In theClinic
In the Clinic
Physician Writers
Meeta Prasad Kerlin, MD, MSCE
Jeffrey L. Tokar, MD
Section Editors
Deborah Cotton, MD, MPH
Darren Taichman, MD, PhD
Sankey Williams, MD
Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013
Acute
Gastrointestinal
Bleeding
Prevention page ITC2-2
Presentation and Diagnosis page ITC2-3
Treatment page ITC2-7
Practice Improvement page ITC2-13
Tool Kit page ITC2-14
Patient Information page ITC2-15
CME Questions page ITC2-16
The content of In the Clinic is drawn from the clinical information and education
resources of the American College of Physicians (ACP), including PIER (Physicians’
Information and Education Resource) and MKSAP (Medical Knowledge and Self-
Assessment Program). Annals of Internal Medicine editors develop In the Clinic
from these primary sources in collaboration with the ACP’s Medical Education and
Publishing divisions and with the assistance of science writers and physician writ-
ers. Editorial consultants from PIER and MKSAP provide expert review of the con-
tent. Readers who are interested in these primary resources for more detail can
consult http://pier.acponline.org, http://www.acponline.org/products_services/
mksap/15/?pr31, and other resources referenced in each issue of In the Clinic.
CME Objective: To review current evidence for the prevention, presentation and
diagnosis, treatment, and practice improvement of acute gastrointestinal bleeding.
The information contained herein should never be used as a substitute for clinical
judgment.
© 2013 American College of Physicians
 cute gastrointestinal (GI) bleeding is common in both the outpatient

A setting and the emergency department. Annual U.S. incidence rates over
the past decade are approximately 90–108 per 100 000 persons (1), lead-
ing to approximately 300 000 hospitalizations annually. Most cases are due to
nonvariceal sources of bleeding (e.g., peptic ulcers) and continue to be associated
1. Hreinsson JP, Kalaitza-
with significant mortality (3–14%) and health economic burden (1–3). The inci-
kis E, Gudmundsson dence of nonvariceal bleeding may be decreasing in the West largely because of
S, Björnsson ES. Up-
per gastrointestinal decreased incidence of Helicobacter pylori infection and increased awareness and
bleeding: incidence,
etiology and out- implementation of ulcer-prevention strategies in users of nonsteroidal anti-
comes in a popula-
tion-based setting.
inflammatory drugs (NSAIDs) (1). However, identifying patients with acute GI
Scand J Gastroen- bleeding who are in danger of serious adverse events and establishing evidence-
terol. 2013;48:439-47.
[PMID: 23356751] based treatment plans are essential in both primary and specialty care.
2. Yavorski RT, Wong RK,
Maydonovitch C, et al.
Analysis of 3,294 cases
of upper gastrointesti-
nal bleeding in mili-
tary medical facilities.
Prevention
Am J Gastroenterol. Who is at risk for acute GI bleeding? with lower GI bleeding include in-
1995;90:568-73.
[PMID: 7717312] Risk factors for acute GI bleeding flammatory bowel disease, infectious
3. Targownik LE, Nabal- vary according to the site and cause. colitis, neoplasia, angioectasias, and
amba A. Trends in
management and Upper GI bleeding most often re- benign anorectal disease (14).
outcomes of acute
nonvariceal upper
sults from peptic ulcer disease (ap-
gastrointestinal bleed- proximately one quarter of all cases), Approximately 10–20% of patients
ing: 1993-2003. Clin
Gastroenterol Hepatol. the primary risk factors for which with GI bleeding have “obscure”
2006;4:1459-1466. include NSAIDs and H. pylori infec- bleeding, defined as an unknown
[PMID: 17101296]
4. McColl KE, el-Nujumi tion and, less commonly, increased cause despite evaluation with esopha-
AM, Chittajallu RS, et
gastric acid production (e.g., the gogastroduodenoscopy (EGD),
al. A study of the
pathogenesis of Heli- Zollinger-Ellison syndrome). Smok- colonoscopy, and radiographic small
cobacter pylori nega-
tive chronic duode- ing (4–6), severe physiologic stress bowel imaging. Approximately half
nal ulceration. Gut. (7), and various host factors (e.g., of these patients have recurrent or
1993;34:762-8.
[PMID: 8314508] genetic polymorphisms affecting cy- persistent bleeding and are further
5. Kurata JH, Nogawa
clooxygenase and prostaglandin pro- subclassified as obscure-overt (pas-
AN. Meta-analysis of
risk factors for peptic duction) (8, 9) may increase risk for sage of visible blood with melena or
ulcer. Nonsteroidal
antiinflammatory peptic ulcers even in persons without hematochezia) or obscure-occult
drugs, Helicobacter concomitant H. pylori infection or (iron-deficiency anemia and/or posi-
pylori, and smoking. J
Clin Gastroenterol. NSAID exposure. Although spicy tive for fecal occult blood) (15).
1997;24:2-17.
foods may cause GI symptoms, there Many such patients have bleeding
[PMID: 9013343]
6. Talamini G, Zamboni are no convincing data that they in- sources in the small intestine, now
G, Cavallini G. Antral
mucosal Helicobacter crease the risk for peptic ulcers. Oth- sometimes referred to as “mid-GI
pylori infection densi- er risk factors for acute upper GI bleeding” (between the ligament of
ty as a risk factor of
duodenal ulcer. Di- bleeding include varices, esophagitis, Treitz and the ileocecal valve). An-
gestion. 1997;58:211-
7. [PMID: 9243115]
vascular abnormalities (e.g., angioec- gioectasia is the most common cause
7. Barkun AN, Bardou M, tasias, arteriovenous malformations, of small-bowel bleeding in the West,
Pham CQ, Martel M.
Proton pump in- Dieulafoy lesions), Mallory–Weiss accounting for 70–80% of cases (16).
hibitors vs. histamine tear from protracted vomiting, and
2 receptor antago- Can acute GI bleeding be prevented?
nists for stress-related benign and malignant neoplasms
mucosal bleeding
(10–13). Prevention of acute GI bleeding de-
prophylaxis in critical-
ly ill patients: a meta- pends on the risk factors and causes.
analysis. Am J Gas- Lower GI bleeding, historically de- For example, reducing use of
troenterol.
2012;107:507-20. fined as bleeding from a source distal NSAIDs and administering antacid
[PMID: 22290403]
8. Arisawa T, Tahara T,
to the ligament of Treitz, also results treatment with H2-inhibitors or pro-
Shibata T, et al. Asso- from several causes with distinct risk ton-pump inhibitors (PPIs) prevents
ciation between ge-
netic polymorphisms factors. Diverticulosis is the most peptic ulcer bleeding. Prophylactic
in the cyclooxyge- common cause of hematochezia, ac- acid suppression should be considered
nase-1 gene promot-
er and peptic ulcers counting for up to half of all cases, in selected hospitalized patients who
in Japan. Int J Mol
Med. 2007;20:373-8.
particularly in patients older than 65 are at increased risk for gastroduode-
[PMID: 17671743] years. Other risk factors associated nal ulceration and bleeding, including

© 2013 American College of Physicians ITC2-2 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 patients who are mechanically venti-
lated or who have coagulopathy or
thrombocytopenia, traumatic brain or
spinal cord injury, or a history of
burns (7, 17). In patients with chronic
liver disease, nonselective beta-
blockers (such as propanolol or
nadolol) to reduce portal hypertension
and endoscopic interventions, such as
band ligation (18), can effectively pre-
vent variceal bleeding. No specific
measures can prevent diverticular
bleeding or bleeding related to an-
gioectasias. Although high-fiber diets
are often recommended, the role of
fiber in the pathogenesis or progres-
sion of diverticulosis is debatable. Sur-
gical intervention for diverticulosis is
not routinely done for prevention, be-
cause risk for diverticular bleeding is
low relative to the morbidity of pro-
phylactic surgical resection (19).

Prevention... Risk factors for, and therefore prevention of, acute GI bleeding
depends on the site and cause of bleeding. In general, minimizing use and appro-
priate prescribing of NSAIDs, antiplatelet agents, and anticoagulants, as well as
judicious primary and secondary prophylactic acid suppression in selected pa-
tients, are effective measures for ulcer-related upper GI bleeding. Nonselective
beta-blockers and endoscopic therapy for esophageal and gastric varices are ef-
fective for primary and secondary prevention of variceal bleeding. Few measures
are helpful in preventing lower GI bleeding, except for reducing exposure to
NSAIDS, anticoagulants, and antiplatelets.

