Analgesic Prescribing for Musculoskeletal Complaints in the Ambulatory Care Setting After the Introduction and Withdrawal of Cyclooxygenase-2 Inhibitors

Article Outline

• Abstract
• Methods
  • Study Design and Population
  • Data Sources
  • Study Sample
  • Outcomes
  • Data Analysis
• Results
  • Analgesic Prescriptions for Musculoskeletal Complaints
• Discussion
  • Study Limitations
• Conclusions
• Acknowledgments
• References
• Copyright

Abstract 

Wilson RD. Analgesic prescribing for musculoskeletal complaints in the ambulatory care setting after the introduction and withdrawal of cyclooxygenase-2 inhibitors.

Objective

To evaluate the analgesic prescribing patterns for musculoskeletal complaints in a nationally representative sample of ambulatory care visits over a dynamic period of pharmaceutical treatments.

Design

Sequential cross-sectional analysis of complex probabilistic survey data with multivariable logistic regression analysis.

Setting

National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey 1999 to 2005.

Participants

Visits to ambulatory care physicians in the United States from 1999 to 2005 with the reason for visit being a musculoskeletal complaint.

Interventions

Not applicable.

Main Outcome Measures

Prescribing of cyclooxygenase-2 (COX-2) inhibitors, nonselective nonsteroidal anti-inflammatory drugs (nsNSAIDs), opiate analgesics, nonnarcotic analgesics, and adjuvant analgesic medications for visits related to musculoskeletal complaints.

Results

There were 41,804 visits for musculoskeletal complaints, representing 789 million visits from 1999 to 2005 in the weighted analyses. Prescribing for any nonsteroidal anti-inflammatory drugs (NSAIDs) remained relatively stable from 1999 to 2005. It appears that COX-2 inhibitors were substituted for nsNSAIDs initially. After safety concerns arose and select COX-2 inhibitors were withdrawn, there were sharp increases in prescribing of nsNSAIDs and nonnarcotic analgesics. Opiate analgesics and adjuvant analgesics increased in usage over the study period, although apparently not in substitution for NSAIDs.

Conclusions

COX-2 inhibitors may have been prescribed as substitutes for nsNSAIDs initially, but nsNSAID prescriptions returned to prior levels by 2005 as COX-2 inhibitor prescriptions declined. An increase in nonnarcotic analgesic prescribing in 2005 may have been caused by a class effect concern for NSAIDs. Prescribing of opiate analgesics and adjuvant analgesics increased over the study period, although apparently not in substitution for NSAIDs.

Key Words: Analgesics, Health care surveys, Musculoskeletal system, Prescriptions, Rehabilitation

List of Abbreviations: COX-2, cyclooxygenase-2, NAMCS, National Ambulatory Medical Care Survey, NHAMCS, National Hospital Ambulatory Medical Care Survey, NSAID, nonsteroidal anti-inflammatory drug, nsNSAID, nonselective nonsteroidal anti-inflammatory drug

ORAL ANALGESIC OPTIONS for the treatment of musculoskeletal complaints have changed since the risks associated with NSAIDs, and COX-2 inhibitors in particular, became evident. Two COX-2 inhibitor analgesics were released in 1999, and in the following year, 2 large trials¹, ² provided support that they had favorable side effect profiles compared with nsNSAIDs. In 2001, a study raised concern that COX-2 inhibitors raised cardiovascular risk,³ a finding that was confirmed in 2004 when a study was halted as a result of drug safety concerns.⁴ Rofecoxib was voluntarily removed from the market in 2004, followed by the removal of valdecoxib in early 2005. Coincidentally, the paradigm for treating pain has also undergone a change. There is greater emphasis on treating pain and a push by regulatory bodies, such as the Joint Commission (formerly the Commission on Accreditation of Healthcare Organizations), that places pain relief as a priority for health care providers.⁵

Studies have described patterns of analgesic use,⁶, ⁷, ⁸, ⁹ but there have been no studies of the prescribing patterns of multiple analgesic medications by physicians in the United States over the period in which COX-2 inhibitors were introduced and rofecoxib and valdecoxib were removed from the market. It remains unclear how physician prescribing adapted to changes in medication availability and with emerging information about drug safety.

