Our study indicates that the release of adverse postmarket-ing safety information in late 2004 and early 2005 dramatically influenced prescribing patterns for NSAIDs. Although the total number of prescriptions dispensed for NSAIDs has decreased overall since the fall of 2004, most of this decrease was attributable to a 66% reduction in prescriptions for the COX-2-selective NSAIDs.

Although the efficacy of COX-2-selective agents is similar to that of the traditional NSAIDs, their GI side effects are fewer in number. Adverse cardiovascular side effects—primarily nonfatal myocardial infarction (MI)—were noted with rofecoxib 50 mg once daily, compared with the usual doses of canadian naproxen (500 mg twice daily), as early as 2000, with the publication of the Vioxx Gastrointestinal Outcomes Research (VIGOR) study.

However, the Celecoxib Long-term Arthritis Safety Study (CLASS) found no increase in cardiovascular side effects twice daily, compared with the usual doses of either diclofenac potassium (Novartis) or ibuprofen.

Retrospective claims data analyses also demonstrated an elevated cardiovascular risk with rofecoxib, especially at high doses, but not with celecoxib. Published data relating specifically to the changes in NSAID prescribing trends that we describe (see Figure 2) are summarized next.

In September 2004, the Data Safety Monitoring Board of the Adenomatous Polyp Prevention on Vioxx (APPROVe) trial reported an increase in cardiovascular events with rofecoxib. Merck stopped the trial early. On September 30, 2004, the company voluntarily withdrew rofecoxib from the market.

In November 2004, greater cardiovascular toxicity was noted in patients after coronary artery bypass graft (CABG) surgery who were given parecoxib, an injectable pro-drug of valdecoxib and oral valdecoxib when compared with patients receiving placebo. These findings led to a contraindication in the valdecoxib label for the treatment of postoperative pain immediately following CABG surgery, and they resulted in a boxed warning for the product.

In December 2004, subsequent data from two similar placebo-controlled polyp-prevention trials with celecoxib were contradictory in terms of their cardiovascular side effects. The

Adenoma Prevention with Celecoxib (APC) trial was terminated early because of a dose-dependent increase in composite endpoint of death from cardiovascular causes, MI, stroke, or heart failure in subjects taking celecoxib compared to those taking placebo. However, results of the Prevention of Spontaneous Adenomatous Polyps (PreSAP) trial did not show an increase in cardiovascular events in subjects taking celecoxib compared with those taking placebo.

In December 2004, the FDA also reported the termination of a National Institutes of Health (NIH) study that was using, celecoxib, or placebo to treat elderly patients at risk for Alzheimer’s disease, the Alzheimer’s Disease Anti-inflammatory Prevention Trial (ADAPT). A preliminary analysis indicated an increased risk of cardiovascular events in the naproxen-treated subjects compared with those treated with placebo and suggested no difference between placebo and celecoxib.

In February 2005, the FDA convened an Arthritis Advisory Council Committee to evaluate all published and non-published information and to make recommendations regarding the marketability, labeling, and overall safety of the COX-2-selective NSAIDs coxib drugs.

On April 6, 2005, the FDA decided that the data did “not clearly demonstrate that the COX-2-selective agents confer a greater risk of serious adverse CV events than nonselective NSAIDs.” The agency also found that both selective and non­ selective NSAIDs demonstrated an increased risk, compared with placebo, but it was unable to rank the drugs with respect to cardiovascular risk.

The FDA recommended a black-box label for all prescription NSAIDs to include warnings of both cardiovascular and GI risk; they also recommended an expanded warning for over-the-counter NSAIDs, including GI bleeding risk, cardiovascular risk, and serious skin reactions. The agency also requested the voluntary withdrawal of valdecoxib because of an increase in reports of serious skin reactions, including Stevens-Johnson syndrome, compared with other selective NSAIDs and increased safety labeling with celecoxib and all nonselective NSAIDs.


Some limitations of our study included the retrospective nature of the claims data, which were restricted to prescribing in New York State. These data also included fewer than half of the pharmacy beneficiaries of an insurance company, and we could not track over-the-counter NSAID use.

Data were available only through June 2005. An additional analysis of prescription claims data beyond this time, as well as the findings from subsequent publications, would be helpful to quantify the sustained impact of these events on overall NSAID use.


Our data demonstrate the effect of adverse events on NSAID prescribing in northern New York State. Celecoxib canadian use increased following the removal of rofecoxib from the market (Figure 2) but then decreased sharply as concerns were raised following the release of the APC trial results. Similarly, naproxen use fell after the NIH terminated its Alzheimer’s disease study (ADAPT).

The use of other NSAIDs, which had received no adverse publicity—the cardiovascular risk has not been investigated in most of these drugs—either increased or remained unchanged. These monthly changes in prescribing patterns reflect the responses of clinicians, health care systems, the pharmaceutical industry, and consumers to these series of events.