Sirs, I read with interest the recently published ‘guidelines’ for appropriate use of non-steroidal anti-inflammatory drugs (NSAIDs), cyclo-oxygenase-2-specific (COX-2) inhibitors, and proton pump inhibitors (PPIs) in chronic NSAID users.1 The authors are to be congratulated for clearly identifying the background of the study and the support for the project. It is implied that the drug company sponsor's motivation was primarily altruistic leading them to invest the large sums required to host the meeting, to select a scientific-writing group to record and draft the manuscript, and pay honorarium and travel for the participants.
The guidelines are described as ‘evidence-based’, which implies the presence of analyzable data and a knowledgeable and unbiased group of judges. The small group of nine participants were chosen to minimize bias based on speciality, practice environment, or region as they generally practiced in different specialties and different parts of the country. One wonders how the nine participants was selected, and how the proportions of specialist and academic physician were arrived. It is not clear that they were either representative of the general medical community or free of major conflicts of interest.
Risk reduction requires defining risk. The authors’ citation for reduction in risk in the introduction (‘which similarly reduces the gastrointestinal risk’) is the Acid Suppression Trial: Ranitidine vs. Omeprazole for NSAID-associated Ulcer Treatment (ASTRONAUT) study which did not assess risk in terms of clinical events such as bleeding or perforation.2 Unfortunately, to date all of the large drug company sponsored studies regarding PPI use for prevention of NSAID-induced gastroduodenal damage have used surrogate endpoints (often as composite endpoints that included endoscopic ulcers, erosions, and symptoms) rather than clinical endpoints such as hemorrhage. In addition the published trials have often used ‘unfair’ comparisons. For example, in the ASTRONAUT study2 and the Omeprazole vs. Misoprostol for NSAID-induced Ulcer Management (OMNIUM) study,3 omeprazole was compared with standard-dose ranitidine (150 mg b.d.) (which parenthetically had repeatedly been shown to be ineffective for NSAID ulcer prevention)4 and to half-dose misoprostol (200 μg b.d.) (equivalent to approximately 600 mg of cimetidine as an antisecretory drug). Post hoc analysis of these studies separating the results for those with and without Helicobacter pylori infection showed that the majority of the advantage among omeprazole users was in the H. pylori infected patients and half-dose misoprostol was actually superior to omeprazole for both healing and prevention of gastric ulcers among those without H. pylori infection.5 The significant advantage for omeprazole over standard-dose ranitidine (ASTRONAUT) was lost in the H. pylori negative patients with rates for gastric ulcer being 21 vs. 30 per 100 patient years and for duodenal ulcer 2.4 vs. 4.2 per 100 patient years).5 These results were confirmed comparing full dose misoprostol and lansoprazole.6
There are no large randomized trials showing a significant reduction in clinical events with a PPI and use of a COX-2 inhibitor alone or the combination of a PPI and a traditional NSAID resulted in an unacceptably low reduction in clinical events among high-risk patients, e.g. those with a recent non-H. pylori NSAID-induced upper gastrointestinal (UGI) bleed?7 Why not misoprostol instead? One wonders whether the ‘guidelines’ have been different if sponsored by the manufacturer of misoprostol. Although misoprostol has a higher potential for side effects, it is important to remember that it was taken successfully by the majority of participants in the large clinical trials. Why not an H2-receptor antagonist, or even sucralfate? Was the focus on a PPI at all related to the sponsor's new combination product consisting of naproxen and lansoprazole (NaproPak)?
Academic physicians are often chosen to ‘edit’ and participate in the manuscript development possibly to add an aura of respectability. Would it not be important for the reader to know if those authors had major potential conflicts of interest? The above is not to say the ‘guidelines’ might not be useful, rather it is to decry the long-standing traditions whereby drug companies use willing, trusting, and possibly naive academic and non-academic physicians to further their own agendas.8