CLINICAL BOTTOM LINE

Presentation and
What are the symptoms and signs it can occur with small-bowel bleed- Diagnosis
of acute GI bleeding? Can they ing and even slowly bleeding right
help localize the site of bleeding? colonic lesions. The source of hema-
GI bleeding can present with myriad tochezia is usually the colon. Up to
signs and symptoms (Table 1). The 10% of upper GI bleeding episodes
most indolent forms may present as present with hematochezia, which is 9. Malaty HM, Graham
DY, Isaksson I, En-
severe anemia. Manifestations include often associated with signs and symp- gstrand L, Pedersen
fatigue; dizziness; pallor; and rarely toms of hemodynamic instability. NL. Are genetic influ-
ences on peptic ulcer
end-organ complications, such as un- dependent or inde-
stable angina. Brisk bleeding from the What are the common causes of pendent of genetic
influences for Heli-
upper or lower GI tract can present upper and lower GI bleeding? cobacter pylori infec-
tion? Arch Intern
with more specific manifestations, Major causes of GI bleeding are Med. 2000;160:105-9.
such as visualized blood, syncope, or shown in the Box. Upper GI bleed- [PMID: 10632311]
10. Boonpongmanee S,
other symptoms of hypotension. ing is approximately 5 times as com- Fleischer DE, Pezzul-
lo JC, et al. The fre-
mon as lower GI bleeding, with quency of peptic ul-
Distinguishing between upper and mortality rates of 10%. Upper GI cer as a cause of
upper-GI bleeding is
lower GI bleeding can help guide ini- bleeding can be classified as non- exaggerated. Gas-
tial diagnosis and therapy. Upper GI variceal or variceal. trointest Endosc.
2004;59:788-94.
bleeding is more likely to cause nau- [PMID: 15173790]
sea and dyspepsia, whereas lower GI Methods allowing visualization and 11. Enestvedt BK, Gral-
nek IM, Mattek N,
bleeding is more likely to result in al- treatment of lesions deep in the Lieberman DA, Eisen
G. An evaluation of
tered bowel habits, lower abdominal small intestine have recently been endoscopic indica-
pain, or rectal discomfort. Hemate- developed. This has led to increasing tions and findings
related to nonva-
mesis is exclusive to upper GI bleed- use of the term “mid-GI bleeding” riceal upper-GI hem-
orrhage in a large
ing. Melena, which can be caused by (MGIB) (20). MGIB can present multicenter consor-
as little as 50 mL of blood, usually re- with occult bleeding (iron-deficiency tium. Gastrointest
Endosc. 2008;67:422-
sults from upper GI bleeding; however, anemia and stools positive for occult 9. [PMID: 18206878]

6 August 2013 Annals of Internal Medicine In the Clinic ITC2-3 © 2013 American College of Physicians

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 Table 1. History and Physical Examination Elements of Patients with Acute Upper Gastrointestinal Bleeding
Category Physical Examination Finding Notes
History Hematemesis Indicates bleeding proximal to the ligament of Treitz
Melena Should not be confused with dark stool of another etiology
(test with guaiac to confirm the presence of blood)
Bloody diarrhea 10% of patients will have bleeding sources proximal to the
ligament of Treitz
Presyncope and/or syncope Indicates loss of significant blood volume
Physical examination Hypotension (systolic blood pressure Indicates severe intravascular volume loss (≥50%),
90 mm Hg) assuming normal baseline systolic blood pressures
Tachycardia (≥120 beats/min) Indicates severe intravascular volume loss (≥50%)
Orthostatic changes in blood pressure Indicates loss of 20–25% of intravascular volume
(≥10 mm Hg) or heart rate (≥30/min)
Nasogastric aspirate shows blood or Indicates an upper gastrointestinal source of bleeding
coffee ground–like material
Pallor Subjective/poor indicator without corroborative evidence
Perioral telangiectasias Suggestive of hereditary hemorrhagic telangiectasia syndrome
Skin abnormalities Stigmata of cirrhosis, pigmented lip lesions, acanthosis
nigricans, vascular anomalies

From Bjorkman DJ, Eisen GM. Gastrointestinal bleeding, non-variceal upper. http://pier.acponline.org/physicians/diseases/d184/tables/d184-thp.html. (Date
accessed 26 May 2009). PIER. Philadelphia: American College of Physicians; 2009.

blood) or overt bleeding (melena, with acute GI bleeding (e.g., outpa-

12. Longstreth GF. Epi-
demiology of hospi- hematochezia). More frequently tient vs. inpatient management) and
talization for acute
upper gastrointesti-
encountered sources of MGIB are provide prognostic information. Pre-
nal hemorrhage: a vascular lesions (e.g., angioectasias, dictors of rebleeding risk and mor-
population-based
study. Am J Gas-
Dieulafoy lesions, arteriovenous tality in patients vary according to
troenterol. malformations), ulcerating disorders the cause of bleeding; however, sev-
1995;90:206-10.
[PMID: 7847286] (e.g., NSAID-enteropathy and in- eral factors portend a poorer prog-
13. Cook DJ, Fuller HD, flammatory bowel disease), and neo-
Guyatt GH, et al. Risk nosis. Adverse outcomes have also
factors for gastroin- plasms (e.g., neuroendocrine tumors, been associated with chronic alco-
testinal bleeding in
critically ill patients. GI stromal tumors, lymphoma, and holism and active cancer (22).
Canadian Critical
Care Trials Group. N
primary or metastatic carcinomas).
Engl J Med. The International Consensus Upper
1994;330:377-81. The most common source of lower Gastrointestinal Bleeding Confer-
[PMID: 8284001]
14. Strate LL. Lower GI GI bleeding is diverticulosis, reflective ence Group advocates use of validat-
bleeding: epidemiol-
ogy and diagnosis.
of the high prevalence of this condi- ed risk-stratification tools to facilitate
Gastroenterol Clin tion in the adult population; however, triage of patients with acute upper
North Am.
2005;34:643-64. only a few patients with diverticulosis GI bleeding by prognostication based
[PMID: 16303575] have diverticular bleeding. Other on risk for poor clinical outcomes
15. Rondonotti E, Mar-
mo R, Petracchini M, common causes of lower GI bleeding (e.g., rebleeding, need for emergent
de Franchis R, Pen-
nazio M. The Ameri-
include colonic polyps, cancer, colitis surgery, or mortality) (22). The
can Society for Gas- (due to inflammatory bowel disease,
trointestinal Rockall scoring system (23) and the
Endoscopy (ASGE) infection, or ischemia), vascular
Glasgow–Blatchford Scale (24)
diagnostic algorithm lesions (angioectasia, radiation procti-
for obscure gastroin- (Table 2) are 2 commonly used risk-
testinal bleeding: tis), and anorectal sources (hemor-
eight burning ques- stratification systems that incorporate
tions from everyday rhoids, anal fissures) (21).
clinical practice. Dig
clinical, laboratory, and/or endoscopic
Liver Dis. Can risk for adverse outcomes be parameters. The Glasgow–Blatchford
2013;45:179-85.
[PMID: 22921043] predicted in patients with acute and the “Clinical” Rockall stratifica-
16. Pennazio M. Tech-
niques in Gastrotin- GI bleeding? Which patients may tion systems are preendoscopic
testinal Endoscopy. be evaluated as outpatients, and scoring systems and use clinical and
2012;14:94-99.
17. Alain BB, Wang YJ. which require the emergency laboratory information to help pre-
Cushing’s ulcer in
traumatic brain in-
department or hospitalization? dict the need for hospitalization or
jury. Chin J Trauma- Risk stratification can help deter- endoscopic intervention. Additional
tol. 2008;11:114-9.
[PMID: 18377716] mine the disposition of a patient risk stratification by incorporating

