LEDA Home Page Harvard Law School LEDA at Harvard Law School

Untangling the Vioxx-Celebrex Controversy: A Story about Responsibility

Lan Tran

Class of 2005

May 4, 2005

This paper is being submitted in satisfaction of both the Food and Drug Law course requirement and the third year written work requirement.

While the withdrawal of prescription drugs is not a new concept, the Vioxx and Celebrex controversy has put a new spin on the issue of withdrawal. Vioxx and Celebrex belong to the same class of drugs and both were linked to increased heart risks. While Merck chose to withdraw Vioxx from the market, Pfizer decided to keep Celebrex put, choosing instead to suspend all direct-to-consumer advertising and make small labeling changes. This paper attempts to construct a story about why these two companies took such contrasting reactions, which was right, and what should be done in the future. But before delving into that analysis, it is important to understand the background of the controversy, including the history and development of Vioxx and Celebrex, why they were approved in the first place, and the basis for such approvals. Next, it is crucial to comprehend the different medical studies that have been conducted, which are the basis on which action has been demanded from the companies. From there, it may then be possible to analyze the different actions taken by Merck and Pfizer and what resulted from such actions, which will allow for better basis in which to evaluate who was right. Finally, in every step of the controversy, the FDA’s actions will be examined, including its formal and informal guidance, regulatory powers, and an analysis of when and why it chooses to take or not take action.




Donna Allen, a 50-year old technology recruiter from Berlin, Massachusetts suffers from rheumatoid arthritis, a condition that makes simple tasks like turning a doorknob or the car ignition key quite tormenting. In the midst of all the negative publicity of the cardiovascular risks of COX-2 inhibitors, drugs approved for the treatment of rheumatoid arthritis, Allen says, “You worry about damage to your cardiovascular system every time you eat a cheeseburger, too. In the meantime, I’m pain free.”[1] In one sentence, Allen has elegantly and poignantly pointed out the heart of the issue relating to these COX-2 inhibitors, which is an issue of cost and benefit.

No drug is completely safe. Yet drugs are prescribed and taken everyday, and regular drug use has become part of a daily routine for many Americans. We all take drugs on the assumption that the benefits outweigh the costs. So what does it mean for a drug to be withdrawn from the market? Does it mean that the FDA has made a huge mistake in allowing the drug to have ever step foot on the market? Does it mean the drug approval process is flawed? Such sentiments and uncertainties often accompany withdrawal of drugs, as seen with Baycol in 2001 and Fen-Phen in 1997. Vioxx, one of the most recent drugs removed from the market has stirred up these questions once again. But the Vioxx story becomes much more complicated than Baycol or Fen-Phen. On September 30, 2004, Merck announced a voluntary withdrawal of Vioxx, a COX-2 inhibitor, after the results of a study showing an association between the drug and increased heart risk.[2] About three months later, on December 19, 2004, Pfizer announced a suspension of all direct-to-consumer advertising for Celebrex, also a COX-2 inhibitor, after the results of a study showing an association between Celebrex and increased heart risk.[3] Pfizer’s decision was a double shock, the first shock pertaining to the increased heart risk and the second shock pertaining to the drastically different action in response to the findings of the risk. All were baffled by the divergent reactions—that is, Merck’s decision to withdraw and Pfizer’s decision to suspend advertising. This paper attempts to construct a story about why these two companies took such contrasting reactions, who was right, and what should be done in the future.

It turns out that between the different decisions lie layers of complexities, including differences in drug components, medical studies, histories, and even FDA guidance and reaction to the new-found risks. In order to understand the situation, it is important to comprehend the history of the drugs, including why they were approved in the first place and the basis for such approvals. Next, it is important to understand the different studies that have been conducted and the value and limitations of each study. From there, it may then be possible to analyze the different actions taken by Merck and Pfizer and what resulted from such actions, which will give us a better basis to evaluate who was right. Of course, the FDA, as the enforcing agency, played a central role in the controversy, and the agency’s actions will be examined accordingly. I am attempting to construct a plausible story out of the hodgepodge of events relating to Vioxx and Celebrex, which will inherently have its limitations, but will hopefully nevertheless serve as guidance for reform.


Vioxx and Celebrex belong to a class of drugs known as nonsteroidal anti-inflammatory drugs or NSAIDs.[4] NSAIDs are commonly prescribed for the inflammation of arthritis and other body tissues, and include drugs such as aspirin, ibuprofen (Motrin), and naproxen (Aleve or Naprosyn).[5] NSAIDs like aspirin and ibuprofen relieve pain by inhibiting a pair of enzymes called COX-1 and COX-2, which produced prostaglandins, chemicals which are believed to cause pain and inflammation of injuries.[6]

Researchers then discovered that it was only the blocking of COX-2 enzymes that relieved pain and that the blocking of COX-1 interfered with blood clotting and caused stomach irritation and bleeding.[7] This discovery explained the increase in gastrointestinal (GI) problems, such as ulcers, associated with long-time users of NSAIDs. Each year, clinical upper GI events occur in 2 to 4 percent of patients taking nonselective NSAIDs.[8] Naturally, pharmaceutical companies began experimentation to selectively inhibit COX-2 without inhibiting COX-2 in order to derive the benefits of relieving pain and inflammation but without the adverse GI side effects.[9]

The outcome of such experimentation was the creation of a new class of NSAIDs known as COX-2 inhibitors. The most popular COX-2 inhibitors include Vioxx and Celebrex. Aside from treating pain and inflammation, many other uses of COX-2 were explored.[10]



In the United States, the approval process for a new drug is an extremely lengthy and costly process, taking about 7 to 13 years and costing about $30 to $50 million dollars, and before a new drug application (NDA) can even be submitted to the FDA, three phases of research and testing must be conducted, taking about 5 to 10 years.[11] These studies form the basis on which the FDA will decide whether or not to approve the drug. Specifically, an FDA advisory committee will discuss these studies and make recommendations to the agency. Although the FDA is not bound by the decision of the committee, for all practical purposes, it should be noted that the FDA usually follows the advice of its advisory committees.[12]

The fact that an NDA is filed for a drug indicates at least sufficiency of merit for approval. Objectively speaking, it means that the drug has passed the Investigation New Drug (IND) review process and Phases I, II, and III of the clinical review process, which is not an easy feat.[13] Moving from the IND process to the clinical review process indicates that there was adequacy of: protection of the human research project, animal studies completed and analyzed, scientific merit, and qualifications of the investigator.[14] Phase I includes pharmacological and toxicological studies on multiple doses to detect adverse effects on one human volunteer.[15] Phase II includes testing on a small number of patients with multiple dosages to study the therapeutic effects, and only drugs that are determined to be safe and effective will enter Phase III.[16] In this phase, hundreds or thousands of patients are studied in a clinical setting (usually randomized, double-blind, and placebo-controlled), where the drug is typically compared with a placebo or standard drug, and any adverse reactions and drug interactions are observed.[17] Today, phases are often combined.[18] After completing at least two adequate and well-controlled Phase III studies, the clinical, pharmacological, and toxicological data is then compiled into a NDA and submitted to the FDA for approval.[19] Thus, by the time Merck and Searle submitted NDA’s for Vioxx and Celebrex, there was a lot of existing information on the drugs. The next step in the process was to wait for approval from the FDA, but perhaps more importantly, to get approval from the FDA Arthritis Advisory Committee.

It should be noted that Searle originally submitted the NDA for Celebrex. The drug was developed by the Searle division of the Monsanto Corporation.[20] Monsanto merged with Pharmacia in 2000.[21] Three years later, in April of 2003, Pharmacia merged with Pharmacia in a 60 billion buyout.[22] This paper will focus on the actions of Pfizer, the company that co-marketed Celebrex with Pharmacia in 2001[23] and finally took full control of the drug after the buyout in 2003.


1. The Framework :

The Arthritis Advisory Committee convened on December 1, 1998 to discuss the NDA for Celebrex (celecoxib) in order to make a recommendation to the FDA, including approval of drug uses, doses, labeling, etc.[24] The crux of the discussion was based on Searle’s NDA[25] , which is supposed to contain all favorable and unfavorable information Searle, the sponsor, obtained through research and studies relating to the safety and effectiveness of the drug.[26] The meeting began with a presentation from Searle researchers followed by a presentation of FDA researchers.[27] Presenters reported, discussed, analyzed, and fielded questions on matters of their expertise, including topics such as animal studies, pharmacokinetics[28] , pharmacology, toxicology, gastrointestinal safety, renal safety, overall safety, pain, and efficacy for specific diseases.[29] An open public hearing followed the presentations where different groups and organizations were allotted time to endorse or raise concerns about Celebrex.[30]

2. Concerns :

Public organizations voiced many concerns during the open public hearing. Dr. Palmer of SmithKline Beecham questioned the benefits of COX-2 inhibitors and berated the media for dubbing it as a super aspirin.[31] He stressed that the benefits of COX-2 inhibitors have been overstated, especially in light of the relatively small number of studies conducted, not to mention that the benefits of the inhibition of COX-1 enzymes have been largely ignored.[32] Dr. Soller of Science and Technology for the Non-prescription Drug Manufacturers Association asked the committee to make efforts to ensure that the data on Celebrex was not misrepresented.[33] Similarly, Dr. Bryant of the Aspirin Foundation underscored the concern for misrepresentation and asked the committee to evaluate Celebrex on its own merit and not comparatively with other drugs, stressing that comparative claims should be validated by rigorous studies.[34] Dr. Wolfe of the Public Citizens Health Research Group was solely concerned with labeling and readily acknowledged the efficacy of Celebrex in treating osteoarthritis and rheumatoid arthritis. Dr. Wolfe did not believe the NSAID class warning label should be dropped unless it was shown that Celebrex, compared with the least dangerous of the NSAIDs, Ibuprofen, had a statistically lower rate of serious GI complications.[35]

Right on cue with Dr. Wolfe, the biggest concern voiced by committee members was related to the issue of labeling, particularly on how to label Celebrex accurately in light of all the studies, some of which were inconclusive or raised concerns of efficacy and safety. The most difficult issue was whether to require Celebrex to carry the traditional NSAID warning template on its label. NSAIDs currently carry a class warning template on their labels cautioning users of an increase risk of adverse GI events. Celebrex is also part of the NSAID class but is the first selective NSAID, (as opposed to non-selective NSAIDs) on the market, and Searle thus was hoping to be able to abandon the class warning template.[36] While studies indicated that endoscopically diagnosed ulcers occur less frequently with patients taking Celebrex compared to patients taking NSAID comparators, studies completed did not include definitive comparisons of clinically significant GI adverse events.[37] After much debate, committee members voted to keep the warning template on the label but to modify it to reflect studies affecting the organ systems studied.[38] This conservative approach to labeling was seen throughout the discussions, and Celebrex was not permitted label comparisons with other NSAIDs.

Other concerns of committee members dealt with dosages and treatment, especially in special populations, such as the elderly or those with hepatic metabolism.[39]

3. Recommendations :

A unanimous committee voted that Celebrex should be approved for the indications of the treatment of the signs and symptoms of osteoarthritis and rheumatoid arthritis.[40] Committee members recommended that any Phase IV studies should include pediatric and bone studies.[41]


1. The Framework :

The Arthritis Advisory Committee convened on April 20, 1999 to discuss the NDA for Vioxx or rofecoxib in order to make a recommendation to the FDA.[42] According to Sandra Kweder, Deputy Director of the Office of New Drugs[43] (a division of the FDA), Vioxx was given a six-month priority review because the drug potentially provided a significant therapeutic advantage over existing NSAIDs, particularly a lower rate of adverse GI events, including bleeding.[44] The format of the meeting was typical: Merck researchers presented, followed by FDA researches, and the meeting concluded with a series of discussion and questions.[45]

The original safety database for Vioxx was about 5000 patients.[46] In the clinical trials conducted, the risk of adverse GI events was determined through the use of endoscopy, which showed a significantly lower risk of GI ulcers in Vioxx treatment compared to Ibuprofen treatment.[47] Vioxx treatment did not show an increase risk in heart attack or stroke.[48]

2. Concerns :

One major concern of committee members was the practice of “dose creep” or patients voluntarily taking more than the approved dose of a drug, which is an off-label use. Although they recognized the inherent problem of dose creep since monitoring patient intake is impossible, the concern was heightened because of suggestion from the data of dose-related toxicity.[49] Researchers for Merck stated that the dose should be set at 12.5 mg per day, which was efficacious for most patients and that some could go up to 25 mg per day based on need.[50] But there was concern that there may have been higher efficacy at 50 mg per day and that patients would be tempted to go that high even though it may not be safe, and there was discussion on whether labeling should reflect these concerns.[51]

Another concern was again an issue of dose, and it was decided that for acute pain, 50 mg could be taken per day up to five days, since there was evidence of adverse events after five days.[52] It was agreed that discontinuation after five days for acute pain should be reflected in the label.[53] A majority of the committee members agreed that the label should convey that dosing at 50 mg per day might be associated with increased peripheral edema and other renal adverse effects.[54]

All issues of labeling were eventually resolved with caution, and the approach adopted was to only state what was known from studies to date.[55] The NSAID class warning template was an issue that came up once again, and the conclusion was similar to that applied to Celebrex—leave the label in place, but modify it with what is known from trials studies.[56] Any comparisons made to other NSAIDs on the label had to be dose-specific and include limitations.[57]

3. Recommendations :

The committee unanimously recommended the approval of Vioxx for the indication of the treatment and signs and symptoms of osteoarthritis at a dose of 12.5 mg to 25 mg per day.[58] The committee recommended that any Phase IV studies should include studies on the elderly, pediatrics, and dose creep at the 50mg level.[59]


On December 31, 1998, Celebrex was approved at doses of 100 mg and 200 mg for the signs and symptoms of osteoarthritis and rheumatoid arthritis.[60] On May 20, 1999, Vioxx was approved for the relief of the signs and symptoms of osteoarthritis, the management of acute pain, and the treatment of primary dysmenorrheal (menstrual cramps).[61] New uses and doses were approved for both Vioxx and Celebrex upon subsequent studies.


Questions regarding whether Vioxx and Celebrex should have been approved in the first place reflect the common criticism that the FDA’s drug approval process is too lax. The approval processes for Vioxx and Celebrex summarized above is a very short rundown of the actually discussions that took place. For each drug, the panelists reviewed an NDA, which is an enormous collection of data. All research data and study results, whether negative or positive, must be included in the NDA. A good portion of the time was dedicated to questions, in which panelists pressed and probed the researchers about different aspects of a drug. Although it is debatable, it has not been shown that the FDA’s so-called “gold standard” has been compromised due to its faster review process. The only legitimate criticism therefore, would be that the FDA’s review process in general is not rigorous enough. As described in subsequent parts of this paper, the FDA’s review process is highly defined and systematic. Approval is a long and arduous process for drug companies, and this is reflected in the fact that although there have been large increases in research and development (R & D) spending in the past decade, fewer drugs have been submitted for approval.[62] Because the FDA applies the same level of scrutiny to all NDAs, any argument that Vioxx and Celebrex should not have been approved because of lax standards must necessarily imply that the FDA must intensify its approval review for all drugs, which is a general criticism of the FDA and not a problem that is idiosyncratic to Vioxx and Celebrex. This does not mean that the approval process should not generally be reevaluated.

The Vioxx and Celebrex approval process also brings up criticisms of the accelerated review process, which has it origins at a time when public outcry demanded that new AIDS medication be approved faster.[63] Vioxx and Celebrex were both given accelerated reviews because of their potential to provide a “significant therapeutic advantage” over existing NSAIDS due to a decrease of adverse GI events.[64] Again, any criticism on the expedited review process is also a general criticism of the FDA and not an issue specific to Vioxx and Celebrex. Still, this does not mean that the expedited process should not be reevaluated. For instance, a question that arises is, should Vioxx and Celebrex have been given an accelerated review at all? With both drugs, the FDA concluded that the decreased rate of adverse GI events was not sufficiently demonstrated or statistically significant in the studies conducted.[65] That is why the NSAID warning label was mandated for both.[66] Later studies by Merck convinced the FDA to remove such warning from Vioxx, but Celebrex was still required to carry the warning.[67] Furthermore, according to Dr. Psaty, if the results of VIGOR, which became available in December of 1999, had been available only seven months earlier, Vioxx may not have been approved.[68]

In the end, the approvals of Vioxx and Celebrex were the result of meticulous cost-benefit analyses. Looking at all the costs and benefits of the drugs, panel experts concluded that the drugs were efficacious for certain indications and warranted warnings for certain side effects, but were safe overall for human use according to the data available at the time. Criticisms that Vioxx and Celebrex should not have been approved are more accurately aimed at FDA systematic procedure and not the judgment of these experts. Perhaps if other procedures were in place, including the requirement of more, longer, or larger studies, the experts’ judgment would have rested on a more complete data collection, which could have influenced them differently. The question is how much should be required, and this is a criticism of the system, not the specific approval of Vioxx and Celebrex. The experts spoke based on the available and required data, and their answer was approval.


It is important to generally understand the basics of the critical studies that have been performed on Vioxx and Celebrex because they are the basis for much of the controversy. Each study is different in kind, size, procedure, etc.; therefore, although it is useful to have an overview of these studies, comparisons should be made with caution. It is also important to note that no matter how well planned, each study has limitations. The result of some of these studies, particularly VIGOR, APPROVe, and APC have caused huge debates due to their implications regarding increased heart risk but also due to their limitations that shed doubt on their significance. No one study is conclusive, but familiarity with the studies will help clarify the issues and bring awareness to inconsistencies and shortcomings.


The following studies on Vioxx are summarized in chronological order. Each summary gives a general overview of the studies, including, the purposes, methods, results, and limitations.

1. Vioxx Gastrointestinal Outcomes Research (VIGOR)

Purpose :

VIGOR was a clinical trial attempting to assess whether rofecoxib “would be associated with a lower incidence of clinically important upper GI event than is the nonselective NSAID naproxen among patients with rheumatoid arthritis.”[69]

Methods :

The method used was randomly assigning 8076 patients with rheumatoid arthritis who were “at least 50 years of age (or at least 40 years of age and receiving long-term glucocorticoid therapy) to receive either 50 mg of rofecoxib daily or 500 mg of naproxen twice daily.”[70] The primary end point was confirmed upper GI events such as perforations, bleedings, and ulcers.[71]

Results :

The results of the trial showed that rofecoxib and naproxen had similar efficacy against rheumatoid arthritis.[72] During a median follow-up of 9 months, there was about a 50% lower rate of confirmed GI events that occurred with rofecoxib than with naproxen.[73] However, the incidence of myocardial infarction (heart attack) was much lower among patients in the naproxen group than among the rofecoxib group (0.1 percent versus 0.4 percent with 95 percent confidence interval for difference).[74] Both groups faced similar overall mortality rate and rate of death from cardiovascular (CV) causes.[75]

Dr. Bruce Psaty stated before a Congressional committee that there were additional results for the cardiovascular events from the VIGOR study, which were reported to the FDA but not to the New England Journal of Medicine. These results included the following: for combined outcome of all CV deaths, heart attacks, and strokes, occurrences for patients on rofecoxib and patients on naproxen were 1.30 versus 0.67 events per 100 person years, and for heart attack, the rate for patients on rofecoxib was five times higher with patients on naproxen.[76]

Limitations :

A noted limitation of this trial is that certain patients were excluded from the study, particularly those who required or had been receiving treatment with aspirin. Aspirin users were excluded because aspirin is a known selective COX-1 inhibitor, which may contribute to increased rates of GI bleeding, and thus would have confounded the results of the study. However, the exclusion of patients on low dose aspirin may have influenced CV events in the study because low dose aspirin has been shown to reduce CV risk.[77]

Thus, a possible result from the omission of aspirin users is that instead of rofecoxib having an adverse effect on heart attacks, naproxen may have had a beneficial effect on it. Therefore, patients taking rofecoxib with aspirin therapy may have faced lower CV risks. According to Dr. Peter Kim of Merck, “Because naproxen inhibits platelet aggregation in similar fashion to low-dose aspirin, we concluded that the VIGOR results were most likely due to the effects of naproxen.”[78]

If naproxen could have had a beneficial effect on heart attacks instead of rofecoxib having a deleterious effect, why not take naproxen out of the equation and instead use a placebo? Says Dr. Sandra Kweder of the FDA, to do so would have meant that patients with rheumatoid arthritis would have been randomized to receive no pain relief and that would have been intolerable.[79] Many agree that using a placebo would have been unethical. However, without the placebo, it is difficult to thoroughly analyze the increased CV risk from the study.

