BackgroundOsteoarthritis is a chronic conditioninvolving degeneration of cartilage withinthe joints. It is the most common form ofarthritis and is associated with pain,substantial disability, and reduced qualityof life. About 6 percent of U.S. adults aged30 years or older have symptomaticosteoarthritis of the knee, and 3 percenthave symptomatic osteoarthritis of the hip.Osteoarthritis increases with age: theincidence and prevalence increase two- totenfold from age 30 to 65 and continue toincrease after age 65. The total costs forarthritis, including osteoarthritis, may begreater than 2 percent of the grossdomestic product, with more than half ofthese costs related to work loss.

Common oral medications forosteoarthritis include nonsteroidalantiinflammatory drugs (NSAIDs) andacetaminophen. Patients with osteoarthritisalso use over-the-counter supplements notregulated by the U.S. Food and DrugAdministration (FDA) as pharmaceuticals,including glucosamine and chondroitin, aswell as topical agents. Opioid medicationsare also used for selected patients withrefractory, chronic pain but are notrecommended for first-line treatment ofosteoarthritis and therefore not included inthis review. Each class of medication orsupplement is associated with a uniquebalance of risks and benefits. In addition,

efficacy and safety may vary for individualdrugs within a class. Nonpharmacologicinterventions (such as physical therapy,weight reduction, and exercise) also helpimprove pain and functional status inpatients with osteoarthritis.

Comparative Effectiveness and Safety ofAnalgesics for Osteoarthritis

Executive Summary

Effective Health Care

Effective Health Care Program

The Effective Health Care Programwas initiated in 2005 to provide validevidence about the comparativeeffectiveness of different medicalinterventions. The object is to helpconsumers, health care providers, andothers in making informed choicesamong treatment alternatives. Throughits Comparative Effectiveness Reviews,the program supports systematicappraisals of existing scientificevidence regarding treatments for high-priority health conditions. It alsopromotes and generates new scientificevidence by identifying gaps inexisting scientific evidence andsupporting new research. The programputs special emphasis on translatingfindings into a variety of usefulformats for different stakeholders,including consumers.

The full report and this summary areavailable at www.effectivehealthcare.ahrq.gov/reports/final.cfm

Number 4

Effective Health Care

2

A challenge in treating osteoarthritis is deciding whichmedications will provide the greatest symptom reliefwith the fewest serious adverse effects. NSAIDsdecrease pain, inflammation, and fever by blockingcyclo-oxygenase (COX) enzymes. Understanding of thepharmacology of NSAIDs continues to evolve, but it isnow thought that most NSAIDs block three differentCOX isoenzymes, known as COX-1, COX-2, andCOX-3. COX-1 protects the lining of the stomach fromacid. COX-2 is found in joint and muscle, andmediates effects on pain and inflammation. By blockingCOX-2, NSAIDs reduce pain compared to placebo inpatients with arthritis, low back pain, minor injuries,and soft tissue rheumatism. However, NSAIDs that alsoblock the COX-1 enzyme (also called “nonselectiveNSAIDs”) can cause gastrointestinal bleeding. In theUnited States, there are an estimated 16,500 annualdeaths due to NSAID-induced gastrointestinalcomplications, a higher death rate than that for cervicalcancer or malignant melanoma. Theoretically, NSAIDsthat block only the COX-2 enzyme (also called“coxibs,” “COX-2 selective NSAIDs,” or “selectiveNSAIDs”) should be safer with regard togastrointestinal bleeding, but they also appear to beassociated with increased rates of serious cardiovascularand other adverse effects. Less is known about COX-3,which is found in the cerebral cortex and cardiac tissueand appears to be involved in centrally mediated pain.

For this report, we defined the terms “selectiveNSAIDs” or “COX-2 selective NSAIDs” as drugs inthe “coxib” class (celecoxib, rofecoxib, valdecoxib,etoricoxib, lumiracoxib). We defined “partially selectiveNSAIDs” as other drugs shown to have partial in vitroCOX-2 selectivity (etodolac, nabumetone, meloxicam).Aspirin differs from other NSAIDs because itirreversibly inhibits platelet aggregation, and thesalicylic acid derivatives (aspirin and salsalate) wereconsidered a separate subgroup. We defined“nonaspirin, nonselective NSAIDs” or simply“nonselective NSAIDs” as “all other NSAIDs.”

This report summarizes the available evidencecomparing the benefits and harms of analgesics in thetreatment of osteoarthritis.

1 These drugs are currently not approved by the FDA for use in theUnited States (etoricoxib, lumiracoxib, tenoxicam, tiaprofenic acid)or have been withdrawn from the market (rofecoxib and valdecoxib).

Oral agents include:

• Acetaminophen

• Aspirin

• Celecoxib

• Choline magnesium trisalicylate

• Chondroitin

• Diclofenac

• Diflunisal

• Etodolac

• Etoricoxib1

• Fenoprofen

• Flurbiprofen

• Glucosamine

• Ibuprofen

• Indomethacin

• Ketoprofen

• Ketoprofen ER

• Ketorolac

• Lumiracoxib1

• Meclofenamate sodium

• Mefenamic acid

• Meloxicam

• Nabumetone

• Naproxen

• Oxaprozin

• Piroxicam

• Rofecoxib1

• Salsalate

• Sulindac

• Tenoxicam1

• Tiaprofenic acid1

• Tolmetin

• Valdecoxib1

3

Questions addressed in this report are:

1. What are the comparative benefits and harms oftreating osteoarthritis with oral medications orsupplements? How do these benefits and harmschange with dosage and duration of treatment, andwhat is the evidence that alternative dosagestrategies, such as intermittent dosing and drugholidays, affect the benefits and harms of oralmedication use? (Note: The only benefits consideredunder this question are improvements inosteoarthritis symptoms from long-term use.Evidence of harms associated with NSAID useinclude long-term studies of these drugs for treatingosteoarthritis or rheumatoid arthritis and for cancerprevention.)

