CURATA: A Patient Health Management Program for the Treatment of Osteoarthritis in Qu"bec: An Integrated Approach to Improving the Appropriate Utilization of Anti-Inflammatory/Analgesic Medications

, , , ,
The American Journal of Managed Care, August 2004, Volume 10, Issue 8

Objectives: To identify gaps in current osteoarthritis (OA) carein Québec, Canada, and to implement and evaluate interventionsto promote appropriate use of evidence-based medicine.

Study Design: Pretest and posttest; analysis of the Québechealth insurance database.

Methods: CURATA is a patient health management programutilizing an evidence-based approach for OA treatment. Evaluationof the current level of care revealed major gaps in physicians'knowledge of (1) risk factors for gastrointestinal (GI) toxicity associatedwith nonsteroidal anti-inflammatory drugs (NSAIDs); (2)NSAID-induced toxicity associated with long-term administrationand contraindications for NSAID use in patients with hypertension,cardiovascular disease, or renal insufficiency; (3) choice of cytoprotection;and (4) use of nonpharmacologic treatments for OA.The CURATA intervention consisted of educational workshops,with and without presentation of a decision tree regarding appropriateuse of pharmacologic and nonpharmacologic OA treatments.Participating physicians were asked to complete an 8-itemquestionnaire before and after the workshop, as well as 3 and 6months later, to test their immediate and remote knowledge oftreatment choices. The prescribing patterns of GPs also were evaluatedthrough analysis of the Québec health insurance database.

Results: The participating physicians were better immediateand remote risk assessors of GI bleeding and made more appropriatetreatment choices (15.2% improvement relative to mean preworkshopscore).

Conclusion: These evidence-based interventions were successfulnot only in improving the physicians' knowledge regarding thediagnosis and management of OA, but also–more importantly–inchanging their behavior to make more appropriate therapy choicesfor their patients.

(Am J Manag Care. 2004;10:569-575)

Osteoarthritis (OA) is the most prevalent jointdisorder; its primary signs and symptomsof joint pain, swelling, and stiffness reflect theloss of joint cartilage integrity.1,2 There is currently nocure for OA. Available treatments include a range ofpharmacologic and nonpharmacologic interventionsdesigned to reduce pain and inflammation, maintain orimprove mobility, and limit functional disability.2

Osteoarthritis has been shown in a variety of US-basedstudies to be a leading contributor to the cost ofmedical appointments, over-the-counter drug use, andrestricted activity.3-5 Drug therapy represents a largeproportion of the direct costs, with analgesics accountingfor approximately 20% of the total. The treatment ofgastrointestinal (GI) adverse reactions related to non-selectivenonsteroidal anti-inflammatory drug (NSAID)use adds significantly to both direct and indirectcosts.6,7 Nonpharmacologic modalities such as physicaland occupational therapy also contribute to these costs.Indirect costs (eg, absenteeism) are responsible for themajority of the economic burden of OA.

As a result of our aging population, the increaseddemand for healthcare for this disease adds to theseconcerns. There is a critical need to develop strategiesto manage the disease more cost effectively through betterdiagnosis, safer and more effective treatment, andenhanced patient education. This paper describes aninnovative approach using patient health managementto address this need.

CURRENT TRENDS IN THE PHARMACOLOGICTREATMENT OF OSTEOARTHRITIS

Despite their side effects, conventional NSAIDs havebeen widely used and are the agents of choice in thetreatment of OA8 for patients who do not respond toacetaminophen. In Canada, the most commonly prescribedNSAIDs include diclofenac sodium, a combinationof diclofenac and misoprostol, ibuprofen, andnaproxen.6 However, current concerns about thegreater risk of GI toxicity associated with prolonged useof NSAIDs,9-16 including potentially fatal perforation,ulcer, and bleed (PUB),11 have led to a reexamination oftheir role in the treatment of the disease. Risk factorsfor GI toxicity include age more than 65 years, a historyof peptic ulcer or upper-GI bleeding, concomitant useof oral corticosteroids and anticoagulants, comorbiditiessuch as cardiac conditions, and renal dysfunction.11-14 For patients with GI risk factors,coprescription of a gastroprotective (GPA) agent hasbeen advised. More recently, with the introduction ofthe cyclooxygenase-2 (COX-2) inhibitors, it has beenrecommended that patients who require chronic analgesiaand who are at risk for GI adverse events withNSAIDs receive first-line alternative therapy with acetaminophen;if they do not respond, they should receivea COX-2 inhibitor.17

