

4.44 [1.52, 13.00]21.80 [1.27, 373.91]

2.00 [1.41, 2.85]11.67 [0.58, 235.92]

6.23 [4.00, 9.70]2.60 [0.12, 55.77]

12  |     COVENTRY ET al.

heterogeneity(chi‐square=33.96,p = 0.0002; I2=71%)andassuchresults must be interpretedwith caution. Sensitivity analysis wasconductedonfivestudiesthatfollowedaprotocolforwithdrawingbloodfromaPIVC.Thefindingsweresimilar(OR6.46;95%CI,4.21–9.91).Therewasnoevidenceofheterogeneity (chi‐square=1.22,p=0.75;I2=0%).

Haemolysis was measured by either visual techniques (Grant,2003; Lowe et al., 2008; Seemann & Reinhardt, 2000), automatedtechniques (Barnardetal.,2016;Corboetal.,2007;Dietrich,2014;Munnixetal.,2010;Phelanetal.,2018;Wollowitzetal.,2013),orthemeasurementtechniquewasnotreported(Ongetal.,2008;Ortells‐Abuye et al., 2014; Zlotowski et al., 2001). Blood sample rejection

TA B L E 4  Haemolysisassessmentmethods,rejectionratesandauthorsrecommendations

Author Haemolysis Assessment Haemolysis sample rejection (haemolysis rate) Authors recommendations

Barnard etal.(2016)

HaemolysismeasuredbyspectrophotometryHaemolysisdefinedas≥30µmol/lserumHb

Totalsamplerejections:92/844(10.9%)‐ PIVC:84/587(14.3%)‐ Venepuncture:7/257(2.7%) (OR5.63;95%CI,2.49–12.73) Sub‐analyses Side of patient

‐ Right:57/450(12.7%)‐ Left:34/394(8.6)(OR0.68;95%CI,0.42–1.10)Anatomical site‐ Antecubitalfossa:50/637(7.8%)‐ Distaltoantecubitalfossa:41/207(19.8%Significant(OR2.25;95%CI,1.40–3.63)Difficulty of sampling‐ Veryeasy(comparedto):24/393(6.1%)‐ Easy:24/266(9.0%)(OR1.29;95%CI,0.69–2.35)‐ Average:20/106(18.9%)(OR2.95;95%CI,1.51–5.77)‐ Difficult/verydifficult:23/79(29.1)(OR4.36;95%CI,2.04–9.32)

Estimated tourniquet time‐ <1min(comparedto):29/416(7.0%)‐ 1–2min:49/390(12.6%)(OR1.28;95%CI,0.76–2.16)‐ >2min:13/38(34.2%)(OR2.15;95%CI,0.82–5.65)

• WherepracticableallbloodsamplesshouldbeobtainedviavenepunctureratherthanaPIVC

• Theoveralleconomicimpactofseparatingvenepunctureandinser‐tionofPIVCiscomplexandrequiresfurtherevaluation.

Corboetal.(2007)

Notreported. NohaemolysedsamplesNocomplicationsduringaspirationofPIVC

• AspiratingbloodviaPIVCisanac‐ceptablemethodofobtainingbloodsamples

Dietrich(2014)

HaemolysismeasuredbyspectrophotometrySamplesclassifiedas:Usable–haemolysis<200mg/dlRejected–haemolysis>200mg/dlAcceptablerateofsamplerejectionforhaemolysiswasdefinedas2%asperbenchmarkbestpracticebytheAmericanSocietyofClinicalPathology

Totalsamplerejections:58/8,944(0.65%) Samplerejection:‐ PIVCinsertion:41/3,803(1.1%)‐ Venepuncture:3/3,301(0.1%)‐ ExistingIVcatheters:14/1,840(0.8%)

• Measurehaemolysisusingstandard‐izedspectrophotometricmeasure‐mentratherthancolourcharts

• Levelsofhaemolysisrequiredforrejectionshouldbestandardized

• Actualcostsofdelayedlaboratoryresultsshouldbemeasuredagainsttheactualcostsofperformingad‐ditionalvenepuncturesinallpatientswhoalreadyhaveIVaccessestab‐lishedbutinwhomnoadditionalvenepunctureisnecessary.

Grant(2003)

Visual Totalsamplerejections:59/454(13%)Samplerejection:‐ EDPIVCinsertion:50/255(20%)‐ Venepuncture:1/117(<1%)‐ ExistingIVcatheters:8/82(10%) (20%vs.<1%,p<0.001)

Sub‐analysesED PIVC insertion withdrawal methodSamplerejection:‐ Vacutainer:44/195(23%)‐ Syringe:5/60(9%) (22%vs.9%,p=0.02)

• DrawbloodinEDPIVCinsertionusingasyringeinsteadofavacu‐tainerandthentransferbloodtoatubeviatheneedlessconnector

(Continues)

     |  13COVENTRY ET al.

rates for haemolysis varied between collection methods: from ve‐nepuncture between 0‐6.8%; from newly inserted PIVC between0‐20%;fromexistingPIVCbetween0.8‐24.4%;andfromstudiesthatfollowed a protocol between 0‐5.6%. Two studies (Dietrich, 2014;Loweetal.,2008)reportedthattheacceptablerateofsamplerejec‐tionforhaemolysiswasdefinedbya2%benchmarkbestpracticesetbytheAmericanSocietyofClinicalPathology.

Afewstudies(Barnardetal.,2016;Grant,2003;Loweetal.,2008;Munnixetal.,2010;Ongetal.,2008;Phelanetal.,2018;Wollowitzetal.,2013)conductedsub‐analyses;however,inonestudy(Munnixetal.,2010)nostatisticalanalysiswasperformedmakingitdifficulttoascertainthesignificanceoffindings.Twostudies(Grant,2003;Ongetal.,2008)foundthattheuseofavacutainercomparedwithsyringe resulted in higher PIVC haemolysis rates and one study

(Phelanet al., 2018) foundnodifferences. Three studies (Barnardetal.,2016;Phelanetal.,2018;Wollowitzetal.,2013)foundblooddrawnfromtheantecubitalfossawerelesslikelytobehaemolysedwhencomparedwithblooddrawn fromother sites, incontrast toanotherstudy(Loweetal.,2008)whofoundnodifferencesrelatedtoblooddrawsite.Twostudies(Phelanetal.,2018;Wollowitzetal.,2013)foundthattheuseoflargergaugeneedleswerelesslikelytohavehaemolysed samples comparedwith a smaller gaugeneedle,incontrasttoanotherstudybyOngetal.,2008whofoundnodif‐ferences related toneedlegauge size.The samestudy (Wollowitzetal.,2013)alsofoundthatthebloodsamplesweremorelikelytobehaemolysed if thebloodcollectiontubewas less thanhalf full.Twostudies(Phelanetal.,2018;Wollowitzetal.,2013)foundifthetourniquet timewasgreater than1minbloodsamplesweremore

Author Haemolysis Assessment Haemolysis sample rejection (haemolysis rate) Authors recommendations

Loweetal.(2008)

HaemolysisassessedvisuallyAcceptablerateofsamplerejectionforhaemolysiswasdefinedas2%asperbenchmarkbestpracticebytheAmericanSocietyofClinicalPathology

Totalsamplerejections:29/853(3.4%)Samplerejection:‐ EDPIVCinsertion:28/470(5.6%)‐ Venepuncture:1/354(<1%) (5.6%vs.<0.3%,p=0.001)

Sub‐analysesSamplerejectionbysiteVenepuncture; PIVCAntecubital1/309(<1%);4/135(2.9%)Forearm0/18;7/140(5%)Hand0/22;12/99(12%)Multi0/1;0/0Wrist0/2;5/92(5.4%)Nosignificantdifferences

• Venepunctureshouldbethestandardofcarefordrawingbloodsampleswiththeexceptionofhigh‐acuitypatientsandpatientswhohavedifficultvenousaccess.

Munnixetal.(2010)

Haemolysismeasuredbyspectrophotometry

Totalsamplerejections16/100(16%) Samplerejection‐ PIVC:16/100(16%)‐ Venepuncture:0/50(0%)Sub‐analysis

PIVC‐ 1sttube:16/100(16%)‐ 2ndtube:4/100(4%)‐ 3rdtube:4/100(4%)‐ 4thtube:2/100(2%)Difficult PIVC placement‐ No:6/77(8%)‐ Yes:10/23(44%)

Size of needle‐ 18Gauge:5/34(15%)‐ 20Gauge:11/65(17%)Blood collection‐ Needlewithpre‐attachedholder:10/86(12%)‐ Directdrawadaptor:6/12(50%)Site of blood draw‐ Leftantecubital:4/31(13%)‐ Rightantecubital:1/23(4%)‐ Leftforearm:4/20(20%)‐ Rightforearm:2/17(12%)‐ Lefthand:2/3(67%)‐ Righthand:3/5(60%)Nostatisticalanalysesreported

• Thenumberofhaemolysedspeci‐menssenttothelaboratorycanbesignificantlyreducedbyeliminationofthefirsttubeofblood.

