Acute upper gastrointestinal bleeding (UGIB) is a common gastroenterological emergency. A vast majority of these bleeds have nonvariceal causes, in particular gastroduodenal peptic ulcers. Nonsteroidal antiinflammatory drugs, low-dose aspirin use, and Helicobacter pylori infection are the main risk factors for UGIB. Current epidemiologic data suggest that patients most affected are older with medical comorbidit. Widespread use of potentially gastroerosive medications underscores the importance of adopting gastroprotective pharamacologic strategies. Endoscopy is the mainstay for diagnosis and treatment of acute UGIB. It should be performed within 24 hours of presentation by skilled operators in adequately equipped settings, using a multidisciplinary team approach.
Key points
- •
There is a trend toward a decrease in the overall incidence and hospitalization for nonvariceal upper gastrointestinal bleeding (UGIB) worldwide. Peptic ulcer is still the most common cause of hemorrhage.
- •
The changing epidemiology is characterized by an aging population, with multiple comorbidities and increased use of aspirin, nonsteroidal antiinflammatory drugs (NSAIDs), or other antiplatelets/anticoagulants.
- •
Mortality for UGIB is still approximately 5% and is usually related to multiorgan failure, cardiopulmonary conditions, and end-stage malignancy.
- •
Endoscopy is the mainstay in the management of UGIB, allowing for proper diagnosis, risk stratification, and treatment of the bleeding lesion.
- •
Unless contraindicated, endoscopy should be performed within 24 hours of patient presentation to maximize benefits and improve economic outcomes.
Epidemiology of acute nonvariceal upper gastrointestinal bleeding
UGIB is predominantly nonvariceal in origin and remains one of the most common challenges faced by gastroenterologists and endoscopists in daily clinical practice. Despite major advances in the approach to the management of nonvariceal UGIB over the past 2 decades, including prevention of peptic ulcer bleeding, optimal use of endoscopic therapy, and adjuvant high-dose proton pump inhibitors (PPIs), it still carries considerable morbidity, mortality, and health economic burden.
Incidence of Acute Upper Gastrointestinal Bleeding
With more than 300,000 hospital admissions annually in the United States, UGIB is one of the most common gastrointestinal (GI) emergencies. A 2012 update on the burden of GI disease in the United States reports that GI hemorrhage still ranked 7th among the principal GI discharge diagnoses from hospital admissions in 2009, with a 22% increase compared with year 2000, and 10th among causes of death from GI and liver diseases.
The incidence rates of UGIB demonstrate a large geographic variation, ranging from 48 to 160 cases per 100,000 population per year, with consistent reports of higher incidences among men and the elderly. Possible explanations for the reported geographic variations in incidence are differences in definition of UGIB in various studies, population characteristics, prevalence of gastroerosive medications, in particular aspirin and NSAIDs, and Helicobacter pylori prevalence.
Some but not all time-trend studies have reported a significant decline in incidence of all-cause acute UGIB, especially peptic ulcer bleeding, in recent years. In the Netherlands, the incidence of UGIB decreased from 61.7/100,000 in 1993/1994 to 47.7/100,000 persons annually in 2000, corresponding to a 23% decrease in incidence after age adjustment. This was confirmed in a population-based study carried out in Northern Italy in which the overall incidence of UGIB decreased from 112.5 to 89.8/100,000 per year, which corresponds to a 35.5% decrease after adjustment for age. Trends for incidence of hospitalization due to GI complications in the United States from 2001 to 2009 confirm decreases in UGIB (78.4–60.6/100,000) and peptic ulcer bleeding (48.7–32.1/100,000). The reasons for the observed decrease in hospitalizations due to nonvariceal UGIB are not well defined, but it is reasonable to assume that the use of eradication therapy in patients with ulcer disease and the progressive increase in the implementation of preventive strategies in patients taking aspirin and NSAIDs may have played a role.
Outcome data from multicenter observational registries of UGIB, originating from Italy, Canada, and the United Kingdom, reported a mean age of bleeders over 60 years and a prevalence of UGIB in men. In-hospital bleeding (ie, GI hemorrhage that occurs in patients already hospitalized for another medical-surgical condition) occurs in 10% to 25%.
Causes of Acute Upper Gastrointestinal Bleeding
Peptic ulcer bleeding is still the most common cause of nonvariceal UGIB, responsible for approximately 31% to 67% of all cases, followed by erosive disease, esophagitis, malignancy, and Mallory-Weiss tears. In 2% to 8% of cases, uncommon causes, such as Dieulafoy lesion, hemobilia, angiodysplasia, vascular-enteric fistula, and gastric antral vascular ectasia are found ( Table 1 ).
| % | |
|---|---|
| Peptic ulcer | 31–67 |
| Erosive disease | 7–31 |
| Variceal bleeding | 4–20 |
| Esophagitis | 3–12 |
| Mallory-Weiss tears | 4–8 |
| Malignancy | 2–8 |
| Vascular lesions | 2–8 |
| None (no lesion identified) | 3–19 |
In recent years, there has been an overall decrease in the incidence of UGIB related to bleeding peptic ulcers, at least in subjects under 70 years of age, whereas its incidence is stable or even higher among patients of more advanced age. A study from Australia on bleeding ulcers over a 10-year period (1997–2007) confirmed that the number of bleeding ulcers remained unchanged despite a decreased incidence of uncomplicated peptic ulcer. Gastric ulcers increased significantly in both bleeding and nonbleeding patients whereas the proportion of duodenal ulcers fell significantly. The proportion of bleeding ulcers related to NSAIDs or aspirin increased significantly over 10 years, from 51% to 71%. Gastroduodenal ulcers are also the most frequent causes of nonvariceal bleeding in cirrhotic patients (48%–51%).
