Home Feature Upper GI bleeding: Who’s at risk for rebleeds?

Upper GI bleeding: Who’s at risk for rebleeds?

December 2007

Published in the December 2007 issue of Today’s Hospitalist

Here is the good news about upper GI bleeds: As many as 80% to 85% of these patients stop bleeding on their own.

The key issue for hospitalists becomes how to identify those patients and get them out of the hospital as quickly as possible. The other big concern is how to detect and treat the 15% to 20% of patients who are likely to rebleed.

A growing body of evidence supports the practice of treating first-time, low-risk non-variceal bleeders by discharging them as soon as their endoscopic workup is completed, explained Peter K. Lindenauer, MD, a hospitalist and medical director of clinical and quality informatics at Baystate Medical Center in Springfield, Mass.

Speaking at a Spring 2007 Hospitalist CME Series meeting earlier this year, Dr. Lindenauer pointed out that 90% of all rebleeds happen within 72 hours of hospital admission.

Any patient not determined to have low-risk lesions should be watched in the hospital for two to three days, he said. However, quick diagnosis and the timely use of both endoscopic and non-endoscopic therapies can reduce patients’ length of stay.

Who’s at low risk?
Dr. Lindenauer, who is associate professor of medicine at Tufts University School of Medicine, painted a relatively bleak picture of upper GI bleeds. While the condition is less common than heart attacks or pneumonia, upper gastrointestinal bleeding results in 250,000 hospitalizations a year in the U.S.

More common in men than in women, its primary cause is peptic ulcers. Another 11% are due to varices, 10% to esophagitis and ulcers, 7% to Mallory-Weiss tears, and 6% to gastritis and duodenitis. Only 5% are mass lesions, while bleeding occurs as a complication of hospitalization in about 10% of all cases.

To judge which patients are likely to rebleed, Dr. Lindenauer recommended focusing on key comorbidities, such as liver disease, renal failure or cancer. Look for any history of prior abdominal aortic surgery, chest pain or abdominal pain, and whether patients are having chest pain now.

The severity of bleeding matters, and you should consider the possibility of a non-GI bleeding source. “Every year there are one or two patients at our hospital who are admitted as upper GI bleeders who turn out to have a nasopharyngeal source of their bleeding,” he said.

Limit lab tests to hematocrit, platelets, creatinine, liver function and INR. An electrocardiogram is important because cardiac ischemia is a common complication.

And because nasogastric aspiration has a high level of false negative findings “10% to 15% by most accounts “a negative NG aspirate doesn’t rule out a diagnosis of upper GI bleed. (A positive NG, however, does rule it in.) Dr. Lindenauer recommended reserving abdominal X-rays for patients with pain or peritoneal findings.

Signs that a patient has a low risk of rebleeding are normal vital signs, no comorbidities, no melena or syncope, and normal liver and cardiac function.

One clue that bleeding could be serious and require added attention, he said, is when patients present with bloody stools. Although most of the time, patients in the ER with non-variceal upper GI bleeding come in with a combination of hematemesis and melena, the 5% or so of patients who present with bloody stools are sometimes misdiagnosed as having a lower GI bleeding source.

“These are the people that you need to be especially mindful of,” said Dr. Lindenauer, “because hematochezia as a presentation of upper gastrointestinal hemorrhage is usually associated with fairly significant bleeds.”

The issue of endoscopic timing
One study suggests that 20% of patients with non-variceal bleeding have such a low risk of rebleeding that they ought to be treated as outpatients. But Dr. Lindenauer said he believes that the evidence is too slight to tell patients to go home without at least a diagnostic esophagogastroduodenoscopy (EGD).

“The consensus has been that everyone with any kind of significant history of upper GI bleed needs an endoscopy before they can go home,” he said. Endoscopy, Dr Lindenauer said, can help predict what is likely to happen to a patient over the next few days, simply by the endoscopic appearance of the ulcer.

