Interim Analyses of Clinical Trials, page-4

Hi II,

You have been busy & thanks for making a new thread on this topic, as it may not be of interest to everyone here & I do appreciate your efforts after a big day.

What we do know is that interim analyses are pre-defined before commencement of studies in a statistical plan for trial protocol. I appreciate what you have done here as an explainer, but it appears that it is a little more complicated than that, for the DSMB (alternative terminology is DMC Data Monitoring Committee) review & the decisions they make to either stop a study or continue.

So I’m going to add a couple of references here, to build on what you have already done.

This first one I have added, because it is written by biostatisticians in the context for non-biostatisticians so that maybe we can understand complex issues better in Interim Reviews.

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10260346/
So it isn’t all black & white with numbers early on in a trial. Treatment effect actually does come into the equation as does safety obviously with stopping a trial too early. What is very clear though is confidentiality in unblinded data not to impact on the integrity of a clinical trial & from the above article & this is why we & Dimerix will only receive limited information on the outcome of DSMB review at interim one analysis:





The second one is Guidance for Sponsors from the FDA on DSMB (DMC) review. From pp14 maybe of more interest. This is from 2006, there maybe updates, so apologies for using a reference greater than 10 years old.

https://www.fda.gov/media/75398/download

I also did have a look at the references provided, so thank you. There are examples out there of statistical analysis in CKD trials published & as I mentioned previously, there is research ongoing in this space & I had posted a CKD trial meta-analysis recently which also is helpful.

There is going to be differences between indications, trial design & historical trial data & Regulatory requirements to inform current trial design. We saw that early on in ACTION3 design. That is what I have been looking at, not trial designs in general & basics 101 in clinical trials. Oncology interim analyses may be different again in open label trials for example. I think there would have to be a very VALID reason to stop this trial early for overwhelming efficacy on Interim 1, as it may not capture the more long term eGFR slope, which is a longer term confirmatory endpoint in CKD trials.

So maybe that is where we do differ in opinion II, but I do appreciate your input & just take a look at your example again of statistical analysis & numbers, I’m sure it was just a typo, & not nitpicking, but I noticed. I can see what you were trying to explain. Have a good day & hope you didn’t get caught in any storms, bit of a cracker yesterday afternoon around the area.