Reineck, Lora (NIH/NHLBI) [E] <lora.reineck@nih.gov>
Wed 2020-05-20 09:43

I don’t think NHLBI (or NIH) has a policy about unblinded investigators attending committee calls. PETAL has generally avoided having unblinded investigators on committee calls as part of “good clinical practice”. 

Here’s 2006 guidance from FDA that I suspect you’re already familiar with and can/should be used as a guide (Guidance for Clinical Trial Sponsors - Establishment and Operation of Clinical Trial Data Monitoring Committees, pages 29-30): https://www.fda.gov/media/75398

Assigning the primary trial statistician the responsibility for interim analysis and reporting to the DMC can be problematic, however. When issues arise that might suggest changes to the trial design, a statistician performing both the primary collaborative and the interim analysis functions including reporting to the DMC, will probably be the only member of the trial management team with knowledge of the interim data. In considering possible changes, the statistician’s objectivity will inevitably be compromised by the knowledge of the potential impact of such changes on the outcome of the trial. When statisticians with knowledge of interim data participate in trial management meetings in which potential changes to study size, entry criteria, or endpoints are discussed, perceptions of biased decision-making could arise. Even if the statisticians remain silent about the interim data, it is essentially impossible for any opinion they may express not to be influenced by knowledge of these data. When the statistician is present for such discussions and knows which of the alternative courses of action is more likely to result in the experimental intervention being shown effective, even unintentional non-verbal communication may reveal (or may be perceived to reveal) some of that knowledge.

For this reason, when there is a DMC and a formal mechanism for interim analyses, it is advantageous for the statistician performing the interim analysis to be uninvolved in managing the conduct of the trial, especially in regard to making decisions about design modifications. This could mean that a statistician other than the primary trial statistician would take responsibility for the interim analysis and reporting to the DMC. Alternatively, the primary statistician could maintain this role, but forego further involvement in trial management once interim analysis was undertaken. Sponsors may identify and develop other approaches to reduce potential inappropriate influence of interim data on trial management.