2. Do the comparative benefits and harms of oraltreatments for osteoarthritis vary for certaindemographic and clinical subgroups of patients?

• Demographic subgroups include age, sex, andrace.

• Coexisting diseases include hypertension,edema, ischemic heart disease, heart failure;peptic ulcer disease; history of previousbleeding due to NSAIDs.

• Concomitant medication use includesanticoagulants.

3. What are the comparative effects of coprescribingof H2-antagonists, misoprostol, or proton pumpinhibitors (PPIs) on the gastrointestinal harmsassociated with NSAID use?

4. What are the comparative benefits and harms oftreating osteoarthritis with oral medications ascompared with topical preparations? Topicalpreparations include: capsaicin, diclofenac,ibuprofen, ketoprofen, and salicylate.

A summary of the findings is shown in Table A.

Conclusions

Oral NSAIDS

Benefits: improvements in osteoarthritis symptoms

• Nonselective NSAID vs. another nonselectiveNSAID

• Many trials found no clear differences betweenvarious nonaspirin, nonselective NSAIDs orpartially selective NSAIDs (meloxicam,

nabumetone, etodolac) in efficacy for painrelief or improvement in function.

• In one short-term trial, salsalate and aspirin didnot differ significantly in efficacy for painrelief or symptom improvement.

• No studies evaluated the comparative efficacyof salsalate or aspirin vs. a nonaspirin NSAID.

• COX-2 selective NSAID vs. nonselective NSAID

• COX-2 selective NSAIDs and nonselectiveNSAIDs did not clearly differ in efficacy forpain relief, based on many good-quality,published trials.

• COX-2 selective NSAID vs. different COX-2selective NSAID

• Celecoxib and rofecoxib did not differsignificantly in efficacy for pain relief atcommonly used and comparable doses, basedon consistent evidence from six good-qualitytrials.

• No studies compared efficacy of COX-2s otherthan celecoxib and rofecoxib.

Harms: gastrointestinal (GI) and cardiovascular(CV)

• Rofecoxib vs. nonselective NSAID

• In the only large, long-term trial (VIGOR),rofecoxib 50 mg daily caused fewer seriousulcer complications than naproxen 1,000 mgdaily in patients with rheumatoid arthritis butalso significantly increased the risk ofmyocardial infarction. The overall rate ofserious adverse events was higher withrofecoxib than with naproxen.

• There were about 16 fewer symptomaticulcers, including 5.2 fewer serious GIcomplications, for every 1,000 patientstreated with rofecoxib vs. naproxen after amedian of 9 months of treatment.

• There were 3.0 additional myocardialinfarctions for every 1,000 patients treatedwith rofecoxib compared to naproxen inVIGOR.

• Rofecoxib was associated with an increasedrisk of myocardial infarction relative to placeboin the most comprehensive systematic reviewof randomized controlled trials (RCTs).

• About 3.5 additional myocardialinfarctions occurred for every 1,000patients treated for 1 year with rofecoxibcompared to placebo in the systematicreview.

• Rofecoxib was withdrawn from the market inSeptember 2004, primarily because of CVrisks.

• Celecoxib vs. nonselective NSAID or placebo

• It is not clear whether celecoxib has fewerpotential harms than nonselective NSAIDswhen used longer than 3-6 months. In the onlylarge, published trial (CLASS), celecoxib at800 mg daily did not decrease predefinedserious ulcer complications overall comparedwith diclofenac and ibuprofen; the risk ofserious GI events was lower than withibuprofen, but not diclofenac, at 6 months inpatients who did not use aspirin; and there wasno reduction in serious GI events at the end offollowup. The overall rate of serious adverseevents with celecoxib was similar to the ratewith ibuprofen and diclofenac.

• In fair-quality meta-analyses of arthritis trials,most of which evaluated short-term use,celecoxib caused fewer ulcer complicationsthan nonselective NSAIDs and did not increasethe risk of myocardial infarction.

• Celecoxib 400 mg twice daily was associatedwith an increased risk of serious CV events(CV death or myocardial infarction) relative toplacebo in a long-term trial of polypprevention.

• Celecoxib was associated with an increasedrisk of myocardial infarction relative to placeboin the most comprehensive systematic reviewof RCTs. Most of the CV events with celecoxibwere reported in two large polyp-preventiontrials evaluating 200 mg or 400 mg twice daily,or 800 mg once daily.

• About 3.5 additional myocardialinfarctions occurred for every 1,000patients treated for 1 year with celecoxibcompared to placebo.

• Valdecoxib vs. nonselective NSAID or placebo

• Valdecoxib was associated with a lower risk ofupper GI complications compared with

diclofenac, ibuprofen, or naproxen in two fair-quality meta-analyses of published andunpublished trials.