TA B L E 4   (Continued)

(Continues)

14  |     COVENTRY ET al.

likelytobehaemolysedwithonestudy(Barnardetal.,2016)findingnodifferences.Twostudies (Barnardetal.,2016;Wollowitzetal.,2013)foundbloodsamplesweremorelikelytobehaemolysedifthevenepuncturewasdifficult(Table4).

3.3 | Equivalence of blood tests

Meta‐analysiswasconductedforthreestudies(Corboetal.,2007;Hambletonetal.,2014;Zlotowskietal.,2001) thatcompared the

Author Haemolysis Assessment Haemolysis sample rejection (haemolysis rate) Authors recommendations

Ongetal.(2008)

Haemolysisassessedusingvalidatedmethodsinabiochemistrylaboratory.

Totalsamplerejections:45/227(19.8%) Samplerejection‐ PIVC:41/168(24.4%)‐ Venepuncture:4/59(6.8%)Univariableanalysis:(OR4.4;95%CI,1.5–13.0)Univariable Sub analysis‐ Syringe:16/146(11%)‐ Vacutainer:29/81(35.8%) (OR4.5;95%CI,2.3–9.0)

Size of needle‐ ≤21G:15/86(17.4%)‐ >21G:30/141(21.3%)NotsignificantOperator‐ Registrar:2/18(11.1%)‐ Medicalofficer:22/137(16.1%)‐ Consultant:4/18(22.2%)‐ Student/nurse:17/54(31.5%)NotsignificantMultivariable analysis‐ Useofavacutainerwasassociatedwithasignificantlyhigherratesofhaemolysis(adjustedOR,6.0;95%CI,2.3–15.1)

• Drawingbloodwithavacutainerhadincreasedratesofhaemolysis

• Ifasyringeisusedtodrawblood,whetherfromIVcannulaorvenepuncture,aneedlessmethodshouldbeusedforsampletransfer.

Ortells‐Abuyeetal.(2014)

Samplerejection:‐ Venepuncture0/272(0%)‐ PIVC10/272(3.7%)

• Bloodsamplesobtainedbyve‐nepunctureandPIVCcanbeusedroutinelyformostroutinelaboratorytests

Phelanetal.(2018)

Haemolysismeasuredbyspectrophotometry Haemolysis>300serumHb=grosslyhaemolysedandsamplerejected Haemolysis>80≤300serumHb=haemolysedwithcomment

Totalsamplerejections: Combined(haemolysedwithcommentandgrosshaemolysis):5,439/54,531(10%)‐ PIVC:4,821/47,266(10.2%)‐ Venepuncture:33/615(5.4%) Significant

Sub‐analysis:PIVCSite‐ Antecubital:2,117/28,786(7.4%)‐ Peripheral:2,622/17,960(14.6%)‐ SignificantSyringe/ Vacutainer‐ Syringe:92/705(13.0%)‐ Vacutainer:1,825/16,590(11.0%)NotsignificantSize of needle‐ 16–20G:3,882/44,571(9.3%)‐ Other:939/5,633(16.7%)SignificantTourniquet time‐ >1min:532/3,832(13.9%)‐ <1min:1,362/13,162(10.3%)Significant

• Bloodsamplesobtainedbyve‐nepunctureandintheantecubitallocationareassociatedwithreducedhaemolysis

• ForbloodsamplesobtainedbyPIVCshortertourniquettimesandlargergaugeneedleareassociatedwithlowerhaemolysis

Seemannand Reinhardt(2000)

Haemolysisassessedvisually

Totalsamplerejections:4/34(11.8%) Samplerejection:‐ ExistingPIVC4/17(23.5%)‐ Venepuncture0/17(0%)

• PIVCisavalidmethodofproducingviablebloodsamples

TA B L E 4   (Continued)

(Continues)

     |  15COVENTRY ET al.

Author Haemolysis Assessment Haemolysis sample rejection (haemolysis rate) Authors recommendations

Wollowitzetal.(2013)

Haemolysisassessedbymeasurementoffreeserumhaemoglobinlevels

Totalsamplerejections564/4513(12.5%) Samplerejections:‐ PIVC544/3727(14.6%)‐ Venepuncture21/786(2.7%)Sub‐analysisSite of blood draw‐ Antecubitalfossa:306/3160(9.7%)‐ Other:260/1353(19.2%)Needle/catheter gauge‐ 14–1829/373(7.8%)‐ 20406/2922(13.9%)‐ 216/322(1.9%)‐ 2315/464(3.2%)Fullness of collection tubes‐ <halffull147/639(23%)‐ ≥halffull418/3874(10.8%)Tourniquet time‐ >1min214/1221(17.5%)‐ ≤1min352/3,292(10.7%)Difficulty of venepuncture‐ Difficult224/954(23.5%)‐ Notdifficult341/3559(9.6%)

• ThemosteffectivestrategytoreducetherateofhaemolysisintheEDistousebutterflyneedlesforphlebotomyratherthanIVcatheters.

Zlotowskietal.(2001)

Haemolysisfromvenepuncture0/32(0%) HaemolysisfromPIVC2/64(3.1%)

• SupportsuseofbloodsamplesobtainedfromPIVC

TA B L E 4   (Continued)

TA B L E 5  PooledmeanandpooledmeandifferencebetweenbloodtestsobtainedbyPIVCcomparedwithvenepuncture

Studies Lab testNumber of blood tests

Pooled PIVC Mean

Pooled Venepuncture Mean

Pooled Mean Difference [95% CI]

Heterogeneity

Chi p‐value I2

Corboetal.(2007);Hambletonetal.(2014);Zlotowskietal.(2001)

Sodium(mEq/L)

728 139.3 139.2 −0.10[−0.13,0.32] 0.4 0.8 0%

Corboetal.(2007);Hambletonetal.(2014);Zlotowskietal.(2001)

Potassium(mEq/L)

728 3.9 3.9 −0.01[−0.02,0.01] 15.5 <0.001 87%

Corboetal.(2007);Hambletonetal.(2014);Zlotowskietal.(2001)

Chloride(mEq/L)

728 105.2 104.9 0.32[0.09,0.5]) 0.58 0.75 0%

Hambletonetal.(2014);Zlotowskietal.(2001)

Bicarbonate(mmol/L)

582 26.2 26.8 −0.6[−0.8,−0.4] 0.59 0.44 0%

Corboetal.(2007);Hambletonetal.(2014);Zlotowskietal.(2001)

Glucose(mg/dl) 728 116.6 116.4 0.6[−0.4,1.6] 0.88 0.64 0%

Hambletonetal.(2014);Zlotowskietal.(2001)

Albumin(g/dl) 582 3.6 3.6 −0.06[−0.17,−0.05] 0.75 0.39 0%

Corboetal.(2007);Hambletonetal.(2014)

Troponin(µg/L) 664 0.002 0.0017 0.00[−0.00,0.00] 0.79 0.37 0%

Hambletonetal.(2014);Zlotowskietal.(2001)

Hemoglobin(g/dl)

582 12.7 12.8 −0.1[−0.13,−0.07] 0.07 0.8 0%

Corboetal.(2007);Zlotowskietal.(2001)

Hematocrit(%) 210 38.5 38.7 −0.26[−1.31.0.79] 0.19 0.66 0%

Hambletonetal.(2014);Zlotowskietal.(2001)

Platelets(K/µl) 582 208.6 211.0 −2.4[−3.48,−1.32] 0.0 0.98 0%

Hambletonetal.(2014);Zlotowskietal.(2001)

INR 582 1.1 1.2 −0.01[−0.02,0.00] 0.15 0.7 0%

16  |     COVENTRY ET al.

TAB

LE 6

 Equivalenceofbloodresults

Aut

hor

Equi

vale

nce

Aut

hor

reco

mm

enda

tions

Aut

hor r

ecom

men

datio

ns

Corboetal.

(2007)

1. D

iffer

ence

bet

wee

n La

b te

sts a

nd n

umbe

r exc

eedi

ng C

LIA

stan

dard

s and

Bla

nd‐A

ltman

LO

A•Eventhough43/584(7.4%)couldnot

beexplainedbyallowableerrorforCLIA

and35/584(6.0%)felloutsideBland‐

AltmanLOAnoneofthesevalueswould

haverequiredclinicalintervention

•Samplesforhemocrit,electrolytes,

serumglucose,andcardiacenzymescan

betakenfromaPIVCsalinelockdevice

instablepatientsinED

LabTest

Venepuncture

Mean(+

SD)

PIVC

Mean(+

SD)

Mea

n difference

Numberexceed

National

Reco

mm

ende

d CLIARange

a N(%)

Numberexceed

Bland‐Altman95%

LOA1

SD

; N(%)

Na+(m

Eq/L

)140.1(2.9)

140.4(3.0)

−0.36

0(0)

6(8.2)

K+(m

Eq/L

)4.1(2.9)

4.1(0.4)

−0.04

4(5.5%)

2(2.7%)

CO2 (

mEq

/L)

25.5(3.9)

25.5(3.6)

−0.08

15(20.5%)

4(5.5%)

Cl‐

(mEq

/L)

106.1(3.5)

106.3(3.5)

−0.23

05(6.8%)

Glucose(m

g/dl

)115(57.0)

115.9(58.1)

−0.60

7(9.5%)

6(8.2%)

CPK(U

/L)

137.9(223.9)

142.9(224.6)

−5.019

4(5.5%)

2(2.7%)

TroponinI

(µg/

L)0.004(0.01)

0.006(0.016)

−0.0016

12(16.4%)

6(8.2%)

HCT

(%)

38.5(3.9)

38.4(4.0)

0.11

1(1.3%)

4(5.5%)

Total

43(7.4%)

35(6.0%)

2. A

spira

tion

succ

ess r

ates

Cathetersize

Numberofsuccess/

numberofattempts

(95%CI)

186/6

100(54.1–100)

2063/69

91.3(82–96.7)

224/6

66.7(22.3–95.7)

Total

73/8

190.1(81.5–95.6)

(Continues)

     |  17COVENTRY ET al.