Dr. Topol, a prominent cardiologist, has criticized Merck for never publishing another trial conducted by Merck, known as Study 090, which showed significantly higher rates of heart attacks ands strokes among patients treated with rofecoxib compared with patients treated with controls.[80] According to Topol, not only was Study 090 unpublished and only available through FDA memorandum, the VIGOR trial, as presented in the New England Journal of Medicine, suffered from omission of data, erroneous information, and a lack of completeness.[81] He claimed that the data presented to the FDA (six weeks prior to publication of VIGOR article) and the data presented to the journal were different. Among these discrepancies were: actual deaths were not reported even though the article stated that the mortality rate was similar in the two groups, the heart attack rate for rofecoxib was erroneous, and more than half of the thrombotic events were not presented.[82]

Aspirin Debate and the Aftermath of VIGOR :

Merck attempted to solve the puzzle relating to concomitant aspirin use with rofecoxib. In an internal study, known as Study 136, the efficacy of rofecoxib with the concurrent use aspirin was tested with patients treated with a placebo, low-dose aspirin alone, low-dose aspirin and Vioxx, or ibuprofen alone.[83] An endoscope was then used to examine ulcer formation after 12 weeks of treatment.[84] In a November 2001 report, Merck had expected to find that patients on low-dose aspirin and rofecoxib would have a lower rate of ulcer formation than patients on ibuprofen alone. However the study results indicated that the rate was the same for both groups.[85] Merck notes that a large limitation of the study was that fact that there was no comparison between low-dose aspirin and rofecoxib and low-dose aspirin and ibuprofen, due to ethical considerations. A different study had indicated that ibuprofen disrupts the anti-clotting properties of aspirin, meaning that patients taking both aspirin and ibuprofen would not receive the protective cardiovascular benefits of aspirin, and Merck decided that it would not run a study that would possibly put patients at such a risk.[86]

2. Cleveland Clinic Study

Purpose :

This purpose of this retrospective study was to determine whether COX-2 inhibitors are associated with higher or lower rates of CV events by assessing randomized trials that have been conducted on COX-2 inhibitors.[87]

Methods :

The research team searched MEDLINE, the World Wide Web, and relevant submissions to the FDA to identify all published, English-language, randomized, double-blind trials of COX-2 inhibitors published between 1998 and February 2001.[88] The search generated two major trials, VIGOR and Celecoxib Long-term Arthritis Safety Study (CLASS)[89] and two smaller trials, Study 085 and Study 090.[90]

Results :

The VIGOR study indicated a higher risk of developing a confirmed adjudicated thrombotic cardiovascular event, (such as heart attack, unstable angina, resuscitated cardiac arrest, sudden or unexplained death, and ischemic stroke) with rofecoxib treatment compared with naproxen, of 2.38 (95% confidence interval, 1.39-4.00; P =.002).[91] However, the results from the CLASS trial showed that there was no significant difference in CV events between celecoxib and NSAIDS.[92] Cleveland Clinic investigators note that this lower CV rate could be due to the use of low-dose aspirin allowed in CLASS or due to the pharmacological differences in the NSAID agents used as controls in CLASS and VIGOR.[93]

Study 085 was a small, randomized, double-blind, placebo-controlled trial (n=1042) comparing the efficacy and safety of rofecoxib (12.5 mg per day), nabumetone (1000 mg per day), and placebo after six weeks of treatment for osteoarthritis of the knee.[94] Aspirin use for cardio protection was permitted.[95] Of the three total CV events in the trial, 1 event (0.2%) was from the rofecoxib group, 2 events (0.4%) were from the nabumetone group, and there were no events in the placebo group.[96]

Study 090 was a small randomized, double-blind, placebo-controlled trial (n=978) comparing the efficacy and safety of rofecoxib (12.5 mg per day), nabumetone (1000 mg per day), and placebo in patients with osteoarthritis of the knee.[97] Aspirin use for cardio protection was permitted.[98] Out of the nine serious CV events in the trial, 6 (1.5%) events were from the rofecoxib group, 2 (0.5%) events were from the nabumetone group, and 1 (0.5%) event was from the placebo group.[99]

Compared to the heart attack rate of the placebo group (0.52%) in the meta-analysis of the US Physicians’ Health Study, the UK Doctors Study, the Thrombosis Prevention Trial, and the Hypertension Optimal Treatment trials (combined n=48,540), the annualized heart rate for both the VIGOR and CLASS trials were higher, 0.74% with rofecoxib and 0.8% with celecoxib, both significant.[100]

The results of a meta-analysis of four aspirin primary prevention trials with a placebo group with similar cardiac risk factors as the patients in VIGOR suggest that the increased CV rates in VIGOR were due to a prothrombotic effect and not merely a failure to offer the cardio protection of other NSAIDs, like naproxen.[101] The authors acknowledge that comparisons of patient populations in different trials present problems and also that rheumatoid arthritis increases the risks of heart attack, again, making comparisons problematic.[102]

Study 085 and 090 did not demonstrate the CV rate of VIGOR, but these studies were very small, used only 25% of the dose of rofecoxib used in VIGOR, and had very few events for comparison, suggesting that the prothrombotic effect of rofecoxib may be dose dependent.[103] In addition, the comparisons suggest that low-dose aspirin may negate the GI benefits of COX-2 inhibitors, and evidence shows that adverse GI events from aspirin are not dose related.[104]

Limitations :

The authors of the Cleveland Clinic Study acknowledge many limitations of the study. One such limitation is the uncertainty intrinsic to post hoc analysis. Patient populations were heterogeneous, patients with rheumatoid arthritis have a higher risk of heart attack, and only continuous use of COX-2 inhibitors were examined, even though the most frequent pattern of use is intermittent use.[105] Furthermore, definitive evidence of the adverse effect of concomitant use of aspirin would require a prospective randomized clinical trial.[106]

These limitations have been echoed by other commentators. Emphasis is put on fact that the investigators were comparing studies, a technique that inherently has flaws, coupled with the fact that the individual studies themselves contain flaws. The fundamental problems of comparing different patient populations in different studies remain the biggest criticism of the study.[107]

3. Kaiser Study

Purpose :

The purpose of this observational study was to investigate the CV risk of rofecoxib and celecoxib and a variety of non-selective traditional NSAIDs and particularly whether the risk of serious coronary heart disease was increased in patients taking rofecoxib and whether the use of naproxen is protective against this increased risk of heart disease.[108]

Methods :

Dr. David Graham, the principle investigator of this study, used Kaiser Permanente’s California patient database (of six million patients) to conduct the observational study.[109] From January 1, 1999 to December 31, 2001, all patients (ages 18 to 84) who filled at least one prescription for a COX-2 selective or non-selective NSAID were identified for the study.[110] Patients were entered into the study cohort if they had no diagnoses of certain diseases during the screening interval. Cohort members were followed from this entry date until the end of the study period or until the occurrence of a heart attack or death.[111]

Results :

At different times during the study, most patients were exposed to a variety of different NSAIDs including celecoxib, ibuprofen, naproxen, and rofecoxib. Altogether, there were 8,199 incident cardiac events.[112]

There was an increase risk of serious coronary heart disease with rofecoxib use compared to remote NSAID use (1.40 fold increase) and celecoxib use (1.63 fold increase).[113] There was an increased risk of hospitalized acute heart attack and sudden cardiac death (SCD) with high-dose rofecoxib compared to remote use of an NSAID (3.15 fold increase).[114] Still, there are many other comparisons between the CV relative risks among the drugs.

The report submitted by Graham states that the data suggests that: (1) Serious coronary heart disease is increased in patients treated with rofecoxib compared with celecoxib, (2) CV risk is increased in patients treated with high-dose rofecoxib compared with remote use of any NSAID, and (3) Naproxen is not protective against serious coronary heart disease, but may actually confer an increase in risk, which contradicts the cardio protective theory of naproxen in the VIGOR study.[115] In addition, Graham states that three cohort studies show no such protective effect, and three other case-controlled studies, which showed that naproxen had a protective effect against acute heart attack, were flawed.[116]

Graham concludes that rofecoxib use increases the risk of serious coronary heart disease (including heart attack and sudden cardiac death). High-dose rofecoxib use had an increased risk of 3.7 fold and standard dose rofecoxib use had an increased risk of 1.5 fold compared to celecoxib use.[117] Furthermore, the 5-fold increase risk shown in the VIGOR trial cannot be attributed to a protective effect of naproxen because this study and three other cohort studies do not show the effect.[118]

Limitations :

According to Dr. Graham, the most important limitation was that because use of high-dose rofecoxib was low within the population studied, there resulted a relatively small number of exposed cases. Only about 7.4% of all rofecoxib use in the study was high-dose use compared with 17.4% in the US. Another limitation is that medical record review and case validation was not performed.[119]

An extensive critique by Aviel Shatz points out other flaws and limitations of the Kaiser Study. First, according to Shatz, the data source for the study was from Kaiser Permanente, an organization that excluded COX-2 inhibitors from its formulary completely, so patients taking COX-2 inhibitors were uniquely stratified.[120] These “patients had a significantly higher baseline for CV risk than regular patients, who were prescribed on lower cost products in the formulary,” says Shatz. [121]  

Secondly, Graham identified the potential for a significant increase in heart attack and sudden cardiac death based on data collected from only 10 patients on the Study Group, compared to only 8 patients on the Control Group.[122] According to Graham’s data, there was a CV event (any CV event from a broad spectrum) in 3 patients from the 10 Study Group, compared to only 2 patients among the 8 Control Group patients. “Based on this statistically inadequate population,” says Shatz, “Graham calculated an adjusted risk of 3.15, with a confidence interval of 1.14-8.75.”[123]  

Thirdly, even though Graham acknowledged that his conclusion, that CV risk was decreased with celecoxib and increased with standard-dose rofecoxib was not significantwhen compared to remote exposure , he used the risk calculated on 10 Study patients and the 8 Control patients to project 12,940 cases attributed to high dosage rofecoxib and then used this ratio (and the discovery that celecoxib reduced the risk of CV event (OR 0.86)) to project 14,845 at the standard dose.[124]

Finally, Graham’s report published on the FDA’s website includes Table 7, which is the same table as Table 3 (Risk of AMI and SCD) on Graham’s presentation in France, but with the Control column omitted. The omission of this column hides the fact that only 10 Study patients were compared to 8 Control patients.[125]

4. Adenomatous Polyp Prevention on Vioxx (APPROVe)

Purpose :

Because precancerous tissues also produce COX enzymes, researchers conjectured that inhibiting these enzymes might stop or slow cancer growth, and this supposition has led to many studies currently being conducted to assess whether COX-2 drugs can treat or prevent various cancers.[126] APPROVe is one such trial. The APPROVe trial was designed for the purpose of measuring the efficacy of Vioxx 25 mg in preventing the recurrence of colorectal polyps in patients with history of colorectal adenomas (benign tumors in the large intestine).[127] The trial was terminated two months before its official date of completion, and therefore no report was submitted on the trial results or on the efficacy of Vioxx in reducing cancer risk.[128]

Methods :

APPROVe was a randomized trial of 2586 patients (62% male, age = 40) receiving either rofecoxib (25 mg per day) or placebo (stratified by level of use of low-dose aspirin) within 12 weeks of surgical resection of colectoral adenomas.[129] The primary endpoint was the recurrence of colorectal polyps after three years of treatment.[130] Initially, patients taking low-dose aspirin were excluded, but after the results of VIGOR illustrated the importance of testing users of aspirin, the protocol was amended to permit up to 20% of patients taking low-dose aspirin into the study.[131]

CV events were analyzed by specific outcomes, either thrombotic events (such as sudden cardiac death, fatal and non-fatal heart attack, unstable angina, fatal and non-fatal ischemic stroke, and pulmonary embolism) or a composite endpoint (any CVD death, any unknown death, any non-fatal heart attack, and any non-fatal stroke.)[132]

The baseline characteristics of patients (including hypertension and diabetes) were balanced across the two groups, but during the trial, a significantly higher proportion of patients taking rofecoxib required antihyerpensives.[133]

Results :

Patients treated with rofecoxib had a statistically higher incidence of cardiac events and total thrombotic events compared to patients treated with placebo, but differences were not detected until after 18 months of trial.[134] The rate of heart attack and stroke was 1.9% for the placebo group after 18 months of trial compared with 3.5% for the rofecoxib group, indicating almost a 2-fold risk with rofecoxib after 18 months of therapy.[135] There were no significant interactions identified in subgroup analysis according to the baseline characteristics (including use of low-dose aspirin).[136]


Although the APPROVe results indicated that there were no significant interactions detected in subgroup analysis on the use of low-dose aspirin, there has been criticism that the report does not provide any detail on the effect of aspirin (and other NSAIDs) on the study and control CV occurrences.[137] Furthermore, since subjects were post surgery patients, those on placebo could have consumed more aspirin and other NSAIDs than study group patients if pains persisted after taking placebo. This could distort results as aspirin and NSAIDs may have an effect of lowering CV risks, and this should be taken into account in the relative risk (RO) calculation.[138]

Also, CV events as defined in the study were wide in range of categorization, variety and severity. Mortality, which is a precise statistical measurement, was the same among the different treatment groups.[139] Furthermore, among a predominantly male population (62%), the incidence of CV events below 2% is “not abnormal” and variation within that range may be due to chance.[140]

Odd ration (OR) statistical analysis methodology was applied for the risk assessment of CV events in the study, but caution should be used because the events were of low occurrence and the patient population was small.[141]

Dr. Robert Bresalier, an investigator of the APPROVe trial, reported that there were no difference in risk between the two groups for the first 18 months, and the risk was then increased by an unknown trigger.[142] Although it is still undetermined why the risk was not present in the first 18 months, it is verifiable that less then 2.5% of patients taking Vioxx in the United States were consuming Vioxx for less than 18 months, which means that majority of patients taking Vioxx is not affected by the trigger.[143]


The following studies on Celebrex are summarized in chronological order. Each summary gives a general overview of the studies, including, the purposes, methods, results, and limitations.

1. Celecoxib Long-Term Arthritis Safety Study (CLASS)

Purpose :

The trial was designed for the purpose of determining whether of celecoxib is associated with a lower incidence of significant upper GI toxic effects and other adverse effects compared to other traditional non-selective NSAIDs.[144] Serious ulcer complications were defined as bleeding, perforation, or obstructions (POB).[145]

Methods :

The study was a double-blind, randomized, controlled, six-month trial[146] conducted in over three hundred eighty-six clinical sites and included 8059 patients (at least 18 years of age) with osteoarthritis (OA) or rheumatoid arthritis (RA), of which 7968 patients received at least 1 dose of study drug, and 4573 patients (57%) received treatment for six months.[147] The study compared the use of celecoxib, 400 mg twice per day (2 and 4 times the maximum RA and OA FDA approved effective dosages, respectively; n = 3987); ibuprofen, 800 mg 3 times per day (n = 1985), and or diclofenac, 75 mg twice per day (n = 1996).[148] Patients using aspirin for cardiovascular prophylaxis were allowed to participate in the study.[149]

Results :

The annualized incidence of POBs in celecoxib-treated patients was 0.76% versus an incidence of 1.45% for patients taking NSAIDs (P =.09).[150] While the rates favored celecoxib, it did not reach statistical significance. However, when the incidence of symptomatic ulcers was combined with POBs, there was a statistical difference (2.08% and 3.54%, respectively, P =.02).[151]

Patients taking low dose aspirin had similar rates of POBs and symptomatic ulcers in both groups.[152] Corticosteroids did not enhance the risk for POBs.[153] The overall incidence of renal adverse effects and hypertension in particular, were significantly lower in patients treated with celecoxib than in patients treated with NSAIDs.[154]

The incidence of cardiovascular adverse events and heart attack in particular was similar in both treatment groups.[155] For patients not taking aspirin, incidence of heart attack in patients treated with celecoxib was less than in patients treated with NSAIDs.[156]

The withdrawal rate for adverse events and lack of therapeutic efficacy was less in the celecoxib treatment group compared to the NSAID treatment group.[157]

Limitations :

Although this is the first study to query and demonstrate lower rates of serious ulcer complications in patients taking COX-2 compared to patients taking NSAIDs,[158] the difference was not statistically significant and only became statistically significant when the incidence of symptomatic ulcers was added to POBs. A longer study with a larger population of subjects would likely have produced a difference that would have been significant.[159]

GI rates may have been affected because patients using aspirin were permitted into the study, since low-dose aspirin taken concurrently with celecoxib may have negated the protective effect of celecoxib. The data supports this supposition because the lower rates of POBs and POBs plus symptomatic ulcers were both statistically significant in celecoxib patients not concomitantly on low-dose aspirin.[160] This effect can be explained because aspirin is a selective COX-1 inhibitor, which causes stomach irritation and bleeding.[161] Also, COX-1 generated prostaglandins is thought to be gastric protective.[162] Thus, it is important to note that the use of selective COX-2 inhibitors may offer no GI benefit for elderly individuals since they are more likely to suffer from serious ulcer complication but are also more likely to be using low dose aspirin.[163]

2. Adenoma Prevention with Celecoxib (APC)

Purpose :

The APC trial was designed for the purpose of measuring the efficacy and safety of celecoxib (200 mg twice daily and 400 mg twice daily) compared to placebo in preventing recurrence of colorectal polyps in the colon and rectum one year and three years after endoscopic polypectomy with patients with a history of colectoral neoplasia.[164]

Methods :

There were 2035 patients with a history of colorectal neoplasia in the study, and a detailed medical history, including a baseline assessment of CV disease status was obtained for each patient.[165] Baseline characteristics were balanced among the groups.[166] Stratification was utilized according to the use and nonuse of aspirin in the randomization of patients.[167] To evaluate CV risk, a hierarchy of composite end points was defined, including death from CV causes, heart failure, heart attack, and stroke.[168]

Results :

There was a dose-related increase in composite endpoint of CVD death, non-fatal heart attack, stroke, and heart failure in patients treated with continuous use of celecoxib compared to placebo.[169] The concurrent use of aspirin did not affect the measured effect.[170] A composite cardiovascular end point of death from CV causes, heart attack, stroke, or heart failure was reached in 7 of 679 patients in the placebo group (1.0%), as compared with 16 of 685 patients receiving 200 mg of celecoxib twice daily (2.3%), and with 23 of 671 patients receiving 400 mg of celecoxib twice daily (3.4% ).[171] Due to these results, and a similar trend in results for other composite end points, the data and safety monitoring board recommended termination of the use of celecoxib in the study.[172]

Limitations :

The investigators of ACP study note that their findings on CV risk must be read with caution since such risk was based on a small number of events, the study was not specifically designed or statistically powered to evaluate CV risk, and some events may have been unreported.[173]

Also noted by the investigators is that although the finding of increased CV risk in the study is consistent with the findings in the APPROVe study (of rofecoxib use), it contradicts findings of other studies, including CLASS and preSAP.[174] Although CLASS might be distinguished because of the participation of low-risk patients[175] , preSAP and APC used an identical analysis to assess CV risk and was conducted by the same independent review board on safety.[176] One explanation for the differing results between the two trials may be related to the dosing regimen since preSAP only used one dose of celecoxib while ACP called for two daily doses, which supports the hypothesis that the sustained inhibition of prostacyclin may contribute to CV risks.[177] Of course, other differences in patient populations and medical histories could also have contributed to the difference.[178]

3. Prevention of Spontaneous Adenomatous Polyps (preSAP)

Purpose :

The preSAP study compared celecoxib 400 mg once a day (n=933) versus placebo (n=628) and was similar in size and duration to the APC trial.[179] An identical analysis was used to assess cardiovascular risk in the APC trial, which was also conducted by the same independent safety review board.[180] The 3,600 patients enrolled in APC and preSAP were followed for over a five year period, and some patients participated for more than four years.[181] Pfizer estimates that two years of treatment have been completed by about 2,400 patients assessed in the cardiovascular analysis.[182]

Results :

There has been no increased risk for patients taking 400mg of celecoxib daily compared with those taking placebo.[183] There were 11 major events including, CV death, heart attack, and stroke, (1.8%) in the placebo group and 16 events (1.7%) in the celecoxib group.[184] The dosing of preSAP was suspended based on the observations of the increased CV risk in the APC trial.[185]

4. Alzheimer's Disease Anti-inflammatory Prevention Trial (ADAPT)

Purpose :

The ADAPT study was designed to assess the potential benefit of long-term use of naproxen and celecoxib “in decreasing the risk of developing Alzheimer’s disease in people 70 years of age or older who were considered to be at increased risk because of family history, but did not have symptoms of the disease.”[186]


About 2,400 patients were randomly assigned to received naproxen (220 mg twice a day), celecoxib (200 mg twice a day), or placebo for up to three years.[187]


No significant increase in risk for celecoxib was found, but the use of celecoxib and naproxen were suspended in part because of the findings of increased CV risk from the APC trial, [188] as well as an indication in this trial of an increase in CV events among patients taking naproxen when compared with those taking placebo.[189]


What do all these studies actually add up to? The summaries above are only those studies most cited, but there are dozens more for both Vioxx and Celebrex. Common questions that come up include: how are the studies different? Why do some studies show certain risks while others do not? Are some more reliable than others? And taken together, what do they say about the safety of the drugs? While it is difficult to give very precise answers to most of these questions, a better understanding can certainly be reached by looking at the studies separately and together.


There is a long history of a great divide among scientists in the medical field between those who believe that only clinical studies have value and those who believe that observational studies are also important. Clinical studies, which are generally randomized, double-blind, and controlled, are known as the gold standard of tests, chiefly because it is easier to attribute an observed effect to the treatments being compared.[190] Randomization means that patients are grouped based on their baseline risk for the health outcome of interest so that on average this baseline risk will be evenly distributed across the groups.[191] In observational studies, investigators use a patient base that is part of the “real world”, that is, patients and their physicians select the medication, so the factors that lead to the selection of medication may be potential biases.[192] This means that outcomes could be due to differences among patients or specifically, differences between the baseline risks among patients, rather than by the treatments alone. While the traditional view has caused a hierarchy of studies with randomized controlled studies at the top, there has been a new voice calling out for an elevation of the value of observational studies. An article in the New England Journal of Medicine is well-cited for its comparison of various types of studies. Investigators in the study conducted a meta-analysis of different types of studies and concluded that there were no major differences between the estimate of treatment effects in the randomized controlled studies and observational studies, noting that this may have been due to improvements in observational studies in terms of comparable subjects and better design.[193]

The difference between these types of studies is critical to an evaluation of Vioxx because drug companies, including Merck, are often quick to dismiss results of non-clinical studies that show increased risks. In October 2002, Dr. Wayne Ray revealed the results of an epidemiological[194] study of Medicaid patients in Tennessee who were taking high doses of Vioxx [195] , which showed a significant increase in heart attacks and strokes compared to patients not taking high doses of Vioxx.[196] Merck maintained that it remained confident of the safety of Vioxx because it considered such epidemiological studies less reliable than clinical trials.[197] Similarly, the Kaiser study summarized above is also an observational study, and it is likely that Merck did not put much weight on the results of the study, which also indicated an increase heart risk. As noted above, one limitation of this study and, an inherent problem with observational studies in general, is that the baseline risks of patients being compared were different. The subjects in the study were all patients of Kaiser Permanente, which excluded COX-2 inhibitors from its formulary so patients prescribed were uniquely stratified. In addition, they also had a “significantly higher baseline for CV risk than regular patients prescribed on lower cost products in the formulary .”[198]

Along the same vein, the value of meta-analyses of studies has also been questioned. A meta-analysis is usually set up by identifying all studies that examined the same clinical topic. The purpose is to get a general overview of all studies that have been conducted on that topic. For example, the Cleveland Clinic trial summarized above is an example of a meta-analysis looking at randomized trials conducted on COX-2 inhibitors in order to determine whether COX-2 inhibitors are associated with a higher rate of CV events.[199] Again, Merck was likely not shaken by the negative results of such a study because of the limitations that even the authors of the study acknowledge—the uncertainty intrinsic to post hoc analysis, including comparing heterogeneous patient populations with very different baseline risks as well as different periods of use, different doses of Vioxx, and different concomitant uses of other drugs.[200]

This is why Merck took much more seriously the results of VIGOR and APPROVe, two randomized clinical trials that showed an increased heart risk with Vioxx treatment. But as noted above, and as we will see, clinical trials are far from perfect, and these studies were no exception.


An ideal study would have a patient database of millions of subjects, which would nearly eliminate the effect of events happening by chance instead of being caused by the treatment. Unfortunately, there are not enough resources to conduct such studies before a drug is approved or there would be very little incentive for drug companies to create new drugs. Keeping track of the effect of drugs in observational studies is probably the best way to study large patient populations, such as the Kaiser Study, which used Kaiser Permanente’s Northern California database of six million patients.[201] But of course, there are too many limitations with such observational studies. To get more precise (and more controlled) data, we have to turn to clinical trials, but clinical trials also come in many shapes and sizes. What we end up with is a series of randomized controlled trials that are testing different things (efficacy, safety, etc.) and using different patient populations, doses of medication, time periods, etc. The result is a web that must slowly be untangled.

1. Concomitant Use of Aspirin

One of the most controversial issues relating to the studies was whether to allow low-dose aspirin users into the studies, which underscores the difficulty of testing due to the reactions caused by other drugs. The aspirin issue came to the forefront in the aftermath of VIGOR. Merck was able to contend that that the increased heart risks that resulted from VIGOR may have been due to the protective properties of naproxen, instead of Vioxx having a deleterious effect, since aspirin use, which is known to be cardio protective, was not permitted in the study.[202] The reason why aspirin users were not allowed in the study will illustrate why it is difficult to construct clinical trials perfectly. Aspirin was not allowed in the study because it is a COX-1 inhibitor, which is known to increase GI bleedings, which would confound the results of a study testing to see whether adverse GI events would decrease with Vioxx treatment.[203] But then not having aspiring ended up confounding the study as well because aspirin is known to decrease CV events, and since naproxen inhibits platelet aggregation in similar fashion to low-dose aspirin, the increased CV events seen with Vioxx treatment may simply be due to the protective effects of naproxen.[204] So, why not compare Vioxx (with no aspirin) to a placebo, which would not have the cardiac protective effects of naproxen? Because the subjects were people with rheumatoid arthritis (an indicated use of Vioxx), and giving such patients s a placebo that would not relieve their pain is unethical.[205] Consistent with Merck’s conjecture of aspirin confounding the results of GI events, the CLASS study summarized above may also have been affected in that same manner. In CLASS, patient use of low-dose aspirin may have negated the GI protective effect of Celebrex, and the data supports this supposition since lower rates of POBs and POBs plus symptomatic ulcers were both statistically significant in patients taking Celebrex and not concomitantly using aspirin.[206] With concomitant aspirin use, the difference was only statistically significant when the number of symptomatic ulcers were added to POBs.[207] However, in the end, the results beg a harder question: if concomitant use of aspirin negates the GI benefits and the certain patient populations, such as the elderly, are more likely to suffer from adverse GI events but also need aspirin for cardio benefits[208] , then the results do matter. In other words, even if aspirin confounded the results—so what? What good is the efficacy of a drug in isolation when the public will be using it with other drugs?

