Systematic Reviews
Copyright ©The Author(s) 2023.
World J Gastrointest Surg. Jul 27, 2023; 15(7): 1485-1500
Published online Jul 27, 2023. doi: 10.4240/wjgs.v15.i7.1485
Table 1 Development studies of combined and intraoperative multivariate models
Ref.Period + numberSample regionOperationCharacteristicsSource of dataModels testedOutcomes tested
Janowak et al[26], 20152009-2013 (168)United StatesTT/TH/Hybrid/MIEN: 93/168; H: N/ASingle centreeSASMajor morbidity
Yoshida et al[28], 20152005-2013 (365)JapanTTN: 113/250; H: N/ASingle centreYoshida scoreMorbidity +; Major morbidity
Stroyer et al[27], 20172011-2014 (234)DenmarkTTN: 188/234; H: N/ASingle centreModified eSAS, SAS, eSASMajor morbidity
Xi et al[30], 20192015-2018 (194)ChinaTT/Hybrid/MIEN: 48/194; H: N/ASingle centreeSAS nomogramMajor morbidity
Xi et al[29], 20202015-2017 (251)ChinaTT/Hybrid/MIEN: 56/251; H: N/ASingle centreIPF nomogramMajor morbidity
Huang et al[31], 20202016-2018 (330)ChinaTT/Hybrid/MIEN: 41/330; H: 321SCC/9ACSingle centreHuang nomogramAnastomotic leak
Table 2 Validation studies of combined and intraoperative models
Ref.Period + numberSample regionOperationCharacteristicsSource of dataModels testedOutcomes tested
Zafirellis et al[36], 20021990-1999 (204)United KingdomTT/THN: 39/204; H: 156AC/45SCCSingle centrePOSSUMMajor morbidity + Mortality
Lai et al[37], 20072001-2005 (545)ChinaTTN: N/A; H: 545SCC/0ACHong Kong (All)POSSUM; O-POSSUM; P-POSSUMMortality
Lagarde et al[38], 20071993-2005 (663)NetherlandsTT/THN: N/A; H: 476AC/187SCCSingle centreO-POSSUMMortality
Baba et al[39], 20082000-2007 (142)JapanTTN: 15/142; H: 142SCC/0ACSingle centreE-PASSMorbidity
Bosch et al[33], 20111991-2007 (278)NetherlandsTTN: 10/278; H: 235AC/43SCCSingle centreACCI; CCI; O-POSSUM; P-POSSUMMortality
Yoshida et al[40], 20132005-2012 (308)JapanTTN: N/A; H: N/ASingle centreE-PASSMorbidity
Filip et al[34], 20142004-2013 (43)RomaniaTT/THN: 22/43; H: 33SCC/9ACSingle centreACCI; CCI; POSSUM; O-POSSUM; P-POSSUMMortality
Filip et al[32], 20152008-2012 (167)ItalyTT/TH/MIEN: 131/167; H: 105 AC/62SCCSingle centrePNI-multivariate PNI; CCI; ACCI; O-POSSUM; P-POSSUM; Amsterdam scoreMorbidity
Yamana et al[35], 20152005-2013 (251)JapanTT/MIEN: 150/251; H: N/ASingle centreGNRI; PNI; E-PASS; POSSUMRespiratory complications
Xing et al[41], 20162008-2010 (189)ChinaTT/THN: 7/189; H: 141SCC/45ACSingle centreE-SASMajor morbidity
Eto et al[42], 20162007-2015 (399)JapanHybridN: 176/399; H: N/ASingle centreSASMajor morbidity
Nakagawa et al[43], 20172007-2017 (379)JapanTT/Hybrid/MIEN: 129/379; H: 367SCC/12ACSingle centreSASMajor morbidity
Aoki et al[44], 20192011-2015 (246)JapanTT/Hybrid/MIEN: 152/246; H: N/ASingle centreE-SAS; Modified E-SASMajor morbidity
Hayashi et al[45], 20192012-2016 (190)JapanTTN: 115/190; H: N/ASingle centreSASMajor morbidity
Table 3 Clinical credibility of combined and intraoperative models
ModelRef.Oesophageal specificNo thresholdsTimely dataReliable dataEasy to generateUnderstandableUseful rangeTotal
POSSUMCopeland et al[46], 1991NoNoNoYesYesYesYes4
O-POSSUMTekkis et al[48], 2004PartlyNoNoYesYesYesYes4.5
P-POSSUMPrytherch et al[47], 1998NoNoNoYesYesYesYes4
E-PASSHaga et al[49], 1999NoYesNoPartlyNoPartlyYes3
Yoshida scoreYoshida et al[28], 2014YesNoNoYesYesYesNo4
Xi SAS nomogramXi et al[29], 2020YesNoNoPartlyYesYesYes4.5
Xi IPF nomogramXi et al[30], 2019YesNoNoPartlyYesYesYes4.5
Huang ModelHuang et al[31], 2020YesNoNoYesYesYesYes5
