'The Censor's Hand'

Many scholars complain that institutional review boards unreasonably delay or block studies involving human research subjects. A new book, The Censor's Hand: The Misregulation of Human-Subject Research (MIT Press), doesn't see IRBs as fixable, but as an inherently flawed system. The author -- Carl E. Schneider is Chauncey Stillman Professor of Law and Professor of Internal Medicine at the University of Michigan -- responded via email to questions about his book's themes.

Q: Is the problem that IRBs are overzealous or (as you suggest) that the entire system is messed up?

A: The problem is that the IRB system is so fundamentally misconceived that it is virtually a model of how to regulate badly. Good regulation is accountable, but IRBs are effectively answerable to nobody. Good regulation has clearly defined jurisdictional limits, but IRBs may intervene as they wish. Good regulation is guided by clear rules, but IRBs have little more than empty principles. Good regulation is disciplined by fair procedures, but IRBs can ignore every fundamental precept of due process. Good regulation is transparent, but IRBs need not even explain -- much less justify -- their decisions. Good regulation is staffed by experts, but IRB members cannot be competent in all the specialties they regulate. Good regulation has manageable workloads, but IRBs regulate more details of more research in more ways than they can review responsibly, and they have steadily broadened and intensified their hold over research.

In short, the IRB system makes unreliable decisions because it is lawless and unaccountable, because its organization, procedures, membership and imperialism are so inappropriate. The problem is not regulation, it is bad regulation.

Q: Why do you believe IRBs have become censors?

A: IRBs have become censors because they have been told to censor. IRBs may decide what questions researchers can ask, how to ask them, how to analyze answers and how to report findings. IRBs' incentives lead them into the classic censors’ faults -- like constraining too much for too little reason. If IRBs say no, little harm can come to them: researchers may be dismayed, but they cannot afford to alienate regulators who have unreviewable authority over their work. If IRBs say yes, they risk blame for trouble that (justly or not) the research provokes. This trouble can be painful, including institutional embarrassment, lawsuits and federal sanctions.

Q: In recent years, social scientists have been among the vocal critics of IRBs, saying that their research needs a different kind of review than biomedical review. Does your critique apply here?

A: IRB review of social-science research suffers from all the problems of fundamentally misconceived regulation that I just described. But those problems are exacerbated by the fact that the IRB system was designed for biomedical research. Thus IRBs too often ask questions and impose restrictions ill suited to social-science research. In addition, many kinds of social-science research are ideologically controversial, tempting IRBs to use their power to promote their own preferences.

Q: Many scholars complain about IRBs as delaying their work, a bureaucracy, etc. But you say more than this -- that they are hurting science. Would you summarize your evidence?

A: Scholars rightly say that the burdensome bureaucratic processes that IRBs have come to demand and the restrictions that IRBs impose delay, distort, discourage and even prevent research. These impediments reduce the amount of research that can be done, diminish its quality and postpone its benefits. The Censor's Hand spends a chapter detailing the evidence that these costs of the IRB system are high. For example, reports of researchers' experiences with multisite studies indicate that just dealing with IRBs can consume 15 percent of a study's budget. Those reports also show IRBs responding to a single protocol in bafflingly inconsistent -- and sometimes incomprehensible -- ways. Nor is there good evidence that IRBs actually improve research or protect its subjects.

Q: The United States was home to the Tuskegee experiments, and even today, some researchers are accused of doing research abroad that would never be tolerated with Americans. Is there some need for IRBs, even if not in their present form?

A: The response to one social disaster should not be to create another. Yet a social disaster is what the fundamentally misconceived IRB system has turned out to be. Before regulating, government should ask what problem regulation addresses and whether regulation can do more good than harm. Most research -- both biomedical and social science -- poses risks no greater than those people handle routinely without government supervision. Furthermore, research varies enormously, so human-subject research should not be treated as posing one set of problems and requiring one kind of solution. So we should ask which specific threats research actually poses and then identify the best way to handle each kind of threat.

Q: What system would you have to protect research subjects?

A: One approach to that question is to ask how we protect patients from doctors, who (for lack of adequate evidence from research) must daily treat patients with unproved methods and whose misjudgments yearly kill and injure far more people than research has in all its modern history. We do not make doctors get prior approval for every treatment from a doctor review board. Another approach to that question is to ask how we protect the public from journalists, who daily invade privacy and damage reputations in ways researchers would find wholly unethical. We do not make journalists get prior approval for every interview question from a press review board. In the case of both doctors and journalists, we recognize that the harm that regulation did would inevitably exceed the good, just as the harm of the IRB system inevitably exceeds its benefits. We regulate doctors and journalists as we regulate most people, by punishing wrongdoers (administratively, criminally or through private lawsuits). Sanctions can already be visited upon researchers, and if more sanctions or harsher sanctions are needed, they can be supplied.