Channeling and prevalence of cardiovascular contraindications in users of cyclooxygenase 2 selective nonsteroidal antiinflammatory drugs

Authors


Abstract

Objective

To assess use and channeling of cyclooxygenase 2 selective inhibitors (coxibs) over time and to estimate the percentage of coxib users with cardiovascular contraindications.

Methods

The study population comprised all coxib and nonselective nonsteroidal antiinflammatory drug (NSAID) users in the Integrated Primary Care Information project between January 2000 and December 2004. The prevalence of risk factors for NSAID-related upper gastrointestinal ulcer complications, cardiovascular disease, and cerebrovascular disease at the start of treatment was compared between users of coxibs and users of nonselective NSAIDs.

Results

The study population included 72,841 nonselective NSAID users and 10,739 coxib users. The prevalence of risk factors for NSAID-related gastrointestinal complications was higher in coxib users than nonselective NSAID users (odds ratio [OR] 1.18, 95% confidence interval [95% CI] 1.10–1.26). Similarly, the prevalence of prior cardiovascular disease was higher in coxib users than in nonselective NSAID users (OR 1.35, 95% CI 1.28–1.43). Channeling of coxibs to patients with NSAID-related gastrointestinal risk factors declined after 2001 but increased again in 2004, whereas the channeling of coxibs to patients with cardiovascular disease remained constant. Less than 15% of all coxib users had history of ischemic coronary or cerebrovascular disease. Among coxib users with increased risk for NSAID-related gastrointestinal disorders, 27% had history of ischemic coronary or cerebrovascular disease.

Conclusion

This study demonstrates that coxibs were preferentially prescribed to patients with risk factors for NSAID-related gastrointestinal disorders and/or cardiovascular diseases. Only one-quarter of coxib users with increased risk for NSAID-related gastrointestinal complications had cardiovascular conditions compatible with recent European safety contraindications for coxibs.

INTRODUCTION

Nonselective nonsteroidal antiinflammatory drugs (NSAIDs) are commonly used in the treatment of arthritis, pain, and stiffness, but are associated with a 2–3-fold increase in the risk of upper gastrointestinal (GI) complications such as bleeding, ulcers, and perforation (1, 2) due to the inhibition of cyclooxygenase 1 (COX-1). COX-2 selective inhibitors (coxibs) are efficacious pain relievers and reduce the risk of serious GI complication as compared with nonselective NSAIDs (3–5).

Based on evidence from clinical trials and cost-effectiveness analyses, the Dutch General Practice guidelines (2003; in line with international guidelines) recommend that coxibs should be used only in persons with an increased risk of NSAID-related GI complications (6–8). The cardiovascular safety profile of rofecoxib has been discussed since the Vioxx Gastrointestinal Outcomes Research Study trial was published in 2000 (3). In 2002, the Food and Drug Administration (FDA) required a change in the package insert advising that caution should be exerted when rofecoxib is used in patients with a medical history of ischemic disease. The drug was removed from the market in September 2004 because of cardiovascular safety problems (9). In February 2005, the European Medicines Agency contraindicated the use of any coxibs in patients with ischemic heart disease or stroke, and the use of etoricoxib in patients with hypertension (10). An advisory committee recommended in February 2005 that the FDA issue a black box warning regarding the risk of heart attack and stroke for all coxibs. Due to the discussion about cardiovascular safety, the new contraindications, and the clustering of GI risk and cardiovascular conditions, physicians may be confused about who could be prescribed coxibs.

Only a few studies are available on the use of coxibs in the general population, and all of these studies relate to North America (11, 12). In addition, little or no information is available on the concurrent prevalence of NSAID-related GI risk and cardiovascular disease in coxib users.

Evidence of channeling of coxibs to persons with NSAID-related GI risk factors is consistently described in all observational studies that investigate the GI and/or cardiovascular effects of coxibs as compared with nonselective NSAIDs, both in Canada and the US (13–17). A UK study of GI outcomes demonstrated that coxibs are prescribed to patients with a greater disease severity, more comorbid conditions, and higher prevalence of GI risk factors (18). The concurrent existence of cardiovascular disease risk and GI disease risk was not described.

In view of the limited information on use and channeling of coxibs in Europe, the recent safety restrictions, and the current confusion among physicians concerning who might receive coxibs, our objective was to describe the extent of coxib use in the first 5 years of marketing, to assess patterns of channeling of coxibs between 2000 and 2004, and to assess the percentage of coxib users with and without concurrent risk factors for NSAID-related GI complications and cardiovascular disease.