© 2013 American College of Physicians ITC2-4 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 some endoscopic information to the
Table 2. Risk Stratification Tools for Major Causes of
Clinical Rockall scale leads to the
Upper Gastrointestinal Bleeding Gastrointestinal Bleeding
Full Rockall scale, which helps pre-
dict risk for adverse outcomes from Scoring System Points Inflammatory
Rockall Risk: Total score = sum of all category Peptic ulcer disease
acute upper GI bleeding. Additional scores; risk category: high (>5), intermediate Esophagitis or esophageal ulceration
stratification schemes for upper and (3–4), low (0–2) Diaphragmatic hernia (Cameron
lower GI bleeding have been pro- Age, y erosions)
posed (25–28). Outpatient manage- >80 2 Inflammatory bowel disease
ment may be appropriate for persons 60–79 1 Benign and malignant neoplasms
with low risk scores (Rockall score of <60 0 Primary gastrointestinal tract
0–2 or Glasgow–Blatchford score of Shock
neoplasms, at any site
0), whereas those with high risk Metastatic deposits in the
SBP <100 mm Hg 2
gastrointestinal tract, at any site
scores require inpatient care (24, SBP >100 mm Hg 1
29–34). Admission to an intensive Vascular anomalies
No shock 0 Gastroesophageal varices
care unit (ICU) should be considered Comorbid conditions Angioectasias
for patients with evidence of brisk, Renal failure, liver failure, 2 Dieulafoy lesion
active bleeding, such as hemodynam- widespread cancer Gastric antral vascular ectasia (GAVE,
ic instability (systolic blood pressure Cardiac failure, ischemic 1 also known as watermelon stomach)
[BP] <100 or pulse >100, orthostatic heart disease Radiation proctopathy
hypotension, evidence of shock), or No other comorbid conditions 0
Drug-induced
Diagnosis on EGD
other clinical parameters associated Aspirin
with high risk for rebleeding and Upper gastrointestinal cancer 2 Nonsteroidal anti-inflammatory drugs
mortality. To date, there are no wide- All other diagnoses 1 Miscellaneous
ly accepted validated scoring systems No lesion on EGD 0 Colonic diverticulosis
Stigmata of recent hemorrhage Postpolypectomy
specific to acute lower GI bleeding.
Blood in upper gastrointestinal 2
Mallory–Weiss tear
What should the initial diagnostic tract, adherent clot, visible vessel Meckel diverticulum
evaluation for possible acute GI No stigmata 0 From Rajan E, Ahlquist D. Gastro-
bleeding include? Glasgow–Blatchford: Total score = sum of all intestinal bleeding. In: Dale DC,
category scores; a score of zero suggests low risk Federman DD, eds. ACP Medicine, 3rd ed.
The initial evaluation of a patient and the safety of outpatient care New York: WebMD; 2007:863.
with acute GI bleeding should in- Blood urea nitrogen (mmol/L)
clude a focused history and physical >25 6
examination. The history should in- 10–<25 4
clude associated signs and symptoms; 8–<10 3
use of NSAIDs, antiplatelet agents, 6.5–<8 2 18. Garcia-Tsao G,
Sanyal AJ, Grace ND,
anticoagulants, selective serotonin re- <6.5 0 Carey W; Practice
uptake inhibitors, and beta-blockers Hemoglobin (g/dL) Guidelines Commit-
tee of the American
(because of issues of risk assessment <10 in men and women 6 Association for the
with tachycardia and resuscitation im- 10–<12 in men 3
Study of Liver Dis-
eases. Prevention
plications); prior GI bleeding epi- 10–<12 in women/12–<13 in men 1
and management of
gastroesophageal
sodes; and comorbid conditions. Vital ≥12 in women or ≥13 in men 0 varices and variceal
signs on postural changes should be SBP (mm Hg)
hemorrhage in cir-
rhosis. Hepatology.
assessed, and stool should be exam- <90 3 2007;46:922-38.
[PMID: 17879356]
ined. Resting hypotension (systolic 90–99 2 19. McGuire HH Jr. Bleed-
BP ≤90 mm Hg) or tachycardia 100–109 1
ing colonic diverticu-
la. A reappraisal of
(≥120 bpm) indicates severe intravas- >110 0 natural history and
management. Ann
cular volume loss (≥50%). An increase Other markers Surg. 1994;220:653-6.
of ≥30/min in the pulse or severe Cardiac failure 2
[PMID: 7979613]
20. Okazaki H, Fujiwara
lightheadedness when rising from a Hepatic disease 2 Y, Sugimori S, et al.
supine position also indicates signifi- Presentation with syncope 2
Prevalence of mid-
gastrointestinal
cant volume loss (35, 36). Presentation with melena 1
bleeding in patients
with acute overt
Pulse >100/min 1 gastrointestinal
Immediate laboratory tests should bleeding: multi-cen-
ter experience with
include a complete blood count, pro- EGD = esophagogastroduodenoscopy; SBP = 1,044 consecutive
thrombin and partial thromboplastin systolic blood pressure. patients. J Gastroen-
terol. 2009;44:550-5.
times, platelet count, blood type and [PMID: 19360374]

6 August 2013 Annals of Internal Medicine In the Clinic ITC2-5 © 2013 American College of Physicians

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 crossmatch in anticipation of blood requirements, rescue surgery, or
transfusion, and a routine chemistry mortality (41–43), urgent endoscopy
panel. An increased ratio of blood (<12 hours) is indicated in patients
urea nitrogen to creatinine (>25:1 if with suspected variceal bleeding (39,
in mg/dL) suggests an upper GI 44, 45). It also provides valuable in-
source of bleeding (35, 36). formation for appropriate patient
21. Zuckerman GR,
Prakash C. Acute triage (see “Can risk for adverse out-
lower intestinal Nasogastric or orogastric aspiration comes be predicted in patients with
bleeding. Part II: eti- may confirm an upper GI source of
ology, therapy, and acute GI bleeding?”).
outcomes. Gastroin- bleeding and provide prognostic
test Endosc.
1999;49:228-38. information on bleeding activity and What is the role of prokinetic
[PMID: 9925703]
22. Barkun AN, Bardou
severity. However, the procedure is medications before upper
M, Kuipers EJ, et al. uncomfortable; yields false-negative endoscopy in patients with acute
International con-
sensus recommen- results in approximately 15% of GI bleeding?
dations on the man-
agement of patients
patients with active bleeding; and Prokinetic medications (such as
with nonvariceal up- although it may lead to earlier per- erythromycin and, to a lesser extent,
per gastrointestinal
bleeding. Ann Intern formance of endoscopy, it has not metoclopramide) administered
Med. 2010;152:101- been proven to improve clinical out- intravenously 20 to 120 minutes be-
13. [PMID: 20083829]
23. Rockall TA, Logan RF, comes (30, 37). Some expert panels fore upper endoscopy has the poten-
Devlin HB, Northfield
TC. Selection of pa- recommend it in selected patients tial to facilitate clearance of blood
tients for early dis- but do not provide selection criteria and clots from the stomach, thus
charge or outpatient
care after acute up- (22, 38). Others state that it is not improving visualization in patients
per gastrointestinal with upper GI bleeding. However,
haemorrhage. Na-
required in patients with suspected
tional Audit of Acute upper GI bleeding for diagnosis, they do not seem to alter important
Upper Gastrointesti-
nal Haemorrhage. prognosis, visualization, or therapeu- clinical outcomes, including diag-
Lancet.
1996;347:1138-40.
tic effect (39). As a result, use of gas- nostic rates, need for transfusions or
[PMID: 8609747] tric aspiration should be dictated by surgery, or hospital length of stay.
24. Blatchford O, Murray
WR, Blatchford M. A institutional preference and practice. Routine use before upper endoscopy
risk score to predict is not advocated and should be re-
need for treatment
for upper-gastroin- When should a gastroenterologist served for patients with red blood
testinal haemor- be consulted in the evaluation of hematemesis or blood in the naso-
rhage. Lancet.
2000;356:1318-21. acute GI bleeding? gastric aspirate (22, 39, 46).
[PMID: 11073021]
25. Saltzman JR, Tabak A gastroenterologist should be con-
YP, Hyett BH, et al. A sulted early to consider prompt en- What adjunctive tests can help
simple risk score ac-
curately predicts in- doscopy for patients with GI bleeding evaluate (and/or treat) patients
hospital mortality,
length of stay, and
and to facilitate patient triage. EGD with acute GI bleeding without an
cost in acute upper or colonoscopy or both are the initial identified source on EGD or
GI bleeding. Gas-
trointest Endosc. diagnostic tests of choice for identifi- colonoscopy?
2011;74:1215-24.
cation and treatment of specific Patients without a demonstrable
[PMID: 21907980]
26. Strate LL, Orav EJ, bleeding lesions. Clinical presentation bleeding source on good-quality up-
Syngal S. Early pre-
dictors of severity in dictates which procedures are done. per endoscopy or colonoscopy have
acute lower intestin- EGD is appropriate for patients with “obscure” GI bleeding. Small-bowel
al tract bleeding.
Arch Intern Med. melena and hematemesis and a subset barium radiography, “push” enterog-
2003;163:838-43.
[PMID: 12695275] of patients with hematochezia result- raphy, technetium-labeled red blood
27. Strate LL, Saltzman ing from an upper GI source. cell scan, and angiography have his-
JR, Ookubo R, Mutin-
ga ML, Syngal S. Vali- torically been considered as subse-
dation of a clinical In upper GI bleeding, early en- quent diagnostic tests, however with
prediction rule for
severe acute lower doscopy (within 24 hours of admis- low diagnostic yield. Wireless video
intestinal bleeding.
Am J Gastroenterol.
sion) has a greater impact than capsule endoscopy (VCE) has be-
2005;100:1821-7. delayed endoscopy on some clinical come most gastroenterologists’ test of
[PMID: 16086720]
28. Das A, Wong RC. outcomes, such as blood transfusion choice for patients with obscure GI
Prediction of out-
come of acute GI
requirements and hospital length of bleeding. The diagnostic yield ranges
hemorrhage: a re- stay (40). Although emergent en- from 30–50% (7); however, no VCE
view of risk scores
and predictive mod- doscopy (<6–8 hours) does not systems can provide hemostatic inter-
els. Gastrointest En-
dosc. 2004;60:85-93.
seem to further reduce rebleeding, ventions, and many institutions are
[PMID: 15229431] hospital length of stay, transfusion unable to perform urgent inpatient

