Causes of UGIB

Several studies confirm that peptic ulcer is responsible for 28% to 59% of UGIB. In the United States and Greece, the percentage of PUB is high compared with other European populations. Percentages of PUB as high as 59% in the United States and Greece versus 28% to 37% in some European countries (UK, France, Netherlands) have been reported. The reasons for these differences are unclear, but it is suspected that the high proportion of NSAID use and/or the prevalence of H pylori infection may be behind these differences. However, there are no enough data from these countries to confirm that this hypothesis is based on solid grounds. Recent evidence suggests, however, that the incidence of PUB as a cause of acute UGIB either may be decreasing or is underreported. The Analysis of the Clinical Outcomes Research Initiative found that between December 1999 and July 2001 endoscopy performed for acute UGIB found a duodenal or gastric ulcer in only 1610 (20.6%) of the patients included. In this study, the presence of mucosal abnormality was the most common cause of UGIB (40%). Probably, widespread proton pump inhibitors (PPIs) prescribing and H pylori eradication protocols also contribute to this downward incidence trend of PUD causing nonvariceal UGIB. Another important factor might be the use of cyclooxygenase 2 (COX-2) inhibitors, which are associated with both decreased episodes of UGIB and endoscopic lesions when compared with nonselective NSAIDs.

There are few epidemiologic data focusing on bleeding in cirrhotic patients. UGIB is, however, an important cause of morbidity and mortality in liver cirrhosis. Lecleire and colleagues evaluated cirrhotic and noncirrhotic patients using the same data set as in the survey of Czernichow and colleagues. A total of 2133 patients presented with UGIB in a 6-month period in 1996 in 4 French geographic areas, including 468 patients with cirrhosis. Variceal bleeding was the cause of bleeding in 59% of the cirrhotic patients, followed by PUB in 16%. The inhospital mortality rate was significantly higher in cirrhotic patients compared with noncirrhotic patients (34% vs 11%, respectively). In a French study comparing data from 1996 and 2000 concerning endoscopic variceal bleeding management, it was shown that endoscopic variceal ligation therapy is now more often used as the first-choice therapy compared with endoscopic sclerotherapy. Furthermore, there are also increasing trends in the use of terlipressin and somatostatin, 2 vasoactive drugs that have shown to be beneficial in cirrhotic patients with this complication. Epidemiologic studies are still needed to evaluate the most recent outcome trends in cirrhotic patients.

There are other less frequent causes of UGIB. It is estimated that 5% to 15% of all cases of acute UGIB are secondary to Mallory-Weiss tears. Vascular ectasias or angiodysplasias are another source of both acute and chronic nonvariceal UGIB and are estimated to be the cause of UGIB in approximately 5% to 10% of cases. Dieulafoy lesion represents the cause for nonvariceal UGIB in less than 5% of all cases. Malignant and benign neoplasms are another infrequent cause of nonvariceal UGIB and represent less than 5% of all causes. More than one potential cause of UGIB is recorded in 16% to 20% of cases. Eventually, in 7% to 25%, no lesions were found that could have explained the bleeding episodes, despite one or more endoscopic procedures. It is unclear how many of these episodes represent bleeding events from the small bowel, beyond the angle of Treitz.

Risk factors for UGIB

H pylori and NSAID independently increase the risk of gastroduodenal ulcer and ulcer bleeding. The prevalence of gastroduodenal ulcers in patients taking NSAIDs regularly is approximately 15% to 30%, although it has been reported to be up to 45% at 6 months in endoscopy trials in NSAID users in patients with arthritis. Several risk factors have been associated with increased risk of PUB, which are potentiated in the presence of NSAID use. These factors are prior history of complicated ulcers, prior history of uncomplicated ulcer, age of 60 years or more, and concomitant use of NSAIDs with anticoagulants and corticosteroids or multiple NSAID use (including aspirin).

Low-dose aspirin increases the risk of UGIB by approximately 2-fold. When NSAIDs and low-dose aspirin are combined, the risk of bleeding increases approximately 2- to 4-fold as compared with low-dose aspirin or NSAID alone. Other factors have also been reported to be associated with increases in risk of UGIB in some studies but not in all. These factors include severity of rheumatoid arthritis, comorbidities such as cardiovascular disease, and dyspepsia.

