What therapies have replaced rofecoxib in Ireland?

Abstract

What is already known about this subject

• Prior to the worldwide withdrawal of rofecoxib in 2004, due to cardiovascular toxicity with its use, selective COX-2 inhibitors had been shown to be preferentially prescribed to ‘at-risk’ patients in Ireland, including older females or those receiving antiulcer medications or multiple concomitant medications.

What this study adds

• During the 12 months post rofecoxib withdrawal the majority of those previously receiving chronic rofecoxib therapy were prescribed either no further nonsteroidal anti inflammatory drug (NSAID) therapy or short-term NSAID therapy only.

• Prescriptions for analgesic agents were initiated in half of the study group.

• These results suggest that the withdrawal of rofecoxib and the subsequent controversy may have led prescribers to question their use of NSAIDs in the first place and to adhere more closely to published guidelines on their use.

Aims

To examine prescription patterns of nonsteroidal anti-inflammatory drugs (NSAIDs) or analgesics in patients prescribed chronic rofecoxib treatment prior to withdrawal from the Irish market, and to determine the impact on proton pump inhibitor (PPI) co-prescription.

Methods

Using a national prescribing database, adults (≥16 years) prescribed rofecoxib for ≥3 months, but not analgesics, from January to September 2004 were identified. A longitudinal prescribing history was used to determine switching patterns to other cyclooxygenase (COX)-2 inhibitors, NSAIDs or analgesics during 3 and 12 months after withdrawal. Concomitant PPI prescription was examined. Logistic regression was used to determine the likelihood of switching to a COX-2 inhibitor vs. nonselective NSAID and factors influencing concomitant PPI prescription.

Results

After rofecoxib withdrawal, 30.2% (1558) and 17.9% (922) of the 5155 study subjects received no further NSAID prescription during 3 and 12 months, respectively. During the 12-month period, approximately one-third of NSAID prescriptions were for <3 months; 40.7% (2096) received sequential prescriptions for different NSAIDs. Co-prescription of analgesics occurred in 49.3% (2539) of subjects. Neither age nor gender influenced the type of NSAID prescribed in the 12 months post rofecoxib withdrawal. PPI prescription increased by 5.5% during the study, associated with use of nonselective NSAIDs, prior use of PPIs and increasing age.

Conclusions

The majority of those receiving chronic rofecoxib therapy were prescribed either no further NSAID or short-term NSAID therapy only during the 12 months post withdrawal, which suggests the subsequent controversy may have encouraged prescribers to adhere more closely to published guidelines.

Introduction

Rofecoxib was the first selective cyclooxygenase (COX)-2 inhibitor to be authorized for use in Ireland. We have previously shown that COX-2 inhibitors were preferentially prescribed to ‘at-risk’ patients, such as older females or those receiving antiulcer medications or multiple concomitant medications [1, 2].

In September 2004, rofecoxib was withdrawn from the global market because of concerns about cardiovascular toxicity with its use [3, 4]. No changes were recommended for the use of those COX-2 inhibitors remaining on the market in Ireland at that time [5]. However, restrictions on their use were subsequently introduced by the national and European regulatory authorities in February 2005 [6].

The aims of this study were to examine the prescription patterns of nonsteroidal anti-inflammatory drugs (NSAIDs) or analgesics in patients, previously prescribed rofecoxib, after its withdrawal from the market in September 2004, and to determine the impact of these prescription changes on the use of proton pump inhibitors (PPIs).

Methods

The National Shared Services Primary Care Reimbursement Service of the Health Service Executive in Ireland (HSE-PCRS, formerly GMS medical services) pharmacy claims database was used to identify the study population. The HSE-PCRS scheme provides free health services to approximately 30% of the Irish population (approximately 1.2 million people), including provision of medicines without charge [7]. Eligibility for this service is means tested and previous reports have estimated that it accounts for up to 70% of all medicines prescribed in primary care in Ireland [8]. In 2005, the HSE-PCRS drug expenditure was in excess of €831 million [7]. The HSE-PCRS pharmacy database records basic demographic information on all eligible patients (such as age and sex) and full details on all items dispensed in the scheme, including ingredient costs and pharmacist dispensing fees per item dispensed. Medicines are coded using the World Health Organization Anatomical Therapeutic Chemical (ATC) classification system [9]. It does not record data on diagnosis.