The aim of this study was to evaluate the analgesic prescribing patterns for musculoskeletal complaints in a nationally representative sample of ambulatory care visits over a dynamic period of pharmaceutical treatments.

Back to Article Outline

Methods 

Study Design and Population 

This is a sequential cross-sectional analysis of visits to ambulatory care physicians in the United States from 1999 through 2005. Visits to primary care physicians and specialists were included in the analysis. The data are public-use data files that are released for research and do not include any provider or patient identification information. The institutional review board at MetroHealth Medical Center confirmed that this study was exempt from review.

Data Sources 

Data were obtained from the NAMCS and NHAMCS. The NAMCS includes a sample of visits to nonfederally employed office-based physicians (with the exceptions of anesthesiologists, pathologists, and radiologists), whereas the NHAMCS includes ambulatory care services in noninstitutional and nonfederal hospital emergency and outpatient departments. The surveys are conducted annually. Before survey completion, specially trained interviewers visited the physicians or outpatient departments to provide them with survey materials and to teach the physicians and/or their staff members how to complete the forms. Data were obtained on demographic characteristics of patients, expected source or sources of payment, patient complaints, physician diagnoses, diagnostic/screening services, procedures, medication therapy, disposition, types of health care professionals consulted, causes of injury where applicable, and other information. All medical and drug coding and keying operations were subject to quality-control procedures. Quality control involved a 2-way 10% independent verification procedure. Additionally, all patient record forms with differences between coders or illegible entries were reviewed and adjudicated at the National Center for Health Statistics. The multistage, complex survey designs allow estimation of unbiased national estimates of ambulatory care visits.

Study Sample 

Patient visits were identified to be of a musculoskeletal nature by the Reason for Visit Classification code number, developed by National Center for Health Statistics, of the patient's expressed reason for the medical visit. If a complaint related to the musculoskeletal system was coded in any of 3 reasons for visit, then the patient visit was included in the analysis. Acute and chronic musculoskeletal complaints were included in the study. Visits to the emergency department and pre- and postsurgery follow-up were excluded.

Outcomes 

The outcomes were prescriptions for new or continued analgesic medications during visits of musculoskeletal complaints by year from 1999 to 2005. The database allowed up to 6 medications to be prescribed or continued for each visit. Although 8 medications were coded after 2003, we used only the first 6 to keep the analyses standard. The medication categories are opiate analgesics (short- and long-acting opiates), nonnarcotic analgesics (acetaminophen and tramadol), NSAIDs (COX-2 inhibitors and nsNSAIDs), and adjuvant analgesic medications. Adjuvant analgesics are medications primarily developed for indications other than analgesia that, in certain medical conditions, may confer analgesia. In this study, anticonvulsants or tricyclic antidepressants were considered adjuvant analgesics. Combination medications that included opiates were counted as opiates. Aspirin was excluded from the analysis as a result of the large overlap with prevention of cardiovascular and cerebrovascular disorders. A new or continued prescription in one category does not exclude others from being counted because many of these medications are used concurrently.

Data Analysis 

The unit of analysis for all statistical analyses was the ambulatory care visit. Patient weights, sampling strata, and primary sampling units provided by the surveys were used to provide nationally representative estimates of ambulatory care visits in the United States. All analyses took into account the complex survey designs of the NAMCS and NHAMCS by ProcSurveyFreq and ProcSurveyLogistic by SAS version 9.1.^a

The weighted proportions of patients with a musculoskeletal complaint who received different categories of medications for each year were calculated. A logistic regression model was used to model the probability of receiving each medication to account for differences between the cohorts with the cohort of 1999 serving as the referent group. To test changes in prescribing from year to year, a multivariable logistic regression analysis was conducted in which each year was included as the reference group. The multivariable logistic regression analyses adjusted for patient age, race, sex, geographic region, metropolitan statistical area, physician specialty, payer, and anatomic location of complaint, as well as whether the complaint was considered by the physician to be acute, a routine chronic problem, an exacerbation of a chronic problem, or other routine care as documented in the survey.