• There have been too few events reported inRCTs of patients with chronic conditions toaccurately assess CV risk associated withvaldecoxib.

• Two short-term trials in a high-risk post-coronary-artery-surgery setting found thatvaldecoxib was associated with a two- tothreefold higher risk of CV events comparedwith placebo.

• Valdecoxib was withdrawn from the marketdue to life-threatening skin reactions andincreased CV risk.

• Etoricoxib vs. nonselective NSAID

• Etoricoxib was associated with fewer GIadverse events (perforations, symptomaticulcers, and bleeds) than nonselective NSAIDsin a fair-quality meta-analysis of 10 trials.

• In primarily short-term trials, systematicreviews of RCTs suggest that etoricoxib has asimilar CV safety profile compared to otherNSAIDs, with the possible exception ofnaproxen. Definitive conclusions are notpossible because of small numbers of CVevents.

• Lumiracoxib vs. nonselective NSAID

• Results from one large trial (TARGET) foundfewer adverse GI events with lumiracoxib thanwith naproxen and ibuprofen.

• There was no statistically significant differencein rates of serious CV events betweenlumiracoxib relative to naproxen or ibuprofenin TARGET.

• Too few events have been reported in RCTs toaccurately assess CV risk associated withlumiracoxib.

• Partially selective NSAID vs. nonselectiveNSAID

• Meloxicam: There were no significantdifferences in risks of serious GI events inseveral meta-analyses of up to 28 primarilyshort-term clinical trials, and no difference inCV risk in three observational studies.

4

5

• Nabumetone or etodolac: There wasinsufficient evidence to make reliablejudgments about relative GI safety and noevidence on CV safety.

• Nonselective NSAID vs. nonselective NSAID orany COX-2 selective NSAID

• No clear difference in GI safety was foundamong nonselective NSAIDs at commonlyused doses.

• The CV safety of naproxen was moderatelysuperior to that of any COX-2 selective NSAIDin a large systematic review of RCTs.

• There were 3.3 additional myocardialinfarctions for every 1,000 patients treatedwith any COX-2 inhibitor instead ofnaproxen for 1 year.

• The CV safety of nonselective NSAIDs otherthan naproxen (data primarily on ibuprofen anddiclofenac) was similar to that of COX-2selective NSAIDs in a large systematic review.

• In indirect analyses, naproxen was the onlynonselective NSAID associated with neutralCV risk relative to placebo.

• Aspirin

• Aspirin is associated with a lower risk ofthromboembolic events and a higher risk of GIbleeds compared to placebo or nonuse whengiven in long-term prophylactic doses.

• There is insufficient evidence to assess thebalance of GI and CV safety of higher doseaspirin as used for pain relief compared withnonaspirin NSAIDs.

• Salsalate

• Salsalate was associated with a lower risk ofadverse events than other selective andnonselective NSAIDs using broad compositeendpoints in older, poor-quality observationalstudies. In a more recent observational study,salsalate had a similar rate of complicationscompared with other NSAIDs.

• Almost no data are available on CV safety.

Harms: mortality

• Individual trials were not large enough to detectdifferences in mortality between the includeddrugs.

• One meta-analysis of celecoxib found nodifference between celecoxib and nonselectiveNSAIDs, but there were few events.

• In one fair-quality cohort study, nabumetone wasassociated with a lower risk of all-cause mortalitycompared with diclofenac and naproxen, but thisfinding has not been replicated.

Harms: hypertension, congestive heart failure(CHF), edema, and impaired renal function

• All NSAIDs and COX-2 inhibitors can cause oraggravate these conditions.

• There is good evidence from short-term trials that,on average, nonselective NSAIDs raise meanblood pressure by about 5.0 mm Hg (95-percentconfidence interval [CI] 1.2 to 8.7). However,similar average blood pressure changes may notnecessarily correspond with similar likelihoods ofan event requiring withdrawal, medication change,or other clinical consequences.

• Evidence from good-quality observational studiessuggests that rofecoxib is associated with greaterrisks of hypertension, CHF, and edema thancelecoxib. Indirect evidence from various meta-analyses of either rofecoxib or celecoxib vs.nonselective NSAIDs are consistent with thesefindings. Direct randomized trial evidence,however, is limited in quantity and difficult tointerpret because of possible non-equivalentdosing of drugs. Evidence regarding thecomparative risk of renal dysfunction for celecoxiband rofecoxib is sparse.

• There was weak evidence that aspirin and sulindachave less hypertensive effect than othernonselective NSAIDs.

• There were no clear differences among otherselective or nonselective NSAIDs for these adverseevents.

Harms: hepatotoxicity

• Clinically significant hepatotoxicity was rare.

• Among currently marketed NSAIDs, onlydiclofenac was associated with a significantlyhigher rate of liver-related discontinuationscompared with placebo (1 additional case forevery 53 patients treated with diclofenac).

Tolerability

• Relative to nonselective NSAIDs, COX-2 selectiveand partially selective NSAIDs were better or

6

similarly tolerated and aspirin was less welltolerated.

• There were no clear differences in tolerabilityamong COX-2 selective or nonselective NSAIDs.

• Uncertainty remains regarding the comparativetolerability of salsalate and nonselective NSAIDs.Available evidence is somewhat sparse and mixed,with two of three short-term trials suggestingsalsalate is less well tolerated than nonselectiveNSAIDs and older, flawed observational studiessuggesting that salsalate is less toxic thannonselective NSAIDs.

