NSAIDs: misoprostol prevents ulcers better than proton-pump inhibitors or H2 antagonists but causes diarrhoea

Clinical bottom line (level 1a)

  1. Patients on long-term NSAIDs who take misoprostol compared with placebo are less likely to develop gastric ulcers (NNT = 11 at 3-24 months) or duodenal ulcers (NNT = 31 at 3-12 months) , but are more likely to develop diarrhoea (NNH = 10 at months) and stop medication (NNH = 15 at 3-24 months) .
  2. Patients on high-dose misoprostol compared with low-dose misoprostol were less likely to develop endoscopic ulcers, but more likely to develop diarrhoea.
  3. Patients on standard dose H 2 antagonists compared with placebo are less likely to develop duodenal ulcers (NNT = 30 at 3-12 months) , but not clearly less likely to develop gastric ulcers or more likely to stop medication.
  4. Patients on double-dose H 2 antagonists compared with placebo are less likely to develop gastric ulcers (NNT = 7 at 3-12 months) or duodenal ulcers (NNT = 10 at 3-12 months) , but not clearly more likely to stop medication.
  5. Patients on proton-pump inhibitors compared with placebo are less likely to develop endoscopic gastric ulcers (NNT = 8 at 3-12 months) or endoscopic duodenal ulcers (NNT = 12 at 3-12 months) , but not clearly more likely to stop medication.
  6. Patients on misoprostol compared with ranitidine 150 mg are less likely to develop endoscopic ulcers (NNT = 29 at 1-2 months) , but not clearly more likely to stop medication.
  7. Patients on misoprostol compared with proton-pump inhibitors are less likely to develop endoscopic ulcers but not clearly more likely to stop medication (NNT = 8 at 3 months) .
  8. Patients on proton-pump inhibitors compared with H 2 antagonists are less likely to develop endoscopic ulcers (NNT = 7 at 6 months) .
Rostom et al: Cochrane Library 2001; 4 : -
Expires April 2004

The study

Systematic review of all randomised controlled trials of
  • Patients: had taken NSAIDs for at least 3 weeks
  • Intervention: prostoglandin analogs (misoprostol), H 2 receptor antagonists, or proton-pump inhibitors compared with
  • Outcome: NSAID-induced upper gastrointestinal toxicity: ulcers, ulcer complications (haemorrhage, perforation, pyloric obstruction or death), symptoms

    Articles found in all languages using Medline, Embase, Current Contents, Cochrane Controlled Trials Register, 1966 to January 2000 (search terms: ) and searching recent conference proceedings, references lits of potentially relevant articles and reviewand content experts and pharmaceutical companies were contacted

    Selection criteria: by 2 independent reviewers
    Appraisal criteria: by 2 independent reviewers using Jadad criteria
    Articles excluded if:
    • NSAIDs prescribed for 3 weeks or less
    • endoscopic ulcers 3 mm or less in diameter
    • healthy volunteers
    • H2 receptor antagonist doses less than 300 mg twice daily


    35 RCTs found
    • 19 involving misoprostol
    • 7 involving standard dose H-2 drugs
    • 3 involving double-dose H-2 drugs
    • 4 involving proton-pump inhibitors

    Studies were not found to be significantly heterogeneous.

