Don't mix the design with the expectation ... a genius does not mix these ... do not have to be a genius to understand the difference between the two ...
The ">95% power to detect >20% relative risk reduction" was in the IRP between June 2015 and January 2017 and it obviously didn't come from nowhere ... it came from the given (final) design. e.g. they could say ">99% power to detect >30% relative risk reduction" but it is the result of the 7,990 patient and min. 5.2% p rate ... isn't original / pre-SPA design.
Please, cite the source, since I (and as I remember: nobody other than you) do not aware any type of patent / citation / source / etc. about / for it.
??? My quote is B&W ... the reason of the addition of the 1,000 patients was due to only: "To protect against the possibility that the actual placebo event rate is lower than estimated ... without having to modify the other sample size assumptions" (other sample size assumptions include "relatively reduced by 15%", expected RRR wasn't changed).
We do not know what is / was (at the time of the design) the expectation (see / saw at doable) of Amarin, but of course it is / was higher than 15% RRR and higher than 5.9% placebo rate, since nobody design a $100M+, double-blinded trial on the edge ... The expected RRR could be 20%, but could be 40% also ...
e.g. - with 4.8% composite rate (tracking as projected) and 6.0% placebo rate (assuming actual is equal with expected) the RRR is 40% - with 4.8% composite rate (tracking as projected) and 5.32% placebo rate (assuming actual placebo event rate is lower than (originally) estimated, but higher than the final "minimum") the RRR is 20%