SASGawande et al[50], 2007NoNoNoPartlyYesYesYes3.5
eSASJanowak et al[26], 2015YesNoNoPartlyYesYesYes4.5
Modified eSASStroyer et al[27], 2017YesNoNoPartlyYesYesYes4.5
Table 4 Methodological quality (overall performance) for combined and intraoperative models
ModelStudy participation (out of 8)Measurements (out of 4)Analysis (out of 8)Total (out of 20)
POSSUM735.515.5
O-POSSUM84820
P-POSSUM73717
E-PASS7.53616.5
Yoshida score64616.5
Xi SAS nomogram84719
Xi IPF nomogram84618
Huang model7.54616.5
SAS84719
eSAS84416
Modified eSAS84617.5
Table 5 Summary of the performance for each of the combined and intraoperative models
Predictive model (n)Ref.DiscriminationCalibration/other statisticsOutcome
POSSUM (5)Zafirellis et al[36], 20020.55, 0.62HL P < 0.001, 0.002Morbidity + Mortality
Lai et al[37], 2007AUC = 0.776O:E 0.367 (P = 0.001)Mortality
Filip et al[34], 2014AUC = 0.826O:E 1.11Mortality
Filip et al[32], 2015AUC = 0.59HL P value (0.08)Morbidity
Yamana et al[35], 2015AUC = 0.601Not reportedResp complications
O-POSSUM (4)Lai et al[37], 2007AUC = 0.676O:E 0.505 (P = 0.002)Mortality
Lagarde et al[38], 2007AUC = 0.60O:E 0.29 (P < 0.001)Mortality
Bosch et al[33], 2011AUC = 0.756HL P value (0.035)Mortality
Filip et al[34], 2014AUC = 0.740O:E 1.70Mortality
P-POSSUM (3)Lai et al[37], 2007AUC = 0.776O:E (P = 0.814)Mortality
Bosch et al[33], 2011AUC = 0.766HL P value (0.529)Mortality
Filip et al[34], 2014AUC = 0.607O:E 4.28Mortality
E-PASS (3)Baba et al[39], 2008CRS > 1 (P < 0.001)N/AMorbidity
Yoshida et al[40], 2013CRS > 0.9 (P 0.029)N/AMorbidity
Yamana et al[35], 2015AUC = 0.659N/AResp complications
Yoshida (1)Yoshida et al[28], 2015P value = < 0.001, 0.037Not reportedMorbidity + Major morbidity
Xi SAS nomogram (1)Xi et al[30], 2019AUC = 0.903, 0.967“Good”Major morbidity
Xi IPF nomogram (1)Xi et al[29], 2020AUC = 0.921“Sufficient”Major morbidity
Huang (1)Huang et al[31], 2020AUC = 0.69“Good”Anastomotic leak
SAS (4)Eto et al[42], 2016AUC = 0.68Cut off = 6 Pulm, SSI, ALMajor morbidity
Stroyer et al[27], 2017AUC = 0.561All comp P = 0.148Major morbidity
Nakagawa et al[43], 2017Not reportedCut off > 5, P < 0.0001Major morbidity
Hayashi et al[45], 2019Not reportedCut off ≥ 6, P = 0.032Major morbidity
eSAS (4)Janowak et al[26], 2015AUC = 0.614Cut off = 6, Com P = 0.01Major morbidity
Xing et al[41], 2016AUC = 0.62Cut off > 7 LOSMajor morbidity
Stroyer et al[27], 2017AUC = 0.572All comp P = 0.460Major morbidity
Aoki et al[44], 2019Optimal cut-off 4 (Youden’s index = 0.069)Major morbidity (P = 0.29)Major morbidity
Modified eSAS (2)Stroyer et al[27], 2017AUC = 0.532All comp P = 0.096Major morbidity
Aoki et al[44], 2019Optimal cut-off 4 (Youden’s index = 0.180)Major morbidity (P = 0.004)Major morbidity
Table 6 Summary of the combination and intraoperative models across the five categories
ModelRef.Clinical credibility (out of 7)Methodological quality (out of 20)Model performance (overall utility)External validationClinical effectiveness
POSSUMCopeland et al[46], 1991415.5MortalityYesNo
O-POSSUMTekkis et al[48], 20044.520NoYesNo
P-POSSUMPrytherch et al[47], 1998417MortalityYesNo
E-PASSHaga et al[49], 1999316.5NoYesNo
SASGawande et al[50], 2007416.5NoYesNo
eSASJanowak et al[26], 20154.519NoYesNo
Modified eSASStroyer et al[27], 20174.518NoYesNo
Yoshida scoreYoshida et al[28], 2015516.5NoNoNo
Xi SAS nomogramXi et al[29], 20203.519Major morbidityNoNo
Xi IPF nomogramXi et al[30], 20194.516Major morbidityNoNo
HuangHuang et al[31], 20204.517.5NoNoNo