NorthWest Biotherapeutics Inc (NWBO)

[Lykiri]

Maybe this will help. On page three, the red box was added as a step in recent years. This must be what DI meant when he said they are checking 3,4,5 times. https://www.phusewiki.org/docs/2009%20PAPERS/DH02.pdf

flipper44,

thanks for sharing that paper: “Clinical Database acceptance: what statistical review checks are necessary to validate a database?”

These statistical checks can be performed after that all queries are solved, to see whether there is any other outstanding data issue that could have an impact on the statistical analysis, or could be done during the course of the trial at the same time as the production of queries, to identify the issues as soon as possible and get feedback from the investigators at the same time.

The statistical review can be done as an ongoing basis during the course of the trial. But the process of running checks and reviewing the outputs is generally time-consuming and may need a lot of resources. In addition to resources, the risk of review of data issues on an ongoing basis may lead to unblind the study treatments. The biostatistician needs to be careful to the summaries performed to maintain the blind of the trial.

CONCLUSION (HEADER 1)
Performing some statistical checks before the database snapshot / lock is recommended to enable any inconsistencies in the data that could induce issues in statistical analysis. It is critical to allow time for this step before database lock to get the database as clean as possible. The biostatistician needs to be involved in the whole process of data cleaning, to provide advice on checks and to take time to review the data on an ongoing basis.

The questions to have at the beginning of the process are: what type of statistical checks is needed on this trial? When do we need to perform these statistical checks? Could it be done just before database lock, or is there a real gain to do this on an ongoing basis, knowing that this review is time-consuming?

The biostatistician needs to stay vigilant to perform checks that will keep the blind of the data. Review of data during the course of the trial is always difficult and the biostatistician needs to be aware of the risk of looking at summaries of data.

Following this, the database should be provided to the biostatistician and statistical programmer in better quality!