Practical Approach to Acute Gastrointestinal Bleeding

Christopher S. Huang MD Assistant Professor of Medicine Boston University School of Medicine Section of Gastroenterology Boston Medical Center

Learning Objectives

• • UGIB – – Nonvariceal (PUD) and variceal Resuscitation, risk assessment, pre-endoscopy management – Role of endoscopy – Post-endoscopy management LGIB – Risk assessment – Role and timing of colonoscopy – Non-endoscopic diagnostic and treatment options

Definitions

• • • Upper GI bleed – arising from the esophagus, stomach, or proximal duodenum Mid-intestinal bleed – arising from distal duodenum to ileocecal valve Lower intestinal bleed – arising from colon/rectum

Stool color and origin/pace of bleeding

• • •

Guaiac positive stool

– Occult blood in stool – Does not provide any localizing information – Indicates slow pace, usually low volume bleeding

Melena

– Very dark, tarry, pungent stool – Usually suggestive of UGI origin (but can be small intestinal, proximal colon origin if slow pace)

Hematochezia

– Spectrum: bright red blood, dark red, maroon – Usually suggestive of colonic origin (but can be UGI origin if brisk pace/large volume)

Case Vignette – CC:

• 68 yo male presents with a chief complaint of a large amount of “bleeding from the rectum”

Case Vignette - HPI

• • • Describes bleeding as large volume, very dark maroon colored stool Has occurred 4 times over past 3 hours He felt light headed and nearly passed out upon trying to get up to go the bathroom

Case Vignette - HPI

• • • • • Denies abdominal pain, nausea, vomiting, antecedent retching No history of heartburn, dysphagia, weight loss No history of diarrhea or constipation/hard stools No prior history of GIB Screening colonoscopy 10 years ago – no polyps, (+) diverticulosis

Case Vignette – PMHx, Meds

• • • • • • • Hepatitis C CAD – h/o MI PVD AAA – s/p elective repair 3 years ago HTN Hypercholesterolemia Lumbago • Medications: – – Aspirin Clopidogrel – Atorvastatin – Atenolol – Lisinopril

Case Vignette – Physical Exam

• Physical examination: – BP 105/70, Pulse 100, (+) orthostatic changes – Alert and mentating, but anxious appearing – Anicteric – Mid line scar, benign abdomen, nontender liver edge palpable in epigastrium, no splenomegaly – Rectal examination – no masses, dark maroon blood

Case Vignette - Labs

• Labs – Hct 21% (Baseline 33%) – Plt 110K – BUN 34, Cr 1.0

– Alb 3.5

– INR 1.6

– ALT 51, AST 76

Initial Considerations

• •

Differential diagnosis?

– What is most likely source? – What diagnosis can you least afford to miss?

How sick is this patient? (risk stratification)

– Determines disposition – Guides resuscitation – Guides decision re: need for/timing of endoscopy

Differential Diagnosis – Upper GIB

• • • • • • • • •

Peptic ulcer disease Gastroesophageal varices

Erosive esophagitis/gastritis/duodenitis Mallory Weiss tear Vascular ectasia Neoplasm Dieulafoy’s lesion Aortoenteric fistula Rare, but cannot afford to miss Hemobilia, hemosuccus pancreaticus Most common

Differential Diagnosis – Lower GIB

• • • • • • •

Diverticulosis

Angioectasias Hemorrhoids Dieulafoy’s lesion Most common diagnosis Colitis (IBD, Infectious, Ischemic) Neoplasm Post-polypectomy bleed (up to 2 weeks after procedure)

History and Physical

• • • • • • • • •

History

Localizing symptoms History of prior GIB NSAID/aspirin use Liver disease/cirrhosis Vascular disease Aortic valvular disease, chronic renal failure AAA repair Radiation exposure Family history of GIB • • • • •

Physical Examination

Vital signs, orthostatics Abdominal tenderness Skin, oral examination Stigmata of liver disease Rectal examination – Objective description of stool/blood – – Assess for mass, hemorrhoids No need for guaiac test

History and Physical

• • • • • • • • •

History

Localizing symptoms History of prior GIB NSAID/aspirin use Liver disease/cirrhosis Vascular disease Aortic valvular disease, chronic renal failure AAA repair Radiation exposure Family history of GIB • • • • •

Physical Examination

Vital signs, orthostatics Abdominal tenderness Skin, oral examination Stigmata of liver disease Rectal examination – Objective description of stool/blood – – Assess for mass, hemorrhoids No need for guaiac test

Take Home Point # 1

Always get objective description of stool

Avoid noninformative terms such as “grossly guaiac positive”

Take Home Point #2

If you need a card to tell you whether there’s blood in the stool, it’s NOT an

acute GIB

Narrowing the DDx: Upper or Lower Source?