The COX-2 inhibitors have recently emerged as analternative class of anti-inflammatory drugs with efficacyin reducing pain and inflammation equivalent to thatof NSAIDs 18-26 and a clearly reduced potential for upper-GI injury relative to NSAIDs.18,22-24,26-28 TraditionalNSAIDs inhibit both the COX-1 and COX-2 forms of thecyclooxygenase enzyme. Prostaglandins involved in theprotection of the GI mucosa are produced by COX-1,whereas those that mediate inflammation and painare induced by COX-2.23 The COX-2 inhibitors rofecoxiband celecoxib (coxibs) selectively inhibit COX-2,with little effect on COX-1.25-29 However, they are notfree of side effects: dyspepsia has been reported withcoxib use, but at a lower frequency than withNSAIDs.30,31

A retrospective cohort study conducted in Québecbetween April and November 2000 reported that coxibswere more frequently prescribed than acetaminophenor NSAIDs to elderly (≥65 years) patients with musculoskeletaldiseases. The study also found that elderlypatients with a history of GI events were more frequentlygiven a coxib or acetaminophen as opposed toNSAIDs. The study also reported that coxibs were 47%less likely to be prescribed with a GPA than wereNSAIDs.32

BACKGROUND

The Healthcare System in Québec

The Régie de l'assurance maladie du Québec (RAMQ)applies and administers the provincial health insuranceplan established by the Health Insurance Act. The publicprescription drug insurance plan, instituted in 1997,guarantees basic coverage for all Québec residents fordrugs purchased in Québec and listed on the provincialformulary (liste de médicaments). Adults with access toa private insurance plan through their workplace (4.2million individuals; 57% of the population) must pay toreceive coverage for themselves and their dependents.All others are covered by the provincial plan, includingthose receiving employment assistance and persons age65 years and over.33-36 Celecoxib became eligible forreimbursement by RAMQ in October 1999, followed inApril 2000 by rofecoxib. In the first half of 2000, nopublished guidelines or consensus existed for the utilizationof these new anti-inflammatory drugs. Clinicalexperience was still developing.

Profile of NSAID Use in Québec

In the planning stage of this study, we conducted aliterature review to profile and identify the care gaps inNSAID and coxib utilization in Québec.

A study conducted in 1997 by the Québec College ofPhysicians using administrative data obtained fromRAMQ revealed inappropriate utilization of NSAIDs.The study found that general practitioners (GPs) oftenignored drug interactions, dosages, and adverse eventsin prescribing NSAIDs and did not prescribe GPAs asadvised by the guidelines.37,38 In a second study,Choquette et al surveyed the NSAID prescription patternsof all GPs in the province of Québec (D.Choquette, MD unpublished data, 1999). The results ofthis self-reported questionnaire, based on 2328 participants(30% response rate), corroborated the finding ofthe Québec College of Physicians and demonstrated alack of understanding of the indications, contraindications,dosage, and duration of NSAID treatment.

A recent study by Rahme et al concluded that the useof NSAIDs significantly increased the risk of GI adverseevents by 2.48-fold, requiring an additional cost of 0.66Canadian dollars per patient-day of treatment withNSAIDs.39

Tamblyn et al40 conducted a blinded, office-basedstudy using simulated patients, in which 2 clinical cases(chronic hip pain due to early OA and NSAID-inducedgastropathy) were used to assess the management decisionsof 112 physicians for elderly patients with arthritis.In 41.7% of office visits, unnecessary prescriptionsfor NSAIDs or other drugs were given. Gastropathyrelated to NSAID use was properly diagnosed in 93.4% ofcases, but acceptably managed in only 77.4% of cases.The risk for suboptimal management of NSAID-relatedadverse events was increased 16.6-fold with an incorrectdiagnosis versus a correct diagnosis.

