T ea men and in enon‐ o‐ ea p opensi y sco e analysis o e alua e he impac o ideo‐
assis ed ho acic su ge y on 90‐day mo ali y ae ana omical esecon o lung cance
Au ho s
JoseLuisRecue o‐Díaz¹†,IñigoRoyo‐C espo¹,Da idGómezde‐An onio²,Se giCall³,Bo ja
Aguinagalde⁴,Ma íaTe esaGómez‐He nández⁵,Jo geHe nández‐Fe ández⁶,Da idSánchez‐
Lo en e⁷,JulioSesma‐Rome o⁸,Edua doRi o⁹,NicolásMo eno‐Ma a¹⁰,andRaulEmbun¹,¹¹*†
onbehal o heSpanishG oupo Video‐assis edTho acicSu ge y(GEVATS)
†Theseau ho scon ibu edequally o hiswo k.
Affiliaons
1. Depa men o Tho acicSu ge y,Hospi alUni e si a ioMiguelSe e ,Hospi alClínico
Uni e si a ioLozanoBlesa,IISA agón,Za agoza,Spain
2. Depa men o Tho acicSu ge y,Hospi alUni e si a ioPue adeHie oMajadahonda,
Mad id,Spain
3. Depa men o Tho acicSu ge y,Hospi alUni e si a iMú uaTe asa,Uni e si a de
Ba celona,Te asa,Ba celona,Spain
4. Depa men o Tho acicSu ge y,Hospi alUni e si a iodeDonosa,SanSebasán‐
Donosa,Spain
5. Depa men o Tho acicSu ge y,Hospi alUni e si a iodeSalamanca,IBSAL,
Uni e sidaddeSalamanca,Salamanca,Spain
6. Depa men o Tho acicSu ge y,Hospi alUni e si a ioSag a Co ,Ba celona,Spain
7. Depa men o Tho acicSu ge y,Hospi alClínicdeBa celona,Ins u oRespi a o io,
Uni e si yo Ba celona,Ba celona,Spain
8. Depa men o Tho acicSu ge y,Hospi alGene alUni e si a ioAlican e,Alican e,
Spain
9. Depa men o Tho acicSu ge y,Hospi alUni e si a ioSanagodeCompos ela,
SanagodeCompos ela,Spain
10. Depa men o Tho acicSu ge y,Hospi alRamónyCajal,Mad id,Spain
11. Depa men o Su ge y,Facul yo Medicine,Uni e si yo Za agoza,Za agoza,Spain
*Co esponding au ho
This is a p e-copyedi ed, au ho -p oduced e sion o an a icle accep ed o publica ion in Eu opean Jou nal o Ca dio ho acic
Su ge y ollowing pee e iew. The e sion o eco dJose Luis Recue o-Díaz, Iñigo Royo-C espo, Da id Gómez de-An onio, Se gi Call,
Bo ja Aguinagalde, Ma ía Te esa Gómez-He nández, Jo ge He nández-Fe ández, Da id Sánchez-Lo en e, Julio Sesma-Rome o,
Edua do Ri o, Nicolás Mo eno-Ma a, Raul Embun, on behal o he Spanish G oup o Video-assis ed Tho acic Su ge y (GEVATS),
T ea men and in en ion- o- ea p opensi y sco e analysis o e alua e he impac o ideo-assis ed ho acic su ge y on 90-day
mo ali y a e ana omical esec ion o lung cance , Eu opean Jou nal o Ca dio-Tho acic Su ge y, Volume 62, Issue 3, Sep embe
2022, ezac122 is a ailable online a : h ps://doi.o g/10.1093/ejc s/ezac122.
RaulEmbun
Depa men o Tho acicSu ge y
Hospi alUni e si a ioMiguelSe e
PaseoIsabellaCa ólica1,Za agoza50009,Spain
Tel:+34‐635492179
Email: aulem[email p o ec ed]
Abs ac
OBJECTIVES:Theaimo hiss udywas oknow he ea men effec o ideo‐assis ed ho acic
su ge y(VATS)on90‐daymo ali yae ana omicallung eseconbasedonanaonwide
coho .
METHODS:Thisisamulcen ep ospec ecoho o 2721ana omical esecons o lung
cance omDecembe 2016 oMa ch2018.T ea men andin enon‐ o‐ ea (ITT)analyses
we epe o medae in e sep obabili ysco eweighnganddiffe en p opensi ysco e
ma chingalgo i hms.Co a ia ebalancewasassessedbys anda dizedmeandiffe ences.The
esma o s epo edwe e hea e age ea men effec , hea e age ea men effec on he
ea edandodds aosae condionallogiscmodelswi h95%con idencein e als.The
uncon oundednessassumponwase alua edbysensi i yanalysis o a e age ea men
effec (c‐dependence)anda e age ea men effec on he ea ed(Γ).