For the purpose of this study the HSE-PCRS pharmacy claims database was used to identify a cohort of patients, aged ≥16 years, who had been prescribed rofecoxib, but not analgesic agents, for three or more months (representing chronic use) during January to September 2004. This cohort was followed up by way of longitudinal monthly prescribing history from October 2004 to September 2005 to determine the pattern of NSAID and analgesic prescribing during the first 3- and 12-month periods post rofecoxib withdrawal. The prescribing pattern of concomitant PPIs was examined for the cohort before and after the withdrawal of rofecoxib. The following ATC codes were used: A02B (any PPI), M01A (any NSAID), M01AH (any selective COX-2 inhibitor) N02 (paracetamol- and aspirin-containing analgesic agents, excluding aspirin strengths <300 mg). Nimesulide (ATC M01AX 17) was separately evaluated. Logistic regression was used to determine the likelihood of switching to an alternative COX-2 inhibitor vs. a nonselective NSAID and the factors affecting concomitant prescription of any PPI post rofecoxib withdrawal. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated and are presented. Significance at P < 0.05 is assumed.

Results

The study cohort comprised 5155 patients who had been prescribed chronic rofecoxib, but not analgesic, therapy during January to September 2004. Of these, 2145 (41.6%) had received PPIs in the same period.

Three months post withdrawal

Table 1 outlines the prescribing pattern for NSAIDs and analgesic agents in the study cohort from October to December 2004 (i.e. the first 3 months post withdrawal). The largest single group received no NSAID therapy during this period (n = 1558, 30.2%). The remainder of the group was divided between nonselective NSAIDs, including nimesulide or another COX-2 inhibitor. Some patients received sequential prescriptions for different NSAIDs (classified as ‘sequential NSAIDs’). Co-prescription of analgesics was recorded in 1263 (26.7%) of the study cohort, consisting primarily of paracetamol-containing preparations, including coproxamol (81.4%). Less than 1% of prescriptions were for aspirin preparations. Approximately 30% of the patients receiving no further NSAID therapy were prescribed analgesic agents; the remainder of analgesic prescriptions were equally distributed among the NSAID prescribing groups.

Table 1.

Prescribing pattern of nonsteroidal anti-inflammatory drugs (NSAIDs) and analgesics in the first 3 months post rofecoxib withdrawal in the study cohort

NSAID group	Numbers (%) prescribed NSAIDs	Numbers (%) co-prescribed analgesics*
None	1558 (30.2)	346 (22.2)
Non-selective (excluding nimesulide)	1032 (20.0)	244 (23.6)
Other COX-2 inhibitor	1453 (28.2)	325 (22.4)
Nimesulide	433 (13.2)	114 (26.3)
Sequential NSAIDs	679 (13.2)	234 (34.5)
Total	5155	1263 (26.7)

^*Percentages based on figures in each NSAID.

One year post withdrawal

Table 2 outlines the prescribing pattern from October 2004 to September 2005 (i.e. up to 12 months post withdrawal). The largest group was of those receiving sequential NSAIDs during this time (n = 2096, 40.7%). Less than a third (24.5%) of NSAID prescriptions were for <3 months' duration. A total of 922 (17.9%) received no NSAID therapy. Co-prescription of analgesic agents occurred in 2539 (49.3%) of patients [primarily paracetamol-containing, including coproxamol (76.2%)]; the majority (54%) of these were for short-term (<3 months) use only.

Table 2.

Prescribing pattern of nonsteroidal anti-inflammatory drugs (NSAIDs) and analgesics in the 12 months post rofecoxib withdrawal in the study cohort

NSAID group	Numbers (%)	Chronic prescribing numbers (% total)
None	922 (17.9)
Non-selective (excluding nimesulide)	1105 (21.4)	747 (67.6)
Other COX-2 inhibitor	633 (12.3)	372 (58.8)
Nimesulide	399 (7.7)	266 (66.7)
Sequential NSAIDs	2096 (40.7)
Total	5155	1385 (24.5)

^*≥3 months' prescription.

The likelihood of switching to an alternative COX-2 inhibitor vs. a nonselective NSAID within 12 months of rofecoxib withdrawal was not influenced by the age (OR 1.1, 95% CI 0.76, 1.59) or gender (OR 1.07, 95% CI 0.91, 1.27) of the patient.

