Dr. Greenberg is recipient of a Physician Scientist Development award from the Arthritis Foundation/American College of Rheumatology. Dr. Bingham is recipient of an Arthritis Investigator award from the Arthritis Foundation.
Assessment of coxib utilization by rheumatologists for nonsteroidal antiinflammatory drug gastroprotection prior to the coxib market withdrawals
Article first published online: 27 JUL 2006
Copyright © 2006 by the American College of Rheumatology
Arthritis Care & Research
Volume 55, Issue 4, pages 543–550, 15 August 2006
How to Cite
Greenberg, J. D., Bingham, C. O., Abramson, S. B., Reed, G., Kishimoto, M., Hinkle, K. and Kremer, J. (2006), Assessment of coxib utilization by rheumatologists for nonsteroidal antiinflammatory drug gastroprotection prior to the coxib market withdrawals. Arthritis & Rheumatism, 55: 543–550. doi: 10.1002/art.22095
- Issue published online: 27 JUL 2006
- Article first published online: 27 JUL 2006
- Manuscript Accepted: 7 DEC 2005
- Manuscript Received: 18 JUL 2005
- Nonsteroidal antiinflammatory drug
To examine cyclooxygenase 2 inhibitor (coxib) utilization by rheumatologists for patients receiving nonsteroidal antiinflammatory drugs (NSAIDs) prior to the coxib market withdrawals.
A prospective study of patients with rheumatoid arthritis enrolled in the Consortium of Rheumatology Researchers of North America registry was performed.
Of 1,833 patients receiving prescription NSAIDs, 1,380 (75.3%) received gastroprotection, defined as either coxib monotherapy and/or gastroprotective agent (GPA) cotherapy, and 1,207 (65.8%) received coxibs. The distribution of gastroprotective strategies included 860 (46.9%) patients who were prescribed coxib monotherapy, 347 (18.9%) prescribed dual coxib plus GPA cotherapy, 173 (9.4%) prescribed a nonselective NSAID (NS-NSAID) plus GPA cotherapy, and 453 (24.7%) prescribed an NS-NSAID without GPA cotherapy. For patients with 0, 1, and ≥2 identifiable gastrointestinal (GI) risk factors, coxib prescribing rates as a proportion of NSAID agents were 64.1%, 66.4%, and 68.6%, respectively; among dual aspirin/NSAID users, coxib prescribing rates were 66.2%, 78.3%, and 68.5% of NSAID prescriptions, respectively.
The majority of NSAID users were prescribed a gastroprotective strategy, primarily attributable to coxib utilization. Coxib utilization rates were consistently high across all levels of GI risk, including patients without identifiable risk factors. These data indicate that rheumatologists broadly adopted the coxib class of NSAIDs in a nonselective manner with respect to underlying GI risk and concomitant aspirin use. As novel therapeutic classes are introduced, early evaluation of prescribing patterns using arthritis registries can determine the appropriateness of prescribing patterns and may improve patient outcomes.
Nonsteroidal antiinflammatory drugs (NSAIDs) are among the most frequently used medications in the US (1). However, numerous studies have demonstrated the association of NSAIDs and serious ulcer-related gastrointestinal (GI) complications (2–4). In epidemiologic studies, >70,000 hospitalizations and >7,000 deaths have been attributed to NSAIDs per year (3). When selective cyclooxygenase 2 inhibitors (coxibs) were introduced to the market, they were widely regarded as a novel therapeutic class that would reduce the risk of NSAID-related gastropathy.
For NSAID-induced gastropathy, randomized controlled trials have previously established that the risk of gastropathy can be reduced using cotherapy with a proton-pump inhibitor (PPI) or misoprostol (5). In the past decade, randomized clinical trials of coxib agents have demonstrated that coxib monotherapy is also a highly efficacious strategy to reduce the risk of NSAID gastropathy (5–7). As a result, both of these gastroprotective strategies were incorporated into the treatment guidelines of both the American College of Rheumatology (ACR) and the Arthritis Foundation (8–10). These guidelines clearly state that gastroprotective strategies should be reserved for patients with established GI risk factors, and that not all NSAID users require gastroprotection.