These studies revealed significant gaps between optimaland actual practice. The results raise questionsabout the appropriateness of NSAID use in the elderlypopulation at large and concerns that current prescribingpatterns are contributing to GI morbidity in the elderly,with the associated avoidable health costs.Selective utilization of coxibs in the elderly is perhaps abetter medical practice. Their acquisition cost is muchhigher than that of NSAIDs. However, the cost of treatmentwith an NSAID plus a proton pump inhibitor inCanada is greater than the cost of a COX-2 inhibitoralone.41 In a public healthcare system such as that ofQuébec, the widespread utilization of coxibs andNSAIDs plus proton pump inhibitors, combined withtheir high acquisition costs, makes them affordable onlyat the expense of other healthcare services. Therefore,programs that promote appropriate utilization of thecoxibs and NSAIDs are needed.

The CURATA Program

CURATA (Concertation pour une UtilisationRaisonnée des Anti-inflammatoires dans le Traitementde l'Arthrose [An Integrated Approach to Improving theAppropriate Utilization of Anti-Inflammatory/AnalgesicMedications in the Treatment of Osteoarthritis inQuébec]) is a patient health management programdeveloped by the Association of Rheumatologists andMerck Frosst Canada in 2000. It was developed toaddress the gaps in OA care and implement effectiveinterventions to promote evidence-based medicationprescription for OA patients, with the goal of optimizingthe management of OA. Its objectives are toimprove the quality of life of OA patientsand to encourage the appropriate utilizationof NSAIDs and coxibs in the treatment of thedisease.

The patient health management approachseeks to optimally manage disease throughimprovement of practice patterns andpatient outcomes. It is a patient-centeredapproach in which best practices are comparedwith current practices, and modificationsare introduced to improve therapeuticresults in the most cost-effective manner. Itsgoal is to bring the greatest health benefit tothe largest number of people at the best cost.To be optimally effective, patient healthmanagement mobilizes a wide range ofhealthcare stakeholders: governments, third-partypayers, university researchers, healthcharities, community specialists, familyphysicians, nurses, pharmacists, special-interestgroups, pharmaceutical companies,and above all, patients. These partnershipswill ensure the broadest application of theprogram and its interventions.

The key to success of this process is continuousmeasurement and feedback of outcomes obtained fromthe broadest application of evidence-based therapies.42,43

A community-based approach to research providesthe greatest opportunity for highly committed stakeholdersto identify the gaps in care and to work aspartners to define best practices to achieve optimal outcomesat the lowest possible cost. Another view mightbe that partnership-directed, evidence-based diseasemanagement represents healthcare improvement fromthe ground up, as opposed to the top down. The functionalprocesses are illustrated in Figure 1.

The Continuing Medical Education Department ofthe University of Montréal, in collaboration with delegatesof the Association of Rheumatologists, developedthe CURATA educational tools. The partners in the programinclude representatives from government, academia,charitable organizations, and the pharmaceuticalindustry (Appendix).

METHODS

The program activities were coordinated by using amodified structure similar to that utilized by theNational Institutes of Health to conduct a large-scalestudy.44

The primary objectives of the CURATA program wereto (1) improve the GP's ability to identify OA patientsthrough appropriate questioning and musculoskeletalexamination and (2) to enhance the GP's ability to selectappropriate pharmacologic and nonpharmacologic therapyaccording to a defined decision tree. The underlyinghypothesis was that an integrated program involvingprofessionals and patients whose goals were to enhancethe appropriate use of NSAIDs and GPAs would facilitatebridging the gap that exists between the current state ofcare and optimal practice.

