Citizen science: a new frontier for IRB review
How IRBs can approach citizen science studies
While private companies that conduct human subject research do not fall within the definition of regulated research, the decision whether to entirely forgo approval is not always clear. Though some "citizen science" research can be as simple as gathering data from surveys, some can involve a person’s self-reported health data — and can raise concern in the research world.
Citizen science is the concept of conducing scientific research, usually by amateur or nonprofessional scientists, by way of crowdsourcing. Though not a new concept, it has gained some public interest in recent months, particularly in the research and IRB worlds through start-up company uBiome. The California company created a crowdfunding program to raise money to develop kits to sample a person’s microbiome to learn more about the "good" and "bad" bacteria in the person’s body and how it correlates with the microbiomes of other contributors. For a fee, a person can pre-order a kit, take a swab from his or her gut, mouth, nose, genitals, or skin, and send the sample back to uBiome for analysis and correlation.
Controversy arose among the research ethics community about the project: While uBiome is not required to seek IRB approval, does the company have a moral obligation to gain IRB approval, since health-related human subject research will be conducted? Are consent forms thorough, understandable and reviewed? How will participants’ biological information be used — and how will it be kept private?
"There were a series of questions [about the project], but the primary question is one I would take as a derivative of the therapeutic misconception issue — when does a person go from being a client to being a subject?" says Erica Heath, director, regulatory affairs, Ethical & Independent Review Services, the IRB that approved the uBiome study. "Where is the differentiation, if there is one? This crowdsourcing for funding is a new idea in research, and it seemed to be nowhere close to as crisp as you might get in medicine."
"We wanted to do this in a way that is thoughtful, and take good care of our participants," says Jessica Richman, CEO of uBiome. "We’re all from academic backgrounds and we wanted to get approval. We want to do this ethically, but what is practically, morally, and structurally the right way to do this?"
Richman says they spoke to a few different IRBs before finding one to work with them and "hold our hand" through the approval process. "It was a little hard to find someone to work with us," she says. "They’ve never seen this situation before and didn’t know how to approach it."
Defining research
When the uBiome team first met with Heath and E&I Review for IRB consultations, Heath helped them outline what the research is and what they are trying to accomplish, and the obligations that come with the research, elements of consent forms, and the consent process. "It’s that way with everybody who is a novice, and we get novices a lot," Heath says. "It’s a teaching and learning environment. How do you write something simple and understandable? It goes back to defining what the research is — the research will define consent."
Heath began with the decision tree charts from the Office of Human Research Protections (OHRP), taking Richman and the uBiome team step by step through the decision tree to determine whether the project is research involving human subjects, if any exemptions apply to the research, any privacy and consent form issues, and if expedited review applies. "With each of those came more duties," Heath says. "The first issue we talked about a lot was research and what it was they would be doing. Once we get beyond that, what are the obligations that come along with doing research? We went over the consent process and the elements of consent and how to get them across formats. Defining the research went a long way toward creating what she [Richman] sent to the board," Heath says.
Client vs. participant
The next issue, Heath says, is to identify what a participant will get out of the study once he or she becomes a research subject as opposed to being a client. "It’s unlike an investigational drug, where you get that drug," she says. "What are you getting here, and what are the risks of transition [from client to subject]? If you’re just getting your own report back, that’s one thing. If this is a long association study and can be compared to association studies, there are risks of incidental findings and such."
E&I Review Services was very focused on potential social and psychological risks of the uBiome project, Richman says. Those risks include subjects finding out a result about themselves they wouldn’t like; subjects getting a result they may not like but isn’t a health risk; or other people finding out something about the subject. There was not much concern about physical risks, she says. "The test is not diagnostic; we can’t tell you whether you have a disease," Richman explains. "The main concerns were mostly risks involving information."
There is also a question, Heath says, of reliability of the data associations, and whether there will be enough data to make the study statistically meaningful. "When you start with no data points and you have participants wanting information to interpret their personal results, when is the data enough to be meaningful? This is a statistical question that bedevils scientists."
One of the concerns with the uBiome project is the fee involved, as research subjects typically do not pay to take part in a study. "It’s pretty rare, though not unheard of, to have subjects pay for study participation," Heath says. She cites costs of smoking cessation plans, though participants earn money back through completion. "The next issue was, once a client becomes a subject, what is it they are getting out of the study by being a subject as opposed to being a client?" For example, a subject in a pharmaceutical or therapeutic study receives the experimental drug or control treatment. "Part of the distinction between client and subject was when you make that distinction, you can see what the difference is," Heath says.
Privacy concerns also topped the list. Richman says uBiome focused on privacy from the beginning, and participants can opt out of the data correlation if they desire. "No one else knows your data," she says. "You can have it correlate with other people in the dataset anonymously if you want, and you can opt in to the research study with the IRB." The data correlation will be done by uBiome.
"In any genetic study or association study, you have to look at it in two ways — what is private about it and what harm the data could do, and how is it being protected," Heath says. "Both of those aspects are looked at. What’s there and how harmful is it, or could it be? How is it being protected?"
How IRBs can approach
As these types of crowdfunded citizen science studies are relatively new, IRBs may have questions on how to proceed if approached for review. Currently, IRBs have little to no experience with citizen scientists, Heath says. "The citizen scientist job descriptions range from data gatherers to amateur investigators. uBiome was asked to define and separate their more rigorous studies from these amateur studies over which control was not yet defined," she says. "This could well be an IRB topic for the next decade."
To begin looking at the review process, "I would hope any IRB would start off with definitions and work through what it is they [the researchers] are doing, what’s generalizable about it, how it becomes research, who the subjects are, and what they’re doing," Heath says. "Then you can work through the determinations and consent issues. This is generally a good approach. Again, with different clients you’ll find stopping points at different places, but the approach is a failsafe."
"We [IRBs] have been through all kinds of new things. We’ve seen genetics research, stem cell research, genome sequencing," Heath says. "This is another new thing on the horizon. It’s fun to watch. That’s what IRBs are all about — if these were all easy, we wouldn’t need IRBs."