As summarized above in Study 136, Merck did do a comparison between Vioxx with low-dose aspirin and ibuprofen alone.[209] The results did not come out as Merck had hoped, and adverse GI events for both were similar.[210] Again, the perfect clinical trial could not be designed because Merck couldn’t compare Vioxx with low-dose aspirin with ibuprofen and low-dose aspirin since ibuprofen disrupts the anti-clotting process of aspirin, meaning patients would not receive the cardio protection of aspirin, which again, would be unethical.[211]

2. Minor Differences Matter

So far, we have been examining studies set up differently. Obviously, these differences may affect the results of such studies. What about studies set up similarly but end up with different results? In this situation, looking at even minor design differences may help. Such was the case with APC and PreSAP, two clinical studies testing the efficacy of celecoxib in preventing the recurrence of colorectal polyps in patients with a history of colectoral neoplasia.[212] Identical analyses were used to assess cardiovascular risk in the both trials, which were also conducted by the same independent safety review board.[213] However, the results of APC showed an increased heart risk with Celebrex compared to placebo while PreSAP showed no difference between the two groups.[214] Alas, the most plausible explanation for the difference actually stems from a difference in dosing regimen, in otherwise very similar trials. PreSap patients used only one dose of Celebrex, which supports the hypothesis that the sustained inhibition of prostacyclin[215] may contribute to heart risks.[216] But even if the dosing regimen were the same, the different results may be explained by the difference in patient population and the medical history of the patients.[217]

3. Duration of Studies

There is obvious value in looking at the long-term effects of a drug, even if the drug is not explicitly recommended for such long-term use. The studies summarized above all vary in the number of patients and length of study, which may have affected the results. An editorial comment from John Hopkins Division of Rheumatology online notes that prior to the APC and APPROVe studies, attempts to assess the association between COX-2 inhibitor and heart risks have compared COX-2 inhibitors to non-selective NSAIDs in patients with osteoarthritis and rheumatoid arthritis for twelve or less months of treatment.[218] The APC and APPROVe studies, which showed increases in heart risks with Celebrex and Vioxx treatment respectively, bring an important aspect into the evaluation because they provide for placebo controls and a “much longer duration of treatment”.[219] The CLASS trial showed no increased heart risk with Celebrex, but the trial was only six months long.[220] It took much longer than that for APC to show an increased risk and eighteen months for APPROVe to show an increased risk. VIGOR, another long-term Vioxx study, also showed an increase heart risk. On the other hand, preSAP, like APC, was a long-term study, but did not show an increased heart risk. What does this mean? Drawing a precise conclusion based on the length of the study would be difficult, but it does help to note that all three studies showing an increased heart risk, VIGOR, APPROVe, and APC were relatively long-term studies. And although Celebrex seems safer than Vioxx, so far APPROVe, the study that led Vioxx’s withdrawal, is longer that any single, published study of Celebrex.[221]

The recent results of ADAPT, a relatively long-term study, has actually brought out an interesting development to issue of long-term versus short-term studies and the safety of drugs. Researchers were shocked that the results revealed that naproxen was associated with a 50% increase in heart attacks or strokes compared with placebo whereas Celebrex was associated with no such increases.[222] Part of the shock that followed these results was due to the fact that naproxen was approved by the FDA in the mid 1970s and has for awhile been available over the counter.[223] Kweder of the FDA has dubbed the situation as confusing since this is the first time evidence has indicated an increase in heart risk with naproxen and surmises that it may be due in part to the fact that long-term safety studies on naproxen have not been conducted.[224] After this discovery, APAPT investigators articulated that perhaps increased heart risks may be an issue similar to all NSAIDs but that they’ve never examined the issue before.[225] So it turns out that naproxen, a drug that has thought to be safe for decades, may pose similar alleged heart risks as COX-2 inhibitors. This suggests that Vioxx and Celebrex may have been under the heat simply because they have been conducting long-term studies, something that perhaps all drugs, especially those that have been out there for awhile should undergo, given the new found risks associated with naproxen.

4. Dose

Dr. Gregg Fonarow, Professor of Cardiovascular Medicine at UCLA has supported the safety of COX-2 inhibitors if taken at standard doses and standard durations, reminding all that NSAIDs were approved as "short-term treatments for pain relief, not as drugs to be used on a daily basis at high doses."[226] In fact, the recommended daily doses for Vioxx and Celebrex are quite low, with the maximum daily dose of Celebrex not to exceed 400mg (except for prevention of polyps) and 50mg for Vioxx (not to exceed five days at this dose, otherwise 25mg per day).[227]

So far, CV risk seems to be dose-related. In the APC trial, patients taking 400mg of Celebrex daily had a 2.5 fold increased risk of CV event compared to placebo, and patients taking 800mg of Celebrex daily had a 3.4-fold increased risk compared to placebo.[228] A dose of 800mg is high but was used to test whether inflammation is a cause of cancer.[229] Also, patients in the trial were taking Celebrex for a lengthy average of 33 months.[230] In addition, dosing regimen may also have an effect on CV risk. One explanation for the differing results between the very similar APC and preSAP trials is due to the different dosing regimens since preSAP only used one dose of celecoxib while ACP called for two daily doses, which supports the hypothesis that the sustained inhibition of prostacyclin (from the two doses) may contribute to CV risks.[231] Similarly, a daily dose of 50mg was used in VIGOR, a long-term study, even though a dose of 50mg was only recommended as the maximum daily dosage for certain types of acute pain.[232] Even then, doses at 50mg were not to exceed five days or five migraine treatments per month.[233] Chronic use of Vioxx was not recommended.[234]

Of course, even though many of the doses tested were higher than the recommended doses, as in the issue of duration, people often don’t take the advice of the recommended dose and duration. This is why it is still valuable to do long-term studies and studies with higher drug doses. However, the proper caution (that doses in the study are higher than recommended) should be acknowledged so as to not deter drug companies from conducting such beneficial studies.


In the end, as good as a study may be set up, it cannot be perfect. The results of a subgroup of people will never translate perfectly into the real world. Differences between studies are wide, including the type of study, what is being studied, the number of patients, the medical history of patients, concomitant uses of other drugs, duration of studies, doses of treatment, use of placebo, use of comparators, etc. Sometimes these differences will not matter, and sometimes they can lead to contradictory results, which may be of no consequence or can shed light on new matters. The case of Vioxx and Celebrex is no exception. The studies summarized above illustrate the wide range of studies conducted on each drug with varying results that may indicate risk or may be the result of a flawed study design. Taken together, they probably do suggest some likelihood of an increased heart risk. But all drugs have adverse effects. The important factor is to determine the likelihood of the risk. So far, both drugs have been shown to increase heart risk after continuous long-term use at certain doses, some above the recommended level. When dealing with such immense scientific data, the decision of whether the risk is sufficient to warrant action should be made by those with the expertise and capabilities to make sound scientific decisions. Although Merck and Pfizer arguably have some of the most talented researchers in the world, the decision should of course, come from a neutral third party with the public interest in mind.


The big question on the minds of the many spectators of the Vioxx and Celebrex saga is simply, “Why did Merck pull Vioxx off the shelves while Pfizer was content with simply pulling advertisements and making a few labeling changes to Celebrex?” These actions are not only different but are arguably polar opposites in the spectrum of actions both companies could have taken. Normative questions come to mind. Had Merck done too much or had Pfizer done too little or both? Was one right and therefore the other wrong? Or were the both actions appropriate in their respective situations?


1. Merck’s Surprise Announcement

On September 30, 2004, Merck announced a voluntary withdrawal of Vioxx from the U.S. and worldwide market.[235] A patient letter from Merck explained the withdrawal as a result of the findings of the cancer study, APPROVe, where an increase of CV events such as heart attack and stroke was found in patients taking 25 mg of Vioxx per day compared with patients on placebo.[236] The letter stated that although the incidence of CV events was low, there was an increased risk after 18 months of treatment.[237] The letter concluded with, “We are taking this action because we believe it best serves the interests of patients. That is why we undertook this trial to better understand the safety profile of VIOXX. And it’s why we instituted this voluntary withdrawal upon learning about these data .”[238]

2. FDA Involvement in the Vioxx Withdrawal

A. Did the FDA Order the Withdrawal of Vioxx?

Had the FDA ordered the withdrawal of Vioxx from the market, its actions would have been quite clear to the public. Because Merck’s withdrawal of Vioxx was voluntary, there is much ambiguity surrounding the role of the FDA in the withdrawal. Several questions come to mind. Would the FDA have ordered the withdrawal? If yes, did Merck voluntary withdraw based on this knowledge? If no, why would Merck withdraw when it was not required to withdraw? On a different note, even if it wanted to, did the FDA have the authority to order such a withdrawal? In other words, was withdrawal warranted?[239]

Merck informed the FDA on September 27, 2004 that the Data Safety Monitoring Board for the APPROVe study had recommended that the study be terminated early because of an increased risk of CV events.[240] Based on this information, Merck and FDA officials met on September 28, 2004, where Merck informed the FDA that it was voluntarily withdrawing Vioxx from the market.[241] Since the meeting between Merck and the FDA was private, little could be known about what took place, and FDA intent in the matter remains unclear. Kweder, of the FDA, remarked that agency officials were stunned when they learned about the withdrawal.[242] She has expressed the view that before ordering withdrawal, the FDA would have had to examine studies of the drug closely and perform a cost benefit analysis.[243] However, Acting FDA Commissioner, Lester Crawford[244] has stated that, “Merck did the right thing by promptly reporting these findings to the FDA and voluntarily withdrawing the product from the market. Although the risk that an individual patient would have a heart attack or stroke related to Vioxx is very small, the study that was halted suggests that, overall, patients taking the drug chronically face twice the risk of a heart attack compared to patients receiving a placebo.”[245] Neither Kweder nor Crawford purport to take a stand on what the FDA would have done, and these arguably contradicting statements reveal that that FDA action is at best unclear.

So why did Merck decide to voluntarily withdraw the drug if the FDA had not ordered it? The possible reasons for this and an analysis of its actions will be discussed subsequently, keeping in mind that Merck was fully aware of the option of re-labeling, a path it previously chose to take after the results of a prior study suggested an increase in CV risk.

B. Could the FDA have Ordered the Withdrawal of Vioxx?

Perhaps the easier question is whether the FDA would have had the authority to order the withdrawal of Vioxx. Under Section 505(e) of the Federal Food Drug and Cosmetic Act (the Act), the Secretary shall, after due notice and opportunity for hearing to the applicant, withdraw approval of an application if the Secretary finds new clinical evidence that shows that the drug is not safe for use under the conditions upon the basis of which the application was approved.[246] The Secretary may also withdraw on the basis of new evidence together with evidence available when the application was approved, which shows that there is a lack of evidence that the drug will have the effect it purports.[247] There are many other bases on which the Secretary can withdraw an application including untrue statements of material fact, failing to establish a system for maintaining records, and labeling that is false or misleading.[248]

Section 505(e) suggests several routes for the FDA to claim authority to withdraw. First, the new clinical data from the APPROVe trial indicated an increased risk of CV events with Vioxx treatment, which the FDA could use to deem the drug unsafe for use under the conditions of which the NDA was approved. That the trial used a dose of 25 mg per day, a dose approved for Vioxx, bolsters a claims of withdrawal, but the fact that the increased risk did not show up until 18 months of continuous use may show that the drug is not unsafe for its specified use, which like most NSAIDs, are approved as “short term treatment for pain relief, not drugs to be used daily at high doses”.[249] Secondly, the FDA can make the claim that the new evidence, together with the pre-approval evidence, shows that there is a lack of evidence that the drug will have the effect it purports, for instance, by pointing to evidence that the GI benefits are negated when aspirin is used concomitantly. After all, it was this potential GI benefit that led to excitement for its approval. However, despite this possible negation of GI benefits, what Vioxx purported was to treat osteoarthritis, rheumatoid arthritis, and primary dysmenorrheal, and its efficacy in treating these conditions have not been disproved or doubted. Finally, the FDA can use its misbranding authority to withdraw the drug. Since CV risks and risks of use with other drugs are not on the label or labeling of the drug, it is false and misleading, which means that the drug is misbranded and can thereby be withdrawn.[250] Of course, labeling can easily be fixed.

C. Why Didn’t the FDA Order the Withdrawal of Vioxx?

It seems fairly certain that the FDA had ample statutory support to make a determination of withdrawal for Vioxx. Furthermore, once the FDA decides to give notice of withdrawal, it is quite improbable that a challenge to its decision would be successful.[251] Given such authority and power, it is the FDA’s inaction that may be of importance in this case. Because an investigational new drug tests very few patients before the NDA is approved, it is almost certain that low frequency adverse actions will be discovered when the drug is made available to the public at large.[252] The safety of Vioxx was tested in about 5000 patients, a relatively large number, before the approval of its NDA[253] , but actual use of Vioxx was considerable larger, somewhere along the lines of two million patients.[254] Therefore, although the rate of CV events suggested in the studies was low, with such a huge patient database, the absolute numbers are very large, and it is not surprising then that an FDA report estimated the number of heart attacks or sudden cardiac deaths caused by Vioxx to be nearly 28,000.[255]

In fact, the General Accounting Office found that of 198 drugs approved by the FDA between 1976-1985, about half had serious post-approval problems, as indicated by labeling changes or market withdrawal.[256] Out of these drugs found with serious adverse reactions in the study, all but six were still marketed in 1998 because the FDA determined that the benefits outweighed the risks.[257]

Given the nature of the drug approval process and the expectancy of such adverse reactions, it is highly possible that the FDA was not yet willing to withdraw Vioxx from the market because of the benefits of the drug. According to Kweder, in order to require withdrawal, the agency would have looked closely at the APPROVe study and perform a cost benefit analysis. “We haven’t done that. We wish he had had the opportunity to do that,” says Kweder.[258] Janet Woodcock, Acting Deputy Commissioner of Operations of the FDA, has also stated that agency response to new risks includes asking the manufacturer to re-label, restrict distribution, or remove the product from the market.[259] Says Woodcock, “our action will depend on the frequency of the reports, the seriousness of the diseases or conditions for which the drug provides a benefit, the availability of alternative therapy, and the consequences of not treating the disease.”[260] So, if it can be inferred from the FDA’s inaction that it recognized these benefits, why did it allow Merck to withdraw Vioxx and why would Merck withdraw despite implied FDA support?

3. Was Merck’s Decision to Withdraw Rational?

One plausible hypothesis to explain why Merck withdrew Vioxx from the market could simply be because it was acting rationally in the shadow of the law. If Merck believed that the FDA would have ordered the withdrawal and would have the authority to do so, it is highly rational for Merck to cut its losses, salvage its reputation, and voluntarily pull the drug off the market. This is in fact, a two-factor analysis. In order to make sense, two factors have to be true, and both are to some extent inter-related. First, Merck has to believe that the FDA was decidedly or at least leaning towards ordering the withdrawal. Secondly, Merck had to believe that the FDA would have the authority to do so. What makes these factors inter-related is that the FDA would have the authority to withdraw the drug if there was merit for the withdrawal, which in turn would make it much more likely that they FDA would be inclined to withdraw the drug. In other words, the FDA probably pursues withdrawals when it is likely to have the authority to do so or when it is likely to win on a challenge to its actions. Under this analysis, Merck would have to believe that there was sufficient merit from the studies for the FDA to assert its authority to withdraw Vioxx from the market. Even if the merit of the FDA’s case was not certain, as long as it is above a certain level, Merck may not have been willing to face the challenge and risk losing, especially when the risk is compounded in terms of reputational costs and the cost of giving tort suit plaintiffs the precedence of victory against Merck. In fact, once the FDA gives sufficient notice of withdrawal, the FDA can declare the withdrawal of the NDA effective even before granting any hearings to Merck to challenge the withdrawal.[261] This puts Merck at a considerable disadvantage, again, in terms of cost, reputation, and tort suit strategies.

In fact, the prospect of winning a challenge of withdrawal against the FDA is bleak. First, although the FDA has to give notice of withdrawal to the holder of the NDA, it can deny a hearing to a holder who is challenging the withdrawal if the holder does not supply the FDA with evidence, which the FDA deems sufficient to raise a question about the FDA’s conclusion that the drug lacked safety or efficacy.[262] In other words, the holder of the NDA does not even get a chance to have a hearing if the FDA unilaterally determines that a hearing would be useless based on the evidence. This concept closely parallels summary judgment (if data is insufficient, party can win on issue it asserts, but will lose case because the evidence will still be insufficient to sustain continued marketing of the drug as safe and efficacious[263] ), except that instead of a third party or judge making that decision, the FDA or the opposing party itself is able to make that determination. Secondly, assuming a hearing is reached, even when there is a battle of the experts in the hearing, the standard for the drug company is prohibitively high—that is, the evidence supporting the drug at the time of withdrawal must be as high as the standard of evidence for the approval of an initial NDA.[264] Furthermore, judicial review is highly favorable to the FDA, and courts are greatly deferential to the expertise of the FDA.[265]

Assuming that Merck is a rational actor, a simple economic model could explain its actions. The likelihood of winning a challenge against the FDA’s order to withdraw is extremely low. Three scenarios come to mind. (1) The FDA decides to order the withdrawal. In this scenario, since it is certain that the FDA is ordering withdrawal and is likely to win, it makes sense for Merck to voluntarily withdraw, saving litigation and reputational costs. (2) The FDA decides not to order the withdrawal. In this scenario, Merck should have confidence that its NDA will not be withdrawn and should go ahead and keep the drug on the market. However, Merck could still be concerned with litigation and reputational costs, although the decision of the FDA to keep the drug on the market should mitigate these costs considerably. (3) The FDA makes no decision for the time being and will probably study the case before making a determination on withdrawal. This is the scenario Merck was most likely dealing with. In this situation, Merck can do nothing but await the FDA decision and then act accordingly (as in scenario (1) and (2) above), but there are additional costs imposed by waiting. Instead, Merck could preemptively act by guessing the likelihood of the FDA’s action. It would only make sense for Merck to voluntarily withdraw if it believes that the likelihood that the FDA would order withdrawal is high or if the cost of guessing incorrectly, that the FDA would decide not to withdraw when in fact it will decide to withdraw, is prohibitively high. Whatever the reasoning or combinations of factors involved, an economist could rationalize Merck’s actions by showing that the company was logical in voluntarily withdrawing Vioxx prior to official FDA action because it determined that voluntary withdrawal was the best outcome, given the likelihood of events based on its assessments of the risk and values.

In the end, the only explanation Merck gave for withdrawal was that it was in the best interest of patients.[266] One can say that if this was the reason, ethics trumped all economic factors. On the other hand, this can also fit nicely into a rational behavior model because if the value human life or the interest of patients was high enough, the value of withdrawal substantially increases and the value of leaving the drug on the market substantially decreases.

4. Effects of Vioxx Withdrawal on Patients

Perhaps those most confused and upset with Merck are not those who thought the drug should be withdrawn but rather those who opposed the decision, namely patients who relied on the drug. Their biggest complaint is no freedom of choice. Says Dr. Diwan, “Many people who took Vioxx were those for whom other medications had not worked.”[267]

There are countless stories of patients who gained relief from Vioxx when nothing else worked. Hayes Wilson, a rheumatologist in Atlanta says, "If you live with intractable pain every day of your life, would you take a (small) chance that you would have a heart attack? A lot of my patients would."[268] "With Vioxx,” says Barbara Birmingham, “I could bend my fingers in the morning and walk without problems. I can't do that now."[269] Before Vioxx came on the market, Dimitri Poulos, who was diagnosed with rheumatoid arthritis in 1998, was sometimes unable to sit in a motive theater or dress herself. “I have 40 pills left, 40 days before my life and ability are severely altered... I will assume all liability and sign any waiver. Please give me that option," says Poulos. The nature of arthritis is that it leaves many in a state of near immobility and to these patients, the cardiac risk is underrated.

Whether or not patients should be able to make individualized cost benefit analyses is very debatable, but it is important to note that given the opportunity, many would take the chance. There will be a more thorough discussion of patient issues in subsequent sections of this paper.


1. Review of Labeling

A little known difference is that while a label is what is on the outside package of the drug, labeling refers to the set of communications (e.g. patient package insert) that accompanies the drug.[270] Unless regulations specify that certain information must be on the label, drug labels that cannot carry the sufficient information the FDA requires to be communicated to end users must use labeling to convey the required information.[271] The FDA imposes a high level of accuracy, and labeling is considered to be misbranded if it is false or misleading in any particular.[272]

The FDA’s regulation of accurate labeling is vigorous and robust. As evidenced from the advisory committee meetings before the approval of Celebrex and Vioxx, the labeling of a drug can be quite a contentious issue because of its importance to patients, physicians, drug makers, and the FDA. During the advisory committee meetings, one of the most heated debates was over whether to allow Celebrex and Vioxx to drop the NSAID class warning template and whether to allow comparisons with other NSAIDS on the labeling,[273] which illustrates the tension between accuracy versus over-cautiousness.

2. Re-labeling of Vioxx

In March of 2000, Merck learned of the preliminary results of the VIGOR trial, which showed a five-fold increase in hearth attacks for patients taking Vioxx compared to patients taking naproxen,[274] and promptly disclosed these findings to the FDA.[275] Two months later, the VIGOR data were submitted to the New England Journal of Medicine and was published by the Journal in November of 2000.[276] Meanwhile, in June of 2000, Merck made a supplementary application to include the results of the VIGOR trial in the prescribing information.[277] In February of 2001, a public Advisory Committee meeting was held, and while there was concern about heart risks, no panel members suggested withdrawing Vioxx.[278] Some felt that the drug’s GI benefits far outweighed its possible cardiac risks, but cardiologists argued that such possible harm was still a real issue.[279] All supported the idea that more studies should be conducted.[280] In April of 2002, the FDA approved revised labeling, which reflected both the data from the VIGOR trial and the interim data from an ongoing trial with elderly patients with Alzheimer’s disease, which did not show an increase risk of CV events.[281] The labeling changes were very detailed and included a caution that a dose of 50mg was not recommended for chronic use.[282]

The obvious question that comes to mind is why Merck chose re-labeling instead of withdrawing. Merck officials have claimed that there was still uncertainty surrounding the dangers of Vioxx because of limitations of the VIGOR trial, namely the omission of aspirin users and a placebo group in the study. They concluded that the difference in CV rates were most likely due to the beneficial effects of naproxen in inhibiting platelet aggregation in a similar fashion to low-dose aspirin instead of a deleterious effect of Vioxx.[283] However, when the results of APPROVe came out, Merck promptly withdrew Vioxx from the market.