PATIENTS AND METHODS

Role of the funding source.

Pfizer was involved in the design of a small channeling study that covered the time period of 2000–2003. The decision to extend the data to 2004, to look at time trends, to adjust for physicians, and to look at the percentage of patients with contraindications was made independently by Erasmus University Medical Center. Pfizer had the opportunity to comment on the manuscript within 30 days prior to its submission, however Erasmus University Medical Center had full rights to publication.

Setting.

We conducted a population-based study in the Integrated Primary Care Information (IPCI) database. The IPCI database is a general practice research database in the Netherlands that comprises the complete electronic medical records of >600,000 patients. Details of the database have been published elsewhere (19). In brief, the database contains anonymous data on patients' demographics, signs and symptoms, physical evaluation and findings, diagnoses, prescriptions, referrals, laboratory examinations, and summaries of discharge letters. There is a complete record of all drug prescriptions, their indication, and dose regimen. To maximize completeness of the data, general practitioners (GPs) participating in the IPCI project are not allowed to maintain a system of paper-based records besides the electronic medical records. The IPCI system complies with the European Union guidelines on the use of medical data for research and has been proven valid for pharmacoepidemiologic research (19). The Scientific and Ethical Advisory Board of the IPCI project approved this study.

The source population for description of drug use comprised all patients registered in the IPCI database during the study period (January 2000 to December 2004). Each patient was followed from registration with the GP or start of the study period until death, transferring to another practice, or end of the study period, whichever occurred first. The population for assessment of channeling and prevalence of concurrent risk factors for NSAID-related GI complications and cardiovascular disease comprised all persons with at least 12 months of valid database history.

Drug use was categorized as use of nonselective NSAIDs (all M01A except for COX-2 preferential drugs and COX-2 selective drugs) and use of coxibs (rofecoxib, celecoxib, etoricoxib, valdecoxib). In the assessment of channeling, diclofenac plus misoprostol (in a fixed combination) and COX-2 preferential drugs (meloxicam, nabumetone) were excluded from the nonselective NSAIDs class.

Patient characteristics.

The first time a patient received a nonselective NSAID or coxib prescription during the study period, we assessed age, sex, and the prevalence or history of the following diseases and conditions 12 months prior to the prescription: GI comorbidity (peptic ulcer disease, GI hemorrhage, gastritis, duodenitis, dyspepsia, and abdominal pain), cardiovascular comorbidity (ischemic heart disease, chronic heart failure, hypertension, and hyperlipidemia), stroke, and diabetes mellitus. Prior use of systemic corticosteroids, antacids, histamine H2 receptor antagonists, proton-pump inhibitors, other ulcer-healing drugs, and cardiovascular drugs (positive inotropics, diuretics, beta-blockers, angiotensin-converting enzyme inhibitors plus angiotensin II receptor blockers, vasodilators, anticoagulants, lipid-lowering drugs) issued in the 6-month period prior to the prescription of the nonselective NSAID or coxib was assessed.

As an overall measure of NSAID-related GI risk, the cumulative number of generally recognized risk factors for NSAID-related upper GI ulcer complications was calculated (age >65 years, use of systemic corticosteroids, use of anticoagulants or acetylsalicylic acid, and a history of peptic ulcer or GI bleeding). Patients were considered to have cardiovascular disease (contraindications) if they had a history of stroke, heart failure, or ischemic heart disease.

Statistical analysis.

Use of nonselective NSAIDs and coxibs was described in terms of users per 1,000 person-years of followup. Logistic regression was used for assessment of channeling; in these analyses, the odds of patient characteristics were separately compared between users of coxibs and users of nonselective NSAIDs while adjusting for age and sex. Therefore, the observational unit in the logistic regression was a prescription of either a coxib (outcome 1) or a nonselective NSAID (outcome 0). Approximately 8% of the patients were entered multiple times since they had more than 1 index prescription. To account for the effect of prior use of another study drug, we considered prior use as covariate in the logistic regression analysis. To examine patient characteristics as a determinant for use of coxibs while adjusting for prescribing preference within a GP practice in addition to adjusting for age and sex, we used logistic regression with practice as a random effect using PROC NLMIXED in SAS (SAS Institute, Cary, NC). Calendar year stratified analyses were conducted to verify trends in channeling over time. All analyses were performed with SPSS version 11 (SPSS, Chicago, IL) and SAS Release 8.2 (SAS Institute).