• Predictors of UGI source: – Age <50 – Melenic stool – BUN/Creatinine ratio • If ratio ≥ 30, think upper GIB J Clin Gastroenterol 1990;12:500 Am J Gastroenterol 1997;92:1796 Am J Emerg Med 2006;24:280

Utility of NG Tube

• • • Most useful situation: patients with severe hematochezia, and unsure if UGIB vs. LGIB – Positive aspirate (blood/coffee grounds) indicates UGIB

Can provide prognostic info:

– Red blood per NGT – predictive of high risk endoscopic lesion – Coffee grounds – less severe/inactive bleeding Negative aspirate – not as helpful; 15-20% of patients with UGIB have negative NG aspirate Ann Emerg Med 2004;43:525 Arch Intern Med 1990;150:1381 Gastrointest Endosc 2004;59:172

Take Home Point #3

Upper GI bleed must still be considered in patients with severe hematochezia, even if NG aspirate

negative

Initial Assessment

• • Always remember to assess A,B,C’s Assess degree of hypovolemic shock

Class I

Blood loss (mL) 750 Blood volume loss (%) Heart rate SBP < 15% <100 No change Urine output (mL/hr) Mental status >30 Alert

Class II

750-1500 15-30% >100 Orthostatic change 20-30 Anxious

Class III

1500-2000 30-40% >120 Reduced 10-20

Class IV

>2000 >40% >140 Very low, supine <10 Aggressive/dro wsy Confused/unco nscious

Resuscitation

• • • • IV access: large bore peripheral IVs best (alt: cordis catheter) Use crystalloids first Anticipate need for blood transfusion • Threshold should be based on underlying condition, hemodynamic status, markers of tissue hypoxia • Should be administered if Hgb ≤ 7 g/dL • • 1 U PRBC should raise Hgb by 1 (HCT by 3%) Remember that initial Hct can be misleading (Hct remains the same with loss of whole blood, until re-equilibration occurs) Correct coagulopathy

Resuscitation

• • • • IV access: large bore peripheral IVs best (alt: cordis catheter) Time Anticipate need for blood transfusion • 40% IVFs Threshold should be based on underlying condition, hemodynamic status, markers of tissue hypoxia • Should be administered if Hgb ≤ 7 g/dL • • 1 U PRBC should raise Hgb by 1 (HCT by 3%) Remember that initial Hct can be misleading (Hct remains the same with loss of whole blood, until re-equilibration occurs) Correct coagulopathy

Transfusion Strategy

• Randomized trial: – 921 subjects with severe acute UGIB – Restrictive (tx when Hgb<7; target 7-9) vs. Liberal (tx when Hgb<9; target 9-11) – Primary outcome: all cause mortality rate within 45 days NEJM 2013;368;11-21

Restrictive Strategy Superior

Mortality rate Rate of further bleeding Overall complication rate

Restrictive

5% 10% 40%

Liberal

9% 16% 48%

P value

0.02

0.01

0.02

Benefit seen primarily in Child A/B cirrhotics NEJM 2013;368;11-21

Resuscitation

• • • • IV access: large bore peripheral IVs best (alt: cordis catheter) Use crystalloids first

Weigh risks and benefits of reversing anticoagulation

Anticipate need for blood transfusion • •

Assess degree of coagulopathy

Threshold should be based on underlying condition, hemodynamic status, markers of tissue hypoxia Should be administered if Hgb ≤ 7 g/dL • • Remember that initial Hct can be misleading (Hct remains the same with loss of whole blood, until re-equilibration occurs) Correct coagulopathy

Resuscitation

•

Early intensive resuscitation reduces mortality

– Consecutive series of patients with hemodynamically significant UGIB – First 36 subjects = Observation Group (no intervention) – Second 36 subjects = Intensive Resuscitation Group (intense guidance provided) – goal was to

decrease time to correction of hemodynamics, Hct and coagulopathy

Am J Gastroenterol 2004;99:619

Early Intensive Resuscitation Reduces UGIB Mortality

(groups are essentially the same) Intervention: Faster correction of hemodynamics, Hct and coags.

