I. Localization by characteristics:

  1. Hematemesis--prox. to lig. of Treitz
  2. Melena--Us. prox to lig. of Treitz, but may be from jej., ileum, or asc. colon, if transit time prolonged
  1. Need 60ml blood in gut for 8h to make 1 black stool
  1. Hematochezia--Us. dist. to lig. of Treitz, but can be prox., if rapid hemorr.

II. Clin. manifestations of blood loss--see under "Hypovolemic Shock"

III. Labs: HCT drop only several hours; mild rise in WBC, platelets, BUN (esp with UGI bleeds) within sev. hrs.

IV. Etiologies of UGI bleeds

  1. Swallowed blood: from oro- or nasopharynx
  2. Peptic Ulcer Disease: most common, us. duodenal, may be initial presentation of ulcer disease
  3. Gastritis
  4. Bleeding venous varices from Portal Hypertension: can be esophageal or gastric; usually sudden, massive hemorrhage
    1. 50% of UGI bleeds in pts with cirrhosis are from non-variceal sources*
  5. Mallory-Weiss: mucosal tear at esoph-gastr. jn, us. h/o retching/nonsanguinous vomiting, then hematemesis
  6. Esophagitis & Esophageal Ca: us. chronic, occult bleeding
  7. Gastric Cancer
  8. Abdominal Aortic Aneurysm fistulizing into gut: when occur us. after repair, bleed out fast
  9. Hematobilia due to hepatic trauma
  10. Uremia: us. chronic, occult bleeding
  11. Primary blood dyscrasias, vasculitis, connective tiss. disorders
  12. Angiodysplasia
  13. Physiologic stress--"Stress Ulcers" in ICU patients
    1. Esophagitis due to bile reflux may be responsible for some "stress-induced" GI bleeding in ICU patients (Gastroent. 116:1293, 1999--JW)
    2. H2-blockers shown to be ass'd with fewer GI ulcers than placebo in ICU patients in a meta-analysis randomized trials (JAMA 275:308, 1996)
    3. A later meta-analysis of 5 randomized trials showed no benefit of H2-blockers in preventing GI bleeding in ICU patients c/w placebo (BMJ 321:1103, 2000--JW)
  14. Lymphoma, Polyps, Diverticula, Leimyoma/-sarcoma: ALL UNCOMMON

V. Etiologies of LGI bleeds

  1. Hemorrhoids, Anal Fissures/Fistulae: us. sm. blood after straining at stool
  2. Proctitis (an IBD variant, or 2? to GC/mycoplasma)
  3. Anal Trauma
  4. Colon Ca/Polyps: us. chronic, slow bleeds
  5. Angiodysplasia: us. descending colon, elderly pts; estrogen-progestin tx has been used but shown to be ineffective in a randomized trial (Gastroent. 121:1073, 2001--JW)
  6. Ulcerative Colitis: us. bloody diarrhea
  7. Inf. Enterocolitides: us. bloody diarrhea; shigella, amoebae, campylobacter, rarely salmonella
  8. Mesenteric Ischemia: us. bloody diarrhea; us. elderly with or young women on OCs
  9. Colonic Diverticuli

VI. Diagnostic approach to Pt with significant GI Bleed

Note--For young person with mild rectal bleeding from probable benign distal source, e.g. Anal Fissure, it's controversial whether & how to workup:
  • In one study of 186 pts with bright red blood per rectum all of whom underwent colonoscopy, 61 were < 55yo; among them, 3 had Ca, 1 of which was > 60cm beyond the anus, i.e. beyond reach of a flexible sigmoidoscope (Am. J. Gastroent. 95:1184, 2000--JW)
  • In one retrospective study of 1766 pts (mean age 57) undergoing colonoscopy for visible rectal bleeding but who had no other GI sx or weight loss, only 2 of the 165 pts < 40yo had cancer found on colonoscopy; 5 had polyps; 7% of pts > 70yo had cancer (Am. J. Gastroent. 97:328, 2002--JW)

Note--For young person with mild rectal bleeding from probable benign distal source, e.g. Anal Fissure, it's controversial whether & how to workup. In one study of 186 pts with bright red blood per rectum all of whom underwent colonoscopy, 61 were < 55yo; among them, 3 had Ca, 1 of which was > 60cm beyond the anus, i.e. beyond reach of a flexible sigmoidoscope (Am. J. Gastroent. 95:1184, 2000--JW)

