Upper-GI Endoscopy, Surgical — Stratify Before Scoping, Treat by Forrest Stigma, and Escalate Failures to Clips, Embolization or Theatre

The upper-GI bleed wants four verdicts before the scope — who goes home, how low the haemoglobin runs, which stigma gets treated, and what rescues failure — because Rockall identifies the dischargeable sixth, restrictive transfusion lifts survival while halving transfusions, oozing rebleeds near placebo rates after haemostasis, and over-scope clips beat standard therapy by forty absolute points on recurrent bleed. Stratify with Rockall and Blatchford, transfuse at seven, scope within a day on high-dose PPI, treat spurting, oozing and visible vessels while watching flat spots and clean bases, double oral PPI for high Rockall, re-look selectively, clip refractory bleed before embolizing, band varices without adding sclero, scope foreign bodies by the hour, survey Barrett by grade, feed neurologically by PEG, pause antithrombotics by risk, and stent selected perforations with surgery as salvage — with every number taken from the papers named beside it.^{[1][2][3][4][13]}

A 68-year-old man with melaena, a urea rise and a clean-based ulcer; a 54-year-old woman with a spurting duodenal ulcer that rebleeds after clips; and a 71-year-old man with complete oesophageal food-bolus obstruction. One needs stratification with discharge arithmetic, one needs failure escalation through clips to embolization or theatre, and one needs the clock — emergent scope within hours with medicines never delaying it. The examiner will watch you score Rockall, defend restrictive transfusion, quote Forrest rebleed bands, sequence PPI around the scope, predict adrenaline-heater-probe failure, choose clips over embolization by death arithmetic, band varices with stricture awareness, time foreign bodies, and justify PEG with infection numbers — with every number taken from the papers named beside it.^[1][13][16]

Rockall sorts the bleed from two English populations: 4,185 cases plus 1,625 validation cases with age, shock, comorbidity, diagnosis, major stigmata and rebleeding as independent death predictors — identifying 15% at presentation and 26% after endoscopy as low rebleed, negligible-death early discharges.^[1] Modern scores sharpen the pre-scope eye: 571 prospective ED patients at 34.9% high-risk Ia-IIb findings with CHAMPS beating Blatchford at 0.889 against 0.831 — complementary to established scores, never replacing them.^[22] Scores know their limits: across 795 ED patients at 142 interventions, Blatchford predicts intervention weakly while AIMS65 and ABC predict death best — with male, haematemesis, melaena, raised urea and low red-cell width flagging intervention.^[23] The strategic arc fits one sentence: stratify, transfuse restrictively, infuse PPI, scope within a day, treat by stigma, double-dose high Rockall, re-look selectively, clip refractory, embolize or operate failures, band varices, time foreign bodies, and stent selected perforations.^{[1][2][3][13][20]}

Stratify before scoping, transfuse restrictively, scope within a day

Transfusion is the first therapy and less is more: 921 randomised severe bleeders at 7 against 9 thresholds with half the restrictive arm never transfused — survival 95% against 91% with rebleed 10% against 16% and adverse events 40% against 48%.^[2] ESGE codifies it: restrictive haemoglobin between 7 and 9 with higher targets only for ischaemic comorbidity; Blatchford 0-1 discharges without early scope; high-dose IV PPI as 80 mg bolus plus 8 mg hourly without delaying the scope; erythromycin for severe active bleeds; scope within 24 hours of resuscitation.^[3]

Read Forrest and treat by stigma

High stigmata earn haemostasis: spurting, oozing and visible-vessel ulcers treated as high persistent-bleed risks.^[3] Low stigmata earn discharge: flat spots and clean bases untreated as low rebleed risks, home on once-daily oral PPI selected.^[3] Oozing dogma falls on numbers: placebo-arm rebleeds at 22.5% spurting, 17.6% clot, 11.3% vessel and 4.9% oozing — with oozing equal on PPI or placebo after haemostasis — so high-dose IV PPI after oozing haemostasis needs re-evaluation.^[4] The classification still predicts rebleed but never death: 397 registry patients at 18.6% rebleed with 59% in spurting disease and similar Ib-IIc odds — simplified to spurting / increased / low with matched test characteristics.^[5] The flat spot lies at high Rockall: 140 second-look IIc patients at 18.6% against 2.9% rebleed days 4-14 and 24.3% against 4.3% days 4-28 for Rockall 6-plus — so the low-risk label needs a Rockall check.^[6]