Phases of the CURATA Program

In the first phase of the program, the current level ofcare was evaluated in an effort to identify potential discrepancieswith optimal evidence-based treatment.Studies identified in this stage of the program aredescribed in the section titled Profile of NSAID Use inQuébec. In summary, there were major gaps in physicians'knowledge of (1) risk factors for GI toxicity associatedwith NSAIDs; (2) NSAID-induced toxicityassociated with long-term administration and contraindicationsfor NSAID use in patients with hypertension,cardiovascular disease, or renal insufficiency; (3)choice of cytoprotection; and (4) use of nonpharmacologictreatments for OA.37-40

In the second phase, interventions were introduced.To provide guidelines for the appropriate utilization ofNSAIDs and coxibs as well as GPAs, the CURATAdevelopment committee created a decision tree withan algorithm for optimizing OA treatment approaches(Figure 2).45 To maximize adoption of this decisiontree, the committee developed a workshop including 3clinical cases on the diagnosis and treatment of OA. Ithas been demonstrated that small-group, interactive,case-based workshops ideally facilitate the disseminationof such material and provide a forum for peer discussionand emulation of the desire to excel.46

For each clinical case, a participant worked individuallyto answer questions, then shared his or her answerswith all others in a small group. The group produced asummary of individual responses on transparency for discussionin a plenary session. In this session, expertsshared their experience with all participants. This formatallowed participants to reflect on their own processes ofdata interpretation in making clinical decisions and toexamine those of experts. The workshop was accreditedby the College of Family Physicians of Canada and provided1.5 hours of MAINPRO-C credit to its participants.

A large group session was held for pharmacists andother health professionals. A patient follow-up sheetwas created for pharmacists, which included questionsregarding patients' use of homeopathic and other alternativemedicine products. A series of 5 brochuresdetailing the various joints affected by OA was developedfor patients. In addition, there was a brief presentationby a physiotherapist and an occupationaltherapist to describe their role in patient managementof OA. A multidisciplinary panel including a physician,pharmacist, physiotherapist, and occupational therapistthen discussed a case study.

Evaluation of the Educational Intervention

To assess the impact of CURATA, 2 measurementswere undertaken: one to evaluate the impact on medicationprescription and the other to determine theimmediate and remote impact on retention of knowledgeregarding treatment selection for OA. AlthoughCURATA was implemented throughout the province ofQuébec, the evaluation was undertaken in 8 cities.

Impact on Medication Prescription

. The interventionwas comprised of the workshop and a decision treein the form of a laminated sheet. There were 4 interventiontypes: workshop and tree, workshop only, decisiontree only, and control group; each type wasassigned to 2 cities.

Records of all patients age 65 years or older who hadfilled a prescription for a coxib, NSAID, or acetaminophenbetween May 2000 and June 2001 (the date whenthe data were requested) from an eligible GP (GP with≥2 NSAID prescriptions) were retrieved from the RAMQdatabase. Only new prescriptions preceded by a diagnosisof OA in the previous 3 years were considered. Ingeneral, prescriptions for a chronic condition are writtenwith the possibility of 3 refills. New prescriptionswere either a new treatment or a continued treatmentfilled for the first time (nonrenewal). Because a prescriptioncould have been written before the workshoptook place and refilled later, refills were not assessed. A0/1 score was given to every filled prescription accordingto the instruction on the decision tree. All acetaminophenprescriptions were given a score of 1, andall patients were presumed to have used acetaminophenbefore NSAIDs or coxibs prescriptions.

Mean scores were calculated for each GP in thepreintervention and postintervention periods. Theanalyses were done both by "intent-to-treat" (includingall GPs who practice in 1 of the 8 cities) and "per protocol"(including only GPs who attended the workshopsin the intervention groups) and were compared with thescores of all GPs in the control group.