Proton pump inhibitor therapy

A total of 2145 patients (41.6% of the cohort) were coprescribed PPIs and rofecoxib at the time of entry into the study. Overall, there was a 5.5% increase in PPI prescribing post rofecoxib withdrawal, compared with the period before withdrawal. The PPI prescription rate was 37.7% (n = 1784) during the first 3 months and 47.1% (n = 2427) during the 12 months post withdrawal of rofecoxib (Table 3). Patients receiving no further NSAID therapy were significantly more likely to discontinue PPIs compared with the rest of the cohort, whereas those prescribed nonselective NSAIDs or a combination of different NSAIDs were significantly more likely to commence on PPI therapy during the two study periods (Table 3). Overall, prescribing of PPIs post rofecoxib withdrawal was related to increasing age, use of nonselective or sequential NSAIDs and prior prescription of PPIs (Table 4). Gender had no influence on subsequent PPI prescription.

Table 3.

Prescribing pattern of proton pump inhibitors (PPIs) at 3 and 12 months postrofecoxib withdrawal

NSAID group	Number (%)*initiating PPIs (3 months)	Number (%)*continuing PPIs (3 months)	Number (%)*initiating PPIs (12 months)	Number (%)*continuing PPIs (12 months)
None	41 (4.3)	360 (59.4)	67 (12.5)	248 (64.1)
Non-selective (excl. nimesulide)	57 (9.0)	326 (81.5)	150 (21.7)	360 (87.4)
Other COX-2 inhibitor	48 (5.9)	503 (79.2)	63 (16.8)	237 (91.5)
Nimesulide	16 (6.6)	151 (79.5)	37 (16.9)	160 (88.9)
Sequential NSAIDs	47 (12.9)	235 (74.8)	319 (26.8)	786 (86.7)
χ2P-value	<0.0001	<0.0001	<0.0001	<0.0001

^*Percentages based on figures for each subgroup in Table 2

Table 4.

Odds ratio and 95% confidence intervals for any proton pump inhibitor (PPI) prescribing in the 12 months after rofecoxib withdrawal

Variable		Odds ratio	95% CI
Age	45–64 vs. 16–44 years	1.20	0.8, 1.80
	≥65 vs. 16–44 years	1.89	1.29, 2.77*
Gender male vs. female		1.03	0.86, 1.23
Type of NSAID:	nonselective vs. none	1.59	1.23, 2.05*
	other Cox-2 vs. none	1.24	0.99, 1.56
	nimesulide vs. none	1.32	0.95, 1.83
	sequential vs. none	1.49	1.12, 1.97*
Previous prescription for PPI		43.1	36.0, 51.5*

^*P < 0.05.

Discussion

In this study prescribers were less likely to switch patients, previously receiving rofecoxib therapy, to another COX-2 inhibitor irrespective of age or gender, either in the short or longer term post rofecoxib withdrawal. These results contrast with findings from studies undertaken in this population prior to rofecoxib's withdrawal [1, 2], which showed that COX-2 inhibitors were preferentially prescribed to those considered to be ‘at-risk’ patients (older female patients and those receiving multiple medications), and that prescribers were more likely to switch older female patients and those with a past history of peptic ulcer disease from nonselective NSAIDs to selective COX-2 inhibitors. The change in prescribing pattern seen in the current study suggests that prescribers viewed the cardiovascular safety concerns reported with rofecoxib, which received widespread media attention, as indicative of a class effect with COX-2 inhibitors, despite initial reassurances from the national regulatory authority [5]. It may also reflect the patients' unwillingness to take such medicines, as a result of the negative media attention around the time of rofecoxib withdrawal. Since the changes in prescribing patterns predated the recommendations of the EU and national regulatory authorities [6], the results also highlight the potentially greater influence of the media in matters relating to drug safety compared with regulatory authorities. The importance of negative publicity on subsequent prescribing of hormone replacement therapy has been shown previously in this population [10].

