Occult GI Bleeding in NSAID Users—The Base of the Iceberg!

Abstract

The first account of the adverse effects of aspirin on the stomach appeared in The Lancet in 1938.1Douthwaite A.H. Lindtott G.A.M. Gastroscopic observation of the effect of aspirin and certain other substances on the stomach.Lancet. 1938; 2: 1222-1225Abstract Google Scholar Shortly afterward, Hurst and Lindtott2Hurst A. Lindtott G.A.M. Aspirin as a cause of hematemesis a clinical and gastroscopic study.Guys Hosp Reports. 1939; 89: 173-176Google Scholar noted aspirin’s then-unexplained tendency to cause gastrointestinal bleeding. Another 20 years passed before Duggan in Australia3Duggan J.M. Aspirin in chronic gastric ulcer an Australian experience.Gut. 1976; 17: 378-384Crossref PubMed Scopus (38) Google Scholar noted retrospectively a close connection between aspirin consumption and the occurrence of an epidemic in young women in the 1940s of gastric but not duodenal ulcers. These observations soon were confirmed in many other countries. In these studies, affected patients usually were consuming relatively large amounts of aspirin. On later analysis by Graham and Lacey-Smith,4Graham D.Y. Lacey-Smith J. Aspirin and the stomach.Ann Intern Med. 1986; 104: 390-398Crossref PubMed Scopus (203) Google Scholar the association between clinical gastric ulcers—as diagnosed by radiology or surgery—was not seen in those consuming fewer than 14 tablets a week, and was not statistically significant at fewer than 22 tablets per week (3 tablets/day).The first careful evaluation of epidemiologic evidence that consumption of aspirin was associated with serious gastrointestinal (GI) bleeding was by Langman5Langman M.J.S. Epidemiological evidence for the association of aspirin and acute gastrointestinal bleeding.Gut. 1970; 11: 627-634Crossref PubMed Scopus (42) Google Scholar in 1970. He believed that although the association was real, the strength of the evidence was short of that needed to conclude that aspirin caused bleeding. He noted, however, that subgroups of patients with bleeding who were radiographically negative contained a higher proportion of aspirin takers than did subgroups of patients with visible ulcers. Furthermore, in 4 of 5 studies of aspirin-using patients with bleeding ulcers, duodenal ulcers (whose occurrence was not linked to aspirin use) were more commonly the source of bleeding than were gastric ulcers. The data suggested that aspirin-associated bleeding occurred whether or not an ulcer was present, and that ulcers not caused by aspirin could bleed during aspirin therapy.Others in the United Kingdom6Parry D.J. Wood P.H.N. Relationship between aspirin taking and gastro-duodenal hemorrhage.Gut. 1967; 8: 301-307Crossref PubMed Scopus (44) Google Scholar also showed (by using chromium-labeled red blood cells) that occult bleeding was “an inescapable concomitant of taking the drug” in all normal patients, and suggested that occult (very common) and serious (rare) hemorrhage might result from different pathogenetic mechanisms.6Parry D.J. Wood P.H.N. Relationship between aspirin taking and gastro-duodenal hemorrhage.Gut. 1967; 8: 301-307Crossref PubMed Scopus (44) Google Scholar Throughout the 1960s, Davenport7Davenport H.W. Gastric mucosal hemorrhage in dogs effects of acid, aspirin and alcohol.Gastroenterology. 1969; 56: 439-449Abstract Full Text PDF PubMed Scopus (168) Google Scholar and others described the effects of aspirin on the stomach, including aspirin-induced bleeding from the gastric mucosa. Davenport’s7Davenport H.W. Gastric mucosal hemorrhage in dogs effects of acid, aspirin and alcohol.Gastroenterology. 1969; 56: 439-449Abstract Full Text PDF PubMed Scopus (168) Google Scholar findings supported the concept that gastric mucosal injury by aspirin was a major source of blood lost into the GI tract, whether or not a frank gastric ulcer was present. Only in 1968 was the inhibitory effect of aspirin on platelet function first described8O’Brien J.R. Effects of salicylates on human platelets.Lancet. 