Other oral agents: benefits and harms

• Acetaminophen

• Acetaminophen was modestly inferior toNSAIDs for pain and function in foursystematic reviews.

• Pain severity ratings averaged less than 10points higher for acetaminophencompared to NSAIDs on 100-point visualanalog scales.

• Compared with NSAIDs, acetaminophen hadfewer GI side effects (clinical trials data) andserious GI complications (observationalstudies).

• Acetaminophen may be associated with modestincreases in blood pressure and renaldysfunction (observational studies).

• One good-quality, prospective observationalstudy found an increased risk of CV eventswith heavy use of acetaminophen that wassimilar to the risk associated with heavy use ofNSAIDs.

• Acetaminophen at therapeutic doses does notappear to be associated with an increased riskof hepatotoxicity compared to nonuse inpatients without underlying liver disease.

• Glucosamine and chondroitin

• In one large, good-quality trial the combinationof pharmaceutical-grade glucosaminehydrochloride plus chondroitin (not currentlyavailable in the United States) was not superiorto placebo among all patients studied. Neitherglucosamine nor chondroitin alone wassuperior to placebo. In an analysis of a smallsubgroup of patients with at least moderate

baseline pain, there was a modest benefit forpain relief, but this did not appear to be apreplanned analysis.

• Systematic reviews of older trials foundglucosamine modestly superior to oral NSAIDsand placebo in most trials, but there was someinconsistency between trials, most trials hadsome flaws and results may not be directlyapplicable to the United States because thepositive trials primarily evaluatedpharmaceutical-grade glucosamine available inEurope.

• Only 2 of 20 placebo-controlled trials assessedeffects of glucosamine on radiologic diseaseprogression. One fair- and one good-qualitytrial found pharmaceutical-grade glucosaminesuperior to placebo for progression of kneejoint space narrowing over 3 years.

• Glucosamine and chondroitin were generallywell tolerated and no serious adverse eventswere reported in clinical trials.

Effect of dosage and duration of treatment onthe benefits and harms of oral medication use

• We found no studies evaluating the GI or CVsafety of alternative dosing strategies (such asalternate day dosing, once daily versus twice dailydosing, or periodic drug holidays).

• The risk of GI bleeding increases with higherdoses of nonselective NSAIDs.

• The most comprehensive systematic review ofRCTs found no clear association between durationof exposure and CV risk of COX-2 inhibitors.However, estimates of CV risk with shorterduration of exposure are imprecise due to lownumbers of events.

• The most comprehensive systematic review ofRCTs found higher doses of celecoxib associatedwith increased CV risk, but could not determinethe effects of dose on CV risk associated withrofecoxib due to low numbers of events at lowerdoses. Most trials of nonselective NSAIDsinvolved high doses.

Differences in demographic and clinicalsubgroups

• GI and CV complication rates are higher amongolder patients and those with predisposingcomorbid conditions, but there is no evidence that

7

the relative safety of different NSAIDs variesaccording to baseline risk.

• Compared to nonuse of NSAIDs, oneadditional death per 1 year of use occurred forevery 13 patients treated with rofecoxib, 14with celecoxib, 45 with ibuprofen, and 24 withdiclofenac in one large, population-basedobservational study of high-risk patients withacute myocardial infarction.

• There is no evidence that the comparative safety orefficacy of specific selective or nonselectiveNSAIDs varies depending on age, gender, or racialgroup, although data are sparse.

• Among patients who had a recent episode of upperGI bleeding, there is good evidence that rates ofrecurrent ulcer bleeding are high (around 5 percentafter 6 months) in patients prescribed celecoxib ora nonselective NSAID plus a PPI.

Concomitant anticoagulant use

• Concomitant use of anticoagulants (e.g., warfarin)and any nonselective NSAID increases the risk ofGI bleeding three- to sixfold compared toanticoagulants alone.

• Reliable conclusions about the safety of selectiveNSAIDs used with anticoagulants are not possibledue to flaws in existing observational studies,although there are case reports of serious bleedingevents, primarily in the elderly.

Concomitant aspirin use

• In the CLASS studies, there was no difference inrates of ulcer complications between celecoxib andnonselective NSAIDs in the subgroup of patientswho took aspirin.

• Concomitant low-dose aspirin use increased therate of endoscopic ulcers by about 6 percent inboth patients on celecoxib and those onnonselective NSAIDs in one meta-analysis.

• Rofecoxib plus low-dose aspirin or ibuprofenalone were associated with similar risks ofendoscopic ulcers (16-17 percent), which weresignificantly higher than those for placebo (6percent) or aspirin alone (7 percent).

• The most comprehensive systematic review ofRCTs found that compared to nonuse of aspirin,concomitant aspirin use did not ameliorate theincreased risk of vascular events associated withCOX-2 selective NSAIDs.

Effects of coprescribing H2-antagonists,misoprostol, or PPIs

• Consistent evidence from good-quality systematicreviews and numerous clinical trials foundcoprescribing of PPIs to be associated with thelowest rates of endoscopically detected duodenalulcers relative to gastroprotective agents.

• Coprescribing of misoprostol is associated withsimilar rates of endoscopically detected gastriculcers as coprescribing of PPIs.

• While misoprostol offers the advantage of beingthe only gastroprotective agent to reduce rates ofperforation, obstruction, or bleeding, there is ahigh rate of withdrawals due to adverse GIsymptoms.

