• Paul A Scuffham, Michael J Yelland, Jane Nikles, Eva Pietrzak, and David Wilkinson.
• School of Medicine, Griffith University, Meadowbrook, Qld, Australia.
• Value Health. 2008 Jan 1; 11 (1): 97-109.

ObjectiveTo explore the economic viability of N-of-1 trials for improving access to selected high cost medications in Australia.MethodsCost and effectiveness estimates were derived from two N-of-1 trials conducted by The University of Queensland from 2003 to 2005-celecoxib versus sustained-release paracetamol for osteoarthritis in a general practice setting and gabapentin versus placebo for chronic neuropathic pain in a hospital setting. Effectiveness was determined by the proportion of responders to each medication. The costs of trials were offset against the savings generated by subsequent changes in prescribing. Decision analysis models with semi-Markov processes were used to compare different scenarios of N-of-1 trials versus usual care.ResultsThe fixed cost of performing N-of-1 trials was approximately AUS$23,000 for each trial and the variable cost was approximately AUS$1300 per participant. Clinical outcomes favored celecoxib over paracetamol in 17% of participants and gabapentin over placebo in 24% of participants. Modeling these results showed that the cost-offsets from efficient use of medications were less than the cost of running a trial; however, the incremental costs per quality-adjusted life-year gained were AUS$6,896 and AUS$29,550 for the gabapentin/placebo and celecoxib/paracetamol trials, respectively, over a 5-year horizon. Key factors affecting the viability were the time horizon modeled, the variable cost per participant, the probability of response to the intervention medication, and rates of use in nonresponders and the usual care alternative.ConclusionsThe N-of-1 strategy offers a realistic and viable option for increasing access to selected high cost medications where the medications are used for the symptomatic treatment of chronic disease, have rapid onset of action, and clinical response is unpredictable without a trial.

43 keywords...

hide…