The identification of H pylori infection as a factor in the development of peptic ulcer has raised the question of a possible synergistic relation between the presence of H pylori infection and NSAID use. Although several studies have found these 2 factors to be independent, the actual size of the interaction remains undefined. A meta-analysis of epidemiologic studies concludes that there is a clear interaction of both factors for the occurrence of PUB. Randomized trials comparing H pylori eradication with noneradication in patients receiving NSAID showed that H pylori eradication reduces the incidence of peptic ulcer in the overall population, but it is especially effective in drug-naive patients. Nonetheless, H pylori eradication seems less effective than treatment with a PPI for the prevention of both uncomplicated and complicated peptic ulcers. A recent study by Venerito and colleagues concluded that for primary ulcer prevention, H pylori eradication before starting an NSAID therapy reduces the risk of NSAID-induced gastric ulcer and virtually abolishes the risk of duodenal ulcer. H pylori eradication alone is not sufficient for secondary prevention of NSAID-induced gastric and duodenal ulcers. H pylori infection seems to further increase the protective effects of PPI to reduce the risk of ulcer relapse. H pylori eradication does not influence the healing of both gastric and duodenal ulcers if NSAID intake is discontinued.

COX-2 inhibitors were introduced in clinical practice because in 2 long-term outcome studies they were associated with reduced risks of upper gastrointestinal (GI) complication and symptomatic gastroduodenal ulcers compared with nonselective NSAIDS. However, a large Spanish case-control study including more than 1500 cases showed that there was still a modest increased risk with the use of COX-2 selective inhibitors (relative risk [RR], 2.6; 95% confidence interval [CI], 1.9–3.6) compared with nonuse, although less than that seen with nonselective NSAIDS. In the absence of aspirin use, the estimated overall RR associated with current use of COX-2 selective inhibitors was 0.6 (95% CI, 0.4–0.9) when compared with nonselective NSAIDs. This advantage over NSAIDs was abolished when using additional aspirin. In a randomized controlled study among 2587 patients, PUB occurred significantly more often in the rofecoxib group than in the placebo group (RR, 4.9%; 95% CI, 1.2–14.5).

Risk factors for UGIB

H pylori and NSAID independently increase the risk of gastroduodenal ulcer and ulcer bleeding. The prevalence of gastroduodenal ulcers in patients taking NSAIDs regularly is approximately 15% to 30%, although it has been reported to be up to 45% at 6 months in endoscopy trials in NSAID users in patients with arthritis. Several risk factors have been associated with increased risk of PUB, which are potentiated in the presence of NSAID use. These factors are prior history of complicated ulcers, prior history of uncomplicated ulcer, age of 60 years or more, and concomitant use of NSAIDs with anticoagulants and corticosteroids or multiple NSAID use (including aspirin).

Low-dose aspirin increases the risk of UGIB by approximately 2-fold. When NSAIDs and low-dose aspirin are combined, the risk of bleeding increases approximately 2- to 4-fold as compared with low-dose aspirin or NSAID alone. Other factors have also been reported to be associated with increases in risk of UGIB in some studies but not in all. These factors include severity of rheumatoid arthritis, comorbidities such as cardiovascular disease, and dyspepsia.

The identification of H pylori infection as a factor in the development of peptic ulcer has raised the question of a possible synergistic relation between the presence of H pylori infection and NSAID use. Although several studies have found these 2 factors to be independent, the actual size of the interaction remains undefined. A meta-analysis of epidemiologic studies concludes that there is a clear interaction of both factors for the occurrence of PUB. Randomized trials comparing H pylori eradication with noneradication in patients receiving NSAID showed that H pylori eradication reduces the incidence of peptic ulcer in the overall population, but it is especially effective in drug-naive patients. Nonetheless, H pylori eradication seems less effective than treatment with a PPI for the prevention of both uncomplicated and complicated peptic ulcers. A recent study by Venerito and colleagues concluded that for primary ulcer prevention, H pylori eradication before starting an NSAID therapy reduces the risk of NSAID-induced gastric ulcer and virtually abolishes the risk of duodenal ulcer. H pylori eradication alone is not sufficient for secondary prevention of NSAID-induced gastric and duodenal ulcers. H pylori infection seems to further increase the protective effects of PPI to reduce the risk of ulcer relapse. H pylori eradication does not influence the healing of both gastric and duodenal ulcers if NSAID intake is discontinued.

COX-2 inhibitors were introduced in clinical practice because in 2 long-term outcome studies they were associated with reduced risks of upper gastrointestinal (GI) complication and symptomatic gastroduodenal ulcers compared with nonselective NSAIDS. However, a large Spanish case-control study including more than 1500 cases showed that there was still a modest increased risk with the use of COX-2 selective inhibitors (relative risk [RR], 2.6; 95% confidence interval [CI], 1.9–3.6) compared with nonuse, although less than that seen with nonselective NSAIDS. In the absence of aspirin use, the estimated overall RR associated with current use of COX-2 selective inhibitors was 0.6 (95% CI, 0.4–0.9) when compared with nonselective NSAIDs. This advantage over NSAIDs was abolished when using additional aspirin. In a randomized controlled study among 2587 patients, PUB occurred significantly more often in the rofecoxib group than in the placebo group (RR, 4.9%; 95% CI, 1.2–14.5).

