Variceal Bleeding: The Liver Emergency That Cannot Wait — and How to Prevent It
Variceal bleeding is a medical emergency. If someone with cirrhosis vomits blood or passes black, tarry stools, a swollen vein in the food pipe has most likely burst — go to a hospital with an endoscopy unit immediately. Even with modern treatment, about 15–20% of patients die within six weeks of a bleed. Most first bleeds can be prevented with beta-blockers or band ligation.
— Dr Chetan Kalal, DM Hepatology, Gleneagles Hospital Mumbai
Key takeaways
- Varices are swollen veins in the oesophagus or stomach caused by portal hypertension, and they cause no symptoms until they bleed.
- Vomiting blood, black stools or sudden dizziness in a person with cirrhosis is an emergency — the first hour belongs to a hospital.
- Six-week mortality after a variceal bleed is around 15–20%. A MELD score of 19 or more predicts 20% or higher; a MELD below 11 predicts under 5%.
- Four things happen within hours of admission: a drug to lower portal pressure, an antibiotic, cautious transfusion, and endoscopy with band ligation within 12 hours.
- In the highest-risk patients, an early TIPS raised one-year survival from 61% to 86% in the landmark randomised trial.
- Prevention works: beta-blockers (carvedilol preferred) cut decompensation from 27% to 16% in the PREDESCI trial; after a bleed, beta-blocker plus banding is standard.
A composite of a call I receive several times a year: a 48-year-old man with alcohol-related cirrhosis vomits a basin of dark blood at two in the morning. His family’s first instinct is to wait until the clinic opens. Their second is to drive him across the city to the hospital where he is “known”. Both instincts are understandable. Both are wrong.
Variceal bleeding — haemorrhage from swollen veins in the oesophagus or stomach — is the most dramatic complication of cirrhosis and one of the few in hepatology where the clock runs in minutes. It is also, and patients rarely hear this, largely preventable. This article covers how to recognise a bleed, what a hospital will do in the first twelve hours, how dangerous the episode is, and the two layers of prevention I use in my practice at Gleneagles Hospital, Mumbai — written for patients with cirrhosis and for the families who will make the two-in-the-morning decision.
What are oesophageal varices, and why do they bleed?
Varices are enlarged veins in the lower oesophagus or upper stomach, and they exist for one reason: the liver has become too stiff for blood to pass through it easily. Blood from the intestines and spleen normally drains through the liver via the portal vein. In cirrhosis, scar tissue raises the resistance, pressure in the portal system climbs — this is portal hypertension — and blood looks for other routes back to the heart. The thin veins beneath the lining of the oesophagus become one of those routes, and they were never built for that pressure — so they stretch, swell and eventually rupture.
A varix does not hurt — no heartburn, no difficulty swallowing, nothing. Most patients have no idea they have varices until an endoscopy finds them — or until one bleeds. That silence is why screening matters, and why I spend so much of every cirrhosis consultation on it.
The pressure, not the vein, is the real target. The Baveno VII consensus (Journal of Hepatology, 2022) centres everything on “clinically significant portal hypertension”, a portal pressure gradient of 10 mmHg or more — the threshold above which varices form, ascites follows and the risk of death rises. That is why a drug that lowers portal pressure can protect a patient who has never had a single varix banded.
What are the warning signs of variceal bleeding?
Three signs, and any one of them is enough. First, vomiting blood — fresh red, or dark and granular like coffee grounds. Second, black, sticky, foul-smelling stools, which are digested blood from higher up the gut. Third, sudden dizziness, fainting or a racing pulse with cold, clammy skin — enough blood lost to drop the blood pressure. Blood vomiting in liver cirrhosis is variceal until proven otherwise, and even if it turns out to be an ulcer, the emergency is the same.
Do not wait to see whether it settles. Call an ambulance or go to the nearest hospital with an emergency department and an endoscopy service, and say the words “cirrhosis” and “varices” at the door — it changes what the team does in the first ten minutes. A patient actively vomiting blood may need the airway protected before endoscopy — Baveno VII recommends intubation in exactly this situation — and no clinic can offer that.
