IRB Independence

I recently participated in a discussion on the IRBForum regarding the involvement of the Institutional Official (IO) in the policies and procedures of the IRB. This prompted me to expand on my thoughts here.

Many people believe that the IRB must be a completely autonomous body that operates without any interference from the institution. I not only believe that this is a mistaken interpretation, but it is also dangerous and puts subjects at risk. The independence required for the IRB to fulfill its mission is independence in decision making. In Sections 112 of 45 CFR 46 and 21 CFR 56 the regulations state that institutional officials cannot approve research if it has not been approved by the IRB. Thus maintaining the independence of IRB decisions, at least negative decisions. This is not saying that IRBs are completely independent from the institution. In fact, Section 112 also says that research approved by the IRB can be subject to further review by the institution. The assurance process in 45 CFR 46 gives the responsibility for establishing policies and procedures for the protection of human subjects to the institution. Therefore, it is the institution, generally through the IO, which sets policy, not the IRB. So, the belief that the IRB is completely independent is mistaken.

Why is this belief dangerous? Because a completely independent IRB runs the risk of having an adversarial relationship develop with the investigators. Without some oversight over its policies and procedures, the IRB can wind up losing sight of the burden that its requirements place on investigators. When that happens (and I have seen it develop several times as a consultant), an adversarial relationship can develop between the IRB and the investigators. This type of relationship increases the chances that investigators will not be forthcoming with the IRB or will even conduct research without IRB approval. This is when subjects are put at risk.

This is not to say that IRBs should become "rubber stamps" or that institutional polices and procedures should be lax. Rather, that the institution has the responsibility to ensure that the policies and procedures of the IRB are reasonable. As I said in the discussion on the IRBForum, the IRB needs to follow the principle of beneficence, similar to that in the Belmont Report, in its relations with investigators. It needs to make sure that the benefits of imposing requirements on investigators (primarily in terms of protecting subjects) justify the cost (in time, money, paperwork, etc.) to the investigators.

So, I do believe that the IO must be involved with the way that the IRB performs its function. The IO cannot and should not intervene in IRB decisions, but should oversee the policies and procedures that govern how the IRB functions and ensure that they balance the need to protect subjects with the needs of the investigators.