The Forrest Classification: The Evidence Behind the Tool

It is half past eleven at night in the endoscopy room. A 46-year-old lorry driver from Hubballi has had black stools for two days. For a week before that he took a painkiller for his back, bought at a pharmacy without a prescription. His haemoglobin is 8.1 g/dL. In the duodenal bulb there is an ulcer, and on its floor sits a dark clot. The resident beside you says, “That is a IIb, adherent clot. Do we leave it?” This is an illustrative case, not a real patient.

The Forrest Classification tool takes what you see at endoscopy, gives the class, and prints what the European and American guidelines recommend for it: the endoscopic therapy, the acid suppression, the Helicobacter test and the next seventy-two hours. It keeps every one of those. It leaves out where the classes came from and how reliably they are applied. It also leaves out what the registries found, and why the acid and the adrenaline matter. Those are here. By the end you will know why the resident should wash that clot before naming it, and what the class can and cannot tell you about this man.

This article is about the class and the ulcer. For the whole bleed, the scores before endoscopy, the drugs, varices and discharge, read Upper GI Bleeding: The Evidence Behind the Pathway.

1. Where the six classes came from

James Forrest, Niall Finlayson and David Shearman published Endoscopy in gastrointestinal bleeding in the Lancet in August 1974. It is indexed as a comparative study of endoscopy against barium radiology, not as a proposal for a grading scheme, and the six-class lettering used every day, Ia through III, is not the lettering in that paper. It accumulated afterwards, in the hands of the people who needed to say which ulcers to inject.

“So who decided what each class gets?” Not the 1974 paper. That is why the tool attributes every recommendation to the guideline that prints it: the European Society of Gastrointestinal Endoscopy (ESGE) guideline of 2021 with its 2026 update, and the American College of Gastroenterology (ACG) guideline of 2021. Both use the modern lettering, Ia to III, and both were read at source.

2. Two endoscopists, one ulcer

Before you trust a class, ask how often two experienced people give the same one. Forty-seven experienced endoscopists in Piedmont classified the same 25 videotaped examinations (Mondardini 1998). Agreement beyond chance was good where blood was moving and poor where it was not.

The number used is kappa. It measures how much two observers agree beyond what chance alone would give. A kappa of 1 is perfect agreement; 0 is no better than tossing a coin. Above about 0.6 is usually read as good, and 0.4 to 0.6 as moderate.

The authors opened by pointing out that the wide range of published rebleeding rates for the same stigma is itself evidence that the definitions are not applied uniformly. That is the honest frame for every percentage in this tool, including the ones it prints.

Look at where the weak agreement sits: the stigmata classes, IIa to IIc, which are exactly the classes our lorry driver could fall into. A clot is the easiest thing to misread. That is why the tool insists on one step before you classify: wash the clot vigorously. Studies that irrigated properly found most clots come away, and what is underneath decides the class.

3. What the class predicts, and what it does not

A prospective registry of 397 patients with bleeding peptic ulcers, followed for 30 days, gives the most quotable modern figures (de Groot 2014). Rebleeding occurred in 74 of the 397, 18.6 per cent overall, and was highest in Forrest Ia at 59 per cent. This is how often each class was seen.

Three findings from the same registry are worth carrying in your head.

• The odds of rebleeding among Ib, IIa, IIb and IIc were similar to one another. The middle of the scale is not a graded slope; it is a plateau.
• Prediction was more reliable for gastric ulcers than for duodenal ones.
• The authors proposed collapsing the scheme to three bands, high risk Ia alone, increased risk Ib through IIc, low risk III. It has not been adopted, and both current guidelines still write the six classes.

“If the class predicts rebleeding, surely it predicts death?” It does not. The same registry found the classification did not predict mortality at all. A patient with a clean base can still die of the illness that put the ulcer there. So the tool prints no mortality figure by class. What decides whether this man survives is his age, his other illnesses and how much blood he has lost, not the letter on the ulcer.

4. Why 30 mL of adrenaline, and why never adrenaline alone

Injected adrenaline works by volume tamponade and by narrowing the vessel, and both wear off. It clears the field and buys time, but on its own it leaves the artery as it was. That is why both guidelines want a second modality with it, thermal or mechanical.

No guideline read for this tool gives a dose. The 30 mL of 1:10,000 the tool prints comes from a randomised trial of 228 patients (Liou 2006). Twenty millilitres left a rebleeding rate of 20.3 per cent, against 5.3 per cent at 30 mL. Forty millilitres bought a further 2.5 percentage points, at the price of a significantly higher perforation rate.