Nonvariceal UGIB is not just about peptic ulcers. According to different registries, nonvariceal nonulcer bleeding accounts for 34% to 64% of all presenting cases of nonvariceal UGIB. Recent data from Italy show that patients with Dieulafoy lesions have high rebleeding rates (19.1%); although rare causes of nonvariceal UGIB, they can cause torrential bleeding and can be difficult to locate. Patients with Dieulafoy lesions were more likely to present with hematemesis, shock, syncope, and a lower hemoglobin concentration and require blood transfusion compared with patients presenting with other endoscopic diagnoses. Of greater concern was the reported rebleeding rate for Mallory-Weiss tears (6.3%), traditionally considered benign, low-risk, and self-limiting lesions. It is possible that this may represent endoscopic undertreatment because of the perceived low-risk nature of Mallory-Weiss tears but also raises questions regarding uniform diagnostic criteria.
The source and outcomes of UGIB in oncologic patients are poorly investigated. The causes of UGIB in oncologic patients seem to be different from those in the general population. Retrospective data on 324 patients with cancer referred for endoscopy due to UGIB showed that tumor was the most common cause of bleeding (23.8%), followed by varices (19.7%), peptic ulcer (16.3%), and gastroduodenal erosions (10.9%). If considering only patients with tumors outside the GI tract, however, the most common causes of UGIB are similar to those in the general population, that is, peptic ulcer, gastroduodenal erosions, and varices. On the other hand, even in patients with tumors outside the GI tract, metastases were the source of bleeding in a significant number of patients (11%).
Risk Factors for Acute Nonvariceal Upper Gastrointestinal Bleeding
Risk factors for peptic ulcer bleeding are H pylori infection, use of NSAIDs, use of low-dose aspirin, and other antiplatelet medications or oral anticoagulants.
Helicobacter pylori infection
H pylori infection is found in 43% to 56% of peptic ulcer bleeding patients. The true H pylori prevalence in bleeding peptic ulcer is probably underestimated: in a recent meta-regression on 71 studies, including 8496 patients, the mean prevalence of H pylori infection in peptic ulcer bleeding was 72%. The most significant variables associated with a high prevalence of H pylori infection were the use of a diagnostic test delayed until at least 4 weeks after the bleeding episode (odds ratio [OR] 2.08; 95% CI, 1.10–3.93) and a lower mean age of patients (OR 0.95 per additional year; 95% CI, 0.92–0.99). The low prevalence of H pylori infection reported in peptic ulcer bleeding may be due to the methodology of the studies and to patient characteristics. These data also support the recent recommendations of an international consensus on nonvariceal UGIB regarding the performance of a delayed diagnostic test when H pylori tests carried out during the acute bleeding episode are negative.
The incidence of H pylori –negative idiopathic bleeding ulcers (ie, those not related to NSAIDs or other gastroerosive medications) is rising. These bleeding ulcers account for 16.1% of patients admitted for UGIB and 42.4% of patients who bled while in the hospital. These ulcers are also prone to recurrent complications at 12 months: 13.4% (95% CI, 7.3%–19.5%) versus 2.5% (95% CI, 0.4%–4.6%) in patients with H pylori –positive ulcers who received eradication therapy.
H pylori –negative bleeding ulcers are also associated with poorer outcomes regardless of use of NSAIDs. In a study from the University of Texas, patients without H pylori infection had significantly more comorbid medical conditions and higher Charlson index comorbidity scores than those with H pylori . Recurrent bleeding within 30 days was more frequent (11% vs 5%, P = .009) and hospital length of stay was significantly longer compared with H pylori –positive patients. Such outcome data confirm previous findings of significantly higher incident rates of rebleeding and death in patients with H pylori -negative idiopathic ulcers than in controls with H pylori –positive bleeding ulcers. Moreover, gastroprotective agents, such as PPIs or H 2 -receptor antagonists, did not reduce the risk of recurrent bleeding or mortality for patients with H pylori –negative idiopathic bleeding ulcers.
Current guidelines recommend testing for H pylori infection among users of low-dose aspirin who are at high risk for developing ulcers, because the long-term incidence of rebleeding with aspirin use is reduced after H pylori is eradicated. This was confirmed by a recent prospective study from Hong Kong, in which the incidence of ulcer bleeding (per 100 patient-years) in the H pylori –eradicated cohort did not differ significantly from that of the average-risk cohort (no history of ulcers). Aspirin users without current or past H pylori infections who develop ulcer bleeding have a 5-fold incidence of recurrent bleeding.
Low-dose aspirin, antiplatelets, and nonsteroidal antiinflammatory drugs
Long-term use of aspirin is recommended for prevention of cardiovascular events among patients with prior cardiovascular disease or multiple risk factors. Regular aspirin use is associated with an increased risk of major GI bleeding. A recent meta-analysis found an approximately 2-fold higher risk of GI bleeding among individuals regularly using aspirin compared with placebo, with no difference between 75 to 162.5 mg/d and greater than 162.5 to 325 mg/d. Such a magnitude of risk is confirmed by a large prospective study of 87,680 women in which the relative risk (RR) of major UGIB requiring hospitalization or blood transfusion was 1.43 (95% CI, 1.29–1.59) over a 24-year follow-up period. Furthermore, when used for primary prevention of cardiovascular disease, the absolute harms of aspirin seem to exceed its benefits: a meta-analysis of 27 studies showed that 60 to 84 major cardiovascular events per 100,000 person-years were averted, whereas 68 to 117 GI bleeds were incurred. There was a nonsignificant change in total cardiovascular disease, whereas risks were increased by 37% for GI bleeds. Risk factors for UGIB among low-dose aspirin users include (1) a history of peptic ulcer disease or GI bleeding; (2) older age; (3) concomitant use of NSAIDs, including cyclooxygenase (COX)-2 inhibitors; (4) concomitant use of anticoagulants or other platelet aggregation inhibitors; (5) the presence of severe medical comorbidities; and (6) high aspirin dose. H pylori and aspirin seem to be independent risk factors for peptic ulcer bleeding, so that in patients with a history of peptic ulcer disease, H pylori infection should be assessed and treated if present.