Given that the APPROVe study had recognized limitations,[284] which arguably could have justified keeping Vioxx on the market, why did Merck choose to ignore these limitations, especially in light of its defense of Vioxx after the VIGOR trial, a study also marked with limitations? Even Kweder acknowledged that the FDA would have had to look very closely at the studies before deciding on withdrawal and bemoaned not having the opportunity to do so before Merck’s sudden withdrawal.[285] Simply put, Merck re-labeled once, why not twice?

Many possible reasons may explain Merck’s departure from re-labeling toward a more drastic measure. Obviously a second study showing an increase in cardiac risk means further proof of the dangers of the drug. If Merck believed that the evidence showed sufficient danger, it could have made an ethical or an economic decision to pull the drug. But even if Merck did not believe that there was sufficient evidence of danger, it still must take into account the perceptions of the public and of the FDA. From the results of VIGOR to the results of APPROVe, something must have changed (whether risks, values, perceptions, etc.) to lead Merck down a different path. The only explanation Merck gave for withdrawal was that it was in the best interest of patients[286] , which would arguably make its actions consistent since a second negative study should have increased Merck’s concern for patients, making withdrawal a natural choice in the progression of reforms.

3. FDA’s Re-labeling Authority and Vioxx’s Re-labeling Delay

Significant criticisms of the ineffectiveness of re-labeling have stemmed from the huge labeling delays surrounding Vioxx. Merck first learned of the VIGOR results in March of 2000. It submitted a supplementary application to the FDA in June of 2000 to include the results of VIGOR. Apparently, conflict in the labeling language led to a public advisory committee held in February of 2001, about eight months after the submission of the supplementary application and about eleven months after Merck learned of the VIGOR results. But this is not the end of the story because it was not until April that the FDA approved a revision of the prescribing information, which would bring the count up to ten months after the supplementary application submission and fourteen months after Merck learned of the results of VIGOR. What could possibly explain such delays? This is the same questions with which many congressmen have struggled. Before the Senate Health, Education, Labor, and Pensions Committee, Sandra Kweder of the FDA testified that the delay was caused by the FDA’s lack of authority to order specific labeling changes. According to Kweder, the FDA did not have the authority “to tell a company, this is how your label has to look. This is the language that needs to go into your label...We have to negotiate with the company the specific language of how things should be worded, placement...”[287] Two days later, before the same Senate Committee, Janet Woodcock of the FDA, overruled Kweder’s assertion with her testimony that the FDA indeed had the explicit authority to dictate labeling changes.[288]

Who was right in regards to the FDA’s authority? A thorough examination of the issue will show that the FDA had both the explicit and implicit authority to dictate specific labeling changes. This explicit authority was recognized by Senator Enzi during the Senate Committee hearing when he rebutted Dr. Kweder’s claim that the FDA had no such authority.[289] Under Section 502(a) of the Federal Food, Drug, and Cosmetic Act (the “Act”), a drug is misbranded “if its labeling is false or misleading in any particular.”[290] Under Section 201(n)[291] of the Act, in determining whether labeling is misleading, one must take into account not only representations made or suggested but also if the labeling “fails to reveal material facts...with respect to consequences which may result” from the use of the product as labeled, or under uses that are “customary or usual”.[292] Under this standard, the FDA can easily assert an increased heart risk as a material fact that was not revealed, especially because the increased heart risk in the APPROVe study resulted from a recommended dose of the drug.[293] Even if the increased heart risk had only resulted from a dose higher than that recommended on the label (as it was in the case of Celebrex), such a dose may still be argued to be a use that is “customary or usual” since dose creep is a known phenomenon in the medical field. Likewise, even though the increased heart risk was not shown until 18 months of continuous use, and long-term use has not been affirmatively recommended on the labeling of Vioxx, such long-term use can also be argued to be “customary and usual.” And once the FDA has made the determination that a label is misleading, it is extremely difficult to bring a challenge against such a determination because the agency receives great deference for its “technical analyses of data”.[294]

Once a label is determined to be misleading, the drug is by definition misbranded, which is a prohibited act, and the FDA can take numerous actions against a drug company under its misbranding authority,[295] which are judicial remedies provided elsewhere in the Act and other mechanisms.[296] Former FDA general counsel, Peter Hutt, has in fact expressed the view that “except where expressly prohibited, I believe the Food and Drug Administration is obligated to develop whatever innovative and creative regulatory programs are reasonable and are appropriate to achieve the fundamental objectives laid down by Congress.”[297] Therefore a company that sells a misbranded drug and refuses to correct its label can face regulatory measures like withdrawal or even criminal and civil penalties. However, it is extremely important to note the severe limitations of the FDA’s misbranding power with regards to dictating specific labeling changes, in that, because the FDA only has authority to withdraw, it has to use this as a back route towards labeling changes. This will make it difficult when the agency wants labeling changes but is unwilling to withdraw the drug.[298]

Notwithstanding explicit statutory authority, the FDA also has a lot of implicit authority to force specific labeling changes. To start with, a threat of withdrawal should be enough to force any prudent company into compliance. Again, not only can the FDA effortlessly deny hearings but will receive great deference in any judicial review. Moreover, the FDA can always threaten to take the issue to the media, which will no doubt be eager to pick up the story. The FDA can also slow down any review process for NDAs not yet approved or for future NDAs. The ability of the FDA to use implicit authority to force labeling changes is thus constrained only by what it explicitly cannot do.

Since it is quite clear that the FDA has the authority, whether explicit or implicit, to dictate labeling changes, this suggests that the focus and heart of the matter is more accurately placed on the question of why the FDA did not force Merck to adopt specific labeling changes but instead chose to enter into multiple drawn-out negotiations with the drug company, which resulted with a compromised labeling change nearly ten months after initial negotiations. Two explanations persist. One popular view is that the FDA had been bought out by the drug industry or more accurately that the agency has transferred power to the drug companies in exchange for monetary funding from the industry. But this kind of conspiracy theory doesn’t explain why the FDA sometimes chooses to use its authority and why sometimes it does not. Clearly, we have seen the FDA use its authority to dictate specific labeling changes only months before the Vioxx withdrawal, with pediatric use of antidepressants and again with Bextra, only weeks before the Celebrex controversy.[299] Therefore, the second supposition, that the FDA clearly saw the benefits of the drug and did not want to use its authority to take it off the market, may be a more likely explanation. Because the FDA cannot force the labeling changes directly but rather indirectly by way of withdrawal of the drug[300] , forcing a reluctant Merck to change the label may have resulted in an outcome that was undesirable—that is, Merck refusing to change the label and the FDA withdrawing it from the market. Therefore, if the FDA valued having Vioxx on the market, it would rationally choose to negotiate with Merck rather than assert its authority and risk having to make good on its threat and withdraw the drug from the market. Unfortunately, these guess will remain only speculation unless the labeling negotiations process becomes more transparent.

4. Pfizer’s Response: Change Effected Through Consumer Advertising and Labeling

A. Introduction—A Parallel Situation

On December 19, 2004, nearly three months after Merck’s withdrawal of Vioxx, Pfizer announced that it was halting all advertising of Celebrex.[301] For comparison purposes, such a move could almost not have resulted from a more parallel situation, for it was also the result of a clinical trial, APC, which led to Pfizer’s decision. In fact, both trials were cancer studies, and both showed an unexpected increase in CV risk leading to their early terminations.

B. A Look at Pfizer’s Response

According to a Pfizer spokesperson, the halting of advertising is indefinite and includes television, radio, newspaper and magazine ads, and other promotions to consumers, but did not extend to the marketing of the drug to doctors.[302] Celebrex’s label was changed only to emphasize that there is no problem if used for treating arthritis in recommended doses,[303] but did not reflect the results of the APC study. It remains unclear why such important results were not reflected in the labeling change, especially because the discussions and negotiations between Pfizer and the FDA following the results of APC, are confidential and not available to the public.

Several reasons could explain Pfizer’s seemingly weak and indirect re-labeling efforts, and all lead to the conclusion that Pfizer may have simply been waiting for specific FDA directions before taking action. After all, when the results of APC were released and before Pfizer agreed to halt consumer advertising, FDA acting commissioner, Lester Crawford stated that the FDA had great concerns about Celebrex and was considering taking action, including withdrawing the drug or placing severe warnings on its label.[304] Furthermore, only a few weeks earlier, the FDA had already ordered Pfizer to place a black box warning, the most severe warning label, on its other COX-2 drug, Bextra.[305] Pfizer was also aware that an advisory committee panel was going to convene in early February to decide the fate of COX-2 inhibitors, including Celebrex. Thus Pfizer may have been waiting for further direction before changing its label to reflect the study.

But is waiting justified when there is a risk to consumers? In the end, waiting is consistent with Pfizer’s claim from the outset that it believed that the results of the APC study were an anomaly[306] and that no other completed study of Celebrex showed any increased heart risks.[307] Pfizer also asserts that heart risks found in the APC study occurred when patients took Celebrex at two to four times that recommended dose for long periods and that Celebrex has not been shown to increase heart risks when taken at the recommended dose level for arthritis patients.[308] In fact, Dr. Friedman from the Cornell Medical Center has acknowledged that, “the risk appears statistically significant at a certain dose, but like most adverse drug effects, the risk could well be distributed on a continuum with dose.”[309] Taking these two facts together—that the risk from the APC study occurred with abnormal use of Celebrex and that no risk has occurred with normal use--can likely explain why the change in label only emphasized that there is no problem if used for treating arthritis in recommended doses. One may view such a labeling change as weak or as reflecting the beliefs of a drug company that regards its product to be overwhelmingly safe.

C. FDA Involvement?

There is however, a tension between Pfizer’s emphasis on the safety of Celebrex, in light of the totality of the studies on the drug, and its decision to stop all direct-to-consumer advertising of the drug. In other words, if Pfizer believed Celebrex was overwhelmingly safe, why stop advertising the drug? If it believed that the heart risks shown in the APC trial were an anomaly, why make contradictory statements such as advertising will be stopped until the seriousness of the drug’s health risks can be determined?[310] Pfizer’s decision to halt advertising but continue selling the drug has baffled and confused many, and had led Dr. Friedman to ask, rather poignantly, “If a drug is truly safe and effective, then a drug company has little to fear from advertising it to the public. Pfizer’s action will only foster the public’s growing perception that the drug companies place profit ahead of public safety.”[311] The most common explanation put forward is that the suspension of consumer advertising was the result of negotiations between Pfizer and the FDA. Newspapers across the nation have reported the event slightly differently, and again, because these meetings are confidential, we can only guess as to the content of the discussions between Pfizer and FDA. Some reported that Pfizer had “little choice”[312] but to halt consumer advertising because the FDA said that it was considering severe re-labeling or withdrawal of the drug. Along the same vein, others believe that Pfizer suspended consumer advertising “under pressure from the FDA”[313] . Last, there is also the portrayal that that decision was made by Pfizer and approved by the FDA.[314] While this may all be a matter of semantics[315] , it is important to note that the FDA was to some extent involved in the decision.

The highlighted role of the FDA in the Celebrex controversy entreats a query into its seemingly absent role in the Vioxx withdrawal. Three days after Merck informed the FDA of the results of the APPROVe trial, the drug company announced its decision to withdraw Vioxx from the market. There is no indication that any communications or negotiations between Merck and the FDA during those three days yielded this result. In fact, Dr. Kweder of the FDA said the agency was stunned by the decision and regretted not having the opportunity to conduct further investigations of the drug to cast an agency decision on its safety.[316] It seems rather odd that a company that was not under pressure (at least not publicly) from the FDA would voluntarily withdraw its product whereas a company under arguably immense pressure from the FDA would make a few concessions but for the most part, stand by its product. In the end, the answers may lie in the respective confidence of the companies. Merck said it did what was in the best interest of patients. Pfizer asserts the same. Henry A. McKinnell, chief executive of Pfizer, has said that the company has no plans to stop selling Celebrex because “it is an appropriate option for many, many patients.”[317]

D. Evaluation of Pfizer’s Decision

The dangers of direct-to-consumer advertising have always been justified by its potential benefits. The perils lie in that repetitive effective messages disseminated to the general public may infiltrate the minds of many who do not need such drugs and create an artificially high demand for them. The benefit of course is that the public will be educated about cutting edge new drugs of which they would otherwise not be informed. Janet Woodcock stated in 2001 before a House Committee that, “At present, FDA is not aware of any evidence that the risks of DTC [direct-to-consumer] promotion outweigh its benefits. [318] The FDA has jurisdiction over the advertising of prescription drugs to consumers and the general public, an authority that runs with the drugs, not with the market to which the advertisements are directed.[319]

All advertising material for prescription drugs are submitted to the FDA for scrutiny.[320] Drug companies understand that courts will not overrule the FDA’s expertise when the agency makes a determination of inadequate efficacy data or misleading claims.[321] Furthermore, no sensible drug company would challenge the FDA on advertising determinations for fear of withdrawal of an NDA or stalling of future NDAs.[322] There are a few underlying principles for the advertisements of prescription drugs. The general rule is that advertising claims must be balanced and cannot be false or misleading.[323] Other important principles include: claims should be based on the approved labeling of the drug[324] , negative information about contraindications and side effects must be clearly communicated in “brief summaries”[325] , and exaggeration of claims are not permitted.[326]

Even this general overview of consumer advertising brings up relevant questions concerning Pfizer. First, if the FDA had put pressure on Pfizer to halt advertising, what was its basis for doing so? Perhaps the agency felt Celebrex’s effectiveness claim was now exaggerated with respect to its dangers. Or perhaps it was due to the absence of warnings of the possible heart risks in the advertising materials. But none of these explanations make sense because the FDA allowed Pfizer to continue selling Celebrex, which presumably is accompanied with the labeling, the information of which must be basis of any consumer advertising. Also, Pfizer never stopped marketing the drug to physicians. Therefore, the “discovery” of increased heart risk cannot logically be the basis for the halting of consumer advertising if this information continued to be permitted to be disseminated to the public through labeling and physicians. The agency’s action maintains consistency only if it believes that consumer advertising is a more effective or pervasive form of communication than through labeling or physician advertising. This conjecture is supported by Janet Woodcock’s concern that changes in labeling are not an effective means of communicating new information to the public either because patients are not as aware of labeling changes or the nature of paper labeling makes it so that old labels continue to be in circulation for some time.[327] According to Woodcock, there is considerable difference between labeling changes and information that will “reach directly” to prescribers and patients.[328] One FDA-endorsed idea is to inform the public of new information from studies via a drug watch web page.[329]

Thus, if we assume that the FDA pressured Pfizer to stop consumer action but allowed it to continue selling the drug with its labeling information and allowed it to continue marketing the drug to physicians, its actions are only justified if the agency believes that consumer advertising is a more pervasive and dangerous form of communication, which should be cut off until the agency comes to a final decision on the matter. On the other hand, Pfizer’s decision to stop consumer advertising despite its confidence in the safety of Celebrex can easily be explained by its attempt to save the drug from the fate of more severe FDA action. As noted earlier, no prudent drug company will make challenges on advertising when the FDA possesses a nearly unchallenged power to withdraw drugs.

This is not to say that Pfizer did not suffer from the advertising suspension. In 2003 alone, Pfizer spent over $87.6 million in major media advertisements in an industry that is a $3.8 billion a year business.[330] But in the end, its decision likely saved Celebrex. The suspension of advertising seemed sufficient to cool down the FDA while all awaited the meeting of a special advisory committee that would meet two months later and determine the fate of the COX-2 inhibitors.


An examination and comparison of the different approaches taken by Merck and Pfizer seem somewhat incomplete without an evaluation of the approaches themselves. The inquiry is whether and how effective these approaches are in communicating to the public the risk of taking these drugs. Clearly, the withdrawal of Vioxx was a clear and direct message to the public of the dangers of Vioxx. While an effective communication, it may have been too high of a price to pay, since informing the public of a risk of a drug does not have to go so far as to take the drug away from them. Nevertheless, if Merck believed that the dangers of Vioxx outweighed the benefits, short of recall, withdrawal is the most effective means of effectuating the message.

Turning to re-labeling and to the suspension of advertising, the effectiveness of such communication is less clear. Both methods are supposed to serve as secondary sources of communication, the primary source of which is supposed to come from the prescribing physician. In regards to labeling, it is unclear whether patients take the time to read the lengthy and convoluted patient insert that is the heart of the label. First, even though the FDA holds strict standards on the readability of labels, patients with varying degrees of vision problems can see have difficulties reading the labels.[331] Secondly, the prominence of the information concerning risks or warnings affects whether it is effectively communicated to the patient. For instance, when Merck re-labeled Vioxx to reflect the results of the VIGOR study, the results were simply added as part of the clinical studies section of the labeling[332] , and the warnings about the increased heart risks were not highlighted. In fact, quite recently, an editorialist in the Circulation Journal chided the FDA for not requiring a more prominent display of a renal risk warning for the drug Natrecor[333] , saying, “the FDA acted as if practicing physicians would carefully read the precautions in the package insert and that this information would influence clinical practice.”[334] Dr. Cecil Wilson, a member of the American Medical Association’s Board of Trustees, testified before a Senate Committee that although the best way to inform doctors about the risks and benefits of a drug is through the labeling (patient insert), it is usually very difficult to locate particular information in the very lengthy material, and information like “dosage, contraindications, major risks, and potential drug interactions are often buried within the highly technical text of the document.”[335]Unfortunately,” says Dr. Wilson, “today's package insert has become a long and complicated, really legal document rather than a useful resource for physicians.” [336] The lack of faith on the part of physicians reading the label would no doubt extend to patients as well. Black box warnings are the most severe labeling change because warnings are enclosed in a box and listed first on the labeling. Such labeling was used with Bextra. According to Dr. Fassler, there is data indicating a very significant decline in the prescription rate of certain anti-depressant drugs required to carry the black box warning.[337] Since neither Vioxx nor Celebrex used a black box warning, it can be asserted that the most effective means of labeling communication was not used. A related issue beyond the scope of this paper concerns the language used on the labeling. Words like “myocardial infarction” were used to describe the increased heart risk of Vioxx, and it is doubtful that the lay person would understand this to mean “heart attack”.

Direct-to-consumer advertising has long been thought to be an effective source of communication to the public. Many attribute the popularity of drugs like Vioxx and Celebrex to consumer advertising fueling the superficial demand for them. In fact, these ads have become so commonplace that there was concern over whether such ads were losing their impact.[338] Says Peter J. Pitts, a former FDA public affairs official, “Direct-to-consumer had developed to a point that some drug advertising was barely distinguishable from marketing for Cheerios.”[339] In light of this, it is unclear what effect a suspension on advertising would have without affirmative warnings of the risk of the drugs. Pfizer probably lost out on new clientele base (from continued marketing) for its drug, but what about patients already hooked on Celebrex? Were they sufficiently warned of the new found risks? There is the possibility of remedial advertising, which requires communications to the same audience, but this remedy is rare, and has only been used “where it is felt that the violative promotion has left a misleading impression of lasting effect,” on the audience.[340]

A brief look at the different methods used to communicate the risks of a drug to the public leaves us still searching for an effective method that will allow the public to be informed of risks while maintaining the choice of making an educated decision on whether to use the drug. The case of Vioxx is one where the remedy may have gone too far but Celebrex may not have gone nearly far enough.


Perhaps the different decisions between Merck and Pfizer can be explained, at least in part, by their different litigation strategies. Merck is presently under Congressional and Justice Department investigations, and faces potentially thousands of tort lawsuits that could cost the company billions of dollars.[341] So far, hundreds of individual and class action lawsuits have been filed in federal courts against Merck, which will be consolidated in a federal court in New Orleans before a Judge Eldon E. Fallon, a choice that pleased many plaintiff’s lawyers, although Judge Fallon is also known to be unbiased in such tort cases.[342] Merck has set aside $675 million in reserves for legal expenses.[343] It is believed that Merck may have withdrawn Vioxx from the market in anticipation of lawsuits after the public release of the results of APPROVe. However, new lawsuits have actually been coming in at an “accelerating pace” since the withdrawal,[344] which supports the belief of the drug industry that Merck “compounded its legal and financial problems” when it withdrew Vioxx from the market.[345] Also, doesn’t common sense dictate that pulling a drug off the market means admittance that the drug was dangerous?

So did Merck make the wrong litigation move? Perhaps not. First, by pulling the drug off the market, Merck made a good faith attempt to mitigate harms of the drug. The drug company can justify the move by saying that such a moratorium was put into effect so that a proper determination of drug’s safety could be conducted before more patients were exposed to the observed clinical risks. Secondly, when Merck withdrew Vioxx, the APPROVe study only showed that there was an increase heart risk after 18 months of continuous use. This evidence may knock out the claims of many plaintiffs, some of whom used Vioxx for only a short period of time, such as the case of Keith Stubblefield, who suffered heart failure and died six weeks after taking Vioxx.[346] Along the same vein, many plaintiffs will face huge obstacles proving causation because heart attacks and strokes are caused by many different reasons.[347] Finally, it was serendipitous that the safety of Celebrex, naproxen, and other NSAIDs came into the spotlight because it led to an FDA advisory committee meeting on the safety of NSAIDs and COX-2 in particular. The advisory committee voted that Vioxx was safe enough to bring back to the market, and this judgment is bad news for plaintiffs. Any assertion of Merck being careless or reckless for keeping Vioxx on the market after indications of its dangers surfaced in the VIGOR study “will now have to confront the fact that 17 expert advisers think it is safe enough to put back in circulation.”[348] The fact that Merck was cautious enough to withdraw Vioxx even though expert advisors later thought it was safe enough to put back on the market could be used as a favorable spin on the withdrawal.

The case of Celebrex may simply be one of learning from the error of others. The fact remains that lawsuits piled up exceedingly fast once Merck withdrew Vioxx from the market. Pfizer’s position from the outset was to claim that the results of the APC study were an anomaly and that all other completed studies showed no such increase heart risk. Its litigation strategy can be seen as a parallel of its marketing and public relations move—promote the safety of the drug. Thus, deciding not to withdraw was a natural extension of its stance to stand by its drug. Unlike Merck, Pfizer does not have to admit any perceived risk of its drug in any litigation because it did not remove the drug from the market. Its suspension of direct-to-consumer advertising may pose the same issues as the withdrawal, although to a lesser degree and can be justified in the same matter—that is, as merely a good faith attempt not to unduly influence consumers until the safety of the drug could be determined. With the FDA advisory committee vote overwhelmingly supporting keeping Celebrex on the market (31 to 1), it seems Pfizer made the right litigation move by standing by its drug. With 31 experts supporting Celebrex’s safety compared to Vioxx’s 17, Pfizer seems to be in quite a relatively favorable litigation position.


To be fair, Merck and Pfizer may have taken different routes because of their arguably highly different situations. Perhaps the choice between re-labeling/advertising suspension versus withdrawal was due to the different safety perceptions of the drugs, and thus perhaps, both Merck and Pfizer acted accordingly in their respective circumstances. It is important to consider whether the difference in the perceptions and reputations of Vioxx and Celebrex were caused by the drugs’ differing risks (since Celebrex has been recognized to be less dangerous for the heart) or whether the difference in perceptions and reputations were an effect of the withdrawal of Vioxx or both.

The argument that the different safety perceptions of Vioxx and Celebrex may have been an effect of Merck’s withdrawal of Vioxx has its basis in the belief that the dangers of one drug will elevate the safety or use of another. For instance, when the ADAPT study results came out showing naproxen to be associated with an increased heart risk, Pfizer’s stock increased by 4%.[349] Also, Dr. Kenneth Brandt has noted that the difference in labeling between Vioxx and Celebrex, after Vioxx’s labeling change from VIGOR, resulted in greater sales for Celebrex.[350] Extending this logic, it may be the case that Merck’s previous negative actions of re-labeling and then withdrawal boosted Celebrex’s safety perception so much that although an increased heart risk was later found with Celebrex, a satisfactory response did not need to rise to the same level as Merck’s response. In the end, these arguments are based on mere speculation. At most, the argument that Celebrex was boosted by Vioxx’s reputation can only be part of the story.