Time to endoscopy similar Am J Gastroenterol 2004;99:619

Early Intensive Resuscitation Reduces UGIB Mortality

• • Observation group – 5 MI – 4 deaths Intense group – – 2 MI 1 death (sepsis) Am J Gastroenterol 2004;99:619

• • • •

Causes of Mortality in Patients with Peptic Ulcer Bleeding

Patients rarely bleed to death

Prospective cohort study >10,000 cases of peptic ulcer bleed Mortality rate 6.2%

80% of deaths not related to bleeding

Am J Gastroenterol 2010;105:84

•

Causes of Mortality in Patients with Peptic Ulcer Bleeding

Most common causes of non-bleeding mortality: – Terminal malignancy (34%) – Multiorgan failure (24%) – Pulmonary disease (24%) – Cardiac disease (14%) Am J Gastroenterol 2010;105:84

Take Home Point #4

Early resuscitation and supportive measures are critical to reduce mortality from UGIB

Risk Stratification

• • • Identify patients at high risk for adverse outcomes Helps determine disposition (ICU vs. floor vs. outpatient) May help guide appropriate timing of endoscopy

Rockall Scoring System

• • Validated predictor of mortality in patients with UGIB 2 components: clinical + endoscopic

Variable 0 1 2 3 Age

<60

Shock Comorbidity

No SBP ≥ 100 P<100 No major 60-79 Tachy SBP ≥ 100 P>100 ≥ 80 Hypotension SBP <100 Cardiac failure, CAD, other major Renal failure, liver failure, malignancy Gut 1996;38:316

Clinical Rockall Score – Mortality Rates

60% 50% 40% 30% 20% 10% 0% 0 1 2 3 4 5 6 7

AIMS65

• Simple risk score that predicts in-hospital mortality, LOS, cost in patients with acute UGIB lbumin <3.0

NR > 1.5

ental status altered ystolic BP <90 + years old Gastrointest Endosc 2011;74:1215

AIMS65

Gastrointest Endosc 2011;74:1215

Blatchford Score

• • • Predicts need for endoscopic therapy Based on readily available clinical and lab data Can use UpToDate calculator Lancet 2000;356:1318

Blatchford Score

Gastrointest Endosc 2010;71:1134

Blatchford Score

• • Most useful for safely discriminating low risk UGIB patients who will likely NOT require endoscopic hemostasis “Fast track Blatchford” – patient at low risk if: BUN < 18 mg/dL Hgb > 13 (men), 12 (women) SBP >100 HR < 100

Pre-endoscopic Pharmacotherapy

• For Non-Variceal UGIB – IV PPI: 80 mg bolus, 8 mg/hr drip – Rationale: suppress acid, facilitate clot formation and stabilization – Duration: at least until EGD, then based on findings

Pre-endoscopy PPI

• • • Reduces the proportion of patients with high risk endoscopic stigmata (“downstages” lesion) Decreases need for endoscopic therapy Has not been shown to reduce rebleeding, surgery, or mortality rates High risk Low risk

Endoscopic treatment required:

Omeprazole – 19% (23% of PUD) Placebo – 28% (37% of PUD) N Engl J Med 2007;356:1631

Endoscopy - Nonvariceal UGIB

• • Early endoscopy (within 24 hours) is recommended for most patients with acute UGIB Achieves prompt diagnosis, provides risk stratification and hemostasis therapy in high risk patients J Clin Gastroenterol 1996;22:267 Gastrointest Endosc 1999;49:145 Ann Intern Med 2010;152:101

• •

When is Endoscopic Therapy Required?