Back to Article Outline

Results 

There were 399,787 ambulatory care visits in the NAMCS and NHAMCS outpatient survey during the study period of 1999 to 2005. Of these, 45,303 patients (11.3%) had a musculoskeletal complaint as 1 of 3 reasons for the visit, and 3499 (7.7%) of the visits were excluded because they were for visits for pre- and postsurgery or injury follow-up. The final sample with acute or chronic musculoskeletal complaints comprised 41,804 (10.5%) of the ambulatory care visits. The musculoskeletal complaint was the primary reason for visit in 32,906 (78.7%) of those with musculoskeletal complaints. When we used the survey weights from the NAMCS and NHAMCS, there were an estimated 789 million (95% confidence interval, 729–849 million) visits related to musculoskeletal complaints to US ambulatory care physicians from 1999 to 2005.

Analgesic Prescriptions for Musculoskeletal Complaints 

There was little change in the proportion of ambulatory care visits with a musculoskeletal complaint provided as a reason for visit during the study period (11.5% in 1999 vs 11.4% in 2005). Table 1 provides the weighted proportion of visits for new or continued analgesic medication by year.

Table 1. Weighted Percentage of Visits With Musculoskeletal Complaints Receiving Prescription for Analgesic Medications

NOTE: Unweighted visits, n=41,804. Values are provided as percentages.

As can be seen in table 1, the proportion of visits for a musculoskeletal complaint that was associated with a new or continued prescription for any NSAID increased 15.9% from 1999 to the peak prescribing in 2000, followed by a decrease of 24.4% in prescribing of NSAIDs through 2005. Prescriptions for COX-2 inhibitors increased 107.4% from 1999 to 2001 while prescriptions for nsNSAIDs decreased 62.6%. After 2001, prescriptions for COX-2 inhibitors declined 69.5% and nsNSAIDs were increasingly prescribed through 2005. From 1999 to 2005, there was an increase in prescriptions for opiate analgesics (52.0%), nonnarcotic analgesics (36.9%), and adjuvant analgesics (30.6%).

Differences in prescribing remained after adjusting for age, race, sex, metropolitan statistical area status, geographic region, physician specialty, payer, and anatomic location of the complaint (table 2). Compared with 1999, the odds for a prescription for NSAIDs were not significantly different in the years 2000 through 2005. The odds for prescribing COX-2 inhibitors were significantly higher in the years 2000 to 2004 compared with 1999, and significantly lower in 2005. nsNSAIDs were significantly less likely to be prescribed in the year 2001 through 2004 than in 1999, and 2005 was not significantly different. The odds for prescribing opiate analgesics were significantly more likely from 2002 through 2005 than in 1999, and more likely for adjuvant analgesics in 2001 to 2005 compared with 1999. Nonnarcotic analgesics trended to be more likely in all years compared with 1999 except in 2004, although the differences were not significant. The odds of prescribing any analgesic were higher only in the year 2000.

Table 2. Weighted Odds Ratios (95% Confidence Interval) for Receiving Analgesic Prescription for Ambulatory Care Visits for Musculoskeletal Complaints Compared With 1999