The assertion that the different safety perceptions of Vioxx and Celebrex were caused by their differing heart risks is reflected in many studies and is highly supported in the medical community. Although both Vioxx and Celebrex are COX-2 inhibitors, they are also “distinct molecules with specific biological purposes”.[351] The general view seems to be that Vioxx poses more of an increased heart risk than Celebrex. This perception stems from the fact that so far, only one completed clinical trial of Celebrex, APC, has shown an association with increased heart risks compared with placebo. Moreover, the results of APC was in stark contrast to other Celebrex studies, including preSAP, a trial conducted in parallel to APC, had an identical analysis used to assess CV risk, and was conducted by the same independent safety review board.[352] Says Pfizer chief executive Henry McKinnell, APC is “inconsistent with the great body of data that we’ve accumulated over 10 years.”[353] On the other hand, Vioxx’s association with increased heart risk has been indicated in three independent randomized trials and a cumulative-meta analysis.[354] Dr. Topol of the Cleveland Clinic has also noted that several epidemiologic studies involving a large number of patients taking NSAIDs or COX-2 inhibitors have suggested an increased risk for Vioxx, especially at high doses, but not for Celebrex..[355] According to Dr. Topol, in order to really determine drug differences there needs to be head-to-head comparisons of Vioxx versus Celebrex.[356] Without such direct comparisons, we can only look at different studies of each drug and compare them to each other, but such comparisons are very limited because of important differences such as patient populations, drug doses, and duration of therapy.[357]

Although no head-to-head comparisons of Vioxx versus Celebrex have been conducted, a case control study, observational studies, and meta-analyses have been performed. Dr. Stephen Kimmel and colleagues performed a case control study to determine the effect of Vioxx, Celebrex, and other non-selective NSAIDs on risk for heart attacks.[358] Their case control study included more than 8000 patients taking COX-2 inhibitors at 36 hospitals in a 5-county area, where 1718 patients suffered heart attacks and 6800 did not.[359] Their findings indicate that patients taking Vioxx faced an increased heart attack risk of 2.72 (statistically significant) compared with patients taking Celebrex.[360] Also, patients taking non-selective NSAIDs faced a lower risk of heart attacks than patients not taking such NSAIDs.[361] Kimmel also looked at the results of 3 published observational studies and 2 meta-analyses that compared Vioxx and Celebrex for heart risk. While none of the studies indicated an increased heart risk for Celebrex, some showed a risk for Vioxx.[362] Limitations recognized were the possibility of recall bias, the fact that only 50% of eligible participants completed the telephone interviews, and the possibility that the lower heart attack rates of Celebrex occurred by chance.[363]

Dr. Kimmel has articulated what seems to be the general view of COX-2 inhibitors. “Current data provide conflicting evidence of the potential risks and benefits of different COX-2 inhibitors, but, overall, they suggest that there may be differences between them. We sought to determine the association between COX-2 inhibitors (specifically rofecoxib and celecoxib) and risk for nonfatal MI.”[364] Whether or not the difference in risks between Vioxx and Celebrex are real, the manifestations from differing perception may be equally important. Kaiser, whose procedures have been a prototype for other health care providers has suspended the use of Pfizer’s other COX-2 inhibitor, Bextra, and has ordered its pharmacies to stop dispensing the drug, but Kaiser doctors will continue prescribing Celebrex.[365] This move by Kaiser is the first of its kind—that is, banning a drug that has been approved by the FDA.[366] This suggests that if Vioxx was still on the market, it may have faced the same ban. After all, the increased heart risk of Bextra has not been known to be as high as the risk from Vioxx. Kaiser doctor, David Campen has commented that at low doses, Celebrex does not seem to have the same risks as Bextra or Vioxx.[367]

Finally, although the FDA Arthritis Advisory Panel that reviewed the safety of COX-inhibitors and other NSAIDs, recommended that both Vioxx and Celebrex were safe enough to be sold on the market, there was overwhelming support for Celebrex as opposed to quite marginal support for Vioxx. In fact, the committee votes were 17-15 in favor of Vioxx and 31-1 in favor of Celebrex.[368] Moreover, two New York Times reporter did their own recount, in which they took out the votes cast by panel members[369] that had conflict of interests, mainly ties to the industry. The recount would not have affect Celebrex with its 31-1 vote but it would have meant that the vote to return Vioxx to the market would only have been 8-14, [370] a marked different from 17-15, and definitely not enough votes for its return to the market.

Whether or not the differences between Celebrex and Vioxx are real, and notwithstanding the reasons for such varying views, there is certainly differing perceptions of their safety, and this may in the end be what matters in terms of the actions of companies. This is not to say that Merck and Pfizer both acted in the right manner, and a more thorough analysis of normative questions will be discussed later, but in any comparison between the two drugs, it is important to note that the two situations were not equal. Merck and Pfizer may still be responsible for acting accordingly but to hold their situations constant may be undeserved.


Amidst the COX-2 inhibitor controversy has been a backlash against the FDA. In fact, the agency has received more criticism than the drug companies themselves. Perhaps this is a natural result as the FDA is regarded as the protector of America’s health, “an icon of trust...a certifier of safety...an enabler of innovation...[and] a repository of information,”[371] whereas drug companies are in the business of profit. Criticisms of the agency due to the Vioxx and Celebrex controversy include, close ties to the drug industry, not enough authority to dictate changes, and drug approval standards that are too tax. Reforms has been demanded, but are they warranted?


1. Industry Funding: Speeding Up Approvals

When critics refer to the FDA working for the drug industry, what they are usually alluding to is the 1992 agreement (and renewal in 1997 and 2002) between the agency and the industry, known as the Prescription Drug User Fee Act (PDUFA)[372] , which was an arrangement that consisted of the industry giving millions of dollars in funding to the agency in return for the agency spending a certain annual amount of money on new drug approvals.[373] The agreement came about at a time when the FDA was severely under-funded and simply did not have enough reviewers to review important drugs, many of which were to treat AIDS, cancer, and heart disease.[374] The industry agreed to pay user fees, which provided for much of the budget of the Center for Drug Evaluation (CDER), the main office that conducts drug reviews and safety.[375] But what the industry wanted in return was a guarantee that government financing of new drug reviews would never fall below 1992 levels (later amended to 1997 levels) to ensure that the industry money wasn’t simply replacing government money but was actually used to hire new reviewers, but as less and less Congressional dollars are coming in, money from other FDA departments are being shifted into new drug reviews in order to meet the minimal level provided pursuant to the agreement.[376] According to Dr. Graham of the FDA, this heavy focus on approvals, instead of safety, has led to problems like Vioxx and will inevitably lead to similar problems in the future.[377]

Were the dangers of Vioxx and Celebrex caused by being approved too quickly? While quicker approvals surely enhance the industry’s agenda, this is not necessarily bad. FDA officials maintain that drug reviews are as vigorous as they were a decade ago, which is supported by the fact that the ratio of drugs withdrawn compared to those approved have not changed over decades.[378] In fact, there is no reason to think that quicker reviews mean less rigorous reviews if money is being spent to hire new reviewers and fund the reviews in general. It may even be more logical to surmise that that reviews are becoming more stringent, given that so much of the budget (about four-fifths) goes into such reviews. According to Dr. Raymond Woosley, over the past ten years, the drug industry has increased R & D spending by 250 percent, but the number of significant new products submitted to the FDA for review has actually fallen by 50 percent.[379] In fact, the number of approved NDAs remain relatively low, for instance, 78 in 2002 and 73 in 2003.[380] Vioxx and Celebrex underwent the same process as other drugs in order to be approved, completing the requisite initial process and clinical studies.

In addition to adhering to a rigorous standard of review, quicker reviews also mean quicker access to much needed and even life-saving drugs. In fact, Vioxx and Celebrex were both given priority reviews because of their potential to reduce adverse GI events.[381] A 1984 law allowing the generic drug industry to capitalize on sales once drug patents expired, put further pressure on drug companies to have quicker reviews.[382] Drug companies were already spending 12-15 years and $1 billion for each successfully developed drug, and slower reviews simply meant the time and money were not worth it.[383] In a catch-22 situation where faster reviews may mean compromising thoroughness but slow reviews may mean no new drug development, the only way to come out of the situation was to increase funding that would allow for quicker but also thorough reviews and still keep the proper incentives for drug companies to continue researching and developing new drugs. The issue of safety, therefore, is better focused not on the approval process (which is arguably over-funded) but on the post review process, which seems to be under-funded. Approving a lot of drugs is theoretically good, but is it good if the FDA cannot keep up with post approval mechanisms to ensure the safety of all these approved drugs? This issue will be analyzed further in the discussion of the FDA’s authority to require post approval studies.

To be fair, some have defended the 1992 amendment as having saved the industry.[384] Because of the desperate need to approve new drugs, Republican lawmakers proposed legislation that would have cut the FDA out of the process and essentially allow drug companies to skip agency review.[385] The lawmakers justified such drastic measures on the drug lag and the serious need for new drug approvals.[386] Luckily the agreement was adopted and put into place, allowing review time to be cut in half, and with it, severing all arguments of a drug lag.[387] So in fact, the 1992 agreement, though not perfect, had saved the integrity of the approval process. FDA’s approval process is quite stringent and still considered to be the best in the world. Again, approval is not so much the problem as what happens or more accurately, what doesn’t happen, after the approval.

2. FDA Advisory Committees and Conflict of Interests

The use of advisory committees has become an integral part of the pre and post drug approval process. Although the FDA is not legally bound to adopt the recommendations of the panel, it usually does.[388] Use of such committees has been so widespread that there are now many different committees for specifics types of drugs. Vioxx and Celebrex were both reviewed by the FDA Arthritis Committee. Within the Vioxx and Celebrex controversy alone, we have seen the committee meet four times, once for each of the approvals, once after the results of VIGOR, indicating an increased heart risk with Vioxx, and finally, after the results of APC, indicating an increased heart risk with Celebrex. The idea behind these advisory committees was that it was good and beneficial for the FDA to solicit outside opinions of experts and physicians to review the efficacy and safety of new drugs.[389] The only problem is that panel members often have ties to the drug industry, which makes sense because these experts are also highly sought after by the industry as consultants or principal investigators of clinical trials or because many rely on grants from the industry to conduct research, especially in light of decreasing government funds.[390] According to Elizabeth Glode, “the result is that nearly all of the experts asked to serve on advisory committees by FDA are conflicted to some degree.”[391] It comes as no surprise then that many of the committee members on the Arthritis Committee had conflicts of interest.

In order to minimize the effect of conflict of interests, the FDA requires panel members to disclose any financial ties to the industry.[392] The FDA also publicly discloses such conflicts of interest to the public by having an FDA executive secretary read a statement acknowledging members on the panel with conflicts of interest, and the type of waiver, if any, the agency has granted to those members.[393] Waivers are granted when the agency determines that the benefits of having that member on the panel outweigh the risks.[394] Independent reviews on all waivers are conducted by the FDA’s Ethics Staff.[395] Such precautions are in place to ensure mitigation of conflicts of interest and to bolster public confidence in the agency.

Although such criticisms in regards to advisory committees are not new, the Vioxx and Celebrex situation has put a different spin on the issue. While the criticism related to advisory committees have focused on drug approval, the focus here was on a post-approval advisory committee evaluating the safety of COX-2 inhibitors in the aftermath new found data indicating heart risk and even a voluntary withdrawal from the market. It turns out that ten of the thirty-two panel members had ties to Merck or Pfizer, with conflicts varying from consulting fees, to research support, to even honorariums for speaking.[396] The vote to keep Celebrex on the market was 31-1, so with or without the conflicts, Celebrex was overwhelmingly supported.[397] But because the margin of support for Vioxx was so slim, the importance of these conflicts of interest was amplified. The vote for Vioxx was a close 17-15, in favor of bringing Vioxx back on the market.[398] Casting out the votes of members with conflicts would make the make 8-14, and the withdrawal would have been sustained.[399] Looking at it from a different angle, of the 10 conflicted members, 9 to 1 voted to bring Vioxx back.[400] While this suggests that industry ties do influence committee members, it should also be balanced with the fact that conflicts are almost inevitable and that the FDA waiver process is firmly in place to mitigate industry influence. Eight of the ten panel members stated that they were not influenced by past ties, and several of the panel members on the Arthritis panel stated that most or all of their financial conflicts consisted of money that was sent to their universities or institutions.[401]

The benefits of having these panels are not really in question. Such panels are cost-effective, especially as medicine is becoming increasingly specialized and complex, and allows for fresh perspectives, provides insulation for FDA staff from industry pressures, and most importantly, eases the FDA of its heavy workload.[402] The only issue is whether the benefits outweigh the cost of having panel members potentially being influenced by the drug industry, whose incredibly important decisions affect millions of consumers. For general approvals, perhaps such conflicts are not as essential because drug trials are few and fairly simple at that point. But when we are talking about something like determining the safety of drugs that have been shown to increase heart risks, where there are dozens of studies performed, many of which are conflicting, perhaps there needs to be further scrutiny into conflicts. For instance, financial conflicts is defined broadly, encompassing ties to competing products, and the financial interests of a spouse, minor children, general partners, and any organization in which the expert serves as officer, director, trustee, or current prospective employee, and as found in this situation, being an employee of an institution that has received grant money from a relevant drug company.[403] Further inquiry should allow members with fairly innocuous or tenuous connection to the relevant drug company to serve on the panel while screening out experts with closer ties to the industry. Again for a contentious issue like the COX-2 inhibitors, perhaps the FDA should open up room for public opposition by disclosing ahead of time the names of those appointed on the panel and allow for a procedure to bring to the FDA’s attention any strong opposition to certain members. In addition, the FDA should remind the public that it is not legally bound to adopt the recommendations of the panel. On controversial issues, there should be a formal FDA review process of the panel’s recommendation, and such review should be made public. Because there are immense benefits of having these panels, the FDA should try to maintain at least the appearance of propriety by implementing extra procedures when dealing with panels addressing contentious issues of public interest and importance.


As was discussed earlier, under Section 502(a) of the Federal Food, Drug, and Cosmetic Act (“the Act”), the FDA certainly has the authority to force specific labeling changes. However, this power is severely limited because it is an indirect authority and can only be asserted by way of withdrawal. Because a label that does not contain sufficient warning of a risk, such as an increased heart risk, is considered misleading under Section 201(n) of the Act, and thereby “misbranded” under Section 502(a), the FDA has the authority to take regulatory actions against the drug company, including withdrawing the drug. But William B. Schultz, former FDA deputy commissioner for policy has astutely noted that, “if the FDA finds new information about a drug like Vioxx, it shouldn’t just have the option of withdrawing the drug from the market or bringing this branding (ph) action.”[404] It is in fact, this lack of direct authority, not having the ability to tell a drug a company to change its label without having to go through other routes, which weakens the FDA.[405]

We have seen that this lack of direct authority has caused the FDA to enter into negotiations with Merck presumably because it wanted to keep Vioxx on the market. When drug companies know that there is value to keeping a drug on the market and that the FDA wants it to stay on the market, the agency’s power is greatly weakened because it will not make use of its misbranding authority and will not make good on its threat of withdrawal. Moreover, the FDA needs the power to make labeling changes in all circumstances, regardless of any other actions either by the drug company or the FDA. In other words, even if a drug company decides to withdraw its drug or if the FDA has ordered such withdrawal, the agency should still have the ability to tell drug companies to change the label and take remedial action with regards to drugs that are in circulation or in possession of patients. So, while the FDA certainly does not lack the authority to force a label change, its power greatly diminishes when it doesn’t have the direct power to specify labeling changes without risking the loss of the drug. Interestingly, the FDA’s implicit authority may prove more influential than its explicit authority. Since the FDA may not want to actually go through with its explicit powers to withdraw the drug, it may be more effective to go through other routes. If the agency threatens to inform the media that a company refuses to change its label, reputational concerns may bring a company to action. Similarly, if the FDA were to slow down the company’s future NDA’s, this may be too hefty of a price to pay. The bottom line is that the FDA still has a lot of weapons to force labeling changes, although explicit direct authority would certainly make it a lot easier.


In the aftermath of the Vioxx and Celebrex controversy, no other call to reform has resounded as loudly as the cry to give the FDA the authority to require post approval testing of drugs, especially longer-term studies on safety. During the Congressional hearing on Vioxx and the Senate Committee meeting on drug safety, expert after expert, and congressmen after congressmen all expressed the same sentiments: give the FDA more power, especially the power to require post-approval testing. But it’s hard to imagine that the broad powers conferred to the agency under the Federal Food, Drug, & Cosmetic Act (“the Act”) would not cover such authority. In fact, a more thorough examination of the Act reveals the possibility that indeed the agency does have the authority to require post approval studies to be conducted by drug companies. But even if this is true, the FDA should consider the consequences of exercising such authority.

1. Requirement of Post-Approval Studies upon Condition of Approval

There are circumstances in which the FDA clearly has the authority to require post approval studies from drug companies. One such instance is in the situation of an expedited review process. Under Section 506(a) of the Act, the FDA may expedite the review process if the drug is intended to treat a “serious or life-threatening condition” and such a drug “demonstrates the potential to address unmet medical needs for such a condition.”[406] But under section (b) of the same statute, the FDA can condition such expedited review by requiring the sponsor to conduct post-approval studies.[407] Not only is this an extremely explicit grant of authority, it also fits nicely with the rest of the statute. The FDA created an expedited process to allow much needed drugs to reach patients quicker, but in exchange, the agency retained many safeguards to keep the process in check. For instance, when a drug goes under expedited review, the drug sponsor has less procedural protections if the FDA decides the drug should be suspended or withdrawn.[408] Thus, drug companies that go under an expedited review accept the “easy come, easy go” approach.[409]

Fast track review usually means that Phase II and III of the IND are collapsed so that the shorter safety-oriented Phase II studies are merged with the longer more detailed Phase III studies.[410] In theory, the quality of the data should not be diminished because post-market studies can be required from fast track drug sponsors.[411] Both Vioxx and Celebrex were granted expedited review by the FDA.[412] By this fact alone, the FDA clearly had the authority to condition approval on the requirement of post-approval studies. The FDA clearly knows how to assert this authority because between 1970 and 1984, the FDA required post-approval research for more than 50 new drugs before approving such drugs.[413]

In addition to fast track review, under Regulations §310.303(a), the FDA has the authority to require additional studies upon the approval of certain drugs. The regulation states that with drugs that must be used for long periods of time, even a life-time, extensive animal and clinical studies are required as a condition of approval, but if the therapeutic usefulness of such drugs warrants that approval should not be delayed upon completion of such studies, the FDA may approve the drug on condition that the such studies will be conducted post-approval. Although the authority for conditional approval is clear under this regulation, it is doubtful that Vioxx and Celebrex would have qualified as such a drug under the regulation since it contemplates drugs that must be used for long period of time, and Vioxx and Celebrex were not approved for long-term or chronic use.

It is unclear whether Phase IV studies were required upon approval of Vioxx and Celebrex. The approval letter for Celebrex states, “We remind you of your Phase 4 commitment specified in your submission dated December 29, 1998. This commitment is to study the effects of Celebrex on using a protocol agreed to by the review Division.”[414] Although referring to a Phase IV commitment, the letter does not specify what the commitment is, which is in contrast to a subsequent approval letter for Celebrex, when the drug was approved for use in the reduction of adenomatous colorectal polyps. That letter clearly and explicated stated the type of Phase IV studies that were required, including a randomized controlled trial with adolescents aged 12 to 19.[415] On the other hand, no post-approval studies were mentioned in the approval letter of Vioxx.[416] So, it seems as though at least some obligation of post-approval studies was required for Pfizer, but it also appears that none were required for Merck. Why not? Perhaps the agency felt that the efficacy and safety were amply demonstrated during the review process of Vioxx. But members of Arthritis Committee that had recommended the approval of Vioxx and Celebrex clearly stated the certain post approval studies should be conducted if the FDA would require post approval studies, including pediatric studies, studies on the elderly, and studies on dose creeps.[417] Conceivably, the agency may have felt that the standard post approval process was sufficient. According to Janet Woodcock, the agency remains involved with a drug even after the drug is approved and are involved as new trials are conducted.[418] In regards to post-approval, Woodcock notes that “there’s a very active process going on with new indications being added, new warnings. It is really a continuum between the pre-market and the post market.”[419]

While the FDA clearly had the authority to condition the approval of Vioxx and Celebrex on the requirement of certain post-approval studies, it chose not to do so for Vioxx, and it is unclear how much it required of Pfizer. Even assuming studies had been required for both, it may not have made any difference at all within the Vioxx and Celebrex controversy. Even if the FDA had conditioned approval on the requirement of post-approval studies, short of requiring broad studies done on safety, which would be wide-ranging and costly, there is no indication that agency would have requested that studies on cardiovascular safety to be conducted.[420] In fact, the Arthritis committees that approved the drugs did not specifically recommend Phase IV studies on CV risk. Therefore, the power conferred to the FDA under Section 506(b)(2) may prove not so useful after all if the agency was not aware of certain risks at the time of approval. Given that many safety issues are unforeseeable at the time of approval, the more important question becomes whether the FDA has the authority to require post approval studies as issues arise. Specifically, did the FDA have the authority to require Merck and Pfizer to conduct post approval studies after there was indication of an increased heart risk? The answer to this question is less clear.

2. Requirement of Post Approval Studies as Issues Arise

Although conditional approval of expedited drugs upon post-approval research is an authority explicitly conferred by statute to the FDA, there was substantial debate on whether the statute conferred authority on the agency to require post-approval studies in general. Even before Section 506(b)(2), which was enacted in 1997, the agency from time to time had required post-approval studies under Sections 505(e) and (k), but even these instances were in the context of a condition upon approval.[421] However, Sections 505(e) and (k) simply grants authority to the Secretary to require records and reports after the approval of an NDA to determine whether such approval should be withdrawn and makes no mention of conditional approval. The old debate was over what was contemplated by “records and reports,” and whether it includes clinical trials and other studies or whether it pertains only to data collected in the normal course of business.[422]

Recently promulgated regulations for Section 505 of the Act have shed new light to this issue. Under 21 CFR § 310.303, titled, “Continuation of long-term studies, records, and reports on certain drugs for which new drug applications have been approved,” there is a clear statement that “a proposal to require additional or continued studies for a drug for which an NDA has been approved may be made by the Commissioner.”[423] So clearly, the Commission has the authority to require these additional studies for a drug that has been approved. However, we come back to the old debate over whether the Commissioner is limited to in his ability to assert such authority to certain contexts, since part (a) of the regulation talks about additional studies in the context of conditional approval of a drug.

A literal reading of part (b) the regulation suggests that the Commissioner is not limited to assert this authority only in the context of conditional approval because the language is in the past tense. “A proposal to require additional or continued studies with a drug for which a new drug application has been approved may be made by the Commissioner...”[424] If a drug has already been approved, a proposal to require additional studies by definition can no longer be a condition of the approval.