At-risk low-dose aspirin users are recommended to take gastroprotective agents. PPIs seem superior to eradication only to prevent recurrent ulcer bleeding in patients using low-dose aspirin. Moreover, in patients with acute coronary syndrome or myocardial infarction receiving aspirin, clopidogrel, enoxaparin, or thrombolytics, PPIs are superior to H 2 -receptor antagonists.
Population-based epidemiologic data confirm that the use of low-dose aspirin, NSAIDs, warfarin, and the combination thereof are significantly more common among UGI bleeders than nonbleeders. An increasing proportion of patients are nowadays hospitalized due to medication-related UGIB. Comparing data from Italy and the United Kingdom, the proportion of UGIB patients taking NSAIDs or antiplatelet agents and that of patients on anticoagulants was almost identical (46.4% vs 44.2% and 11.7% vs 13.2%, respectively). The increased risk of UGIB associated with NSAIDs seems lower than previously reported: in a Spanish primary care research database study of more than 660,000 subjects, increased risks were found with current use of NSAIDs (RR 1.72; 95% CI, 1.41–2.09), low-dose aspirin (RR 1.74; 95% CI, 1.37–2.21), other antiplatelet drugs (RR 1.73; 95% CI, 1.27–2.36), and oral anticoagulants (RR 2.00; 95% CI, 1.44–2.77). The use of oral corticosteroids, selective serotonin reuptake inhibitors, or acetaminophen was not associated with an increased risk.
In patients at risk for GI bleeding and using NSAIDs, a protective drug is recommended. Acid-suppressing drugs reduce the risk among users of NSAIDs (OR 0.58; 0.39–0.85), particularly in users with antecedent peptic ulcer (OR 0.16; 0.05–0.58). Unfortunately, despite several society and national guidelines that have been formulated, these are poorly followed in clinical practice and gastroprotection is largely underused. Data from Europe show that although approximately one-half of all patients with peptic ulcer bleeding were using NSAIDs or aspirin, an effective gastroprotective agent was only used in 10% to 15% of at-risk patients. A multinational study from the United Kingdom, Italy, and the Netherlands confirmed that the risk of UGIB is significantly higher in nonselective NSAID users with gastroprotective nonadherence. Among 618,684 nonselective NSAID users, nonadherers to gastroprotection had a 2-fold risk of UGIB (OR 1.89; 95% CI, 1.09–3.28). Selective cyclooxygenase (COX)-2 inhibitors are associated with a lower risk of clinically significant UGIB than nonselective NSAIDs ; however, concerns over the possible cardiovascular adverse effects of some of these agents should be taken into account. Moreover, switching to selective COX-2 inhibitors in patients with previous bleeding is not completely risk-free, and concomitant PPI therapy may also be needed. The combination of COX-2 inhibitors with PPIs promotes the greatest risk reduction for NSAID-related UGIB.
A retrospective study among general practitioners in France documented that within 2 years after prescribing a PPI, physicians did not renew this prescription for approximately 33% of those patients at risk for GI events (>65 years, past history of GI ulcer, or receiving antiplatelet agents) receiving continuous NSAIDs. Predictors for no longer receiving a prescription for a PPI included switching to a COX-2 selective inhibitor or to a nonselective NSAID and female gender. The risk for upper GI injury was higher among patients with discontinued PPI prescriptions (OR 1.45; 95% CI, 1.06–2.09).
UGIB is a rare but serious potential side effect of bisphosphonate therapy. In a population-based nested cohort study in Canada within an exposure cohort of 26,223 subjects, 117 individuals suffered a serious UGIB within 120 days of starting a bisphosphonate (0.4%). Age greater than 80 (adjusted OR 2.03; 95% CI, 1.40–2.94) and a past history of serious UGIB (adjusted OR 2.28; 95% CI, 1.29–4.03) were the strongest predictors. Men were 70% more likely to suffer an UGIB compared with women (adjusted OR 1.69; 95% CI, 1.05–2.72).
Patients with UGIB have increasing non-GI comorbidities. Results of a matched case-control study on 16,355 patients with nonvariceal UGIB and 81,636 controls showed that non-GI comorbidity had a strong association with UGIB; the adjusted OR for a single comorbidity was 1.43 (95% CI, 1.35–1.52) and for multiple or severe comorbidity was 2.26 (95% CI, 2.14–2.38). The additional population attributable fraction for comorbidity (19.8%; 95% CI, 18.4–21.2) was considerably larger than that for any other measured risk factor, including aspirin or NSAIDs use (3.0% and 3.1%, respectively). Non-GI comorbidity is an independent risk factor for UGIB and contributes to a greater proportion of patients with bleeding in the population than other recognized risk factors. These findings could explain why the incidence of nonvariceal bleeding remains high in older populations.