Recent evidence from placebo-controlled clinical trials indicates that all 3 coxib agents approved by the Food and Drug Administration may increase the risk of cardiovascular events (11–13). Based on the results of these clinical trials and other safety concerns, both rofecoxib and valdecoxib were withdrawn from the market. Although the risk may differ among the various coxib agents, these studies and their associated drug withdrawals have highlighted concerns regarding the appropriate utilization of coxibs for patients with increased GI risk.
Much of the initial rationale for prescribing coxib agents for patients with arthritis was the high mortality rate associated with NSAID gastropathy in the precoxib era (3). The initial observational studies, undertaken soon after the introduction of coxib agents, reported underutilization of NSAID gastroprotective strategies (14–16). However, more recent population-based studies have observed that rates of gastroprotection have increased, and that coxib prescribing has become increasingly prevalent and nonselective (17, 18). Studies examining NSAID gastroprotective prescribing patterns of rheumatologists, however, have been extremely limited (19). As we have previously demonstrated, patients with arthritis who receive NSAIDs are frequently coprescribed aspirin for cardiovascular prophylaxis (20). Studies to date have not examined gastroprotective strategies prescribed for dual aspirin/NSAID users.
Observational studies examining adherence to clinical guidelines and quality indicators have demonstrated that specialists are more likely to adhere to specialty-based guidelines and quality indicators than generalists (21–24). We have previously demonstrated that cardiovascular risk was not an independent determinant of coxib utilization by rheumatologists (20). Using the Consortium of Rheumatology Researchers of North America (CORRONA) registry, a multicenter prospective study, we sought to evaluate the selectivity of coxib utilization by rheumatologists for patients with varying levels of GI risk prior to the rofecoxib withdrawal.
PATIENTS AND METHODS
Data sources and data collection.
A total of 2,690 consecutive patients with rheumatoid arthritis (RA) in the CORRONA registry who were prescribed NSAIDs and/or low-dose aspirin from March through September 2004 were included in the study. These 2,690 patients were treated by 114 rheumatologists at 47 rheumatology practices across the US. Details of the CORRONA registry and participating rheumatologists have been previously described (20). Briefly, the CORRONA registry is a prospective registry of patients with various forms of arthritis who are under the care of rheumatologists. Since March 2002, both academic rheumatologists and private practice rheumatologists in the US have joined the consortium. Completed questionnaires are entered directly online through a secure Web site. Approval for the CORRONA registry was obtained from the respective institutional review boards of participating academic sites and a central private institutional review board for patients from private practice sites.
The study population included consecutive patients with RA who were enrolled in CORRONA with updated clinical information obtained during the 7-month study period between March 1, 2004 and September 30, 2004. This study period was selected to ensure that complete data on GI risk factors were available. In March 2004, the CORRONA questionnaire forms were updated to include more comprehensive patient data, including detailed questions on prescription anticoagulants and antiplatelet agents. The end of the study period was selected based on the date of withdrawal of rofecoxib from the worldwide market (25). A total of 2,690 patients with RA were included in the study. These patients were recruited into the CORRONA registry by 114 treating rheumatologists who were located at 47 sites across the US.
Medication and clinical data.
Data were prospectively collected during the study period from patient and physician questionnaires administered during routine clinical encounters. All medication data were obtained from patient questionnaires that were completed during clinical encounters. Specifically, use of prescription NSAIDs, including nonselective NSAIDs (NS-NSAIDs), coxibs, PPI agents, misoprostol, oral corticosteroids, warfarin, and aspirin, was reported on the patient questionnaires. Meloxicam was categorized as an NS-NSAID for the purposes of this study, distinct from the coxib class. The coxib class included celecoxib, rofecoxib, and valdecoxib. Low-dose aspirin was defined as either 81 mg or 325 mg daily. For each patient, medication use was based on the most recent clinical encounter during the study period. Patient comorbidities were defined based on physician-reported diagnoses that were included in the physician questionnaires.