  1. First steps: VS, type/cross-match, IVF to maintain hemodynamics
  1. Pts with hematemesis have us. bled >1 liter, twice the amount (and twice the mortality) of melena alone*
  1. Hx:
  1. Physiologic "stress" (see under "gastritis" in list above)
  2. Recent EtOH, NSAID: for gastritis
  3. Chronic EtOH: for varices
  4. Retching? : for Mallory-Weiss
  5. Bloody diarrhea? : for UC, inf. enterocolitis, ischemic colitis
  6. Fmly h/o GI dis./Bleeding disorders
  1. Px:
  1. Orthostatic hypotn (>20% blood loss)
  2. Mouth, nose, throat for source of bleeding
  3. Stigmata of liver disease
  4. Lymph nodes, abd. masses (Ca in GI tract)
  5. Skin telangiectasias (Osler-Weber-Rendu Sd., see under "Coagulation Disorders")
  6. Peri-oral pigmentation (Peutz-Jaehger's Sd)
  7. Dermal fibromas (neurofibromatosis)
  8. Sebaceous cysts/Bone tumors (Gardner's Sd.)
  9. Palpable purpura (vasculitides)
  10. Diffuse pigmentation (?) (hemochromatosis)
  1. Labs to order: CBC with RBC morphology and Platelets, PT/PTT
  2. Further w/u:
  1. Perform NG lavage, esp. if suspect UGI bleed:
  1. If lavage fluid comes back clear and active bleeding, can remove
  2. If lavage fluid comes back clear and no active bleeding, leave for sev. hours or do EGD; may have duod. bleed
  3. If lavage fluid comes back bilious and no active bleeding, can remove (indicate pylorus is open, and no thus no duod. bleed)
  4. If get blood orcoffee-ground like material, saline irrigation to get idea of rate of bleeding and to clear blood--ice water lavage is useless
  5. In a retrospective study of 520 pts with UGI bleeding and nasogastric aspirate documented before endoscopy, bloody aspirates were significantly associated with high-risk lesions (OR 4.82) compared with clear/bile aspirates.  Absence of either blood or coffee grounds reduced the probability of a high-risk lesion to 15% (Gastroint. Endosc. 59:172, 2004--JW)
  1. If + blood on NG lavage:
  1. If pt is stable and bleeding is slow or stopped, either
  1. UGI Barium study (would impede subsequent EGD or angio) or
  2. EGD (most sens, esp. for Mallory-Weiss and erosive gastritis)
  1. If bleeding persists, then
  1. EGD emergently (may employ endo. cauterization)
  2. Angio may be nec. for heavy bleeds
  1. If - blood on NG lavage, i.e. LGI (dist. to lig. of Treitz) bleed:
  1. Digital rectal exam
  2. Anoscopy-Sigmoidoscopy-Colonoscopy
  3. Arterial Angio--if brisk bleeding (>0.5 ml/min; with this loss should be orthostatic)
  1. Therapeutic angio: esp. good for M-W or ulcers: vasoconstrictors (e.g. ADH), embolism*
  1. Tc-labelled RBC scan--sensitive (>0.1 ml/min bleed); can use to guide angio
  2. Ba enema--limited role; impedes subsequent colonoscopy or angio
  3. In a study in 100 pts with lower GI bleeding randomized to urgent colonoscopy vs. "standard care" (technitium RBC scanning for ongoing bleeding then visceral angiography if positive), the urgent colonoscopy group had sig. higher likelihood of identifying a definitive bleeding source (42% vs. 22%) but no diff. in early or late rebleeding rates, mortality, length of hospital stay, ICU days, units transfused, or need for surgery. (Am. J. Gastroent. 100:2395, 2005--JW)
  1. When both upper endoscopy and colonoscopy don't show the source, need to evaluate the small intestine Options for this:
    1. Capsule endoscopy
    2. Enteroscopy
    3. In a study randomizing 78 pts with obscure GI bleeding to either of the above 2 approaches, ID of a definite source of bleeding was sig. more common in CE group (26% vs. 12%) (Gastroent. 132:855, 2007--JW)