1968; 1: 779-783Abstract PubMed Google Scholar; this effect later was found to be shared with all classic nonselective nonsteroidal anti-inflammatory drugs (NSAIDs).During the 1980s, there was rapid acceptance of the concept that acute or chronic use of either aspirin or nonaspirin NSAIDs was accompanied by the unpredictable development of gastrointestinal hemorrhage,9Somerville K. Faulkner G. Langman M.J.S. Non-steroidal anti-inflammatory drugs and bleeding peptic ulcer.Lancet. 1986; 1: 462-464Abstract PubMed Scopus (456) Google Scholar less commonly perforation,10Coggon D. Lambert P. Langman M.J.S. 20 years of hospital admission for peptic ulcer in England and Wales.Lancet. 1981; 1: 1302-1304Abstract PubMed Scopus (132) Google Scholar, 11Collier D.S.J. Pain J.A. Non-steroidal anti-inflammatory drugs and peptic ulcer perforation.Gut. 1985; 26: 359-363Crossref PubMed Scopus (232) Google Scholar or, rarely, death.12Catford J.C. Simpson R.J. Confidential enquiry into deaths from peptic ulcer.Health Trends. 1986; 18: 37-40Google Scholar, 13Griffin M.R. Ray W.A. Schaffner W. Non-steroidal anti-inflammatory drugs and deaths from peptic ulcer in elderly adults.Ann Intern Med. 1988; 109: 359-363Crossref PubMed Scopus (393) Google Scholar The population risk of sustaining one of these adverse events was estimated by the US Food and Drug Administration to be of the order of 2%–6% per annum.14O’Neill R.T. Testimony to the arthritis advisory committeeFDC reports. FDA, Rockville, MD1987: 8-9Google Scholar Most clinicians assumed that the observed adverse events were complications of gastric or duodenal ulcers that had been caused by the drugs and then bled. Other factors possibly contributing to GI hemorrhage largely were ignored. These included the universal effect of aspirin/NSAIDs, serious platelet dysfunction, and the possibilities that the drugs could initiate GI bleeding from either pre-existing GI conditions or from nonpeptic lesions caused by the drugs in more distal sites.However, such assumptions were challenged by results of the VIGOR trial,15Bombardier C. Laine L. Reicin A. et al.Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis.N Engl J Med. 2000; 343: 1520-1528Crossref PubMed Scopus (3699) Google Scholar which showed that of 31 GI bleeds in rofecoxib-treated patients, only 12 were attributable to gastroduodenal ulcers, and of 82 bleeds in naproxen-treated patients, only 32 similarly were attributable to gastroduodenal ulcers. A later publication from this trial confirmed that much of the overt bleeding originated distal to the stomach and duodenum,16Laine L. Connors L.G. Reicin A. et al.Serious lower gastrointestinal clinical events with non-selective NSAID or coxib use.Gastroenterology. 2003; 124: 288-292Abstract Full Text PDF PubMed Scopus (326) Google Scholar although the study could not assess the relative contributions to bleeding from underlying GI disease, comorbid illness, or co-therapy with other medications.The Base of the Iceberg of Nonsteroidal Anti-inflammatory Drug–Induced Gastrointestinal BleedingIn this issue of the journal, Bowen et al17Bowen B. Yuan Y. James C. et al.Time course and pattern of blood loss with ibuprofen treatment in healthy subjects.Clin Gastroenterol Hepatol. 2005; 3: 1075-1082Abstract Full Text Full Text PDF PubMed Scopus (19) Google Scholar report that the use of moderate doses of ibuprofen (800 mg 3 times/day) for 4 weeks also leads to an increase in mean daily fecal blood loss in all patients, similar to the blood loss reported with aspirin. The increase started on the third day of therapy. The amount of bleeding also increased episodically to a variable degree: episodes occurring 1 to 7 times in 26 of 31 patients, over 4 weeks of observation, the duration of each episode lasting approximately 3 days. The estimated total blood loss over the 28 days averaged 71.35 mL, but 5 patients lost between 100 and 300 mL.Although the development of anemia was not studied formally, it is reasonable to assume that blood loss that continues at this rate in long-term NSAID users could lead to significant anemia in at least some patients. In a community setting, the effects of age, iron intake, Helicobacter pylori infection, co-use of acid-suppressing drugs, low- or regular-dose aspirin, a second NSAID, steroids, clopidogrel, ticlopidine or another antiplatelet drug, or an anticoagulant all could be expected to influence the rate of blood loss and its impact on the patient; thus far, these factors have not been studied. The study reported here was relatively small, of short duration, and performed in healthy volunteers aged 18–60 years. A number of additional questions need to be addressed. What is the mechanism responsible for the bleeding? Where in the gut is the source of bleeding? Is the magnitude of the effect the same for all NSAIDs, and is it dose dependent?The report by Bowen et al17Bowen B. Yuan Y. James C. et al.Time course and pattern of blood loss with ibuprofen treatment in healthy subjects.Clin Gastroenterol Hepatol. 