© 2013 American College of Physicians ITC2-6 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 VCE. Another option is angiogra- approximately 50–70% and 25–70%
phy, which allows for therapeutic in- of cases, respectively (16). Some cen-
tervention (embolization) if a lesion ters offer advanced deep enteroscopic
is localized; however, it requires ac- procedures, which are low-risk and
tive bleeding to be diagnostic. Com- enable visualization and therapeutics
puted tomography angiography or deep within the small intestine, es-
29. Chen IC, Hung MS,
computed tomography/magnetic sentially eliminating the need for Chiu TF, Chen JC,
resonance enterography are alterna- high-risk intraoperative enteroscopy Hsiao CT. Risk scor-
ing systems to pre-
tive diagnostic options that identify for patients with suspected small dict need for clinical
intervention for pa-
sources of obscure bleeding in bowel pathology. tients with nonva-
riceal upper gas-
trointestinal tract
bleeding. Am J
Presentation and Diagnosis... GI bleeding can present with myriad signs and Emerg Med.
2007;25:774-9.
symptoms, ranging from asymptomatic to overt hematemesis or hematochezia, [PMID: 17870480]
and can be due to a number of causes virtually anywhere along the GI tract. Ini- 30. Hwang JH, Fisher
DA, Ben-Menachem
tial evaluation, including history and physical examination and routine laboratory T, et al. The role of
tests, can help to narrow the differential diagnosis. endoscopy in the
management of
acute non-variceal
upper GI bleeding.
CLINICAL BOTTOM LINE Gastrointest Endosc.
2012;75:1132-8.
[PMID: 22624808]
31. Masaoka T, Suzuki H,
Hori S, Aikawa N,
Hibi T. Blatchford

Treatment scoring system is a

useful scoring sys-
tem for detecting
patients with upper
What interventions should be cardiovascular function, with the goal gastrointestinal
started immediately in patients bleeding who do
of normal, stable vital signs. not need endoscop-
with acute GI bleeding? ic intervention. J
Regardless of the source, initial man- Observational studies and small con- Gastroenterol Hepa-
tol. 2007;22:1404-8.
agement of a patient with acute GI trolled trials have suggested that blood [PMID: 17716345]
32. Srirajaskanthan R,
bleeding involves rapid diagnostic as- transfusion to a target helmoglobin Conn R, Bulwer C,
Irving P. The Glas-
sessment and aggressive resuscitation level of 9–10 mg/dL might actually gow Blatchford scor-
with isotonic fluids via 2 large-bore be detrimental for some patients with ing system enables
accurate risk stratifi-
(18-gauge or larger) peripheral IV hypovolemic anemia, including those cation of patients
with upper gastroin-
catheters. Early intensive resuscita- with GI bleeding. testinal haemor-
rhage. Int J Clin
tion has been associated with de- Pract. 2010;64:868-
A recent randomized controlled trial (49)
creased mortality from GI bleeding 74. [PMID: 20337750]
compared a “restrictive” vs. a “liberal” transfu- 33. Stanley AJ. Update
(48). The goal of resuscitation is to sion strategy in patients with acute upper GI on risk scoring sys-
tems for patients
maintain tissue perfusion until bleed- bleeding. The threshold for transfusion of red with upper gastroin-
ing resolves spontaneously or is con- blood cells was a hemoglobin value <7 g/dL testinal haemor-
rhage [Editorial].
trolled by more definitive therapy. in the restrictive transfusion group (target World J Gastroen-
terol. 2012;18:2739-
Smaller catheters (e.g., 20- or 22- range, 7–9 g/dL) and 9 g/dL in the liberal 44. [PMID: 22719181]
gauge) do not allow rapid fluid ad- transfusion group (target range, 9–11 g/dL). 34. Stanley AJ, Ashley D,
Dalton HR, et al. Out-
ministration. Longer central venous Overall, the restrictive transfusion strategy patient manage-
was associated with a significantly greater ment of patients
catheters increase resistance to fluid with low-risk upper-
flow, thereby slowing the rate at probability of survival at 6 weeks, lower inci- gastrointestinal
haemorrhage: multi-
which fluids may be given. Patients dence of further bleeding, fewer adverse centre validation
with emesis who are unable to pro- events, fewer total units of blood transfused, and prospective
evaluation. Lancet.
and less need for rescue therapy. Subgroup
tect the airway from aspiration 2009;373:42-7.
analysis revealed that the survival advan- [PMID: 19091393]
should be intubated. Isotonic IV flu- 35. Cappell MS, Friedel
tage of the restrictive strategy was limited to D. Initial manage-
ids, such as normal saline or lactated Child–Pugh class A and B cirrhosis (not class ment of acute upper
Ringer solution, should be given im- C). Patients with “massive exsanguinating
gastrointestinal
bleeding: from initial
mediately, with the goal of replenish- bleeding,” the acute coronary syndrome, evaluation up to
gastrointestinal en-
ing intravascular volume as quickly as symptomatic peripheral vasculopathy, doscopy. Med Clin
blood has been lost and as rapidly as stroke, transient ischemic attack, and lower North Am.
2008;92:491-509.
is tolerated by the patient’s underlying GI bleeding were excluded from the trial, and [PMID: 18387374]

6 August 2013 Annals of Internal Medicine In the Clinic ITC2-7 © 2013 American College of Physicians

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 a restrictive transfusion strategy may not be vessels, or active bleeding are associ-
applicable to every patient. ated with a high-risk for continued
or recurrent bleeding and warrant
Although coagulopathy has been pharmacologic treatment and such
36. Zuccaro G Jr. Man- shown to be a risk factor for non- endoscopic interventions as thermal
agement of the adult variceal upper GI bleeding, data are methods; injection with vasoconstric-
patient with acute
lower gastrointestinal limited and conflicting (48, 50, 51). tive agents, such as epinephrine; and
bleeding. American
College of Gastroen-
Acknowledging the paucity of pub- mechanical hemostasis with various
terology. Practice Pa- lished data on the topic, the recently clips or sealants; or some combina-
rameters Committee.
Am J Gastroenterol. published International Consensus tion (53, 54). Strong evidence
1998;93:1202-8.
[PMID: 9707037]
Recommendations advise that indicates that monotherapy with
37. Huang ES, Karsan S, coagulopathy in patients receiving epinephrine injection alone is less
Kanwal F, et al. Im-
pact of nasogastric anticoagulants be treated, but that effective than combination therapy
lavage on outcomes
in acute GI bleeding.
management should not delay thera- (injection plus thermal therapy, or
Gastrointest Endosc. peutic endoscopy unless the INR is clips followed by injection) (53, 54).
2011;74:971-80.
[PMID: 21737077] supratherapeutic (>2.5). Furthermore,
38. Barkun A, Bardou M,
Marshall JK; Nonva-
this approach should not be general- In patients with peptic ulcer bleed-
riceal Upper GI ized to patients with cirrhosis be- ing, PPI therapy (bolus followed by
Bleeding Consensus
Conference Group. cause the INR, at any threshold, infusion) before endoscopy has been
Consensus recom- cannot predict bleeding risk in these shown to decrease the likelihood of
mendations for
managing patients patients (22). In addition, a systemat- high-risk stigmata on subsequent
with nonvariceal up-
per gastrointestinal ic review of published studies report- endoscopy and to reduce the likeli-
bleeding. Ann Intern ed a lack of data on which to hood of requiring an intervention
Med. 2003;139:843-
57. [PMID: 14623622] recommend optimal platelet counts during endoscopy. It may also be as-
39. Laine L, Jensen DM.
Management of pa-
for patients with GI bleeding. Based sociated with shorter length of stay
tients with ulcer on expert opinion, they recommend- in the hospital. However, preendo-
bleeding. Am J Gas-
troenterol. ed targeting values of 50 000/µL in scopic PPI therapy has not been
2012;107:345-60; the absence of platelet dysfunction,
quiz 361. shown to reduce mortality or the
[PMID: 22310222] or 100 000/µL if functional platelet need for rescue surgery or rebleeding
40. Lin HJ, Wang K,
Perng CL, et al. Early dysfunction is suspected (52). rates when endoscopic therapy is
or delayed en-
doscopy for patients
consistently delivered (55, 56) (Ap-
with peptic ulcer How should acute upper GI pendix Figure, available at www.an-
bleeding. A prospec- bleeding due to peptic ulcer nals.org). Thus, preendoscopic PPIs
tive randomized
study. J Clin Gas- disease be managed? should be considered but should not
troenterol.
1996;22:267-71.
The initial diagnostic approach to delay early endoscopy (within the
[PMID: 8771420] bleeding peptic ulcers is the same as first 24 h) or replace resuscitation
41. Sarin N, Monga N,
Adams PC. Time to for other causes of GI bleeding. En- because its effect on outcomes is
endoscopy and out-
comes in upper gas-
doscopy is nearly 100% specific and minor at best (57). After endoscopy,
trointestinal bleed- >90% sensitive; can be done at the hemodynamically stable patients
ing. Can J
Gastroenterol. bedside in the emergency depart- without serious comorbid conditions
2009;23:489-93.
[PMID: 19623332]
ment, endoscopy unit, or ICU; and who are found to have low-risk ul-
42. Tai CM, Huang SP, allows biopsy to assess the cause of cers (e.g., clean-based, flat pigment-
Wang HP, et al. High-
risk ED patients with the ulcer (e.g., H. pylori infection or ed spot) can generally be discharged
nonvariceal upper
gastrointestinal
an ulcer within an adenocarcinoma). early on once-daily oral PPIs (58).
hemorrhage under- Endoscopists use the Forrest classifi-
going emergency or
urgent endoscopy: a cation to describe peptic ulcers and A meta-analysis of randomized trials of
retrospective analy- to predict rebleeding risk associated high-dose IV PPI therapy (80 mg bolus then 8
sis. Am J Emerg
Med. 2007;25:273-8. with different ulcer appearances. mg/h infusion for 72 h after endoscopic ther-
[PMID: 17349900]
Approximately 50% of ulcers have an apy) after successful endoscopic therapy for
43. Targownik LE,
Murthy S, Keyvani L, appearance associated with a low ulcers in patients with high-risk stigmata (i.e.,
Leeson S. The role of active bleeding, nonbleeding visible vessel or
rapid endoscopy for probability of rebleeding (i.e., clean
high-risk patients adherent clot, or Forrest classification IA, IB,
with acute nonva-
ulcer base or flat pigmented spot in IIa, or IIb) (54, 59) concluded that mortality,
riceal upper gas- the ulcer base). Patients with these need for surgery, and rebleeding is reduced
trointestinal bleed-
ing. Can J ulcers require only pharmacologic (relative risk 0.41, 0.43, and 0.40, respectively)
Gastroenterol.
2007;21:425-9.
treatment. By contrast, ulcers with compared with endoscopy alone. Another
[PMID: 17637943] adherent clots, nonbleeding visible meta-analysis also concluded that for