~80% bleeds spontaneously resolve Endoscopic stigmata of recent hemorrhage major

Stigmata Active bleeding Nonbleeding visible vessel

Adherent clot

Continued/rebleeding rate 55-90% 40-50%

Variable, depending on underlying lesion: 0-35% Flat pigmented spot Clean base 7-10% < 5%

Major Stigmata – Active Spurting

Kelsey, PB (Dec 04 2003). Duodenum - Ulcer, Arterial Spurting, Treated with Injection and Clip. The DAVE Project. Retrieved Aug, 1, 2010, from

http://daveproject.org/viewfilms.cfm?film_id=39

Major Stigmata - NBVV

Adherent Clot

• • • Role of endoscopic therapy of ulcers with adherent clot is controversial Clot removal usually attempted Underlying lesion can then be assessed, treated if necessary

Minor Stigmata

Flat pigmented spot Clean base

Low rebleeding risk – no endoscopic therapy needed

Endoscopic Hemostasis Therapy

• • • Epinephrine injection Thermal electrocoagulation Mechanical (hemoclips) • Combination therapy superior to monotherapy Kelsey, PB (Nov 08 2005). Stomach - Gastric Ulcer, Visible Vessel. The DAVE Project.

Retrieved Aug, 1, 2010, from http://daveproject.org/viewfilms.cfm?film_id=306

Baron, TH (May 01 2007). Duodenum - Bleeding Ulcer Treated with Thermal Therapy, Perforation Closed with Hemoclips. The DAVE Project. Retrieved Aug, 1,

2010, from http://daveproject.org/viewfilms.cfm?film_id=620

Nonvariceal UGIB – Post-endoscopy management

• • Patients with low risk ulcers can be fed promptly, put on oral PPI therapy.

Patients with ulcers requiring endoscopic therapy should receive PPI gtt x 72 hours – Significantly reduces 30 day rebleeding rate vs placebo (6.7% vs. 22.5%) – Note: there may not be major advantage with high dose over non-high dose PPI therapy N Engl J Med 2000;343:310 Arch Intern Med 2010;170:751

Nonvariceal UGIB – Post-endoscopy management

• • • Determine H. pylori status in all ulcer patients Discharge patients on PPI (once to twice daily), duration dictated by underlying etiology and need for NSAIDs/aspirin In patients with cardiovascular disease on low dose aspirin: restart as soon as bleeding has

resolved

– RCT demonstrates increased risk of rebleeding (10% v 5%) but decreased 30 day mortality (1.3% v 13%) Ann Intern Med 2010;152:1

Nonvariceal UGIB – Post-endoscopy management

• • Determine H. pylori status in all ulcer patients Discharge patients on PPI (once to twice daily), duration dictated by underlying etiology and • dose aspirin: restart as soon as bleeding has

resolved

– RCT demonstrates increased risk of rebleeding (10% v 5%) but decreased 30 day mortality (1.3% v 13%) Ann Intern Med 2010;152:1

Variceal Bleeding

• • • • Occurs in 1/3 of patients with cirrhosis 1/3 initial bleeding episodes are fatal Among survivors, 1/3 will rebleed within 6 weeks Only 1/3 will survive 1 year or more

Predictors of large esophageal varices

• • • • Severity of liver disease (Child Pugh) Platelet count < 88K Palpable spleen Platelet count/spleen diameter (mm) ratio <909 Gut 2003;52:1200 J Clin Gastroenterol 2010;44:146 J Gastroenterol Hepatol 2007;22:1909 Arch Intern Med 2001;161:2564 Am J Gastroenterol 1999;94:3103

leed

asoconstrictor therapy ntibiotics esuscitation U level care ndoscopy ternative/Rescue therapies eta blockade

asoconstrictor therapy

• • • • • Goal: Reduce splanchnic blood flow Terlipressin – only agent shown to improve control of bleeding and survival in RCTs and meta-analysis – Not available in US Vasopressin + nitroglycerine – too many adverse effects Somatostatin – not available in US Octreotide (somatostatin analogue) • Decreases splanchnic blood flow (variably) • Efficacy is controversial; no proven mortality benefit • Standard dose: 50 mcg bolus, then 50 mcg/hr drip for 3-5 days Gastroenterology 2001;120:946 Cochrane Database Syst Rev 2008;16:CD000193 N Engl J Med 1995;333:555 Am J Gastroenterol 2009;104:617

ntibiotics

• • • Bacterial infection occurs in up to 66% of patients with cirrhosis and variceal bleed Negative impact on hemostasis (endogenous heparinoids)

Prophylactic antibiotics reduces incidence of bacterial infection, significantly reduces early rebleeding