• The risk of endoscopic duodenal ulcers forstandard-dose H2 blockers was lower thanplacebo, similar to misoprostol, and higher thanomeprazole. Standard dosages of H2 blockerswere associated with no reduction of risk forgastric ulcers relative to placebo.

• Double (full) dose H2 blockers were associatedwith a lower risk of endoscopic gastric andduodenal ulcers relative to placebo. It is unknownhow full-dose H2 blockers compare to otherantiulcer medications because head-to-head trialsare lacking.

Comparison of oral medications with topicalpreparations

• Topical NSAIDs: efficacy

• Studies of topical NSAIDs typically evaluatedproprietary formulations not approved by theFDA.

• Topical NSAIDs were similar to oral NSAIDsfor pain relief in trials primarily of patientswith osteoarthritis of the knee, with topicaldiclofenac (often with dimethyl sulphoxide[DMSO], a drug not approved for use inhumans in the United States) best studied.

• Topical ibuprofen was superior to placebo inseveral trials.

• Topical NSAIDs: safety

• Consistent evidence from good-quality trials,systematic reviews, and observational studiesfound topical NSAIDs to be associated withincreased local adverse events compared withoral NSAIDs.

8

• Total adverse events and withdrawal due toadverse events were similar.

• Data from one good-quality trial found topicalNSAIDs superior to oral NSAIDs for GIevents, including severe events, and changes inhemoglobin.

• Topical salicylates and capsaicin

• Topical salicylates were no better than placeboin higher quality placebo-controlled trials.

• Compared to placebo, one additional patientachieved pain relief for every eight that usedtopical capsaicin in a good-quality meta-analysis, but capsaicin was associated withincreased local adverse events and withdrawalsdue to adverse events.

Balance of evidence and harms

Each of the analgesics evaluated in this report wasassociated with a unique set of benefits and risks. Eachwas also associated with gaps in the evidence necessaryto determine the true balance of benefits vs. harms. Therole of selective and nonselective oral NSAIDs andalternative agents will continue to evolve as additionalinformation emerges. At this time, although theamount and quality of evidence vary, no currentlyavailable analgesic reviewed in this report wasidentified as offering a clear overall advantagecompared with the others. This is not surprising, giventhe complex tradeoffs between the many benefits (painrelief, improved function, improved tolerability, andothers) and harms (CV, renal, GI, and others) involved.

Individuals are likely to differ in how they prioritize theimportance of the various benefits and harms oftreatment. Adequate pain relief at the expense of anincrease in CV risk, for example, could be anacceptable tradeoff for some patients. Others mayconsider even a marginal increase in CV riskunacceptable. Factors that should be considered whenweighing the potential effects of an analgesic includeage (older age being associated with increased risks forbleeding and CV events), comorbid conditions, andconcomitant medication use (such as aspirin andanticoagulation medications). As in other medicaldecisions, choosing the optimal analgesic for anindividual with osteoarthritis should always involvecareful consideration and thorough discussion of therelevant tradeoffs.

Remaining Issues• The CV safety of nonselective NSAIDs has not

been well studied in large, long-term clinical trials.Naproxen, in particular, may be associated withfewer CV risks than other NSAIDs and should beinvestigated in long-term, appropriately poweredtrials.

• Large observational studies assessing the safety ofNSAIDs have been helpful for assessingcomparative benefits and harms but have generallyhad a narrow focus on single adverse events.Observational studies that take a broader view ofall serious adverse events would be substantiallymore helpful for assessing the overall tradeoffsbetween benefits and harms.

• The CV risks and GI benefits associated withdifferent COX-2 selective NSAIDs may vary.Large, long-term trials with active and placebo-controlled arms would be needed to assess thesafety and benefits of any new COX-2 selectiveanalgesic.

• Meta-analyses of the risks associated withselective COX-2 inhibitors need to continue toassess the effects of dose and duration as moredata become available; current estimates of risks atlower doses and with shorter duration of exposureare less precise than estimates at higher doses andlonger duration of exposure because of smallnumbers of events.

• Large, long-term trials of the GI and CV safetyassociated with full-dose aspirin, salsalate, oracetaminophen compared with nonaspirin NSAIDsor placebo are lacking. Recent observational datasuggesting an increased CV risk with heavy use ofacetaminophen highlight the need for long-term,appropriately powered clinical trials.

• Given the large number of patients who meetcriteria for aspirin prophylaxis for CV events,more trials evaluating the dose-related effects ofaspirin 50-1500 mg on GI benefits and CV safetyare needed.

• The effects of alternative dosing strategies such asintermittent dosing or drug holidays have not beenassessed. Studies evaluating the benefits and risksassociated with such strategies compared withconventional dosing could help clarify the effectsof these alternative dosing strategies. In addition,although there is speculation that once daily versus

9

twice daily dosing of certain COX-2 inhibitorscould reduce CV risk, this hypothesis has not yetbeen tested in a clinical trial.

• Most trials showing therapeutic benefits fromglucosamine were conducted usingpharmaceutical-grade glucosamine not available inthe United States and may not be applicable tocurrently available over-the-counter preparations.Large trials comparing currently available over-the-counter preparations of glucosamine andchondroitin with oral NSAIDs are needed, as theseare likely to remain available even if the FDAapproves pharmaceutical-grade formulations.