RESULTS

In the source population of >470,000 patients during 2000–2004, use of nonselective NSAIDs decreased from 161 to 131 users per 1,000 person-years, whereas use of coxibs increased from 0 to 32 users per 1,000 person-years (Figure 1). The subpopulation for assessment of channeling comprised 77,177 patients. Nonselective NSAIDs were used by 72,841 patients and coxibs were used by 10,739 patients. A total of 6,403 patients received more than 1 prescription for a study drug and therefore were entered more than once in the study. The mean percentage of coxib prescriptions over the total number of NSAIDs per practice increased from 6% in 2000 to 28% in 2004; 10% of the practices never prescribed coxibs during the study.

Figure 1.

Use of nonselective nonsteroidal antiinflammatory drugs (NSAIDs; diamonds) and coxibs (squares) between 2000 and 2004. The X-axis represents the year of observation and the Y-axis represents the number of NSAID users per 1,000 person-years (PY) of observation.

Patients starting coxibs were more likely to be female (65.7% versus 56.8%) and were older than those starting nonselective NSAIDs (mean ± SD age 58.3 ± 17.0 years versus 46.1 ± 17.7 years) (Table 1). Coxib users also received NSAIDs (31.5% versus 10.7%) and acid suppressive drugs (16.6% versus 6.8%) more frequently in the previous 6 months than nonselective NSAID users (Table 1). Use of cardiovascular drugs was 2.5 times higher among coxib users than nonselective NSAID users. The differences in prior or concomitant drug use diminished substantially after adjustment for age, sex, and practice, but remained statistically different (Table 1). In line with the observed differences in drug use, the prevalence/history of GI disease, cardiovascular disease, cerebrovascular disease, and diabetes mellitus was higher among coxib users than NSAID users (Table 1). Patients with ischemic heart disease and history of stroke were 32% and 23% more likely to receive a coxib, respectively. The most important determinants for prescription of a coxib were use of NSAIDs in the previous 6 months (odds ratio [OR] 3.23, 95% confidence interval [95% CI] 3.07–3.41), use of acid suppressive drugs (OR 2.01, 95% CI 1.89–2.15), and a history of GI disorders (OR 1.82, 95% CI 1.72–1.92).

Table 1. Characteristics of coxib and nonselective NSAID users and the association between use of coxibs and aggregated gastrointestinal risk factors, medication, and cardiovascular disease*
 Coxib (n = 10,739)NSAID (n = 72,841)Population average OR (95% CI)Practice adjusted OR (95% CI)
  • *

    Values are the number (percentage) unless otherwise indicated. NSAID = nonsteroidal antiinflammatory drug; OR = odds ratio; 95% CI = 95% confidence interval; PPIs = proton-pump inhibitors; CV = cardiovascular; ACE = angiotensin-converting enzyme; GI = gastrointestinal.

  • OR adjusted for age and sex.

  • OR adjusted for age and sex, calculated with PROC NLMIXED in SAS (SAS Institute, Cary, NC).

  • §

    GI risk factors comprised age >65 years, use of systemic corticosteroids, use of anticoagulants or acetylsalicylic acid, and a history of peptic ulcer or GI bleeding; cardiovascular risk factors comprised a history of stroke, heart failure, and ischemic heart disease.