RESULTS:VATSwas heinialapp oachin1911paen s(70.2%), hough273cases(14.3%)had
obecon e ed o ho aco omy.Nine y‐daymo ali y a eswe e: ea men analysis(VATS
1.16% sopen3.9%,P<0.001),ITTanalysis(VATS1.78% sopen3.36%,P=0.012).Ae
in e sep obabili ysco eweighngandp opensi ysco ema ching,in he ea men analysis,
VATSmean absolu e isk educonsbe ween2.25%and2.96%and ela e isk educons
be ween65%and70%(OR=0.34,95%con idencein e al0.15–0.79,allP‐ alues<0.004).
Howe e ,all heesma o s u nedou obenon‐signi ican in heITTanalyses.Ahigh
sensi i y ounobse ablecon ounde swasp o ed(c‐dependence0.135,Γ=1.5).
CONCLUSIONS:VATScan educe he isko 90‐daymo ali yae ana omicallung esecon.
Howe e , heimplicaonso con e sion o ho aco omy,compa ingITT e sus ea men
analysis,and hepo enalimpac o hiddenbiasshoulddese e u he aenonin he u u e.
Keywo ds:Video‐assis ed ho acicsu ge y,90‐Daymo ali y,In enon‐ o‐ ea analysis,
Tho acicsu ge y,Ana omicallung esecon,Lungcance
Abb e iaons
ATE‐A e age ea men effec
ATT‐A e age ea men effec on he ea ed
CIs‐Con idencein e als
GEVATS‐SpanishG oupo Video‐assis edTho acicSu ge y
ITT‐In enon‐ o‐ ea
IPSW‐In e sep obabili ysco eweighng
PSM‐P opensi ysco ema ching
SMD‐S anda dizedmeandiffe ences
VATS‐Video‐assis ed ho acicsu ge y
In oducon
Al hough he ea emulple e ospec ese ies ha ha eshownalowe a eo complicaons
andpos ope a es ayinpaen sope a edonby ideo‐assis ed ho acicsu ge y(VATS), he e
a emo edisc epancieson heimpac ha hesu gicalapp oachcouldha eonpos ope a e
mo ali y.In his ega d,a ecen andomizedclinical ial,designed oe alua esho ‐ e mand
oncologicefficacyo VATS, ailed odemons a ediffe encesinpos ope a emo ali y
be weenVATSand ho aco omy[1].
Rega dingpos ope a emo ali y, ecen se iesha eshown ha 90‐daymo ali ycould
double30‐dayo in‐hospi almo ali yae lung esecon[2–4].Al houghmos se ies
compa ingVATSwi h ho aco omydono menon90‐daymo ali y,someimpo an s udies
ha eno shownsigni ican diffe encesinei he in‐hospi al,30‐dayo 90‐daymo ali y[5–7].A
ecen publicaon om heESTSda abasedidshowsigni ican diffe encesinmo ali ya
discha ge(VATS1% s ho aco omy1.9%,P=0.020),no menoningmo ali ya 90days[8].
Fews udiesha epe o medanin enon‐ o‐ ea (ITT)analysiscompa ingVATSand
ho aco omy o lung esecon,ande enlessha ecompa ed he esul sob aineddepending
on hes a egyo analysispe o med[9,10].Howe e ,sinceanon‐negligiblep opo ono
paen sunde goingVATSmus be inallycon e ed o ho aco omy,ITTanalysisseems obe
hemos app op ia es a egy oe alua e ea men effec s ela ed oVATSina ealscena io
[11].
In iewo hecon licng esul sand heunde epo edITTanalysisin heli e a u e, he
objec eo ou s udywas ode e mine heimpac o hesu gicalapp oachon90‐day
mo ali y,compa ing ea men andITTanalysisinpaen swhounde wen andana omical
lung esecon o lungcance in henaonalcoho by heSpanishG oupo Video‐assis ed
Tho acicSu ge y(GEVATS)[12].
Ma e ials and Me hods
E hical s a emen
Thisp ojec wasapp o edbyall helocale hicscommieesandin o medconsen was
ob ained om he ec ui edpaen s ouse hei clinicalda a o scien icpu poses(App o al
byE hicsCommieeo A agonHeal hResea chIns u eon20May2015PI15/0072).
Da a sou ce
TheGEVATSo heSpanishSocie yo Tho acicSu ge yisap ospec e olun a ymulcen e
obse aonals udywi ha o alo 33Tho acicSu ge yDepa men pa cipang.TheCen es
we eno selec edbasedon hei expe ienceinVATSo anyo he c i e ion.Thecoho
includedpaen swhounde wen anana omicallung esecon om20Decembe 2016 o20
Ma ch2018.TheGEVATSobjec eswe e oknow heimpac o su gicalapp oachonsho ‐
andlong‐ e mou comes.Theme hodo heGEVATS,includingsamplesizejus icaonand
heaudi p ocesspe o med,was ecen lypublished[12].
Paen allocaonin oVATSo ho aco omydependedonclinicalp acce,expe ienceand
belie so eacho hepa cipangsu geonsanddepa men s.