A substantive reading of the statute separately and as a whole (parts (a) and (b)) also indicates that the Commissioner is not limited in his authority. It is true that § 310.303(a) describes the situation of conditional approval. It states that in cases where drugs need to be consumed for long periods of time, even a lifetime, extensive animal and clinical tests are required as a condition of approval, but because of the therapeutic usefulness of such drugs, approval should not be delayed pending completion of such long-term studies. In these cases, the FDA may approve the drug on condition that the long-term studies will be conducted. The regulation goes on to state that the procedures required by § 310.303(b) will be followed in order to list such a drug. Thus, the regulation states that conditional approvals of drugs upon post-approval studies are okay in certain situations, and that such conditional approvals must follow the procedures of part (b), that is, the proposal must be proposed by the Commissioner, the applicant will have an opportunity to meet with FDA officials, and requirements will be published pursuant to § 310.304.

But the substantive rule of part (a) has nothing to do with the substantive rule of part (b). Part (b) unequivocally states that additional studies can be required after an NDA has been approved . Essentially, this is describing the situation of when issues arise after a drug has been approved . If this happens, follow the procedure, which is delineated in the rest of part (b), which is the same procedure required for additional studies from part (a). Thus, conditional studies from part (a) must follow the procedure of part (b), and additional studies from part (b) must also follow the procedure of part (b).

Any interpretation that part (a) delineates the types of additional studies that can be required, which are conditions upon approval, and that part (b) simply sets out the procedure for those studies, or that the studies stated in part (b) are the same studies stated in part (a) is wrong. Part (a) simply gives the Commissioner the authority to require additional studies as a condition upon approval of the drug. Part (b) gives the Commissioner the authority to require additional studies after the approval of the drug. Part (b) also lays out the procedure for requiring additional studies whether as a condition upon approval or after an NDA has been approved. Reading both parts together make sense because it contemplates all situations where additional studies may be required--as a condition upon approval and after a drug has already been approved.

Of course, regulations are the FDA’s own interpretation of Congressional statutory provisions, and regulations are often challenged in court. When a court evaluates an agency interpretation’s of statutory law, Chevron criteria is used. Under Chevron , there is a two step inquiry: (1) Has Congress addressed the precise question at issue, and if not, (2) is the agency’s interpretation reasonable?[425] Suffice it to say that the Chevron test is highly deferential to an agency’s interpretation of law, and if challenged, a court would likely find that the FDA’s interpretation of Section 505(k) of the Act by way of 21 CFR § 310.303 is reasonable. However, there still exists the possibility that Regulation § 310.303(b) is an unreasonable interpretation of Section 505 of the Act because it is overbroad. Under Section 505(k), records and reports must be submitted so that the Secretary can make a determination for withdrawal of the drug under Section 505(e). But Section 505(e) allows for withdrawal in specific instances, mainly dealing with the safety and efficacy of the drug based on new information. Therefore, 21 CFR §310.303(b) may not be a reasonable interpretation of the FDA’s authority under Section 505(k) of the Act if it allows for the Secretary to require post-approval studies without limiting such requirements to test the safety or efficacy of the drug. The reasons for requiring post-approval studies are not stated in 21 CFR §310.303(b) and therefore seem limitless. However, the FDA can also assert that assuming 21 CFR §310.303(b) is limited to safety and efficacy studies is a reasonable reading of the regulation. To be safe, the regulation should be amended to narrow the authority to require testing to only issues of safety and efficacy of the drug in order to avoid being overbroad and thus an unreasonable interpretation of Section 505(k). Under the weak requirements of Chevron , the agency probably does not have to go so far as to show reasonable cause or why it is requiring specific safety or efficacy studies be conducted so long as they are related to safety and efficacy.

3. Incentive For Compliance through the Tort System

Unless the FDA asserts its authority to require post-approval studies, it has to rely on the voluntary cooperation of drug companies, which have every incentive to delay testing because testing can hurt sales, lead to withdrawal, and expand the duty to warn.[426] Additionally, testing data may be subject to litigation in the future.[427] If safety issues arise, and drug companies do not want to conduct additional studies, the FDA’s only credible threat at that point would be to withdraw the drug[428] , an outcome that may be undesirable for all.

Regardless of legal authority, Neil Cave has suggested that the agency should set up incentives for drug companies to voluntarily perform post-approval studies, and this may in the end be the most efficient means to accomplish such studies. Cave notes that the current general tort system is set up in opposition to these incentives. Because there is no independent duty to test, and because current products liability law generally does not hold a drug company liable for injuries on the basis of information that did not exist at the time of injury, the drug company has no incentive to discover adverse information.[429] Cave’s solution is a form of specific immunity to liability so that following completion of a “conclusive” test, a drug company should be immune to liability for injuries that are the subject of that test, since the FDA will either withdraw the drug or re-label the drug, thus informing the public and reducing risks.[430] Specific immunity is efficient because as new patient injury reports come in and drive the cost of injury above the cost of testing, the drug company will have an incentive to test the drug quickly to avoid the accumulating cost of liability.[431]

Cave’s argument is interesting but relies too much on economic incentives. Because the post-approval regime right now is a passive surveillance system, many adverse events go unreported. Therefore, it may take much longer for injuries to rise to the level where it is above the cost of conducting a study, especially given the large costs of such studies, and in particular, useful long-term studies. There is also an adminstrability problem because the idea of reforming the tort system, a system that differs from state to state, may not be realistic. While giving incentives to encourage voluntary post-approval studies is important, it needs to be complemented with statutory authority to require such studies when incentives don’t work. Last, economic analyses may not be proper in the context of safety because the FDA does not put a dollar value on human life. For the agency, it may be the case that one life taken away is too costly, and therefore all serious injuries should be examined, not simply those injuries that add up to a value drug companies consider to be a quantifiably sufficient value.

4. The Cost of Asserting Authority

A practical consideration that should not be forgotten is the role of cost. Asserting the authority to require post-approval studies will be extremely costly on the industry. Before a Senate Committee, Dr. Raymond Woosley has expressed great apprehension about the role of cost and the fate of the drug industry and what it would mean for the country’s health. According to Dr. Woosley, drug companies are now investing more money than ever on R & D, but the number of significant new products submitted to the FDA has fallen by 50 percent.[432] Tens of billions of dollars were lost when 17 drugs were withdrawn from the industry, and hundreds of lawsuits threaten the future of the drug industry and therefore the innovation of new medicine.[433] Dr. Woosley implored the Committee to remember that actions may have unintended consequences. Adding the requirement for post-approval studies without other important changes to the entire system to balance it out, “could be catastrophic”.[434] While he absolutely supports a more active and rigorous post-marketing surveillance system, Dr. Woosley suggests that the added costs of post-approval testing should be offset by simultaneous lower costs during the approval process.[435]

The FDA cannot be unaware of the enormous costs of conducting post-approval studies, especially in light of the sky-high costs of successfully bringing a drug to the market, not to mention that cost of drugs that fail to be approved. The case of Vioxx and Celebrex was no different. Presently, Merck’s Vioxx –related lawsuits threaten the fate of the company, although it is among the best and most innovative drug companies in the world. Perhaps the FDA’s reluctance to require post-approval studies has been a recognition of the cost of such studies, which could make the cost of creating and selling drugs exorbitantly high.

Thus, the call for reform of the FDA should be followed by a call for more funding for the agency. The current passive surveillance post marketing regime is relatively cheap because it depends on self-reporting by patients and physicians. An active surveillance system would help the FDA gain a more accurate understanding of the risks of drugs. In this way, the agency can better require a more narrow set of post-approval studies, which would help drive down the cost of these studies. Furthermore, increased funding would mean that the FDA can conduct some its of own safety studies (or co-fund some of the studies), which would not only ease the financial burdens of the drug companies but would also allow for studies to be performed by “neutral” outside experts.



In the face of COX-2 troubles, Merck and Pfizer took very different actions. One pulled its drug off the market while the other suspended all direct-to-consumer advertising and made a small labeling change. Who made the right decision? In making this determination, two points must be kept in mind. First, no drug is risk free. The choice on whether a drug should be available is necessarily based on a cost-benefit analysis, and this is the proper inquiry in evaluating the decisions of Merck and Pfizer. Secondly, it must be remembered that a caveat remains in that Vioxx and Celebrex are two different drugs, and therefore did not have identical risks and benefits.


1. A Public Assessment of Costs and Benefits

Within the life of a drug, the guidance and decisions of the FDA are essentially cost-benefit analyses, some of which are implicit and some of which are explicit. The first step however, is left to the sponsor of the drug. In pursuing a new drug, the sponsor has to decide whether the benefits are likely to exceed the risks.[436] The sponsor will then make a proposed human study and submit it to the Institutional Review Board (IRB) of a clinic or hospital where the first human experiments will take place.[437] Then an investigational new drug application (IND) is filed with the FDA, and a FDA medical officer will determine whether the research will have benefits exceeding the risk, and if the decision is positive, the study may proceed.[438] If the decision is negative, the study is stopped.[439] When the IND stage is passed, a new drug application (NDA) is filed, and medical reviewers, often an advisory committee will review the NDA, and look to the costs and benefits in order to recommend approval or not.[440] The FDA usually adopts the recommendation of the reviewers.[441] In post-approval stages, FDA epidemiologists also apply a risk benefit analysis to adverse experiences from reports and post-marketing studies.[442] When risks outweigh the benefits, the FDA will withdraw the drug or the sponsor may voluntarily withdraw.[443]

This short review of the guidance of the FDA illustrates that the agency is constantly making new cost-benefit analyses as it accumulates new information on a drug. Whereas most people do not have the time, resources, and money to make a thorough cost benefit analysis before taking a drug, the FDA is in the best position (in terms of expertise, resources, and mission) to make such an analysis, and public confidence has been conferred upon the agency to make well-informed decisions. Because this authority is delegated to the FDA, the question arises of whether anyone else should be able to make such a cost-benefit analysis. Can drug companies make this analysis to decide what goes on and what goes off the market? Can patients make this analysis to decide what drugs they will take? More specifically, can patients override the decision of the FDA to make their own cost-benefit analysis specific to them? On the one hand, it is good to let the FDA perform the analysis because they are looking at the nation as a whole, but on the other hand, patients may want individualized cost-benefit analyses because they are affected more severely.

2. A Private Assessment of Costs and Benefits

Many believe that Merck did the public a disservice by overstepping the role of the FDA and withdrawing Vioxx from the market. Say New York Times reporter Gardiner Harris, “Merck decided by itself that the drug should be withdrawn.”[444] Even Kweder of the FDA admits that the agency would have looked more closely at the APPROVe trial and performed a cost benefit analysis to determine if withdrawal was warranted.[445] Says Kweder, “We haven’t done that. We wish he had had the opportunity to do that.”[446] Moreover, the Arthritis Committee recently recommended that Vioxx was safe enough to bring back on the market. The Committee reviewed a tremendous amount of data from studies and reports on the various drugs, assessed the risks and benefits, and determined that the benefits of Vioxx and Celebrex both outweighed the risks. During the first two days of the meeting, the panel members said that they believed that all COX-2 inhibitors carry heart risks.[447] Then the panel turned to the benefits to compare with the risks.[448] In the end, the panel decided by varying margins that the potential benefits of the drugs outweighed the increased heart risks.[449]

Dose the Arthritis panel’s support to bring back Vioxx imply that Merck made the wrong decision? The answer remains unclear because the vote to bring back Vioxx was marginal, 17-15, as opposed to Celebrex’s overwhelming support of 31-1.[450] When the votes are this close, Merck’s own cost-benefit analysis leaning toward withdrawal seems more reasonable. However, the appropriate inquiry may be whether Merck should ever be able to make this decision. While the FDA cannot force a drug company to sell a product, there certainly are dangers of introducing a drug on the market, having patients rely on it, and then withdrawing it from the market. Merck after all, is a self-interested for-profit corporation, and therefore, its cost benefit analysis is by nature different from that of the FDA. It is concerned with reputation, tort liability, the cost of post-approval studies, and a host of other factors. Says Alan Reynolds, a fellow at the Cato Institute, "I am now denied the opportunity to use Vioxx responsibly, not to eliminate even the slightest health risk (which is impossible), but in the hope of reducing litigation risk."[451] Reynolds poignantly points out the heart of the issue—Merck and other drug companies have a completely different assessment of costs and benefits because of private costs, and therefore, in the end, the decision to withdraw a drug from the market, should be left in the neutral hands of the FDA. Alternatively, because these extra costs change the cost-benefit analysis and because of patient reliance, perhaps drug companies should not be allowed to unilaterally withdraw a drug unless FDA agrees to such withdrawal or when the company can show severe financial distress. After all, cost should never be wholly left out of the equation, or the incentive to innovate will suffer.

It seems like Celebrex, on the other hand, waited for the FDA to do a cost benefit analysis before taking action. In a decision involving the FDA, the company suspended all direct-to-consumer advertising. It also made a minor labeling change. Such concerted efforts convey the sense that Pfizer allowed the FDA to call the shots. However, the company may have in fact, performed its own cost-benefit analysis before allowing the FDA to take over. In defending its decision to keep Celebrex on the market, Chief executive of Pfizer, McKinnell stated that the drug was an appropriate drug for many patients and that the results of APC were “inconsistent with the great body of data...accumulated over 10 years.”[452] Would Pfizer have acted in the same manner if the risks were higher? For instance, if three studies showed an increased heart risk, would the company have taken the situation into its own hands and withdrawn the drug? Regardless of the answers to these questions, it seems quite likely and natural for drugs companies to make their own cost benefit analysis in light of new found risks. But should they be allowed to? While this is difficult to evaluate in the Celebrex context, the Vioxx situation is a good illustration of when this question becomes important, since the risks and benefits were so close in that context, and given the opportunity, it seems likely that a self-interested company will always lean towards the decision that is less costly.

3. An Individual Assessment of Costs and Benefits

Finally, should patients be able to make a cost-benefit analysis for themselves? During the FDA Arthritis Committee meeting in February of 2005, patients told panelists that Celebrex and Vioxx were the only drugs that worked for them.[453] In fact, many patients have echoed this sentiment all over the country. The website www.bringvioxxback.com is solely dedicated to the goal of bring Vioxx back on the market.[454] The website is an “independent patient-driven organization” that believes that Vioxx should return to the market with black box warnings accompanying its labeling.[455] The site contains many statements of individual assessments of costs and benefits and proclamations of the freedom of choice to make such individual analyses. One such statement says, “If you live with intractable pain every day of your life, would you take a (small) chance that you would have a heart attack?”[456] There are many similar statements by doctors and editorialists on the website, along with a collection of individual stories and powerful statements made by patients around the country.[457]

Should these patients be allowed to make their own choices and trade off risks and benefits in the limited situation where only one drug works for them? I do not purport to extend the question to include the situation of patients being able to choose even when other drugs are effective for them. Although it seems logical to let the FDA make an overall decision that is good for the entire nation (which speaks to the point of having a federal drug regulatory agency in the first place), the heavy price of such an over-enforcing policy will fall squarely on the shoulders of a limited group, who are thereby making a sacrifice for the health of others. On the other hand, if dangerous drugs were allowed to stay on the market, it would elevate the interests of the limited group over the health of the nation, which may be an intolerable under-enforcing policy. Proposals by the FDA panel suggest a compromise on the matter. According to New York Times reporters, “Cost-benefit analyses for drugs are rarely as explicit as they were yesterday [at the panel].”[458] They were referring to the panel’s recommendation that no medicine is risk free and that sometimes patients should be allowed to choose medicines for non-life threatening diseases, even when there are dangerous side effects.[459] Panel chair, Dr. Alastair Wood, expects severe restrictions on the use of Vioxx, with distribution confined to very limited patient populations.[460] Other panel members agreed that cox-2 inhibitors should only be prescribed as “last resorts,” although what exactly this entailed was not agreed upon.[461] The overall message was that cox-2 drugs should only be available to patients who need them most, which would strike a balance between the interests of those who need Vioxx and the health of the nation. The importance of such balancing (through labeling and safety surveillance) has been heavily advocated by Abbey Meyers, the President for National Organization for Rare Disorders, as the split interest between what she has dubbed as the two FDA constituencies, those who are healthy and those who have serious diseases.[462]

4. Conclusion

In the end, any realistic patient freedom of choice method will be operated within the boundaries of FDA regulation. Even if certain populations of patients are allowed to take a drug, it is the FDA that defines the boundaries of such uses and restrictions. The decisions of whether a drug company is allowed to market a drug or whether patients should be allowed to take a drug is properly regulated by the FDA because the agency is in the proper position to balance the health interests of all. But we should not forget that patients are always allowed to make individualized cost-benefit analyses when they decide to stop taking a drug. We are only left with the question of whether drug companies should be allowed to make their own cost benefit analysis and withdraw a drug, especially in light of repercussions on patients. The FDA clearly has allowed drug companies to voluntarily withdraw in the past, but the justifications are unclear. It seems precarious to allow a drug company to make a cost-benefit analysis when its interests are completely different from the agency’s interest, namely monetary interest as opposed to health interests. Perhaps the FDA has allowed such withdrawals because in the end, it is the drug company that is held liable for any actions regarding the safety of the drug. On the other hand, perhaps it should be emphasized that drug companies should take into account the extra implied costs of being in the drug industry, including a responsibility to take into account patient health concerns in withdrawing a drug.


During the FDA Arthritis Committee proceedings in February of 2005, Merck announced that it might bring Vioxx back on the market.[463] Merck’s research chief, Dr. Peter Kim, stated that if the panel concluded that similar drugs to Vioxx were also associated with increased heart risks, then Merck would take that information into consideration, implying that Vioxx might be relatively safe enough to bring back.[464] When asked whether Merck would bring Vioxx back because there was a conclusion of a COX-2 class effect on increased heart risks, Dr. Kim did not answer directly but stated that, “there are unique benefits to Vioxx.”[465]

Does Merck’s possible intention of bringing Vioxx back imply that the company is conceding its mistake of pulling it off the market in the first place? Perhaps. It seems logical to think that Merck has now realized that the withdrawal of Vioxx may have been a drastic measure, especially in light of Celebrex’s less radical response. On the other hand, Merck may have acted prudently with the reasonable belief in September of 2004 that only Vioxx produced an increased heart risk. With the discovery that Celebrex and Bextra also produced such risks, Merck was then free to retract its severe reactions. But if the risks are still the same, why should Merck act any differently? If the company believed it was dangerous enough to pull off the market in September of 2004, why should the discovery of the similar risks of other similar drugs make any difference? Perhaps this tension can be mitigated if Merck was merely overcompensating or was purposely overly-cautious when it pulled Vioxx off the market, believing that though not terribly dangerous, there were other safer alternatives to Vioxx, including Celebrex and Bextra; therefore withdrawal was best at the time . But with the finding that both Celebrex and Bextra were also associated with increased heart risks, such drugs were no longer safer alternatives but simply competitors. Regardless of the reasoning and whether withdrawal was the right decision at the time, the market return of Vioxx will necessarily imply that the drug was always safe enough to stay on the market.


The overwhelming support to keep Celebrex on the market suggests that Pfizer did the right thing by leaving the drug on the market despite the results of the APC study. On the other hand, the recent panel recommendations indicate that Pfizer probably had not taken enough action at the time. One recommendation from the panel was to issue harsher warnings about heart risks on the label. Pfizer never changed its label to reflect the results of the APC study and opted for a weaker label change, which merely emphasized that there is no problem if used for treating arthritis in recommended doses.[466] Although Vioxx did not promptly place the results of VIGOR and the finding of an increased heart risk under the “warnings” section of its label, the information can be found on the label, albeit in a more inconspicuous place on it.[467] Another recommendation of the panel was to ban consumer advertising, and Pfizer had already pulled all direct-to-consumer advertising of Celebrex. Last, the panel recommended that each drug should be accompanied by a guide outlining the risks. Again, Pfizer fell short here because it did not clearly outline or even list the risks of the increased heart risks on the labeling. Merck, at least, put it out there.

Despite these similar and dissimilar approaches, in the end, the reactions of Merck and Pfizer were as different as day and night. While Merck arguably went too far, Pfizer doesn’t seem to have gone far enough. Did Pfizer simply make a lucky guess? Or rather, did Pfizer make a better guess, taking into account the possible risks and benefits? Unfortunately, we can only examine concrete actions and end results, and such actions and results indicate that Pfizer suffered less reputationally and financially, compared to Merck, by keeping Celebrex on the market. From the patient aspect, it is difficult to evaluate who did the right thing. Pfizer allowed patients to keep taking the drug, but there was probably harm caused by leaving it on the market, especially in light of arguably inadequate warnings. Merck did not give patients a choice when it came to drug use, but it probably prevented harm by pulling the drug. However, looking from a regulatory and national good viewpoint, it was a decision better left up to the FDA, so even from the patient viewpoint, Pfizer arguably did what was better for patients by allowing the FDA to make the decision for the good of the nation.

Pfizer made a decision to stand by its drug when things looked bad for Celebrex. Some thought it was a very risky not to pull Celebrex off the market and some even believed that Pfizer was multiplying its legal and financial problems by not withdrawing the drug.[468] In the end, Pfizer’s CEO Henry A. McKinnell seems to have been most accurate in defending keeping Celebrex on the market by saying that it was an appropriate option for many patients.[469] This seems to parallel the sentiments of the FDA panel, which agreed that many patients benefited uniquely from the drug and decided that controls on usage and control on patient population would be the best methods of ensuring safety.



The common question in the mind of physicians, patients, regulators, Congressmen, and the general public is, “How can we prevent future recurrences similar to the COX-2 problem?”. It seems everyone has their own view on what should be done. A congressional hearing was held after Vioxx was withdrawn from the market, and more recently a Senate committee hearing was held after the FDA Arthritis Committee convened to determine the fate of the COX-2 inhibitors. During these hearings, researchers, regulators, and Congressmen, all made suggestions for reform, in order to prevent another COX-2 problem in the future. Although many suggestions overlapped, there were also conflicting viewpoints, especially in regards to the duties of the FDA. At the forefront of blame in the controversy, the FDA has received a lot of criticism for its role in the approval and post-approval monitoring of Vioxx and Celebrex.

1. FDA Should Assert Its Authority in Regards to Re-labeling and Post-Approval Studies

One of the most common calls for reform is to give the FDA more authority in terms of dictating labeling changes and requiring drug manufacturers to perform post-approval studies. It can be argued that the FDA already has the authority to do both. However, it is important to note that in the COX-2 controversy, the FDA has chosen not to exercise its authority in terms of both. This could imply that the agency simply did not believe that such labeling changes and post-approval studies were necessary. Or it could also imply that the agency did not want to assert its authority for other reasons. Regarding labeling changes, part of the problem is that the FDA does not have direct authority to dictate labeling changes. Although the agency has the authority to withdraw a drug that is misbranded and non-properly labeled drugs are by definition misbranded, thereby allowing the FDA to threaten withdrawal unless labels are changed, this threat is not credible unless the FDA is willing to go through with withdrawal, and it may not be worth it for the agency to withdraw a drug because the maker will not change the label in a specific way. I believe this is why the FDA chooses rather to negotiate these labeling changes. In this way, it can keep the drug on the market and have labeling changes as well. I also believe that the FDA has the authority to require post-approval studies, but there are practical problems of cost, which may be the best explanation for the agency’s reluctance to assert this authority.