Prescriptions	1999	2000	2001	2002	2003	2004	2005
NSAID⁎	Ref	1.23(1.00–1.50)	0.89(0.68–1.17)	0.88(0.70–1.11)	0.94(0.73–1.21)	0.87(0.67–1.12)	0.79(0.62–1.02)
COX-2 inhibitor	Ref	1.79(1.24–2.58)	2.09(1.38–3.18)	2.03(1.34–3.07)	1.85(1.21–2.83)	1.48(0.97–2.26)	0.56(0.35–0.89)
nsNSAID	Ref	0.98(0.79–1.22)	0.55(0.41–0.74)	0.54(0.41–0.71)	0.67(0.50–0.88)	0.70(0.51–0.94)	0.94(0.73–1.21)
Opiate	Ref	1.22(0.98–1.54)	1.30(0.93–1.83)	1.73(1.21–2.46)	1.64(1.14–2.37)	1.90(1.30–2.77)	1.97(1.44–2.71)
Nonnarcotic	Ref	1.01(0.69–1.47)	1.06(0.73–1.56)	1.06(0.72–1.54)	1.03(0.72–1.49)	0.86(0.57–1.29)	1.45(0.98–2.13)
Adjuvant	Ref	1.10(0.80–1.52)	1.70(1.10–2.62)	1.75(1.05–2.93)	2.04(1.30–3.20)	1.81(1.13–2.93)	1.87(1.22–2.85)
Any analgesic	Ref	1.24(1.01–1.53)	1.05(0.80–1.39)	1.11(0.88–1.41)	1.16(0.89–1.52)	1.08(0.81–1.45)	1.22(0.95–1.57)

NOTE: Unweighted visits n=41,804. Analyses adjusted for patient age, race, sex, metropolitan statistical area status, geographic region, physician specialty, payer, and anatomic location of complaint, as well as whether the problem was acute, chronic, or unrelated to reason for visit.

Abbreviation: Ref, reference group.

Changes in prescribing comparing year to year and adjusting for age, race, sex, metropolitan statistical area status, geographic region, physician specialty, payer, and anatomic location of the complaint showed fewer statistically significant differences (table 3). There was an increase in the odds for prescribing COX-2 inhibitors from 1999 to 2000, and a decrease from 2004 to 2005. The opposite was seen in nsNSAIDs, where there was a significant decrease in the odds for prescribing from 2000 to 2001 followed by an increase from 2004 to 2005. The only other statistically significant changes in the odds for prescribing was an increase in nonnarcotic analgesics from 2004 through 2005 and the odds of prescribing any analgesic in the year 2000 compared with 1999.

Table 3. Weighted Odds Ratios (95% CI) for Receiving Analgesic Prescription for Musculoskeletal Complaints Compared to the Prior Year

Prescriptions	2000 vs 1999	2001 vs 2000	2002 vs 2001	2003 vs 2002	2004 vs 2003	2005 vs 2004
NSAID⁎	1.23(1.00–1.50)	0.82(0.66–1.00)	0.99(0.83–1.19)	1.07(0.88–1.31)	0.92(0.74–1.14)	0.91(0.70–1.19)
COX-2 inhibitor	1.79(1.24–2.58)	1.17(0.84–1.62)	0.97(0.77–1.22)	0.91(0.70–1.18)	0.80(0.63–1.02)	0.37(0.27–0.53)
nsNSAID	0.98(0.79–1.22)	0.56(0.41–0.76)	0.98(0.76–1.27)	1.23(0.97–1.56)	1.05(0.82–1.34)	1.35(1.02–1.79)
Opiate	1.22(0.98–1.54)	1.06(0.75–1.51)	1.33(1.00–1.77)	0.95(0.71–1.27)	1.16(0.86–1.56)	1.04(0.74–1.46)
Nonnarcotic	1.01(0.69–1.47)	1.05(0.71–1.55)	0.99(0.69–1.42)	0.98(0.70–1.36)	0.83(0.60–1.16)	1.68(1.20–2.36)
Adjuvant	1.10(0.80–1.52)	1.53(1.00–2.33)	1.02(0.70–1.62)	1.15(0.79–1.67)	0.88(0.57–1.35)	1.02(0.71–1.46)
Any analgesic	1.24(1.01–1.53)	0.85(0.64–1.13)	1.06(0.87–1.28)	1.04(0.85–1.29)	0.93(0.73–1.20)	1.13(0.87–1.46)

NOTE. Unweighted visits n=41,804. Analyses adjusted for patient age, race, sex, metropolitan statistical area status, geographic region, physician specialty, payer, and anatomic location of complaint, as well as whether the problem was acute, chronic, or unrelated to reason for visit.

Abbreviation: CI, confidence interval.

