OA imposes a significant functional burden on affected individuals. NHANES III (the Third National Health and Nutrition Examination Study) revealed that more than 8% of US adults have symptomatic hand OA, and there were no sex-specific differences 4. Furthermore, these individuals had difficulty lifting 10 pounds, dressing and eating, and had significantly increased use of analgesics, particularly acetaminophen. There was no increase in the use of nonsteroidal anti-inflammatory drugs (NSAIDs). A French study 5 found that more than 80% of clinical OA patients reported limitations in their daily lives, including basic tasks, work and leisure activities, and these limitations affected patients who were retired as well as those still working. Finally, Heuts and coworkers 6 measured the association between pain-related fear and daily living in patients with OA and found that pain levels and fear of (re)injury were both associated with significant functional limitations. Interestingly, radiological findings were not significant predictors as compared with pain-related fear, emphasizing the need for effective analgesics that can be used in the long term.

OA imposes a significant economic burden for both patients and health-care systems. Rabenda and colleagues 7 measured the direct and indirect costs to individuals with OA. The mean direct cost to patients was €44.5 per patient-month, and this included hospitalizations, drugs, and physician examinations. In addition, patients also required a mean 0.8 days sick leave per patient-month, resulting in substantial costs to employers, whereas unaffected care givers required an average 0.2 days sick leave per patient-month. A study conducted in the USA 8 revealed that patients with OA incur direct costs of more than $2,600 per year, approximately twice as much as do individuals without OA. Similarly, a Canadian study 9 confirmed that 60% of patients experienced costs directly related to their OA and that much of this was due to time lost from work and leisure activities for patients and unpaid care givers.

Health care systems also bear an economic burden due to OA. A study 10 calculated that from July 1993 to June 1994, the annual cost to an individual health maintenance organization for 10,101 patients with OA was more than $4.7 million, including hospital care, medications, and ambulatory care. The authors pointed out that the OA care was infrequent and that the largest portion of the expenditure was for hospitalizations, which affected only 5% of the OA patients; this indicates that there is a need for improved OA management. A French study 11 calculated that in 2002 the direct costs of OA in France were €1.6 billion, which comprised about 1.7% of expenditure for the French health care system. Expenditures in 2002 were 156% higher than in 1993, primarily due to an increased number of patients. These numbers also reinforce the need for effective strategies to manage the symptoms of OA and decrease the number of patients with functional disability.

As the number of individuals with OA increases, effective strategies to manage the pain and disability of OA will become even more important. Guidelines for the management of OA have been published recently by both the OsteoArthritis Research Society International (OARSI) 1213 and the European League Against Rheumatism (EULAR) 141516. Experts from multiple disciplines, who analyzed and weighed the scientific and clinical evidence as well as previous guidelines, developed recommendations for the management of knee, hip, and hand OA. The guidelines indicate that a combination of pharmacological and nonpharmacological modalities is the most effective strategy to manage the pain and disability associated with OA.

Pharmacological guidelines for the management of OA recommend acetaminophen up to 4 g/day as first-line therapy and suggest that alternative pharmacological therapy should be used only in the presence of an inadequate response and severe pain 131415. This is because of a perceived lower risk for adverse events from acetaminophen as compared with NSAIDs. Both OARSI and EULAR guidelines recommend a switch to NSAIDs if acetaminophen cannot adequately control symptoms or if there are signs of clinical inflammation (Table 1) 13. Once the decision has been made to use NSAIDs, OARSI and EULAR guidelines both recommend the lowest effective dose and, in patients with increased gastrointestinal risk, consideration of a gastroprotective agent such as proton pump inhibitors or misoprostol. All of the guidelines recommend monitoring of individual patients after initiation of NSAID treatment. Finally, the OARSI and EULAR guidelines recommend that if patients do not respond to oral analgesics, then they should receive intra-articular injections of either corticosteroids or hyaluronate for OA of the knee or hip, followed by the use of opioids and narcotics only when all other pharmacological options have been considered. Surgery, including joint replacement, is recommended only as a last resort.

Despite the recommendation that acetaminophen be used as first-line therapy, there is evidence that NSAIDs may be more effective than acetaminophen for the control of pain in patients with OA. Several published studies have suggested that acetaminophen is not as effective as either NSAIDs 1718 or placebo 19. In addition, a systematic review conducted by the US Agency for Healthcare Research and Quality 20 indicated that although there does not appear to be efficacy differences between selective and nonselective NSAIDs, all NSAIDs appear to be superior to both aceta-minophen and placebo for the treatment of OA symptoms. Although the recent EULAR and OARSI guidelines recognize differences in efficacy, the recommendations for initial treatment with acetaminophen remain based on issues of safety 131415.