Assuming that the FDA has the authority to dictate labeling changes and to require post-approval studies, it is difficult to surmise as to why the agency has chosen not to assert such authority. One line of thought conjectures that the agency is acting in a utilitarian way, doing what is most efficient and easy. The other line of thought is that the agency is simply not aware of its authority. The repeated statement by the deputy director of the Office of New Drugs that the FDA does not have the authority to dictate labeling changes suggests that agency officials are not aware of the agency’s statutory authority.[470] Numerous comments about the lack of FDA authority to require post-approval studies were voiced during the Senate hearing, and not one agency official responded to the contrary, and in fact, Dr. Kweder of the FDA repeated the same sentiments.[471] It is this sense of unawareness that indicates that there may be more going on than simply a utilitarian and strategic assertion of authority.

I would recommend that the FDA assert its authority to establish precedence. In the future, whether the agency chooses to assert its authority in specific situations may be a matter of utilitarian discretion, but bargaining power is always higher when agency action and threats are backed by statutory authority. Harvard Medical School Professor, Dr. Avon, has stated that, more than money, the agency needs to have courage to insist that drug companies take action in response to questions of safety.[472]

2. Improving Communication of Safety Information to the Public

Another recommendation I would make is to improve the lines of communication from drug companies and the FDA to the public. With all the talk about labeling changes, there is still a real question in regards to the effectiveness of these labels. Do patients read the labels? If yes, do they understand them? Do they take warnings seriously? Do they in fact make their own individual cost benefit analyses? Perhaps even more importantly, do doctors read them? Dr. Cecil Wilson, a member of the board of trustees of the AMA, has proposed several reforms in this respect. First, the FDA should finalize their proposed rule on modifications to the format and content of the insert in order to make them more user-friendly.[473] Secondly, although the package insert and post-approval surveillance should be the risk management plan for most drugs, when this is insufficient, communication to physicians should be the primary means to reduce the risks of drugs.[474] Last, high level risk minimization tools, such as patient forms or performance-linked accessed systems, should be used as last resorts.[475] Assuming that labeling is an effective source of communication, these proposed reforms should be adopted.

In addition to labeling, there is still a need for faster communication. Labeling changes are slow, and it takes some time for a change to settle in because labels have to be printed and shipped out, while old labels are still floating about in the general public. This issue has been identified by Janet Woodcock of the FDA as a reason for her proposal to open more direct channels of communications.[476] Her proposals include a new Drug Watch webpage that will post new information about serious side effects and other safety risks for previously and newly approved drugs.[477] I would endorse this proposal with the qualification that updated information, such as results from studies, are well-balanced and accurately reported so as to not create an epidemic of fear every time new information is released. Along with this proposal, I would add the suggestion that all studies should be available to the public. Drug companies have long held studies from public scrutiny by asserting limitations or flaws of such studies. But, as long as those limitations are noted, there should be no problem with disclosing the results of such studies.

3. No Outside Independent Review of NDAs

The Bush administration has recently announced that it will set up an independent Drug Safety Oversight Board to monitor FDA-approved drugs and inform physicians and patients of any new safety information about the drugs.[478] The board’s primary function will be to monitor drugs on the market, but it will have some authority to challenge the safety of drugs that have yet to be approved.[479] The importance of the board is that it will deliver findings quickly to the public by posting new safety information on its webpage.[480] The board will include members from of the FDA, the Department of Health & Human Services, and other governmental units. Consultants will include medical experts and consumer groups.[481] With enough funding, I believe that this board is a step in the right direction in terms of better and quicker communication to the public.

I would however, oppose, independent review of NDAs from any outside group, including this board, although the proposed oversight board is not technically an outside group since members of the FDA will also serve on the board. I do not believe that an outside group would add much value to the approval process in terms of reviewing safety. As Dr. Fleming has noted, FDA experts have experience and familiarity with the drug approval process, including the limitations of the authority of the FDA.[482] There are also serious conflict of interest issues for those outside the FDA who may serve on a separate safety group.[483] Dr. Woosley, who once supported an independent drug safety agency has changed his mind, and now only supports the idea of an agency that would look at a drug’s use in general, not one that conducts safety analyses.[484] His altered opinion is due to his belief that the determination of drug safety, which requires a cost-benefit analysis, should properly be left up to the FDA, which makes assessments of costs and benefits on an ongoing basis and has a “corporate memory” of the drug and issues relating to the drug.[485] Also, the FDA is able to better assess a particular drug because it has knowledge about the safety of other drugs related to it and in the same class of drugs.[486]

4.Better Drug Studies and More Congressional Funding

The final two changes I would recommendation go hand in hand--conducting larger and longer studies of the safety of drugs and appropriating more funds to the FDA. There is simply no way of knowing the risks of a drug unless the drug is tested. Pre-approval studies with a patient database in the thousands will never be as accurate as post-approval studies with a patient database in the millions. The FDA’s post-approval system should be one of active surveillance, not the passive reporting surveillance system currently in place. The most cost effective way to gather information on millions of patients is to have a database system that will track the use of drugs. Large observational studies are another way of analyzing a large patient population. Clinical studies are perhaps the most accurate but will always have much smaller patient populations and are the most expensive studies to conduct. A combination of all three of these types of active post-surveillance methods should be employed. However, funding must be a priority for any of this to be possible. The FDA has been historically under-funded relative to its function, duties, and goals. Cries for reform simply cannot happen without the appropriations from Congress. There is of course industry support, but that can only go so far. The reason why Phase IV studies are often not required is because of their costs, which would strain pharmaceutical companies financially, which would lead to other negative consequences, including a disincentive for the innovation of new drugs. Thus, there is always a fine balance between innovation and safety, and pharmaceutical companies, already faced with enormous development costs, should not have to fund drug safety alone. This is a shared responsibility between drug companies, who profit from the sales of drugs, the government, who is responsible to the public, and the people, who benefit from the innovation of new drugs.


When Merck pulled Vioxx off the market, there was quite a stir among the FDA, the drug industry, Congress, and the public. But it was the same kind of commotion that accompanies any drug withdrawal. It was not until Celebrex came into the picture where the story took an expected turn and became interesting. The results of two different cancer treating efficacy studies both indicated that Vioxx and Celebrex were associated with an increased heart risk. However, while Merck chose to withdraw Vioxx from the market, Pfizer decided to keep Celebrex put, choosing instead to suspend all direct-to-consumer advertising and make small labeling changes. Such drastically different approaches beg an analysis of the situation, including why Merck and Pfizer chose their respective actions, what was at stake, what resulted from such actions, and finally, who was right. At the forefront of drug approval and safety, the FDA was also implicated in the controversy, and the agency’s actions have come under intense scrutiny.

I have attempted to construct a plausible story for what happened by examining the different parts of the timeline of events. I started by looking at the backdrop of the controversy, including the approval process of Vioxx and Celebrex, summary of medical studies, and an analysis of such studies. Next I analyzed the different approaches taken by Merck and Pfizer, namely, withdrawal, re-labeling, and suspension of advertising. Using a combination of deconstruction tools, including, cost-benefit analyses, limitations of FDA authority, and economic incentives, I constructed possible theories for the actions taken by Merck, Pfizer, and the FDA. In the end, it must be noted, and I acknowledge the vast limitations in such theories. Accordingly, I turned my attention to a normative analysis, starting with an evaluation of the actions of Merck and Pfizer and ending with inevitably, a critical review of the actions of the FDA and concluded with a series of suggested reforms for the agency.

On a superficial level, Merck seems to have come out the loser. Its withdrawal appears to have been done in haste in light of the recent FDA panel vote, which concluded that Vioxx was safe enough to bring back on the market. The list of wrongs could be extended. After all, Merck withdrew Vioxx when the FDA had not had the opportunity to make its judgment on the matter. It withdrew Vioxx although re-labeling was clearly an option before any FDA decision. It withdrew Vioxx in spite of the number of patients who relied on the drug when nothing else would work for them. It withdrew Vioxx despite the proven GI benefits of the drug. The story ends with a statement from Merck that the company is thinking about bringing Vioxx back on the market. One has to wonder whether this is a concession that withdrawal was wrong. One also has to wonder whether Merck has been thinking with its pocketbook rather than the good of the public. After all, there is an inherent problem in allowing a drug company to make voluntary withdrawals because it implicates a personal cost-benefit analysis. Recall the major difference between the FDA’s cost benefit analysis in determining withdrawal and one conducted by a drug company is that the drug company will introduce private cost factors, including reputational and litigation costs, not considered by the FDA. In the end, a determination of withdrawal is best and most neutrally made by the FDA, and this is the extent of my criticism of Merck.

While Pfizer appears to have come out the winner in the controversy, it can be argued that the company did not take enough action. Although Pfizer probably made the right move by waiting for agency directions and suspending direct-to-consumer advertising, its initial labeling changes were pitiable. It is also unfair to chide Merck for possibly making a cost-benefit analysis and not Pfizer. It was simply easier to evaluate Merck’s decision. The big difference is that with voluntary withdrawal, Merck’s cost-benefit calculations are implied by way of formal company-sponsored actions, but with non-withdrawal, any calculations made by Pfizer are masked by formal FDA action.

I believe that Merck and Pfizer were in two different situations. Vioxx’s link with heart risk came first, giving Pfizer the opportunity to learn from Merck’s mistakes. Vioxx’s association with increased heart risk was found in several published studies, while only one published study found such an association with Celebrex. Although Pfizer’s initial labeling change was insufficient, the company made the right decision by allowing the FDA to step in and take over the situation. Merck’s ultimate mistake was acting unilaterally and clearly acting pragmatically depending on the climate of support for Vioxx and COX-2 drugs generally. The drug industry and the FDA should take away many valuable lessons from the Vioxx Celebrex controversy. No drugs are completely safe. However, careful monitoring and responsible action can do much to alleviate a public crisis. With regards to drug safety issues, informing the public should be the top priority, and this can only be done effectively with cooperation between the FDA and the drug industry. Both should work towards getting new safety information out quickly to the public. After that, FDA should step up as the trusted agency that it is and assert its authority to dictate labeling changes, require post-approval studies, or order withdrawal. Vioxx and Celebrex is not a story about the safety of drugs because all drugs are all unsafe in some respect. It is a story about responsibility, the responsibility of drug companies and the FDA to handle safety issues for the good of the public. What should have been handled quickly and routinely was dragged out in a frenzy of confusion because power lines were not drawn and bureaucratic complexity blurred events and decisions. I am hopeful that the problems with Vioxx and Celebrex will only lead to a better post-surveillance drug safety system and drug safety generally.

Update on Controversy :

On April 7, 2005, the FDA asked Pfizer to voluntary withdraw Bextra, its other COX-2 drug, and the drug company complied with the request.[487] The agency is also asking all makers of marketed prescription NSAIDS to re-label by adding a boxed warning and Medication Guide highlighting potential heart risks as well as adverse GI risks of such drug use.[488] Makers of non-prescription NSAIDS are being asked to re-label to include more specific information about potential heart and GI risks.[489] These labeling changes do not apply to aspirin.[490]

The analysis of such agency actions would be quite lengthy, and is beyond the scope of this paper. Whether these actions are warranted or not, they suggest a more assertive FDA, and it will be interesting to see if and how the agency will transform itself after the aftermath of Vioxx and Celebrex.

[1] Christopher Rowland, With Vioxx Gone, Now What Do I Do? Arthritis Sufferers are Blitzed with Ads from Advil, [sic] and Other[sic] Conflicting Commercials are Driving Patients into Their Doctors’ Offices for Advice , THE BOSTON GLOBE , Nov. 7, 2004 at C1.

[2] Gina Kolata, A Widely Used Arthritis Drug is Withdrawn , N.Y. TIMES , Oct. 1, 2004 at A1.

[3] Alex Berenson, Pfizer to Halt Its Advertising of Celebrex to Consumers , N.Y. TIMES , Dec. 20, 2004 at C1.

[4] Anahad O’Conner and Denise Grady, Pfizer and Celebrex: The Patients; Problems May Send Patients Back to Age-Old Aspirin , N.Y. TIMES , Dec. 18, 2004 at C1.

[5] Anonymous, Vioxx Three Times as Dangerous as Celebrex, Study Indicates , MEDICAL NEWS TODAY, December 7, 2004 at http://www.medicalnewstoday.com/printerfriendlynews.php?newsid=17435 . (Lasted visited Feb. 19, 2005).

[6] CDER 1999 Report to the Nation at 8; O’Conner, supra note 4, at C1.

[7] Id .

[8] Bombardier et al., Comparison of Upper Gastrointestinal Toxicity of Rofecoxib and Naproxen in Patients with Rheumatoid Arthritis , 343 NEW ENGLAND J. MED. 1520, 1520 (2000).

[9] O’Conner, supra note 4, at C1.

[10] Id .


[12] Gardiner Harris, F.D.A. is Advised to Let Pain Pills Stay on Market , N. Y. TIMES , Feb. 19, 2005, at A1.

[13] See HUTT , supra note 11, at 516.

[14] Hutt, supra note 11, at 515.

[15] Id . at 516.

[16] Id .

[17] Id .

[18] Pfizer at http://www.arthritis.ca/look%20at%20research/clinical%20trials/learning/default.asp?s=1 . (Last visited Mar. 25, 2005).

[19] Hutt, supra note 11, at 516.

[20] Anonymous, University of Rochester Sues for Celebrex Royalties , CHICAGO TRIBUNE , Apr. 13, 2000 at N2.

[21] Id .

[22] Bruce Jaspen and Susan Kuczka, Pfizer Cutting 1500 Area Job; Skokie, Mt. Prospect, Elk Grove Village Hit , CHICAGO TRIBUNE , Apr. 30, 2003 at C1.

[23] Rita Rubin, Pills on a pedestal Celebrex and Vioxx have been heralded as wonder arthritis drugs. But the FDA has safety concerns , USA TODAY , Feb. 6, 2001 at 1D.

[24] See FDA CDER, Transcript of Arthritis Advisory Committee, Dec. 1, 1998 available at http://www.fda.gov/ohrms/dockets/ac/98/transcpt/3480t1.rtf . (Last visited Feb. 18, 2005).

[25] See id .

[26] HUTT , supra note 11, at 519.

[27] See FDA CDER, supra note 24.

[28] Pharmacokinetics is defined as, “characteristic interactions of a drug and the body in terms of its absorption, distribution, metabolism, and excretion.” Merriam Webster Online at http://www.m-w.com/cgi-bin/dictionary?book=Dictionary&va=pharmacokinetics . (Last visited Mar. 25, 2005).

[29] See FDA CDER, supra note 24.

[30] See id .

[31] Id .

[32] Id .

[33] Id .

[34] Id .

[35] Id .

[36] For discussion of selective and non-selective NSAID, see infra Scientific Background for COX-2.

[37] FDA CDER, supra note 24.

[38] Id .

[39] Id .

[40] Id .

[41] Id .

[42] See FDA CDER, Transcript of Arthritis Advisory Committee, Review of NDA #21-042, Apr. 20, 1999 at http://www.fda.gov/ohrms/dockets/ac/99/transcpt/3508t1.rtf . (Last visited Feb. 18, 2005).

[43] The Office of New Drugs is part of the FDA Center for Drug Evaluation (CDER).

[44] Testimony of Sandra Kweder, Congressional Finance Committee, Nov. 18, 2004.

[45] See FDA, supra note 42.

[46] Kweder, supra note 44.

[47] Id .

[48] Id .

[49] See FDA, supra note 42.

[50] Id .

[51] Id .

[52] Id .

[53] Id .

[54] Id .

[55] Id .

[56] Id . Again, this labeling decision reflected the committee’s conservative tendencies. Although studies indicated that endoscopically diagnosed ulcers occur at a lower rate with Vioxx as compared to an NSAID comparator, comparison rates of clinically significant adverse GI events were less clear because of the small number of such events in the studies.

[57] See FDA, supra note 42.

[58] Id .

[59] Id .

[60] FDA Drug Approval List available at www.fda.gov .

[61] Id .

[62] Raymond Woosley, Transcript of Senate Health, Education, Labor, and Pensions Committee, Mar. 1, 2005.

[63] HUTT , supra note 11, at 552.

[64] Kweder, supra note 44.

[65] Barry Meier et al., Medicine Fueled by Marketing Intensified Trouble for Pain Pills , N.Y. TIMES , Dec. 19, 2004 at 1.

[66] Id .

[67] Id .

[68] Testimony of Dr. Bruce M. Psaty, Congressional Finance Committee, Nov. 18, 2004.

[69] Bombardier et al., supra note 8, at 1520.

[70] Id .

[71] Id .

[72] Id .

[73] See id . There were 2.1 confirmed GI events per 100 patient-years with rofecoxib compared with 4.5 with naproxen.

[74] Id . at 1523.

[75] Id . at 1526.

[76] Psaty, supra note 68. This heart attack comparison rate is also supported at Murray Charters, Part 1: Vioxx Withdrawal Prompts Reevaluation of COX-2 Inhibitor Safety , Parkinsn Online, at http://www.parkinsons-information-exchange-network-online.com/parkmail1/2004d/msg00141.html . (Last visited Mar. 18, 2005; Testimony of David J. Graham, Congressional Finance Committee, Nov. 18, 2004.

[77] Kweder, supra note 44.

[78] Correspondence, Rofecoxib, Merck, and the FDA , 351 NEW ENGLAND J. MED . 2875, 2875 (2004).

[79] Kweder, supra note 44.

[80] Correspondence, supra note 78, at 2877; Study 090 is available at http://www.fda.gov/ohrms/dockets/ac/01/briefing/3677b2_06_cardio.pdf . (Last visited Mar. 19, 2005).

[81] Correspondence, supra note 78, at 2878.

[82] Id .

[83] Barry Meier, Questions Are Seen on Merck’s Stance on Pain Drug’s Use , N.Y. TIMES , Nov. 24, 2004 at A1.

[84] Id .

[85] Id .

[86] Id .

[87] Debabrata Mukherjee et al., Risk of Cardiovascular Events Associated With Selective COX-2 Inhibitors , 286 J.AM. MED. ASSOC. 954 , 954 (2001) available at JAMA Online. (Last visited Mar. 19, 2005).

[88] Id .

[89] See infra III. SUMMARY OF STUDIES, B. STUDIES ON CELEBREX, 1. Celecoxib Long-Term Arthritis Safety Study (CLASS), for a detailed analysis of the study.

[90] Mukherjee, supra note 87.

[91] Mukherjee, supra note 87.

[92] Id .

[93] Id .

[94] Id .

[95] Id .

[96] Id .

[97] Id .

[98] Id .

[99] Id .

[100] Id .

[101] Id .

[102] Id .

[103] Id .

[104] Id .

[105] Id .

[106] Id .

[107] See Aviel Shatz, Vioxx Pain: A Story of Business and Statistics Error at http://www.interconus.com/Vioxx_Pain.doc . (Last visited Apr. 24, 2005).

[108] David J. Graham, Internal FDA Memorandum, available at http://www.fda.gov/cder/drug/infopage/vioxx/vioxxgraham.pdf . (Last visited Mar. 18, 2005).

[109] Id .

[110] Id .

[111] Id .

[112] Id . Of course, patient exposure to different drugs and treatments were different among the different groups.

[113] Id .

[114] Id .

[115] Id .

[116] Id .

[117] Id .

[118] Id .

[119] Id .

[120] Shatz, supra note 107.

[121] Id .

[122] Id .

[123] Id .

[124] Id .

[125] Id .

[126] O’Conner, supra note 4, at C1.

[127] Shatz, supra note 107.

[128] Id .

[129] Id ; Anonymous, Cardiovascular Risk of Selective COX-2 Inhibitors...Fact or Fiction? , John Hopkins Division of Rheumatology, at http://www.hopkins-arthritis.som.jhmi.edu/news-archive/2005/cardiac_risk_cox2.html#approve . (Last visited Mar. 19, 2005).

[130] Anonymous, supra note 129.

[131] Psaty, supra note 68.

[132] Anonymous, supra note 129.

[133] Id .

[134] Id . This suggests that continuous use of rofecoxib under 18 months will not produce increased cardiac events. It is important to note that rofecoxib was not approved for long-term or chronic use.

[135] Summary of COX-2 Inhibitor Trials With Cardiovascular (CV) Event Data at http://www.google.com/search?hl=en&q=summary+of+COX-2+inhibitor+trials+with+cardiovascular+(CV)+event+data&spell=1 (Last visited Mar. 19, 2005).

[136] Anonymous, supra note 129.

[137] Shatz, supra note 107.

[138] Id .

[139] Id .

[140] Id .

[141] Id .

[142] Id .

[143] Id . Again, Vioxx was not approved for long-term or chronic use. Contrary to the small 2.5% figure, many have argued that given the large population of Vioxx users (about two million), the actual number exposed and thus, the population affected can be rather large.

[144] Fred E. Silverstein et al., Gastrointestinal Toxicity With Celecoxib vs. Nonsteroidal Anti-inflammatory Drugs for Osteoarthritis and Rheumatoid Arthritis: The CLASS Study: A Randomized Controlled Trial , 284 J. AM. MED. ASSOC. 1247, 1247 (2000), available at JAMA Online. (Last visited Mar. 19, 2005).

[145] Anonymous, Celecoxib (Celebrex) Long-term Arthritis Safety Study (CLASS) , John Hopkins Division of Rheumatology, at http://www.hopkins-arthritis.som.jhmi.edu/news-archive/2000/class_study.html . (Last visited Mar. 19, 2005).

[146] Id .

[147] Silverstein et al., supra note 144, at 1247.

[148] Id .

[149] Id .

[150] Id .

[151] Anonymous, supra note 145.

[152] Id .

[153] Id .

[154] Silverstein et al., supra note 144, at Table 4.

[155] Id .

[156] Id .

[157] Anonymous, supra note 145.

[158] Id . Previous studies have used endoscopic ulcer rates to show that ulcer complication rates are considerably lower in patients taking COX-2 inhibitors compared to patients taking NSAIDs.

[159] Id .

[160] Id . The data (that excluded patients taking concomitant aspirin) shows that rate of POBs for celecoxib patients was 0.44% and for 1.27% for NSAIDs patients with P =0.4. Rates of POBs and symptomatic ulcers were 1.40% and 2.90%, respectively, P =0.2.

[161] O’Conner, supra note 4, at C1.

[162] Anonymous, supra note 145.

[163] Id .

[164] Solomon et al., Cardiovascular Risk Associated with Celecoxib in a Clinical Trial for Colectoral Adenoma Prevention , 352 NEW ENGLAND J. MED. 1071 , 1071-1072 (2005).

[165] Id . at 1072.

[166] Anonymous, supra note 129.

[167] Solomon et al., supra note 164, at 1072.

[168] Id . at 1072-1073.

[169] Anonymous, supra note 129.

[170] Id .

[171] Solomon et al., supra note 164, at 1071.

[172] Id .

[173] Id . at 1078.

[174] Id . at 1077.

[175] Id .

[176] Anonymous, Celebrex Suspended , News-Medical.net, at http://www.news-medical.net/?id=6881 . (Last visited Mar. 18, 2005).

[177] Solomon et al. supra note 164, at 1078.

[178] Id .

[179] Anonymous, Updated Advisory for Healthcare Professionals: Celecoxib (Celebrex®) and Related Drugs (2004), Health Sciences Authority, at http://www.hsa.gov.sg/docs/safetyalert_celecoxib_27Dec04.pdf . (Last visited Mar. 18, 2005).

[180] Anonymous, supra note 176.

[181] Id .

[182] Id .

[183] Id .

[184] Anonymous, supra note 176.

[185] Id .

[186] Use of Non-Steroidal Anti-Inflammatory Drugs Suspended in Large Alzheimer's Disease Prevention Trial , National Institutes of Health, at http://www.nih.gov/news/pr/dec2004/od-20.htm . (Last visited Mar. 18, 2005).