“So why not just inject more?” Because the gain from 30 to 40 mL is small and the perforation is not. Thirty millilitres is where the benefit and the harm balance.

5. Why the acid matters for seventy-two hours

Here is the mechanism behind the high-dose proton pump inhibitor. Clot stability depends on gastric pH. Pepsin digests a clot below about pH 6 and platelet aggregation fails below about pH 5.4, so profound and continuous acid suppression is not adjuvant treatment here: it is what keeps the endoscopic result in place while the vessel organises.

Think of the clot over the treated vessel as wet cement. It needs a few days to set, and acid keeps washing it away. Seventy-two hours is the window in which almost all rebleeding happens, so that is how long the acid must be held down.

“Does it have to be an infusion?” Not necessarily. For intermittent dosing, a meta-analysis of randomised trials in endoscopically treated high-risk ulcers found it non-inferior, with a risk ratio for seven-day rebleeding of 0.72 and an absolute difference of minus 2.64 per cent, comfortably inside the pre-specified margin. (Sachar 2014). Non-inferior means the trials were designed to show that the simpler regimen is not meaningfully worse, and it was not. Both 2021 guidelines now permit either. If your ward cannot run a syringe pump reliably for three days, the intermittent regimen is a defensible choice.

6. The urease test that came back negative

Our lorry driver has a rapid urease test on a biopsy from the antrum, and it is negative. The common mistake is to write “H. pylori negative” in the notes and move on. It is easy to see why: the test is quick and the result looks definite. But blood in the stomach and the proton pump inhibitor already running both suppress the organism and the test. A meta-analysis pooled every method in this specific setting (Gisbert 2006).

Sensitivity is the share of infected patients the test calls positive. Specificity is the share of uninfected patients it calls negative. A test with a sensitivity of 0.67 misses one infected patient in three.

Hence the European recommendation to retest anyone negative at the index endoscopy. In a series of patients whose in-hospital breath test was negative, 15 of 18 turned positive when repeated after stopping the acid suppression, and the only factor associated with the false negative was how long after admission the test was done. Eradication is what stops this ulcer bleeding again, so a missed diagnosis is a readmission later.

7. What we know from India

A search of the published literature on 3 September 2026 found no Indian cohort reporting outcomes by Forrest class. So every per-class figure above is European. What Indian data there is describes the illness rather than the class. A prospective series of 138 adults at a teaching hospital in Dehradun found (Singh 2026):

• Peptic ulcer disease the commonest source at 56.5 per cent, ahead of variceal bleeding at 25.4 per cent.
• Most patients between 31 and 50 years old.
• Alcohol use in 43.5 per cent, NSAID exposure in 36.2 per cent.
• Haemoglobin below 7 g/dL in 29 per cent, and 63.8 per cent transfused.
• Endoscopic haemostasis successful in 87 per cent, 13 per cent rebled, 9.4 per cent died.

Notice the age. These patients are a generation younger than the European registry patients, and they bleed after alcohol and painkillers. Ask about painkillers by name. Our lorry driver did not think of the tablets from the pharmacy as medicine until he was asked.

Back to the endoscopy room

The resident washes the clot firmly. It lifts, and underneath is a raised, pigmented protuberance on the ulcer floor. That is a non-bleeding visible vessel, Forrest IIa, not an adherent clot. The plan changes from “do we leave it?” to endoscopic therapy: thermal, mechanical or a sclerosant, alone or with adrenaline, and a high-dose proton pump inhibitor for seventy-two hours. His negative urease test is recorded as a test, a result and a date, with a breath test booked after the acid suppression stops. And his painkiller goes in the notes by name.

Try this

A 52-year-old woman has a gastric ulcer with a flat, pigmented spot on its base. Nothing is raised, and washing does not change it. What class is it, and does it need endoscopic therapy? Illustrative case.

Answer. A flat pigmented spot is Forrest IIc. The distinction from a visible vessel is elevation, not colour. The European guideline recommends against endoscopic haemostasis for a flat pigmented spot or a clean base, so she needs no endoscopic therapy, an oral proton pump inhibitor, and a Helicobacter test that is believed only if it is positive.

