If you were cycling or driving, you’d probably know what the stopping rules were. Traffic not moving, big red sign, large goose with malevolent glare (Lincolnshire speciality).
What if you’re doing a clinical trial?
There are a variety of things what have been described, some of them are qualitative (SUSAR – sudden, unexpected, serious adverse reactions) and some statistical. The latter have with them a set of maths that leads to reasons to discontinue, either for proven benefit or futility.
The maths is argued over by clever people, but they all conceptually do the same thing. They ask “what is the likelihood of this difference being a chance finding?” and balance this with “and if we keep looking and looking, eventually we will see something”.
This means that the trial designers need to decide what their significance level and power are that they are aiming for, their minimally clinically important difference and then how many times they are going to look. (The simplest idea is that if you look twice, you have to halve the p-value you’re going to accept as significant, cause you’ve used up one lot of assessing.) This then produces a size estimation for the study and boundaries for “better – stop!” and “futile – stop!” at each reassessment.
The usual maths of doing this is with the ‘Flemming-O’Brien alpha spending function’, but a more efficient variant is the Triangular Test, which recalibrates as it ammasses data.
(PS – This post was requested by – You can ask too!)
@drbobphillips can you do a blog piece on ‘triangular test?’ Google not helping. Authors closed study because of results of this? ?
— XOXO بنت القيل (@hhassan90210) July 13, 2015
(If you want a longer & more intellectual read – try here.)