Our study sample consisted of 3 cohorts: patients receiving prescription NSAIDs, patients receiving low-dose aspirin, and patients receiving both prescription NSAIDs and low-dose aspirin. The index date was defined as the most recent encounter during the study period in which a prescription NSAID or low-dose aspirin was prescribed. Within the prescription NSAID cohort, we defined 4 mutually exclusive subcohorts based on patients' medication use at their last encounter during the study period: NS-NSAID users without gastroprotective agent (GPA) cotherapy, NS-NSAID users with GPA cotherapy, coxib users without GPA cotherapy, and coxib users with GPA cotherapy. In our study, only PPI agents and misoprostol were included as GPA cotherapy for NSAID users based on these medications' proven efficacy in randomized trials and their inclusion in treatment guidelines (8–10, 26). H2 receptor antagonists were excluded based on their lack of efficacy in preventing NSAID-induced ulceration and exclusion from various treatment guidelines (26). Similarly for aspirin users, only PPI agents and misoprostol were included as GPA cotherapy using the same criteria, namely, proven efficacy in clinical trials and inclusion in clinical guidelines (8, 26–28). Gastroprotection for dual users of aspirin and NSAIDs was defined as PPI or misoprostol cotherapy, but not coxib use, based on recent consensus recommendations and the results of 2 large coxib trials (7, 8, 29). Selected professional organizations' recent consensus recommendations for NSAID and aspirin gastroprotection are summarized in Table 1.
|NSAID users||Aspirin users||NSAID/aspirin users|
|Cotherapy with PPI/misoprostol|
|American College of Gastroenterology (26)||√||√||NA|
|American College of Rheumatology (9, 10)||√||NA||NA|
|Arthritis Foundation (8)||√||NA||√|
|Utilization of a coxib agent|
|American College of Gastroenterology (26)†||NA||NA||NA|
|American College of Rheumatology (9, 10)||√||NA||NA|
|Arthritis Foundation (8)||√||√||X|
We defined individual risk factors for NSAID-related gastropathy based on published meta-analyses, consensus guidelines, and quality indicators. We selected the following risk factors: patient age >65 years, lifetime history of peptic ulcer disease and/or GI bleeding, concomitant warfarin use, concomitant steroid use, and lifetime history of serious cardiovascular comorbidities. Risk factors were assessed from physician-reported diagnoses recorded at study entry and visits prior to the index date. We used the same GI risk factor definitions for both the dual aspirin/NSAID cohort and the aspirin cohorts. Serious cardiovascular comorbidities were defined as physician-reported history of cardiovascular disease including coronary disease, myocardial infarction, or stroke.
We compared the patient characteristics of the 3 cohorts using chi-square and analysis of variance (ANOVA) tests to compare categorical and continuous variables, respectively. Rates of gastroprotection for the 3 cohorts were calculated overall and stratified by GI risk factor. For both the NSAID cohort and the dual NSAID/aspirin cohort, we also determined the distribution of gastroprotective strategies stratified by number of GI risk factors. In addition, we calculated the proportion of NSAID users who were prescribed a coxib agent. Utilization rates between medication groups were compared using chi-square tests, and utilization rates were compared across number of GI risk factors using logistic regression, with risk factor groups as an ordered variable to test for trend. All analyses were performed using Stata version 8.2 software (Stata, College Station, TX).
Our study sample of 2,690 patients consisted of 1,833 (68.1%) NSAID users, 538 (20.0%) aspirin users, and 319 (11.9%) dual NSAID and aspirin users (Table 2). Among demographic characteristics of the 3 cohorts, age and sex were significantly different (P < 0.001), but race was not (P = 0.556). Patients in the aspirin and aspirin/NSAID cohorts were older than those in the NSAID cohort (mean ± SD age 67.0 ± 11.5 years, 65.0 ± 10.8 years, and 57.8 ± 12.6 years, respectively; ANOVA P < 0.001). The mean ± SD modified Health Assessment Questionnaire score was 0.38 ± 0.5 for the NSAID cohort, 0.37 ± 0.5 for the aspirin cohort, and 0.49 ± 0.5 for the dual aspirin/NSAID cohort. The proportion of patients with RA functional class 3 or 4 (30) ranged from 13.4% among NSAID users to 20.8% among dual aspirin/NSAID users. There was no significant difference in the lifetime prevalence of peptic ulcer disease (P = 0.267) or gastroesophageal reflux disease (P = 0.144) among the 3 cohorts. However, the lifetime prevalence of cardiovascular disease varied among the 3 cohorts (P < 0.001), ranging from 5.5% of the NSAID cohort to 27.0% of the aspirin cohort.