The American Heart Association (AHA) recently issued a controversial statement that included a stepped care approach to the management of musculoskeletal symptoms 21. The AHA recommends that for patients with symptoms that cannot be controlled by nonpharmacological approaches, acetaminophen, full doses of aspirin, nonacetylated salicylates, or short-term narcotics should be used as initial therapy, and only those individuals who cannot tolerate the previous treatments or who require long-term or high-dose therapy should receive NSAIDs. Nonselective NSAIDs should be used first, and only if symptoms are not controlled should NSAIDs with 'increasing degrees of COX-2 [cyclo-oxygenase-2] inhibitory activity' be used before 'ultimately concluding with the COX-2 selective NSAIDs.' The AHA rationale for this approach is that some COX-2 selective NSAIDs might carry increased risk for cardiovascular events compared with nonselective NSAIDs.

Concerns about the AHA recommendations were addressed in an editorial published in 2007 22. First, Moskowitz and coworkers 22 wrote that nonacetylated salicylates tend to be less effective than NSAIDs and may take longer to benefit patients. There is also no evidence to support the contention that nonacetylated salicylates have an improved cardiovascular safety profile compared with NSAIDs. In addition, the authors of the editorial stated that full-dose aspirin confers increased risk for gastrointestinal bleeding as well as the potential for increased hemorrhagic stroke. The authors suggested that the AHA reconsider their recommendations. Finally, because there is no clinical support for the idea that it is possible to differentiate cardiovascular risk among selective and nonselective NSAIDs on the basis of in vitro COX-2/COX-1 inhibition ratios, the editorial suggested that COX-2 selective agents should be considered, particularly in patients with increased gastrointestinal risk, although NSAID drugs should only be prescribed after evaluation of the risk/benefit balance for the individual patient 22.

Moskowitz and coworkers 22 also disagreed with the AHA regarding the early use of opioids to treat patients with musculoskeletal pain, arguing that long-term use is not recommended by either the American Pain Society or the American Academy of Pain Management. In addition, the editorial points out that patients with OA tend to be older and therefore more likely to experience adverse events associated with opioid use; these include dizziness, confusion, and risk for falling, which could in turn increase the risk for hip fracture.

A number of studies have been conducted to determine the cardiovascular risks associated with both selective and nonselective NSAIDs. An epidemiological study conducted by McGettigan and Henry 23 identified increased risk for cardiovascular events with rofecoxib, whereas Hernandez-Diaz and colleagues 24 observed increased risk with rofecoxib and diclofenac, and presented data suggesting an increase in risk for cardiovascular events with ibuprofen. In their meta-analysis, Kearney and coworkers 25 reported that the COX-2 selective NSAIDs rofecoxib and celecoxib, as well as high doses of the nonselective NSAIDs ibuprofen and diclofenac, were all associated with moderately increased risk for cardiovascular events. In contrast, high-dose naproxen was not associated with increased vascular risk. The authors went on to state that there was insufficient data to determine whether naproxen is protective, as has been suggested in other studies.

Although the meta-analysis reported by Kearney and coworkers was unable to determine whether cardiovascular risk differed between aspirin users and nonusers, Strand 26 discussed epidemiological studies reporting that concomitant aspirin therapy decreases the risk for cardiovascular events with COX-2 inhibitors as well as some nonselective NSAIDs, but not with ibuprofen 2728. That investigator also indicated that these events were confirmed in the CLASS (Celecoxib Long-term Arthritis Safety Study) 29 and TARGET (Therapeutic Arthritis Research and Gastrointestinal Event Trial) 30 trials.

In addition to an increase in the risk for cardiovascular events, NSAIDs increase blood pressure. Two meta-analyses 3132 concluded that NSAIDs increase both systolic and diastolic blood pressures, and a later clinical trial 33 indicated that although treatment with rofecoxib significantly elevated 24-hour systolic blood pressure, this was not the case in patients receiving celecoxib or naproxen. However, the authors noted that all three NSAIDs destabilized hypertension control, reiterating the need for careful monitoring of patients with hypertension.

