"Note: Some rounding has been done so values above are +/- 0.5 month. Based on above, a crude estimate of cumulative patient years is 4244, which divided by 2400 patients gives 1.77 years (21.2 months) as the average dosing period. However, this is a crude estimate that does not account for patient deaths or patient dropouts; the "wave" approach creates a lot of uncertainty too."

I had some thought provoking private messages with a few folks this afternoon. It got me thinking more about the degree of patient drop outs in BETonMACE. There are always patient drop outs, especially in long clinical trials. Some of it is because patients die, but for most cases it is some kind of non-fatal adverse event that leads to patient voluntary discontinuation/withdrawal of consent or discontinuation at the advisement of the physician. I looked back at a few recent diabetic CVOT studies and the typical range is 20-30% discontinuation. SUSTAIN-6, EXAMINE, EMPA-REG OUTCOMES, CANVASS, ELIXA. Like I said, some of this is due to patient death. But a majority is for reasons other than death. 

So thinking back to my patient year estimate of 4244......if we shave off 20% due to patient discontinuation, this gives us closer to 3400 patient years. The percentage of discontinuation may be higher too. However, we don't know how long the patient may have been on placebo or study drug before discontinuation. So a straight cut of 20% may be an over or underestimate. Therefore the events per patient years (annual event rate) estimates, such as those that George provided earlier,may be underestimates due to this discontinuation variable.

BDAZ