© 2013 American College of Physicians ITC2-8 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 patients with ulcer bleeding, PPIs reduce re- monitored closely for adverse effects
bleeding and need for surgery or repeated of volume replacement, including
endoscopic procedures and improve mortal- pulmonary edema. Targeting a
ity in the highest–risk patients (59). hemoglobin level between 7–8 g/dL 44. Hearnshaw SA, Lo-

The optimal dose and route of acute might prevent excessive restitution of gan RF, Lowe D, et al.
Use of endoscopy for
PPI administration remain unclear, blood volume and subsequent in- management of
acute upper gastroin-
but patients with high-risk lesions creases in portal pressure (62). testinal bleeding in
the UK: results of a
having endoscopic therapy should re- There are no definitive guidelines for nationwide audit.
Gut. 2010;59:1022-9.
ceive in-hospital therapy (IV, high- management of coagulopathy and [PMID: 20357318]
dose therapy remains the regimen thrombocytopenia in patients with
45. Lim CH, Ahmed MM.
The optimal timing
with the best evidence [22]) for 3 acute variceal hemorrhage. Reversal for urgent en-
days, mirroring the period during doscopy in nonva-
of coagulopathy requires administra- riceal upper gas-
which risk for rebleeding is greatest trointestinal
tion of fresh frozen plasma, although bleeding [Letter]. En-
(60, 61). In the absence of compara- doscopy.
the volume necessary may be prohibi-
tive data, a once-daily oral PPI is rec- 2011;43:1018.
tive. Recombinant factor VIIa as an [PMID: 22057771]
ommended after completion of 72 46. Barkun AN, Bardou
alternative means of normalizing the M, Martel M, Gralnek
hours of IV therapy (22, 30, 39).
prothrombin time has been proposed, IM, Sung JJ. Prokinet-
H-receptor blockers are not as effec- ics in acute upper GI
but randomized, controlled trials have bleeding: a meta-
tive and should not replace PPIs for analysis. Gastrointest
not shown a consistent advantage
acute management of bleeding ulcers. Endosc.
(62). Because prothrombin time–INR 2010;72:1138-45.
[PMID: 20970794]
How should acute esophageal is not a reliable indicator of the 47. Pasha SF. Diagnostic
yield of deep en-
variceal bleeding be treated? coagulation status in patients with teroscopy tech-
Acute bleeding from esophageal cirrhosis, recommendations regarding niques for small-
bowel bleeding and
varices is frequently life-threatening management of coagulopathy/throm- tumors. Tech Gastro
Endosc. 2012;14:100-
because it can be severe, difficult to bocytopenia cannot be made on the 105.
control, and rarely resolves sponta- basis of available data (63). 48. Baradarian R, Ramd-
haney S, Chapala-
neously. It usually occurs in patients madugu R, et al. Ear-
with end-stage liver disease. Eso- Antibiotic prophylaxis reduces infec- ly intensive
resuscitation of pa-
phageal varices are caused by signifi- tious complications and decreases tients with upper

cant portal hypertension; therefore, rebleeding with acute variceal bleed- gastrointestinal
bleeding decreases
bleeding occurs under high pressure ing. A short-term quinolone course mortality. Am J Gas-
troenterol.
and is often brisk. In addition, pa- (e.g., up to 7 days of norfloxacin or 2004;99:619-22.
tients have synthetic liver dysfunc- ciprofloxacin) is recommended. Data [PMID: 15089891]
49. Villanueva C, Colo-
tion and coagulopathy. Under these from a few studies suggest that cef- mo A, Bosch A, et al.
Transfusion strate-
circumstances, acute management triaxione may prevent bacterial infec- gies for acute upper
requires rapid and aggressive inter- tions better than norfloxacin, and IV gastrointestinal
bleeding. N Engl J
ventions, including fluid resuscita- ceftriaxone (1 g/day) may be prefer- Med. 2013;368:11-
21. [PMID: 23281973]
tion, bleeding control, and efforts able in patients with advanced cir- 50. Wolf AT, Wasan SK,
aimed at reducing portal pressures. rhosis and/or in regions with high Saltzman JR. Impact
of anticoagulation
prevalence of quinolone-resistant on rebleeding fol-
Intravascular volume replacement organisms (62, 64). lowing endoscopic
therapy for nonva-
should occur as with any patient riceal upper gas-
with acute GI hemorrhage. Resusci- Variceal bleeding can often be con- trointestinal hemor-
rhage. Am J
tation of patients with end-stage trolled by both medical and endo- Gastroenterol.
2007;102:290-6.
liver disease is often difficult due to scopic means. Medical therapy [PMID: 17100959]
hypoalbuminemia and extravasation primarily involves infusion of 51. Shingina A, Barkun
AN, Razzaghi A, Mar-
of fluid from the intravascular space. octreotide, a somatostatin analogue tel M, Bardou M,
Gralnek I; RUGBE In-
Some experts advocate preferential that causes splanchnic vasoconstric- vestigators. System-
use of blood products and albumin tion and reduced portal pressure. atic review: the pre-
senting international
for intravascular volume resuscita- Numerous trials of octreotide have normalised ratio
(INR) as a predictor
tion, because crystalloid fluids deliver yielded equivocal results, but it is of outcome in pa-
a sodium load to patients who are recommended in combination with tients with upper
nonvariceal gastroin-
typically already total-body sodium endoscopic therapies to control testinal bleeding. Ali-
overloaded and tend to eventually variceal bleeding and reduce risk for ment Pharmacol
Ther. 2011;33:1010-8.
exacerbate ascites. Patients should be recurrence (18). Octreotide should [PMID: 21385193]