Details of the methodology adopted in this study andthe results have been published.45

RESULTS

P

P

In the intent-to-treat analysis, the study demonstratedan adjusted significant improvement on mean scoresof 8% ( = .003) between the control group and theworkshop and workshop and tree groups combined. Nosignificant improvement was reported in the tree groupversus the control group. A per protocol analysisrevealed a 12% ( = .008) improvement in mean scores.

In summary, the intervention had a positive impact onthese physicians' prescription patterns for OA. In view ofthe widespread use of coxibs, nonselective NSAIDs, andacetaminophen, and given that more than 50% of theseprescriptions were for the treatment of OA, the positiveimpact should reflect considerable cost savings.

Immediate and Remote Impact on Retention ofKnowledge.

t

P

The GPs were asked to fill a questionnaireconsisting of 8 multiple-choice questions before (pretest)and after (posttest) the workshop, as well as 3 and 6months later. (For a copy of this questionnaire, pleasecontact the author.) Each question was scored 1 or 0according to whether or not the checked answer agreedwith the workshop content. Overall pretest and posttestscores for each GP were calculated and compared byusing a paired test. Questions were divided into 2 types,those concerning patients with no GI risk factors andthose concerning patients with GI risk factors. Of 52 GPsinvited to the workshop, 26 completed the questionnaire.Nonparametric tests were used to compare the medianpretest and posttest scores for the 2 types of questions.Overall median scores (pretest, posttest, and value)were 2.0, 3.5, and <.0001, respectively, for questionsdescribing patients with no GI risk factors (4 questions)and 3.0, 3.5, and <.01, respectively, for questions describingpatients with GI risk factors (4 questions). At the 6-month evaluation, the knowledge improvement wassustained; scores were similar to those seen on theimmediate posttest. Thus, physicians who attended theworkshop were, in fact, better assessors of GI bleeding,both immediately afterwards and 6 months later; andthey selected a more appropriate therapeutic regimen.

On completion of this workshop, the GPs were ableto adopt a structured approach to the assessment ofjoint pain, identify the principal signs of OA on clinicalexamination, use a decision tree to manage patientswith OA, and select an anti-inflammatory medicationappropriate to the patient's clinical condition.

In light of these encouraging results, 50 small-groupsessions and 2 large-group sessions were held, with atotal of 1000 physicians attending. The decision treewas disseminated to 7500 physicians in June 2001. Weanticipate that the positive effect of CURATA will beeven greater once the intervention for pharmacists andother health professionals is implemented.

CONCLUSION

The results of the initial evaluations have demonstratedthat these evidence-based interventions weresuccessful not only in improving physicians' knowledgeregarding the diagnosis and management of OA, butalso&#150;more importantly&#150;in changing their behavior tomake more appropriate therapy choices for theirpatients. The observed modification of their prescriptionpatterns reflects an improvement in their medical practice,which may lead to better patient outcomes and generategreater cost efficiencies for the health care system.

The immediate impact of this patient health managementapproach was demonstrated in late 2001, whenthe reimbursement policy for coxibs was being reevaluatedby the Qu&#233;bec Ministry of Health and transition toexception-medication status was under consideration.This reevaluation was motivated by concerns aboutincreased consumption and the associated costs, as wellas questions regarding the appropriateness of coxib use.Results of the CURATA evaluation provided sufficientevidence for the government to decide to maintain anopen reimbursement policy for this new class of drugs.

The widespread implementation of the interventionsof the CURATA program will further encourage theappropriate use of coxibs, and through these efforts evidence-based guidelines will be developed. The capacityfor programs such as CURATA to shape the healthcareenvironment is further validated by the currentrequirement of the Ministry of Health to establishpatient health management programs as a measure toensure appropriate utilization of various classes ofdrugs. In addition, a concurrent drug utilization reviewmust be carried out. The Ministry also has required thecollaborative involvement of multiple pharmaceuticalindustry partners in these programs. This collaborationrepresents the next challenge.