Mortality from Acute Upper Gastrointestinal Bleeding
Despite advances in endoscopic hemostasis and adjuvant pharmacologic treatment, the overall mortality from UGIB remains 5% to 14%, although most studies from the United States, Europe, and Asia place that figure closer to 5%. A systematic review of 18 studies (10 using administrative databases and 8 using bleeding registries) showed mortality rates from acute UGIB ranging from 1.1% in Japan to 11% in Denmark. The 28-day mortality after nonvariceal UGIB in England decreased from 14.7% in 1999 to 13.1% in 2007. A recent analysis of the trends for incidence of hospitalization and death due to UGIB in the United States from 2001 to 2009 confirmed that age/gender-adjusted case fatality owing to bleeding is low (2.5%) and increases with age but remains less than 5% even in elderly patients. These discrepancies in reported mortality rates of nonvariceal UGIB are attributable to differences in study methodologies and populations studied (heterogeneous definitions of case ascertainment, differing patient populations with regard to severity of presentation and associated comorbidities, varying durations of follow-up, and different health care system-related practices). More uniform standards in reporting would enable a better understanding of causes of death and the apparent discrepant outcome in endoscopic end points versus clinical end points. Nonvariceal UGIB is now predominantly a disease of the elderly, with more than 60% of patients over the age of 60 years and approximately 20% over the age of 80 years. Because elderly patients have more comorbid illness, are more likely users of aspirin and NSAIDs, and are less tolerant of hemodynamic insult, the management of this high-risk population is a major challenge. Current evidence indicates that most peptic ulcer bleeding–linked deaths are not a direct sequela of the bleeding ulcer itself. Instead, mortality derives from multiorgan failure, cardiopulmonary conditions, or end-stage malignancy, suggesting that improving further current treatments for the bleeding ulcer may have a limited impact on mortality unless supportive therapies are developed for the global management of these patients. In a prospective cohort study enrolling more than 10,000 cases of peptic ulcer bleeding at the Prince of Wales Hospital in Hong Kong, one of the most reputed bleeding centers worldwide, overall mortality was 6.2%. The study reported that approximately 80% of deceased patients died of non–bleeding-related causes.
Risk factors for mortality after nonvariceal UGIB are
- •
Increasing age
- •
Hemodynamic instability on admission
- •
Presence of severe and life-threatening comorbid medical conditions
Recent data from Italy also identified an American Society of Anesthesiologists (ASA) score of 3 or 4 versus 1 or 2 as the variable with the greatest OR for predicting mortality (OR 3.92; 95% CI, 2.37–6.50). One or more comorbidities are present in almost two-thirds of UGIB patients. Underlying comorbidity is consistently associated with an increased short-term mortality in patients with peptic ulcer bleeding. A systematic review and meta-analysis of 16 studies showed that the risk of 30-day or in-hospital mortality was significantly greater in patients with comorbidity than in those without (RR 4.44; 95% CI, 2.45–8.04). Patients with 3 or more comorbidities had a greater risk of dying than those with 1 or 2 (RR 3.46; 95% CI, 1.34–8.89). Among individual comorbidities that significantly increased the risk of death, RRs were higher for hepatic, renal, and malignant disease (RR range, 4.04–6.33) than for cardiovascular and respiratory disease and diabetes (2.39, 2.45, and 1.63, respectively). Coagulation disorders are also independently associated with more than a 5-fold increase in the odds of in-hospital mortality.
Cirrhotic patients suffering a nonvariceal bleed have significantly greater in-hospital mortality than noncirrhotic patients. The presence of cirrhosis independently increased mortality (adjusted OR 3.3; 95% CI, 2.2–4.9). Decompensated cirrhosis had higher mortality than compensated cirrhosis. Cryptogenic etiology of cirrhosis, renal dysfunction, actively bleeding ulcers on hospital admission, concurrent presence of duodenal ulcer and erosive disease, and bleeding from vascular lesions are all independent predictors of mortality.
Impaired renal function is an independent risk factor for UGIB. Crude rates of acute nonvariceal UGIB among patients undergoing dialysis have not decreased in the past 10 years. Although 30-day mortality related to UGIB declined, the burden on the end-stage renal disease population remains substantial. Overall 30-day mortality is 4.8% to 13.7%, with a 2-fold risk of death in peptic ulcer bleeding patients with end-stage renal disease compared with those without. Such a high mortality in patients undergoing dialysis was significantly correlated with older age, female gender, infection during hospitalization, single episodic UGIB, abnormal white blood cell count, and albumin level less than or equal to 3 g/dL.
Mortality is also increased in patients who are already admitted in hospital for another medical problem. In-hospital bleeding accounts for 10% to 25% of the overall nonvariceal bleeding population and is associated with poorer outcome, with mortality rates as high as 26%. Nonetheless, the reasons for increased mortality in this subgroup of patients have not been consistently identified. Guidelines on optimal management of inpatients who develop nonvariceal UGIB have been derived essentially from studies on outpatient bleeding, whereas few data are available that focus on in-hospital bleeding and its management. Recent data from Italy shed some light on the issue, showing that the mortality rate for in-hospital bleeding was significantly higher than that of outpatient bleeding (8.9% vs 3.8%; OR 2.44; 95% CI, 1.57–3.79). Hemodynamic instability on presentation and the presence of severe comorbidity were the strongest predictors of mortality for in-hospital bleeders.
The risk of mortality increases with rebleeding, which is thus another major outcome parameter. Rebleeding rates, using a combination of endoscopic hemostasis and adjuvant acid-suppressing therapy, have been shown to be reduced to less than 10% and it would be difficult, if not impossible, to reduce this further. Because a majority of nonvariceal UGIB patients today are of advanced age, users of NSAIDs and aspirin, and with multiple comorbid illnesses, the “endeavor to further reduce this rebleeding rate is an uphill battle.”