2005; 3: 1075-1082Abstract Full Text Full Text PDF PubMed Scopus (19) Google Scholar discussed ibuprofen-associated hemorrhage and appears to assume that the occult microbleeding and severe but rare episodic bleeding are both in some way caused by the same process. This assumption may or may not be correct, but it seems likely that additional factors must be operative in the small minority of ibuprofen users who experience serious bleeds. Such overt GI bleeding is but the tip of the iceberg, and the contributions of pre-existing disease and other factors to its genesis cannot be ignored. The universal occurrence of the microbleeding shared by ibuprofen and aspirin suggests that it results from a common pharmacodynamic effect of these drugs. The inhibition of platelet cyclooxygenase-1 (Cox-1) activity and platelet function is the most likely common action. This inhibition is irreversible in the case of aspirin, even with doses as low as 10 mg/day,18Lee M. Cryer B. Feldman M. Dose effects of aspirin on gastric prostaglandins and stomach mucosal injury.Ann Intern Med. 1994; 120: 184-189Crossref PubMed Scopus (80) Google Scholar but is dose dependent and reversible in the case of ibuprofen. When ibuprofen is given at a dose of 600 mg every 8 hours, platelet dysfunction normalizes within 24 hours of discontinuing therapy19Goldenberg N.A. Jacobson L. Manco-Johnson M.J. Duration of platelet dysfunction after a 7-day course of ibuprofen.Ann Intern Med. 2005; 142: 506-509Crossref PubMed Scopus (69) Google Scholar; normalization of function probably occurs much more rapidly at lower doses, especially if the drug is used only intermittently. It is unknown how rapidly the risk for GI bleeding returns to baseline after ceasing ibuprofen ingestion.The observation that sparing or preservation of Cox-1–dependent platelet function is important in preventing gastrointestinal bleeding is supported by many observations. First, related compounds that do not inhibit platelet function (eg, meloxicam, nabumetone, etodolac, or salsalate) cause much less overt GI bleeding. Second, after aspirin use, the risk for GI bleeding returns to baseline after 5 days, a time period equivalent to the half-life of the platelet.20Levy M. Miller D.R. Kaufmann D.W. et al.Major upper gastrointestinal bleeding relation to the use of aspirin and other non-narcotic analgesics.Arch Intern Med. 1988; 148: 281-285Crossref PubMed Scopus (98) Google Scholar Third, clopidogrel is an antiplatelet compound that remains active for the lifespan of the platelet and is believed to be nonulcerogenic: nevertheless, clopidogrel use is accompanied by an increased incidence of GI bleeding, both from ulcers (in those with previously healed ulcers) and also from lesions in the small or large intestine, with a frequency comparable to that seen in those using aspirin.21Chan F.K.L. Ching J.Y.L. Hung L.C.T. et al.Clopidogrel versus aspirin and omeprazole to prevent recurrent ulcer bleeding.N Engl J Med. 2005; 352: 238-244Crossref PubMed Scopus (516) Google Scholar The contributions of both hemostasis and mucosal injury to GI bleeding have been examined in detail by Hawkey et al,22Hawkey C.J. Hawthorne A.B. Hudson N. et al.Separation of hemostasis by aspirin from mucosal injury to the human stomach.Clin Sci (Colch). 1991; 81: 565-573Google Scholar, 23Hawkey C.J. Aspirin and gastrointestinal bleeding.Aliment Pharmacol Ther. 1994; 8: 141-146Crossref PubMed Scopus (49) Google Scholar but uncertainty remains as to whether bleeding can occur from thrombocytopenia or platelet dysfunction in the absence of a mucosal break.Cox-1 also is constitutive in most of the intestine and inhibition of its action alone, or combined with inhibition of Cox-2, might indeed cause mucosal lesions. This hypothesis, that Cox inhibition might cause mucosal injury, is supported by several recent studies of small intestinal mucosal abnormalities that occurred in response to NSAID therapy,24Graham D.Y. Opekun A.R. Field F.W. et al.Visible small intestinal mucosal injury in chronic NSAID users.Clin Gastroenterol Hepatol. 