The establishment of effective patient health managementprograms relies heavily on the commitment andcooperation of a broad range of partners representingacademia, government, industry, and most importantly,patients.42,43 The general public has become increasinglyaware of health concerns in recent times and is eager tobe involved in such innovative approaches to optimizingthe quality of health and healthcare delivery. Similarly,there is an evolving role for the pharmaceutical industry.Beyond an interest in profits, the pharmaceutical industryshares the commitment of its partners as well as itsexpertise in the development of programs that willenhance patient adherence to optimal therapies, whilecontrolling costs and improving health outcomes. Thepromising results of programs such as CURATA demonstratetheir potential to achieve the ultimate goal ofpatient health management: to provide the best health forthe greatest number of people at the best possible cost.

Acknowledgment

We would like to acknowledge the collaboration of MarilynKrelenbaum, MSc, in the writing and editing of the manuscript.

From Merck Frosst Canada Ltd, Kirkland, Qu&#233;bec, Canada (MB, RC); Institut deRhumatologie de Montr&#233;al, Montr&#233;al, Canada (DC); Division of Clinical Epidemiology,McGill University Health Center, Montr&#233;al General Hospital, Montr&#233;al, Canada (ER); andHopital R&#233;gional de Rimouski, Rimouski, Qu&#233;bec, Canada (LB).

This work has been done as a project of the program CURATA with the financial supportof Merck Frosst Canada Ltd.

Address correspondence to: Michele Beaulieu, MEd, Merck Frosst Canada Ltd, PatientHealth Management, 16711 Trans Canada Highway, Kirkland, Qu&#233;bec H9H 3L1. E-mail:michele_beaulieu@merck.com.

ArthritisRheum.

1. Lawrence EC, Helmick CG, Arnet FR, et al. Estimates of the prevalence ofarthritis and selected musculoskeletal diseases in the United States. 1998;41:778-799.

Arthritis Rheum.

2. Hochberg MC, Altman RD, Brandt KD, Clark BM, et al. Guidelines for themedical management of osteoarthritis, part I: osteoarthritis of the hip. AmericanCollege of Rheumatology. 1995;38:1535-1540.

J Rheumatol.

3. Felts W, Yelin EH. The economic impact of the rheumatic diseases in the UnitedStates. 1989;16:867-884.

RheumatoidArthritis: Pathogenesis, Assessment, Outcome and Treatment.

4. Yelin EH. Work disability and rheumatoid arthritis. In: Wolfe F, ed. New York, NY:Marcel Decker; 1994.

Arthritis Rheum.

5. Yelin EH, Felts W. A summary of the impact of musculoskeletal conditions inthe United States. 1990;33:750-775.

Am J Med.

6. Bloom BS. Direct medical costs of disease and gastrointestinal side effects duringtreatment of arthritis. 1988;88(suppl 2A):20-24.

J Rheumatol

7. Hochberg MC. Association of nonsteroidal anti-inflammatory drugs with uppergastrointestinal disease: epidemiologic and economic considerations. .1992;19(suppl 36):63-67.

Osteoarthritis: Diagnosis and Medical/SurgicalManagement.

8. Batchlor EE, Paulus HE. Principles of drug therapy. In: Moskowitz RW, HowellDS, Goldberg VM, Mankin HJ, eds. Philadelphia, Pa: WB Saunders; 1992.

Arch Intern Med.

9. Roth SH. NSAIDs gastropathy. A new understanding. 1996;156:1623-1628.

Arch Intern Med.

10. Singh G, Ramey DR, Mortfeld HS, et al. Gastrointestinal tract complications ofnonsteroidal anti-inflammatory drugs. 1996;156:1530-1536.

New Engl J Med.

11. Wolfe MM, Lichtenstein DR, Singh G. Gastrointestinal toxicity of nonsteroidalanti-inflammatory drugs. 1999;340:1888-1899.

CMAJ.

12. Tannenbaum H, Davis P, Russell AS, et al. An evidence-based approach toprescribing NSAIDS in musculoskeletal disease: a Canadian consensus. 1996;155:77-88.