|NSAID only (n = 1,833)†||Aspirin only (n = 538)‡||NSAID/aspirin (n = 319)§||P|
|Age ≥65 years||545 (31.6)||287 (58.2)||158 (52.3)||< 0.001|
|Female sex||1,358 (77.6)||326 (64.2)||204 (67.5)||< 0.001|
|White race||1,470 (84.8)||439 (86.8)||258 (85.4)||0.556|
|Current smoker||221 (12.3)||57 (10.9)||44 (14.1)||0.397|
|RA functional class ≥3||242 (13.4)||97 (18.9)||65 (20.8)||< 0.001|
|History of GERD||371 (20.2)||115 (21.4)||80 (25.1)||0.144|
|History of PUD||189 (10.3)||68 (12.6)||38 (11.9)||0.267|
|History of CV disease¶||100 (5.5)||145 (27.0)||82 (25.7)||< 0.001|
|Concomitant steroids||643 (35.1)||222 (41.3)||114 (35.7)||0.031|
|Concomitant warfarin||32 (1.7)||18 (3.3)||9 (2.8)||0.060|
Rates of coxib use as a proportion of overall NSAID prescriptions were stratified by number of GI risk factors (Table 3). Overall, coxibs were prescribed to dual aspirin/NSAID users more frequently than NSAID users not receiving aspirin (72.4% versus 65.8%; P = 0.02). Among dual aspirin/NSAID users, coxibs were prescribed to the majority (72.4%) of patients and did not vary based on number of GI risk factors (P = 0.95 for trend). For patients with 0, 1, and ≥2 GI risk factors, coxib prescribing was consistently high and was prescribed to 66.2%, 78.3%, and 68.5% of patients, respectively. Among NSAID users not prescribed aspirin, coxib utilization did increase with the number of GI risk factors but was not statistically significant (P = 0.13 for trend), and the range was relatively small. Rates of coxib use were 64.1%, 66.4%, and 68.6% for patients with 0, 1, and ≥2 GI risk factors, respectively. Across all levels of GI risk in both cohorts, coxibs were prescribed to the majority of NSAID users. These results were confirmed in a subcohort of patients who were newly prescribed NSAIDs (n = 608).
|All NSAID agents (n = 1,833)||New NSAID agents (n = 608)||NSAID plus aspirin (n = 319)|
|No risk factors†||478/746 (64.1)‡||113/219 (51.6)§||43/65 (66.2)¶|
|1 risk factor†||497/749 (66.4)‡||137/256 (53.5)§||112/143 (78.3)¶|
|≥2 risk factors†||232/338 (68.6)‡||73/133 (54.9)§||76/111 (68.5)¶|
|All patients||1,207/1,833 (65.8)||323/608 (53.1)||231/319 (72.4)|
We compared rates of gastroprotection for the 3 cohorts overall, as well as stratified by number of risk factors (Table 4). Gastroprotection rates were highest in the prescription NSAID cohort. Overall, 1,380 (75.3%) of 1,833 patients receiving prescription NSAIDs received gastroprotection (either coxib and/or GPA cotherapy). Among NSAID users with no identifiable risk factors (n = 746), a total of 537 (72.0%) received gastroprotection. The rate of gastroprotection among NSAID users increased with progressively greater number of defined GI risk factors (P = 0.002 for trend). Among patients with 1 identifiable GI risk factor (n = 749), 572 (76.4%) received gastroprotection. The rate of gastroprotection for patients prescribed NSAIDs with ≥2 GI risk factors increased to 80.2%. We repeated the analyses using a broader definition of GI risk factors including concomitant aspirin use, current smoking status, and general health status defined as ACR functional class ≥3 (30). Rates of gastroprotection were similar using the broader GI risk definition. Among patients with 0, 1, and ≥2 risk factors, the rates of gastroprotection were 71.6%, 73.9%, and 81.3%, respectively (P < 0.001 for trend), using the broader definitions.
