An estimated 50 million Americans now take aspirin regularly to prevent cardiovascular disease. To reduce the risk of gastrointestinal hemorrhage, lower dosages of aspirin have been recommended, and expensive modifiedrelease formulations have been developed. However, the comparative safety of the alternatives has not been established. Derry and Loke conducted a meta-analysis of randomized controlled trials to establish the relative risk of gastrointestinal hemorrhage during long-term aspirin therapy.

The authors used electronic and traditional search strategies to identify randomized, controlled clinical trials with at least 50 patients in each treatment group. Studies were excluded if they were restricted to selected groups of patients, such as pregnant women, children or patients with platelet disorders. The analysis included trials that compared aspirin with placebo or no treatment, but excluded trials that compared different dosages of aspirin or that compared aspirin with other antiplatelet or anticoagulant medications. All trials were double-blinded and placebo-controlled, and were checked independently by two reviewers for eligibility. Data from eligible studies were pooled to calculate odds ratios for adverse outcomes.

The meta-analysis included 24 randomized, controlled trials that involved 65,987 patients. About 74 percent of study participants were middle-aged men. The aspirin dosage ranged from 50 to 1,500 mg per day, and the mean duration of therapy was 28 months. All trials excluded patients at high risk of gastrointestinal hemorrhage and those with contraindications to aspirin therapy. The indications for aspirin therapy ranged from primary prevention in healthy persons to secondary prophylaxis following stroke.

The overall rate of gastrointestinal hemorrhage was 2.47 percent in patients taking aspirin compared with 1.42 percent in those taking placebo. The authors calculate an odds ratio of 1.68 for the occurrence of gastrointestinal bleeding during aspirin use. They estimate that for one patient to be harmed, 106 patients needed to be treated for 28 months. To investigate any reduced risk with lower dosages of aspirin, the eight trials that used dosages of 50 to 162.5 mg per day were analyzed separately, and regression analyses were performed relating the aspirin dosage to hemorrhage risk. Neither of these investigations conclusively demonstrated that lower dosages of aspirin were safer than conventional dosages. The regression analyses estimated a reduction of 1.5 percent for each 100-mg dosage reduction, but this finding was not significant.

The authors conclude that long-term aspirin therapy is associated with a risk of gastrointestinal hemorrhage of about one per 100 patients treated for 28 months. Currently, no good evidence supports the use of lowerdosage or modified-release preparations to reduce this risk.