    The evidence

    Outcome Time to outcome CER OR
    (95% CI)    		 NNT
    (95% CI)
    misoprostol v. placebo: gastric ulcers 3-24 months 221/1709
    (12.9%)    		 0.25
    (0.19 to 0.32)     		11
    (10 to 12)
    misoprostol v. placebo: duodenal ulcers 3-24 months 85/1394
    (6.1%)    		 0.45
    (0.32 to 0.65)     		31
    (25 to 49)
    misoprostol v. placebo: stopped medication 3-24 months 1936/6564
    (29.5%)    		 1.36
    (1.26 to 1.46)     		-15
    (-20 to -12)
    misoprostol v. placebo: diarrhoea months 114/1539
    (7.4%)    		 2.73
    (2.21 to 3.38)     		-10
    (-13 to -7)
    Low-dose H 2 antagonists v. placebo: gastric ulcers 3-12 months 52/495
    (10.5%)    		 0.71
    (0.46 to 1.10)     		36
    (NNT = 110 to infinity;
    NNH = 19 to infinity)
    Low-dose H 2 antagonists v. placebo: duodenal ulcers 3-12 months 27/487
    (5.5%)    		 0.38
    (0.19 to 0.73)     		30
    (23 to 70)
    Low-dose H 2 antagonists v. placebo: stopped medication 3-12 months 145/592
    (24.5%)    		 0.77
    (0.58 to 1.01)     		22
    (NNT = 540 to infinity;
    NNH = 12 to infinity)
    high-dose H 2 antagonists v. placebo: gastric ulcers 3-12 months 38/147
    (25.9%)    		 0.37
    (0.20 to 0.67)     		7
    (5 to 14)
    high-dose H 2 receptor antagonists v. placebo: duodenal ulcers 3-12 months 20/147
    (13.6%)    		 0.26
    (0.12 to 0.60)     		10
    (9 to 20)
    high-dose H 2 antagonists: stopped medication 3-12 months 27/140
    (19.3%)    		 0.84
    (0.45 to 1.55)     		39
    (NNT = 13 to infinity;
    NNH = 10 to infinity)
    PPI v. placebo: endoscopic gastric ulcers 3-12 months 65/311
    (20.9%)    		 0.31
    (0.20 to 0.48)     		8
    (6 to 10)
    PPI v. placebo: endoscopic duodenal ulcers 3-12 months 34/311
    (10.9%)    		 0.19
    (0.10 to 0.36)     		12
    (10 to 15)
    PPI v. placebo: stopped medication 3-12 months 16/155
    (10.3%)    		 1.18
    (0.64 to 2.19)     		-61
    (NNT = 10 to infinity;
    NNH = 29 to infinity)
    misoprostol v. ranitidine 150 mg: stopped medication 1-2 months 80/300
    (26.7%)    		 1.30
    (0.91 to 1.85)     		-18
    (NNT = 7 to infinity;
    NNH = 55 to infinity)
    misoprostol v. ranitidine 150 mg: endoscopic ulcers 1-2 months 13/269
    (4.8%)    		 0.28
    (0.10 to 0.75)     		29
    (23 to 86)
    misoprostol v. PPI: stopped medication 3 months 50/296
    (16.9%)    		 0.67
    (0.42 to 1.07)     		20
    (NNT = 100 to infinity;
    NNH = 11 to infinity)
    misoprostol v. PPI: endoscopic ulcers 3 months 154/296
    (52.0%)    		 0.59
    (0.43 to 0.83)     		8
    (5 to 21)
    PPI v. H 2 antagonist: endoscopic ulcers 6 months 44/215
    (20.5%)    		 0.28
    (0.16 to 0.48)     		7
    (6 to 11)

    • Patients on misoprostol compared with placebo were more likely to develop diarrhoea, vomiting, abdominal pain or flatulence that led to stopping medication.
    • A dose-effect was noted for misoprostol. 800 microgram was more effective than m400 microgram at preventing endoscopic ulcers (p = 0.0055), but increased the risk of diarrhoea (p < 0.001).

    Comments

    1. Only misoprostol has been shown to reduce the clinically-symptomatic ulcers compared with placebo. It is unclear whether misoprostol is better than the other types of drug in reducing symptomatic ulcers or GI bleeding.

    Citation

    1. Rostom A, Wells G, Tugwell P, et al: prevention of NSAID-induced gastroduodenal ulcers. Cochrane Library 2001; 4 : -
    Search Terms:
    Contributor: Chris Ball, April 2002
    Reviewer:

    Clinical Question.
    Patient on NSAIDs long-term
    Intervention or Exposure misoprostol, H2 antagonists, proton pump inhibitors
    Outcome peptic ulcer
    		•