IRBs and scientific validity (August 12, 2005)

This page is moving to a new website.

One of the regular posters on the edstat-l mailing list shared a comment that is commonly held among many of the people I have talked to.

IRB's should NOT be in the business of trying to make studies good. IRBs are not sufficiently experts in all the different kinds of proposals that come across their desks. They are there to make sure that the principles of informed consent and lack of coercion ... have been faithfully taken care of.

This comment is a reaction to the tendency of some IRBs to nitpick. While I am somewhat sympathetic to overly intrusive reviews that we all have to endure, it is important to remember the underlying rationale for IRB review of scientific validity.

Suppose the research study involves collection of cerebrospinal fluid through a painful procedure known as lumbar puncture. If I were going to participate in such a study, I'd like to have some assurance that the study was "good." And I wouldn't want to trust the researchers themselves. Far too often they have a financial stake in the research or they are so enthusiastic about their work that they lose a sense of perspective. In my mind, some sort of external review of the quality of the study would be mandated anytime human subjects are involved, and the level of that review should be commensurate with the risks that I am undertaking.

It's hard for an IRB to find the right balance between making sure that the research is of sufficient quality to justify the risks and inconveniences that a research subject has to endure. Too much review and too little review are both bad.