Enhancing crossover trial design for rare diseases: limiting ineffective exposure and increasing study power by enabling patient choice to escape early

Contemp Clin Trials. 2014 Jul;38(2):204-12. doi: 10.1016/j.cct.2014.05.001. Epub 2014 May 12.


Background: Addressing the two most important considerations in designing clinical trials, i.e. maximizing study power and minimizing patient exposure to ineffective treatment, is particularly challenging for trials of rare diseases. The familial Mediterranean fever (FMF) rilonacept trial (Hashkes et al., Ann Intern Med 2012;157:533-41) demonstrates a novel crossover design by enabling patient choice to early escape for rare disease.

Purpose: To investigate the effect on study power, exposure to the ineffective treatment arm and dropout rate by implementing early escape to crossover design, and to propose a Bayesian modeling approach.

Method: Based on the FMF trial data, simulation studies compared study power and dropout rate among three types of designs for crossover trial: traditional without early escape, early escape per-patient-choice, and early escape per-protocol.

Results: The early escape per patient choice or per protocol design achieved 0.89 ± 0.12 and 0.78 ± 0.20 of the study efficiency when compared to the traditional crossover design assuming no dropout. Early escape per patient choice compared to early escape per protocol improved power by 1.29 ± 0.26, and reduced the dropout rate by 8-29%, but with greater patient exposure to the less effective treatment arm.

Conclusions: The results of the FMF trial and simulation studies suggest that allowing early escape in crossover trial enhanced the design by minimizing patient's exposure to the ineffective treatment arm while maintaining a reasonable study power, which is particularly important for rare disease trials. Choice between the two types of early escape presents tradeoff between study power and exposure to ineffective treatment.

Trial registration: ClinicalTrials.gov NCT00582907.

Keywords: Early escape; Missing data prevention; Patient decision; Randomized clinical trial; Small trial; Statistical power.

Publication types

  • Randomized Controlled Trial
  • Research Support, U.S. Gov't, P.H.S.

MeSH terms

  • Bayes Theorem
  • Cross-Over Studies*
  • Double-Blind Method
  • Familial Mediterranean Fever / drug therapy
  • Humans
  • Rare Diseases / drug therapy*
  • Recombinant Fusion Proteins / therapeutic use
  • Research Design*
  • Sample Size


  • Recombinant Fusion Proteins
  • rilonacept

Associated data

  • ClinicalTrials.gov/NCT00582907