Two points families miss. A bleed can begin quietly — a single black stool, a day of weakness, a faint on the way to the bathroom — and these “herald” bleeds deserve the same urgency because the next one may be large. And confusion or drowsiness after a bleed is not tiredness: blood in the gut produces ammonia, and that encephalopathy needs hospital treatment in its own right.
What happens in hospital when varices bleed?
Four interventions, run in parallel, within hours. Baveno VII and the 2024 AASLD guidance agree on all four.
A drug to lower portal pressure. Vasoactive medicines (terlipressin, somatostatin or octreotide, per Baveno VII) are started the moment variceal bleeding is suspected — before the endoscopy, not after — and continued for two to five days.
An antibiotic. Bleeding lets gut bacteria into the bloodstream, and infection kills many of these patients in the days that follow. An updated Cochrane analysis of 12 randomised trials and 1,241 patients found that antibiotic prophylaxis cut bacterial infections by about two-thirds (relative risk 0.35), reduced rebleeding (relative risk 0.53) and lowered mortality (relative risk 0.79). It is given from admission, not once a fever appears.
Cautious blood transfusion. Counter-intuitively, over-transfusing makes variceal bleeding worse, because extra volume raises portal pressure. In a randomised trial of 921 patients with severe upper gastrointestinal bleeding (NEJM, 2013), those transfused only when haemoglobin fell below 7 g/dL had better six-week survival than those transfused below 9 g/dL (95% versus 91%) and less further bleeding (10% versus 16%). A lower haemoglobin than you expect is deliberate, not neglect.
Endoscopy within 12 hours. Once the patient is resuscitated, an upper GI endoscopy is performed within 12 hours of presentation — sooner if unstable. The endoscopist places small rubber bands over the bleeding varix (band ligation), strangling the vessel and stopping the flow; gastric varices are treated with tissue glue instead. A CT scan follows to exclude a portal vein clot or liver cancer as the trigger.
How dangerous is a variceal bleed — and who needs an early TIPS?
Dangerous. Even with everything above done well, roughly 15–20% of patients die within six weeks of a variceal bleed. In a Barcelona cohort of 178 patients, six-week mortality was 16%, and the MELD score at admission predicted it: a MELD of 19 or above carried 20% or higher mortality, a MELD below 11 less than 5%.
That stratification matters because the highest-risk group has a treatment that changes the outcome. A TIPS (transjugular intrahepatic portosystemic shunt) is a covered stent placed through the neck veins into the liver, letting portal blood bypass the scarred tissue and dropping the pressure at once. Used early — a “pre-emptive” TIPS within 72 hours, ideally within 24 — it works. In the landmark randomised trial (NEJM, 2010) of 63 high-risk patients, one-year survival was 86% with early TIPS against 61% with standard drug-plus-banding therapy, and 97% of the TIPS group remained free of rebleeding against 50%.
Baveno VII defines “high-risk” precisely: Child-Pugh C cirrhosis scoring below 14, or Child-Pugh B above 7 with active bleeding at the first endoscopy. The decision is made on the day of admission — one more reason the right hospital matters, since not every centre can place a TIPS at short notice. A variceal bleed is also a decompensating event, and Baveno VII states plainly that patients with decompensated cirrhosis should be considered for liver transplantation — a conversation for the weeks after the bleed, not years later.
Can variceal bleeding be prevented?
Yes — and this is where the field has moved most. Baveno VII changed the question: the goal is no longer “prevent the first bleed” but “prevent decompensation of any kind” — ascites, encephalopathy or bleeding — and the tool is a non-selective beta-blocker started as soon as clinically significant portal hypertension is identified, whether or not varices have yet appeared.