Infuse PPI around the scope, re-look selectively

Pre-scope omeprazole downgrades the lesion: 638 randomised bleeders at 80 mg bolus plus hourly infusion with endoscopic therapy needed in 19.1% against 28.4% — fewer active ulcers, more clean bases, shorter stays, equal transfusion, rebleed, surgery and death.^[7] Post-infusion double oral PPI protects high Rockall: double against standard oral esomeprazole after 3-day infusion with rebleed 10.8% against 28.7% days 4-28 at Rockall 6-plus.^[9] Second-look stays selective: 699 high-risk patients at 9.2% rebleed with NSAIDs, 5-plus-unit transfusion and skipped second-look as predictors — routine re-look off-guideline, selected re-look risk-lowering.^[10][3]

Predict failure and escalate through clips to embolization or theatre

Failure announces itself: 1,144 adrenaline-plus-heater-probe patients at 98.6% initial haemostasis with 8.2% rebleed — hypotension, low haemoglobin, fresh blood, active bleeding and large ulcers predicting — so big severe bleeders earn alternate plans upfront.^[8] Clips rescue recurrence: 66-patient 9-centre randomised trial of recurrent ulcer bleed with further bleeding 57.6% standard against 15.2% over-scope — surgery needed once per arm, deaths equal.^[13] Clips beat embolization on death: 128 propensity-adjusted refractory patients at comparable success with longer embolization ICU stays and matched death 22.5% against 5% — clips first where feasible, embolization or surgery on second failure per the ESGE ladder.^[14][3]

Band varices, never lone adrenaline

Banding wins the varix: 95 randomised bleeders with equal acute control — banding obliterates in fewer sessions with 6.4% against 20.8% rebleed and zero against 10.4% strictures — at the price of higher variceal recurrence at 28.7% against 7.5%.^[11] Combination adds nothing: 8-study meta-analysis of band-plus-sclero against band alone with no difference on rebleed, death or sessions and a stricture signal — so band alone stands.^[12] Adrenaline never stands alone: epinephrine forbidden as monotherapy, always paired with a second modality.^[3]

Time foreign bodies, scope caustic, survey Barrett

Foreign bodies run on hours: emergent within 6 hours for complete obstruction, sharp objects and batteries; urgent 24 hours for other oesophageal bodies; CT for suspected perforation; never barium; never let medicines delay the scope.^[16] Push the bolus first: gentle oesophageal push into stomach, retrieve on failure, always work up underlying disease histologically.^[16] Caustic sorts by stability: stable equivocal 3a disease gets diagnostic laparoscopy; unstable perforated 3b disease gets duodenal damage-control washout over primary closure on injured tissue.^[17] Barrett surveillance follows grade: post-ablation progression 6.3% high-grade against 1.3% low-grade with surveillance numbers-needed of 21 against 90 — retain high-grade and cancer in programmes.^[18]

Feed, pause and close perforations

PEG feeds the neurologically devastated: 644 single-centre patients at 71.5% neurological indications with 11.2% complications led by 5.7% peristomal infection — antibiotic absence and diabetes raising infection.^[19] Pause antithrombotics by risk: low-thrombotic high-risk procedures stop P2Y12 five days out; high-thrombotic cases keep aspirin with cardiology liaison; DOAC high-risk stops 48 hours out, low-risk skips the morning dose.^[15] Boerhaave splits three ways evenly: 216 European patients at endoscopic, surgical and combined thirds with 8-day ICU stays flat, 12% hospital and 90-day death, and longest stays combined.^[20] Iatrogenic perforation stents selected: 25 studies with 596 patients at 9.75% pooled death with interventional causes dominant — stenting matches surgery on death in selected patients with surgery as salvage.^[21]

