NPC Archive Item: Rofecoxib is associated with an early increase in cardiovascular events, which persists for a year after stopping treatment

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo

Baron J, Sandler RS, Bresalier RS, et al. Cardiovascular events associated with rofecoxib: final analysis of the APPROVe trial. The Lancet. Early online publication 14th October 2008. DOI:10.1016/S0140-6736(08)61490–7

In the APPROVe trial, rofecoxib increased the risk of the combined primary endpoint (stroke, myocardial infarction [MI] or death) from soon after treatment was started until a year after it was stopped.

There is good evidence that coxibs cause a small increased risk of thrombotic events in comparison with placebo. This risk appears to increase with dose and persists throughout treatment. These new rofecoxib data indicate that the increased risk of CV events persists for a year after stopping the coxib.

Some traditional non-steroidal anti-inflammatory drugs (NSAIDs) (e.g. diclofenac 150mg/day) carry the same increased risk of thrombotic events as coxibs. Low-dose ibuprofen (<1200mg/day), and naproxen 1000mg/day do not appear to share these cardiovascular (CV) side effects, and most data indicate their rateof serious adverse gastrointestinal (GI) events is at the lower end of the NSAID spectrum – especially if their use is accompanied by a proton pump inhibitor (PPI) as recommended by the NICE osteoarthritis guideline.

Both NSAIDs and coxibs should be used only when other, safer treatments are ineffective or not tolerated. They should be prescribed at the lowest effective dose for the shortest possible period of time. When choosing an individual agent, prescribing should be based on the safety profiles of individual NSAIDs/coxibs and on individual patient risk factors.

What is the background to this?
The MHRA reported the immediate worldwide withdrawal of rofecoxib in September 2004 after results from the Adenomatous Polyp Prevention on Vioxx® (APPROVe) study showed an increased risk of serious thrombotic events (including MI and stroke) compared to placebo, following long-term use.

The APPROVe study was a multi-centre, randomised controlled trial (RCT) to compare 3 years treatment with rofecoxib or placebo for prevention of recurrence of neoplastic polyps of the large bowel in almost 2,600 patients with a history of colorectal adenoma. More patients taking rofecoxib experienced a confirmed serious thrombotic event, compared with placebo (45 vs. 25, respectively). The absolute event rates were approximately 3 per 400 patient years for placebo and 6 per 400 patient years for rofecoxib i.e. an absolute increase in risk of approximately 3 thrombotic events per 400 patient years of treatment.

The difference in event rates only appeared to be apparent after 18 months of treatment. However, the analysis of the APPROVe study was criticised, mainly because patients were only followed up for 14 days after stopping treatment. This means that some adverse events related to the study drug may not have been recorded, and the analysis was not strictly intention to treat. Subsequently, patients from the APPROVe study were followed up in an attempt to obtain data on CV events for at least one year after patients had stopped study treatment.

What does this study claim?
The intention to treat analysis in this follow-up study found that the chance of having a non-fatal MI or stroke, or dying from a CV, haemorrhagic, or unknown cause, was almost doubled in patients taking rofecoxib, compared with placebo. Fifty nine patients in the rofecoxib group experienced one of these events, compared with 34 in the placebo group (hazard ratio [HR] 1.79, 95% confidence interval [CI] 1.17 to 2.73; P=0.006).  Participants on rofecoxib had a significant two-fold increase in the risk of MI (HR 1.94, 95% CI 1.09 to 3.43; P=0.02) and a non-significant two-fold increase in the risk of stroke (HR 2.17, 95% CI 0.98 to 4.80; P=0.05).

As the authors of the study and an accompanying editorial point out, this new analysis does not give us much information about how any CV hazard evolves over time because the number of events was very small. However, the hazard ratio for the combined primary endpoint (stroke, MI or death) did not vary substantially over time suggesting an early increase in risk that seems to persist for one year after stopping treatment. Although the data was not statistically definite, it is interesting that individuals with risk factors for CV disease seemed to have a higher risk of CV events with rofecoxib compared with placebo, than those without CV risk factors.

