Central IRBs: Consistency increases, but so does confusion
Pros and cons are weighed
As the use of central IRBs grows nationally, these models are increasing consistency in IRB review, but they also are causing some confusion for institutional IRBs, research ethics and IRB directors say.
"Centralized IRB models can increase the consistency of human subject protection," says James Riddle, MCSE, CIP, CPIA, assistant director of Fred Hutchinson Cancer Research Center in Seattle.
"Having one location where the research is consistently reviewed helps ensure consistent protections apply across all of the institutions doing the protocol," he adds.
Research institutions increasingly are awarded federal grants that require a centralized IRB model out of a desire for greater consistency, Riddle says.
One of the driving forces behind the centralized IRB model was the perceived lack of IRB review consistency between sites reviewing the same multisite study. But now that centralized models have become more accepted and consistency is improving, a new administrative problem has arisen, Riddle says.
In recent years, research institutions working with central IRBs have discovered that the efficiencies gained from having a multisite study use one main IRB have not trickled down to the local IRBs.
"Just because you’re using a central IRB doesn’t mean it’s less work for the IRB office," Riddle says.
"You still have to keep track of the central IRB and what the rules are," Riddle says. "So you might improve study time and efficiency of the overall IRB system, but you may also end up adding administrative burden to keep track of all of those different centralized IRBs and their unique requirements."
For larger research institutions like Fred Hutchinson Cancer Research Center, there often are three or more central IRBs being used at one time, Riddle says. (See story about different central IRB models, page 112.)
"Each institution has to come up with their own mechanisms for working with central IRBs," he says.
The human subjects protection world is in transition with centralized IRBs, and while this makes things more confusing, it will evolve and improve, says Kathryn Flynn, PhD, an associate professor at the Medical College of Wisconsin in Milwaukee. Flynn has studied the central IRB model and published research about it.
"Right now, in this transition period, it might mean extra paperwork for IRBs, but hopefully this is only for the transition period," Flynn adds.
The ideal is for both local and central IRBs to agree on terms and a workable model, she says.
When the first centralized IRBs emerged in the 2000s, institutions and local IRBs were concerned about whether involvement of central IRBs would affect the local and institutional IRBs’ ability to reflect community standards, but this has not proven to be a problem, says Robert Klitzman, MD, professor of psychiatry and director of the bioethics masters and online certificate programs at Columbia University in New York City.
"Centralization has a lot of advantages, and what I’ve found in research is that there are several advantages to having local IRBs, as well," Klitzman says. "One advantage is that local IRBs have local knowledge of subjects and researchers."
For instance, members of a local IRB might have a curbside consult with investigators, he says.
"An investigator might see an IRB member in the parking lot or hallway and say, I’m thinking about doing this with my study, what do you think?’" Klitzman says. "The IRB member might answer, Make sure you’re aware of XYZ.’"
The researcher takes that informal advice and incorporates changes in the protocol before sending it to the IRB for review. That informal process can save a month or more of IRB review time and back-and-forth between investigators and the IRB, Klitzman explains.
"A parking lot is not a place for an official decision, but any informal network can be helpful," he adds.
The research community can do more to improve local IRB and central IRB relationships and to increase efficiency when centralized IRBs are used, Flynn says.
Flynn was the lead author of a study that looked at perceived barriers to using centralized IRB review and discussed the need to clarify terms.1
"Ten years ago there was a conference that outlined main barriers to centralized IRB review, and we were trying to present some potential solutions to these barriers," Flynn says. "We talked with a number of different stakeholders, received their feedback, and presented solutions to barriers."
Flynn and co-investigators spoke with more than 40 experts, including IRB directors, federal regulators, and others.
"We held a two-day expert meeting to get all of these different perspectives," Flynn says.
One barrier identified was the feasibility of working with multiple outside IRBs when each required different forms. The solution Flynn and co-investigators suggested was for the research community to identify standard data elements to facilitate review and reporting across disparate systems.1 (See story about IRB responsibilities on this page.)
The centralized IRB model still is new, so it’s too soon to say whether it’s the best model for human subjects research review, Flynn says.
"Theoretically it looks like it should be the better model, but it remains to be seen," she adds.
When research institutions form a partnership with a central IRB, the human subjects protection community should be cautious of expecting too many benefits from this relationship, Klitzman says.
"The problem is that people are looking to central IRBs to solve significant discrepancies in IRBs’ decision-making," he notes.
But a central IRB is not the answer to all IRB standardization problems, he adds.
Klitzman’s research has shown that discrepancies in IRB reviews can occur within a local institution and even within a single IRB, as well as between IRBs that are entirely different. This suggests that most IRB review inconsistency is due more to individual board members’ opinions than to differences in community standards, as is usually suggested, Klitzman says.2
IRB review is a human process, and this makes consistency more challenging, he says.
While CITI and other human subjects training are widely used by researchers, clinical trial staff, and IRB workers, such training is rarely required of actual IRB members, Klitzman says.
"Why isn’t there a required test to be an IRB member?" he says. "If researchers have to take a test, IRB members should take a test."
The field’s dilemma is similar to that of psychiatric research 30 years ago, when one psychiatrist would say a patient is depressed, but another would think that was overstating the problem, he explains.
"So the psychiatric field developed standardized rating scales that tell us if someone is depressed or not," Klitzman says.
The same sort of rating scales could be developed for the IRB process, he adds.
"What we need is open, transparent discourse," Klitzman says. "It’s in the absence of this that people believe central IRBs are the answer."
Making the IRB process more consistent would require both a checklist that IRB staff can use to make certain each submitted study follows all regulations, and a review rating scale to look at issues where personal opinions can sway a decision one way or the other, he says.
An IRB review rating scale could look at the informed consent process, the concept of undue influence, the balance of risks and benefits, and other issues.
"I would argue it’s still a human process and we can’t automatize everything, but there needs to be much more of an attempt to do so," Klitzman says. "The differences between IRB reviews are not related to community values but are due to idiosyncrasies, and that’s an unacceptable reason for inconsistency."
- Flynn KE, Hahn CL, Kramer JM, et al. Using central IRBs for multicenter clinical trials in the United States. PLoS One. 2013:Jan. 30, 2013.
- Klitzman R. How local IRBs view central IRBs in the US. BMC Med Ethics. 2011;12:13. n