6 August 2013 Annals of Internal Medicine In the Clinic ITC2-9 © 2013 American College of Physicians

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 be given as a bolus dose of 50 µg For patients with bleeding gastric
followed by continuous IV infusion varices, initial clinical and phar-
of 50 µg/hour for a total of 3 to 5 macologic management is similar
days, although many experts cur- to that for esophageal varices.
rently prefer a 5-day regimen (63). However, decision making regard-
ing definitive treatment options,
Endoscopy should be done within particularly for isolated gastric
12 hours in patients with known or varices, is more complex. Band lig-
suspected varices who present with ation, injection sclerotherapy, and
upper GI bleeding. There are 2 pri- TIPS have been reported and are
mary endoscopic interventions: sometimes used in clinical prac-
sclerotherapy and band ligation. tice. More recent therapeutic
Sclerotherapy involves injection of developments for treatment of
a sclerosing agent (e.g., sodium gastric varices include endoscopic
morrhuate or ethanolamine) into injection of biological glues and a
varices. Band ligation involves radiologic procedure called bal-
wrapping elastic bands over the loon-occlusion retrograde transve-
varices in the distal esophagus. nous obliteration. Recently,
Clinical trials have shown that self-expanding esophageal stents
compared with sclerotherapy, band- have been used to staunch bleed-
ing has more prolonged benefit; ing, particularly in refractory
lower rates of rebleeding, mortality, cases; however, further studies are
52. Razzaghi A, Barkun
and complications (such as needed (63).
AN. Platelet transfu- esophageal stricture formation,
sion threshold in pa-
tients with upper perforation, and mediastinitis); and How should patients with acute
gastrointestinal
bleeding: a system-
reduced need for endoscopic treat- lower GI bleeding from colonic
atic review. J Clin ments (65). Thus, band ligation has diverticulosis be treated?
Gastroenterol.
2012;46:482-6. become the acute treatment of Initial management of colonic diver-
[PMID: 22688143] choice in most institutions. ticular hemorrhage is the same as
53. Barkun AN, Martel
M, Toubouti Y, that for other causes of GI bleeding,
Rahme E, Bardou M. For patients with variceal bleeding including fluid resuscitation, blood
Endoscopic hemo-
stasis in peptic ulcer refractory to medical and endoscopic transfusion, and diagnostic testing.
bleeding for patients
with high-risk le-
therapy, balloon tamponade (e.g., Colonoscopy should be done to try
sions: a series of with a Sengstaken–Blakemore tube) to localize the bleeding, although
meta-analyses. Gas-
trointest Endosc. may be used as a temporizing meas- with brisk hemorrhage, localization
2009;69:786-99.
[PMID: 19152905]
ure. The balloons should be inflated can be difficult due to poor visuali-
54. Laine L, McQuaid KR. for no more than 12 hours. More zation. The timing of the colonos-
Endoscopic therapy
for bleeding ulcers: definitive treatments include trans- copy remains controversial; however,
an evidence-based jugular intrahepatic portosystemic it is usually done within 12–24
approach based on
meta-analyses of shunt (TIPS) placement with a hours of presentation with a rapid
randomized con-
trolled trials. Clin
PTFE stent and surgical interven- colonic preparation that seems to be
Gastroenterol Hepa- tions. In patients at high risk for safe in this setting (66). The most
tol. 2009;7:33-47.
[PMID: 18986845] treatment failure (e.g., Child–Pugh important contribution of colonos-
55. Lau JY, Leung WK,
Wu JC, et al.
class C score < 14 or class B with ac- copy is usually exclusion of other
Omeprazole before tive bleeding), TIPS should be causes. In rare cases, it can also be
endoscopy in pa-
tients with gastroin- placed within 72 hours (63). Salvage therapeutic if a visible vessel or ad-
testinal bleeding. N
Engl J Med.
(rescue) TIPS placement is associat- herent clot is noted. More typically,
2007;356:1631-40. ed with a high mortality rate. These however, bleeding occurs from mul-
[PMID: 17442905]
56. Sreedharan A, Mar- therapies are effective and often tiple sites and may be intermittent,
tin J, Leontiadis GI, render surgical interventions unnec- so the opportunity for intervention
et al. Proton pump
inhibitor treatment essary. Surgical options include por- is limited. Other studies that may
initiated prior to en-
doscopic diagnosis
tosystemic shunting, esophageal localize the bleeding include nuclear
in upper gastroin- transaction, and liver transplantation. imaging and angiography (19).
testinal bleeding.
Cochrane Database Mortality approaches 80% for pa-
Syst Rev. tients with continued bleeding from Bleeding resolves spontaneously in
2010:CD005415.
[PMID: 20614440] esophageal varices. approximately 75% of patients with

© 2013 American College of Physicians ITC2-10 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 diverticular hemorrhage; recurrent How should therapy for acute GI
bleeding occurs in 25–35% of pa- bleeding be monitored?
tients. If bleeding does not resolve Continued monitoring is required to
spontaneously or with angiographic assess whether bleeding has stopped.
intervention, surgical resection is Tachycardia may be an early warning
the remaining therapeutic option. of recurrent bleeding, followed soon
Approximately 20% of patients by hypotension. Hemoglobin levels
hospitalized for diverticular bleed- should be checked on a serial basis at
ing require surgery. Segmental/ least every several hours initially to
subtotal colectomy may be required assess for stability after blood trans-
if bleeding can be localized and fusions; levels that do not increase by
endscopic therapy is not feasible. It approximately 1 g/unit of transfused
may also be necessary if lower GI packed red blood cells may indicate
bleeding remains nonlocalizable, ongoing blood loss. Additional blood
despite the risk for ongoing bleed- transfusions and diagnostic testing
ing (due to failure to resect the ac- (e.g., repeated endoscopy) should be
tual bleeding site) and may be considered if the patient has evidence
associated with significant morbidi- of ongoing blood loss. Similarly,
ty and mortality (67). platelet count and coagulation should
be measured serially to assess the
What is the role of angiography? need for repeated transfusions. Pa- 57. Barkun AN. Should
Angiographic interventions include tients requiring multiple transfusions every patient with
local administration of vasopressin suspected upper GI
of red blood cells should be moni- bleeding receive a
and embolization of the source. Va- tored for hypocalcemia. proton pump in-
hibitor while await-
sopressin is a potent vasoconstrictive ing endoscopy? [Edi-
agent, and local instillation can con- When should a surgeon be torial]. Gastrointest
Endosc.
trol bleeding in up to 80% of pa- consulted for the management of 2008;67:1064-6.
[PMID: 18513549]
tients; however, rebleeding occurs on a patient with acute GI bleeding? 58. Gralnek IM, Barkun
cessation of the infusion in up to Advances in medical and endoscopic AN, Bardou M. Man-
agement of acute
50% of patients. Therefore, vaso- therapies have resulted in fewer pa- bleeding from a
peptic ulcer. N Engl J
pressin may be most useful as a tem- tients requiring surgery for acute GI Med. 2008;359:928-
porizing measure. Although there are bleeding. However, surgical consulta- 37. [PMID: 18753649]
59. Leontiadis GI, Shar-
no absolute contraindications, vaso- tion should be considered early in the ma VK, Howden CW.
pressin should be used with caution evaluation and management of pa- Proton pump in-
hibitor therapy for
in patients with coronary artery dis- tients with severe bleeding. Surgery is peptic ulcer bleed-
ing: Cochrane col-
ease or peripheral vascular disease be- indicated when life-threatening laboration meta-
cause of the risk for vasoconstriction bleeding continues, hemodynamic analysis of
randomized con-
at sites other than the target lesion. compromise continues despite initial trolled trials. Mayo
Clin Proc.
Vasopressin can also cause cardiac ar- aggressive resuscitation, or bleeding 2007;82:286-96.
rhythmia and hyponatremia. cannot be stopped by endoscopic or [PMID: 17352364]
60. Sung JJ, Barkun A,
angiographic means. The choice of Kuipers EJ, Mössner
Embolization is a common inter- surgery depends on bleeding location J, et al. Intravenous
esomeprazole for
vention for control of active hemor- and comorbid conditions. Localiza- prevention of recur-
rhaging. It involves injection of tion of the bleeding site is critical for
rent peptic ulcer
bleeding: a random-
sealant materials, such as gel foam surgical planning. ized trial. Ann Intern
Med. 2009;150:455-
or polyvinyl alcohol, or mechanical 64. [PMID: 19221370]
devices, such as coils and temporary What follow-up outpatient 61. Lau JY, Chung SC,
Leung JW, et al. The
balloons. Embolization is effective in evaluations are required in evolution of stigma-
ta of hemorrhage in
up to 80% of cases. The primary patients who have experienced bleeding peptic ul-
contraindication is poor collateral acute GI bleeding? cers: a sequential
endoscopic study.
blood supply. The major complica- Specific guidelines for management Endoscopy.
1998;30:513-8.
tions are intestinal ischemia and me- of patients after discharge other than [PMID: 9746158]
chanical complications of the arterial those addressing secondary prophy- 62. Bari K, Garcia-Tsao G.
Treatment of portal
cannulation (such as local laxis are lacking. In general, decisions hypertension. World
hematoma, arterial dissection, or regarding timing of postdischarge J Gastroenterol.
2012;18:1166-75.
pseudoaneurysm). outpatient visits and repeated blood [PMID: 22468079]