So what?
This study adds to existing data on the CV risks of NSAIDs. As discussed in a MeReC Extra on the CV and GI safety of NSAIDs, all coxibs are associated with a small excess risk of thrombotic events compared with no treatment (about three per 1000 users treated for one year), and they are contraindicated in patients with established CV disease. However, some traditional NSAIDs may also be associated with an increased risk of thrombotic events. Diclofenac 150mg/day appears to be associated with a similar excess risk to that of coxibs, whereas low-dose ibuprofen (≤1200mg/day) and naproxen 1000mg/day appear to be associated with a lower risk.

Where NSAIDs are required, prescribing should be based on the safety profiles of individual NSAIDs and on individual patient risk factors. All NSAIDs should generally be used at the lowest effective dose and for the shortest period of time necessary to control symptoms. The ideal anti-inflammatory prescribing choice will vary from patient to patient, depending on individual risk factors, therapeutic response, patient preference, and the patient’s attitude to the risk of adverse events.

Coxibs, because of their risk of adverse CV effects, would appear to have a limited role in clinical practice. Co-prescription of a traditional NSAID with a PPI appears at least as effective as a coxib alone in reducing GI side effects, and is a less expensive option. For patients with osteoarthritis, the NICE guidance suggests prescribing an NSAID or coxib (other than etoricoxib 60mg) in combination with a PPI. However, there is, as yet, no good evidence to suggest a coxib plus PPI is a more beneficial, equivalent or worse option then a traditional NSAID plus PPI with regard to GI side effects. And of course, compared with ibuprofen 1200mg per day or less or naproxen 1000mg per day, use of a coxib may put the patient at greater risk of a CV event.

Low dose ibuprofen (400mg three times a day) is an appropriate first choice NSAID in view of its low CV and GI risks at this dose. Naproxen 1000mg/day is an alternative to ibuprofen 1200mg/day and may be preferable to NSAIDs other than ibuprofen 1200mg/day in patients at significant risk of adverse CV effects. Co-prescription of a PPI seems advisable especially in patients at particular risk of serious GI effects. The apparently increased rate of CV events with diclofenac and high dose ibuprofen suggests that, like coxibs, these are less appropriate options for patients in whom CV risk is a significant consideration in decision making.

Further information is available in the MeReC Extra and on the musculoskeletal pain section of NPC.

Study details
Design: Follow-up of the APPROVe RCT participants

Patients: 2,587 patients, aged at least 40 years, with a history of colorectal adenomas recruited at 108 centres worldwide during 2000 and 2001. Patients with uncontrolled hypertension; angina or congestive heart failure; history of MI, coronary angioplasty, or coronary arterial bypass grafting within the past year; or history of stroke or transient ischaemic attack within the past 2 years, were excluded.

Intervention: Rofecoxib 25mg vs. placebo

Comparison: Participants were followed for adverse events while on treatment and during the following 14 days during the original study. However, after early termination of treatment because of cardiovascular toxicity, this study attempted to follow up all randomised patients for at least 1 year after stopping study treatment.

Outcomes: External committees blindly assessed potential serious CV events. The focus of the analysis was the combined incidence of non-fatal MI, non-fatal stroke, and death from CV, haemorrhagic, and unknown causes.

Results: Post-treatment CV follow-up data was obtained from 84% of participants, and extended mortality follow-up from 95%.

Participants with events on treatment and in the following 14 days

Full intention to treat analysis, including extended follow-up



Unadjusted HR (95% CI)



Unadjusted HR (95% CI)

Combined MI, stroke or death




(1.20 to 3.74)




(1.17 to 2.73)





(1.21 to 5.75)




(1.09 to 3.43)





(0.67 to 4.02)




(0.98 to 4.80)

Ischaemic stroke







Haemorrhagic stroke







Vascular death




(0.40 to 4.30)




(0.61 to 2.62)

Overall mortality




(0.34 to 3.29)




(0.80 to 2.15)

* 4 additional patients in each group died more than 14 days after cessation of treatment because of an adverse event that took place on or within 14 days of treatment

Sponsorship: Merck Research Laboratories

Please comment on this blog in the NPC discussion rooms, or using our feedback form.