Characterizing new users of NSAIDs before and after rofecoxib withdrawal

Abstract

What is already known about this subject

• Public concern regarding the cardiovascular safety of the COX-2 inhibitors began with the withdrawal of rofecoxib from the market in September 2004.

• Since then, a myriad of evidence has pointed towards an adverse cardiovascular effect of other COX-2 inhibitors in persons with cardiovascular risk.

• This study examines the impact of the clinical trial publicity and drug regulatory advice on nonsteroidal anti-inflammatory drug (NSAID) prescribing in patients in general practice with known cardiovascular risk both before and after the rofecoxib withdrawal.

What this study adds

• Results from this study indicate that the profile of new users of NSAIDs and COX-2 inhibitors did not change despite new information becoming available.

• This may highlight the uncertainty experienced by prescribers of treatment alternatives available and the lack of unbiased information at this time for at-risk groups.

Aims

To characterize patients initiated on nonsteroidal anti-inflammatory drugs (NSAIDs), pre and postrofecoxib withdrawal, by age, gender and concomitant cardiovascular (CV) therapy.

Methods

A national primary care prescription database was used to identify patients who initiated NSAID therapy pre and postrofecoxib withdrawal. Patients receiving CV therapy were identified in the same periods also. Adjusted odds ratios (OR) and 95% confidence intervals are presented.

Results

Female patients [OR = 1.15 (1.11, 1.19)], those over 65 years [OR = 2.76 (2.65, 2.86)] and those at CV risk [OR = 1.72 (1.67, 1.79)] were more likely to start on celecoxib (over a nonselective NSAID) than male patients, those under 65 years and those not at CV risk. Similar results were found for rofecoxib and nimesulide. Postwithdrawal analysis showed results comparable to the prewithdrawal period.

Conclusion

The results highlight a possible uncertainty experienced by prescribers of treatment alternatives available and a lack of unbiased information at this time for at-risk groups.

Introduction

Selective cyclooxygenase 2 inhibitors (COX-2i's) were one of the most rapidly adopted new drug classes of recent years because of their reported better gastrointestinal safety profile compared with nonselective (ns) NSAIDs. Public concern regarding the cardiovascular (CV) safety of rofecoxib arose following reports from the Vioxx Gastrointestinal Outcomes Research (VIGOR) study [1] and this CV risk was more definitively seen in the APPROVe trial [2], which ultimately resulted in the withdrawal of rofecoxib. This widely publicized recall attracted great media attention and prompted concern regarding a potential class effect of the COX-2i's. Evidence has since emerged of chronic use of the COX-2i's [3] despite early warnings of problems with these new drugs [4].

At this time, little information was available to assist general practitioners (GPs) in their decision on alternative therapeutic options available for this group of patients. The aim of this study was to characterize patients new to NSAIDs, before and after the withdrawal of rofecoxib, by age, gender and concomitant CV therapy.

Methods

The General Medical Services (GMS) prescription database was used to identify the study population and has been described in detail elsewhere [5]. Therapies examined (based on sales figures) included specifically diclofenac, ibuprofen, naproxen and mefenamic acid, i.e. nsNSAIDs; celecoxib and rofecoxib, i.e. selective COX-2i, and nimesulide, a nsNSAID with some preferential selectivity for COX-2 inhibition in vivo [6]. Patients were identified in the prewithdrawal period, i.e. between October 2003 and May 2004, as not having received NSAID medication in the 6 months prior to initiation of therapy (n = 154 407 patients). Patients initiating therapy 1 year later (between October 2004 and May 2005), postwithdrawal (n = 139 515), were identified in a similar manner. Patients at risk of CV disease were defined as those having received a prescription for any of the following: nitrate and/or aspirin, β-blockers, renin–angiotensin system blockers, statins, fibrates, diuretics and calcium channel blockers. SAS (version 9) was used for statistical analysis (SAS Inc., Cary, NC, USA). Odds ratios (OR) and 95% confidence intervals were calculated, adjusted for age and gender, using logistic regression analysis. Significance at P < 0.05 is assumed throughout.

Results

The number of new patients receiving prescriptions for NSAIDs pre and postrofecoxib withdrawal is shown in Table 1. Overall, the nsNSAIDs were the most commonly prescribed, followed by nimesulide. Figures show that while the overall number of new nsNSAID and nimesulide users increased postwithdrawal of rofecoxib, the use of celecoxib decreased dramatically.

Table 1.

Total number of patients who received their first prescription for the four types of nonsteroidal anti-inflammatory drugs (NSAIDs) pre and postrofecoxib withdrawal periods

Drug	Prerofecoxib withdrawal	Postrofecoxib withdrawal	% change
Nonselective NSAIDs	92 515	94 269	+1.9%
Nimesulide	33 628	35 127	+4.5%
Celecoxib	16 721	9 530	−43%
Rofecoxib	11 543	–*	–

^*Patients were identified as having received a prescription for rofecoxib in the postwithdrawal period.

Patients starting on a COX-2i were significantly more likely to be female, be over 65 years and at risk of CV disease. When each drug was considered individually, this effect was maintained (Table 2). Analysis yielded similar results postrofecoxib withdrawal, whereby patients starting on a COX-2i were significantly more likely to be female, over 65 years and at risk of CV disease (Table 2). A test for interaction showed that there was no significant difference in the likelihood of patients starting on nimesulide (over a nsNSAID) either pre or postwithdrawal (P = 0.32). However, there was a significant difference in prescribing of celecoxib compared with nsNSAIDs, pre and postrofecoxib withdrawal, with significantly fewer patients prescribed celecoxib postwithdrawal (P < 0.0001).