Patients with ulcers characterized as having a “clean base,” which account for 42% of ulcers, “have a rebleeding risk of only 5% and a mortality rate of 2% or less,” regardless of any other patient characteristic, he said.

However, patients whose ulcers are described as a “visible vessel,” where a vessel protrudes above the crater base, have a rebleeding rate of 43%.

“Identifying the ulcer appearance is really important in deciding what needs to be done and how long patients need to stay in the hospital,” he said.

In addition, endoscopy lets you see if there are stigmata of recent hemorrhage. One well-accepted risk-prediction scoring system adds points that “nicely predict mortality in patients with upper GI bleed.” (The scoring system is online.) Using this scoring system, researchers have found that about one-third of patients have “essentially no risk of mortality.”

The timeliness of endoscopy is also critically important, not so much in terms of outcomes and mortality, but of early discharge and reduced lengths of stay. Having a diagnostic endoscopy sooner “on the first day, rather than waiting several days “can cut the length of a patient’s hospitalization by 30% to 50%, Dr. Lindenauer said.

Monotherapy vs. combination
Based on their diagnostic EGD, some patients will end up being good candidates for endoscopic therapy. Although details generally fall in the domain of a gastroenterologist, hospitalists should understand that nearly all hemostatic techniques “thermal coagulation, epinephrine therapy, laser therapy or fibrin sealant “are “generally equivalent” as monotherapies.

Moreover, Dr. Lindenauer said, combination therapy tends to be more effective than monotherapy. The most common combination is thermal coagulation with injection therapy. Keep in mind that the success of any technique depends on the skill of the gastroenterologist doing it, so “you want to go with what your gastroenterologist is more comfortable doing.” A routine “second-look” EGD is not associated with improved outcomes, he pointed out, so is not recommended.

H2 blockers and PPIs
What does fall in the purview of hospitalists’ care is non-endoscopic therapies. These start with antisecretory therapies.

The bottom line on H2 receptor antagonists, said Dr. Lindenauer, is “they should not be used for these patients.” Both a meta-analysis of more than 25 trials and a subsequent large randomized controlled trial failed to show that H2 blockers reduced the risk of rebleeding or death.

Proton pump inhibitors (PPIs), on the other hand, should be in a hospitalist’s arsenal for these patients. Studies continue to show that PPIs are a “very potent therapy” in reducing the risk of rebleeds following endoscopic therapy, both for patients with variceal and non-variceal hemorrhage. (While studies have shown that PPIs reduce both the risk of rebleeding and the need for surgical intervention, Dr. Lindenauer said, their effect on mortality is less clear-cut.)

What about ordering PPIs empirically, even before diagnostic endoscopy? Until recently, Dr. Lindenauer noted, there was insufficient evidence to support this common practice.

However, a study published in the April 19, 2007, New England Journal of Medicine found that patients randomized to receive intravenous infusions of omeprazole prior to endoscopy had a reduced need for endoscopic treatment and a shorter length of stay than those on placebo. Based on these new data, Dr. Lindenauer said he feels that empiric PPI use is warranted in patients awaiting endoscopy.

He noted that while PPIs are generally considered very safe, the drugs do interact with some other commonly used medications in the hospital, including diazepam, warfarin, phenytoin, ketoconazole, ampicillin, iron salts, atazanavir and tacrolimus. PPIs are also implicated as a risk factor for Clostridium difficile diarrhea among hospital patients, he said.

Frozen plasma and platelets
In addition to antisecretory agents, Dr. Lindenauer said that hospitalists need to include hematologic products as part of treatment.

He recommended using packed red blood cells to maintain a hematocrit of between 25 and 30, and he urged hospitalists to anticipate the nadir some hours after the initial crit in the emergency room. Fresh frozen plasma can keep patients’ INR below 1.5 and prevent dilutional coagulopathy.