Female7,054 (65.7)41,354 (56.8)1.36 (1.30–1.42)1.47 (1.41–1.54)
Age, mean ± SD years58.3 ± 17.046.10 ± 17.7  
Age categories, years    
 0–20158 (1.5)5,612 (7.7)  
 21–401,559 (14.5)23,560 (32.3)  
 41–604,117 (38.3)28,067 (38.5)  
 61–803,904 (36.4)13,446 (18.5)  
 ≥801,001 (9.3)2,156 (3.0)  
Indication    
 Osteoarthritis1,130 (10.5)932 (1.3)Reference 
 Rheumatoid arthritis165 (1.5)333 (0.5)0.58 (0.47–0.72)0.94 (0.79–1.12)
 Other musculoskeletal disorders5,435 (50.6)31,311 (43.0)0.29 (0.27–0.32)0.41 (0.38–0.43)
 Other indications682 (6.4)10,776 (14.8)0.12 (0.10–0.13)0.15 (0.14–0.17)
 Unknown3,327 (31.0)29,489 (40.5)0.16 (0.14–0.17)0.19 (0.17–0.21)
NSAID use in 6 months prior3,381 (31.5)7,805 (10.7)3.02 (2.88–3.18)3.23 (3.07–3.41)
Acid suppressive agents and ulcer-healing drugs1,786 (16.6)4,973 (6.8)1.97 (1.85–2.09)2.01 (1.89–2.15)
 Antacids91 (0.9)436 (0.6)1.11 (0.88–1.40)1.17 (0.91–1.50)
 H2 receptor antagonist488 (4.5)1,782 (2.5)1.40 (1.26–1.56)1.52 (1.36–1.70)
 PPIs1,271 (11.8)2,982 (4.1)2.26 (2.11–2.43)2.21 (2.05–2.39)
 Other ulcer-healing drugs54 (0.5)144 (0.2)2.04 (1.47–2.83)2.14 (1.51–3.04)
CV drugs3,727 (34.7)12,674 (17.4)1.28 (1.22–1.35)1.28 (1.21–1.35)
 Acetylsalicylic acid515 (4.8)1,678 (2.3)1.08 (0.97–1.20)1.17 (1.05–1.31)
 Beta-blocker1,399 (13.0)5,064 (7.0)1.18 (1.10–1.26)1.17 (1.09–1.26)
 Positive inotropics148 (1.4)429 (0.6)0.90 (0.75–1.10)1.02 (0.98–1.06)
 Diuretics1,281 (11.9)3,697 (5.1)1.14 (1.06–1.22)1.17 (1.08–1.26)
 ACE inhibitors1,304 (12.1)3,936 (5.4)1.32 (1.23–1.42)1.32 (1.23–1.42)
 Other antihypertensive drugs1,063 (9.9)3,083 (4.2)1.24 (1.15–1.34)1.19 (1.09–1.29)
 Antiplatelet drugs and anticoagulants1,294 (12.1)3,871 (5.3)1.18 (1.10–1.27)1.21 (1.12–1.31)
 Lipid-lowering drugs887 (8.3)2,958 (4.1)1.37 (1.26–1.48)1.30 (1.19–1.41)
Other drugs    
 Oral corticosteroids543 (5.1)1,681 (2.3)1.53 (1.38–1.69)1.69 (1.51–1.88)
Prevalence of GI comorbidity2,574 (24.0)9,379 (12.9)1.92 (1.83–2.02)1.82 (1.72–1.92)
 Peptic ulcer255 (2.4)712 (1.0)1.96 (1.68–2.27)1.60 (1.46–1.75)
 GI bleeding102 (1.0)243 (0.3)2.03 (1.59–2.59)1.72 (1.33–2.22)
 Gastritis and duodenitis1,479 (13.8)4,789 (6.1)2.11 (1.98–2.25)1.86 (1.74–2.00)
 Other disorders of stomach and duodenum1,773 (16.5)6,695 (9.2)1.82 (1.72–1.93)1.75 (1.64–1.86)
Prevalence of CV comorbidity3,733 (34.8)12,835 (17.6)1.39 (1.32–1.46)1.35 (1.28–1.43)
 Ischemic heart disease939 (8.7)2,628 (3.6)1.34 (1.24–1.46)1.32 (1.21–1.44)
 Heart failure433 (4.0)901 (1.2)1.32 (1.16–1.49)1.25 (1.09–1.42)
 Hypertension1,972 (18.4)6,616 (9.1)1.32 (1.24–1.40)1.27 (1.20–1.36)
 Hyperlipidemia1,129 (10.5)3,850 (5.3)1.38 (1.28–1.48)1.29 (1.20–1.40)
Prevalence of other diseases    
 Diabetes853 (7.9)2,747 (3.8)1.38 (1.27–1.50)1.33 (1.22–1.45)
 Cerebrovascular disease489 (4.6)1,442 (2.0)1.21 (1.09–1.35)1.23 (1.10–1.39)
GI risk factors    
 05,920 (55.1)57,839 (79.4)  
 13,476 (32.4)11,717 (16.1)1.12 (1.05–1.20)1.09 (1.02–1.17)
 ≥21,343 (12.5)3,285 (4.5)1.29 (1.18–1.41)1.31 (1.19–1.44)
Combined GI and CV risk factors§    
 No GI and no CV risk factors5,654 (52.7)56,802 (78.0)  
 At least 1 GI and 1 CV risk factor1,297 (12.1)3,379 (4.6)1.36 (1.24–1.49)1.47 (1.34–1.60)
 No GI and at least 1 CV risk factor266 (2.5)1,037 (1.4)1.87 (1.62–2.15)2.68 (2.58–2.78)
 At least 1 GI risk factor and no    
  CV risk factor3,522 (32.8)11,623 (16.0)1.16 (1.08–1.24)3.02 (2.89–3.15)

Aggregation of risk factors showed that an increasing number of risk factors for NSAID-related GI complications increased the probability of receiving a coxib (Table 1). The probability of receiving a coxib if at least 1 NSAID-related GI risk factor was present decreased from a 40% increase to no increased probability in 2003 (Figure 2). Patients with cardiovascular disease were 20–30% more likely to receive a coxib than patients without cardiovascular disease, and this pattern did not change over time (Figure 3). In the absence of GI disease, cardiovascular disease remained an independent risk factor for receiving coxibs (Table 1).