• No topical NSAIDs are FDA approved in theUnited States, yet compounding of NSAIDs iswidely available. Although recent trials of topicalNSAIDs are promising, most have been conductedusing a proprietary formulation of diclofenac withDMSO, which is not approved in the United Statesfor use in humans. Cohort studies using largeobservational databases may be required toadequately assess CV risk.

AddendumAs this report was going to press, two relevant meta-analyses on risks associated with NSAIDs werepublished. We were unable to fully incorporate thesestudies into this report, but found their results generallyconsistent with our conclusions:

• A fair-quality meta-analysis of arrhythmia andrenal event (peripheral edema, hypertension, orrenal dysfunction) risk from 114 randomized trialsof COX-2 selective NSAIDs found rofecoxibassociated with increased risks of arrhythmia(primarily ventricular fibrillation, cardiac arrest, orsudden cardiac death) and renal dysfunction(peripheral edema, hypertension, or renaldysfunction) relative to control treatments(placebo, other NSAIDs, or mixed/other) . Theincreased risk was equivalent to approximately 1.1additional arrhythmia events per 1,000 patientstreated with rofecoxib. Celecoxib was associatedwith lower risks of renal dysfunction andhypertension than control treatments, although

there was no difference for the prepecified,primary composite renal outcome of peripheraledema, hypertension, renal dysfunction, orarrhythmia. There was no clear associationbetween other COX-2 inhibitors(valdecoxib/parecoxib, etoricoxib, or lumiracoxib)and either arrhythmia or renal events (noarrhythmia events reported with lumiracoxib).

• A good-quality meta-analysis of cardiovascularrisk (primarily myocardial infarction) from 23observational studies was largely consistent withour qualitative assessment of the observationalliterature. It found rofecoxib associated with adose-dependent, increased risk of cardiovascularevents that was detectable during the first monthof treatment. Of the other NSAIDs, diclofenacwas associated with the highest risk, followed byindomethacin and meloxicam. Celecoxib,naproxen, piroxicam, and ibuprofen were notassociated with increased risks. Assessments ofincreased risk were modest (relative risks all

10

Table

A.

Sum

mary

of

Find

ings

on C

ompara

tive

Effe

ctiv

enes

s and

Safe

ty o

f A

nalg

esic

s fo

r O

steo

art

hriti

s, w

ith S

tren

gth

of

Evid

ence

Spec

ial c

onsi

der

atio

ns in

Trea

tmen

tBen

efits

: sy

mpto

m r

elie

fH

arm

s: g

ast

roin

test

inal,

card

iova

scul

ar, a

nd o

ther

subgro

ups

CO

X-2

sel

ectiv

e•

Goo

d ev

iden

ce C

OX

-2•

GI:

Fair

to g

ood

evid

ence

of

few

er s

erio

us G

I ev

ents

with

•G

ood

evid

ence

that

N

SAID

sse

lect

ive

NSA

IDs

are

CO

X-2

sel

ectiv

e N

SAID

s co

mpa

red

to n

onse

lect

ive

NSA

IDs,

risk

of

GI

blee

ding

com

para

ble

in e

ffic

acy

at le

ast i

n th

e fi

rst 6

mon

ths

of tr

eatm

ent.

and

CV

eve

nts

incr

ease

s (p

ain

relie

f) to

non

sele

ctiv

e •

CV

:C

ompa

rativ

e da

ta o

n C

V r

isks

of

CO

X-2

sel

ectiv

e vs

.w

ith a

ge.

NSA

IDs.

no

nsel

ectiv

e an

d pa

rtia

lly s

elec

tive

NSA

IDs

are

spar

se,

•G

ood

evid

ence

that

•G

ood

evid

ence

CO

X-2

sel

ectiv

ew

ith a

few

exc

eptio

ns (

see

belo

w).

Fai

r ev

iden

ce th

at C

OX

-2ri

sk o

f G

I bl

eedi

ng is

NSA

IDs

are

com

para

ble

inse

lect

ive

NSA

IDs

are

asso

ciat

ed w

ith in

crea

sed

risk

s of

ser

ious

grea

ter

in p

atie

nts

with

effi

cacy

to e

ach

othe

r.C

V e

vent

s (p

rim

arily

myo

card

ial i

nfar

ctio

n) c

ompa

red

to p

lace

bo.

prio

r bl

eedi

ng e

piso

des.

CV

ris

ks m

ay in

crea

se w

ith g

reat

er d

osag

es a

nd d

urat

ions

of

•Fa

ir e

vide

nce

that

tr

eatm

ent,

but e

stim

ates

of

risk

s at

low

er d

oses

and

with

sho

rter

risk

s of

CV

and

ren

aldu

ratio

ns o

f tr

eatm

ent a

re im

prec

ise

due

to s

mal

l num

bers

of

even

ts.

even

ts a

re h

ighe

r in

• R

ofec

oxib

was

with

draw

n fr

om th

e m

arke

t in

Sept

embe

r 20

04,

patie

nts

with

car

diac

and

prim

arily

bec

ause

of

CV

ris

ks.

rena

l com

orbi

ditie

s.•

Cau

tions

abo

ut C

V r

isk

appl

y pr

imar

ily to

rof

ecox

ib a

nd c

elec

oxib

,as

CV

saf

ety

data

are

less

pre

cise

(du

e to

sm

all n

umbe

rs o

f ev

ents

)fo

r va

ldec

oxib

, eto

rico

xib,

and

lum

irac

oxib

. •

Oth

er•

Val

deco

xib

was

with

draw

n fr

om th

e m

arke

t due

to li

fe-t

hrea

teni

ngsk

in r

eact

ions

and

incr

ease

d C

V r

isk.

