“Research” as not-Therapy (or: Why “research” is dangerous and must be controlled): An IRB origin tale

In his comment on my earlier post, Kerim wrote that it would be interesting to see how IRB regulations work for “examples which are supposed to be paradigmatic” (medical research). I agree. Doing so really exposes the fundamental impotence of the rules (despite mighty efforts by droves of well-intentioned, clever rule-writers). While ethnographers and biomedical experimenters do face analogous disconnections between regulatory prescriptions and the inevitable surprises of actual research practice, exploring the problems that medical folks in particular face brings us back to the regulatory system’s “origin place”. It brings us back to the problem that the rules were meant to solve but never did.
I’ll summarize the argument—part of the forthcoming PoLAR article, “Comparative ‘Research’” (mentioned in earlier posts)—versions of which also appear less centrally in my 2006 AE paper.
I begin with the observation that there is no such thing as Research-in-general: there is no suite of ideals or traits shared by all the practices to which the term “research” usually refers. Instead, there are diverse, historically specific, socially organized ways of knowing the world, in shifting partial relations with one another, as well as ambivalent relations with particular non- and quasi-research activities (from journalism or travel writing to stage magic or, as we will see below, therapy).
Bureaucratic regulation cannot tolerate such incoherence. To control human-subjects research, variation must be bracketed so as to create a stable object with clear boundaries. Criteria must be found for deciding when instances of “research” begin and end, and for deciding what is and is not “research”.
However, research involving human beings persistently resists reductive objectification in the interest of consistency. This is true not just among the ornery humanistic social studies but also in biomedicine and other unambivalently science-identified fields whose research models fit the Common Rule definition.
So why bother? Why did “research” with human participants come to be viewed as in need of regulation (and therefore agonizing definition)?
The IRB definition of research was designed originally to provide a means of eradicating a potentially lethal ambiguity, specific to medical practice, concerning when “therapy” ends and “research” begins. This danger, publicized in the 1960s and early 1970s, is central to the history of the US federal regulations (for lots of bibliographic leads, see that November AE again).
But the design was flawed: the original ambiguity is still with us.
Human-subjects research regulations (45 CFR, Pt. 46, known as the Common Rule: see Chris Kelty’s Savage Mind 2/8/06, on which I hope to comment soon!)—formally set in place in the late 1970s/early 1980s—derive from earlier NIH guidelines that reflect specifically biomedical background assumptions. These assumptions go something like this:
1. Since it concerns directly improving the wellbeing of particular individuals, medical therapy is appropriately evaluated in terms of individual patient interests.
2. In contrast, since medical research concerns the production of knowledge “generalizable” beyond individual cases, it is appropriately evaluated in terms of the interests of science and society at large.
3. It follows that—since a concern for individuals isn’t inherent to the definition of “research” the way it is to the definition of “therapy”—even though physical risks to persons are associated with both medical therapy and research, the risks to individuals are qualitatively more serious in research
4. Consequently, “research” (in this sense) needs special oversight.
These specific background assumptions are evident in regulatory provisions that, firstly, limit risk/benefit calculi to research “emas distinguished from risks and benefits of therapies/em” and that, secondly, exclude from consideration the “long range effects of applying knowledge” (45 CFR 46.111[a][2]). “Research” is not “therapy”: not the emapplication/em of medical knowledge to individual persons (applications normally enabled when research results are published and made available to practicing physicians). The medical rationale of regulatory distinctions is just as evident in the definition of “human subject”, but I won’t discuss that here.
Now, while Common Rule regulations were applied widely to federally-financed non-biomedical research from the start, the idea of research-as-not-therapy does not translate readily outside of biomedicine. For example, cancer patients participating in clinical trials of new cancer therapies can compare the likelihood of a direct personal benefit from new knowledge against the risk of harm in the research process itself. In contrast, social research rarely promises practical payoffs to its participants. Its benefits tend to be indirect and its harms are located less in the research process than in its products.
Not only is research-as-not-therapy untranslatable across fields, but the regulatory definition did not even do away with the ambiguities within medical practice. Heroic efforts of clarification can be found in works that interpret the Common Rule for IRBs, like the IRB Guidebook (www.hhs.gov/ohrp/irb/irb_guidebook.htm). Nevertheless, to this day it continues to be a frequent topic of debate in IRB circles.
I’ll stop here for now. In the full length treatment of this argument, I offer an extended example of problem solving discussions among experienced biomedical IRB insiders concerning how the medical therapy/research distinction might be operationalized in particular cases. Such discussions—which happen all the time—demonstrate very clearly that it is effectively impossible in practice to distinguish clearly between when “therapy” ends and “research” begins: when innovations in the care of a particular patient with an unusual condition ought to be treated as research (worth publishing, in need of an IRB review). The regulations offer no solution to this problem.
More soon.

I'm Professor of Anthropology at Princeton University. I've blogged here about the disconnection between ethnographic field ethics and ethics regulation compliance, and advocated hither and ton for humanistic social studies in regulatory ethics contexts. Otherwise, I've done fieldwork in Highland Papua New Guinea (sociality, gender, exchange, historicity) and on knowledge cultures in the US academy (anthropology, historiography, sociology, experimental psychology).

2 thoughts on ““Research” as not-Therapy (or: Why “research” is dangerous and must be controlled): An IRB origin tale

  1. a good example of the therapy/research enigma from my own IRB experience: one of the longest conversations I’ve ever experienced concerned the question of what should be done if, in the course of a standard cognitive science experiment involving fMRI scans of a subject’s brain, “something” “is” found”–the quotes indicate just how much discussion there was. The gist of it was something like, if there is evidence of a tumor, does the researcher have to tell the subject, but if they do so are they acting as a doctor, but if they don’t, are they violating that person’s right to know, but if they do, and it turns out to be benign, do they not cause undue psychological suffering, but if they dont… you get the picture. I’m sure Rena has seen plenty of similar discussions–but I never quite hit on the therapy/research distinction as the root of the problem, so this is clarifying.

  2. Yes I think this post is enormously helpful Rena. Thanks!

    On ‘therapy': do we know how therapists (i.e., shrinks) regulate their ethics and research? I know that many therapists use the knowledge they gain from therapy sessions for their professional publications. Who exercises oversight in that case?

Comments are closed.