Epidemiology of PUB

Some epidemiologic reports have shown that there is a strong decrease in the incidence and admission of symptomatic uncomplicated gastric and duodenal ulcer disease in the past decades. Many published studies often rely on retrospective data from case notes or on nonvalidated discharge codes or summaries made without endoscopy in most cases. Relying on such data results in potential weaknesses in the validity and applicability of the epidemiologic data. Routine endoscopy in patients presenting with UGIB has been only performed in the last 2 decades. Complete case ascertainment is important in this setting to avoid missing values and to give the possibility of discussed bias. During the period 1970 to 1979, an American study showed a decline of 26% in hospital admission for peptic ulcers. This decline was mainly because of a decline in duodenal ulcers because admissions for gastric ulcers remained constant. In another study carried out by Brown and colleagues, in a 14-year period from 1958 to 1972, it was noted that admissions related to peptic ulcers in the United Kingdom decreased by 26% (16% for duodenal ulcers and 41% for gastric ulcers). An American, English, and Scottish survey showed a decline in admission rates for ulcer disease between 1970 and 1999, although among the elderly, admission rate for PUD increased. In contrast to these reports, an English study evaluating the period between 1972 and 1984 did not show any reduction in admission rates in patients with uncomplicated duodenal ulcers. Despite the introduction and observed increasing use of H ₂ receptor antagonists, the emergency admission for uncomplicated duodenal ulcer remained stable at about 88 cases per 100,000 people. For the period 1994 to 1998, Lewis and colleagues showed a decline in patients receiving H ₂ receptor antagonists (from 68%–41%) but an increase in patients receiving PPIs (from 46%–66%) with an overall decline in PUD. Besides a significant decrease in peptic ulcer admissions, the total number of operations for ulcer disease decreased by more than 80%. Another epidemiologic data from Asia, Australia, North America, Japan, South America, and Europe revealed an evident decline in mortality from duodenal and gastric ulcer disease since the beginning of the twentieth century to the beginning of the twenty-first century. At present, it seems clear that the characteristics of patients with PUD are changing because those being hospitalized for duodenal or gastric ulcer disease are, in general, older than 65 years.

There are more controversial data concerning rates of UGIB or complicated peptic ulcers, as shown in several recent population-based epidemiologic studies evaluating cause and outcome of UGIB. One of the highest reported rates came form West Scotland, where the annual incidence of UGIB was put at 172 per 100,000 adults. The incidence in Denmark, Sweden, and The Netherlands has been reported to be much lower between 37 and 48 per 100,000 adults. There are several aspects to consider to understand these strong differences between countries. One aspect is the prevalence of H pylori infection, which may differ between countries. Immigration patterns vary between European countries, and immigrants in Western countries show high H pylori infection rates. Socioeconomic differences might also play a role because low socioeconomic class and crowded living conditions in childhood are shown to be related to H pylori infection. Prescription patterns of both ulcer-healing and ulcer-promoting drugs could be another factor, as well as compliance to medication, because of costs for medication for the patients. Different selection criteria, for example, exclusion of inhospital patients, might lead to selection bias in estimating incidence and population-based mortality. One study by Lanas and colleagues published in abstract form reporting outcomes and predictive factors for bleeding continuation/rebleeding and mortality of UGIB in clinical practice in different European countries concluded that the differences in the outcomes of UGIB in clinical practice across some European countries are explained mainly by patient-related factors and not management factors.

Time trends of UGIB

Opposing trends in peptic ulcer complications such as bleeding or perforation have been reported in different countries, and no decrease or increase in hospitalizations due to PUB complications have been observed. Post and colleagues investigated time trends in the incidence of and hospital admission rates for peptic ulcer in the Netherlands. They concluded that the incidence of histopathologically confirmed gastric ulcer halved between 1992 and 2003 in the Netherlands. Because the number of gastric biopsies increased in this period, a true decrease was likely. Another study by Sadic and colleagues evaluated possible changes in the incidence of bleeding peptic ulcer, treatment, and mortality over time. The investigators showed that the incidence rate for bleeding gastric or duodenal ulcers decreased by one-half in men and by one-third in women and emergency operations decreased significantly (9.2%, 7.5%, and 5.7% during the 3 periods between 1987 and 2004 [ P <.05]). The postoperative mortality tended to decrease (9.7%, 2.4%, and 3.7%) and the 30-day mortality rates in the whole material were less than 4% and have not changed over time.

A recently published systematic review by Sung and colleagues evaluating the current global incidence and prevalence of PUD by systematically reviewing the literature published over the last decade concluded that the annual incidence rates of PUD were 0.10% to 0.19% for physician-diagnosed PUDs and 0.03% to 0.17% when based on hospitalization data. The 1-year prevalence based on physician diagnosis was 0.12% to 1.50% and that based on hospitalization data was 0.10% to 0.19%. Most studies reported a decrease in the incidence or prevalence of PUD over time. PUD remains a common condition, although reported incidence and prevalence are decreasing.