[187] Id .

[188] Id .

[189] Id .

[190] Stuart Barton, Which Clinical Studies Provide the Best Evidence? , GENERAL MED. J., Jul. 29, 2000 at http://bmj.bmjjournals.com/cgi/content/full/321/7256/255 (Last visited Mar. 29, 2005).

[191] See Psaty, supra note 68.

[192] Id .

[193] Barton, supra note 190.

[194] “The most common types of epidemiological studies are case-control studies; cohort studies, and cross-sectional studies.” http://www.greenfacts.org/glossary/def/epidemiological-studies.htm (Last visited Apr. 24, 2005); for more information, please go to http://www.ncabr.org/biomed/FAQ_general/faq_gen_6.html . (Last visited Apr. 24, 2005).

[195] Patients were taking Vioxx at a dose higher than the maximum recommended dose of 25 mg per day. Alex Berenson et al., Dangerous Data—Retracing a Medical Trail; Despite Warnings, Drug Giant Took Long Path to Vioxx Recall , N.Y. TIMES , Nov. 14, 2004 at C1.

[196] Id .

[197] See id .

[198] Shatz, supra note 107.

[199] Mukherjee, supra note 87.

[200] See id .

[201] Graham, supra note 108.

[202] Correspondence, supra note 78, at 2875.

[203] Kweder, supra note 44.

[204] Correspondence, supra note 78, at 2875.

[205] Kweder, supra note 44.

[206] Anonymous, supra note 145. http://www.hopkins-arthritis.som.jhmi.edu/news-archive/2000/class_study.html .

[207] Id .

[208] See id .

[209] Meier, supra note 83, at A1.

[210] Id .

[211] Id .

[212] Anonymous, supra note 145.

[213] Anonymous, supra note 176. http://www.news-medical.net/?id=6881 .

[214] Solomon et al., supra note 164 at 1077.

[215] The more doses per day, the longer the inhibition of prostacyclin is sustained.

[216] Id . at 1078.

[217] Id .

[218] Anonymous, Editorial Comments, supra note 129.

[219] Id .

[220] Id .

[221] Jennifer Couzin, Nail-Biting Time for Trials of COX-2 Drugs , 306 SCIENCE 1673, 1675 (2004).

[222] Victoria Porter, NSAIDs Still Under Surveillance—Celecoxib, Valdecoxib, and Naproxen have Been Added to the List of Suspects , MedScape from WebMD, at http://www.medscape.com/viewarticle/496951_1 . (Last visited Mar. 31, 2005).

[223] Id .

[224] Id .

[225] Id .

[226] Id , citing Heavy, FDA Keeping Options Open on Naproxen , REUTERS HEALTH , Dec. 21, 2004.

[227] Label of Celebrex, FDA CDER, at http://www.fda.gov/cder/foi/label/2005/020998s017lbl.pdf . (Last visited Mar. 23, 2005); Label of Vioxx, FDA CDER, at http://www.fda.gov/cder/foi/label/2004/021052s026_021042s019lbl.pdf . (Last visited Mar. 23, 2005).

[228] Porter, supra note 222.

[229] Id .

[230] Id .

[231] Solomon et al. supra note 164, at 1078.

[232] Label.

[233] Id .

[234] Id .

[235] FDA Public Health Advisory: Safety of Vioxx, FDA CDER, at http://www.fda.gov/cder/drug/infopage/vioxx/PHA_vioxx.htm . (Last visited Mar. 26, 2005).

[236] Patient letter, Important Information for Patients Taking VIOXX (rofecoxib) , Merck website at http://www.vioxx.com/vioxx/documents/english/information_for_patients.pdf . (Last visited Mar. 26, 2005).

[237] Id .

[238] Id . (Emphasis added).

[239] That is, since we know that the FDA can order withdrawal in certain situations.

[240] FDA CDER, supra note 235. http://www.fda.gov/cder/drug/infopage/vioxx/PHA_vioxx.htm .

[241] Id .

[242] Alex Berenson et al., supra note 195, at C1.

[243] Id .

[244] Dr. Lester Crawford has been nominated by President George W. Bush to be the FDA Commissioner, but his confirmation has been postponed due to a controversial over-the-counter emergency contraception drug issue. Anne A. Kornblut, Senate Leaders Postpone Vote on F.D.A. Chief , N.Y. Times, April 14, 2005 at A1.

[245] FDA Issues Public Health Advisory on Vioxx as its Manufacturer Voluntarily Withdraws the Product at http://www.fda.gov/bbs/topics/news/2004/NEW01122.html . (Last visited Mar. 26, 2005).

[246] Federal Food, Drug & Cosmetic Act, 21 USCS § 355(e) (2005).

[247] Id .

[248] See id .

[249] Porter, supra note 222.

[250] See 21 USCS § 352(a).

[251] For overview of challenges to FDA withdrawal, see infra V. WITHDRAWAL, ADVERTISING, AND RELABELING, A. WITHDRAWAL, 3. Was Merck’s Decision to Withdraw Rational?

[252] HUTT , supra note 11, at 547.

[253] Kweder, supra note 44.

[254] Meier, supra note 83, at A1.

[255] There is dispute over the validity of the number. According to Dr. David Graham, applying the risk levels from VIGOR and APPROVe would generate a number in the range of 88,000 to 139,000 Americans, and Dr. Topol has estimated the number to be nearly 160,000. Graham, supra note 76.

[256] HUTT , supra note 11, at 547.

[257] Id .

[258] Berenson et al., supra note 195, at C1.

[259] Testimony of Dr. Janet Woodcock, Senate Health, Education, Labor, and Pensions Committee, Mar. 3, 2005.

[260] Id .

[261] JAMES T. O’REILLY , FOOD AND DRUG ADMINISTRATION (West Group, 2d ed. 1993) at § 13.18, citing John D. Copanos & Sons Inc. v. FDA, 854 F2d. 510 (D.C. Cir. 1988).

[262] Id .

[263] Id .

[264] Id .

[265] Id .

[266] See patient letter, supra note 236.

[267] Mary Duenwald, For Pain Management, Doctors Prescribe Caution , N.Y. TIMES , Feb. 20, 2005, at 1.

[268] Heather Won Tesoriero, Patients Weigh Pain Vs. Risk , WALL STREET J., Dec. 21, 2004 at http://www.sfgate.com/cgi-bin/article.cgi?f=/news/archive/2004/12/21/financial1010EST0056.DTL&type=health . (Last visited Mar. 26, 2005).

[269] Bring Back Vioxx at http://www.bringvioxxback.com/life_after_vioxx/index.html . (Last visited Mar. 26, 2005).

[270] O’REILLY , supra note 261, at § 15.09.

[271] Id .

[272] 21 USCS § 352(a) (2005).

[273] For a more detailed description, see infra II. THE APPROVAL OF VIOXX AND CELEBREX, parts B. and C.

[274] Graham, supra note 76.

[275] Correspondence, supra note 78, at 2875.

[276] Id .

[277] Id .

[278] Berenson et al., supra note 195 at C1.

[279] Id .

[280] Id .

[281] Correspondence, supra note 78, at 2875.

[282] Kweder, supra note 44.

[283] See Correspondence, supra note 78, at 2875.

[284] For a full discussion of the limitations of APPROVe, see infra III. SUMMARY OF STUDIES, A. STUDIES ON VIOXX, 4. APPROVe, Limitations.

[285] Berenson et al., supra note 195, at C1.

[286] See patient letter, supra note 236.

[287] Testimony of Dr. Sandra Kweder, Senate Health, Education, Labor, and Pensions Committee, Mar. 1, 2005.

[288] That is, labeling authority by way of withdrawal. Transcript of Senate Education, Labor, and Pensions Committee, Mar. 3, 2005.

[289] See id .

[290] 21 USCS § 352(a) (2005).

[291] 21 USCS §321(n) (2005).

[292] Id .

[293] The dose of Vioxx used in the APPROVE study was 25 mg per day.

[294] O’REILLY , supra note 261, at § 14.11

[295] 21 USCS § 352(a) (2005).

[296] See HUTT , supra note 11 at 1040.

[297] Id . at 1041-42.

[298] For a discussion of this problem see infra VI. REFORM OF FDA, B. LACK OF LABELING AUTHORITY.

[299] See Nat Ives, Madison Ave. Sharing Drug Maker’s Pain , N.Y. Times, Dec.21, 2004 at C1.

[300] The FDA cannot force Merck to sell Vioxx, which would theoretically require labeling changes or the drug could be misbranded. If Vioxx decides not to change the label, then the FDA may have no choice but to withdraw the drug.

[301] Alex Berenson, supra note 3, at C1.

[302] Id .

[303] Andrew Leckey, Stock Volatility a Side Effect of Health-Care Marketing , CHICAGO TRIBUNE , Jan. 23, 2005 at C7.

[304] Gardiner Harris et al., Pfizer and Celebrex: The Overview, Drug Trial Finds Big Health Risks in 2nd Painkiller , N.Y. TIMES, Dec. 18, 2003 at A1.

[305] Meier et al., supra note 65, at 1.

[306] Id .

[307] Alex Berenson and Gardiner Harris, Pfizer Says 1999 Trial Revealed Risks With Celebrex , N.Y. TIMES , Feb. 1. 2005 at C1.

[308] Berenson, supra note 3, at C1.

[309] Letter to the editor, Story of Celebrex: Pain and Profit , N.Y. Times, Dec. 21, 2004 at A28.

[310] Associated Press, The Markets: Stocks and Bonds , N.Y. TIMES , Dec. 21, 2004 at C9.

[311] Letter to the editor, supra note 309, at A28.

[312] See Berenson, supra note 3, at C1.

[313] See Alex Berenson, Ban or No Ban, Celebrex Sales are Expected to Dwindle , N.Y. TIMES , Dec. 21, 2004 at C1.

[314] See Pfizer to Halt All Ads For Pain Drug Celebrex , N.Y. TIMES , Dec. 20, 2004 at A1.

[315] After all, an innocuous statement like, “Pfizer decided to stop consumer advertising, and its actions were approved by the FDA,” could in reality mean that the FDA put enough pressure on Pfizer that it decided to stop consumer advertising, and of course, the FDA would have approved a plan, the creation of which it had influenced.

[316] See Alex Berenson et al., supra note 195, at C1.

[317] Berenson, supra note 3 at C1.

[318] Testimony of Dr. Janet Woodcock, House Energy and Commerce Committee, Jun. 13, 2001.

[319] See 21 USCS § 352(n) (2005).

[320] O’Reilly, supra note 261, at § 15.11.

[321] Id .

[322] Id .

[323] See 21 CFR § 202.1(e)(5).

[324] O’REILLY , supra note 261, at §15.11.

[325] 21 USCS § 352(n). (2005)

[326] O’REILLY , supra note 261, at §15.11.

[327] Woodcock, supra note 295.

[328] Id .

[329] Two days earlier, Kweder of the FDA had suggested to the same Senate Committee the idea of issuing public health advisories through a drug watch webpage to put out information about clinical trials when they became available to alert the public before the label of a drug is changed to reflect such trials.

[330] Nat Ives, supra note 226, at C2.

[331] See Eric Nagourney, A Prescription for Patient Error , N.Y. TIMES , Nov. 23, 2004 at F8.

[332] The results of the study were added to the “Precautions” section under “Cardiovascular Effects” instead of “Warning”, and the language was phrased in a favorable manner.

[333] Stephanie Saul, Popular Heart Drug Impairs Kidney Action , N.Y. Times, Mar. 22, 2005 at C2.

[334] Id .

[335] Cecil Wilson, Transcript of Senate Education, Labor, and Pensions Committee, Mar. 3, 2005.

[336] Id .

[337] David Fassler, Transcript of Senate Education, Labor, and Pensions Committee, Mar. 1, 2005.

[338] Ives, supra note 299, at C1.

[339] Id .

[340] O’REILLY , supra note 261, at § 15.10 citing FDA Regulatory Letter to Ciba-Geigy Corp, Oct. 16, 1984, reported in FDA Quarterly Activities Report, First Quarter, Fiscal Year 1985, at 17 (1985).

[341] Berenson et al., supra note 195, at C1.

[342] Barnaby J. Feder, Federal Panel Consolidate Vioxx Suits , N.Y. TIMES , Feb. 17, 2005 at C1.

[343] Id .

[344] Id .

[345] Anonymous, New Doubts About Vioxx , N.Y. TIMES , Dec. 18, 2004 at A18.

[346] Feder, supra note 342, at C1.

[347] Berenson et al., supra note 195, at C1.

[348] Anonymous, The Experts’ Verdict on Painkillers , N.Y. TIMES , Feb. 19, 2005 at A14. However, the vote to bring Vioxx back on the market was only 17-15 compared to Celebrex’s 31-1, and plaintiffs may point out this difference. Additionally, there have been assertions that the panel was unduly influenced by the drug industry due to conflicts of interests, and with such a close vote, a court may likely take this into consideration.

[349] Porter, supra note 222.

[350] Berenson, supra note 307, at C1.

[351] Porter, supra note 222, citing Eric Topol, Arthritis Medicines and COX-2 Inhibitors: “House of Coxibs” , 293 J. AMER. MED. ASSOC. 366 , 366 (2005).

[352] Anonymous, supra note 176. http://www.news-medical.net/?id=6881 .

[353] Porter, supra note 222, citing Hensley S et al., As Safety Issues Hits Celebrex, Pfizer Decides to Hang Tough, WALL STREET J. , Dec. 20, 2004.

[354] Eric Topol, Arthritis Medicines and COX-2 Inhibitors: “House of Coxibs” , 293 J. AMER. MED. ASSOC. 366 , 367 (2005).

[355] Dr. Topol notes that this has led researchers to conclude that Celebrex was not associated with increased heart risks but that in randomized trials, “a signal for potential cardiovascular risk with celecoxib was present.” Porter, supra note 222.

[356] Porter, supra note 222.

[357] See id.

[358] Stephen Kimmel et. al., Patients Exposed to Rofecoxib and Celecoxib Have Different Odds of Nonfatal Myocardial Infarction , 142 ANNALS OF INTERNAL MED. 157 (2005) available at Annals of Internal Medicine Online. (Last visited Mar. 23, 2005).

[359] Id .

[360] Id .

[361] Id . Vioxx had 3.39 higher risk of heart attack than naproxen. Celebrex had 0.77 higher risk of heart attack than ibuprofen or diclofenac.

[362] Kimmel, supra note 358.

[363] Id .

[364] Id .

[365] Gardiner Harris, Big Health Plan Suspends Use of Painkiller , N.Y. TIMES , Jan. 29, 2005 at C1.

[366] Id .

[367] Id .

[368] Gardiner Harris, F.D.A. Panel Says Pain Reliever Should Remain on Market , N.Y. TIMES , Feb. 18, 2005.

[369] Ten of the thirty-two panel members were found to have conflicts. Gardiner Harris and Alex Berenson, 10 Voters on Panel Backing Pain Pills Had Industry Ties , N.Y. TIMES , Feb. 25, 2005 at A1.

[370] Id .

[371] Nancy Davenport-Ennis, Transcript of the Senate Education, Labor, and Pensions Committee, Mar. 1, 2005.

[372] Woodcock, supra note 259.

[373] Gardiner Harris and John Schwartz, Regulation Redefined: The F.D.A. Shifts Focus; At FDA, Strong Drug Ties and Less Monitoring , N.Y. TIMES , Dec. 6, 2004 at A1.

[374] Id .

[375] Id .

[376] Id .

[377] Id .

[378] Id .

[379] Raymond Woosley, supra note 62.

[380] Don Monroe, Cooperation Between Industry, FDA Leads to Rebound in Drug Approval , Genomics and Proteomics, Feb. 10, 2004, at http://www.genpromag.com/ShowPR_Print~PUBCODE~018~ACCT~1800000100~ISSUE~0402~RELTYPE~PR~ORIGRELTYPE~RLSN~PRODCODE~00000000~PRODLETT~H~CALLFROM~RELPGM.html . (Last visited May 2, 2005).

[381] Robert Steyer, FDA Gives Green Light to Merck-Monsanto Race; Vioxx Gets Accelerated Look on Heels of Celebrex , ST. LOUIS POST-DISPATCH , Jan. 12, 1999, at C7; Robert Steyer, FDA Oks Drug from Monsanto; Company Says Drug is Easier on Stomach , ST. LOUIS POST-DISPATCH , Jan. 1, 1999, at A1.

[382] Harris, supra note 373, at A1.

[383] See Woosley, supra note 62.

[384] See id .

[385] Harris, supra note 373, at A1.

[386] Id .

[387] Id .

[388] Harris, supra note 12, at A1.

[389] Elizabeth Glode, Advising Under the Influence?: Conflicts of Interest Among FDA Advisory Committee Members , 57 FOOD & DRUG L. J. 293 , 294 (2002).

[390] Id .

[391] Id . at 301.

[392] Id . at 294.

[393] Id . at 299.

[394] Id . at 294.

[395] Id . at 304.

[396] Anonymous, Conflicts Cited for 10 Voters on FDA Panel , CHICAGO TRIBUNE , Feb. 26, 2005, at C14.

[397] Harris, supra note 369 at A1.

[398] Id .

[399] Id .

[400] Id .

[401] Id . Two panel members did not respond to email or telephone messages.

[402] Glode, supra note 388, at 294-95.

[403] Id . at 301.

[404] William Schultz, Transcript of Senate Education, Labor, and Pensions Committee, Mar. 1, 2005.

[405] In fact, Senator Michael Enzi, speaker at the Senate Education, Labor, and Pensions Committee on March 3, 2005, did not realize this distinction. In response to Kweder’s claim, two days earlier, that the FDA did not have explicit authority to dictate specific labeling changes, Senator Enzi had his staff look into the FDA’s authority in the matter and concluded that since the FDA can pull the drug off the market, it seemed to him like “quite a bit of authority.”

[406] 21 U.S.C.A. § 356(a). (2005)

[407] 21 U.S.C.A. §356(b)(2). (2005)

[408] See O’REILLY , supra note 261, at § 13.13.

[409] See id .

[410] Id .

[411] Id .

[412] Steyer, supra note 381, at C7, Steyer, supra note 381, at A1.

[413] HUTT , supra note 11, at 537.

[414] Approval letter for NDA 20-998, FDA CDER, at http://www.fda.gov/cder/foi/nda/98/20998AP_appltr.pdf . (Last visited Apr. 1, 2004).

[415] Approval letter for NDA 20998/S007, FDA CDER, at http://www.fda.gov/cder/foi/nda/99/21156-S007_Celebrex_approv.pdf . (Last visited Apr. 1, 2004).

[416] See approval letter for NDA 021042 and 021052, FDA CDER, at http://www.fda.gov/cder/foi/nda/99/021042_52_vioxx_appltr.pdf . (Last visited Apr. 1, 2004).

[417] See FDA CDER, supra note 24; see FDA CDER, supra note 42.

[418] Janet Woodcock, Transcript of the Senate Education, Labor, and Pensions Committee, Mar. 3, 2005.

[419] Id .

[420] Dr. Gurkipal Singh testified before a Congressional Committee that one of the members of the Arthritis Committee Panel noticed that that thromboembolic events were more frequent in patients taking Vioxx compared with placebo and that a larger database is needed to assess with certainty the risk of cardiovascular and thromboembolic events in patients taking Vioxx. It is uncertain whether this observation, among the many others risk observations from the committee, would have alerted the FDA to require post-approval testing of such risks. Testimony of Gurkipal Singh, Congressional Finance Committee, Nov. 18, 2004.

[421] See id . Two instances where FDA asserted this authority were with levodopa and methadone.

[422] HUTT , supra note 11, at 538.

[423] 21 C.F.R. § 310.303. (2005) (Emphasis added)

[424] Id . (Emphasis added).

[425] See Chevron U.S.A. Inc., v. Natural Resources Defense Council, Inc. 467 US 837 (1984).

[426] Neil Cave, Post-Approval Testing by Pharmaceutical Companies , Food and Drug Law: An Electronic Book of Student Papers, Apr. 30, 2002 at http://leda.law.harvard.edu/leda/data/324/Cave2.html . (Last visited Apr. 1, 2005).

[427] Id .

[428] Id .

[429] Id .

[430] Id .

[431] Id .

[432] Raymond Woosley, supra note 62.

[433] Id .

[434] Id .

[435] Id .

[436] O’REILLY , supra note 261, at § 14.05.

[437] Id .

[438] Id .

[439] Id .

[440] Id .

[441] Harris, supra note 12, at A1.

[442] O’REILLY , supra note 222, at § 14.05.

[443] Id .

[444] Gardiner Harris, Ideas & Trends: Shades of the Vioxx Case for Another Drug , N.Y. TIMES , Nov. 21, 2004 at 3.

[445] Berenson, supra note 195, at C1.

[446] Id .

[447] Gardiner Harris, Merck May Resume Sales of Painkiller, Official Says , N.Y. TIMES , Feb. 18, 2005 at A18.

[448] Id .

[449] Harris, supra note 368.

[450] Id .

[451] Bring Vioxx Back, supra note 269, citing Painkiller Panic , TOWN HALL , Dec. 30, 2004.

[452] Porter, supra note 222, citing Hensley S et al., As Safety Issues Hits Celebrex, Pfizer Decides to Hang Tough, WALL STREET J. , Dec. 20, 2004.

[453] Harris, supra note 12, at A1.

[454] Bring Vioxx Back, supra note 269.

[455] Id .

[456] Id .

[457] Id .

[458] Berenson and Feder, A Reminder that No Drug is Risk-Free , N.Y. TIMES , Feb. 19, 2005 at C1.

[459] Id .

[460] The Associated Press, FDA Seeks Drug—Warning—Label Authority , N.Y. TIMES , Mar. 1, 2005.

[461] Harris, supra note 368. (Remain on Market)

[462] Abbey Meyer, Transcript of Senate Education, Labor, and Pensions Committee, Mar. 1, 2005.

[463] Harris, supra note 447, at A18.

[464] Id .

[465] Id .

[466] Leckey, supra note 303, at C7.

[467] The results of VIGOR can be found in the “precautions” section, under “cardiovascular effects”. Psaty, supra note 68.

[468] Anonymous, supra note 345, at A18.

[469] Berenson, supra note 3, at C1.

[470] Transcript of Senate Health, Education, Labor, and Pensions Committee, Mar. 1, 2005.

[471] Id .

[472] Harris, supra note 373, at A1.

[473] Testimony of Dr. Cecil B. Wilson, Senate Health, Education, Labor and Pensions Committee, Mar. 3, 2005.

[474] Id .

[475] Id .

[476] Testimony of Dr. Janet Woodcock, Senate Health, Education, Labor and Pensions Committee, Mar. 3, 2005.

[477] Id .

[478] The Associated Press, supra note 459.

[479] Editorial, FDA Oversight Board Is the Right First Step; Independent panel will monitor drug safety and provide better public information , THE DETROIT TIMES , Feb. 27, 2005 at 18A.

[480] Id .

[481] Id .

[482] Transcript, supra note 469. Mar. 1, 2005.

[483] Id .

[484] Transcript of Senate Health, Education, Labor, and Pensions Committee, Mar. 3, 2005.

[485] Id .

[486] Id .

[487] FDA Public Health Advisory, FDC CDER, at http://www.fda.gov/cder/drug/advisory/COX2.htm. (Last visited Apr. 8 2005).

[488] Id .

[489] Id .

[490] Id .