After consulting with the blood bank, prescribe platelets for patients with active bleeding and a platelet count of less than 50, including patients on antiplatelet therapy. DDAVP and estrogens can help patients with uremic platelet dysfunction.

And because of the high risk of sepsis in these patients, Dr. Lindenauer said that data now show a place for prophylactic antibiotics. About 45% of patients with variceal hemorrhage, for instance, will develop an infectious complication during their hospitalization, such as spontaneous bacterial peritonitis, pneumonia, urinary tract infection, aspiration or sepsis.

Although the optimal prophylactic antibiotic regimen is unknown, Dr. Lindenauer said the usual recommendation is for a five-day dose of an oral quinolone.

Rebleeds and follow up
According to Dr. Lindenauer, patients at low risk of rebleeding can resume a normal diet immediately. Patients at higher risk should be kept NPO for between 24 and 48 hours and be monitored closely.

“Once you have rebled,” he pointed out, “your risk of mortality multiplies about 10-fold.”

Knowing the specific cause of the bleeding, meanwhile, will determine follow-up and long-term care. If the peptic ulcer was caused by H. pylori, for instance, therapy will likely include two weeks of PPIs plus two antibiotics, such as clarithromycin and amoxicillin or metronidazole. If the ulcer was caused by use of nonsteroidal anti-inflammatory drugs (NSAIDs), treatment will include four weeks of PPIs plus discontinuing the NSAID (or adding misoprostol), which can cure 75% of cases after eight weeks.

For patients whose ulcers are caused by neither of these, hospitalists should look for malignancies or for Z-E syndrome. Once these are ruled out, Dr. Lindenauer said, these patients will need life-long acid suppression therapy.

And if rebleeding occurs, repeating endoscopic hemostasis works in about two-thirds of cases. If it doesn’t, patients will need surgery.

Given the higher complication rate with surgery, Dr. Lindenauer said he considers it reasonable to try do a repeat EGD first, recognizing that some patients will ultimately need surgery anyway.

Variceal bleeding
Patients with variceal bleeding, by comparison, are “exceptionally high-risk patients,” Dr. Lindenauer said. “All require intervention.”

With these patients, the risk of rebleeding remains high for six weeks after the initial bleed, and underlying comorbidities are just as likely to drive the outcome as the bleeding. Mortality, he said, is usually the result of aspiration, sepsis, coma and renal failure, not exsanguination or persistent bleeding.

Gastroesophageal varices vary with the extent of liver disease, such that 40% of patients with Childs A cirrhosis and more than 80% of patients with Childs C cirrhosis have varices.

Variceal hemorrhage is responsible for about one-third of all deaths in cirrhosis patients, and the risk of rebleeding in the first two days can top 50%.

Patients with bleeding varices also require different endoscopic interventions. The primary choices, Dr. Lindenauer said, are sclerotherapy, which usually achieves initial hemostasis 75% to 100% of the time, but results in complications in up to 20% of patients.

The other choice “band ligation “has a lower complication rate. That’s why, he said, band ligation has “superseded sclerotherapy when technically possible.”

Effective non-endoscopic therapies are the vasopressin analogues somatostatin and octreotide. And if sclerotherapy or band ligation is not immediately available, octreotide is a reasonable alternative, he added. Studies show that a combination of octreotide and sclerotherapy or band ligation “is superior to either modality alone.”

Moreover, octreotide should be given empirically to patients with upper GI bleeding in the setting of advanced liver disease, whether or not the patient has a history of variceal bleeding.

“Given the terrible outcomes in variceal hemorrhage, this is a very low-cost, low risk therapy,” he said. While it is not recommended for non-variceal hemorrhage, “it is also not going to do any harm.” The drug seems to have its best effect in the short-term, to control a hemorrhage immediately, but it is not a long-term solution.

“Recent meta-analyses,” he said, “temper expectations.”

Deborah Gesensway is a freelance writer reporting on U.S. health care from Toronto, Canada.