Do NSAIDs cause dyspepsia? a meta-analysis evaluating alternative dyspepsia definitions

Abstract

Nonsteroidal anti-inflammatory drugs (NSAIDs) may be associated with dyspepsia, but the relationship is obscured by variations in the terminology used to report GI symptoms. Using alternative definitions, we assessed the relationship between NSAID use and dyspepsia. We searched MEDLINE, EMBASE, HEALTHSTAR, and BIOSIS databases (1966-1997) and New Drug Application reviews, identifying randomized, placebo-controlled trials (5 days or more duration) of any NSAID, reporting original data on GI complications. Based upon reported terms describing upper GI symptoms, we derived three definitions: strict, using terms synonymous with epigastric pain/discomfort; loose, (containing the strict definition plus terms for heartburn, nausea, bloating, anorexia, and vomiting); and a loose definition without heartburn terms (the loose-less-heartburn definition). Using each definition, we performed a random-effects model meta-analysis of the relationship between NSAID exposure and dyspepsia. Fifty-five published and 37 unpublished controlled NSAID trials met our inclusion criteria. The mean duration of the trials was 33.2 days (SD 40 days). Application of the strict definition resulted in a pooled risk ratio of dyspepsia for NSAIDs compared with placebo of 1.36 (95% CI = 1.11-1.67). For the loose definition, the pooled risk ratio was 1.13 (95% CI = 0.98-1.32). The loose-less-heartburn definition yielded a pooled risk ratio of 1.19 (95% CI = 1.03-1.39). In the placebo-treated control groups, the percent of patients reporting dyspepsia ranged from 2.3% (strict definition) to 4.2% (loose and loose-less-heartburn definitions). Using the strict definition, based solely on epigastric pain-related symptoms, NSAIDs increased the risk of dyspepsia by 36% (p < 0.05). These findings may be useful in creating a standardized definition of NSAID-related dyspepsia.