NSAIDs are also associated with gastrointestinal events, including dyspepsia and heartburn (which affect up to 60% of NSAID users), as well as gastric and duodenal ulcers 34. Garcia Rodriguez and Hernandez-Diaz 35 found that the risk for symptomatic but uncomplicated peptic ulcer increased from 1.03 per 1,000 person-years in individuals who did not use NSAIDs or aspirin to 4.0 per 1,000 person-years for nonaspirin NSAID users. In addition, this risk increased slightly when patients used NSAIDs for more than 6 months. Furthermore, the use of aspirin either with or without other NSAIDs increased the risk for uncomplicated peptic ulcer. A systematic review of epidemiological studies 36 showed that the pooled relative risk for a complicated peptic ulcer was 3.8 for nonselective NSAID use, although the review did find some differences between specific NSAIDs. An observational study of elderly patients conducted to determine the risk for complicated gastrointestinal events in patients receiving NSAIDs 37 found that the incidence of complicated upper gastrointestinal bleeding with nonselective NSAIDs was 4.0, whereas the risk among those using celecoxib was 1.0. This illustrates that COX-2 selective inhibitors do decrease the risk for gastrointestinal complications.

A decreased risk for gastrointestinal events has also been observed in clinical trials. For example, the CLASS study 29 revealed that celecoxib users had fewer symptomatic ulcers and complications than did individuals who received therapeutic doses of nonselective NSAIDs, although the statistical significance was not reached at 1 year. In addition, the MEDAL (Multinational Etoricoxib and Diclofenac Arthritis Long-term) 38 study demonstrated that individuals who received etoricoxib had significantly fewer uncomplicated gastrointestinal events than those who received diclofenac. Patient risk factors that increase the risk for developing gastrointestinal side effects include age, a history of ulcer or dyspepsia, Helicobacter pylori status, and concurrent diseases (for review 39).

The question of whether low-dose aspirin abrogates the gastrointestinal protection of COX-2 inhibitors is continuing to be addressed. Strand 26 argued that although evidence from the CLASS and TARGET studies does not show a significant benefit, the MEDAL study identified significantly fewer complicated ulcer events with etoricoxib and aspirin than with diclofenac and aspirin.

Moore and coworkers 40 calculated the absolute risk for cardiovascular and gastrointestinal events and determined that although gastrointestinal complications occur more frequently with nonselective NSAIDs than with selective COX-2 inhibitors, serious cardiovascular events occur at approximately equal rates. A retrospective cohort study examining administrative data from elderly patients who filled a prescription for NSAIDs or acetaminophen 41 showed that among patients who did not use aspirin, the risk for hospitalization for acute myocardial infarction/gastrointestinal bleeding was greater with naproxen, but the risk associated with celecoxib was comparable to that with acetaminophen. In aspirin users, both naproxen and celecoxib appeared to be the least toxic. Based on these data, Jones and colleagues 39 developed evidence-based recommendations according to the gastrointestinal risks and the cardiovascular risks at the individual level (Table 2).

Renal adverse events associated with NSAID use include decreased renal perfusion, edema, increased blood pressure, and interstitial nephritis 4243. Although these events are relatively rare, the number of patients receiving NSAIDs makes the risks clinically relevant, and patients at increased risk include those with age-related declines in glomerular filtration rate, those with hypovolemia, and those with congestive heart failure, cirrhosis, or nephrosis (for review 44). Based on the adverse events associated with rofecoxib, it is possible that adverse events are attributable to a class effect. However, a meta-analysis performed to quantify the risks for renal events (including renal dysfunction, hypertension, and peripheral edema) associated with COX-2 inhibitors 45 found that although rofecoxib was associated with increased renal events, no increase was seen with other COX-2 selective NSAIDs.

Despite published recommendations stating that acetaminophen has a better safety profile than NSAIDs 1315, recent reports have suggested that there are more adverse events associated with acetaminophen than was previously believed. For example, a UK nested case-control analysis including more than 950,000 individuals between 1993 and 1998 46 showed that individuals who used more than 2 g/day acetaminophen had a relative risk of 3.6 for upper gastrointestinal complications versus relative risks of 2.4 and 4.9 for low/medium and high doses of NSAIDs, respectively. These findings were supported by other studies. One 47 found that among patients aged 65 years or older, those receiving higher doses of acetaminophen had an increased risk for gastrointestinal events compared with those receiving lower doses. Another 48 revealed that use of a combination of a nonselective NSAID plus acetaminophen may increase the risk compared with either agent alone. In addition, there are studies in the literature that suggest that acetaminophen use may be associated with a modest increase in risk for renal functional decline in women consuming more than 3 g/day 49 and that in patients with early-stage chronic renal failure acetaminophen use may exacerbate the effects of renal failure 50. Furthermore, studies suggest that acetaminophen use is associated with a moderate increase in risk for incident hypertension in both men 51 and women 52. Given these results, clinicians must consider the safest therapeutic agent for individual patients with OA.