VII. Treatment

  1. Tx for bleeding ulcers:
  1. Proton-Pump Inhibitors
    1. Omeprazole 40 PO BID x 5d better than placebo in pts not treated with endoscopic interventions (cautery, etc.) in a study from India (NEJM 336:1091, 1997-JW)
    2. In a meta-analysis of 21 randomized trial comparing PPIs with placebo or H2 blockers in 2,915 pts with endoscopically verified bleeding ulcer, PPI tx was not associated with sig. decrease in mortality but was associated with sig. reduction in incidence of rebleeding or surgery (number needed to treat 10 and 25, respectively) (BMJ 330:568, 2005--AFP)
  2. Endoscopic treatment modalities for bleeding peptic ulcer:
    1. Injection therapy with epinephrine or saline
    2. Heater probe thermocoagulation (often done after epinephrine injection)
    3. Electrocautery (often done after epinephrine injection)
    4. Metallic clips ("Hemoclips")
    5. Comparisons among these modalities
      1. In a study in 47 pts with UGI bleeding and "high risk stigmata" (active bleeding or nonbleeding visible vessel), all of whom underwent EGD and received epinephrine injection of any adherent clots randomized to electrocautery vs. Hemoclips, there were no sig. diffs. in incidence of rebleeding, length of hospital stay, transfusion requirements, complications, surgery, or mortality (Am. J. Gastroent. 100:1503, 2005--JW)
    6. In a study of 156 pts with UGI bleeding and inactive bleeding ulcers (with visible vessels or clots) randomized to sham endoscopy vs. active endoscopic tx (with epinephrine injection, thermocoagulation, and clot removal), all of whom received omeprazole, 30d incidence of recurrent bleeding was sig. lower in active-tx group (1% vs. 12%) (Ann. Int. Med. 139:237, 2003--JW)
  3. Arterial embolization
    1. In a retrospective study of 60 pts with refractory bleeding peptic ulcer who underwent arterial embolization, 57 had initial success, and 41 had initial success and no rebleeding within 1mo. (Clin. Gastroent. Hepatol. 7:515, 2009-JW)
  4. Follow-up treatment after patient is stabilized
    1. Proton-Pump Inhibitors
      1. 240 pts s/p endoscopic treatment of bleeding peptic ulcers randomized to omeprazole 80mg IV then 8mg/h x 72 vs. placebo; all pts also received omeprazole 20mg QD x 8wks.  IV omeprazole group had sig. lower 30d incidence of recurrent bleeding (7% vs. 20%) (NEJM 343:310, 2000--AFP)
  1. See section on Portal Hypertension for management of bleeding esophageal varices
  1. Treatment for nonvariceal UGI bleeding:
    1. Tx with somatostatin or octreotide (a somatostatin analogue) for 48-72h reduced risk of continued or repeat bleeding (RR 0.53) and of requiring surgery (RR 0.71) when c/w placebo or an H2-blocker in one meta-analysis of 14 trials involving 1,800 pts with endoscopically confirmed nonvariceal UGI bleed (Ann. Int. med. 127:1062, 1997--JW)
    2. In a meta-analysis of endoscopic & pharmacologic treatments for pts with nonvariceal UGI bleeding and actively bleeding ulcers, nonbleeding visible vessels, or adherent clots, combination of epinephrin + (coagulation, sclerosants, thrombin, or clips) was superior to epinephrine alone in reducing risk for recurrent bleeding (RR 0.34) and proton pump inhibitor therapy also reduced the risk of rebleeding (RR 0.4) (Clin. Gastroent. Hepatol. 7:33, 2009-JW)
  1. It may be safe to manage selected elderly pts with UGIB as oupatients (not orthostatic and obvious but low-risk finding on upper endoscopy, e.g. small ulcer) (Am. J. Gastroent. 94:1242, 1999--JW)
  1. Tx of diverticular bleeding
    1. Often tx'd with partial colectomy
    2. Successful endoscopic tx (e.g. with epinephrine injections and/or bipolar coagulation) has been reports (NEJM 342:78, 2000--JW)
  1. Treatment for Mallory-Weiss tears
    1. In a study in 34 pts with actively bleeding Mallory-Weiss tears randomized to endoscopic band ligation vs. endoscopic epiniephrine injection, there was no sig. diff. in primary hemostasis or recurrent bleeding (Gastrointest. Endoscp. 60:22, 2004--abst)