Aut

hor

Equi

vale

nce

Aut

hor

reco

mm

enda

tions

Aut

hor r

ecom

men

datio

ns

Hambleton

etal.

(2014)

1. D

iffer

ence

bet

wee

n La

b te

sts a

nd n

umbe

r exc

eedi

ng la

bora

tory

acc

epte

d sy

stem

atic

err

or a

nd B

land

‐Altm

an L

OA

•CollectingbloodsthroughPIVCisvalid

whenanalyzingformostcommonly

studiedbloodparametersinED.

•Allparametersshoweddifferences

belowthelaboratory'sacceptedsystem‐

aticerrorexceptforvenousbloodgases.

•Remainsvalidregardlessofthetypeof

druginfusionsadministered.

•Theminimumdiscardofbloodistwice

thedeadspacevolumeofthedevice.

•EDnursesshouldconsiderusingPIVCas

afirstoptionforpatients’blooddraws.

LabTest

Venepuncture

Mean(+

SD)

PIVC

Mean(+

SD)

Meandiff

Differencesbelow

thelaboratory

acceptedsystem‐

aticerror

b

Numberexceed

Bland‐Altman95%

LOA

2

SD; N(%)

Glucose

120.

812

1.4

0.66

Yes

8(3.5)

Urea

41.9

41.8

−0.05

Yes

16(7.7)

Creatinine

1.09

1.07

−0.01

Yes

7(3.0)

Na+

138.

4138.5

0.11

Yes

9(3.9)

K+3.

93.

9−0.05

Yes

11(5.1)

Cl‐

104.

9105.2

0.33

Yes

9(6.34)

Ca2+

8.8

8.7

−0.08

Yes

14(6.51)

Albumin

3.5

3.5

−0.05

Yes

15(6.64)

Amylase

51.9

52.4

−0.54

Yes

4(1.96)

CreatinKinase

152.97

157.01

4.04

Yes

1(0.5)

Bilir

ubin

0.7

0.6

−0.01

Yes

10(5.6)

pH7.

47.

40.

01Yes

4(1.78)

pCO

243

.842

.01

−1.78

No

15(6.52)

pO2

38.6

45.6

7.01

No

20(9.1)

HCO

326.6

25.97

−0.63

No

16(6.9)

Troponin

1.1

1.1

0.04

Yes

1(2.0)

Osmolality

292.

829

2.2

−0.58

Yes

2(2.6)

Leucocytes

9.1

8.99

−0.07

Yes

15(6.2)

RedBloodCells

4.3

4.3

−0.03

Yes

9(3.8)

Hemoglobin

12.7

12.6

−0.11

Yes

17(7.0)

Platlets

209.

3206.9

−2.34

Yes

16(6.6)

aPTT

33.1

32.4

−0.63

Yes

20(8.7)

INR

1.2

1.2

0.01

Yes

6(2.6)

TAB

LE 6

 (Continued)

(Continues)

18  |     COVENTRY ET al.

Aut

hor

Equi

vale

nce

Aut

hor

reco

mm

enda

tions

Aut

hor r

ecom

men

datio

ns

Himberger

and

Himberger,

(2001)

1. D

iffer

ence

bet

wee

n La

b te

sts a

nd n

umbe

r exc

eedi

ng la

bora

tory

clin

ical

acc

epte

d in

tern

val a

nd B

land

‐Altm

an L

OA

•Bloodscanbecollectedfromvenepunc‐

tureandPIVCinterchangeably

•Eventhoughafewvaluesexceededthe

LOAthesevalueswouldnotresultin

clinicaltreatment

•Strictadherencetoprocedureprotocols

isnecessary

•Eachspecimenshouldbemonitoredfor

haemolysis

LabTest

Venepuncture

Mean(+

SD)

PIVC

Mean(+

SD)

Mea

n difference

Numberexceed

National

Reco

mm

ende

d CLIARange

a N(%)

Numberexceed

Bland‐Altman95%

LOA1

SD

; N(%)

WBCx109 /L

Notreported

Notreported

−0.02

0/46(0%)

0(0%)

RBCx10

6 /µl

0.09

0/46(0%)

0(0%)

Hct

−1.5

0/46(0%)

2(4.4%)

Hgbg/dl

0.1

2/46(4.4%)

3(6.5%)

Pltx10

9

−0.2

1/46(2.2%)3

(6.5%)

SodiummEq/L

0.4

1/47(2.1%)

3(6.4%)

PotassiummEq/L

0.05

2/47(4.3%)

3(6.4%)

ChloridemEq/L

0.3

1/47(2.1%)

3(6.4%)

C0 2m

Eq/L

0.15

2/47(4.3%)

2(4.3%)

Glucosemg/dl

−2.44

3/47(6.4%)

2(4.3%)

Creatininemg/dl

0.17

1/47(2.1%)

2(4.3%)

SUNmg/dl

0.1

1/47(2.1%)

1(2.1%)

2.Overallsuccessrate

58/64(90.7%)ofaspirat‐

ingbloodfromaPIVC

3.Therewerenoreported

complicationswithIVsite

withanystudypar‐

ticipantsindicatingthis

isasafeandeffective

methodforobtaining

bloodspecimens

TAB

LE 6

 (Continued)

(Continues)

     |  19COVENTRY ET al.

Aut

hor

Equi

vale

nce

Aut

hor

reco

mm

enda

tions

Aut

hor r

ecom

men

datio

ns

Ortells‐

Abuye

etal.(2014)

1.DifferencebetweenLabtestsandnumberexceedinglaboratoryclinicalacceptedinternvalandBland‐

AltmanLOA

•VenepunctureandPIVCcanbeconsid‐

eredequivalentformostroutinelabora‐

torytestsbutnotpCO2andpO2

•Non‐equivalenceofpCO2andpO2may

beduetohandlingandtransferofblood

sampletobloodgassyringe

•IfpatientshaveexistingPIVC,usingthat

forblooddrawsispreferable

•PIVCcouldbeusedforblooddraws

inpatientswhoarebleedingorhave

infectiousdiseaseandrequiremultiple

requestsforhaemograms

LabTest

Venepuncture

Mean(+

SD)

PIVC

Mean(+

SD)

Mea

n difference

Numberexceed

National

Reco

mm

ende

d CLIARange

c N(%)

Numberexceed

Bland‐Altman95%

LOA1

SD

; N(%)

AmylaseU/L

Notreported

Notreported

Not

reported

1/265(0.4%)

1/265(0.4%)

Calciummg/dl

2/266(0.8%)

7/266(2.6)%

Totalcholesterol

12/269(4.5%)

6/269(2.2%)

Creatininemg/dl

4/271(1.5%)

4/271(1.5%)

CreatininekinaseU/L

13/262(5.0%)

9/262(3.4%)

Basalglucosemg/dl

17/271(6.3%)

9/272(3.3%)

Aspartateaminotrans‐

ferase(SGOT)U/L

7/269(2.6%)

11/269(4.1%)

PotassiummEq/L

19/269(7.1%)

13/269(4.8%)

SodiummEq/L

4/271(1.5%)

12/271(4.4%)

Ureamg/dl

9/269(3.3%)

7/269(2.6%)

Redbloodcells106/µl

3/268(1.1%)

12/268(4.5%)

Haemoglobing/dl

9/268(3.4%)

8/268(3.0%)

Leucocytes103 /µl

4/268(1.5%)

5/268(1.9%)

Platelets103/µl

3/267(1.1%)

7/267(2.6%)

Prothrombinratio(%)

22/269(8.2%)

5/269(1.9%)

VenousCO

2potentialpH

2/260(0.8%)

11/260(4.2%)

VenousCO

2partialpres‐

surepCO

2mmHg

55/260(21.2%)

10/260(3.8%)

VenousO

2partialpressure

pO2mmHg

190/260(73.1%)

10/260(3.8%)

TAB

LE 6

 (Continued)

(Continues)

20  |     COVENTRY ET al.