2005; 3: 55-59Abstract Full Text Full Text PDF PubMed Scopus (484) Google Scholar, 25Maiden L. Thjodleifsson B. Theodore A. A quantitative analysis of NSAID-induced small bowel pathology by capsule endoscopy.Gastroenterology. 2005; 128: 1172-1178Abstract Full Text Full Text PDF PubMed Scopus (424) Google Scholar, 26Goldstein J.L. Eisen G.M. Lewis B. et al.Video capsule endoscopy to prospectively assess small bowel injury with celecoxib, naproxen plus omeprazole and placebo.Clin Gastroenterol Hepatol. 2005; 3: 1-9Abstract Full Text Full Text PDF PubMed Scopus (574) Google Scholar as assessed by video capsule endoscopy (VCE). Although there remains considerable uncertainty as to how such observed lesions should be classified and interpreted, some preliminary statements can be made. First, among asymptomatic patients volunteering to enter drug trials, and thought not to be on NSAIDs, 10%–23% have existing prior lesions on VCE, necessitating their exclusion from studies; this is regarded as a relatively high population prevalence of a lesion of unclear clinical significance. Second, among those who enter a drug trial after a normal baseline VCE, 55%–70% develop visible lesions when given NSAIDs; however, most of the lesions are of uncertain significance and not bleeding at the time of observation.24Graham D.Y. Opekun A.R. Field F.W. et al.Visible small intestinal mucosal injury in chronic NSAID users.Clin Gastroenterol Hepatol. 2005; 3: 55-59Abstract Full Text Full Text PDF PubMed Scopus (484) Google Scholar, 25Maiden L. Thjodleifsson B. Theodore A. A quantitative analysis of NSAID-induced small bowel pathology by capsule endoscopy.Gastroenterology. 2005; 128: 1172-1178Abstract Full Text Full Text PDF PubMed Scopus (424) Google Scholar, 26Goldstein J.L. Eisen G.M. Lewis B. et al.Video capsule endoscopy to prospectively assess small bowel injury with celecoxib, naproxen plus omeprazole and placebo.Clin Gastroenterol Hepatol. 2005; 3: 1-9Abstract Full Text Full Text PDF PubMed Scopus (574) Google Scholar The prevalence of these lesions is not reduced (and even may be increased) by co-administration of a proton-pump inhibitor.25Maiden L. Thjodleifsson B. Theodore A. A quantitative analysis of NSAID-induced small bowel pathology by capsule endoscopy.Gastroenterology. 2005; 128: 1172-1178Abstract Full Text Full Text PDF PubMed Scopus (424) Google Scholar, 26Goldstein J.L. Eisen G.M. Lewis B. et al.Video capsule endoscopy to prospectively assess small bowel injury with celecoxib, naproxen plus omeprazole and placebo.Clin Gastroenterol Hepatol. 2005; 3: 1-9Abstract Full Text Full Text PDF PubMed Scopus (574) Google Scholar In 1 large, well-executed, placebo-controlled study, the incidence of lesions developing during coxib therapy was decreased markedly when compared with proton-pump inhibitor and NSAID therapy; however, the risk was not abolished. Numerous observations reviewed elsewhere24Graham D.Y. Opekun A.R. Field F.W. et al.Visible small intestinal mucosal injury in chronic NSAID users.Clin Gastroenterol Hepatol. 2005; 3: 55-59Abstract Full Text Full Text PDF PubMed Scopus (484) Google Scholar implicate NSAIDs in causing increased small intestinal permeability, inflammation, ulceration, stricture development, protein loss, blood loss, and iron deficiency or other anemia. However, it still is unclear how these relate to observed mucosal changes on VCE, or to other markers of enteric inflammation (eg, fecal calprotectin excretion or abnormal mucosal uptake of 111indium-labeled white blood cells), as detected by radionuclide scanning. Whether mucosal lesions occur episodically, or simply bleed episodically, awaits clarification.The contribution of large-bowel lesions to microbleeding currently is unknown. Until 1992, the medical literature on the contribution of aspirin or NSAID use to colonic bleeding had consisted largely of case reports and reviews that had attracted little attention. In that year, however, 3 publications strongly suggested that microbleeding might arise in large-bowel lesions. In a seminal study, using biochemical methods to detect the use of NSAIDs or aspirin in the 5 days before admission, Lanas et al27Lanas A. Sekar C. Hirschowitz B.I. Objective evidence of aspirin use in both ulcer and non-ulcer, upper and lower gastrointestinal bleeding.Gastroenterology. 