More recently, 3 studies suggested that there has been a marked decrease in the incidence of upper GI complications and a slight increase in the incidence of lower GI complications. One of them concluded that the incidence per 100,000 person-years of hospitalizations due to upper GI ulcer bleeding and perforation decreased over time (from 54.6 and 3.9 in1996 [R ² = 0.944] to 25.8 and 2.9 in 2005 [R ² = 0.410]). Based on data extracted from the validation process, NSAID and low-dose aspirin uses were more prevalent in PUB and colonic diverticular bleeding, respectively. Another one of these studies showed that lower GI complications increased from 20 per 100,000 to 33 per 100,000 between 1996 and 2005. Overall, mortality rates decreased, but the case fatality remained constant over time. Lower GI events had a higher mortality rate (8.8% vs 5.5%), a longer hospitalization (11.6 ± 13.9 vs 7.9 ± 8.8 days), and higher resource utilization than did upper GI events. When comparing upper GI events with lower GI events, the investigators of the study found that male gender (adjusted odds ratio [OR], 1.94; 95% CI, 1.70–2.21) and recorded NSAID use (OR, 1.92; 95% CI, 1.60–2.30) were associated to a greater extent with upper GI events, whereas older age (OR, 0.83; 95% CI, 0.77–0.89), number of comorbidities (OR, 0.91; 95% CI, 0.86–0.96), and having a diagnosis in recent years (OR, 0.92; 95% CI, 0.90–0.94) were all associated to a greater extent with lower GI events than with upper GI events after adjusting for age, sex, hospitalization, and discharge year. The exact source of lower GI complications are often more difficult to identify than upper GI complications because of the anatomic complexity of the lower gut and available diagnostic tests. Among the causes of lower GI bleeding, colonic diverticula and angiodysplasia are 2 lesions that could explain, at least in part, the recent trends. Recently, a study by Lanas and colleagues confirmed a dramatic decrease in the number of hospitalizations due to both PUB and peptic ulcer perforation, whereas the number of cases of colonic diverticular and angiodysplasia bleeding have increased ( Fig. 1 ). Pérez-Aisa and colleagues studied the clinical trend of PUD in the high-prevalence H pylori population. The investigators concluded that the incidence of peptic ulcer and associated complications were declining rapidly. This decline was associated with a reduction of the prevalence of H pylori infection in the young and a widespread use of PPIs. The increase in the use of NSAIDs had not changed the tendency.

Estimated number of peptic ulcer, colonic diverticular, and angiodysplasia bleeding events per 100,000 person-years in Spain.

( Data from Lanas A, García-Rodríguez LA, Polo-Tomás M, et al. The changing face of hospitalisation due to gastrointestinal bleeding and perforation. Aliment Pharmacol Ther 2011;33(5):585–91.)

Rebleeding

Rebleeding is the most important predictive factor of mortality, which occurs in 6% to 10% of patients. In 80% of patients, the bleeding stops spontaneously, but 20% require endoscopic therapy, which is initially effective in more than 90% of patients. Clinical factors found to be predictors of rebleeding include age, history of PUB, shock, comorbid illnesses, low initial hemoglobin level, transfusion requirement, fresh blood (in the emesis, in the nasogastric aspirate, or on rectal examination), source of the bleed, ulcer size, endoscopic stigmata of recent or active hemorrhage (spurting bleeding, oozing bleeding, visible vessel, adherent clot), and lack of PPI use following endoscopy and duodenal ulcer.

There is a growing rate of patients developing an UGIB event when taking anticoagulants. A study by Wolf and colleagues in 233 patients showed that the rebleeding rate was 23% in the patients who underwent anticoagulation and 21% in those with international normalized ratios (INRs) less than 1.3. On multivariable analyses, INR was not a predictor of rebleeding, transfusion requirement, surgery, length of stay, or mortality. The investigators concluded that mild to moderate anticoagulation does not increase the risk of rebleeding after endoscopic therapy for nonvariceal upper GI hemorrhage. However, a more recent published study had demonstrated that the postprocedure use of intravenous or low–molecular-weight heparin increases significantly the risk of rebleeding ( P = .0014). Another 5 independent risk factors were identified, including failure to use a PPI postprocedure, endoscopically demonstrated bleeding, peptic ulcer as the bleeding source, treatment with epinephrine monotherapy, and moderate or severe liver cirrhosis ( P = .032). The risk of rebleeding increased as the number of risk factors present increased reaching the 100% of risk when 4 risk factors were present.