Almost half of the visits for musculoskeletal complaints (48.5%; 95% confidence interval, 47.0–49.9) were chronic in nature. Stratifying by acute or chronic nature of the musculoskeletal complaint produced similar patterns (data not shown).

Back to Article Outline

Discussion 

After 2 COX-2 analgesics were released in 1999, two large trials, the Celecoxib Long-Term Arthritis Safety Study¹ and the Vioxx Gastrointestinal Outcomes Research Study,² concluded that they had favorable side effect profiles compared with nsNSAIDs. Prescribing of COX-2 inhibitors for musculoskeletal complaints more than doubled in the first 2 years of availability (see table 1). At the same time, nsNSAIDs were being prescribed less frequently, which may indicate that COX-2 inhibitors were substitutes for nsNSAIDs, as can be seen graphically in figure 1. The increase in COX-2 prescriptions from 1999 to 2001 is approximately equal to the decrease in prescriptions for nsNSAIDs from 2000 to 2001. Further support that substitution may have occurred is garnered by the relatively stable proportion of visits for which any NSAID (COX-2 inhibitor or nsNSAID) was prescribed from 1999 to 2001, and the stability in the adjusted analyses over the same period. This is similar to the reduction in nsNSAID prescribing after the introduction in COX-2 inhibitors shown in earlier studies.⁸, ⁹

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  Ambulatory care prescription trends for musculoskeletal complaints in 1999 to 2005. Percentages are weighted to represent national estimates.

During the peak of COX-2 inhibitor prescribing for musculoskeletal conditions in 2001, a meta-analysis provided a cautionary note that COX-2 inhibitors raised cardiovascular risk.³ After 2001, the trends in COX-2 inhibitor and nsNSAID prescribing reversed, although not drastically as can be seen in the year-to-year adjusted analyses (see table 3). From this information, taken into account with the concomitant release of the third COX-2 inhibitor, valdecoxib, one might infer that physicians took note of the warning.

The Adenomatous Polyp Prevention on Vioxx Study⁴ confirmed that cardiovascular risk was associated with COX-2 inhibitors, and in 2004, rofecoxib was removed from the worldwide market. Although there was a 22% decrease in new or continued prescriptions of COX-2 inhibitors for musculoskeletal complaints, as can be seen from table 2, the relative prescriptions were still higher than in 1999, and there was no significant difference in 2004 compared with 2003 (see table 3). In 2005, the year in which valdecoxib was withdrawn from the market, there was a further decrease in COX-2 inhibitor prescriptions for musculoskeletal complaints, while prescriptions for nsNSAIDs increased. The adjusted analysis shows a significant decrease in COX-2 inhibitor prescriptions from 2004 to 2005, and a significant decrease compared with 1999. Prescriptions for nsNSAIDs returned to a level not significantly different than 1999, though significantly higher odds for 2005 than 2004. A study of pharmacy claims has provided evidence that the decrease in COX-2 inhibitor use after the withdrawal of rofecoxib was not strictly the result of fewer prescribing options but also the result of a fear of a class effect that also affected prescribing of remaining COX-2 inhibitors.¹⁰ This study shows a trend for a decrease in the odds of receiving an nsNSAID prescription in 2005 compared with 1999 (see table 2), which may be due to the class effect concern or premature study endpoint. The increase in prescriptions for nsNSAIDs to near-1999 levels suggests that nsNSAIDs were in turn being provided in place of COX-2 inhibitors, which also occurred in Scotland, Germany, Australia, and the Netherlands after the withdrawal of rofecoxib.¹¹, ¹², ¹³, ¹⁴

After 2001, the odds for a prescription for opiate analgesics were significantly higher than in 1999 (see table 2), although there were no significant year-to-year changes (see table 3).

It could be inferred from this study that the events that occurred with COX-2 inhibitors did not alter opiate prescribing habits. Indeed, it is consistent with findings of greater prescribing of opiate analgesics since at least 1980.⁷ The potential exists that substitution did occur but was either small in number or delayed so that changes were unnoticed.