The cohorts behind the numbers run Rockall 5,810 English bleeders, Villanueva 921 randomised transfusions, Gralnek ESGE panel, Jensen international oozing reassessment, de Groot 397-registry simplification, Yang 140-patient IIc cohort, Lau 638-patient pre-scope PPI trial, Wong 1,144-patient failure cohort, Cheng high-Rockall double-PPI randomisation, Kim 699-patient second-look cohort, Sarin 95-patient variceal trial, Karsan 8-study combination meta-analysis, Schmidt 66-patient STING randomisation, Kuellmer 128-patient clip-embolization comparison, Veitch antithrombotic panel, Birk foreign-body panel, Di Saverio caustic series, Cotton US RFA registry model, Turan 644 PEG patients, Hauge 216-patient Boerhaave cohort, Papaconstantinou 596-patient perforation review, Önür 571-patient score comparison, and Işık 795-patient score comparison — randomised where randomisable, nationwide and multicentre where rare, pooled where small.^{[1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23]}

Stratify, infuse, scope, treat by stigma, double-dose high Rockall, re-look selectively, clip refractory before embolizing, band varices alone, time foreign bodies by the hour, scope caustic by stability, survey Barrett by grade, feed neurologically by PEG, pause antithrombotics by risk, and stent selected perforations with surgery as salvage.^{[1][3][6][10][13][11][16][18][19][15][21]}

140-patient second-look cohort at IIc after 48-72 h infusion; Rockall 6-plus rebleeds 18.6 vs 2.9% at days 4-14 and 24.3 vs 4.3% at days 4-28 — high Rockall rescues the low-risk label (PMID 31222633).^[3] Predict failure exactly: 1,144-patient adrenaline-plus-heater-probe cohort at 98.6% initial haemostasis with 8.2% rebleed and 9.6% failure; hypotension, low haemoglobin, fresh blood, active bleeding and large ulcers predict — big severe bleeders fail (PMID 11839708).^[8] Clip refractory exactly: 66-patient 9-centre randomised trial of recurrent ulcer bleed; further bleeding 57.6 vs 15.2% favouring over-scope clips with 42.4% absolute difference — surgery needed in one patient per arm (PMID 29803838).^[13] Embolize vs clip exactly: 128-patient propensity-adjusted refractory comparison at 74.2 vs 59.7% success; embolization costs ICU days and matched death at 22.5 vs 5% — clips first where feasible (PMID 34432392).^[14] Price Boerhaave exactly: 216-patient 23-centre European cohort at 32% endoscopic, 34% surgical, 31% combined; ICU 8 days flat with 12% hospital and 90-day death and longest stays combined (PMID 41324551).^[20] Stent selected exactly: 25-study 596-patient iatrogenic review at 9.75% pooled death with 71% interventional causes; stenting 11.37% vs surgery 11.58% death — endoscopy central, surgery salvage (PMID 42377532).^[21]

