Sample size re-evaluation

In Chapter 8 we spoke about the calculation of sample size and in Section 8.5.3 revisiting this sample size calculation as the trial data accumulates. 

Maintaining the blind is critical in these re-evaluations, otherwise a price in terms of the type I error, a, may need to be paid as looking at unblinded data could be viewed as a formal interim comparison of the treatments. There may 

CH14 INTERIM ANALYSIS AND DATA MONITORING COMMITTEES 

Generally speaking these sample size considerations are independent of any formal interim comparisons of the treatments, be they efficacy or futility. Given the need to maintain blinding it is also advisable not to involve the DMC in these re-evaluations their potential knowledge of the unfolding data in an unblinded way could well influence their view on a change in the sample size - they know too much  

Example 14.2 (continued) ESTAT trial in acute stroke 

---

Undertaking interim analyses for either efficacy or futility or both, together with sample size re-evaluation already provides a range of flexibility in the design and planned correctly these approaches may cover many situations of practical interest. What more could be done Phillips and Keene (2006) list a range of potential adaptations including the following ... 

Sample size re-evaluation based on unblinded data. 

Phase III is the confirmatory stage and major adaptation outside of pre-planned interim analyses and blinded sample size re-evaluation could potentially undermine the confirmatory nature of the trial. 

---