Aspirin and gastrointestinal haemorrhage: a methodologic assessment.

The literature on the association between Aspirin (ASA) use and gastrointestinal haemorrhage (GIH) has been reviewed. Thirteen control-studies were selected and examined for (i) the type of design (ii) the choice of outcome measures, and (iii) the adequacy of satisfying defined causation criteria and the influence of sampling strategies on one of these criteria--strength of association. Of the 13 studies seven were large scale randomized control trials (RCTs) and six were case-control studies. Among the outcome measures utilized in these studies and currently available for the assessment of GIH, we found none were uniquely satisfactory for use as a "gold standard". Furthermore, in none of these studies was the particular chosen outcome measure sufficiently satisfactory to allow firm conclusions on the issue of causation. None of the studies adequately met all the defined causation criteria or adequately eliminated the potential biases in the chosen sampling strategies. Thus the evidence that aspirin causes GIH fails to stand up to critical evaluation. This, of course, does not mean that ASA does not cause GIH. However, it is common practice for physicians to advise their patients with a prior history of GIH and or gastrointestinal symptoms to refrain from ASA use. This is probably sound advice in the setting where ASA is being used as a casual or short term analgesic or anti-inflammatory agent. But in the context of severe inflammatory joint disease when the use of ASA is clinically indicated, it should not be withheld on the basis of the risk of GIH. There is a widespread belief in medicine that by implication aspirin and other nonsteroidal anti inflammatory drugs are a common cause of bleeding from the gastrointestinal tract, yet it has been pointed out frequently [1-3] that the evidence to support the belief is weak and circumstantial. Despite this the perceived association of aspirin with gastrointestinal bleeding influences both pharmaceutical and clinical practice and often hinders or prevents the prescription of anti-inflammatory medication to patients with severe inflammatory joint disease. It was felt that a proper critical appraisal of the evidence was essential. It is unfortunate that the evidence most frequently cited implicating ASA as a cause of GIH is almost entirely related to the casual use of the drug. Thus our conclusions must be viewed with some caution in the patient group where this knowledge would be of most importance, that is in patients with inflammatory joint disease.