Take-home points

• Wash the clot before you name the class; what lies beneath decides the treatment.
• Agreement between endoscopists is good for bleeding lesions and only moderate for the stigmata classes.
• The class predicts rebleeding, mainly for Ia, and does not predict death.
• Adrenaline is never treatment on its own; 30 mL of 1:10,000 with a second modality, then seventy-two hours of high-dose acid suppression, infusion or intermittent.
• A negative urease test during a bleed is not a diagnosis; retest, and ask about painkillers by name.

Declaration: the author built and maintains the tool. It is free to use. The figures in this article are those the tool carried in its Evidence tab on 4 November 2026, each from the source the tool names. The illustrative cases follow the tool’s rules; check the tool’s result for a real patient.

References

1. Forrest JA, Finlayson ND, Shearman DJ. Endoscopy in gastrointestinal bleeding. Lancet. 1974;2(7877):394-397.
2. Gralnek IM, Morris J, Laursen SB, et al. Endoscopic diagnosis and management of peptic ulcer bleeding: European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Update 2026. Endoscopy. 2026;58(8):899-924.
3. Correction: Endoscopic diagnosis and management of peptic ulcer bleeding: European Society of Gastrointestinal Endoscopy (ESGE) Guideline - Update 2026. Endoscopy. 2026;58(8):C12.
4. Gralnek IM, Stanley AJ, Morris AJ, et al. Endoscopic diagnosis and management of nonvariceal upper gastrointestinal hemorrhage (NVUGIH): European Society of Gastrointestinal Endoscopy (ESGE) Guideline. Update 2021. Endoscopy. 2021;53(3):300-332.
5. Laine L, Barkun AN, Saltzman JR, Martel M, Leontiadis GI. ACG Clinical Guideline: Upper Gastrointestinal and Ulcer Bleeding. Am J Gastroenterol. 2021;116(5):899-917.
6. Barkun AN, Almadi M, Kuipers EJ, et al. Management of Nonvariceal Upper Gastrointestinal Bleeding: Guideline Recommendations From the International Consensus Group. Ann Intern Med. 2019;171(11):805-822.
7. de Groot NL, van Oijen MGH, Kessels K, et al. Reassessment of the predictive value of the Forrest classification for peptic ulcer rebleeding and mortality: can classification be simplified? Endoscopy. 2014;46(1):46-52.
8. Kahi CJ, Jensen DM, Sung JJY, et al. Endoscopic therapy versus medical therapy for bleeding peptic ulcer with adherent clot: a meta-analysis. Gastroenterology. 2005;129(3):855-862.
9. Mondardini A, Barletti C, Rocca G, et al. Non-variceal upper gastrointestinal bleeding and Forrest’s classification: diagnostic agreement between endoscopists from the same area. Endoscopy. 1998;30(6):508-512.
10. Jensen DM, Kovacs TOG, Ohning GV, et al. Doppler endoscopic probe monitoring of blood flow improves risk stratification and outcomes of patients with severe nonvariceal upper gastrointestinal hemorrhage. Gastroenterology. 2017;152(6):1310-1318.
11. Sachar H, Vaidya K, Laine L. Intermittent vs continuous proton pump inhibitor therapy for high-risk bleeding ulcers: a systematic review and meta-analysis. JAMA Intern Med. 2014;174(11):1755-1762.
12. Gisbert JP, Abraira V. Accuracy of Helicobacter pylori diagnostic tests in patients with bleeding peptic ulcer: a systematic review and meta-analysis. Am J Gastroenterol. 2006;101(4):848-863.
13. Liou TC, Lin SC, Wang HY, Chang WH. Optimal injection volume of epinephrine for endoscopic treatment of peptic ulcer bleeding. World J Gastroenterol. 2006;12(19):3108-3113.
14. Koh JH, Anna O, Teng JJR, et al. Over-the-scope clip versus standard endoscopic treatment in patients with acute nonvariceal upper gastrointestinal bleeding: a systematic review and meta-analysis. J Gastroenterol Hepatol. 2025;40(10):2373-2390.
15. Singh S, Akram M, Singh S. Predictors of mortality and rebleeding in acute upper gastrointestinal bleeding: a prospective hospital-based study. Bioinformation. 2026;22(2):1109-1112.
16. Rout G, Sharma S, Gunjan D, Kedia S, Nayak B, Shalimar. Comparison of various prognostic scores in variceal and non-variceal upper gastrointestinal bleeding: a prospective cohort study. Indian J Gastroenterol. 2019;38(2):158-166.