P114: Introducing the Independent Safety Analysis Team (iSAT): a Sponsor-Employed, Fully Unblinded, Independent Analysis Team
Poster Presenter
Joseph Scherer
Expert Biostatistician
Boehringer Ingelheim United States
Objectives
We present a model where an internal team carries out planned (and unplanned) unblinding activities. Initially focused on anticipated events reporting, it quickly was used for DMCs and IAs as well. Based on 5 years of experience, the advantages and disadvantages of this approach are discussed.
Method
The independent Safety Analysis Team (iSAT) was established in 2017 to primarily support Pharmacovigilance for reporting of anticipated events and has grown to handle other unblinded activities such as DMCs / Interim Analyses (IA). The team consists currently of 2 statisticians and 4 programmers.
Results
A small team of experts was recruited to setup and define a process which allowed for unblinded data handling of ongoing trials in 2017. Firewalls - both physically and electronically - were defined and established to ensure that only the iSAT has access to unblinded information:
1) By default: no read access to any ongoing trial data except to those iSAT is explicitly assigned to
2) Use of separated, secure folders for each study with access rights controlled by the iSAT – nobody outside iSAT has access to these folders
3) Hierarchical separation from all study members organizationally and physically by either working remotely or maintaining a separate office space for iSAT members only
4) Minimization of communication between study team members and iSAT once iSAT is unblinded for a study
For unblinded, anticipated events analysis – which potentially includes pivotal trials - iSAT only receives a limited set of data explicitly excluding any efficacy data. For IA requests, the blinded trial team will program the outputs and send these programs to the iSAT per agreed-upon timelines to maximize efficiency. Then iSAT unblinds the data and runs these programs in the secured environment described above. The summary results are then shared with those who have access rights according to the Logistics and Access Plan which is approved prior to any unblinding activities.
For DMCs, the iSAT acts as iSTAT and iPROG and participates in the closed sessions of the DMC. Outputs are usually based on source data (instead of SDTM). This permits very aggressive timelines of less than 2 weeks from data snapshot until the DMC meeting, if required. iSAT also handles the programming of additional requests which can be turned around much quicker (sometimes even during a DMC closed session) in a confidential manner.
The iSAT model has been successfully applied to over 15 trials with either IND safety reporting, a DMC or an IA with very positive feedback from all parties involved.
Conclusion
In randomized, double-blind clinical trials, it is occasionally necessary to unblind the trials for the purpose of:
1) interim analyses for internal decision making
2) Data Monitoring Committees
3) IND safety reporting
These activities are often outsourced to groups external to the company to allow personnel responsible for the monitoring of the trial to remain blinded throughout study conduct. Regulatory guidance for DMCs discuss having the independent statistician for DMCs being an internal representative from the company. In general, the guidelines caution against this model although the language stresses the need to remain fully independent for pivotal studies. Except for anticipated events, the iSAT model is generally only applied to non-pivotal studies.
Having these activities supported by an internal group has many advantages, including:
1) The iSAT are expert programmers and statisticians with many years of experience working as trial and project statisticians and programmers across multiple therapeutic areas
2) Direct consultation with the trial team thus no need for lengthy contracting discussions or complicated communication lines
3) The internal team is familiar with company procedures and guidelines
4) No transfers of data to external parties
5) Flexible working arrangement – no need for discussions on contract updates or costs associated with extra measures, thus overall improving the quality of the output for a DMC meeting
6) Quick turnaround times – no need for SDTM datasets or discussions with vendors about increased resources needed for a particular DMC
The only disadvantage of this approach is the potential negative external perception of an internal team having access to unblinded trial results. However, by limiting the studies the iSAT works on to non-pivotal studies this perception is not a concern. The iSAT has the responsibility of protecting the blind from the study team and takes this role very seriously.