|NSAID only (n = 1,833)||Aspirin only (n = 538)||NSAID/aspirin (n = 319)||P|
|No risk factors†||537/746 (72.0)‡||18/97 (18.6)§||28/65 (43.1)¶||< 0.001|
|1 risk factor†||572/749 (76.4)‡||58/221 (26.2)§||49/143 (34.3)¶||< 0.001|
|Age ≥65 years||218/276 (79.0)||21/112 (18.8)||16/70 (22.9)||< 0.001|
|History of PUD||57/66 (86.4)||4/13 (30.8)||4/7 (57.1)||< 0.001|
|History of CV disease||14/19 (73.7)||5/18 (27.8)||10/20 (50.0)||0.020|
|Current steroid use||280/385 (72.7)||27/77 (35.1)||18/43 (41.9)||< 0.001|
|Current warfarin use||3/3 (100.0)||1/1 (100.0)||1/3 (33.3)||0.155|
|≥2 risk factors†||271/338 (80.2)‡||81/220 (36.8)§||51/111 (45.9)¶||< 0.001|
|Overall||1,380/1,833 (75.3)||157/538 (29.2)||128/319 (40.1)||< 0.001|
In contrast to the consistently high rates of gastroprotection among NSAID users, utilization of gastroprotection among dual NSAID/aspirin users was markedly lower. Using either PPI or misoprostol cotherapy as the criteria for gastroprotection, a total of 40.1% of dual NSAID/aspirin users received gastroprotection. Using the same criteria, 29.2% of aspirin users were prescribed a gastroprotective strategy. Despite the absence of supporting evidence for efficacy, we observed that 72.4% of the dual NSAID/aspirin cohort were prescribed a coxib agent, including 66.2% of patients without traditional GI risk factors.
The distribution of specific gastroprotective strategies including coxib utilization among NSAID users not receiving aspirin was stratified by number and type of GI risk factor (Table 5). Overall, there was a 5-fold preference for coxib monotherapy (46.9%) relative to NS-NSAID plus GPA cotherapy (9.4%). Among the 1,833 NSAID users, 860 (46.9%) received coxib monotherapy and 173 (9.4%) received an NS-NSAID plus GPA, representing a 5-fold preference for coxib monotherapy. An additional 347 NSAID users (18.9%) received a dual regimen of coxib plus GPA cotherapy.
|Gastrointestinal risk factors||NS-NSAID without GPA (n = 453)||NS-NSAID with GPA (n = 173)||Coxib agent without GPA (n = 860)||Coxib agent with GPA (n = 347)|
|No risk factors (n = 746)†||209 (28.0)||59 (7.9)||363 (48.7)||115 (15.4)|
|1 risk factor (n = 749)†||177 (23.6)||75 (10.1)||354 (47.3)||143 (19.1)|
|Age ≥65 years (n = 276)||58 (21.0)||21 (7.6)||158 (57.3)||39 (14.1)|
|History of PUD (n = 66)||9 (13.6)||9 (13.6)||25 (37.9)||23 (34.9)|
|History of CV disease (n = 19)||5 (26.3)||1 (5.3)||11 (57.9)||2 (10.5)|
|Current steroid use (n = 385)||105 (27.3)||43 (11.2)||158 (41.0)||79 (20.5)|
|Current warfarin use (n = 3)||0 (0.0)||1 (33.3)||2 (66.7)||0 (0.0)|
|≥2 risk factors (n = 338)†||67 (19.8)||39 (11.5)||143 (42.3)||89 (26.3)|
|All patients (n = 1,833)||453 (24.7)||173 (9.4)||860 (46.9)||347 (18.9)|
A total of 478 (64.1%) of the 746 NSAID users with no identifiable GI risk factors were prescribed a coxib agent. The distribution of gastroprotective strategies for patients without identifiable risk factors included 363 (48.7%) receiving coxib monotherapy, 115 (15.4%) receiving dual coxib/GPA therapy, and 59 (7.9%) receiving GPA cotherapy with an NS-NSAID. Only 209 (28.0%) of the 746 patients without identifiable GI risk factors did not receive gastroprotection. Among patients without identifiable GI risk factors, we further stratified by age group. For patients age 18–44 years without GI risk factors, a gastroprotective strategy was prescribed to 103 (67.3%) of 153 patients. Coxib agents were prescribed to 89 (86.4%) of 103 patients in this low-risk cohort receiving gastroprotection. Across all levels of GI risk, regardless of the number of GI risk factors, coxib monotherapy was the predominant gastroprotective strategy.