Aut

hor

Equi

vale

nce

Aut

hor

reco

mm

enda

tions

Aut

hor r

ecom

men

datio

ns

Zlotowski

etal.(2001)

1.MeandifferencebetweenLabtests,Bland‐Altman99%LOA,Clinicalsignificantvalueandequivalence

•BloodsamplesforCBC;bloodurea;

nitrogen;creatinine;liverfunctionand

PT/INRweredeterminedclinically

equivalentwhenfromPIVC.

•Bloodsamplesforpotassium,bicarbo‐

nateandglucosewerenotdetermined

clinicallyequivalentwhenfromPIVC.

LabTest

Venepuncture

Mean(+

SD)

PIVC

Mean(+

SD)

Mea

n difference

Numberexceed

National

Reco

mm

ende

d CLIARange

d N(%)

Numberexceed

Bland‐Altman95%

LOA1

SD

; N(%)

Nammol/L

143.6(1.6)

143.5(1.8)

0.09

−2.55–2.74

5.0

Kmmol/L

3.8(0.22)

3.6(0.2)

0.17

−0.25–0.60

0.3

Cl m

mol

/L101.9(2.1)

102.6(2.2)

−0.69

−3.50–2.12

8.0

HCO

3 mm

ol/L

28.7(2.0)

27.7(2.0)

0.97

−3.14–5.08

3.0

Glucose

83.1(10.0)

81.4(10.4)

1.72

−16.0–19.43

5.0

BUNmg/dl

13.7(3.3)

13.1(3.6)

0.53

−2.24–3.31

5.0

Creatininemg/dl

0.9(0.1)

0.9(0.2)

0.01

−0.13–0.16

0.3

Albuming/dl

4.4(0.2)

4.3(0.3)

0.10

−0.18–0.38

0.4

TotalBilimg/dl

0.4(0.2)

0.4(0.2)

−0.01

−0.15–0.14

0.4

DirectBilimg/dl

0.1(0.03)

0.12(0.06)

−0.01

−0.11–0.09

0.2

ALPU/L

63.8(17.8)

62.2(17.7)

1.53

−5.14–8.20

20.0

ALTU/L

21.0(13.8)

21.2(13.4)

−0.25

−6.32–5.82

15.0

ASTU/L

27.9(13.0)

27.1(13.3)

0.78

–9.7

4–11

.31

15.0

WBCK/µl

5.5(1.4)

5.6(1.4)

−0.01

−0.70–0.68

1.0

Hbg/dl

13.7(1.5)

13.5(1.6)

0.24

−0.47–0.94

1.0

Hct(%)

39.4(3.7)

38.8(3.8)

0.67

−1.33–2.66

4.0

PltK/µl

224.9(47.2)

222.2(46.7)

2.69

−12.98–18.35

30.0

PT(s)

13.1(1.5)

13.2(0.5)

−0.07

−0.67–0.54

1.5

INR

1.0(0.07)

1.0(0.07)

−0.01

−0.10–0.08

0.2

2.Overallsuccessrate32/32(100.0%)ofaspiratingbloodfromaPIVC

a CLIA,‘ClinicallaboratoryImprovementAmendments’areasetofregulationsthataresetoutbytheCentreforDiseaseControlandFoodandDrugAdministrationintheUnitedStates,andareusedto

provideindustrystandardsforlaboratorytestingquality.

b LaboratoryAcceptedSystematicError,UsedtheInternationalOrganizationforStandardization.ISO15189:2007:Medicallaboratories—particularrequirementsforqualityandcompetence.Available

fromtheISOwebpagehttps://www.iso.org/standard/42641.html.

c ClinicallyAcceptedIntervalwasdefinedbythestudyinvestigatorstobetheminimalclinicallysignificantdifferencebetweenvenepunctureandPIVC.

d ClinicallySignificantValuewasbasedonthemedianvaluesoftheexpertopinionof5emergencystaffphysicians.

TAB

LE 6

 (Continued)

     |  21COVENTRY ET al.

equivalenceofbloodtestsbetweenPIVCandvenepuncture(Table5).Therewerenosignificantmeandifferencesinmostbloodtestswiththeexceptionofplateletsandbicarbonate(meanvalueswerelowerin the PIVC group compared with the venepuncture group) andchloride(meanvaluewashigher inthePIVCgroupcomparedwiththe venepuncture group). Statistical heterogeneity was not pre‐sent in any pooled analyses except potassium,where the I2 value was87%.This result showed the substantial heterogeneitywhichmustbe interpretedwithcareasthere isconsiderablevariation inthecombinedorpooledresultsanditmaybemisleadingtoreportacombinedsummarymeasure.Twostudies(Himberger&Himberger,2001;Ortells‐Abuyeetal.,2014)wereunable tobecombined formeta‐analysisandthefollowingdataareanarrativesynthesisofthefindingsofallfivestudiesreportingbloodtestequivalence.

Itisworthnotingthat,studiesdefinedtheclinicallyacceptedinter‐valdifferently;twostudies(Corboetal.,2007;Himberger&Himberger,2001)usedtheClinicalLaboratoryImprovementAmendments(CLIA),thatareasetofregulationssetoutbytheCentreforDiseaseControlandtheFoodandDrugAdministration,thatofferindustrystandardsforlaboratorytestingquality.Onestudy(Hambletonetal.,2014)usedtheLaboratoryAcceptedSystematicError;inanotherstudy(Ortells‐Abuye et al., 2014), the investigators defined the clinically accept‐ableinterval;andinthelaststudy(Zlotowskietal.,2001);anexpertpaneloffiveemergencyphysiciansdefinedtheclinicallyacceptableinterval.Similarly,fourstudies(Corboetal.,2007;Hambletonetal.,2014;Himberger&Himberger,2001;Ortells‐Abuyeetal.,2014)usedBland–Altman95%levelofagreement(LOA)andonestudy(Zlotowskietal.,2001)usedBland–Altman99%LOA.

Twostudies(Corboetal.,2007;Himberger&Himberger,2001)summarized the resultsasnot requiringclinical intervention,even

though some values were outside the laboratory allowable errorandwereoutsideBland–AltmanLOA.Onestudy(Hambletonetal.,2014)summarizedtheresultsasallparameterswerewithinthelab‐oratory's accepted error except for venous blood gases. Similarly,another study (Ortells‐Abuye et al., 2014) also summarized bloodresults,whichcouldbeconsideredequivalentwiththeexceptionofvenousbloodgases.Incontrast,onestudy(Zlotowskietal.,2001)foundbloodsamplesforpotassium,bicarbonateandglucosewerenotclinicallyequivalent.

In addition, three studies (Corbo et al., 2007;Himberger&Himberger, 2001; Zlotowski et al., 2001) reported that the as‐piration of PIVC success rates were between 90% and 100%;with one study (Corbo et al., 2007) further analysing aspira‐tion success for 18‐, 20‐ and 22‐gauge needles (100%, 91.3%,66.7% respectively). Another study (Hambleton et al., 2014)reportedbloodsamplesfromPIVCswithandwithout infusionsand venepuncture were similar; and one study (Himberger &Himberger,2001)reportednocomplicationswiththePIVCwithanyofthestudyparticipantsandconcludedwithdrawingbloodfromaPIVCwassafeandeffectivemethodofobtainingbloodsamples(Table6).

3.4 | Contamination of blood cultures

Twostudies(Kelly&Klim,2013;Selfetal.,2012)examinedtherateof contaminationofblood cultures if theblood samplewas takenfromaPIVCcomparedwithvenepuncture(Table7).Onestudy(Kelly& Klim, 2013) reported blood cultures could be taken accuratelyfromaPIVCwithin1hrofPIVCinsertionwhencomparedwithve‐nepuncture.Incontrast,theotherstudy(Selfetal.,2012)reported

TA B L E 7  Contaminationofbloodcultures

Author

Results

Author recommendationsBlood cultures

KellyandKlim(2013) 1.Numberofpositivecultures:N=65/472,(13.8%)

2.Numberoftruepositivecultures:N=49/65,(75.4%)

3.Numberoffalsepositivecultures:N=16/65,(24.6%)Falsepositiveviavenepuncture:N=8/224(3.6%) FalsepositiveviaPIVC:N=8/248(3.2%)

4.OddsratioforcontaminatedculturesinPIVC:(OR,0.9;95%CI,0.33–2.44)

• BloodculturescanbeaccuratelytakenfromaPIVCwithin1hrofinsertioninanEDwheninfectioncontrolproce‐duresarefollowed.

Selfetal.(2012) 1. Overall2.PIVCcontaminated:33/505(6.5%)3.Venepuncturecontaminated:18/505(3.6%)

4.RelativeriskofcontaminationPIVCcomparedwithvenepuncture(RR1.83;95%CI,1.08–3.11)UseofPIVCcomparedwithvenepunctureresultedin2.97(95%CI,0.29–7.51)additionalcontaminatedculturesper100culturescollected

• ThisstudysuggeststhatcollectingbloodculturesfromPIVCsincreasestheriskofcontaminationcomparedwithvenepuncture

22  |     COVENTRY ET al.

takingbloodculturesfromPIVCincreasestheriskofcontaminationandfalsepositiveresultscomparedwithvenepuncture.