The evidence is the PREDESCI trial (Lancet, 2019). Two hundred and one patients with compensated cirrhosis and significant portal hypertension, but no high-risk varices, were randomised to a beta-blocker (propranolol or carvedilol) or placebo. Decompensation or death occurred in 16% on beta-blockers against 27% on placebo — a hazard ratio of 0.51 — driven mainly by less ascites. A 2022 individual-patient meta-analysis of four randomised trials (352 patients) then showed that carvedilol halved the risk of decompensation (sub-hazard ratio 0.506) and reduced death (sub-hazard ratio 0.417) compared with no treatment or banding alone.
Carvedilol for varices is therefore my default, and Baveno VII’s: it lowers portal pressure more effectively than propranolol, tends to be better tolerated, and has survival data behind it. Banding for varices is reserved for patients who cannot take a beta-blocker and who have large or red-signed varices on endoscopy. Dose and titration belong to your hepatologist; do not start, stop or adjust a beta-blocker on the strength of this article.
After a bleed, the rules tighten. Baveno VII’s first-line therapy to prevent a second haemorrhage is a beta-blocker and band ligation repeated until the varices are gone; TIPS is the treatment of choice if bleeding recurs despite both. And in alcohol-related cirrhosis, no tablet or band matches complete abstinence — I set out why in my article on alcohol-related liver disease.
| Situation | Baveno VII recommendation (2022) | Key evidence |
|---|---|---|
| Compensated cirrhosis with clinically significant portal hypertension, no previous bleed | Non-selective beta-blocker, carvedilol preferred, to prevent decompensation; screening endoscopy not needed once on treatment | PREDESCI: decompensation or death 16% vs 27% (HR 0.51); carvedilol meta-analysis: decompensation SHR 0.506, death SHR 0.417 |
| High-risk varices in a patient who cannot take a beta-blocker | Endoscopic band ligation | Baveno VII statement 5.19 |
| After a first variceal bleed (secondary prophylaxis) | Beta-blocker (traditional or carvedilol) plus band ligation | Baveno VII statement 7.8 |
| Rebleeding despite beta-blocker plus banding | TIPS | Baveno VII statement 7.9 |
| Acute bleed in Child-Pugh C (<14) or Child-Pugh B (>7) with active bleeding at endoscopy | Pre-emptive TIPS within 72 h, ideally <24 h | NEJM 2010 RCT: 1-year survival 86% vs 61% |
Do I need a screening endoscopy?
Often, no. Baveno VII allows a FibroScan reading and a platelet count to stand in for the endoscope in two directions. A liver stiffness of 15 kPa or below together with a platelet count of 150 × 10⁹/L or more rules out clinically significant portal hypertension, with a negative predictive value above 90%. At the other end, a stiffness of 25 kPa or more is enough to rule it in — and therefore to start a beta-blocker — in viral, alcohol-related and non-obese fatty-liver cirrhosis.
In between, it depends on whether a beta-blocker is on board: if you take one to prevent decompensation, Baveno VII says a screening endoscopy will not change management and is not required. If you cannot take one, endoscopy is needed when stiffness is 20 kPa or more or platelets are 150 × 10⁹/L or lower; below both thresholds, a yearly FibroScan and platelet count suffice. These cut-offs apply to compensated cirrhosis only — anyone with ascites, a previous bleed or encephalopathy sits outside them.
What I tell every patient with varices: keep the beta-blocker going unless a doctor stops it — the tablet you feel nothing from is the one doing the work. Know the three warning signs, and decide which hospital you will go to before you need it. And if you drink, stop.
Frequently asked questions
Is vomiting blood with liver cirrhosis always from varices?
No, but it must be treated as variceal until an endoscopy proves otherwise. Ulcers and portal hypertensive gastropathy also bleed in cirrhosis. Whatever the source, vomiting blood or passing black stools with cirrhosis is an emergency: go to a hospital with an endoscopy service immediately, by ambulance if possible.
What is the survival rate after variceal bleeding?