References23ShowHide

1. [1]Rockall TA, et al. Risk assessment after acute upper gastrointestinal haemorrhage. Gut, 1996.PMID 8675081
2. [2]Villanueva C, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med, 2013.PMID 23281973
3. [3]Gralnek IM, et al. Diagnosis and management of nonvariceal upper gastrointestinal hemorrhage: European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Endoscopy, 2015.PMID 26417980
4. [4]Jensen DM, et al. Reassessment of Rebleeding Risk of Forrest IB (Oozing) Peptic Ulcer Bleeding in a Large International Randomized Trial. Am J Gastroenterol, 2017.PMID 28094314
5. [5]de Groot NL, et al. Reassessment of the predictive value of the Forrest classification for peptic ulcer rebleeding and mortality: can classification be simplified? Endoscopy, 2014.PMID 24218308
6. [6]Yang EH, et al. The recurrent bleeding risk of a Forrest IIc lesion at the second-look endoscopy can be indicated by high Rockall scores ≥6. Surg Endosc, 2020.PMID 31222633
7. [7]Lau JY, et al. Omeprazole before endoscopy in patients with gastrointestinal bleeding. N Engl J Med, 2007.PMID 17442905
8. [8]Wong SK, et al. Prediction of therapeutic failure after adrenaline injection plus heater probe treatment in patients with bleeding peptic ulcer. Gut, 2002.PMID 11839708
9. [9]Cheng HC, et al. Double oral esomeprazole after a 3-day intravenous esomeprazole infusion reduces recurrent peptic ulcer bleeding in high-risk patients: a randomised controlled study. Gut, 2014.PMID 24658598
10. [10]Kim SB, et al. Risk Factors Associated with Rebleeding in Patients with High Risk Peptic Ulcer Bleeding: Focusing on the Role of Second Look Endoscopy. Dig Dis Sci, 2016.PMID 26297133
11. [11]Sarin SK, et al. Prospective randomized trial of endoscopic sclerotherapy versus variceal band ligation for esophageal varices: influence on gastropathy, gastric varices and variceal recurrence. J Hepatol, 1997.PMID 9126795
12. [12]Karsan HA, et al. Combination endoscopic band ligation and sclerotherapy compared with endoscopic band ligation alone for the secondary prophylaxis of esophageal variceal hemorrhage: a meta-analysis. Dig Dis Sci, 2005.PMID 15745108
13. [13]Schmidt A, et al. Over-the-Scope Clips Are More Effective Than Standard Endoscopic Therapy for Patients With Recurrent Bleeding of Peptic Ulcers. Gastroenterology, 2018.PMID 29803838
14. [14]Kuellmer A, et al. Over-the-scope clip versus transcatheter arterial embolization for refractory peptic ulcer bleeding-A propensity score matched analysis. United European Gastroenterol J, 2021.PMID 34432392
15. [15]Veitch AM, et al. Endoscopy in patients on antiplatelet or anticoagulant therapy, including direct oral anticoagulants: British Society of Gastroenterology (BSG) and European Society of Gastrointestinal Endoscopy (ESGE) guidelines. Endoscopy, 2016.PMID 26890676
16. [16]Birk M, et al. Removal of foreign bodies in the upper gastrointestinal tract in adults: European Society of Gastrointestinal Endoscopy (ESGE) Clinical Guideline. Endoscopy, 2016.PMID 26862844
17. [17]Di Saverio S, et al. Different possible surgical managements of caustic ingestion: diagnostic laparoscopy for Zargar's grade 3a lesions and a new technique of Duodenal Damage Control with 4-tubes ostomy and duodenal wash-out as an option for extensive 3b lesions in unstable patients. Updates Surg, 2015.PMID 26141256
18. [18]Cotton CC, et al. Surveillance After Treatment of Barrett's Esophagus Benefits Those With High-Grade Dysplasia or Intramucosal Cancer Most. Am J Gastroenterol, 2022.PMID 35435856
19. [19]Turan UF, et al. Evaluation of 644 Percutaneous Endoscopic Gastrostomy Patients in a Single Center. Cureus, 2023.PMID 37261172
20. [20]Hauge T, et al. Treatment strategies for Boerhaave syndrome: multinational retrospective cohort study. Br J Surg, 2025.PMID 41324551
21. [21]Papaconstantinou D, et al. Management of iatrogenic esophageal perforations: a systematic review of non-surgical causes. Surg Endosc, 2026.PMID 42377532
22. [22]Önür A, et al. Comparison of pre-endoscopic risk scores for identifying high-risk Forrest lesions in emergency department patients with suspected upper gastrointestinal bleeding. Am J Emerg Med, 2026.PMID 42632219
23. [23]Işık Nİ, et al. Comparative Performance Analysis of Aims65, Pre-Endoscopic Rockall, Glasgow-Blatchford, and Modified Age Blood Test Comorbidity Scores in Emergency Department Non-Variceal Upper Gastrointestinal Bleeding. J Emerg Med, 2025.PMID 40774003