To our knowledge, this study represents the largest examination of NSAID gastroprotection and coxib utilization conducted among rheumatologists in the US. During the 6-month study period from March through September 2004, we found consistently high rates of gastroprotection for NSAID users across all levels of GI risk, predominantly due to coxib use. Moreover, despite the absence of a proven safety benefit among aspirin users who receive coxibs, coxibs were prescribed by rheumatologists to the majority of aspirin users requring NSAID therapy. Overall, our results indicate that rheumatologists prescribed coxibs to the majority of their patients in a nonselective manner, including those without identifiable risk factors.
In light of the growing evidence of cardiovascular risk associated with the coxib class since the publication of the VIGOR trial (6), it is important to examine possible explanations for the observed trends of broad, nonselective coxib prescribing among rheumatologists. Certainly, marketing efforts by pharmaceutical companies, including direct-to-consumer campaigns, may have contributed to the high rates of coxib utilization. However, the prescribing decisions for coxib agents ultimately remained in the hands of physicians. Similarly, it is reasonable to assert that the risk/benefit profile of coxibs was not clearly defined prior to the more definitive trial of rofecoxib for polyp prevention with a placebo arm (11). Therefore, although some patients may have represented relatively low risks of GI bleeds, the possibility of averting bothersome GI symptoms and further reducing these patients' risk of GI bleeding may have been attractive to both rheumatologists and their patients. However, this rationale presumes there were no other drug-related toxicities. Although this approach to medicine may seem attractive, the number of patients needed to treat with a coxib agent, and its associated cost effectiveness, suggests that this approach is problematic (31). Moreover, as we have learned, certain drug-related toxicities, particularly rare adverse events such as myocardial infarctions, require years of postmarketing surveillance before they are evident.
The high rates of NSAID gastroprotection utilization observed in our study contrast with earlier population-based studies that reported a minority of NSAID users with GI risk factors who were receiving gastroprotection (14, 16). Smalley et al studied gastroprotection among recurrent NSAID users in the Tennessee Medicaid program shortly after the introduction of celecoxib and rofecoxib to the US market (14). Among NSAID users with ≥2 GI risk factors, the rate of gastroprotection was only 30%. Similarly, a population-based study in the Netherlands reported that only 19% of NSAID users with ≥2 GI risk factors were prescribed gastroprotective agents (16). One possible explanation for the discrepancies between these population-based studies and our findings is that our study was limited to patients treated by rheumatologists. It has previously been shown that specialists, including rheumatologists, are more likely to follow evidence-based guidelines within their specialty (21–23). A more likely explanation, however, is the different time periods. The studies indicating low rates of NSAID gastroprotection and coxib utilization were conducted soon after the introduction of the coxib class. In contrast, a recent longitudinal study corroborated our findings, observing that since market release there has been progressively higher prescribing rates of coxib agents, particularly for patients with low GI risk (18). Therefore, the challenges associated with limiting diffusion of novel therapeutic agents to broader patient populations are likely to be challenges that cross subspecialty boundaries within the US health care system.