About 80–85% of patients survive the six weeks after a variceal bleed with modern treatment; roughly 15–20% do not. Risk depends on liver function: a MELD score of 19 or more predicts 20% or higher six-week mortality, while a MELD below 11 predicts under 5%. Early TIPS in the highest-risk group raised one-year survival from 61% to 86% in a randomised trial.
Is carvedilol better than propranolol for varices?
Baveno VII names carvedilol as the preferred beta-blocker in compensated cirrhosis because it lowers portal pressure more effectively and tends to be better tolerated. An individual-patient meta-analysis of four randomised trials found carvedilol halved decompensation and reduced death compared with no treatment or banding alone. The choice and dose belong to your hepatologist.
Can I avoid a screening endoscopy if my FibroScan is low?
Often, yes. Under Baveno VII, a liver stiffness of 15 kPa or below with a platelet count of 150 × 10⁹/L or more rules out clinically significant portal hypertension. Patients already on a beta-blocker do not need a screening endoscopy. Those who cannot take one need endoscopy if stiffness is 20 kPa or more or platelets are 150 × 10⁹/L or lower.
Does a variceal bleed mean I need a liver transplant?
Not automatically, but a variceal bleed is a decompensating event, and Baveno VII states that patients with decompensated cirrhosis should be considered for liver transplantation. Your hepatologist will weigh liver function, MELD score, the cause of cirrhosis and your response to treatment, and the discussion belongs in the weeks after the bleed, not years later.
Book a consultation with Dr Chetan Kalal at Gleneagles Hospital, Mumbai — call +91 83294 76669 or use the appointment form on this site. International and NRI patients may request a teleconsultation.
References
- de Franchis R, Bosch J, Garcia-Tsao G, Reiberger T, Ripoll C; Baveno VII Faculty. Baveno VII – Renewing consensus in portal hypertension. J Hepatol. 2022;76(4):959–974. PMID 35120736. doi:10.1016/j.jhep.2021.12.022
- Kaplan DE, Ripoll C, Thiele M, et al. AASLD Practice Guidance on risk stratification and management of portal hypertension and varices in cirrhosis. Hepatology. 2024;79(5):1180–1211. PMID 37870298. doi:10.1097/HEP.0000000000000647
- Reverter E, Tandon P, Augustin S, et al. A MELD-based model to determine risk of mortality among patients with acute variceal bleeding. Gastroenterology. 2014;146(2):412–419.e3. PMID 24148622. doi:10.1053/j.gastro.2013.10.018
- Villanueva C, Colomo A, Bosch A, et al. Transfusion strategies for acute upper gastrointestinal bleeding. N Engl J Med. 2013;368(1):11–21. PMID 23281973. doi:10.1056/NEJMoa1211801
- Chavez-Tapia NC, Barrientos-Gutierrez T, Tellez-Avila F, et al. Meta-analysis: antibiotic prophylaxis for cirrhotic patients with upper gastrointestinal bleeding – an updated Cochrane review. Aliment Pharmacol Ther. 2011;34(5):509–518. PMID 21707680. doi:10.1111/j.1365-2036.2011.04746.x
- García-Pagán JC, Caca K, Bureau C, et al. Early use of TIPS in patients with cirrhosis and variceal bleeding. N Engl J Med. 2010;362(25):2370–2379. PMID 20573925. doi:10.1056/NEJMoa0910102
- Villanueva C, Albillos A, Genescà J, et al. β blockers to prevent decompensation of cirrhosis in patients with clinically significant portal hypertension (PREDESCI): a randomised, double-blind, placebo-controlled, multicentre trial. Lancet. 2019;393(10181):1597–1608. PMID 30910320. doi:10.1016/S0140-6736(18)31875-0
- Villanueva C, Torres F, Sarin SK, et al. Carvedilol reduces the risk of decompensation and mortality in patients with compensated cirrhosis in a competing-risk meta-analysis. J Hepatol. 2022;77(4):1014–1025. PMID 35661713. doi:10.1016/j.jhep.2022.05.021




