Tada wrote: "The 30% RRR reduction with 80% power was for 2400 patients with a total of 3600 patient years of dosing to get 250 events to achieve that statistical significance with 80% power. We had 2425 patients that went a little bit longer than the anticipated 18 months on average, likely having more than 3600 patient years. There seems to be a few more than 250 events (we'll know the exact number soon) so we may find that statistical significance will be achieve with less than 30%RRR."

10BagR wrote: "So any additional patients in the trial and or additional cumulative time in the trial in terms of patient hours would therefore provide for greater than 80% confidence.   But the 25-30% RRR remains the same regardless."

Tada summed it up nicely. I'd like to emphasize that a couple big unknowns are the number of patient years and the dropout/discontinuation rate. So while it may be true that the more patients and patient hours accumulated may increase the power and confidence, we cannot conclude that BETonMACE has more patients and patient years than planned due to the unknowns about dropout/discontinuations and therefore patient years. All just my opinion.

Also, as Tada suggested, the primary endpoint is only successful if it acheives statistical significance, regardless of whether it is greater or less than 30%. In general (I'm no biostatistician), a trial will be more likely to achieve statistically significant difference between groups if the magnitude of the effect size (i.e. %RRR) matches or exceeds what the trial was powered for. Conversely, a trial will be less likely to achieve statistically significant differences between groups if the magnitude of the effect size is less than what the trial was powered for. Like I said, I'm no biostatistician and I probably can't even spell that word correctly without looking it up. It would be nice to achieve a robust, impressive p-value. It would be heartbreaking to miss on statistical significance with a borderline p-value.

BearDownAZ