Little is known about mortality associated with nonulcer, nonvariceal UGIB. This may stem from a perception among clinicians that, other than those patients with a GI malignancy as the causative bleeding lesion, outcomes of patients with nonulcer bleeding are favorable. Therefore, it is possible that the usual algorithms of postendoscopy care, pharmacotherapy, and monitoring that are applied to patients in whom endoscopic hemostasis has been achieved after peptic ulcer bleeding may not be applied with the same rigor to those with nonulcer bleeding. A large prospective, multicenter study of 3207 patients with documented nonulcer UGIB have a risk of death similar to bleeding peptic ulcers. Mortality was 9.8% for neoplasia, 4.8% for Mallory-Weiss tears, 4.8% for vascular lesions, 4.4% for gastroduodenal erosions, 4.4% for duodenal ulcer, and 3.1% for gastric ulcer. Frequency of death was not different among benign endoscopic diagnoses. The strongest predictor of mortality was the overall ASA score. After adjusting for ASA score, the endoscopic diagnosis had no impact on mortality, suggesting that nonulcer causes of bleeding carry a risk of mortality similar to that of peptic ulcers. Therefore, even a bleed from what is perceived to be a minor lesion in a high-risk patient can be a sufficient precipitant to initiate a downward cascade of clinical events, resulting in multiorgan failure and death. The message could not be clearer: treat the patient and not just the source of GI bleeding.
Epidemiology of acute nonvariceal upper gastrointestinal bleeding
UGIB is predominantly nonvariceal in origin and remains one of the most common challenges faced by gastroenterologists and endoscopists in daily clinical practice. Despite major advances in the approach to the management of nonvariceal UGIB over the past 2 decades, including prevention of peptic ulcer bleeding, optimal use of endoscopic therapy, and adjuvant high-dose proton pump inhibitors (PPIs), it still carries considerable morbidity, mortality, and health economic burden.
Incidence of Acute Upper Gastrointestinal Bleeding
With more than 300,000 hospital admissions annually in the United States, UGIB is one of the most common gastrointestinal (GI) emergencies. A 2012 update on the burden of GI disease in the United States reports that GI hemorrhage still ranked 7th among the principal GI discharge diagnoses from hospital admissions in 2009, with a 22% increase compared with year 2000, and 10th among causes of death from GI and liver diseases.
The incidence rates of UGIB demonstrate a large geographic variation, ranging from 48 to 160 cases per 100,000 population per year, with consistent reports of higher incidences among men and the elderly. Possible explanations for the reported geographic variations in incidence are differences in definition of UGIB in various studies, population characteristics, prevalence of gastroerosive medications, in particular aspirin and NSAIDs, and Helicobacter pylori prevalence.
Some but not all time-trend studies have reported a significant decline in incidence of all-cause acute UGIB, especially peptic ulcer bleeding, in recent years. In the Netherlands, the incidence of UGIB decreased from 61.7/100,000 in 1993/1994 to 47.7/100,000 persons annually in 2000, corresponding to a 23% decrease in incidence after age adjustment. This was confirmed in a population-based study carried out in Northern Italy in which the overall incidence of UGIB decreased from 112.5 to 89.8/100,000 per year, which corresponds to a 35.5% decrease after adjustment for age. Trends for incidence of hospitalization due to GI complications in the United States from 2001 to 2009 confirm decreases in UGIB (78.4–60.6/100,000) and peptic ulcer bleeding (48.7–32.1/100,000). The reasons for the observed decrease in hospitalizations due to nonvariceal UGIB are not well defined, but it is reasonable to assume that the use of eradication therapy in patients with ulcer disease and the progressive increase in the implementation of preventive strategies in patients taking aspirin and NSAIDs may have played a role.
Outcome data from multicenter observational registries of UGIB, originating from Italy, Canada, and the United Kingdom, reported a mean age of bleeders over 60 years and a prevalence of UGIB in men. In-hospital bleeding (ie, GI hemorrhage that occurs in patients already hospitalized for another medical-surgical condition) occurs in 10% to 25%.
Causes of Acute Upper Gastrointestinal Bleeding
Peptic ulcer bleeding is still the most common cause of nonvariceal UGIB, responsible for approximately 31% to 67% of all cases, followed by erosive disease, esophagitis, malignancy, and Mallory-Weiss tears. In 2% to 8% of cases, uncommon causes, such as Dieulafoy lesion, hemobilia, angiodysplasia, vascular-enteric fistula, and gastric antral vascular ectasia are found ( Table 1 ).
| % | |
|---|---|
| Peptic ulcer | 31–67 |
| Erosive disease | 7–31 |
| Variceal bleeding | 4–20 |
| Esophagitis | 3–12 |
| Mallory-Weiss tears | 4–8 |
| Malignancy | 2–8 |
| Vascular lesions | 2–8 |
| None (no lesion identified) | 3–19 |
In recent years, there has been an overall decrease in the incidence of UGIB related to bleeding peptic ulcers, at least in subjects under 70 years of age, whereas its incidence is stable or even higher among patients of more advanced age. A study from Australia on bleeding ulcers over a 10-year period (1997–2007) confirmed that the number of bleeding ulcers remained unchanged despite a decreased incidence of uncomplicated peptic ulcer. Gastric ulcers increased significantly in both bleeding and nonbleeding patients whereas the proportion of duodenal ulcers fell significantly. The proportion of bleeding ulcers related to NSAIDs or aspirin increased significantly over 10 years, from 51% to 71%. Gastroduodenal ulcers are also the most frequent causes of nonvariceal bleeding in cirrhotic patients (48%–51%).
Nonvariceal UGIB is not just about peptic ulcers. According to different registries, nonvariceal nonulcer bleeding accounts for 34% to 64% of all presenting cases of nonvariceal UGIB. Recent data from Italy show that patients with Dieulafoy lesions have high rebleeding rates (19.1%); although rare causes of nonvariceal UGIB, they can cause torrential bleeding and can be difficult to locate. Patients with Dieulafoy lesions were more likely to present with hematemesis, shock, syncope, and a lower hemoglobin concentration and require blood transfusion compared with patients presenting with other endoscopic diagnoses. Of greater concern was the reported rebleeding rate for Mallory-Weiss tears (6.3%), traditionally considered benign, low-risk, and self-limiting lesions. It is possible that this may represent endoscopic undertreatment because of the perceived low-risk nature of Mallory-Weiss tears but also raises questions regarding uniform diagnostic criteria.