Figure 2.

Probability of receiving coxibs for patients with at least 1 risk factor for nonsteroidal antiinflammatory drug–related gastrointestinal complications compared with patients without these risk factors (odds ratio adjusted for practice, age, and sex). The X-axis represents the year of observation and the Y-axis represents the odds ratio calculated with PROC NLMIXED in SAS (SAS Institute, Cary, NC).

Figure 3.

Probability of receiving coxibs for patients with cardiovascular disease compared with patients without cardiovascular disease (odds ratio adjusted for practice, age, and sex). The X-axis represents the year of observation and the Y-axis represents the odds ratio calculated with PROC NLMIXED in SAS (SAS Institute, Cary, NC).

Approximately 45% of coxib users had ≥1 baseline risk factors for NSAID-related upper GI complications, whereas these risk factors were present in 20.2% of nonselective NSAID users. Overall, <15% of coxib users would have had a cardiovascular contraindication when we applied the February 2005 safety restrictions to the study. Among persons with at least 1 risk factor for NSAID-related GI complications, 27% also had a contraindicated cardiovascular condition.

DISCUSSION

This study demonstrated that coxib use gradually increased at the expense of nonselective NSAID use during the first 5 years of marketing in the Netherlands. In line with the diffusion of coxib use, channeling related to the presence of a risk factor for NSAID-related GI complications decreased until 2004. We estimated that 85% of all coxib users and 75% of persons with increased risk of NSAID-related GI complications have no history of contraindicated cardiovascular disease and would remain eligible for a coxib prescription.

In view of debates about the cardiovascular safety of coxibs, it is noteworthy that persons with cardiovascular disease (even in the absence of GI risk factors) were more likely to receive a coxib (20). Also, despite channeling to persons with GI risk factors, most coxib users are persons without risk factors for NSAID-related GI disorders. Our findings are in line with studies in the UK and US, which have shown that coxibs are mostly prescribed to persons without increased risk of NSAID-related GI disorders (18, 21).

Our data on channeling of coxibs are in line with group differences in observational studies that compared GI or cardiovascular safety between coxibs and NSAIDs (13–18). MacDonald et al (18) and Schneeweiss et al (21) demonstrated that coxib users in the US and in the UK also had a higher prevalence of GI risk factors. Cohort studies (US and Canada) that compared cardiovascular outcomes between coxibs and nonselective NSAIDs showed that coxib users received more cardiovascular comedication and had more comorbidity, higher prior use of NSAIDs, and more prescriptions at baseline (13–17), which is in line with our data.

The differences between the population average and the practice-adjusted ORs indicate that some, but not all, of the observed overall channeling is due to patient characteristics. This observation is consistent with studies in Medicare (21, 22) showing that first-time selective COX-2 inhibitor prescribing is dependent on both GI risk factors and physicians' preference in the first 2 years of the availability. Our data also showed an impact of physicians' preference in relation to the other determinants, although the impact was less substantial in the 5 years of observation.

Our study has some limitations that should be considered when reviewing the results. First, we conducted a cohort study to which a patient could contribute multiple times if he or she received different NSAIDs during the study. Approximately 8% of persons entered the study twice. Sensitivity analyses in which patients were included only once had only a minor effect on the results and did not change the conclusion. Second, the IPCI database did not contain enough coxib users to enable us to analyze individual coxibs. As a result, we combined the individual coxibs into one group. Lastly, there could be dependency in the data regarding general practices that preferentially prescribe a certain NSAID. Even though we attempted to adjust for this phenomenon with nonlinear mixed model, it is no guarantee that residual confounding by practice has been eliminated completely.

In conclusion, this study demonstrated that the use of coxibs has increased substantially between 2000 and 2004. The baseline prevalence of these factors was ∼50% among users of coxibs and 20% among users of nonselective NSAIDs. Only one-quarter of coxib users with increased risk of NSAID-related GI complications had cardiovascular conditions compatible with recent European safety contraindications for coxibs.

Ancillary