– Ceftriaxone 1 g IV QD x 5-7 days – Alt: Norfloxacin 400 mg po BID Hepatology 2004;39:746 J Korean Med Sci 2006;21:883 Hepatogastroenterology 2004;51:541

esuscitation

• • • Promptly but with caution Goal = maintain hemodynamic stability, Hgb ~7-8, CVP 4-8 mmHg Avoid excessively rapid overexpansion of volume; may increase portal pressure, greater bleeding

ndoscopy

• • • Should be performed as soon as possible after resuscitation (within 12 hours) Endotracheal intubation frequently needed Band ligation is preferred method Layer, L. & Jaganmohan, S. & Raju, GS & DuPont, AW (Oct 28 2009). Esophagus Band Ligation of Actively Bleeding Gastroesophageal Varices. The DAVE Project. Retrieved Aug, 2, 2010, from http://daveproject.org/viewfilms.cfm?film_id=715

ternative/Rescue therapies

• • • TIPS – Transjugular Intrahepatic Portosystemic Shunt Early placement of shunt (within 24-72hrs) associated with improved survival among high-risk patients Preferred treatment for gastric variceal bleeding (rule out splenic vein thrombosis first) Fan, C. (Apr 25 2006). Vascular Interventions in the Abdomen: New Devices and Applications. The DAVE Project. Retrieved Aug, 2, 2010, from http://daveproject.org/viewfilms.cfm?film_id=497 Hepatology 2004;40:793 Hepatology 2008;48:Suppl:373A N Engl J Med. 2010 Jun 24;362:2370

TIPS+embolization of gastric varices

ternative/Rescue therapies

Sengstaken-Blakemore Tube

• • • • Very effective for immediate, temporary control High complication rate – aspiration, migration, necrosis + perforation of esophagus Use as bridge to TIPS within 24 hours Airway protection strongly recommended

ternative/Rescue therapies

Self-Expanding Metal Stent

• • • • Specially designed covered metal stent Tamponades distal esophageal varices Removable; does not require airway protection Very limited data Gastrointest Endosc 2010;71:71

eta blockade

• •

Reduces risk for recurrent variceal hemorrhage

Use nonselective beta blocker (e.g. Nadolol – splanchnic vasoconstriction, decrease cardiac output) and titrate up to maximum tolerated dose, HR 50-60 – Start as inpatient, once acute bleeding has resolved and patient shows hemodynamic stability

Lower GI Bleed

• • Bleeding arising from the colorectum In patients with severe hematochezia, first consider possibility of UGIB – 10-15% of patients with presumed LGIB are found to have upper GIB

Lower GI Bleed

• Differential Diagnosis - Diverticulosis (# 1 cause) Large volume, painless - Angioectasias - Hemorrhoids - Colitis (IBD, Infectious, Ischemic) Smaller volume, pain, diarrhea - Neoplasm - Post-polypectomy - Dieulafoy’s lesion

LGIB – Risk Stratification

• Predictors of severe * LGIB: 0 factors: ~6% risk        HR>100 SBP<115 Syncope 1-3 factors: ~40% bleeding during first 4 hours of evaluation aspirin use >2 active comorbid conditions * Defined as continued bleeding within first 24 hours (transfusion of 2+ Units, decline in HCT of 20+%) and/or recurrent bleeding after 24 hours of stability Arch Intern Med 2003;163:838 Am J Gastroenterol 2005;100:1821

LGIB – Risk Factors for Mortality

• • •

Age Intestinal ischemia Comorbid illnesses

• • • • • Secondary bleeding (developed during admission for a separate problem) Coagulopathy Hypovolemia Transfusion requirement Male gender Clinical Gastro Hepatol 2008;6:1004

Role of Colonoscopy

• • Like UGIB, ~80% of LGIBs will resolve spontaneously; of these, ~30% will rebleed Lack of standardized approach – Traditional approach: • elective colonoscopy after resolution of bleeding, bowel prep – low therapeutic benefit • Angiography for massive bleeding, hemodynamically unstable patient – Urgent colonoscopy approach • Similar to UGIB – identify stigmata of hemorrhage, perform therapy

Urgent Colonoscopy

• • • • Within 6-12 hours of presentation Requires rapid “purge” prep with 5-6 L Golytely administered 1L every 30-45 minutes Colonoscopy performed within 1 hour after clearance of stool, blood and clots Need for bowel prep and risks of procedural sedation may be prohibitive in unstable patient

Endoscopic Therapy

Srinivasan, R. & Luthra, G. & Raju, GS (Jul 17 2007). Colon - Endoscopic Hemostasis of Diverticular Bleed.