New or continued prescriptions for nonnarcotic analgesics remained relatively stable until 2005. The adjusted analyses show significantly higher odds of receiving a prescription for a nonnarcotic analgesic in 2005 than in 2004 (see table 3), although a nonsignificant trend exists toward an increase compared with 1999 (see table 2). Similarly, a slight increase in acetaminophen prescriptions in 2005 occurred in Australia in 2005 after the withdrawal of rofecoxib.¹³

The odds for a prescription for adjuvant analgesic medications for visits with a musculoskeletal complaint did not change significantly year over year (see table 3), although they were significantly higher in 2001 to 2005 compared with 1999 (see table 2). Although there is no clear change coinciding with changes in NSAID prescribing, as with opiates, adjuvant analgesics might have been prescribed in substitution for COX-2 inhibitors or nsNSAIDs in small numbers, or the change in care was delayed.

As noted previously, there has been effort to improve treatment of pain. It does not appear that physicians in the United States decreased prescribing analgesics for musculoskeletal complaints in response to the withdrawals of rofecoxib and valdecoxib. Conversely, it does not appear that physicians are more likely to prescribe an analgesic to patients with musculoskeletal complaints. With the exception of the year 2000, there were no significant differences in the odds for prescribing an analgesic. In the year 2000, there was an increase compared with 1999, but that appears to be due to an increase in overall NSAID prescribing from the appearance of COX-2 inhibitors on the market (see Table 1, Table 2). In table 3, it can be seen that the odds of being prescribed an analgesic lowered in 2001, and table 2 shows no statistical difference. What is clear is that in 2005, the odds of being prescribed an opiate, nonnarcotic, or adjuvant analgesic for a musculoskeletal complaint were higher than in 1999, unchanged for nsNSAIDs, and lower for COX-2 inhibitors.

Study Limitations 

There are limitations to this study that should be kept in mind. First, sequential cohort analyses from surveys such as the NAMCS and NHAMCS present a challenge for drawing inferences on behavior over time. The unadjusted analyses of differing cohorts of patients and physicians can be difficult for year-to-year comparisons because of the heterogeneity of each cohort. The systematic random sampling design of the NAMCS and NHAMCS is intended to supply objective, reliable information about the provision and use of ambulatory medical care,¹⁵ although some bias may persist. Logistic regression analyses were used in an attempt to control for cohort heterogeneity and reduce biases to improve year-to-year comparisons. Second, the reporting of a small number of new or continued prescribed medications may introduce bias. It is possible that analgesic medications that were prescribed were not listed in the survey because the actual number of prescriptions exceeded the allotted space for reporting. This would be a greater concern for those visits in which the patient has a greater number of comorbid illnesses. Additionally, the frequency of nsNSAIDs and nonnarcotic analgesic prescriptions may not represent the actual usage of these medications because they are available as over-the-counter medications. On the other hand, overestimation is more likely for COX-2 inhibitors, opiate analgesics, or adjuvant analgesic medications because some of the prescriptions that were provided may not have been filled. Third, it is not possible to identify the prescriptions associated with each visit as being prescribed for the musculoskeletal complaint itself. This is a particular limitation for adjuvant analgesics because the prescribing is often as an off-label indication. Finally, the NAMCS and NHAMCS surveys have limited clinical detail for each visit. Physicians consider many patient-specific factors when prescribing medications for which adjustment cannot be made. Differences among dosages, dosing schedules, or duration of treatment could not be taken into account.

A strength of this study is that it includes a large sample of visits to community- and hospital-based ambulatory care providers for musculoskeletal complaints. The data allow a sequential analysis of prescribing practices over a time period in which there were many changes in prescribing options. The survey design allows generalization to all nonemergent ambulatory care visits for musculoskeletal complaints in the United States from 1999 to 2005.

Back to Article Outline

Conclusions 

Analgesic prescribing practices of ambulatory care physicians for musculoskeletal complaints changed from 1999 to 2005. There was an increase in prescribing of COX-2 inhibitors after their release that coincided with a decrease in prescribing for nsNSAIDs. As safety of COX-2 inhibitors became a concern and rofecoxib and valdecoxib were withdrawn from the market, prescriptions for nsNSAIDs returned to nearly previous levels. This may indicate a substitution effect. Prescriptions for nonnarcotic analgesics remained relatively stable with a rapid increase in 2005. Opiate analgesics and adjuvant analgesics increased in use over the study period, although apparently not in substitution for NSAIDs.

