Statistics Collaborative - Design and analysis for biomedical research


Adaptive designs, conditional power, and futility

Statistics Collaborative, Inc. (SCI) has many years of experience reporting interim results to Data Monitoring Committees (DMCs). A DMC evaluates interim data periodically during a clinical trial to assess whether the trial should continue as designed, be modified, or discontinue early.

A trial may be discontinued before its originally planned conclusion when interim data provide strong evidence of either the efficacy or the futility of the experimental agent. Conditional power is the probability of observing a statistically significant treatment effect at the end of a trial, conditional on the data observed thus far.

SCI works with clients and DMCs to present the data necessary for the DMC to make a decision regarding the efficacy or futility of an experimental agent.

Examples of SCI’s work in adaptive designs, conditional power, and futility:

  • Sample size recalculation: Midway through a Phase 3 trial evaluating treatment in patients at risk of acute attacks of hereditary angioedema (HAE), SCI recalculated the trial’s sample size on the basis of the observed variability of the treatment effect. This ensured that the trial included enough subjects to maintain adequate power to test the hypothesized efficacy of the drug.
  • Conditional power: For a large Phase 3 trial investigating a therapy to reduce mortality in subjects with severe sepsis, the trial’s DMC recommended early discontinuation of the trial because of futility. The DMC’s decision was based on SCI’s calculation of the trial’s conditional power, ad hoc confirmatory analyses of the trial’s other efficacy measures, and subgroup analyses to rule out a potential benefit in a subset of patients – all of which showed that the therapy was unlikely to be proven effective with enrollment of a further 1,200 patients.

Examples of publications:

  • Downs M, Christ-Schmidt H. Visualizing conditional power. [Abstract] Controlled Clinical Trials 2002; 23:68S.
  • Wittes J, Lachenbruch P. Opening the adaptive toolbox. Biometrical Journal 2006; 48:598-603.
  • Wittes J. Adaptive designs for clinical trials. P Armitage, T Colton (eds). Encyclopedia of Biostatistics, Second Edition. New York: John Wiley, 1:51-55, 2005.
  • Wittes J. On changing a long-term clinical trial midstream. Stat in Med 2002; 21:2789-2795.