The DAVE Project. Retrieved Aug, 3, 2010, from http://daveproject.org/viewfilms.cfm?film_id=63

Urgent Colonoscopy

• • • Limited high quality evidence of benefit Establishes diagnosis earlier, shorter length of stay “Landmark” study supporting urgent colonoscopy for diverticular bleed published in 2000 – 2 consecutive prospective, non-randomized studies – Group 1 (n=73): urgent colonoscopy, surgical therapy – Group 2 (n=48): urgent colonoscopy, endoscopic therapy N Engl J Med 2000;342:78

Urgent Colonoscopy

• • Group 1: 17 pts with definite diverticular bleed – 9 had recurrent/persistent bleeding – 6 required emergency surgery Group 2: 10 pts with definite diverticular bleed – All 10 patients treated endoscopically – 0 had recurrent bleed, complications, further transfusions, or surgery N Engl J Med 2000;342:78

Urgent Colonoscopy

Standard Management Algorithm

• • Two RCTs published to date Compared urgent colonoscopy (within 8 hours) vs. standard management Am J Gastroenterol 2005;100:2395

Urgent Colonoscopy – RCT#1

Definite bleeding source identified more frequently (42% vs 22%) But no significant difference in important outcomes (but underpowered) Am J Gastroenterol 2005;100:2395

Urgent Colonoscopy – RCT#2

• • • • 85 patients with serious hematochezia (hemodynamically significant, Hgb drop > 1.5 g/dL, blood transfusion) EGD performed within 6 hours If EGD negative, randomized to urgent (<12 hr) or elective (36-60 hr) colonoscopy Primary endpoint= further bleeding Am J Gastroenterol 2010;105:2636

Urgent Colonoscopy – RCT#2

• EGD positive in 15% • No evidence of improved clinical outcomes with urgent colonoscopy – but prespecified

sample size not reached

Am J Gastroenterol 2010;105:2636

Urgent Colonoscopy

• • • In published series, endoscopic therapy is applied in 10-40% of patients undergoing colonoscopy for LGIB Taken together, evidence suggests that colonoscopy should be performed within 12 24 hours in stable patients However, it is unclear how faster timing affects major clinical outcomes

Radiographic Studies

• • • •

Tagged RBC scan

Noninvasive, highly sensitive (0.05-0.1 ml/min) Ability to localize bleeding source correctly only ~66% More accurate when positive within 2 hours (95 100%) Lacks therapeutic capability Coordinate with IR so that positive scan is followed closely by angiography

Radiographic Studies

• •

Angiography

Detects bleeding rates of 0.5-1 ml/min • microcoils, polyvinyl prepped for colonoscopy (or have had colonoscopy with failure to hematomas, thromboses, dissection

Radiographic Studies

• • • •

Multi-Detector CT (CT angio)

Readily available, can be performed in ER within 10 minutes Can detect bleeding rate of 0.5 ml/min Can localize site of bleeding (must be active) and provide info on etiology Useful in the actively bleeding but hemodynamically stable patient Gastrointest Endosc 2010;72:402

Role of Surgery

• • • Reserved for patients with life-threatening bleed who have failed other options General indications: hypotension/shock despite resuscitation, >6 U PRBCs transfused Preoperative localization of bleeding source important

Algorithmic Evaluation of Patient with Hematochezia

Hematochezia Treat lesion Positive positive EGD active NG lavage Assess activity of bleed inactive Prep for Colonoscopy Negative Risk for UGIB negative No risk for UGIB Hemodynamically stable?

Algorithmic Evaluation of Patient with Hematochezia

Active Lower GIB Hemodynamically stable?

No Angiography (+/- Tagged RBC scan) Or Surgery if life threatening Yes Consider “urgent colonoscopy” vs. traditional approach

Take Home Points

• Always get objective description of stool color (best way – examine it yourself) • Don’t order guaiac tests on inpatients • Severe hematochezia can be from UGIB, even if NG lavage is negative

Take Home Points

• All bleeding eventually stops (and majority of

nonvariceal bleeds will stop spontaneously, with the patient alive)

• Early resuscitation and supportive care are key to reducing morbidity and mortality from GIB