• F

air

evid

ence

sug

gest

s th

at r

ofec

oxib

is a

ssoc

iate

d w

ith g

reat

er r

isk

of h

yper

tens

ion,

CH

F, e

dem

a, a

nd c

ardi

oren

al e

vent

s th

an c

elec

oxib

.

NSA

IDs

: •

Goo

d ev

iden

ce n

onse

lect

ive

•G

I:G

ood

evid

ence

that

all

nons

elec

tive

NSA

IDs

are

asso

ciat

ed w

ith•

Goo

d ev

iden

ce th

at r

isk

ofno

nsel

ectiv

e an

d pa

rtia

lly s

elec

tive

NSA

IDs

com

para

ble,

dos

e-de

pend

ent i

ncre

ases

in r

isk

of s

erio

us G

I ev

ents

GI

blee

ding

and

CV

(inc

ludi

ng

are

com

para

ble

in e

ffic

acy

com

pare

d to

non

use.

Goo

d ev

iden

ce th

at c

opre

scri

ptio

n of

even

ts in

crea

ses

with

age

.na

prox

en),

to

eac

h ot

her.

mis

opro

stol

or

PPIs

can

atte

nuat

e th

is r

isk,

but

mis

opro

stol

is•

Goo

d ev

iden

ce th

at r

isk

ofpa

rtia

lly s

elec

tive

less

wel

l tol

erat

ed.

GI

blee

ding

is g

reat

er in

• N

o cl

ear

evid

ence

(fa

ir f

or m

elox

icam

and

poo

r fo

r et

odol

ac

patie

nts

with

pri

or

and

nabu

met

one)

that

par

tially

sel

ectiv

e N

SAID

s ar

e bl

eedi

ng e

piso

des.

asso

ciat

ed w

ith d

ecre

ased

ris

k re

lativ

e to

non

sele

ctiv

e N

SAID

s.•

CV

:D

ata

on C

V r

isks

of

nons

elec

tive

and

part

ially

sel

ectiv

e•

Fair

evi

denc

e th

at r

isks

of

NSA

IDs

are

spar

se, w

ith a

few

exc

eptio

ns:

CV

and

ren

al e

vent

s ar

e•

Fai

r ev

iden

ce th

at h

igh

dose

s of

ibup

rofe

n an

d di

clof

enac

hi

gher

in p

atie

nts

with

ca

rry

sim

ilar

risk

s of

ser

ious

CV

eve

nts

com

pare

d to

CO

X-2

card

iac

and

rena

lse

lect

ive

NSA

IDs.

com

orbi

ditie

s.

Spec

ial c

onsi

der

atio

ns in

Trea

tmen

tBen

efits

: sy

mpto

m r

elie

fH

arm

s: g

ast

roin

test

inal,

card

iova

scul

ar, a

nd o

ther

subgro

ups

• F

air

evid

ence

that

nap

roxe

n is

ass

ocia

ted

with

a lo

wer

ris

k•

Fair

evi

denc

e th

at u

sing

of C

V e

vent

s th

an C

OX

-2 s

elec

tive

NSA

IDs

and

no e

xces

sN

SAID

s co

ncom

itant

ly

risk

com

pare

d to

pla

cebo

.w

ith a

ntic

oagu

lant

s •

Oth

er:

Fair

evi

denc

e th

at d

iclo

fena

c is

ass

ocia

ted

with

hig

her

incr

ease

s G

I bl

eedi

ng r

isk

rate

s of

am

inot

rans

fera

se e

leva

tions

than

oth

er N

SAID

s.th

ree-

to s

ixfo

ld.

Asp

irin

/•

No

evid

ence

com

pari

ng•

Goo

d ev

iden

ce th

at a

spir

in 5

0-15

00 m

g (f

or th

rom

botic

eve

nt•

Goo

d ev

iden

ce th

atsa

lsal

ate

effi

cacy

of

aspi

rin

orpr

ophy

laxi

s) is

ass

ocia

ted

with

gre

ater

ris

ks o

f se

riou

s G

I co

ncom

itant

use

of

aspi

rin

sals

alat

e to

CO

X-2

s or

.ev

ents

com

pare

d to

pla

cebo

or

whe

n ad

ded

to w

arfa

rin.

atte

nuat

es o

r el

imin

ates

the

NSA

IDs

•G

ood

evid

ence

that

low

-dos

e as

piri

n is

eff

ectiv

e fo

r pr

even

ting

GI

bene

fits

of

CO

X-2

CV

eve

nts.

sele

ctiv

e N

SAID

s.•

Insu

ffic

ient

evi

denc

e to

ass

ess

GI

and

CV

ris

ks a

ssoc

iate

d•

Fair

evi

denc

e th

at

with

hig

her

dose

s of

asp

irin

for

pai

n co

ntro

l or

with

sal

sala

te.

conc

omita

nt u

se o

f lo

w-

dose

asp

irin

doe

s no

tel

imin

ate

CV

ris

ks w

hen

adde

d to

NSA

IDs

.