3.5 | Publication bias

Afunnelplotwasusedtoassesspublicationbiasforthestudiesonhaemolysis. The plot is not symmetrical, suggesting that publica‐tionbiasmaybeofconcern.Figure3displaysthefunnelplotforthepooledORofhaemolysis.

4  | DISCUSSION

Thisreviewsynthesizedthestudiesontheeffectofobtainingbloodsamples fromaPIVCcomparedwithvenepuncture.Sixteenstudiesmettheinclusioncriteria,with12studiesexamininghaemolysisrates,fivestudiesexaminingequivalenceofbloodtestresultsandtwostud‐iesexaminingcontaminationratesofbloodcultures.WedidnotfindanystudythatinvestigatedriskofbloodstreaminfectionandrisktothepatencyofthecannulaafterwithdrawingbloodsamplesfromthePIVC.Majorfindingsofthisreviewsuggestthathaemolysisratesarehigher inbloodsampledfromaPIVCcomparedwithvenepuncture.Withregardtoequivalenceofbloodtestresults,eventhoughsomere‐sultswereoutsidethelaboratoryallowableerrorandwereoutsidetheBland–AltmanLOA,noneofthesevalueswouldhaverequiredclinicalintervention.Somestudiesdidnotconsidervenousbloodgaseswereequivalentandasinglestudyfoundbloodsamplesforpotassium,bi‐carbonateandglucosewerenotclinicallyequivalent.Withregardtocontaminationratesofbloodcultures,theresultswereequivocal.

In this systematic review,we highlighted variations in draw‐ingbloodfromaPIVC(oninsertion,newlyinserted,oranexistingPIVC),inhowtheoutcomeofhaemolysiswasmeasured(visuallyorautomated)andsomestudiesdidnotcontrolforconfounding(e.g.

vacutainervs.syringe,needlegauge,siteofblooddrawnetc.).Thevisualmethodofdetectinghaemolysisissubjectiveanddependson the individual's visual acuity and colour perception (Dietrich,2014). The outcome of equivalence was measured differentlyamong the studies (e.g. clinical acceptable intervals and Bland–Altmanplots).Thesevariationscertainly impede the strengthofrecommendationsthatcanbedrawnacrossstudies.Nonetheless,there was sufficient homogeneity to allowmeta‐analysis of thestudiesofhaemolysis.

Meta‐analysis foundtheoddsofhaemolysiswere4.58timesmore likely in blood samples obtained via PIVC compared withvenepuncture.Thisfinding issimilartoothersystematicreviews(Heyeretal.,2012;McCaugheyetal.,2017).Inourstudy,haemo‐lysisratesforbloodobtainedviavenepuncturewerelowandlessthan2.7%innineof10studies.Interestingly,thehaemolysisratesforbloodobtainedviaPIVCvariedgreatlyalsobetween0%and24.4%,withfivestudies(Corboetal.,2007;Dietrich,2014;Loweetal.,2008;Ortells‐Abuyeetal.,2014;Zlotowskietal.,2001)thatfollowed a protocol forwithdrawingblood reportinghaemolysisratesbetween0‐5.6%.Eventhoughoursensitivityanalysiscon‐ductedon the five studies that followedaprotocolwere similar(OR6.46)wecontendhaemolysisrateslessthan5%areapproach‐ingtheAmericanSocietyofClinicalPathologybenchmarkof2%.Acceptinghaemolysisratesoflessthan5%inpatientsknowntobeadifficultvenepunctureorwhorequiremultipleblooddrawsmaybeconsideredapragmaticoption. Inaddition,onestudy (Grant,2003)thatreportedahighhaemolysisrate(20%)implementedaclinicalpracticechangeandencouragedphlebotomiststosamplebloodwithasyringeinsteadofavacutainerandthentransferthebloodtoatubeviaaneedlelessconnector.Audits followingthispracticechangeshowedhaemolysisrateshaddecreasedbetween4‐5%.Other variables thatmay be important regarding haemo‐lysisratesincludesiteoftheblooddrawn,theneedlegauge,the

F I G U R E 3  FunnelplotforthepooledORofhaemolysis.Abbreviations:SE(log[OR]):StandardError(logarithm[OddsRatio])[Colourfigurecanbeviewedatwileyonlinelibrary.com]

Abbreviations: SE(log[OR]): Standard Error (logarithm[Odds Ratio])

Phelan 2018 Wollowitz 2013

Barnard 2016

Ong 2008 Dietrich 2014

Zlotowski 2001Seaman 2000

Lowe 2008

Grant 2003

Munnix 2011

Ortells-Abuye 2012

OR

1001010.10.012

1.5

1

0.5

0SE(log[OR])

     |  23COVENTRY ET al.

fullnessofthecollectiontube,tourniquetuseandifthevenepunc‐turewasconsidereddifficult.

Mostofthestudiesconsideredbloodsamplesfromvenepunc‐tureandPIVCwereequivalent.Irrespectiveofthelaboratoryclin‐icallyacceptederrororBland–Altmananalyses it seems logical toevaluateequivalencewithwhetherthedifferenceintestswouldre‐quireclinicalintervention.Non‐equivalenceofvenousbloodgaseshasbeensuggestedduetohandlingerror.Inthat,contactwithairmaycausechangesinbloodresults.Thebloodsampleneedstobetransferredfromasyringetoabloodgassyringe,thebloodgassy‐ringeneeds tobe filledwith thecorrectamountofbloodandex‐cessairneedstoberemoved.Thestudy(Zlotowskietal.,2001)thatreportednon‐equivalence forpotassium,bicarbonateandglucosesuggestedthismayberelatedtohaemodilution,astheycomparedtheresultsafteradministeringanormalsalinesolutionbolus.

We only found two studies that evaluated contamination ofblood cultures between venepuncture and PIVC.One study sup‐portedobtainingbloodculturesfromPIVCandtheotherstudydidnot.Consideringanothermeta‐analysis(Snyderetal.,2012)evaluat‐ingvenepuncturewithintravenouscathetersrecommendedagainstobtainingbloodfromanintravenouscatheterduetoincreasedcon‐taminationrates,wealsosupportthisrecommendation.Thismeta‐analysis(Snyderetal.,2012)wasdifferenttooursinthatitincludedintravenouscatheterscomprisingofcentrallines,arteriallinesandportacathetersandincludedstudieswithpaediatricpatients.

4.1 | Limitations

This reviewhas some limitations.Somestudiesexaminingequiva‐lenceofbloodtestresultswereexcludedastheirdataanalysesre‐portedpairedttestsandcorrelationcoefficients.Itwasdetermineda priori the most appropriate analyses were the Bland–Altmanmethod(Bland&Altman,1986).ThisreviewwaslimitedtoEnglishlanguage studies, a limitation that may also introduce bias. EventhoughwefollowedtheMeta‐analysisofObservationalStudies inEpidemiology guidelines (Stroup et al., 2000) there remains somesubjectivityinconsensusagreementforratingstudyqualityforin‐clusionandgradingtheoverallstrengthoftheevidence.

The range of settings in the reviewed studies has implicationsforclinicalandstatisticalheterogeneitywithsystematicreviewsandmeta‐analysesbutenhancesgeneralizability.Theresultsofthisreviewhavegeneralizabilitylimitedtoadultpatientsinacutecareandemer‐gencysettings.Limitationsoutsidethecontrolofthereviewauthorsincluded:allthestudieswerefromsingleinstitutions;somestudieshadsmallsamplesizes;manystudiesdidnotincludeunstablepatients;andmostofthelaboratoryresultsanalysedfellinsidethenormalrange.Inaddition,awidevarietyofpracticeswereobservedfordrawingbloodfromaPIVCandnotallstudiescontrolledforconfoundingvariables.

4.2 | Recommendations for practice

The resultsof this reviewcanhelpguide clinical practice in sev‐eral ways. This systematic review showed that five studies with

haemolysisrates lessthan5%usedaprotocoltowithdrawbloodfromaPIVCandonestudyhadlowerratesofhaemolysisafterim‐plementingaprotocoltowithdrawbloodfromaPIVC.SomeofthesuggestionsflowingonfromthisreviewuntilsupportedbyfurtherresearchsuggestthataPIVCprotocolshouldinclude:strictaseptictechnique;haltinfusionofsolutionforatleast2minpriortoblooddraw;usea20‐gaugeorlargercatheter;andthequantityofbloodtobediscardedshouldbedouble thedeadspace.Other sugges‐tionsincludedusinganeedlelessconnectortodrawbloodfromthePIVC,thusreducingtheopportunityofapotentialneedlestickin‐jury,useasyringetoaspiratethebloodnotavacutainerandavoidexcessiveaspirationforceanddonotunder‐fillthebloodtubes.

