Question from Andrew Spieker 2021-07-17

I am involved in a multi-center randomized trial led by Dr. Natasha Halasa (CC’ed) and we are seeking to understand the point at which it becomes appropriate to unblind ourselves to the vaccination groups so that we can start analyzing and reporting the data. We hope you may be able to provide us some guidance given your experience and expertise in conduct and analysis of trial data.

Here’s a brief status of what has and hasn’t been done:

• All participant study visits have been completed.
• The data/sample collection has been completed.
• The statistical analysis plan has been finalized and approved by the FDA.
• The table shells for submission to clinicaltrials.gov are nearing completion in a blinded fashion (stratified as A/B without knowledge of what A/B are).
• The lab samples for the first three years of data (primary analysis) have been analyzed and returned to us.
• The lab samples for the fourth year of data (a smaller group consisting of subjects who re-enrolled from year three – intended for an exploratory analysis) have been sent for analysis and have not yet been returned to us (and we have no real sense of when they will be).

If we can avoid it, we don’t want the lab samples from Year 4 to hold us back from moving forward with the analytic plan and preparing the main manuscript (which will only feature data from Years 1 through 3). Since data for all primary analyses are cleaned and processed, we would like to be able to lock those data and schedule a meeting with our DSMB to obtain authorization to unblind ourselves and move forward while the Year 4 data are still being processed. Do you think this is appropriate at this stage? We’d love to hear your thoughts – we’re trying to make sure that we’re going about this in a way that does not compromise the integrity of the trial.

Answer from Chris Lindsell 2021-07-19

I don’t believe there is a singular moment in a trial that is the only time it is right for unblinding. Generally it occurs at the time of data lock for the main analysis. Ancillary studies may be ongoing, just as your fourth year of data collection, but this ongoing data collection needn’t undermine your main analytical approaches. The key is that any future data collection must occur blinded to assignment (keep your lab people away from treatment assignment information). Then, it is ok to lock the main trial data and report out given that all of your main data are collected, your main outcomes have been measured, and the main data have gone through their cleaning processes.

The one possible concern would be that analysis of year 4 data is guided by knowledge of what happened in earlier data on the same participants. As long as analytical plans are always pre-specified, future data are collected blinded to treatment assignment, and there is transparency in how those data are managed, cleaned and analyzed, then you should be good to go.