In hisp ospec eobse aonalcoho s udy,we ied ospeci icallyelucida e heimpac o
hesu gicalapp oachon90‐daymo ali yae ana omicallung esecon o lungcance .Fo
hispu pose, hosepaen swi hadiagnosiso he hanlungca cinomaand hosewho
unde wen pneumonec omyo ex endedlung eseconwe eexcluded.Ex ended esecon
wasconside edincaseo ches wall,diaph agmo slee e esecon.Ou manusc ip is
epo edacco ding o heSTROBE ecommendaonsandESTSS ascalP ime o p opensi y
sco eanalysis.
S ascal analysis
Adouble ypeo analysiswasca iedou , ea men andITT,dependingon heneedo
con e sion omVATS oopen h oughou hep ocedu e.Theassociaono hesu gical
app oach(open e susVATS)as he ea men a iable,wi hbaselineoncologicandsu gical
a iables ha couldin luenceon heou come a iable(90‐daymo ali y)and heapp oach o
bechosen,wasanalysedby wo‐ aileds ascalhypo hesis esng,usingMann–Whi neyand
Chi‐squa e es sands anda dizedmeandiffe ences(SMD).Those a iableswi haP‐ alueo
less han0.2and/o s anda dizeddiffe encesg ea e han0.1we e heco a ia esused o
build hep opensi ysco e oco ec o seleconbias.
Missingda awe edeal bycasewisedeleonanalysiswhenless han5%o paen shad
incomple e egis ies.
Thep opensi ysco ewasesma edbyalogi modeland heo e lapassumponwasassessed
ondensi yplo s o ea men andITTanalysis.The ea men effec swe ee alua edbasedon
hein e sep obabili ysco eweighng(IPSW)and hep opensi ysco ema ching(PSM)
h ough henea es ‐neighbo me hodwi handwi hou eplacemen ,usingdiffe en callipe
wid hs(0.035,0.05and0.1)andma ching aos(1:1,1:2and1:3).
Co a ia ebalancedwasassessedbySMD,be o eandae weighngo ma ching.SMDless
han0.1o 0.05we econside edgoodo excellen , espec ely, oexclude esidualimbalance
[14].Balanceo co a ia eswasdisplayedondo plo s o IPSWandPSM,sepa a ely.
The ea men effec swe eesma edbyweigh edmeanandma chingou comemodels
epo ng hea e age ea men effec (ATE),basedon hediffe enceinpo enalou come
means,and hea e age ea men effec on he ea ed(ATT).ATEandATTwe e epo edas
absolu e isk educons(IPSWandPSM)and ela e isk educons(IPSW).Inaddion,in
caseo PSM1:1wi hou eplacemen andcallipe 0.035,acondionallogisc ixed‐effec s
eg essionmodelwasconduc ed.In hiscase, ea men effec swe e epo edasodds aos.
Theimpac o su geonexpe ienceinVATSp ocedu es(≤50 e sus>50cases),su geonsenio i y
( esiden e sus acul y<10yea s e sus acul y10–20yea s e sus acul y>20yea s),su gical
olumeandVATS a ebydepa men (disc e e a iables)we eused oadjus heodds aos
epo edby hecondionallogiscmodelsp e iouslydesc ibed.Tocompu esu gical olume
andVATS a ebyins uon h oughou he15‐mon h ec ui men pe iod,weconside ed he
epo ssubmiedby heHeadso heAdminis a eDepa men s omeachIns uon.
These epo swe eusedin heaudi p ocesswep e iouslypublished[12].95%con idence
in e als(CIs)we ecalcula ed om obus s anda de o sandP‐ aluesless han0.05we e
conside eds ascallysigni ican .
Thecondionalpa aldependenceme hodp oposedbyMas ene al.wasused oe alua e
hesensi i yo conclusionsabou heATE.Boundson heATEgi enase o c‐dependence
alues(be ween0and1)and heb eakdownpoin (maximum alueo hec‐dependence
pa ame e unde which heconclusionsllholds)we e epo ed.Inaddion, heimpac o
hiddenbiason heATTesma o ae PSM(1:1,callipe 0.035,no eplacemen )wasassessed
wi h heboundingapp oachp oposedbyRosenbaum.TheΓpa ame e andco espondingP‐
alueswe eused omeasu e hesensi i yo ATT ounobse ablecon ounde s.
TheT ea men Effec sSui einS a a/MP16.0and heS a apackagesS ddiff,Psma ch2,
Calipma ch,Tesensi i yandMhboundswe eused o hes ascalanalysis.TablauDesk op
2020.3.1wasused o plo ep esen aono heco a ia ebalance.
Resul s
A o alo 3533paen swe e ec ui ed,including1917VATScases(54.3%).Ae exclusiono
paen swi hadiagnosisdiffe en olungcance (448paen s,12.7%),pneumonec omy(236
paen s,6.7%)andex ended esecon(165,4.7%),2721paen s(77%o heen eGEVATS
coho )me heinclusionc i e ia.The ypeso eseconsincludedwe e:2444lobec omies
(90%),111bilobec omies(4%)and166ana omicalsegmen ec omies(6%).