Hospitals shouldalsobeencouraged toaudithaemolysis ratesregularlyintheirdepartments,notonlytoincreasestaffawareness,but also to potentially implement clinical practice change to de‐creasehaemolysisratesifrequired.

4.3 | Recommendations for research

Large randomized controlledmultisite trials are required to defini‐tivelycompareeffectivenessofPIVCblooddrawscomparedwithve‐nepuncture.Aclusterdesignisrecommendedtoinvestigatetheeffectofablooddrawprotocol.Theclusterdesignwillmanagetheriskofcontaminationoftheblooddrawprotocolbetweentheinterventionandcontrolgroup.Allstudiesneedtoclearlyarticulate ifthebloodwassampledfromthePIVCon‐insertion,newlyinsertedorfromanexistingPIVC.ThestudiesneedtoevaluateifdrawingbloodfromaPIVCinfluencesprematurecannulafailure,causephlebitis,leadingtoblood‐streaminfectionsandeconomicanalysesshouldbeconducted.

Morestudiesarerequiredthatanalyseabnormallaboratoryval‐ues,thatis,valuesoutsidethenormalrange.Analysisofequivalenceofbloodtestresultsshouldbereportedusingclinicalacceptablelab‐oratoryerror,Bland–Altmanplotsandmoreimportantlywouldtheresultofchangedclinicaltreatment.

Further research is requiredto investigate ifdrawingbloodfromaPIVCisofbenefitforspecificpatientpopulationsandinothersettingsbesidestheemergencydepartment.Someexam‐plesincludepatientswhoareknowntobeadifficultvenepunc‐ture; who have limited venous access; require multiple blooddraws;whoareobese,dehydratedoroedematous;andpatientsonanticoagulation therapywhoareat increased riskofbleed‐ing.Moreover,therehasbeenarecentsinglestudy(Mulloy,Lee,Gregas,Hoffman,&Ashley,2018)intoadevicethatattachestothe PIVC and threads a sterile catheter through the PIVC intothe vein allowing needle‐free blood draws. This study shouldbe replicated indifferentpatientpopulationsandaneconomicanalysisconducted.

5  | CONCLUSION

Hospitalizedpatientsoftenrequiremultiplebloodteststoassistindiagnosis andmanagement of their conditions. Findings from this

24  |     COVENTRY ET al.

reviewsuggestbloodsamples forPIVCcomparedwithvenepunc‐turehavehigherhaemolysisrates;however,someindividualstudiesdemonstrated that if aprotocolwas followed, these ratesmaybelower. Blood test results may be considered equivalent as differ‐ences in resultswouldnotaffectclinical treatmentandbloodcul‐turesshouldnotbetakenfromPIVC.Furthermore,drawingbloodfromPIVCsmaybethebestavailableoptioninsomepatientgroups,however,furtherresearchisrequiredtoinformtheevidenceforbestpracticerecommendations.

FUNDING S TATEMENT

Thisresearchdidnotreceiveanyspecificgrantfromfundingagen‐ciesinthepublic,commercialornot‐for‐profitsectors.

CONFLIC T OF INTERE S T

SK and LC aremembers of theAVATAR group. AVATAR researchis supported by competitive government, university, hospital andprofessional organization research grants as well as industry un‐restricted donations, investigator initiated research/educationalgrants and occasional consultancy payments from the followingcompanies: 3M, Adhezion, Angiodynamics, Bard, Baxter, BBraun,BD, Carefusion, Centurion, Cook, Entrotech, Flomedical, Hospira,Mayo,Medtronic,ResQDevices,Smiths,Teleflex,Vygon.Thismanu‐scriptisindependentlypreparedandreflectsnocommercialentitynorpromotesparticularproductsunlessthesearesupportedbyre‐searchdata.Noconflictofinteresthasbeendeclaredbytheotherauthors.

AUTHOR CONTRIBUTIONS

LC, AJ, HD, LS, SK, EJ: Made substantial contributions to con‐ception anddesign, or acquisitionof data, or analysis and inter‐pretationofdata;LC,AJ,HD,LS,SK,EJ:Involvedindraftingthemanuscriptor revising itcritically for important intellectualcon‐tent; LC,AJ,HD, LS, SK,EJ:Given final approvalof theversiontobepublished.Eachauthorshouldhaveparticipatedsufficientlyintheworktotakepublicresponsibilityforappropriateportionsofthecontent;LC,AJ,HD,LS,SK,EJ:Agreedtobeaccountablefor all aspectsof thework in ensuring thatquestions related totheaccuracyorintegrityofanypartoftheworkareappropriatelyinvestigatedandresolved.

ORCID

Linda L. Coventry https://orcid.org/0000‐0002‐3598‐9942

Alycia M. Jacob https://orcid.org/0000‐0003‐2458‐6715

Hugh T. Davies https://orcid.org/0000‐0002‐0867‐2288

Samantha Keogh https://orcid.org/0000‐0002‐2797‐4388

Elisabeth R. Jacob https://orcid.org/0000‐0002‐3506‐8422

R E FE R E N C E S

Alexandrou, E., Ray‐Barruel, G., Carr, P. J., Frost, S. A., Inwood, S.,Higgins,N.,…Rickard,C.M.;OMGStudyGroup.(2018).Useofshortperipheral intravenouscatheters:Characteristics,managementandoutcomes worldwide. Hospital Medicine, 13(5), E1–E7, https://doi.org/10.12788/jhm.3039

Barnard,E.B.G.,Potter,D.L.,Ayling,R.M.,Higginson,I.,Bailey,A.G.,&Smith,J.E. (2016).Factorsaffectingbloodsamplehaemolysis:Across‐sectional study. European Journal of Emergency Medicine, 23,143–146.https://doi.org/10.1097/mej.0000000000000195

Bland, J.M.,&Altman,D.G. (1986). Statisticalmethods for assessingagreementbetweentwomethodsofclinicalmeasurement.Lancet,1,307–310.https://doi.org/10.1016/S0140‐6736(86)90837‐8

Buowari, O. (2013). Complications of venepuncture. Advances in Bioscience and Biotechnology,4, 126–128. https://doi.org/10.4236/abb.2013.41A018

ClinicalExcellenceCommission. (2013).Guideline for PIVC insertion and post insertion care in adult patients. (GL2013_013). Sydney, NSW:NSWGovernmentHealth.Retrievedfromhttp://www1.health.nsw.gov.au/pds/ActivePDSDocuments/GL2013_013.pdf.

Corbo,J.,Fu,L.,Silver,M.,Atallah,H.,&Bijur,P.(2007).Comparisonoflaboratoryvaluesobtainedbyphlebotomyversussalinelockdevices.Academic Emergency Medicine,14,23–27.https://doi.org/10.1197/j.aem.2006.06.053

Dietrich,H. (2014).One poke or two:Can intravenous catheters pro‐videanacceptablebloodsample?Adatasetpresentation,reviewofpreviousdatasetsanddiscussion.Journal of Emergency Nursing,40,575–578.https://doi.org/10.1016/j.jen.2012.11.002

Gordis,L.(2009).Epidemiology(4thed).Philidelphia:SaundersElsevier.Gorski,L.,Hadaway,L.,Hagle,M.E.,McGoldrick,M.,Orr,M.,&Doellman,

D. (2016). Infusion Therapy Standards of Practice (Vol.39 Supp 1S).Norwood,MA:InfusionNursesSociety.

Government of Western Australia Department of Health. (2017).Insertion and management of peripheral intravenous cannulae in Western Australian Healthcare Facilities Policy. Western Australia:Department of Health. Retrieved from http://www.health.wa.gov.au/circularsnew/pdfs/13345.pdf.

Grant,M.S.(2003).TheeffectofblooddrawingtechniquesandequipmentonthehemolysisofEDlaboratorybloodsamples.Journal of Emergency Nursing,29(2),116–121.https://doi.org/10.1067/men.2003.66

Guyatt,G.H.,Oxman,A.D.,Vist,G.E.,Kunz,R.,Falck‐Ytter,Y.,Alonso‐Coello,P.,&Schünemann,H.J.(2008).GRADE:Anemergingconsen‐susonratingqualityofevidenceandstrengthofrecommendations.BMJ,336,924–926.https://doi.org/10.1136/bmj.39489.470347.AD

Hambleton, V. L., Gomez, I. A., & Bernabeu Andreu, F. A. (2014).Venipuncture versus peripheral catheter: Do infusions alter labo‐ratoryresults?Journal of Emergency Nursing,40,20–26.https://doi.org/10.1016/j.jen.2012.03.014

Heyer,N.J.,Derzon,J.H.,Winges,L.,Shaw,C.,Mass,D.,Snyder,S.R.,…Liebow,E.B.(2012).Effectivenessofpracticestoreducebloodsam‐plehemolysisinEDs:Alaboratorymedicinebestpracticessystem‐aticreviewandmeta‐analysis.Clinical Biochemistry,45,1012–1032.https://doi.org/10.1016/j.clinbiochem.2012.08.002

Higgins,J.P.T.,&Green,S.(Eds).(2011).Cochrane handbook for systmatic reviews of interventions version 5.1.0[updatedMarch2011].Retrievedfromhttp://handbook.cochrane.org.