The source and outcomes of UGIB in oncologic patients are poorly investigated. The causes of UGIB in oncologic patients seem to be different from those in the general population. Retrospective data on 324 patients with cancer referred for endoscopy due to UGIB showed that tumor was the most common cause of bleeding (23.8%), followed by varices (19.7%), peptic ulcer (16.3%), and gastroduodenal erosions (10.9%). If considering only patients with tumors outside the GI tract, however, the most common causes of UGIB are similar to those in the general population, that is, peptic ulcer, gastroduodenal erosions, and varices. On the other hand, even in patients with tumors outside the GI tract, metastases were the source of bleeding in a significant number of patients (11%).
Risk Factors for Acute Nonvariceal Upper Gastrointestinal Bleeding
Risk factors for peptic ulcer bleeding are H pylori infection, use of NSAIDs, use of low-dose aspirin, and other antiplatelet medications or oral anticoagulants.
Helicobacter pylori infection
H pylori infection is found in 43% to 56% of peptic ulcer bleeding patients. The true H pylori prevalence in bleeding peptic ulcer is probably underestimated: in a recent meta-regression on 71 studies, including 8496 patients, the mean prevalence of H pylori infection in peptic ulcer bleeding was 72%. The most significant variables associated with a high prevalence of H pylori infection were the use of a diagnostic test delayed until at least 4 weeks after the bleeding episode (odds ratio [OR] 2.08; 95% CI, 1.10–3.93) and a lower mean age of patients (OR 0.95 per additional year; 95% CI, 0.92–0.99). The low prevalence of H pylori infection reported in peptic ulcer bleeding may be due to the methodology of the studies and to patient characteristics. These data also support the recent recommendations of an international consensus on nonvariceal UGIB regarding the performance of a delayed diagnostic test when H pylori tests carried out during the acute bleeding episode are negative.
The incidence of H pylori –negative idiopathic bleeding ulcers (ie, those not related to NSAIDs or other gastroerosive medications) is rising. These bleeding ulcers account for 16.1% of patients admitted for UGIB and 42.4% of patients who bled while in the hospital. These ulcers are also prone to recurrent complications at 12 months: 13.4% (95% CI, 7.3%–19.5%) versus 2.5% (95% CI, 0.4%–4.6%) in patients with H pylori –positive ulcers who received eradication therapy.
H pylori –negative bleeding ulcers are also associated with poorer outcomes regardless of use of NSAIDs. In a study from the University of Texas, patients without H pylori infection had significantly more comorbid medical conditions and higher Charlson index comorbidity scores than those with H pylori . Recurrent bleeding within 30 days was more frequent (11% vs 5%, P = .009) and hospital length of stay was significantly longer compared with H pylori –positive patients. Such outcome data confirm previous findings of significantly higher incident rates of rebleeding and death in patients with H pylori -negative idiopathic ulcers than in controls with H pylori –positive bleeding ulcers. Moreover, gastroprotective agents, such as PPIs or H 2 -receptor antagonists, did not reduce the risk of recurrent bleeding or mortality for patients with H pylori –negative idiopathic bleeding ulcers.
Current guidelines recommend testing for H pylori infection among users of low-dose aspirin who are at high risk for developing ulcers, because the long-term incidence of rebleeding with aspirin use is reduced after H pylori is eradicated. This was confirmed by a recent prospective study from Hong Kong, in which the incidence of ulcer bleeding (per 100 patient-years) in the H pylori –eradicated cohort did not differ significantly from that of the average-risk cohort (no history of ulcers). Aspirin users without current or past H pylori infections who develop ulcer bleeding have a 5-fold incidence of recurrent bleeding.
Low-dose aspirin, antiplatelets, and nonsteroidal antiinflammatory drugs
Long-term use of aspirin is recommended for prevention of cardiovascular events among patients with prior cardiovascular disease or multiple risk factors. Regular aspirin use is associated with an increased risk of major GI bleeding. A recent meta-analysis found an approximately 2-fold higher risk of GI bleeding among individuals regularly using aspirin compared with placebo, with no difference between 75 to 162.5 mg/d and greater than 162.5 to 325 mg/d. Such a magnitude of risk is confirmed by a large prospective study of 87,680 women in which the relative risk (RR) of major UGIB requiring hospitalization or blood transfusion was 1.43 (95% CI, 1.29–1.59) over a 24-year follow-up period. Furthermore, when used for primary prevention of cardiovascular disease, the absolute harms of aspirin seem to exceed its benefits: a meta-analysis of 27 studies showed that 60 to 84 major cardiovascular events per 100,000 person-years were averted, whereas 68 to 117 GI bleeds were incurred. There was a nonsignificant change in total cardiovascular disease, whereas risks were increased by 37% for GI bleeds. Risk factors for UGIB among low-dose aspirin users include (1) a history of peptic ulcer disease or GI bleeding; (2) older age; (3) concomitant use of NSAIDs, including cyclooxygenase (COX)-2 inhibitors; (4) concomitant use of anticoagulants or other platelet aggregation inhibitors; (5) the presence of severe medical comorbidities; and (6) high aspirin dose. H pylori and aspirin seem to be independent risk factors for peptic ulcer bleeding, so that in patients with a history of peptic ulcer disease, H pylori infection should be assessed and treated if present.
At-risk low-dose aspirin users are recommended to take gastroprotective agents. PPIs seem superior to eradication only to prevent recurrent ulcer bleeding in patients using low-dose aspirin. Moreover, in patients with acute coronary syndrome or myocardial infarction receiving aspirin, clopidogrel, enoxaparin, or thrombolytics, PPIs are superior to H 2 -receptor antagonists.
