Interim and final analysis in clinical trials

[ask.biostat]

Hello,
I would like to know the opinion of biostatisticians with experience in clinical studies with adaptive design.
I participate in a clinical trial with adaptive design as a main biostatistician. The trial is double blinded placebo-controlled study. My responsibility is to provide final statistical analysis plan. There is one interim analysis of main efficacy endpoint in this trial. The purpose of the interim analysis is to provide basis for early termination of the study due to excessive efficiency, or to provide basis for recalculating (increasing if necessary) the sample size. The statisitical analysis plan for interim analysis will be provided by an independent biostatistician. As a main biostatistician, I can’t see the interim statistical analysis plan and can’t have any information about details of the interim analysis. My question is: should an independent biostatistician review the final analysis plan to ensure that the planned analysis at the interim stage is consistent with the planned analysis at the final stage? If not, how will it be appropriate to ensure such exact correspondence? If the study will be stopped due to excessive efficiency, the final analysis of main efficacy endpoint should provide exactly the same results as the interim analysis results.
I will be grateful for any useful information on this issue.

 

[hlsmith]

Hmm, havent thought about that. Probably depends on if they are examining effect or risk. If risk of adverse event is low and they are examining if at a midpoint the primary endpoint can be meet, etc, yeah i would imagine they should. If it does not have to do with the primary outcome, but an adverse risk, maybe not.

Good question.

 

[ask.biostat]

Dear hlsmith, thank you for your response!
This is a study of phase 3, main purpose is to assess efficacy. A phase 2 study showed that the study drug was reasonably safe — a very small number of adverse events with low severity. So, premature termination is not expected due to unsatisfactory safety of the investigational drug. And such a scenario is not planned.
My goal is to ensure complete matching of the methods for analyzing the primary endpoint in the interim analysis and in the final analysis. So far I have 2 options, and both of them are not very happy with me and are not completely clear to me. The first is that an independent biostatistician can review the final SAP, but then he, in my opinion, becomes not entirely independent. The second is to include in the study another independent biostatistician, only to provide this matching, but is there not too many independent biostatisticians in this case?

 

[hlsmith]

Are you referring to say a Data Safety Monitoring Boards or a version of it? If study treatment is randomized, I would imagine the study analytics will be very rudimentary by the DSMB, as long as they know what you all are deeming a significant effect size/alpha, etc. at the end.

I did not follow what your two options were, both just read as though there would be two biostatisticians running final study analytics on the primary outcome. Please add more details to distinguish the scenarios.