Our study also examined the rate of gastroprotection for aspirin users. We found that GI prophylaxis rates for aspirin-induced gastropathy were relatively low. Even among aspirin users with ≥2 GI risk factors, only approximately one-third (36.8%) of patients received gastroprotection. Multiple epidemiologic studies have confirmed the association of low-dose aspirin use with peptic ulcer disease and its complications (32, 33). Given these GI risks, quality improvement efforts to increase utilization of gastroprotective agents among aspirin users are needed. In fact, the Arthritis Foundation's recent consensus recommendations to measure quality of care include recommendations for prescribing a concomitant gastroprotective agent, but not a coxib, for aspirin users at increased risk of gastropathy (8). Given the lack of efficacy of coxibs for gastroprotection among aspirin users in the pivotal clinical trials, our finding that the majority of dual NSAID/aspirin users at increased GI risk were prescribed a coxib agent, but not gastroprotective cotherapy, is concerning. Our study indicates that patients with RA with GI risk factors who are prescribed aspirin (with or without a coxib) are not receiving adequate gastroprotection.
The strengths of our study include the quality of the clinical information and medication data collected in the CORRONA registry. Unlike observational studies using administrative databases, our study prospectively collected clinical diagnostic data that included lifetime histories of relevant GI risk factors. Thus, even remote diagnoses of peptic ulcer disease and related risk factors that may have contributed to the decision to prescribe a coxib were included in our analyses. Moreover, the inclusion of over-the-counter aspirin in the CORRONA registry enabled us to determine rates of gastroprotection for aspirin users and dual aspirin/NSAID users, 2 cohorts that have not previously been studied.
There are a number of limitations as well. First, the fact that CORRONA is an arthritis registry of rheumatologists may limit the generalizability of our findings; however, our study provides an important self-assessment of prescribing patterns for our specialty. In addition, the withdrawal of rofecoxib and valdecoxib from the market, coupled with concern regarding the cardiovascular safety of celecoxib from the Adenoma Prevention with Celecoxib trial, has likely already reduced the nonselective prescribing of coxibs by rheumatologists (13). However, we believe that our study's relevance extends beyond our findings of nonselective prescribing of the coxib class, because novel therapeutic agents for patients with arthritis continue to be developed and introduced to the market.
An additional limitation of our study is that our study population was limited to patients with RA. However, the ACR guidelines for RA management, similar to other organizations' guidelines, do not recommend gastroprotection for all patients with RA (9). Moreover, RA has not been observed to be an independent risk factor for NSAID-related gastropathy (34) and therefore has not been included as a risk factor by other organizations (8, 26). In fact, the rate of gastroprotection for patients with RA (75.4%) in our study was lower than the rate we had previously observed in a smaller sample of patients with psoriatic arthritis (82.2%) and osteoarthritis (84%). Therefore, our findings indicate that rheumatologists prescribed coxib agents broadly to patients with varying arthritis diagnoses (35). A final limitation of our study is that we chose to define risk factors for gastropathy based on consensus recommendations of professional organizations. However, comparable prescribing patterns were observed when we broadened our definition of GI risk factors to include smoking status, poor functional status, and concomitant aspirin use. Nevertheless, other candidate risk factors that were not included may have influenced the decision to prescribe gastroprotection.
The relative GI safety of the coxib class represented a major therapeutic advance for patients at increased GI risk who require long-term NSAID therapy. Our data indicate that coxib prescribing by rheumatologists prior to the coxib market withdrawals was widespread and not limited to a higher risk patient population. As novel therapeutic classes are introduced, early evaluation of prescribing patterns using arthritis registries can determine the appropriateness of prescribing patterns and may improve patient outcomes.
- 25Merck. Merck announces voluntary worldwide withdrawal of Vioxx. URL: http://www.vioxx.com/vioxx/documents/english/hcp_notification_physicians.pdf.
- 26Members of the Ad Hoc Committee on Practice Parameters of the American College of Gastroenterology. A guideline for the treatment and prevention of NSAID-induced ulcers. Am J Gastroenterol 1998; 93: 2037–46., and theDirect Link:
- 35Utilization of gastrointestinal prophylaxis for aspirin and NSAID-related gastropathy [abstract]. Arthritis Rheum 2004; 50 Suppl 9: S316., , , , , .