1992; 103: 862-869Abstract PubMed Google Scholar found that in patients with lower GI bleeding, current NSAID use was detected in 80% of bleeders vs 24.3% of controls (odds ratio, 13.7). They postulated that regardless of the lesion that bled, the drugs caused the bleeding. Riddell et al28Riddell R.H. Tanaka M. Mazzoleni G. Non-steroidal anti-inflammatory drugs as possible causes of collagenous colitis a case-control study.Gut. 1992; 33: 683-686Crossref PubMed Scopus (291) Google Scholar reported that collagenous colitis was almost 5 times more common among NSAID users than among affected nonusers. Gibson et al29Gibson G.R. Whitacre E.B. Ricotti C.A. Colitis induced by non-steroidal anti-inflammatory drugs.Arch Intern Med. 1992; 152: 625-632Crossref PubMed Scopus (174) Google Scholar drew attention to the clinical picture of NSAID colitis, the ease with which the diagnosis was missed, and the need for heightened awareness among clinicians. Laine et al,16Laine L. Connors L.G. Reicin A. et al.Serious lower gastrointestinal clinical events with non-selective NSAID or coxib use.Gastroenterology. 2003; 124: 288-292Abstract Full Text PDF PubMed Scopus (326) Google Scholar in a later review, discussed the published reports of bleeding from many different sites in the colon and concluded that it seemed likely that “inhibition of platelet aggregation could lead to bleeding from GI lesions that are already present or to bleeding from lesions caused by the NSAIDs.” The colonic conditions at risk for overt bleeding on NSAID or aspirin therapy are diverticulosis, colonic ulcers (in collagenous colitis), inflammatory or ischemic colitis, vascular ectasias, hemorrhoids, or carcinoma.16Laine L. Connors L.G. Reicin A. et al.Serious lower gastrointestinal clinical events with non-selective NSAID or coxib use.Gastroenterology. 2003; 124: 288-292Abstract Full Text PDF PubMed Scopus (326) Google Scholar However, the contribution of these various conditions to occult colonic blood loss has not been studied.The observation that lesions and bleeding in the lower bowel might have different causes is reminiscent of the results of a meta-analysis of ulcer disease30Huang J.-Q. Sridhar S. Hunt R.H.H. pylori infection and non-steroidal anti-inflammatory drugs in peptic ulcer disease a meta-analysis.Lancet. 2002; 359: 14-22Abstract Full Text Full Text PDF PubMed Scopus (718) Google Scholar that concluded that H pylori infection is the main cause of peptic ulcers, but that aspirin/NSAIDs are the main causes of ulcer bleeding. This conclusion also could be explained by platelet dysfunction. The frequency of NSAID-induced microscopic injury to the large bowel, of the type seen on VCE in the small bowel, still is unknown but now could be examined with magnification chromoendoscopy.We have improved considerably the technique used for isotopic estimation of fecal blood loss, but there still is a paucity of information on the comparative effects of aspirin and various NSAIDs when studied head-to-head using this method of estimating fecal blood loss. Dose-response studies have been largely neglected except in the case of aspirin, for which the effect of dose appears small.31Derry S. Loke Y.K. Risk of Gastrointestinal hemorrhage with long term use of aspirin a meta-analysis.Br Med J. 2000; 321: 1183-1187Crossref PubMed Scopus (642) Google Scholar In older studies, aspirin appears to cause more GI microbleeding than ibuprofen, fenoprofen, naproxen, or indomethacin, but blood loss with rofecoxib, etoricoxib, nabumetone, and etodolac appeared similar to that with placebo.Thus, bleeding caused by platelet dysfunction may be the base of the iceberg of all NSAID-induced bleeding from the gut. In contrast, rarer overt bleeds from small or large bowel form the tip of the iceberg, occurring with much