Table 2.

Adjusted odds ratios and 95% confidence intervals for initiating a selective COX-2 inhibitor by age, gender and cardiovascular (CV) risk pre and postrofecoxib withdrawal

	Nimesulide vs. nsNSAIDs*	Celecoxib vs. nsNSAIDs*	Rofecoxib vs. nsNSAIDs*
Female vs. male*	1.09 (1.06, 1.12)	1.15 (1.11, 1.19)	1.14 (1.10, 1.19)
	1.11 (1.09, 1.14)	1.15 (1.10, 1.20)	–
>65 vs. <65 years*	1.68 (1.63, 1.73)	2.76 (2.65, 2.86)	2.32 (2.21, 2.42)
	1.63 (1.58, 1.67)	2.21 (2.10, 2.32)	–
CV risk vs. no CV risk*	1.28 (1.25, 1.32)	1.72 (1.67, 1.79)	1.61 (1.54, 1.69)
	1.32 (1.28, 1.35)	1.70 (1.60, 1.79)	–

^*Indicates the reference category in the logistic regression. Postwithdrawal results indicated in bold. nsNSAID, Nonselective nonsteroidal anti-inflammatory drug.

Discussion

In the present study, the nsNSAIDs were the most commonly initiated NSAIDs after withdrawal of rofecoxib, followed by nimesulide and then the COX-2i's. This trend is similar to that observed in an earlier study, where the relatively low prescribing rate for the COX-2i's at that time was attributed to prescriber unfamiliarity [7]. However, in the present study celecoxib and rofecoxib, the first authorized COX-2i's, were more established in the Irish market after their introduction in June and October 2000, respectively. It should be noted that although the prescription rates for the COX-2i's have increased in Ireland since their launch, they have not increased at the same rate in Ireland as in the USA. This may be a reflection of caution on the part of most prescribers, as more information became available on the long-term effects of these products. Nimesulide, which has been available in Ireland since 1995, still retains a major share of the NSAID market. This is probably due to the intense marketing by its manufacturer regarding its selective COX-2i properties [8]. Therefore, although not licensed as a selective COX-2i, it may have been perceived as such by many GPs. Another influential factor may be that it is available in two dosage formulations (tablets and granules) and is manufactured locally, factors which may have been exploited when marketing the product.

The incidence of COX-2i use in new users was higher for females and those patients aged >65 years both before and after the withdrawal of rofecoxib. This trend in COX-2i use has been seen elsewhere [3] and may reflect a perceived greater need for their use in this population, possibly due to the higher risk of gastointestinal toxicity from nsNSAIDs in vunerable populations [9].

Finally, prescription of COX-2i's was also noted in patients with probable CV comorbidity, in each phase of the study. Statements issued by the national and European regulatory authorities (Irish Medicines Board and European Medicines Agency, respectively) in October 2004 reiterated to prescribers the advice for precaution in those patients with a history of CV disease. At this time, all COX-2i's available in Europe already contained warnings regarding heart problems. It is now accepted that persons at risk for CV disease should avoid these drugs. Our study demonstrates that despite these regulatory warnings, rofecoxib withdrawal and studies highlighting the associated risks, patients at risk of CV disease still continued to be prescribed the COX-2i's. There are numerous reasons why this prescribing behaviour may have continued: patients were unwilling to change therapies, a belief on behalf of prescribers that this was not a ‘class effect’, continued marketing strategies from the pharmaceutical companies, which may have capitalized on these beliefs, and a lack of unbiased information available to prescribers at the time. Ultimately the situation highlights the uncertainty experienced by prescribers of treatment alternatives available at this time for these at-risk groups.

Our study has some limitations. NSAID utilization has been described through analysis of a large prescription database, which relates to prescriptions dispensed. Therefore we cannot be sure that the medications were taken as prescribed. Also, with reference to the database, no information on patient diagnosis is recorded, nor do we have information on other potential risk factors pertaining to NSAID use, such as smoking, alcohol consumption or, indeed, CV risk factors [9]. However, the prescribing of certain disease-specific therapies has been used previously with great specificity and sensitivity (e.g. prescribing of aspirin and a nitrate together as a reliable marker of ischaemic heart disease [10]). The database also comprises large numbers of patients who can be followed on a longitudinal basis and where detailed information is available on all medications dispensed on an individual level. Regarding over-the-counter (OTC) use of NSAIDs, such as ibuprofen, it should be noted that the database cannot account for such use. However, patients covered by the GMS scheme are obliged to pay for nonprescription items and the use of nonprescription (OTC) items is therefore likely to be small.

Conclusions

The results of our study indicate that nsNSAIDs constitute the majority of NSAID prescribing for those initiating therapy in Ireland after rofecoxib withdrawal. The results also show that, while prescribing of celecoxib fell by almost 50%, female patients, those over 65 years and at risk of CV disease continued to be preferentially prescribed a COX-2i both before and after withdrawal of rofecoxib. The similar pattern in prescribing observed both before and after the withdrawal of rofecoxib may highlight the uncertainty experienced by prescribers of treatment alternatives available at this time for these at-risk groups.