Background/Purpose: Because older patients with osteoarthritis (OA) and multiple comorbidities face high risk of toxicity from nonselective non-steroidal antiinflammatory drugs (NSAIDs) and Cox-2 inhibitors, opiates (including tramadol) have been proposed as an analgesic strategy. We evaluated clinical and economic outcomes of using prototypic medications (tramadol, naproxen, and celecoxib) in such patients.

Methods: We used the Osteoarthritis Policy Model, a validated computer simulation of knee OA, to project long-term clinical outcomes, costs and incremental cost-effectiveness ratios (ICERs) of OA treatment strategies in patients with mean age 70, knee OA, diabetes and coronary heart disease whose pain persists after initial therapy with acetaminophen, steroid injections and physical therapy (PT). We examined four treatment strategies: 1) continuing PRN acetaminophen; 2) tramadol; 3) naproxen and 4) celecoxib. Pain was the primary determinant of quality of life and its relief was assessed with the Western Ontario McMaster (WOMAC) Pain Scale. Treatment efficacies and toxicities were estimated from published literature and influenced time on regimen. Mean WOMAC change was 22 points for tramadol; 15 for naproxen and celecoxib. Annual medication costs and major toxicities for the first and subsequent years are shown in the Table. Toxicities included CVD and – for tramadol — fractures. We adopted a societal perspective, discounting outcomes at 3%, and assumed a willingness to pay (WTP) of $100,000 per quality adjusted life year (QALY) gained. ICERs below this threshold defined cost-effective strategies.  

Results: Patients remained on tramadol for 1.98 years and the other regimens for 2.36 years. Twice as many experienced major toxicity with tramadol as with the other agents.  The Table lists the cost-effectiveness results. The ICER for tramadol exceeded that for naproxen; thus, we compared naproxen directly to PRN acetaminophen and observed that naproxen had an ICER of $78,848/QALY. However, tramadol cost effectiveness was highly sensitive to its toxicity. When tramadol toxicity was reduced by just 10% it became cost-effective (ICER $53,968/QALY), and ICERs for the naproxen-based strategy then exceeded the WTP threshold (ICER=$102,000/QALY) compared to the tramadol-based strategy.

Conclusion: In patients with OA and multiple comorbidities who have pain despite acetaminophen, steroid injection and PT, naproxen was cost-effective at a WTP=$100,000. Tramadol became cost effective following a 10% reduction in its overall toxicity. The cost of celecoxib precluded its offering acceptable value. The impact of tramadol toxicity on these estimates underscores the need for further research on toxicity of opiates in frail patients; the limited number of years on regimen highlights the need for therapies with better efficacy and toxicity profiles.