lower incidence. These serious events may manifest mainly in patients with pre-existing diseases, or on high doses of NSAIDs. Alternatively, patients may have major risk factors such as old age; a history of ulcers, bleeding, or perforation occurring anywhere in the GI tract; consumption of anti-ulcer drugs before NSAID therapy; or concomitant use of anti-coagulants, steroids, or platelet inhibitors. In clinical practice, there is likely to be considerable variation in the sites of origin of bleeding among individuals with pre-existing GI diseases. The problem of microbleeding seems likely to increase as the use of coxibs decreases, and the use of over-the-counter and generic nonselective NSAIDs increases. It also appears that nonulcer bleeding will not be prevented by co-therapy with proton-pump inhibitors. The effects of nitrosated aspirin or NSAIDs, or of dual cox-lipox (lipoxygenase) inhibitors such as liclofenone, on the prevention of GI bleeding have not been studied. The size and shape of this iceberg are likely to change on further study. The first account of the adverse effects of aspirin on the stomach appeared in The Lancet in 1938.1Douthwaite A.H. Lindtott G.A.M. Gastroscopic observation of the effect of aspirin and certain other substances on the stomach.Lancet. 1938; 2: 1222-1225Abstract Google Scholar Shortly afterward, Hurst and Lindtott2Hurst A. Lindtott G.A.M. Aspirin as a cause of hematemesis a clinical and gastroscopic study.Guys Hosp Reports. 1939; 89: 173-176Google Scholar noted aspirin’s then-unexplained tendency to cause gastrointestinal bleeding. Another 20 years passed before Duggan in Australia3Duggan J.M. Aspirin in chronic gastric ulcer an Australian experience.Gut. 1976; 17: 378-384Crossref PubMed Scopus (38) Google Scholar noted retrospectively a close connection between aspirin consumption and the occurrence of an epidemic in young women in the 1940s of gastric but not duodenal ulcers. These observations soon were confirmed in many other countries. In these studies, affected patients usually were consuming relatively large amounts of aspirin. On later analysis by Graham and Lacey-Smith,4Graham D.Y. Lacey-Smith J. Aspirin and the stomach.Ann Intern Med. 1986; 104: 390-398Crossref PubMed Scopus (203) Google Scholar the association between clinical gastric ulcers—as diagnosed by radiology or surgery—was not seen in those consuming fewer than 14 tablets a week, and was not statistically significant at fewer than 22 tablets per week (3 tablets/day). The first careful evaluation of epidemiologic evidence that consumption of aspirin was associated with serious gastrointestinal (GI) bleeding was by Langman5Langman M.J.S. Epidemiological evidence for the association of aspirin and acute gastrointestinal bleeding.Gut. 1970; 11: 627-634Crossref PubMed Scopus (42) Google Scholar in 1970. He believed that although the association was real, the strength of the evidence was short of that needed to conclude that aspirin caused bleeding. He noted, however, that subgroups of patients with bleeding who were radiographically negative contained a higher proportion of aspirin takers than did subgroups of patients with visible ulcers. Furthermore, in 4 of 5 studies of aspirin-using patients with bleeding ulcers, duodenal ulcers (whose occurrence was not linked to aspirin use) were more commonly the source of bleeding than were gastric ulcers. The data suggested that aspirin-associated bleeding occurred whether or not an ulcer was present, and that ulcers not caused by aspirin could bleed during aspirin therapy. Others in the United Kingdom6Parry D.J. Wood P.H.N. Relationship between aspirin taking and gastro-duodenal hemorrhage.Gut. 1967; 8: 301-307Crossref PubMed Scopus (44) Google Scholar also showed (by using chromium-labeled red blood cells) that occult bleeding was “an inescapable concomitant of taking the drug” in all normal patients, and suggested that occult (very common) and serious (rare) hemorrhage might result from different pathogenetic mechanisms.6Parry D.J. Wood P.H.N. Relationship between aspirin taking and gastro-duodenal hemorrhage.Gut. 1967; 8: 301-307Crossref PubMed Scopus (44) Google Scholar Throughout the 1960s, Davenport7Davenport H.W. Gastric mucosal hemorrhage in dogs effects of acid, aspirin and alcohol.Gastroenterology. 