VATSwas heinialapp oachin1911paen s(70.2%), ep esenng he ea men g oupin
heITTanalysis.Howe e ,273cases(14.3%)had obecon e ed o ho aco omyand,
he e o e, heVATSa min he ea men analysisconsis edo 1638paen s(60.2%).The
unadjus edanalysisshowedanimpo an associaonbe weenin‐hospi almo ali yand90‐day
mo ali ywi h hesu gicalapp oachpe o med(Table1).Theou come a iable(90‐day
mo ali y)wasmissingin11cases(0.4%) ha we eno conside edin heanalysiso ea men
effec s.Thepe cen ageso missing alueswe enegligible o all hecon ounde sexcep o
DLCO.ThemainanalysisexcludedDLCOascon ounde ,so14 a iableswe eused obuild he
p opensi ysco einacomple e‐caseanalysissinceonly1.8%o hepaen shadmissing alues
insomeo heco a ia esincluded(Table2).Thep opensi ysco edis ibuon op o e he
o e lapassumponisshowninadensi yplo (Figu e1).Only9paen sin heVATSg oup
(0.6%)hadap opensi ysco ehighe han hemaximump opensi ysco e aluein heopen
g oup,whileonly2paen sin heopeng oup(0.2%)hadap opensi ysco elowe han he
minimum aluein heVATSg oup.
Ae IPSW, he aoso VATS o ho aco omywe e1339:1324paen sin he ea men
analysisand1347:1317paen sin heITTanalysis.Co a ia ebalancewasp o ed(Figu e2).
PSM1:1wi hou eplacemen andacallipe wid ho 0.035yielded872ma chedpai sin he
ea men analysisand705ma chedpai sin heITTanalysis.Mos o heco a ia es eached
anexcellen balanceae ma ching(Table3).Thep opo onso casesma ched o1con ol
we e54.2%( ea men analysis)and37.6%(ITTanalysis).
Ae PSMwi h eplacemen ,2664paen s(1608VATSand1056open)in he ea men
analysisand2663paen s(1877VATSand786open)in heITTanalysiswe ema ched o1,2
o 3coun e pa s.Independen lyo hema ching aosandcallipe wid hs,mos o he
co a ia es eachedanexcellen balance(Figu e3).
Basedon he ea men analysis,VATSwasconsis en lyassocia ed oalowe 90‐daymo ali y
a eacco ding o heATEandATTesma o s,wha e e hep opensi ysco e echniqueo
algo i hmused.Howe e ,in hecaseo ITTanalysis, heslllowe mo ali yae VATSwas
associa ed onon‐signi ican ATEandATT,in e mso absolu eand ela e isk educons,in
all hecases(Table4).
In hecondionallogiscmodelae PSM1:1(no eplacemen andcallipe wid h0.035), he
esul sob ainedae ea men analysis(OR=0.38;95%CI0.20–0.73;P=0.004)andITT
analysis(OR=0.71;95%CI0.38–1.33;P=0.283)we econsis en wi h heATEandATT
esma o s.Ae adjusng he ixed‐effec modelsbysu geonexpe ience(numbe o VATS
p ocedu esandsenio i y)anddepa men expe ience(su gical olumeandVATS a e) heodds
aoswe esimila o hosein henon‐adjus edma ched ea men analysis(OR=0.34;95%CI
0.15–0.79;P=0.012)andlowe , houghsllnon‐signi ican ,in heITTanalysis(OR=0.41;95%
CI0.13–1.25;P=0.117).
In hesensi i yanalysis, heclose oze oc‐dependence aluesunde which heconclusions
sllheld( ea men analysis0.135andITTanalysis0.09)mean ahighsensi i yo heATE
esma o ohiddenbias(Figu e4).Inaddion, hemaximumΓ aluea whichsigni ican P‐
alueswe eob ained(Γ=1.5)showed ha heATTesmaono hiss udyissensi e obias
(unobse ed a iables)able oinc ease heoddso ecei ingVATSinmo e han50%.
In hemainanalysis,wedidno conside DLCObecauseo hehigh a eo missing alues
(15.2%).Howe e ,inaseconda yanalysisincludingDLCOasaco a ia e obuild hep opensi y
sco e, heATEandATTesma o sae IPSWwe eequi alen o hoseob ainedin hemain
analysis.Mo eo e , hesensi i yanalysisshowedane enhighe po enalimpac o
unmeasu edo unobse ablecon ounde s(Γ=1.15).
Discussion
Ou main indingwas ha VATScan educe90‐daymo ali yae anana omicallung esecon
o lungcance .Howe e , hisbene icialeffec dec easedand u nedou obenon‐signi ican
whenanITTanalysiswaspe o med.No ewo hy, hese indingswe econsis en alongall he
p opensi ysco ealgo i hmsca iedou .Finally, hehighsensi i yo ou analysis ohidden
biashighligh s ha hepo enalimpac o unmeasu edandunobse ablecon ounde sin
obse aonals udiescompa ingVATSandopensu ge ycouldbede e minan .