Arthritis Rheum.

13. Hochberg MC, Altman RD, Brandt KD, et al. Guidelines for the medicalmanagement of osteoarthritis, part II: osteoarthritis of the knee. American Collegeof Rheumatology. 1995;38:1535-1546.

ArthritisRheum.

14. American College of Rheumatology. Ad Hoc Committee on ClinicalGuidelines. Guidelines for the management of rheumatoid arthritis. 1996;39:713-722.

Gastroenterology.

15. Lanza FL. A guideline for the treatment and prevention of NSAID-inducedulcers. 1998;93:2037-2046.

Arch Intern Med.

16. Hernandez-Diaz S, Garcia Rodriguez LA. Association between nonsteroidalanti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: anoverview of epidemiologic studies published in the 1990's. 2000;160:2093-2099.

Can J Clin Pharmacol.

17. Second Canadian Consensus Guidelines on the Use of NSAIDs inOsteoarthritis and Rheumatoid Arthritis. 2000;7(3 suppl A).

Aliment Pharmacol Ther.

18. Emery P, Zeidler H, Kvien TK, et al. Emergency admissions for upper gastrointestinaldisease and their relation to NSAID use. 1997;11:283-291.

Mayo ClinProc.

19. Bensen WG, Fiechtther JJ, McMillen JI, et al. Treatment of osteoarthritis withcelecoxib, a cyclooxygenase-2 inhibitor: a randomized controlled trial. 1999;74:1095-1105.

J Rheumatol.

20. Ehrich EW. Effect of specific COX-2 inhibition in osteoarthritis of the knee: a6-week double blind placebo-controlled pilot study of rofecoxib. 1999;26:2438-2447.

Arthritis Rheum.

21. Cannon GW, Caldwell JR, Holt P, et al. Rofecoxib, a specific inhibitor ofcyclooxygenase 2, with clinical efficacy comparable with that of diclofenac sodium. 2000;43:978-987.

JAMA.

22. Simon LS, Weaver Al, Graham DY, et al. Anti-inflammatory and upper gastrointestinaleffects of celecoxib in rheumatoid arthritis: a randomized controlledtrial. 1999;282:1921-1928.

Am JGastroenterol.

23. Goldstein JL, Silverstein FE, Agrawal NM, et al. Reduced risk of upper gastrointestinalulcer complications with celecoxib, a novel COX-2 inhibitor. 2000;957:1681-1690.

JAMA.

24. Langman MJ, Jensen DM, Watson DJ, et al. Adverse upper gastrointestinaleffects of rofecoxib compared with NSAIDs. 1999;282:1929-1933.

Rheumatology.

25. Brooks P, Emery P, Evans JF, et al. Interpreting the clinical significance of thedifferential inhibition of cyclooxygenase-1 and cyclooxygenase-2. 1999;388:779-788.

JAMA.

26. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxibvs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoidarthritis: the CLASS study: a randomized controlled trial. Celecoxib Long-termArthritis Safety Study. 2000;284:1247-1255.

N Engl J Med.

27. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinaltoxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGORStudy Group. 2000;343:1520-1528.

Arthritis Rheum.

28. Hawkey CJ, Laine L, Simon LS. Comparison of the effects of rofecoxib (acyclooxygenase 2 inhibitor), ibuprofen and placebo on gastroduodenal mucosa ofpatients with osteoarthritis. 2000;43:370-377.

Am J Med.

29. Emery P. Cyclooxygenase-2: a major therapeutic advance? 2001;110(suppl 1):S42-S45.

Arch Intern Med.

30. Watson DJ, Harper SE, Zhao PL, et al. Gastrointestinal tolerability of theselective cyclooxygenase-2 (COX-2) inhibitor rofecoxib compared with non-selectiveCOX1/COX-2 inhibitors in osteoarthritis. 2000;160:2998-3003.

Am J Physiol.

