GI bleeding in ICU patients

Published

Cook DJ, et al. "Risk factors for gastrointestinal bleeding in critically ill patients". The New England Journal of Medicine. 1994. 330(6):337-381.
PubMed • Full text • PDF

Clinical Question

In critically ill patients, what are the risk factors for developing gastrointestinal hemorrhage?

Bottom Line

The greatest risk factors for GI bleeding in critically ill patients are coagulopathy and mechanical ventilation.

Major Points

A 1969 case series by Skillman et al. first described the association of critical illness with development of stress ulcers and their subsequent bleeding. A 1991 meta-analysis demonstrated that prophylaxis with antacids, sulcralfate, and H₂ blockers reduced the rate of this complication. However, given how time consuming and expensive these therapies were, identifying which patients would most benefit from their use would inform medical practice. (Additionally, early evidence hinting to an increased risk for nosocomial pneumonias with ulcer prophylaxis was emerging, though a 2009 cohort study suggests that this may only be associated with PPI use.)

This 1994 prospective trial by Cook et al. from the Canadian Critical Care Trials Group followed consecutive patients admitted to an ICU and asked that the clinicians withheld GI prophylaxis. Clinically-significant GI bleeding occurred in 1.5% of patients. Those at the highest risk for GI bleeding had coagulopathy and respiratory failure on a ventilator for >48 hours. The authors advise that prophylaxis be limited to patients with these risk factors. Of note, patients with head trauma, burns of >30% BSA, organ transplant recipients, or known PUD or gastritis were not included in the study.

The use of pharmacologic prophylaxis against stress ulcers is now widespread in both ICU and non-ICU patients, despite lack of indication for their use in the latter. PPIs have become popular agents for stress ulcer prophylaxis though no large, high quality studies exist demonstrating their effectiveness. Furthermore, patients started on PPIs in an ICU setting are often discharged on these medications without a compelling indication. PPIs have been associated with multiple adverse effect including GI infections (Clostridium difficile, Campylobacter, Salmonella, Shigella, and Listeria), pneumonia, fracture, medication interactions, and malabsorption syndromes. More recently, a 2010 meta-analysis demonstrated no benefit from stress ulcer prophylaxis when added to enteric feedings. This finding has not yet been confirmed by a clinical trial.

Guidelines

American Society of Health-Systems Pharmacists Stress Ulcer Prophylaxis (1999)
The following applies to adult patients only.

• Recommend prophylaxis for ICU patients if:
  • Coagulopathy (defined as Plt <50,000/mm³, INR >1.5 or PTT >2x the ULN; strength of evidence C)
  • Require mechanical ventilation for >48 hours (strength C)
  • History of GI ulceration or bleeding in the prior year (strength D)
  • Two of the following (strength D):
    • Sepsis
    • ICU stay >1 week
    • Occult bleeding ≥6 days
    • High-dose corticosteroids (>250 mg/day hydrocortisone or 62.5 mg/day prednisone)
  • Glasgow Coma Score ≤10 or inability to obey simple commands (strength B)
  • Thermal injury to >35% BSA (strength B)
• Prophylaxis may be indicated in ICU patients if:
  • Multiple trauma with Injury Severity score ≥16 (strength D)
  • Transplantation patients in the ICU perioperatively (strength D)
  • Liver failure (strength D)
  • ICU patients with spinal cord injuries (strength D)
• No prophylaxis recommended for non-ICU patients if <2 risk factors for clinically important bleeding (strength B)
• No prophylaxis recommended for non-ICU patients with ≥2 risk factors for clinically important bleeding (strength D)

IHI Ventilator Bundle (2011)

• Peptic ulcer disease prophylaxis
• Elevation of the head of the bed
• Daily sedation vacations and assessment for extubation readiness
• DVT prophylaxis
• Daily oral care with chlorhexidine

Design

• Multicenter, prospective, observational study
• N=2,252
• Setting: Four academic ICUs
• Enrollment: 1990-1991
• Mean follow up: Not identified
• Primary outcomes:
  • Overt bleeding
  • Clinically-important bleeding

Population

Inclusion Criteria

• Admission to the ICU at an involved research site
• Older than 16 years of age

Exclusion Criteria

• Upper GI bleeding in the 48 hours before or 24 hours after admission (defined as hematemesis, NG suction with blood or coffee grounds material, hematochezia, or melena)
• Total gastrectomy
• Facial trauma
• Epistaxis
• Brain death
• Hopeless prognosis
• Death or discharge within 24 hours of admission
• At three sites, enrollment was closed during the weekend

Baseline Characteristics

• Demographics: Age 60 years, male 66.4%
• Primary diagnosis:
  • CVD: 6.3%
  • CV surgery: 48.5%
  • Respiratory disease: 12.1%
  • GI disease: 9.8%
  • GU disease: 4.0%
  • CNS disease: 4.0%
  • Head injury: 1.2%
  • Multiple trauma: 0.8%
  • Sepsis: 1.6%
  • Organ transplantation: 4.8%
  • Other: 6.9%
• APACHE score: 21
• Length of stay: 5 days
• Mortality: 9.7%

Interventions

• Attending physicians encouraged to withhold GI prophylaxis in all patients except:
  • Head injury
  • Burns over >30% BSA
  • Organ transplant recipients
  • Diagnosis of gastritis in the previous 6 weeks
  • Upper GI bleeding three to six weeks before admission
• Patients were followed for bleeding
• Prophylaxis options included H₂ antagonists, antacids, sucralfate, prostaglandin analogues, and omeprazole

Outcomes

Primary Outcome

Overt bleeding

4.4% of all patients (95% CI 3.6-5.6)

Of these, % on GI prophylaxis: 87.0%

Clinically-important bleeding

1.5% of all patients (95% CI 1.0-2.1)

Of these, % on GI prophylaxis: 69.7%

Subgroup Analysis

Comparisons are clinically-important bleeding vs. all others.

Mortality rate

48.5% vs. 9.1% (P<0.001)

Significant risk for bleed by multiple regression analysis

Respiratory failure (OR 15.6; P<0.001)

Coagulopathy (OR 4.3; P<0.001)

Defined as Plt <50,000/mm³, INR >1.5, PTT >2x ULN

Non-significant risk for bleeding by multiple regression analysis

Hypotension (OR 3.7; P=0.08)

Sepsis (OR 2.0; P=0.17)

Liver failure (OR 1.6; P=0.27)

Renal failure (OR 1.6; P=0.26)

Enteral feeding (OR 1.0; P=0.99)

Glucocorticoid administration (OR 1.5; P=0.26)

Organ transplantation (OR 1.5; P=0.42)

Anticoagulant therapy (OR 1.1; P=0.88)

Additional Analyses

Characteristics

Age: 62 vs. 60 years

Male sex: 66.7% vs. 66.4%

APACHE score: 23 vs. 21

Type of prophylaxis used

H₂ antagonist 71.8%

Sucralfate 7.0%

Antacids 4.9%

Prostaglandin 0.6%

Omeprazole 0.3%

Criticisms

• Did not clearly define what comprised a critically ill patient and included many cardiovascular surgical patients -- a group at low risk for complications
• Low rate of sepsis, cardiovascular, or respiratory disease as reason for ICU admissions
• Coagulopathy defined by elevation in fibrin-split products may better define the condition than alterations in PT/aPTT as warfarin and heparin do not increase risk for GI bleeding

Funding

Supported by the Ontario Ministry of Health