Population-based epidemiologic data confirm that the use of low-dose aspirin, NSAIDs, warfarin, and the combination thereof are significantly more common among UGI bleeders than nonbleeders. An increasing proportion of patients are nowadays hospitalized due to medication-related UGIB. Comparing data from Italy and the United Kingdom, the proportion of UGIB patients taking NSAIDs or antiplatelet agents and that of patients on anticoagulants was almost identical (46.4% vs 44.2% and 11.7% vs 13.2%, respectively). The increased risk of UGIB associated with NSAIDs seems lower than previously reported: in a Spanish primary care research database study of more than 660,000 subjects, increased risks were found with current use of NSAIDs (RR 1.72; 95% CI, 1.41–2.09), low-dose aspirin (RR 1.74; 95% CI, 1.37–2.21), other antiplatelet drugs (RR 1.73; 95% CI, 1.27–2.36), and oral anticoagulants (RR 2.00; 95% CI, 1.44–2.77). The use of oral corticosteroids, selective serotonin reuptake inhibitors, or acetaminophen was not associated with an increased risk.
In patients at risk for GI bleeding and using NSAIDs, a protective drug is recommended. Acid-suppressing drugs reduce the risk among users of NSAIDs (OR 0.58; 0.39–0.85), particularly in users with antecedent peptic ulcer (OR 0.16; 0.05–0.58). Unfortunately, despite several society and national guidelines that have been formulated, these are poorly followed in clinical practice and gastroprotection is largely underused. Data from Europe show that although approximately one-half of all patients with peptic ulcer bleeding were using NSAIDs or aspirin, an effective gastroprotective agent was only used in 10% to 15% of at-risk patients. A multinational study from the United Kingdom, Italy, and the Netherlands confirmed that the risk of UGIB is significantly higher in nonselective NSAID users with gastroprotective nonadherence. Among 618,684 nonselective NSAID users, nonadherers to gastroprotection had a 2-fold risk of UGIB (OR 1.89; 95% CI, 1.09–3.28). Selective cyclooxygenase (COX)-2 inhibitors are associated with a lower risk of clinically significant UGIB than nonselective NSAIDs ; however, concerns over the possible cardiovascular adverse effects of some of these agents should be taken into account. Moreover, switching to selective COX-2 inhibitors in patients with previous bleeding is not completely risk-free, and concomitant PPI therapy may also be needed. The combination of COX-2 inhibitors with PPIs promotes the greatest risk reduction for NSAID-related UGIB.
A retrospective study among general practitioners in France documented that within 2 years after prescribing a PPI, physicians did not renew this prescription for approximately 33% of those patients at risk for GI events (>65 years, past history of GI ulcer, or receiving antiplatelet agents) receiving continuous NSAIDs. Predictors for no longer receiving a prescription for a PPI included switching to a COX-2 selective inhibitor or to a nonselective NSAID and female gender. The risk for upper GI injury was higher among patients with discontinued PPI prescriptions (OR 1.45; 95% CI, 1.06–2.09).
UGIB is a rare but serious potential side effect of bisphosphonate therapy. In a population-based nested cohort study in Canada within an exposure cohort of 26,223 subjects, 117 individuals suffered a serious UGIB within 120 days of starting a bisphosphonate (0.4%). Age greater than 80 (adjusted OR 2.03; 95% CI, 1.40–2.94) and a past history of serious UGIB (adjusted OR 2.28; 95% CI, 1.29–4.03) were the strongest predictors. Men were 70% more likely to suffer an UGIB compared with women (adjusted OR 1.69; 95% CI, 1.05–2.72).
Patients with UGIB have increasing non-GI comorbidities. Results of a matched case-control study on 16,355 patients with nonvariceal UGIB and 81,636 controls showed that non-GI comorbidity had a strong association with UGIB; the adjusted OR for a single comorbidity was 1.43 (95% CI, 1.35–1.52) and for multiple or severe comorbidity was 2.26 (95% CI, 2.14–2.38). The additional population attributable fraction for comorbidity (19.8%; 95% CI, 18.4–21.2) was considerably larger than that for any other measured risk factor, including aspirin or NSAIDs use (3.0% and 3.1%, respectively). Non-GI comorbidity is an independent risk factor for UGIB and contributes to a greater proportion of patients with bleeding in the population than other recognized risk factors. These findings could explain why the incidence of nonvariceal bleeding remains high in older populations.
Mortality from Acute Upper Gastrointestinal Bleeding
Despite advances in endoscopic hemostasis and adjuvant pharmacologic treatment, the overall mortality from UGIB remains 5% to 14%, although most studies from the United States, Europe, and Asia place that figure closer to 5%. A systematic review of 18 studies (10 using administrative databases and 8 using bleeding registries) showed mortality rates from acute UGIB ranging from 1.1% in Japan to 11% in Denmark. The 28-day mortality after nonvariceal UGIB in England decreased from 14.7% in 1999 to 13.1% in 2007. A recent analysis of the trends for incidence of hospitalization and death due to UGIB in the United States from 2001 to 2009 confirmed that age/gender-adjusted case fatality owing to bleeding is low (2.5%) and increases with age but remains less than 5% even in elderly patients. These discrepancies in reported mortality rates of nonvariceal UGIB are attributable to differences in study methodologies and populations studied (heterogeneous definitions of case ascertainment, differing patient populations with regard to severity of presentation and associated comorbidities, varying durations of follow-up, and different health care system-related practices). More uniform standards in reporting would enable a better understanding of causes of death and the apparent discrepant outcome in endoscopic end points versus clinical end points. Nonvariceal UGIB is now predominantly a disease of the elderly, with more than 60% of patients over the age of 60 years and approximately 20% over the age of 80 years. Because elderly patients have more comorbid illness, are more likely users of aspirin and NSAIDs, and are less tolerant of hemodynamic insult, the management of this high-risk population is a major challenge. Current evidence indicates that most peptic ulcer bleeding–linked deaths are not a direct sequela of the bleeding ulcer itself. Instead, mortality derives from multiorgan failure, cardiopulmonary conditions, or end-stage malignancy, suggesting that improving further current treatments for the bleeding ulcer may have a limited impact on mortality unless supportive therapies are developed for the global management of these patients. In a prospective cohort study enrolling more than 10,000 cases of peptic ulcer bleeding at the Prince of Wales Hospital in Hong Kong, one of the most reputed bleeding centers worldwide, overall mortality was 6.2%. The study reported that approximately 80% of deceased patients died of non–bleeding-related causes.