The esul so he ews udiescompa ing heimpac o hesu gicalapp oachon90‐day
mo ali ylead ocon adic o yconclusions.Limi eds ascalpowe seconda y osmall
samplesizesandsu p isinglylowmo ali y a escouldpa lyexplainnon‐conclusi e indings
[1,6,7].Howe e ,whenanalysing he esul sob ainedbyla geins uonalcoho s,signi ican
s ascaldiffe encescouldonlyco espond oclinically i ial ea men effec [15].
In he ho acicsu ge yli e a u e,use ul ecommendaonsha ebeenmade oimp o e
epo ngbasedon hep opensi ysco eanalysis,since hisanalycapp oachisno as
s aigho wa das eg essionanalysis[16].Conside ing he ecommendaonsby heEu opean
Socie yo Tho acicSu geonsini ss ascalp ime epo ,weca iedou he2ad ised
me hods,IPSWandPSMwi hcallipe [13].Inaddion, oassess heconsis encyo ou esul s,
wepu in op accediffe en ma chingalgo i hms( ea ed ocon ol aos,callipe wid hs
and eplacemen e susno eplacemen ma ching).
Theuseo 2p opensi ysco es echniquescompa ingou comesae VATSo openlung
eseconisanexcepon[10].Pagèse al.published he esul sbyaF enchnaonwides udy
e aluang heimpac o su gicalapp oachonsho ‐andlong‐ e mou comesae lobec omy
o lungcance .In hisla gecoho (n=24,811) ec ui ed h oughou an8‐yea pe iod,only
4.9%o hecaseswe eope a edonbyVATS.I isno ewo hy ha hesigni icancele el
ob ainedina ewou comesdiffe edbe ween hep opensi ysco es a egiesused(IPSW
e susPSM).Howe e , ega ding30‐daymo ali y,bo hIPSWandPSM ailed odemons a ea
signi ican educonae VATS(PSMOR=0.89,95%CI0.45–1.81;IPSWOR=0.74,95%CI
0.37–1.45).
Theimpo anceo anITTs a egywhene aluang heimpac o hesu gicalapp oachwas
highligh edinoneo he i s me a‐analysespublishedcompa ingVATSand ho aco omy[17].
Howe e , hein o maonabou con e sion omVATS oopensu ge yisno always egis e ed,
e enincaseo naonalandin e naonal egis ies[6,8,10,18,19].Consequen ly,ITTanalysis
isno awidep accein he ho acicsu ge yli e a u eand, he e o e, hebene i con e ed o
VATSwhene aluang ea men effec scouldbemisleading.In ac , hehighe he a eo
con e sion omVATS oopen hehighe hepo enaldisc epancybe ween ea men andITT
analyses.In his ega d,despi eahighe a eo VATSinou coho (60.2%) han heone
epo edbymanynaonwideandmulcen e egis ies, hep opo ono con e sionwea e
epo ng(14.3%)isno ablysupe io compa ed osome ecognizedse ies anging om2% o
9%[11,20,21].Risko con e sionina ecen me analysiswas9.6%(95%CI:6.6–13.9%)[22].
Wecanno concludeabou he easons o ou highe con e sion a e.Howe e ,naonwide
ep esen a eness,da aaudi ha in ol edall hepa cipangcen es,o jus amae o
disposiono su geons,mo ep one os a ng hep ocedu ebyVATSinou coun y,couldbe
someo hehypo heses.Inaddion,al houghanyonecouldexpec ha highe VATS a esa e
pa allel olowe con e sion a esbecauseo expe ienceacquision, his easoningcouldbe
con licngdependingon hescena ioand he ypeo coho inqueson, o examplesingle
cen e e susmulcen e.In his ega d,ou ela elyhighcon e sion a ecouldsimply
ep esen aconsequenceo ou di ec lyp opo onalhighVATS a e.
Al houghsomes udiesa gue ha con e sion omVATS oopendoesno en ailasu gical
ailu e,o he se iesha eassocia edcon e sion owo seou comesin e mso pos ope a e
mo bidi y[20,23].Ou s udywouldsuppo ade imen aleffec associa ed ocon e sion.
Howe e ,mo ede ailedandspeci icanalysescompa ingcon e edVATS os aigh
ho aco omycouldgene a einsighulknowledgeabou heimplicaonso con e sioni sel
and hemos app op ia edisposion owa dss a ng hep ocedu ebyVATS.
ITTanalysisisclaimedas hegolds anda din andomizedcon ol ials o e lec ap agmac
clinicalscena io, omain ainp ognoscbalancegene a ed om heo iginal andomizaonand
ogi eanunbiasedesma eo ea men effec .Howe e ,weconside edimpo an a his
s agewhenmos o hes udiesdealingwi hVATSa esll epo ng hei esul sbasedona
ea men analysis, oinclude hislessconse a es a egyo analysisso ha wecouldsll
compa eou indingswi h hecoe aneousli e a u e.Toou knowledge, hisis he i s
manusc ip ha e alua e heimpac o hesu gicalapp oachonsho ‐ e mmo ali yae
lung esecon,compa inga ea men andITTanalysis.