The study has shown that 30.2% of the study cohort (who had been prescribed chronic rofecoxib therapy) was prescribed no NSAID therapy during the initial 3 months post rofecoxib withdrawal and, in fact, 17.9% were prescribed no NSAID therapy throughout the entire 12-month period of the study. There are two possible explanations for this. The results raise the question of the need for chronic NSAID therapy in the study group in the first place; this is supported by the fact that in addition to those patients who received no further NSAID therapy, short-term (<3 months) use was noted in approximately one-third of each NSAID prescription group during the 12 months of follow-up (Table 2). The efficacy of chronic NSAID therapy in the management of chronic pain is not established [11, 12]. Moreover, current guidelines for the management of osteoarthritis (primarily a non-inflammatory disease) recommend the use of simple analgesics such as paracetamol and restrict the use of NSAIDs to the management of acute exacerbations [13]. Alternatively, the findings may suggest that these patients were being undertreated. This seems unlikely, as only 30 and 35.5%, respectively, of these patient groups were prescribed analgesic therapy during the two periods of study, of which the majority was for short-term use only. Since it is estimated that pain is the most common symptomatic reason to seek medical consultation [14], it is likely that symptomatic patients would have sought further therapy in terms of different NSAIDs or analgesic agents. Our findings suggest that prescribers may have reviewed the need for chronic NSAID therapy in their patients following the controversy surrounding the withdrawal of rofecoxib [15, 16] and the subsequent publication of amended guidance on NSAID usage from the European regulators [6]. This hypothesis is supported by the extent of analgesic prescribing in the study cohort, which had not been recorded prior to rofecoxib withdrawal.

Selective COX-2 inhibitors were developed with the aim of maintaining anti-inflammatory activity while minimizing gastrointestinal toxicity [17]. Their arrival was seen as a significant breakthrough in reducing the burden of NSAID-related gastropathy [18]. However, because of their significantly greater cost compared with nonselective NSAIDs, some experts recommended their use should be restricted to ‘at-risk’ patients such as those with a past history of gastropathy [19]. The fact that almost 50% of our study cohort was prescribed concomitant PPI and rofecoxib therapy appears to support the preferential usage of rofecoxib in at-risk patients. However, such concomitant usage suggests that prescribers still had concerns about the gastroprotective efficacy of COX-2 inhibitors. This is supported by the finding that PPI prescribing fell significantly in those patients who received no further NSAID therapy after rofecoxib withdrawal.

Nimesulide prescribing was evaluated separately during the study because it holds a particular position in the Irish market [2]. Although not licensed as a selective COX-2 inhibitor [20], it may have been perceived by prescribers as having similar properties, as suggested by the company's marketing campaign [21]. Our results show that during the first 3 months post rofecoxib withdrawal, nimesulide prescription was noted in 13.2% of the study cohort. This represented a relatively large proportion of patients receiving any NSAID prescription. However, the prescription rate of nimesulide during the 12-month period fell to just 7.7% of the cohort. This fall mirrors the fall in prescribing of the other COX-2 inhibitors seen between 3 and 12 months and suggests that prescribers viewed nimesulide as having not only the potential benefits of COX-2 inhibitors but also the potential cardiovascular toxicity risks.

Our study has a number of possible limitations that deserve mention. The pharmacy claims database relates to prescriptions dispensed and therefore we cannot be sure that the medications were taken as prescribed. Although the total quantity of drug prescribed is recorded, it is not possible to estimate the actual daily dosage taken by the patient for the purpose of this study, because data on indications for use or dosage regimen are not recorded. However, the database is large and patients can be followed on a longitudinal basis with detailed information available on all medications at an individual level. Patients were included in the study cohort only if they had received at least 3 months' prescription of rofecoxib during the initial prewithdrawal period, to reflect chronic usage, and patients receiving aspirin- and paracetamol-containing analgesic agents during this time were not included. The prescribing pattern was evaluated for 9 months before and 12 months after the withdrawal of rofecoxib in order to identify changes in prescribing patterns over time in the study cohort. The database cannot account for the use of over-the-counter (OTC) NSAIDs such as aspirin, paracetamol or ibuprofen, in addition to prescribed NSAIDs. Other studies evaluating the pattern of NSAID use have also been unable to determine accurately the level of nonprescription usage of NSAIDs [22, 23]. It is important to note that patients covered by the HSE-PCRS free drugs scheme are obliged to pay in full for nonprescription items. Since there is therefore a strong financial incentive not to comedicate with OTC NSAIDs or analgesics, it is likely that such usage is small in our study group.

Conclusions

After the withdrawal of rofecoxib, prescribers were less likely to switch patients to another COX-2 inhibitor compared with nonselective NSAIDs, suggesting that the problems reported with rofecoxib were viewed as a potential class-related effect. Since the majority of those receiving chronic rofecoxib therapy were prescribed either no further NSAID therapy or short-term NSAID therapy only during the 12 months post withdrawal, these results suggest that the withdrawal of rofecoxib and the subsequent controversy may have led prescribers to question their use of NSAIDs in the first place and to adhere more closely to published guidelines on their use.