1969; 56: 439-449Abstract Full Text PDF PubMed Scopus (168) Google Scholar and others described the effects of aspirin on the stomach, including aspirin-induced bleeding from the gastric mucosa. Davenport’s7Davenport H.W. Gastric mucosal hemorrhage in dogs effects of acid, aspirin and alcohol.Gastroenterology. 1969; 56: 439-449Abstract Full Text PDF PubMed Scopus (168) Google Scholar findings supported the concept that gastric mucosal injury by aspirin was a major source of blood lost into the GI tract, whether or not a frank gastric ulcer was present. Only in 1968 was the inhibitory effect of aspirin on platelet function first described8O’Brien J.R. Effects of salicylates on human platelets.Lancet. 1968; 1: 779-783Abstract PubMed Google Scholar; this effect later was found to be shared with all classic nonselective nonsteroidal anti-inflammatory drugs (NSAIDs). During the 1980s, there was rapid acceptance of the concept that acute or chronic use of either aspirin or nonaspirin NSAIDs was accompanied by the unpredictable development of gastrointestinal hemorrhage,9Somerville K. Faulkner G. Langman M.J.S. Non-steroidal anti-inflammatory drugs and bleeding peptic ulcer.Lancet. 1986; 1: 462-464Abstract PubMed Scopus (456) Google Scholar less commonly perforation,10Coggon D. Lambert P. Langman M.J.S. 20 years of hospital admission for peptic ulcer in England and Wales.Lancet. 1981; 1: 1302-1304Abstract PubMed Scopus (132) Google Scholar, 11Collier D.S.J. Pain J.A. Non-steroidal anti-inflammatory drugs and peptic ulcer perforation.Gut. 1985; 26: 359-363Crossref PubMed Scopus (232) Google Scholar or, rarely, death.12Catford J.C. Simpson R.J. Confidential enquiry into deaths from peptic ulcer.Health Trends. 1986; 18: 37-40Google Scholar, 13Griffin M.R. Ray W.A. Schaffner W. Non-steroidal anti-inflammatory drugs and deaths from peptic ulcer in elderly adults.Ann Intern Med. 1988; 109: 359-363Crossref PubMed Scopus (393) Google Scholar The population risk of sustaining one of these adverse events was estimated by the US Food and Drug Administration to be of the order of 2%–6% per annum.14O’Neill R.T. Testimony to the arthritis advisory committeeFDC reports. FDA, Rockville, MD1987: 8-9Google Scholar Most clinicians assumed that the observed adverse events were complications of gastric or duodenal ulcers that had been caused by the drugs and then bled. Other factors possibly contributing to GI hemorrhage largely were ignored. These included the universal effect of aspirin/NSAIDs, serious platelet dysfunction, and the possibilities that the drugs could initiate GI bleeding from either pre-existing GI conditions or from nonpeptic lesions caused by the drugs in more distal sites. However, such assumptions were challenged by results of the VIGOR trial,15Bombardier C. Laine L. Reicin A. et al.Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis.N Engl J Med. 2000; 343: 1520-1528Crossref PubMed Scopus (3699) Google Scholar which showed that of 31 GI bleeds in rofecoxib-treated patients, only 12 were attributable to gastroduodenal ulcers, and of 82 bleeds in naproxen-treated patients, only 32 similarly were attributable to gastroduodenal ulcers. A later publication from this trial confirmed that much of the overt bleeding originated distal to the stomach and duodenum,16Laine L. Connors L.G. Reicin A. et al.Serious lower gastrointestinal clinical events with non-selective NSAID or coxib use.Gastroenterology. 2003; 124: 288-292Abstract Full Text PDF PubMed Scopus (326) Google Scholar although the study could not assess the relative contributions to bleeding from underlying GI disease, comorbid illness, or co-therapy with other medications. The Base of the Iceberg of Nonsteroidal Anti-inflammatory Drug–Induced Gastrointestinal BleedingIn this issue of the journal, Bowen et al17Bowen B. Yuan Y. James C. et al.Time course and pattern of blood loss with ibuprofen treatment in healthy subjects.Clin Gastroenterol Hepatol. 2005; 3: 1075-1082Abstract Full Text Full Text PDF PubMed Scopus (19) Google Scholar report that the use of moderate doses of ibuprofen (800 mg 3 times/day) for 4 weeks also leads to an increase in mean daily fecal blood loss in all patients, sim