If there are concerns about safety/adverse events and human subjects are involved the trial should/must be monitored by a Data Safety Monitoring Board (DSMB) and masking/blinding of that DSMB is a separate issue from blinding study subjects, investigators, and study personnel that collect outcomes. Depending on whether it is about a drug, a device, or a behavioral intervention, true blinding may or may not be feasible. Even if complete blinding is not feasible it is a good idea to be restrictive with whom knows what.
E.g., in a behavioral intervention, the interventionist and the study subject may know group-assignment but the principal investigator, those evaluating adverse events, and those who collect outcome data do not have to know about the assignment specifics and participants can be asked not to reveal the intervention they received. The "ostrich policy" can be adjusted to the needs of the trial but will typically require a well worked in team where all team members exercise and are expected to exercise the integrity needed to make such an approach work (in reality, e.g., the principal investigator (PI) might be blinded and is also the supervisor of the statistician or other study personnel that is not blinded: in such a constellation it is absolutely essential that the PI respects that these subordinates (temporarily) withhold information from the PI in the interest of the scientific integrity of the trial).
DSMB: DSMBs regularly feel that information about group assignment (other than Group A and B, etc., without revealing intervention vs. control status) helps them to assess adverse events or loss-to-follow-up problems. DSMBs often start out being blinded but may request to become un-blinded at any time during the trial. Personally I have the opinion that, if lifting the blind is considered, a DSMB should first discuss what their recommendations are when remaining blinded and to separately discuss what their recommendations would be if un-blinded (assuming a scenario like Group A = active intervention or Group B = active intervention). Only if there is a potentially appreciable difference between the recommendations there is a potential case to be made for un-blinding. My experience is that it is not unusual that a DSMB's action is "watchful waiting" whether blinded or un-blinded, thus there is no case for lifting the blind at that point in time (which typically cannot be "undone" for the remainder of the trial). However, irrespective of my personal decision-oriented view, as the biostatistician on a specific trial I will not question a DSMBs decision to become un-blinded once that decision is truly reached and I will promptly lift the blind upon a direct request by the chair of that DSMB as I feel that my personal opinion does not matter in that specific situation. If the DSMB feels they should become un-blinded I will simply assist them in that quest as they are charged with supervising that trial; I will request, though, that the un-blinding is recorded appropriately, e.g. in the minutes of the DSMB meeting. (Just to be clear: Lifting the blind for the DSMB does not imply that any other personnel will become un-blinded as well - these are separate decisions/issues.)
Fridtjof
------------------------------
Fridtjof Thomas, Ph.D.
Associate Professor Division of Biostatistics
Department of Preventive Medicine
University of Tennessee Health Science Center
fthomas4@uthsc.edu------------------------------
Original Message:
Sent: 04-21-2017 07:46
From: KyungMann Kim
Subject: open-blind trials
I agree whole-heartedly with David Bristol. Despite the claims argued by some sponsors that they can manage the safety monitoring better, there isn't much to be gained from being "open" about the treatment arms. Treating OL trials as blinded only protects sponsors from making mid-course changes that are probably data-driven. This is simply good clinical trial/statistical practice.
KyungMann Kim, PhD
Professor
Department of Biostatistics & Medical Informatics
University of Wisconsin-Madison
256A WARF Office Building E-mail: kyungmann.kim@wisc.edu
610 Walnut Street Phone: (608) 265-6380
Madison, WI 53726-2336 Fax: (608) 265-9768