Limi aons
Themaind awbackinmulcen e olun a y egis iesis ela ed oseleconbiasandda a
quali y.Al hough hede ailso ou audi we ep e iously epo ed,wecanno ejec esidual
biasa hiss age[12].
Asp e iouslypublished, ocalcula eou samplesize,weconside ed2%absolu e isk
diffe encein90‐daymo ali yas ele an inp acce(4% s2%),andanexpec ed25%o cases
byVATS[12].Becauseo heunde esma ed90‐daymo ali y a ediffe encebe weenVATS
andOpen(1.16% s3.9%)and hehighe ‐ han‐expec edVATS a e(60%), hes ascalpowe
o ou 'posi e'conclusionin he ea men analysis(99%)ishighe han hecon enonal
80%.Howe e ,in heITTanalysis, helowe diffe enceinmo ali ybe weenbo happ oaches
(1.78% s3.36%)and helowe p opo ono con olpaen s(29%)couldbe he easons o a
limi edpowe (69%)o ou 'nega e' esul s.
Missing aluesa eano he co ne s oneinobse aonals udies.Al houghDLCOwas heonly
co a ia ewi hamissing a eo highe han2%(415/2721paen s15%), his espi a o y
pa ame e hasbeenp o edasoneo hemos de e minan isk ac o sae lung esecon,
e eninou owncoho [24–26].Despi eequi alen esul sinapos hocanalysisbasedon he
subg oupo paen swi h his aluep esen ,con oundingbiascanno beexcluded.
Despi e henaonal ep esen a enesso heGEVATScoho ,sincei wouldha eincluded
50%o heana omicallung eseconspe o medinSpaino e he15‐mon h ec ui men
pe iod, hehighaudi ed a eo VATS egis e ed(60%)couldcomp omise heexpo aono
ou esul s oo he naonalo localscena ioswi hanimpo an disc epancyinVATS
implemen aon[12].Howe e , heequipoiseinou s udyin e mso hep opo ono
paen sineachg oupcompa ed oo he se iesmakesi lesslikely ha ei he ou openg oup
ep esen sas onglyselec edcollecono challengingcaseso ou VATSg oupas ongly
selec edcollecono a ou ablecases[6,10,27].
PSMisconside ed ojeopa dize hegene alizaono i sconclusionswhenma chedsample
sizeislowe han50%o heo iginalsample.Inou s udybyca yingou PSM1:1wi hou
eplacemen andcallipe wid h0.035, hep opo ono ma chedsamplewas64%and52%
ae ea men andITTanalysis, espec ely.Ne e heless, he esul sob ainedae such
PSMalgo i hmswe eequi alen o hose eachedae makinguseo all hesamplewi h
o he p opensi ysco e echniques,whichseems op ese eou conclusiona amo egene al
seng.
Finally,hiddenbiasisaubiqui ousp oblemine e yobse aonals udy,e enincaseo a
igo ousp opensi ysco eanalysis,andsensi i y es ies oge abee unde s andingo his
obs acle oob ain eliableesma o s.Howe e , his ypeo analysisisexceponalin ho acic
su ge y[10].Thes udypublishedbyPagèse al. om heEpi ho da abase epo edahigh
sensi i y ounobse ablecon ounde swhenesmangpos ope a edea h(Γ=1.6).Ou
esul swe e e ysimila (Γ=1.5),e enae includingDLCOasaco a ia ein hep opensi y
sco eanalysis(Γ=1.2, heclose o1 hemo esensi eou esmaons ohiddenbias).
Consequen ly, he eliabili yo ou esmaonsshouldbecauouslyadmiedunlmo e
obus e idence.
Conclusion
Ha inginmind heexclusionc i e iap e iously e e ed(pneumonec omyandex ended
esecon),VATSwasconsis en lyshown o educe90‐daymo ali yby65–70%ae an
ana omicallung esecon o lungcance in heGEVATScoho .Howe e ,d awnon heITT
analysisincludingpaen scon e ed oopensu ge yin o heVATSg oup, hisbene i
dec eased o heex en o ob ainings ascallynon‐signi ican , houghpo enally
unde powe ed,diffe ences.Unlupcominge idencecompa ingcon e edVATS os aigh
ho aco omy,aneffo shouldbemade oopmize hebene icialeffec o VATSonsho ‐ e m
mo ali yae anana omical esecon o lungcance .In heanalysiso hiddenbias,we
demons a ed ha he ea men effec esma o swe ehighlysensi e ounobse able
con ounde s,whichcouldonlybeo e comeinasufficien lyla gemulcen e andomized ial.
Acknowledgemen s
We hankall helocal esea che swhocon ibu edwi hpaen ec ui men andda aen y.
Coo dina o :RaulEmbun(Uni e si yHospi alMiguelSe e ).