Supplier

Back to Article Outline

Acknowledgments 

Special thanks to David C. Aron, MD, and Neal V. Dawson, MD, for advice and suggestions to improve this study, and to Charles L. Thomas, BA, for statistical advice.

Back to Article Outline

References 

1. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: the CLASS Study: a randomized controlled trial. Celecoxib Long-Term Arthritis Safety Study. JAMA. 2000;284:1247–1255
   • View In Article
   • MEDLINE
   • CrossRef
2. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis (VIGOR Study Group). N Engl J Med. 2000;343:1520–1528
   • View In Article
   • MEDLINE
   • CrossRef
3. Mukherjee D, Nissen SE, Topol EJ. Risk of cardiovascular events associated with selective COX-2 inhibitors. JAMA 200;286:954-9.
   • View In Article
4. Bresalier RS, Sandler RS, Quan H, et al. Cardiovascular events associated with rofecoxib in a colorectal adenoma chemoprevention trial. N Engl J Med. 2005;352:1092–1102
   • View In Article
   • CrossRef
5. Phillips DM. JCAHO pain management standards are unveiled. JAMA. 2000;284:4–5
   • View In Article
6. Clark JD. Chronic pain prevalence and analgesic prescribing in a general medical population. J Pain Symptom Manage. 2002;23:131–137
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (65 KB)
   • CrossRef
7. Caudill-Slosberg MA, Schwarts LM, Woloshin S. Office visits and analgesic prescriptions for musculoskeletal pain in the US: 1980 vs. 2000. Pain. 2004;109:514–519
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (148 KB)
   • CrossRef
8. Dai C, Stafford RS, Alexander GC. National trends in cyclooxygenase-2 inhibitor use since market release: nonselective diffusion of a selectively cost-effective innovation. Arch Intern Med. 2005;165:171–177
   • View In Article
   • MEDLINE
   • CrossRef
9. Olsen Y, Daumit GL, Ford DE. Opioid prescriptions by US primary care physicians from 1992 to 2001. J Pain. 2006;7:225–235
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (209 KB)
   • CrossRef
10. Thiebaud P, Patel BV, Nichol MB. Impact of rofecoxib withdrawal on cyclooxygenase-2 utilization among patients with and without cardiovascular risk. Value Health. 2006;9:361–368
    • View In Article
    • Abstract
    • Full-Text PDF (92 KB)
    • CrossRef
11. Williams D, Singh M, Hind C. The effect of the withdrawal of rofecoxib on prescribing patterns of COX-2 inhibitors in Scotland. Br J Clin Pharmacol. 2006;62:366–368
    • View In Article
    • MEDLINE
    • CrossRef
12. Schüssel K, Schulz M. Prescribing of COX-2 inhibitors in Germany after safety warnings and market withdrawals. Pharmazie. 2006;61:878–886
    • View In Article
    • MEDLINE
13. Barozzi N, Tett SE. What happened to the prescribing of other COX-2 inhibitors, paracetamol and non-steroidal anti-inflammatory drugs when rofecoxib was withdrawn in Australia?. Pharmacoepidemiol Drug Saf. 2007;16:1184–1191
    • View In Article
    • CrossRef
14. Sukel MP, van der Linden MW, Chen C, Erkens JA, Herings RM. Large-scale stopping and switching treatment with COX-2 inhibitors after the rofecoxib withdrawal. Pharmacoepidemiol Drug Saf. 2008;17:9–19
    • View In Article
15. National Center for Health Statistics. Ambulatory Health Care Data. http://www.cdc.gov/nchs/about/major/ahcd/ahcd1.htmAccessed September 26, 2008
    • View In Article

• a SAS Institute, 100 SAS Campus Dr, Cary, NC 27513.