31. Komhoff M, Grone H-J, Klein T. Localisation of cyclooxygenase-1 and -2 inadult and fetal human kidney: implications of renal function. 1997;272:460-468.

Arthritis Rheum.

32. Rahme E, Marentette M, Kong SX, et al. Use of NSAIDs, COX-2 inhibitorsand acetaminophen and associated co-prescriptions for gastroprotective agents inan elderly population in Canada. 2002;47:595-602.

33. Qu&#233;bec government portal: health network. Available at: http://www.gouv.qc.ca/Vision/Sant&#233;/Sant&#233;-enhtml. Accessed August 6, 2002.

34. R&#233;gie de l'assurance maladie de Qu&#233;bec: The R&#233;gie-Mission, mandates andclientele. Available at: http://www.ramq.gouv.qc.ca/crc/eng/reg/miss.shtml.Accessed July 24, 2002.

35. R&#233;gie de l'assurance maladie de Qu&#233;bec: Prescription Drug InsurancePlan&#150;General conditions of the plan. Available at: http://www.ramq.gouv.qc.ca/crc/eng/cit/assmed/cond.shmtl. Accessed July 24, 2002.

36. R&#233;gie de l'assurance maladie de Qu&#233;bec: Prescription Drug Insurance Plan&#150;Changes to the basic prescription drug insurance plan. Available at: http://www.ramq.gouv.qc.ca/crc/eng/cit/assmed/modi.shmtl. Accessed October 21, 2002.

Le Coll&#232;ge

37. Coll&#232;ge des m&#233;decins du Qu&#233;bec. Crit&#232;res d'utilization prolong&#233;e (plus de 30jours cons&#233;cutifs) des anti-inflammatoires non st&#233;roïdiens, [Bulletin ofColl&#232;ge des m&#233;decins du Qu&#233;bec/Qu&#233;bec College of Physicians]. 1996;36(2):39.

Le Coll&#232;ge

38. Coll&#232;ge des m&#233;decins du Qu&#233;bec. Application qu&#233;b&#233;coise d'un mod&#232;le canadienpour la surveillance et l'am&#233;lioration de la performance des m&#233;decins enexercice, [Bulletin of Coll&#232;ge des m&#233;decins du Qu&#233;bec/Qu&#233;becCollege of Physicians]. 1997;36(2)24-28.

Arthritis Rheum.

39. Rahme E, Joseph L, Watson D, et al. Gastrointestinal healthcare resource useand costs associated with nonsteroidal anti-inflammatory drugs versus acetaminophen:retrospective cohort study of an elderly population. 2000;43:917-924.

Ann Intern Med.

40. Tamblyn R, Berkson L, Dauphinee D, et al. Unnecessary prescribing ofNSAIDs and the management of NSAID-related gastropathy in medical practice. 1997;127:429-438.

Pharmacoeconomics.

41. Marshall JK, Pellissier JM, Attard CL, et al. Incremental cost-effectivenessanalysis comparing rofecoxib with nonselective NSAIDs in osteoarthritis. OntarioMinistry of Health perspective. 2001;19(10):1039-1049.

Am J Manag Care.

42. Montague T, Sidel J, Erhardt B, et al. Patient health management: a promisingparadigm in Canadian healthcare. 1997;3:1175-1182.

Hosp Q.

43. Montague T. Improving women's health quality: the value of closing the caregap. 1998;2:36-39.

44. National Heart, Lung, and Blood Institute, National Institutes of Health.Guidelines for data quality in clinical trials and observational studies. Available at:http://www.nhlbi.nih.gov/funding/policies/dataqual.htm. Accessed April 24, 2001.

Can J Clin Pharmacol.

45. Rahme E, Beaulieu M, Choquette D, et al. Impact of an educational interventionon osteoarthritis treatment. 2002;9(1):21-22.

JAMA.

46. Davis DA, Thomson MA, Oxman AD, et al. Changing physician performance:a systematic review of the effect of continuing medical education strategies. 1995;274:700-705.