Scien ic commiee:Da idGómezdeAn onio(Uni e si yHospi alPue adeHie o
Majadahonda,Mad id);Se giCall(Mú uaTe asaUni e si yHospi al,Uni e si yo Ba celona,
Te asa,Ba celona);NicolásMo eno‐Ma a(RamónyCajalUni e si yHospi al,Mad id);
Ma celoF.Jiménez(SalamancaUni e si yHospi al,Uni e si yo Salamanca,IBSAL,
Salamanca);MiguelCong egado(Vi genMaca enaUni e si yHospi al,Se ille);andSe gio
Bolu e ‐Nadal(Gene alUni e si yHospi alo Alican e).
Local esea che s:JoseLuisRecue oandIñigoRoyoC espo(Uni e si yHospi alMiguelSe e
andLozanoBlesa);Bo jaAguinagaldeandIke LópezSanz(DonosaUni e si yHospi al,San
Sebasán‐Donosa);Se gioAmo ‐AlonsoandF anciscoJa ie Mo adiellos‐Díez(Qui onsalud
Mad idUni e si yHospi al,Mad id);MiguelJesúsA a ás(ValencianOncologyIns u e
Foundaon,Valencia);AnaIsabelBlancoO ozco(Uni e si yHospi alVi gendelRocío,Se illa);
Ma cBoadaandDa idSánchez(Hospi alClinico Ba celona,Respi a o yIns u e,Uni e si yo
Ba celona,Ba celona);Albe oCabañe oSánchez(RamónyCajalUni e si yHospi al,Mad id);
IsabelCalVázquezandRamónMo eno‐Balsalob e(Uni e si yHospi alLaP incesa,Mad id);
ÁngelCille ueloRamos(Uni e si yClinicalHospi al,Valladolid);Sil anaC owleyCa asco
(Uni e si yHospi alPue adeHie oMajadahonda,Mad id);ElenaFe nández‐Ma nand
Flo ennoHe nandoT ancho(Hospi alClínicoSanCa los,Mad id);SanagoGa cía‐Ba ajasand
Cip ianoLópezGa cía(Uni e si yHospi alo Badajoz,Badajoz);Ma iaDolo esGa cía‐Jiménez
(Uni e si yHospi alo Albace e,Albace e);JoseMa íaGa cía‐P imandEdua doRi o(Sanago
deCompos elaUni e si yHospi al,SanagodeCompos ela);JoseAlbe oGa cia‐Salcedo(12
deOc ub eUni e si yHospi al,Mad id);JuanJoséGelbenzu‐ZazpeandMa íaElenaRamí ez‐
Gil(ComplejoHospi ala ioo Na a e,Pamplona);Ca losFe nandoGi aldo‐OspinaandRobe o
MongilPoce(Hospi alRegionalUni e si a io,Málaga);Ma íaTe esaGómezHe nández
(SalamancaUni e si yHospi al,Uni e si yo Salamanca,IBSAL,Salamanca);Jo geHenández
andJuanJoséFiblaAl a a(Sag a Co Uni e si yHospi al,Ba celona);Jenni e D.IllanaWol
(Pue adelMa Hospi al,Cádiz);Albe oJau eguiAbula ach(Valld'Heb onUni e si yHospi al,
Ba celona);UnaiJiménezandRa aelRojo‐Ma cos(Uni e si yHospi alC uces,Bilbao);Nés o J.
Ma nez‐He nández(LaRibe aUni e si yHospi al,Alci a,Valencia);Elisabe hMa nez‐Téllez
andJuanCa losT ujilloReyes(San aC euandSan PauHospi al,Au onomousUni e si yo
Ba celona,Ba celona);LucíaMillaCollado(Hospi alA naudeVilano a,Lleida);Se gioB.
Mo enoMe ino(Vi genMaca enaUni e si yHospi al,Se ille);Ca meObiols(Mú uaTe asa
Uni e si yHospi al,Uni e si yo Ba celona,Te asa,Ba celona);Flo encioQue o‐Valenzuela
(Vi gendelasNie esHospi al,G anada);Rica dRamos‐Izquie do(Uni e si yHospi alo
Bell i ge,Hospi ale deLlob ega ,Ba celona);Albe oRod íguez‐Fus e (Hospi aldelMa ,
Hospi aldelMa MedicalResea chIns u e,Ba celona);Lau aSanchezMo eno(Uni e si y
11. DecaluweH,Pe e senRH,HansenH,PiwkowskiC,AugusnF,B unelliAe al.;ESTS
MinimallyIn asi eTho acicSu ge yIn e es G oup(MITIG).Majo in aope a e
complicaonsdu ing ideo‐assis ed ho acoscopicana omicallung esecons:an
in enon‐ o‐ ea analysis.Eu JCa dio ho acSu g2015;48:588–99.
12. EmbunR,Royo‐C espoI,Recue oDíazJL,Bolu e S,CallS,Cong egadoMe al.Spanish
Video‐Assis edTho acicSu ge yG oup:me hod,auding,andinial esul s oma
naonalp ospec ecoho o paen s ecei ingana omicallung esecons.A ch
B onconeumol2020;56:718–24.