Risk factors for mortality after nonvariceal UGIB are
- •
Increasing age
- •
Hemodynamic instability on admission
- •
Presence of severe and life-threatening comorbid medical conditions
Recent data from Italy also identified an American Society of Anesthesiologists (ASA) score of 3 or 4 versus 1 or 2 as the variable with the greatest OR for predicting mortality (OR 3.92; 95% CI, 2.37–6.50). One or more comorbidities are present in almost two-thirds of UGIB patients. Underlying comorbidity is consistently associated with an increased short-term mortality in patients with peptic ulcer bleeding. A systematic review and meta-analysis of 16 studies showed that the risk of 30-day or in-hospital mortality was significantly greater in patients with comorbidity than in those without (RR 4.44; 95% CI, 2.45–8.04). Patients with 3 or more comorbidities had a greater risk of dying than those with 1 or 2 (RR 3.46; 95% CI, 1.34–8.89). Among individual comorbidities that significantly increased the risk of death, RRs were higher for hepatic, renal, and malignant disease (RR range, 4.04–6.33) than for cardiovascular and respiratory disease and diabetes (2.39, 2.45, and 1.63, respectively). Coagulation disorders are also independently associated with more than a 5-fold increase in the odds of in-hospital mortality.
Cirrhotic patients suffering a nonvariceal bleed have significantly greater in-hospital mortality than noncirrhotic patients. The presence of cirrhosis independently increased mortality (adjusted OR 3.3; 95% CI, 2.2–4.9). Decompensated cirrhosis had higher mortality than compensated cirrhosis. Cryptogenic etiology of cirrhosis, renal dysfunction, actively bleeding ulcers on hospital admission, concurrent presence of duodenal ulcer and erosive disease, and bleeding from vascular lesions are all independent predictors of mortality.
Impaired renal function is an independent risk factor for UGIB. Crude rates of acute nonvariceal UGIB among patients undergoing dialysis have not decreased in the past 10 years. Although 30-day mortality related to UGIB declined, the burden on the end-stage renal disease population remains substantial. Overall 30-day mortality is 4.8% to 13.7%, with a 2-fold risk of death in peptic ulcer bleeding patients with end-stage renal disease compared with those without. Such a high mortality in patients undergoing dialysis was significantly correlated with older age, female gender, infection during hospitalization, single episodic UGIB, abnormal white blood cell count, and albumin level less than or equal to 3 g/dL.
Mortality is also increased in patients who are already admitted in hospital for another medical problem. In-hospital bleeding accounts for 10% to 25% of the overall nonvariceal bleeding population and is associated with poorer outcome, with mortality rates as high as 26%. Nonetheless, the reasons for increased mortality in this subgroup of patients have not been consistently identified. Guidelines on optimal management of inpatients who develop nonvariceal UGIB have been derived essentially from studies on outpatient bleeding, whereas few data are available that focus on in-hospital bleeding and its management. Recent data from Italy shed some light on the issue, showing that the mortality rate for in-hospital bleeding was significantly higher than that of outpatient bleeding (8.9% vs 3.8%; OR 2.44; 95% CI, 1.57–3.79). Hemodynamic instability on presentation and the presence of severe comorbidity were the strongest predictors of mortality for in-hospital bleeders.
The risk of mortality increases with rebleeding, which is thus another major outcome parameter. Rebleeding rates, using a combination of endoscopic hemostasis and adjuvant acid-suppressing therapy, have been shown to be reduced to less than 10% and it would be difficult, if not impossible, to reduce this further. Because a majority of nonvariceal UGIB patients today are of advanced age, users of NSAIDs and aspirin, and with multiple comorbid illnesses, the “endeavor to further reduce this rebleeding rate is an uphill battle.”
Little is known about mortality associated with nonulcer, nonvariceal UGIB. This may stem from a perception among clinicians that, other than those patients with a GI malignancy as the causative bleeding lesion, outcomes of patients with nonulcer bleeding are favorable. Therefore, it is possible that the usual algorithms of postendoscopy care, pharmacotherapy, and monitoring that are applied to patients in whom endoscopic hemostasis has been achieved after peptic ulcer bleeding may not be applied with the same rigor to those with nonulcer bleeding. A large prospective, multicenter study of 3207 patients with documented nonulcer UGIB have a risk of death similar to bleeding peptic ulcers. Mortality was 9.8% for neoplasia, 4.8% for Mallory-Weiss tears, 4.8% for vascular lesions, 4.4% for gastroduodenal erosions, 4.4% for duodenal ulcer, and 3.1% for gastric ulcer. Frequency of death was not different among benign endoscopic diagnoses. The strongest predictor of mortality was the overall ASA score. After adjusting for ASA score, the endoscopic diagnosis had no impact on mortality, suggesting that nonulcer causes of bleeding carry a risk of mortality similar to that of peptic ulcers. Therefore, even a bleed from what is perceived to be a minor lesion in a high-risk patient can be a sufficient precipitant to initiate a downward cascade of clinical events, resulting in multiorgan failure and death. The message could not be clearer: treat the patient and not just the source of GI bleeding.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
