What are the risks of long-term NSAIDs and COX-2 inhibitors?

Journal of Family Practice, March, 2003 by Michael DeBisschop

Wright JM. The double-edged sword of COX-2 selective NSAIDs. CMAJ 2002; 167:1131-1137.

* PRACTICE RECOMMENDATIONS

This review presents an interesting new analysis of cyclo-oxygenase-2 (COX-2) inhibitor safety, concluding that long-term use results in more serious adverse events than traditional nonsteroidal anti-inflammatory drugs (NSAIDs).

The nonsystematic and retrospective properties of this analysis limit its validity. However, the fact that an evaluation of long-term data found some small harm to COX-2 inhibitors relative to traditional NSAIDs (number needed to harm=78 over 9 months) should give clinicians pause. Until better meta-analyses or new safety data are published, clinicians should prescribe COX-2 inhibitors long-term only for those patients deemed to be at high risk of ulcer complications.

* BACKGROUND

Much of the widespread use of COX-2 inhibitors is due to the perception that they are safer than traditional NSAIDs. However, in terms of patient-oriented outcomes, their real safety is unknown.

The 2 major studies in this area, the Celecoxib Long-term Arthritis Safety Study (CLASS) and the Vioxx Gastrointestinal Outcomes Research (VIGOR), originally published results based on 6 months of data. However, as published in publicly available Food and Drug Administration (FDA) reports, these studies actually accrued a median of 9 months of data. The author of this review discusses the reasons for the differences in safety between the 2 classes of NSAIDs, and presents new analyses of their safety based on the longer study periods.

* POPULATION STUDIED

The patient population being analyzed is the same as in the CLASS and VIGOR studies: adults with osteoarthritis and rheumatoid arthritis. The author reanalyzed results from these 2 studies only.

* STUDY DESIGN AND VALIDITY

This was not a systematic review; no literature searches were performed or inclusion criteria described. The author apparently extracted data from the FDA's public reports on the full, long-term results of both the CLASS and VIGOR studies. These data were used to provide separate and pooled estimates of adverse events for each study.

A limitation of this review is that it only included 2 studies; however, these 2 studies are the largest trials published on this issue. Also, the author's definition of "serious adverse event" was not well defined, and retrospective examination of the data in the FDA's report could potentially lead to a biased analysis in favor of the author's viewpoint.

The author also did not consider any possible confounding variables or limitations of including the later data in the analysis. For example, in a letter to the editor, (1) the authors of the CLASS study stated that subjects in the traditional NSAID group dropped out earlier in the study. These patients were those at greater risk of complicated ulcers, resulting in a lower-risk group later on and biasing the long-term results. Nevertheless, these studies represent the best long-term data available on the safety of COX-2 inhibitors.

* OUTCOMES MEASURED

The author reported total mortality, serious adverse events (death, hospital admission, life-threatening events), and complicated ulcers for each study individually and a pooled estimate for both studies (median duration of 9 months).

* RESULTS

There was no significant difference in overall mortality between traditional NSAIDs and COX-2 inhibitors for either study or in the pooled estimate. The risk of serious adverse events was significantly higher (absolute risk increase =1.3%; number needed to harm [NNH]=78) in the pooled estimate as well as the rofecoxib study, but not in the celecoxib study. The risk of complicated ulcers was significantly lower in the rofecoxib study (absolute risk decrease=0.52%; NNH=192), but not in the celecoxib study or the pooled estimate. No P values or confidence intervals were presented.

REFERENCES

(1.) Silverstein F, Simon L, and Faich G. Reporting of 6-month vs 12-month data in a clinical trial of celecoxib. JAMA 2001; 268:2398-2399.

Michael DeBisschop, PharmD, University of Wyoming Family Practice Residency, Casper. E-mail: medrx@uwyo.edu.