13. BenedeoU,HeadSJ,AngeliniGD,Blacks oneEH.S ascalp ime :p opensi ysco e
ma chingandi sal e na es.Eu JCa dio ho acSu g2018;53:1112–7.
14. AusnPC.Balancediagnoscs o compa ing hedis ibuono baselineco a ia es
be ween ea men g oupsinp opensi y‐sco ema chedsamples.S a Med2009
10;28:3083–107.
15. Abdelsaa ZM,AllenMS,ShenKR,Cassi iSD,NicholsFC,WigleDAe al.Va iaonin
hospi aladopon a eso ideo‐assis ed ho acoscopiclobec omy o lungcance and
heeffec onou comes.AnnTho acSu g2017;103:454–60.
16. AusnPC.P opensi y‐sco ema chingin heca dio ascula su ge yli e a u e om
2004 o2006:asys emac e iewandsuggesons o imp o emen .JTho ac
Ca dio ascSu g2007;134:1128–35.
17. YanTD,BlackD,BannonPG,McCaughanBC.Sys emac e iewandme a‐analysiso
andomizedandnon andomized ialsonsa e yandefficacyo ideo‐assis ed ho acic
su ge ylobec omy o ea ly‐s agenon‐small‐celllungcance .JClinOncol
2009;27:2553–62.
18. Pu iV,Gaisse HA,Wo mu hDW,G oganEL,Bu eindWR,ChangACe al.Tho acic
su geonspa cipan spe o ming ho acoscopiclobec omy.AnnTho acSu g
2019;107:202–8.
19. PaulS,Al o kiNK,ShengS,LeePC,Ha poleDH,OnaisMWe al.Tho acoscopic
lobec omyisassocia edwi hlowe mo bidi y hanopenlobec omy:ap opensi y‐
ma chedanalysis om heSTSda abase.JTho acCa dio ascSu g2010;139:366–78.
20. BongiolaS,Gon ioA,ViggianoD,Bo gianniS,PoliL,C isciRe al.;I alianVATS
G oup.Risk ac o sandimpac o con e sion omVATS oopenlobec omy:analysis
omanaonalda abase.Su gEndosc20191;33:3953–62.
21. Lau senLØ,Pe e senRH,HansenHJ,JensenTK,Ra nJ,KongeL.Video‐assis ed
ho acoscopicsu ge ylobec omy o lungcance isassocia edwi halowe 30‐day
mo bidi ycompa edwi hlobec omyby ho aco omy.Eu JCa dio ho acSu g
2016;49:870–5.
22. Powe AD,Me iRE,Abdel‐RasoulM,Moffa‐B uceSD,D'SouzaDM,Kneue zPJ.
Esmang he isko con e sion om ideo‐assis ed ho acoscopiclungsu ge y o
ho aco omy—asys emac e iewandme a‐analysis.JTho acDis2021;13:812–23.
23. Fou d ainA,DeDominicisF,IquilleJ,La ieS,Me luscaG,Wie‐P is e Ae al.
In aope a econ e siondu ing ideo‐assis ed ho acoscopydoesno cons u ea
ea men ailu e.Eu JCa dio ho acSu g2019;55:660–5.
24. B unelliA,DineshP,Woodcock‐ShawJ,LilechildD,PompiliC.Nine y‐daymo ali y
ae ideo‐assis ed ho acoscopiclobec omy:incidenceand isk ac o s.AnnTho ac
Su g2017;104:1020–6.
25. GómezdeAn onioD,C owleyCa ascoS,Rome oRománA,RoyuelaA,SánchezCalle
Á,ObiolsFo nellCe al.Su gical isk ollowingana omiclung eseconin ho acic
su ge y:ap ediconmodelde i ed omaSpanishMulcen e Da abase.A ch
B onconeumol2021Feb24:S0300‐2896(21)00070‐3.
26. AguinagaldeB,InsausA,LopezI,SanchezL,Bolu e S,EmbunR.VATSlobec omy
mo bidi yandmo ali yislowe inpaen swi h hesameppoDLCO:analysiso he
da abaseo heSpanishVideo‐Assis edTho acicSu ge yG oup.A chB onconeumol
2021;57:750–6.
27. XuJ,NiH,WuY,CaoJ,HanX,LiuLe al.Pe iope a ecompa isono ideo‐assis ed
ho acicsu ge yandopenlobec omy o pT1‐s agenon‐smallcelllungcance paen s
inChina:amul‐cen e p opensi ysco e‐ma chedanalysis.T anslLungCance Res
2021;10:402–14.
No e:Thisis heau ho 'saccep edmanusc ip e sion.The inalpublisheda clemaycon ain
addionaledi sand o mangchangesmadedu ing hep oduconp ocess.Pleaseci e he
published e sionwhen e e encing hiswo k.
Figu e_1
Figu e_2
Figu e_3
Figu e_4