6 August 2013 Annals of Internal Medicine In the Clinic ITC2-11 © 2013 American College of Physicians

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 work are based on risk-stratification. diverges on this topic, reporting in-
Early outpatient evaluation is appro- adequate outcomes and cost-
priate for patients with high-risk effectiveness to support use of this
endoscopic findings, significant strategy for every patient. Society
comorbid conditions, recent severe guidelines advocate an individualized
bleeding episode, and those in need approach (68). Surveillance en-
of decision-making pertaining to an- doscopy is most appropriate for pa-
tiplatelet or anticoagulant medication tients with suspicious ulcers on initial
use. Conversely, the remaining low- endoscopy in whom initial biopsies
risk patients usually do not require were negative or were not done, per-
early outpatient evaluation. Patients sistent symptoms despite a course of
should be advised to report symp- antisecretory medications, absence of
63. de Franchis R;
toms or signs of recurrent bleeding. a defined cause, or patients from
Baveno V Faculty. Re- With the exception of select clinical demographic regions with high inci-
vising consensus in
portal hypertension: scenarios (see below), empirical re- dence of gastric carcinoma. If H. py-
report of the Baveno
V consensus work-
peated or “surveillance” endoscopy is lori infection has not been assessed
shop on methodolo- not required for most patients partic- via prior gastric mucosal biopsies,
gy of diagnosis and
therapy in portal hy- ularly if the bleeding source is associ- serology, or stool antigen evaluation,
pertension. J Hepa-
tol. 2010;53:762-8.
ated with a low-risk for recurrent gastric mucosal biopsies for H. pylori
[PMID: 20638742] bleeding (e.g., Mallory–Weiss tears, testing should be considered. Litera-
64. Fernández J, Ruiz del
Arbol L, Gómez C, et postpolypectomy bleeding). ture pertaining to detailed manage-
al. Norfloxacin vs ment of patients with H. pylori is
ceftriaxone in the Patients with cirrhosis who survive
prophylaxis of infec- available (69). Bleeding sources that
tions in patients variceal hemorrhage but do not have may require additional endoscopic
with advanced cir-
rhosis and hemor- further therapy are at significant risk therapy include gastric antral vascular
rhage. Gastroen-
terology.
for rebleeding (60%) and death (ap- ectasia (GAVE, or “watermelon
2006;131:1049-56; proximately 33%) within 1–2 years of stomach”) and radiation proctopathy.
quiz 1285.
[PMID: 17030175] the initial bleeding episode. These
65. Laine L, Cook D. En- outcomes are improved using com- How long should antisecretory
doscopic ligation
compared with scle- bined endoscopic variceal band liga- therapy be continued?
rotherapy for treat-
ment of esophageal
tion and beta-blockers (which are Antisecretory therapy for patients
variceal bleeding. A often given once there has been no with small (<1 cm), uncomplicated
meta-analysis. Ann
Intern Med. evidence of hemorrhage for at least ulcers can usually be discontinued
1995;123:280-7.
[PMID: 7611595]
24 hours, barring any contraindica- after 4–6 weeks in the absence of on-
66. Jensen DM, Machi- tions). Patients who have shunt sur- going ulcer-related symptoms. Main-
cado GA, Jutabha R,
Kovacs TO. Urgent gery or TIPS do not require further tenance therapy should be considered
colonoscopy for the
diagnosis and treat-
preventive measures.The American for higher-risk patients (complicated,
ment of severe di- Association for the Study of Liver recurrent, or large ulcers). In patients
verticular hemor-
rhage. N Engl J Med. Disease suggests repeating endoscopy with H. pylori–associated ulcers, ther-
2000;342:78-82. at 7- to 14-day intervals until eradica- apy is often continued until eradica-
[PMID: 10631275]
67. Stollman N, Raskin tion of esophageal varices has been tion of infection is confirmed, or
JB. Diverticular dis-
ease of the colon.
achieved (usually requires 2–4 ses- indefinitely if attempts at H. pylori
Lancet. sions); and once varices are eradicat- eradication have failed. Instructions
2004;363:631-9.
[PMID: 14987890] ed, repeating endoscopy every 3–6 for patients with alternative causes of
68. Banerjee S, Cash BD,
Dominitz JA, et al.
months initially to evaluate for recur- bleeding are made on a case-by-case
The role of en- rence and need for repeated band lig- basis.
doscopy in the man-
agement of patients ation (18). Eventually, the interval can
with peptic ulcer be increased to 6–12 months. What instructions do patients
disease. Gastrointest
Endosc. 2010;71:663- require after acute GI bleeding?
8. [PMID: 20363407] For patients that recover from gastric Specific guidelines for management
69. Malfertheiner P,
Megraud F, O’Morain ulcer bleeding, surveillance en- of patients after discharge are lacking
CA, et al. Manage-
ment of Helicobacter
doscopy 6 to 12 weeks after the last and should be individualized. Patients
pylori infection—- bleeding episode to exclude a non- should be educated regarding the
the Maastricht IV/
Florence Consensus healing ulcer (which raises concern signs and symptoms of recurrent
Report. Gut. for cancer) is common in clinical bleeding and the anticipated benefit
2012;61:646-64.
[PMID: 22491499] practice. However, the literature and duration of targeted therapies.

© 2013 American College of Physicians ITC2-12 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 Patients with bleeding ulcers associat-
ed with H. pylori infection should be
encouraged to complete a full course
of H. pylori therapy and to have test-
ing, such as a urea breath test or
H. pylori stool antigen assay, to con-
firm successful eradication of the bac-
teria. For patients suspected of having
NSAID-related ulcer bleeding, dis-
continuation is preferable but some-
times not feasible. If a patient must
resume NSAIDs, a cyclooxygenase-
2–selective NSAID at the lowest ef-
fective dose plus a daily PPI is
recommended (70). Similarly, patients
with a bleeding ulcer while receiving
low-dose aspirin therapy for second-
ary prevention of established cardio-
vascular disease should resume
therapy as soon as possible after ces-
sation of bleeding (within 1 week,
ideally as early as 3–5 days). Data are
less robust for management of aspirin
therapy after acute bleeding when as-
pirin is being used for primary pre-
vention of cardiovascular events, and
the risk and benefits must be weighed
on an individual basis.

Treatment... Treatment of GI bleeding is highly dependent on the cause and severity.

Initial evaluation and management in all cases should include history and physical ex-
amination, with simultaneous stabilization interventions, such as placement of IV ac-
cess and IV fluid resuscitation. Emergent endoscopy (e.g., within 6 hours) is rarely
indicated, but urgent endoscopy (e.g., within 12 h) in selected circumstances can be
considered, and is mandatory if variceal bleeding is suspected. Administering a PPI is
warranted in patients with suspected peptic ulcer disease but should not delay en-
doscopy. Transfusion practices should target a hemoglobin threshold of 7–8 g/dL. Out-
patient follow-up should be individualized based on the cause of bleeding and the es-
timated risk for rebleeding.

CLINICAL BOTTOM LINE

What do professional organiz-
ations recommend with regard
to the prevention, diagnosis and
treatment of acute GI bleeding?
The International Consensus Upper
Gastrointestinal Bleeding Conference
Group recommends early risk stratifi-
cation and early diagnostic endoscopy
in most patients with upper GI
bleeding.
The American Society of Gastroen-
terology recommends early
colonoscopy for diagnosis of acute
lower GI bleeding, with angiogra-
phy and tagged red blood cell scan-
ning in patients with active bleeding
and nondiagnostic colonoscopies. If
surgical intervention is contemplat-
ed, preoperative localization of
bleeding is desirable.
In patients with obscure acute GI
bleeding in whom EGD and
Practice
colonoscopy have been unrevealing, Improvement
the American Society of Gastroen-
terology recommends early evalua-
tion of the small bowel by VCE or
angiography, with CT angiography,
CT enteroscopy, and deep entero-
scopy as secondary considerations,
depending on availability and ex-
pertise at the institution.
What measures do stakeholders
use to evaluate the quality of care
for patients with acute GI
bleeding?
The Agency for Healthcare Re-
search and Quality has identified
mortality rate as the primary quality 70. Kanwal F, Barkun A,
indicator for care of patients with Gralnek IM, et al.
Measuring quality of
GI bleeding. An expert panel inde- care in patients with
nonvariceal upper
pendently developed a list of 26 gastrointestinal
quality indicators for non-variceal hemorrhage: devel-
opment of an explic-
upper GI bleeding, categorized as it quality indicator
preendoscopic, endoscopic, and set. Am J Gastroen-
terol. 2010;105:1710-
postendoscopic factors (70). 8. [PMID: 20686458]

6 August 2013 Annals of Internal Medicine In the Clinic ITC2-13 © 2013 American College of Physicians

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 In the Clinic PIER Module

In the Clinic
http://pier.acponline.org/physicians/diseases/d184/d184.html

Tool Kit
PIER module on gastrointestinal (GI) bleeding from the
American College of Physicians.
Patient Information
http://pier.acponline.org/physicians/diseases/d184/d184-pi.html
Patient information that appears on the next page for
duplication and distribution to patients.
Acute www.nlm.nih.gov/medlineplus/gastrointestinalbleeding.html
Gastrointestinal www.nlm.nih.gov/medlineplus/tutorials/uppergiendoscopy/
htm/index.htm
Bleeding www.nlm.nih.gov/medlineplus/spanish/tutorials/ upper
giendoscopy/htm/index.htm
Resources related to GI bleeding from the National
Institutes of Health’s MedlinePLUS, including an
interactive tutorial on upper GI endoscopy, in English
and Spanish.
www.gi.org/physician-resources/brochures/
Patient brochure from the American College of
Gastroenterology (ACG) on understanding ulcers,
common pain medications, and GI Bleeding.