Ace

tam

inop

hen

•G

ood

evid

ence

that

•

Goo

d ev

iden

ce o

f lo

wer

ris

k of

GI

com

plic

atio

ns w

ithN

one

acet

amin

ophe

n is

mod

estly

ac

etam

inop

hen

com

pare

d to

NSA

IDs.

infe

rior

in e

ffic

acy

com

pare

d •

Fair

evi

denc

e of

incr

ease

d ri

sk o

f bl

ood

pres

sure

and

ren

alto

NSA

IDs.

dysf

unct

ion

with

ace

tam

inop

hen

com

pare

d to

non

use.

•Po

or e

vide

nce

(a s

ingl

e ob

serv

atio

nal s

tudy

) th

at h

eavy

use

of

acet

amin

ophe

n ca

rrie

s a

sim

ilar

CV

ris

k co

mpa

red

to h

eavy

use

of

NSA

IDs.

Glu

cosa

min

e •

Fair

evi

denc

e (s

ome

•

Goo

d ev

iden

ce th

at g

luco

sam

ine

and

chon

droi

tin a

re w

ell

Non

e(p

harm

aceu

tical

in

cons

iste

ncy

betw

een

tole

rate

d an

d do

not

app

ear

to b

e as

soci

ated

with

ser

ious

grad

e)/

clin

ical

tria

ls)

clin

ical

ad

vers

e ev

ents

.th

at p

harm

aceu

tical

-gra

de

gluc

osam

ine

and

chon

droi

tin a

re n

ot m

ore

effe

ctiv

e th

an p

lace

bo in

un

sele

cted

pat

ient

s, in

clud

ing

one

rece

nt, l

arge

, goo

d-qu

ality

tria

l fin

ding

no

bene

fici

al e

ffec

ts f

rom

gl

ucos

amin

e or

cho

ndro

itin

alon

e or

in c

ombi

natio

n.

11

12

Spec

ial c

onsi

der

atio

ns in

Trea

tmen

tBen

efits

: sy

mpto

m r

elie

fH

arm

s: g

ast

roin

test

inal,

card

iova

scul

ar, a

nd o

ther

subgro

ups

In a

n an

alys

is o

f a

smal

l su

bgro

up o

f pa

tient

s w

ith a

t le

ast m

oder

atel

y se

vere

ba

selin

e pa

in in

the

latte

r tr

ial,

ther

e ap

pear

ed to

be

a m

odes

t ben

efit

for

pain

re

lief

from

the

com

bina

tion,

bu

t thi

s di

d no

t ape

ar to

be

a pr

epla

nned

ana

lysi

s.•

Fair

evi

denc

e of

no

clea

r di

ffer

ence

in e

ffic

acy

betw

een

phar

mac

eutic

al-

grad

e gl

ucos

amin

e or

ch

ondr

oitin

and

NSA

IDs.

•N

o st

udie

s co

mpa

red

gluc

osam

ine

or c

hond

roiti

n to

ace

tam

inop

hen.

Topi

cal

•G

ood

evid

ence

they

are

•

Goo

d ev

iden

ce th

at to

pica

l NSA

IDs

are

asso

ciat

ed w

ith

NSA

IDs

com

para

ble

to o

ral N

SAID

s in

crea

sed

loca

l adv

erse

eve

nts

com

pare

d w

ith o

ral N

SAID

s.fo

r pa

in r

elie

f in

tria

ls

•G

ood

evid

ence

that

topi

cal a

nd o

ral N

SAID

s ar

e co

mpa

rabl

epr

imar

ily o

f pa

tient

s w

ith

in r

ates

of

tota

l adv

erse

eve

nts

and

with

draw

als

due

knee

ost

eoar

thri

tis.

to a

dver

se e

vent

s.•

Mos

t tri

als

of to

pica

l •

Goo

d ev

iden

ce th

at to

pica

l NSA

IDs

are

asso

ciat

ed w

ith f

ewer

NSA

IDs

eval

uate

pro

prie

tary

G

I ev

ents

, inc

ludi

ng s

ever

e ev

ents

, and

cha

nges

in h

emog

lobi

nfo

rmul

atio

ns n

ot a

vaila

ble

com

pare

d to

ora

l NSA

IDs.

in th

e U

nite

d St

ates

.

Topi

cal

•Fa

ir e

vide

nce

that

cap

saic

in,

•G

ood

evid

ence

that

topi

cal c

apsa

icin

is a

ssoc

iate

d w

ithN

one

salic

ylat

esbu

t not

topi

cal s

alic

ylat

es a

re

incr

ease

d lo

cal a

dver

se e

vent

s an

d w

ithdr

awal

s du

e to

and

caps

aici

nsu

peri

or f

or p

ain

relie

f ad

vers

e ev

ents

com

pare

d to

pla

cebo

.co

mpa

red

to p

lace

bo.

Abb

revi

atio

ns:

CH

F =

con

gest

ive

hear

t fai

lure

; CO

X =

cyc

lo o

xyge

nase

; CV

= c

ardi

ovas

cula

r; G

I =

gas

troi

ntes

tinal

; NSA

ID =

non

ster

oida

l ant

iinfl

amm

ator

y dr

ug;

PPI

= p

roto

n pu

mp

inhi

bito

r.

AH

RQ

Pub

. No.

06-

EH

C00

9-1

Sep

tem

ber

200

6