английский => русский Наука Математика и статистика Medical Statistics

In 2011, Mehta and Pocock introduced the “promising zone design”, building on work by Chen et al. in 2004 and making it more user-friendly for easier implementation [5, 6]. At the interim analysis, conditional power is calculated based on observed data and assuming the future treatment effect will follow the current trend observed so far.

At the analysis, the conditional power value will fall in one of the three pre-specified “zones”: “unfavourable” (i.e. not worth the increase in sample size required to maintain originally planned power).

“Promising” (treatment effect may be lower than expected, but not so low that conditional power cannot be recovered through increasing the sample size).

“Favourable” (the treatment effect is “sufficiently favourable”, and sample size does not need to increase to ensure power is maintained).