Clinical Guidelines
http://annals.org/article.aspx?articleid=745521
International consensus recommendations for managing
patients with nonvariceal upper gastrointestinal bleeding,
published in Annals of Internal Medicine in 2010.
www.sign.ac.uk/guidelines/fulltext/105/index.html
Clinical guideline on the management of acute upper and
lower GI bleeding from the Scottish Intercollegiate
Guidelines Network in 2008.
http://circ.ahajournals.org/content/118/18/1894.full
Consensus document on reducing the GI risks of
antiplatelet therapy and NSAID use from the American
College of Cardiology Foundation (ACCF), ACG, and
American Heart Association (AHA) in 2008.
http://content.onlinejacc.org/article.aspx?articleid=1143980
Consensus document on concomitant use of proton pump
inhibitors and thienopyridines, an update from the
ACCF, ACG, and AHA in 2010.

Diagnostic Tests and Criteria

http://pier.acponline.org/physicians/diseases/d184/tables/
d184-tlab.html
List of laboratory and other studies for acute upper GI
bleeding from PIER.
http://pier.acponline.org/physicians/diseases/d184/tables/
d184-t1.html
Glasgow–Blatchford screening tool to assess the likelihood
that a patient with an acute upper GI bleeding will need
medical intervention.
http://pier.acponline.org/physicians/diseases/d184/tables/
d184-t2.html
Rockall scoring system for identifying patients at risk for
adverse outcome after acute upper GI bleeding.

© 2013 American College of Physicians ITC2-14 In the Clinic Annals of Internal Medicine 6 August 2013

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 THINGS YOU SHOULD In the Clinic
Annals of Internal Medicine
KNOW ABOUT
GASTROINTESTINAL
BLEEDING

What is gastrointestinal (GI) bleeding?

• Bleeding in the digestive tract.
• The upper digestive tract includes the esophagus,
stomach, and upper portion of the small intestine.
• The lower digestive tract includes lower portion of
the small intestine, large intestine (also called the
colon), and anus.
• Most causes of bleeding are curable or controllable,
but some may be life-threatening if left untreated.

What causes bleeding in the digestive

tract?
Causes of bleeding in the upper digestive tract include:
• Peptic ulcers from Helicobacter pylori infections or
long-term use of nonsteroidal anti-inflammatory
drugs, such as aspirin and ibuprofen.
• Varices (enlarged veins) in the lower esophagus that
rupture and bleed.
• Tears in the lining of the esophagus or inflammation
in the lining of the stomach (gastritis) or esophagus
(esophagitis).
• Cancerous or noncancerous (benign) growths.

Causes of bleeding in the lower digestive tract include:

• Diverticular disease (diverticula are irregular pouch- • Chronic bleeding (light bleeding that continues for a

es that develop in the colon wall).
• Colitis (inflammation of the colon) or angiodysplasia
(abnormalities in blood vessels of the intestine).
• Hemorrhoids (ruptured veins in the anus or rectum)
or fissures (anal cuts or tears).
• Cancerous or noncancerous (benign) growths.
What are the signs and symptoms?
• Vomiting bright-red blood or vomit that looks like
coffee grounds indicates bleeding in upper digestive
tract.
• Black or tarry stool or stool that contains dark or
bright red blood indicates bleeding in upper or lower
digestive tract.
Patient Information
long time or starts and stops) may lead to fatigue,
lethargy, and shortness of breath over time.
• Acute bleeding (heavy bleeding) may lead to dizzi-
ness or faintness, shortness of breath, abdominal
pain, and shock.
How is it treated?
• Medical imaging techniques, such as endoscopy or
angiography, may be used to locate the source of the
bleeding inside of the digestive tract and to stop the
bleeding.
• Surgery may be needed if these interventions do not
work.
• Your doctor will try to prevent future bleeding by
treating the condition that is causing the bleeding.

For More Information

http://digestive.niddk.nih.gov/ddiseases/pubs/bleeding/
http://digestive.niddk.nih.gov/Spanish/pubs/bleeding/index.aspx
Information on bleeding in the digestive tract from the National
Institute of Diabetes and Digestive and Kidney Diseases
(NIDDK), in English and Spanish.

http://digestive.niddk.nih.gov/ddiseases/pubs/lowergi/Lower_GI
_Series_T_508.pdf
Information on the lower GI x-rays ordered to help diagnose
problems of the large intestine from the NIDDK.

Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013

 CME Questions
1. A 58-year-old man is evaluated in the
emergency department for painless
bright-red blood per rectum that began
3 hours ago. The bleeding was
accompanied by syncope. He has a
history of rheumatoid arthritis. His
current medications are adalimumab,
methotrexate, and ibuprofen.
On physical examination, temperature is
37.2°C (99.0°F), blood pressure is
88/58 mm Hg, pulse rate is 132/min,
and respiration rate is 24/min.
Abdominal examination is normal. Rectal
examination discloses bright-red blood in
the rectal vault. Nasogastric tube
aspirate shows no evidence of blood or
coffee-ground material.
Laboratory studies reveal a hemoglobin
level of 7.3 g/dL (73 g/L).
Emergency intravenous fluid
resuscitation is begun.
Which of the following is the most
appropriate diagnostic test to perform
next?
A. Colonoscopy
B. Tagged red blood cell scan
C. Upper endoscopy
D. Video capsule endoscopy
2. A 46-year-old man is evaluated for a
3-week history of painless occasional
bright-red rectal bleeding. He has no
fatigue, lightheadedness, weight loss, or
abdominal pain. His stools are frequently
firm, occasionally hard, and there is no
change in the frequency or consistency
of bowel movements. He has never been
screened for colorectal cancer.
On physical examination, temperature is
37.2°C (98.9°F), blood pressure is
132/78 mm Hg, and pulse rate is 84/min.
Digital rectal examination yields a stool
sample that is positive for occult blood;
the examination is otherwise normal.
Anoscopy reveals a few internal
hemorrhoids without active bleeding.
Laboratory studies show a blood
hemoglobin level of 14 g/dL (140 g/L).
Questions are largely from the ACP’s Medical Knowledge Self-Assessment Program (MKSAP, accessed at
http://www.acponline.org/products_services/mksap/15/?pr31). Go to www.annals.org/intheclinic/
to complete the quiz and earn up to 1.5 CME credits, or to purchase the complete MKSAP program.
© 2013 American College of Physicians
Downloaded From: http://annals.org/ by a Glasgow Univ Library User on 08/10/2013
Which of the following is the most
appropriate management of this patient?
A. Banding of hemorrhoids
B. Colonoscopy
C. Fiber supplementation without
further evaluation
D. Home fecal occult blood testing
3. A 60-year-old man hospitalized for
advanced cirrhosis complicated by ascites
and encephalopathy is evaluated for
massive hematemesis and hypotension.
The patient’s medications are
spironolactone, furosemide, and
lactulose.
On physical examination, temperature
is 35.6°C (96°F), blood pressure is
80/50 mm Hg, pulse rate is 146/min, and
respiration rate is 20/min. The patient
has just vomited red blood and has
large-volume ascites; the stool is brown
and positive for occult blood. Laboratory
studies show hemoglobin of 9 g/dL
(90 g/L), platelet count of 60 000/µL
(60 × 109/L), and INR of 3.
In addition to rapid volume resuscitation,
which of the following is the most
appropriate management of this patient?
A. Arteriography
B. Esophagogastroduodenoscopy
C. Intravenous nadolol
D. Mesocaval shunt
E. Transjugular intrahepatic
portosystemic shunt
ITC2-16 In the Clinic
4. A 78-year-old woman is evaluated in the
hospital after being admitted 5 days ago
for a 2-week history of abdominal pain
and nausea. She has also had black, tarry
stools for the past 36 hours. On day 1,
esophagogastroduodenoscopy showed a
clean-based bleeding gastric ulcer that
was positive for Helicobacter pylori
infection; the ulcer was treated with
injection therapy and coagulation
therapy with probe cautery, and proton-
pump inhibitor therapy was initiated. The
bleeding did not stop, and esophago-
gastroduodenoscopy was repeated on day
3 with endoclip therapy. The bleeding
continued, and the patient has received
eight 8 of packed erythrocytes.
On physical examination on day 5,
temperature is 37.2°C (99.0°F), blood
pressure is 95/50 mm Hg, pulse rate is
103/min, and respiratory rate is 16/min.
Rectal examination reveals melanotic
stool. Laboratory studies reveal
hemoglobin of 10.8 g/dL (108 g/L); all
other tests, including coagulation
parameters, are normal.
Which of the following is the most
appropriate next step in the management
of this patient?
A. Bleeding scan
B. Helicobacter pylori eradication
therapy
C. Intravenous octreotide
D. Surgery
Annals of Internal Medicine 6 August 2013