Himberger,J.R.,&Himberger,L.C.(2001).Issuesinaccuracyoflabora‐toryresults.Accuracyofdrawingbloodthroughinfusingintravenouslines.Heart & Lung: The Journal of Acute and Critical Care,30,66–73.https://doi.org/10.1067/mhl.2001.110535

Kelly,A.‐M.,&Klim,S.(2013).Takingbloodculturesfromanewlyestab‐lished intravenous catheter in the emergency department does notincreasetherateofcontaminatedbloodcultures.Emergency Medicine Australasia,25,435–438.https://doi.org/10.1111/1742‐6723.12121

     |  25COVENTRY ET al.

Lowe, G., Stike, R., Pollack, M., Bosley, J., O'Brien, P., Hake, A., …Stover, T. (2008). Nursing blood specimen collection techniquesandhemolysis rates in anemergencydepartment:Analysisof ve‐nipuncture versus intravenous catheter collection techniques.Journal of Emergency Nursing,34, 26–32. https://doi.org/10.1016/ j.jen.2007.02.006

McCaughey,E.J.,Vecellio,E.,Lake,R.,Li,L.,Burnett,L.,Chesher,D.,…Georgiou,A.(2017).Keyfactorsinfluencingtheincidenceofhemoly‐sis:Acriticalappraisalofcurrentevidence.Critical Reviews in Clinical Laboratory Sciences, 54, 59–72. https://doi.org/10.1080/10408363.2016.1250247

Mulloy,D.F.,Lee,S.M.,Gregas,M.,Hoffman,K.,&Ashley,S.W.(2018).EffectofperipheralIVbasedbloodcollectiononcatheterdwelltime,bloodcollectionandpatientresponse.Applied Nursing Research,40,76–79.https://doi.org/10.1016/j.apnr.2017.12.006

Munnix, I.C.A., Schellart,M.,Gorissen,C.,&Kleinveld,H.A. (2010).Factors reducinghemolysis rates inbloodsamples from theemer‐gency department. Clinical Chemistry and Laboratory Medicine, 49,157–158.https://doi.org/10.1515/CCLM.2011.012

Ong,M.E.H.,Chan,Y.H.,&Lim,C.S. (2008).Observationalstudytodetermine factors associatedwithblood samplehaemolysis in theemergency department.Annals Academy of Medicine Singapore,37,745–748.

Ortells‐Abuye, N., Busquets‐Puigdevall, T., Díaz‐Bergara,M., Paguina‐Marcos,M.,&Sánchez‐Pérez, I. (2014).A cross‐sectional study tocomparetwobloodcollectionmethods:Directvenouspunctureandperipheralvenouscatheter.British Medical Journal Open,4,e004250.https://doi.org/10.1136/bmjopen‐2013‐004250

Phelan,M.P.,Reineks,E.Z.,Schold,J.D.,Hustey,F.M.,Chamberlin,J.,&Procop,G.W. (2018).Preanalytic factorsassociatedwithhemo‐lysisinemergencydepartmentbloodsamples.Archives of Pathology and Laboratory Medicine, 142, 229–235. https://doi.org/10.5858/arpa.2016‐0400‐OA

Phelan,M.P.,Reineks,E.Z.,Schold,J.D.,Kovach,A.,&Venkatesh,A.(2016). Estimated national volume of laboratory results affectedby hemolyzed specimens from emergency departments. Archives of Pathology and Laboratory Medicine, 140, 621–621. https://doi.org/10.5858/arpa.2015‐0434‐LE

Queensland Government Department of Health (2015). Guideline Peripheral intravenous catheter (PIVC). Brisbane Queensland:Queensland Health. Retrieved from https://www.health.qld.gov.au/__data/assets/pdf_file/0025/444490/icare‐pivc‐guideline.pdf

Review Manager (RevMan). (2014). (Version 5.3). Copenhagen: TheNordicCochraneCentre,T.C.C.

RoyalCollegeofNursing. (2016).Standards for infusion therapy,4thed.London,UK:RoyalCollegeofNursing.

Seemann, S., & Reinhardt, A. (2000). Blood sample collection from aperipheral catheter system compared with phlebotomy. Journal of Intravenous Nursing,23,290–297.

Self,W. H., Speroff, T., McNaughton, C. D.,Wright, P.W., Miller, G.,Johnson, J. G., … Talbot, T. R. (2012). Blood culture collectionthroughperipheralintravenouscathetersincreasestheriskofspec‐imen contamination among adult emergency department patients.Infection Control and Hospital Epidemiology,33,524–526.https://doi.org/10.1086/665319

Snyder, S.R., Favoretto,A.M.,Baetz,R.A.,Derzon, J.H.,Madison,B.M.,Mass,D.,…Liebow,E.B.(2012).Effectivenessofpracticesto reduce blood culture contamination: A Laboratory MedicineBest Practices systematic review and meta‐analysis. Clinical Biochemistry, 45, 999–1011. https://doi.org/10.1016/j.clinbiochem.2012.06.007

Stroup,D.,Berlin,J.,Morton,S.,Olkin,I.,Williamson,D.,Rennie,D.,&Thacker,S.(2000).Meta‐analysisofobservationalstudiesinepide‐miology aproposal for reporting. JAMA,283, 2008–2012. https://doi.org/10.1001/jama.283.15.2008

Tsukuda, Y., Funakoshi, T., Nasuhara, Y., Nagano, Y., Shimizu, C., &Iwasaki,N.(2016).Venipuncturenerveinjuriesintheupperextrem‐ityfrommorethan1millionprocedures.Journal of Patient Safety,1.https://doi.org/10.1097/PTS.0000000000000264

Wollowitz,A.,Bijur,P.,Esses,D.,&Gallagher,J.(2013).Useofbutterflyneedles to draw blood is independentlyly associatedwithmarkedreductioninhemolysiscomparedtointravenouscatheter.Academic Emergency Medicine, 20, 1151–1155. https://doi.org/10.1111/acem.12245

Zlotowski,S.J.,Kupas,D.F.,&Wood,G.C.(2001).Comparisonoflab‐oratory values obtained bymeans of routine venipuncture versusperipheralintravenouscatheterafteranormalsalinesolutionbolus.Annals of Emergency Medicine,38,497–504.https://doi.org/10.1067/mem.2001.118015

How to cite this article:CoventryLL,JacobAM,DaviesHT,StonemanL,KeoghS,JacobER.Drawingbloodfromperipheralintravenouscannulacomparedwithvenepuncture:Asystematicreviewandmeta‐analysis.J Adv Nurs. 2019;00:1–27. https://doi.org/10.1111/jan.14078

APPENDIX 1ELEC TRONIC DATABA SE FIRS T SE ARCH S TR ATEGY ( JANUARY 20 0 0 –APRIL 2017)

MEDLINE search strategy(((MH“phlebotomy”))OR(“directvenouspuncture”)OR(“venepunc‐ture”)AND((MH“cannula”))OR((MH“Catheter”))OR(“intravenouscannula*”) OR (“intravenous catheter*”) OR (“peripheral venouscatheter*”)OR(“peripheralvenouscannula*”)OR(“peripheralcath‐eter*”)OR(“peripheralcannula*”))

CINAHL search strategy(((MH“phlebotomy”))OR(“directvenouspuncture”)OR(“venepunc‐ture”)AND((MH“cannula”))OR((MH“Catheter”))OR(“intravenouscannula*”) OR (“intravenous catheter*”) OR (“peripheral venouscatheter*”)OR(“peripheralvenouscannula*”)OR(“peripheralcath‐eter*”)OR(“peripheralcannula*”))

Cochrane Library(((MH“Phlebotomy”))OR(“venipuncture*”)AND((MH“Catheters”))OR(Cannula)or((MH“Catheterization”))or(PeripheralCatheterization*))

ScopusKeywords((Phlebotomy))OR (venepuncture)) ((Catheter*)OR (Intravenous

Catheter))

ISI Web of Science

TS=(Phlebotom* OR Venepuncture* OR Direct Venous Puncture)ANDTS=(Catheter*OR“IntravenousCatheter*”ORCatheteriz*)

Joanna Briggs (OVID)(sh(BloodSpecimenCollection)OR(Phlebotomy)OR(Venipuncture))AND(sh(Catheter*)OR(Cannula))

26  |     COVENTRY ET al.

