Multiple IRBs, HIPAA concerns hamper project
Researcher wants IRBs to talk to one another, seek expertise on health services research
When a group of researchers set out to study effective ways to screen HMO members for substance abuse problems and to refer them for treatment, they expected that sensitive topic would require strict confidentiality measures.
They didn't expect nearly seven years of starts and stops due to concerns raised by three different IRBs that eventually handled the protocol. After securing approvals from two of them, they restarted the study in 2006, only to discontinue it in 2009, convinced that the IRB-mandated restrictions made the final project unworkable.
"By the time all of this was resolved and we restarted it and started looking at the data that was coming in, we realized that we had a protocol we were testing that could never be used in practice," says Lisaann Gittner, PhD, a research associate at Case Western University's School of Nursing in Cleveland, OH.
Her group's experience is detailed in a recent article in the Journal of Medical Ethics.
Gittner says she believes the problems that her team encountered point to gaps in federal regulations and IRB expertise in two areas – multisite IRB review and the field of health services research.
"The federal regulations are written for biomedical research," she says. "Many IRBs are biomedically focused. Some large academic institutions also have social/behavioral IRBs, but health services research isn't in either one of those categories."
She'd like to see the federal regulations changed to address those issues, and barring that, have IRBs make such changes on their own, to prevent problems like the ones she encountered.
Choosing screening candidates
Gittner's group began in 1999 with an idea to test a process that would screen people for alcohol problems or other substance abuse. Typically, a person is not asked detailed questions about substance abuse by his or her own regular physician, but only encounters them in the emergency department after an accident.
"These screens really should be done by a PCP (primary care physician) during a routine physical or regular visit," she says. "But PCPs are reticent to do it. They're managing three to four other problems and then to add this kind of sensitive discussion on top of it in a 20-minute visit? It's probably not going to happen."
Gittner's group proposed looking at utilization patterns within an HMO's membership, and picking out people who might benefit from alcohol screening. These people, along with a control group picked at random, would be invited by mail to participate in a phone survey that queried them about their use of alcohol and drugs.
It was an opt-out study, meaning that people would receive phone calls to initiate the survey unless they returned a form saying they didn't want to participate. During the phone call, a screener explained the survey to the participant, gained verbal consent and then transferred the participant to an automated system that delivered standard screening questions ("How many alcoholic beverages do you consume in a week?" "Has anyone ever told you that you have an alcohol problem?" etc.).
A participant could withdraw from the survey at any time by hanging up. At the end, those identified as requiring further help were offered three choices: Direct feedback from the automated system, screening results sent to a nurse manager for follow-up or results sent directly to the person's PCP.
A tale of three IRBs
Gittner says she knew the subject was a sensitive one and the study would require safeguards of participants' confidentiality. But she was unprepared for the problems she encountered at the first HMO the group approached in 1999. Its IRB tabled the protocol and sent it to the organization's legal department. Because of legal concerns, changes were requested, but after two revisions, the HMO withdrew its agreement to partner with the group.
The group approached two other IRBs over the next year, and encountered similar problems – issues raised by legal departments about confidentiality concerns, and requests for revisions.
The two IRBs' requirements were often contradictory, she says.
"They were asking for different things and then absolutely refusing to budge," Gittner says. "Whereas one IRB would say, 'You have to do B,' the other IRB would say, 'If you do B, we will not approve this.' And we couldn't get them to sit down to talk together."
Eventually, the researchers obtained approval from both IRBs, but just as data collection was about to begin, HIPAA Privacy Rule took effect in 2003, causing one IRB to raise fresh concerns.
"Nobody had understood what would happen with HIPAA, and so when it did happen, everybody erred on the side of caution," Gittner says.
In order to proceed with the study, she says researchers had to add a number of burdensome requirements. One was a Certificate of Confidentiality from the National Institutes of Health; the study was suspended for eight months until it could be obtained.
In addition to sending a consent form through the mail to participants, the initial phone screener had to read the form to the participant when calling him or her. The screener also had to read a HIPAA privacy notice.
This led to a problem retaining participants during the phone survey.
"By the time we finished re-reading the entire consent form to them – reading them all this stuff that we had already sent them in the mail – and then you hooked them up to the five-minute computer screening, many of them hung up," Gittner says. "At that point, it was already 20 minutes into the phone call."
A 2005 continuing review by one IRB noted the low response rates and the IRB agreed to a shortened informed consent. Getting final approval for the new consent and other minor changes took another six months. This took the group to January 2006 for the restart of the study.
Talking together, seeking experts
Gittner says that many of the group's problems could have been solved if the IRBs involved had been willing to work things out together, in person or by letter, without requiring the researcher to shuttle back and forth between them.
She also believes that IRBs don't understand health services studies, in which the emphasis is not on how an intervention affects an individual patient, but how a change to a health services system affects the use of the system.
"In systems-level research, we're going to do something to the process that you interact with. And what happens to you is not as important as to where does the process go," Gittner says. "How many of these (substance abuse) screens are getting done? How many people are going into treatment? Are we effective in screening the right kinds of people?"
She recommends that when an IRB is presented with a proposal for this type of research, it should bring in a health services researcher as an outside expert.
Ultimately, Gittner would like to see federal regulations requiring such expert opinions, along with regulations that would designate one IRB to take the lead in a multisite study.
In the meantime, she says, experiences such as hers are causing health services researchers to rethink proceeding with studies based on concerns about IRB review.
"A lot of people are now not doing certain streams of research because they think this is going to happen to them or because something like this did start to happen to them and they pulled back," Gittner says.
- Gittner LS, Roach MJ, Kikano G et al. Health service research: the square peg in human subjects protection regulations. J Med Ethics 2010 Nov 11 (Epub)