Borel Fields,
Tanks for sharing your considerations. Not sure if a detailed discussion will be valuable for all the investors, however I would like to comment on some points:
Not sure which discussions ??. I am just sharing my thoughts after doing (personal) calculations as I believe they indeed provide information on trial parameters and possible results. Let me give you an example: HGEN announced that:
Well, with this information one is able to calculate the p-value they had in the interim and, what is the possible outcome of the trial if results continue with the same trend.
This comes partially from ad-hoc power calculation. Is this information valuable ?? Heck yes (for the HGEN investors)
Indeed. And this is precisely the point. But, also provides information on "how are things" currently, doesn't it ??? Shall we "feel" the same if DSMB says: carry on or "maybe is better to include 200 more patients to meet the designed power??"
Of course, of whatever the trial design parameters were. The point is that trial design parameters do not dictate that the trial will conform to them. Results can or cannot conform to the design. In a trial with "tight" parameters this is more probable.
???? I am just answering some questions of interest to our trial. I can only make calculations making assumptions. For example, we do not (exactly) know the number of deaths, or more importantly, the reduction of deaths. But I like to surmise (valuable) information by doing the calculation. That is all. I have tried not to share numbers and all I have said is that I believe we are slightly "under-powered". This, I believe, is because the designers have a high confidence that the death reduction is going to be stellar.
I share this confidence.
Tanks for sharing your considerations. Not sure if a detailed discussion will be valuable for all the investors, however I would like to comment on some points:
Quote:I'm afraid I can't get on board with TechGuru's discussions of post-hoc power. I have never seen it calculated when a result is significant, and there are problems in using it even when p>.05.
Not sure which discussions ??. I am just sharing my thoughts after doing (personal) calculations as I believe they indeed provide information on trial parameters and possible results. Let me give you an example: HGEN announced that:
Quote:The data safety monitoring board (DSMB) composed of independent subject matter experts conducted an interim analysis of the unblinded data for trial sizing and powering and recommended increasing the target number of events (recoveries) from 257 to 402 to maintain the power of the study at 90 percent. The adaptive trial design only allows for the addition of patients if interim data are in the “promising zone” and also pointed out that:
The company remains blinded to the data and based on the recommended number of events, the HR was calculated to be 1.37, an average of 37 percent more recoveries observed in the lenzilumab arm compared to the control arm.
Well, with this information one is able to calculate the p-value they had in the interim and, what is the possible outcome of the trial if results continue with the same trend.
This comes partially from ad-hoc power calculation. Is this information valuable ?? Heck yes (for the HGEN investors)
Quote:Yes, that depended on a certain death rate in the control group. Post hoc you can recalculate, based on actual total deaths, whether the assumption converting sample size to total deaths was wrong. That's what the DSMBs do in deciding if sample size should be increased to "maintain power."
Indeed. And this is precisely the point. But, also provides information on "how are things" currently, doesn't it ??? Shall we "feel" the same if DSMB says: carry on or "maybe is better to include 200 more patients to meet the designed power??"
Quote:KEY POINT: this calculation is still vs. the 30% death reduction target, and it has nothing to do with the observed difference between treatment vs. control death rates.
Of course, of whatever the trial design parameters were. The point is that trial design parameters do not dictate that the trial will conform to them. Results can or cannot conform to the design. In a trial with "tight" parameters this is more probable.
Quote:As TechGuru is calculating it, I'm pretty sure any outcome with P=.05 will calculate to a 50% post-hoc power.
???? I am just answering some questions of interest to our trial. I can only make calculations making assumptions. For example, we do not (exactly) know the number of deaths, or more importantly, the reduction of deaths. But I like to surmise (valuable) information by doing the calculation. That is all. I have tried not to share numbers and all I have said is that I believe we are slightly "under-powered". This, I believe, is because the designers have a high confidence that the death reduction is going to be stellar.
I share this confidence.
7