APPENDIX 2

ELEC TRONIC DATABA SE SECOND SE ARCH S TR ATEGY (1 JANUARY 20 0 0 –31 DECEMBER 2018)

MEDLINE search strategy(((MH“phlebotomy”))OR(“directvenouspuncture”)OR(“venepunc‐ture”)AND((MH“cannula”))OR((MH“Catheter”))OR(“intravenouscannula*”)OR(“intravenouscatheter*”)OR(“peripheralvenouscath‐eter*”)OR(“peripheralvenouscannula*”)OR(“peripheralcatheter*”)OR(“peripheralcannula*”))AND“occlusion”OR(MH“phlebitis”)OR“dislodge*”OR“failure”OR“devicefailure”OR“infection*”OR(MH“Catheter‐Related Infections”)OR “Infiltration”OR “extravasation”OR “blockage” OR “leakage” OR “he#molysis” OR “accuracy” OR“equivalence”OR“contamination”

CINAHL search strategy((MH "Venipuncture") OR (MH "Phlebotomy") OR "venepunctureor venipuncture or phlebotomy") AND ((MH "Catheterization,Peripheral +") OR "catheteri#ation, peripheral" OR "peripheralintravenous catheter" OR "peripheral venous cannula" OR "pe‐ripheral venousdevice"OR "pivc"OR "piv")AND ((MH"CatheterOcclusion +") OR "occlusion" OR (MH "phlebitis+") OR phlebitisOR"dislodgement"ORfailureOR"devicefailure"OR"devicemal‐function"OR(MH"Catheter‐RelatedBloodstreamInfections")OR(MH "Catheter‐Related Infections")OR infectionOR "infiltration"OR"extravasation"OR"blockage"OR"leakage"OR"he#molysisofbloodsamples"OR"he#molysis inbloodtesting"OR"accuracy inbloodtest"OR"equivalenceinbloodtests"OR"contaminationofbloodcultures")

Cochrane Library(MH “Phlebotomy”) OR “venipuncture*”AND (MH “Catheters”) OR“Cannula” or (MH “Catheterization”) or (Peripheral Catheterization*)AND “occlusion” OR (MH “phlebitis”) OR “dislodge*” OR “failure”OR “infection*” OR “Infiltration” OR “extravasation” OR “blockage”OR “leakage”OR “he#molysis”OR “accuracy”OR “equivalence”OR“contamination”

Scopus("Phlebotomy" OR "venepuncture") AND ("Catheter*" OR"Intravenous Catheter") AND ("occlusion" OR "phlebitis" OR "dis‐lodge*"OR"failure"OR"infection*"OR"Infiltration"OR"extravasa‐tion"OR "blockage"OR "leakage"OR "he#molysis"OR "accuracy"OR"equivalence"OR"contamination")

IS I Web of ScienceTS=(Phlebotom* OR Venepuncture* OR Direct Venous Puncture)ANDTS=(Catheter*ORIntravenousCatheter*ORCatheteriz*)ANDTS=(occlusionORphlebitisORdislodge*ORfailureORinfection*ORInfiltrationORextravasationORblockageORleakageORhemolysisORhaemolysisORaccuracyORequivalenceORcontamination)

Joanna Briggs (OVID)(Blood SpecimenCollection)OR (Phlebotomy)OR (Venipuncture))AND(Catheter*)OR(Cannula))AND((occlusionORphlebitisORdis‐lodge*ORfailureORinfection*ORInfiltrationORextravasationORblockageORleakageORhemolysisORhaemolysisORaccuracyORequivalenceORcontamination)

APPENDIX 3

E XCLUDED S TUDIE S AND RE A SONS FOR E XCLUSION

Alexandrou, E., Ray‐Barruel, G., Carr, P. J., Frost, S., Inwood, S.,Higgins,N., Rickard,C.M. (2015). International prevalence of theuseofperipheralintravenouscatheters.Journal of Hospital Medicine, 10(8),530‐533.https://doi.org/10.1002/jhm.2389Reasonforexclusion:Nodirectcomparisonbetweenthegroups,

PIVCandvenepuncture.Burns,E.R.,&Yoshikawa,N.(2002).HemolysisinSerumSamples

Drawn by Emergency Department Personnel versus LaboratoryPhlebotomists. Laboratory Medicine, 33(5), 378‐380. https://doi.org/10.1309/PGM4‐4F8L‐2P1M‐LKPBReason for exclusion: The first part of the study compared ED

with non‐ED setting – unable to ascertain if the comparisonwasbetweenPIVCandvenepuncture.Thesecondpartofthestudyin‐cludedpaediatricpatients.Carraro,P.,Servidio,G.,&Plebani,M.(2000).Hemolyzedspeci‐

mens:areasonforrejectionoraclinicalchallenge?Clinical Chemistry, 46(2),306‐307.Reason for exclusion: Unclear if they compared between the

groups,PIVCandvenepuncture.Cox,S.R.,Dages,J.H.,Jarjoura,D.,&Hazelett,S.(2004).Blood

samples drawn from IV catheters have less hemolysiswhen5‐mL(vs10‐mL)collectiontubesareused.Journal of Emergency Nursing, 30(6),529‐533.https://doi.org/10.1016/j.jen.2004.10.004Reasonforexclusion:Nodirectcomparisonbetweenthegroups,

PIVCandvenepuncture.Dugan,L.,Leech,L.,Speroni,K.G.,&Corriher,J.(2005).Factors

affectinghemolysisratesinbloodsamplesdrawnfromnewlyplacedIV sites in the emergency department. JEN: Journal of Emergency Nursing, 31(4),338‐418.https://doi.org/10.1016/j.jen.2005.05.004Reasonforexclusion:Nodirectcomparisonbetweenthegroups,

PIVCandvenepuncture.Dwyer, D. G., Fry, M., Somerville, A., & Holdgate, A. (2006).

Randomized,singleblindedcontroltrialcomparinghaemolysisratebetween two cannula aspiration techniques. Emergency Medicine Australasia, 18(5‐6),484‐488.https://doi.org/10.1111/j.1742‐6723. 2006.00895.xReasonforexclusion:Nodirectcomparisonbetweenthegroups,

PIVCandvenepuncture.Everts,R. J.,Vinson,E.N.,Adholla,P.O.,&Reller,L.B. (2001).

Contamination of catheter‐drawn blood cultures. Journal of Clinical Microbiology, 39(9), 3393‐3394. https://doi.org/10.1128/JCM.39.9.3393‐3394.2001

     |  27COVENTRY ET al.

Reason for exclusion: Contained data on an excluded group(paediatrics).Fang, L., Fang, S. H., Chung, Y. H., & Chien, S. T. (2008).

Collecting factors related to the haemolysis of blood speci‐mens. Journal of Clinical Nursing, 17(17), 2343‐2351. https://doi.org/10.1111/j.1365‐2702.2006.02057.xReason for exclusion: Contained data on an excluded group

(paediatrics).Prue‐Owens,K.(2006).Useofperipheralvenousaccessdevices

forobtainingblood samples formeasurementofActivatedPartialThromboplastinTimes.Critical Care Nurse, 26(1), 30–38.Reasonforexclusion:Dataanalysisdidnot includeBland–Altman

plots.

Straszewski,S.,Sanchez,L.,McGillicuddy,D.,Boyd,K.,DuFresne,J.,Joyce,N.,...Mottley,J.(2011).UseofseparatevenipuncturesforIVaccessandlaboratorystudiesdecreaseshemolysisrates.Internal and Emergency Medicine, 6,357‐359.Reasonforexclusion:Thisstudyevaluatedapolicychange–we

were unsure if in the baseline data collection if blood could havebeencollectedbyeithervenepunctureorfromaPIVC.Zengin,N.,&Enç,N.(2008).Comparisonoftwobloodsampling

methods in anticoagulation therapy: venipuncture and peripheralvenouscatheter.Journal of Clinical Nursing, 17(3),386‐393.https://doi.org/10.1111/j.1365‐2702.2006.01858.xReasonforexclusion:Dataanalysisdidnot includemeandiffer‐

enceandBland–Altmanplots.

TheJournal of Advanced Nursing (JAN)isaninternational,peer‐reviewed,scientificjournal.JANcontributestotheadvancementofevidence‐basednursing,midwiferyandhealthcarebydisseminatinghighqualityresearchandscholarshipofcontemporaryrelevanceandwithpotentialtoadvanceknowledge for practice, education, management or policy. JAN publishes research reviews, original research reports and methodological andtheoreticalpapers.

Forfurtherinformation,pleasevisitJANontheWileyOnlineLibrarywebsite:www.wileyonlinelibrary.com/journal/jan

Reasons to publish your work in JAN:

• High-impact forum:theworld’smostcitednursingjournal,withanImpactFactorof1.998–ranked12/114inthe2016ISIJournalCitationReports©(Nursing(SocialScience)).

• Most read nursing journal in the world:over3millionarticlesdownloadedonlineperyearandaccessibleinover10,000librariesworldwide(includingover3,500indevelopingcountrieswithfreeorlowcostaccess).

• Fast and easy online submission: onlinesubmissionathttp://mc.manuscriptcentral.com/jan.• Positive publishing experience:rapiddouble‐blindpeerreviewwithconstructivefeedback.• Rapid online publication in five weeks:averagetimefromfinalmanuscriptarrivinginproductiontoonlinepublication.• Online Open:theoptiontopaytomakeyourarticlefreelyandopenlyaccessibletonon‐subscribersuponpublicationonWileyOnlineLibrary,aswellastheoptiontodepositthearticleinyourownoryourfundingagency’spreferredarchive(e.g.PubMed).