IN PURSUIT OF THERAPEUTIC PRECISION: HOW OFTEN DO N-OF-1 TRIALS AFFECT TREATMENT DECISIONS?

Purpose: Randomized controlled trials (RCTs) generate average treatment effects, but patients want to know which treatments will work for them. N-of-1 clinical trials (crossover experiments conducted in single individuals) may be ideal for determining individual treatment effects, potentially enhancing therapeutic precision. However, comprehensive data on how often n-of-1 trials affect treatment decisions is lacking. We performed this study to examine treatment changes resulting from published n-of-1 trials.

Methods: We undertook a systematic review of English-language n-of-1 trials published between 1985 and February 2008, and indexed by PubMed and Web of Science. Included articles were those having individuals as the unit of randomization and reporting individual outcomes. All articles were reviewed by a single investigator with independent over-reading of a 20% sample by a second. We abstracted trial characteristics, along with treatment information for all participants (when provided). Treatment decisions after n-of-1 trial completion were categorized as consistent with trial results, inconsistent, or uncertain. Descriptive statistics are used to illustrate n-of-1 trial results.

Results: Our search identified 477 articles, and 89 trials met our inclusion criteria. These trials represented 1765 participants, of which 1358 (77%) completed the planned n-of-1 trial (89% completed at least one crossover). Treatment change information was available for 465 (34%) of the completers. 54% of participants had subsequent treatment decisions consistent with n-of-1 trial results (15% of the results favored the initial treatment and the patient continued with the initial treatment, while 39% of the results favored the alternative treatment and the patient switched to the alternative treatment). 8% of participants were treated in ways inconsistent with n-of-1 trial results. In 4% of cases, the patient either switched to or continued with the treatment not favored in the n-of-1 trial, while in 4% the patient stopped all treatment when a clear "winner" was determined. Among the 38% of trials in which neither treatment was clearly better, 20% of participants remained on the initial treatment, 13% switched to the alternative treatment, and 5% stopped all treatment.

Conclusions: Our results indicate that the majority of n-of-1 trials confirm prior treatment decisions or prompt new ones and are therefore a useful tool for enhancing therapeutic precision in selected conditions. To facilitate future meta-analysis, researchers reporting n-of-1 trial results should clearly describe individual data on treatment change.