Objective: We examined to what extent the evidence of the relative gastrointestinal toxicity with non-steroidal anti-inflammatory drugs (NSAIDs) was implemented in clinical practice in Bologna, Italy, Funen, Denmark, and Stockholm, Sweden, areas with accurate computerised information on prescriptions purchased by defined populations.
Methods: We ranked each NSAID by purchased volume in defined daily doses during September 1996 and compared it with the ranking of gastrointestinal complications from a meta-analysis of controlled epidemiological studies published between 1986 and 1994. We restricted our comparison to those NSAIDs that accounted for 90% of the use and within this DU90% segment we determined the proportion of "high risk" (azapropazone, ketoprofen, piroxicam) and "low risk" (ibuprofen, diclofenac) drugs with respect to gastrointestinal toxicity.
Results: In Funen, Denmark, we found the best NSAID profile (63% low risk/11% high risk) while Bologna, Italy, had the other extreme (26% low risk/38% high risk), with Stockholm, Sweden, in between (43% low risk/20% high risk).
Conclusion: Our study suggests that factors other than evidence-based medicine had a dominating impact on the use of prescription NSAIDs in 1996.