Monday, December 18, 2006

Carome, Cohen, and Normal Science

Dr. Jeffrey Cohen reports the following:

At the PRIM&R conference in Washington last week I moderated a panel with Michael Carome and Julie Kaneshiro from OHRP on "When is it Human Subjects Research?" In his presentation, Dr. Carome finally clarified OHRP's position on oral history. As many of you know, in 2003 Dr. Carome wrote a letter stating that OHRP concurred with the position that oral history activities in general do not involve research as defined by the HHS regulations. Many oral historians took that to say that oral history was excluded from IRB review, including the Oral History Association. In his presentation at PRIM&R Dr. Carome clarified that this was meant in the same sense that drawing blood "in general" was not research.

Two points worth noting:

1. Dr. Carome chose to announce his latest views on historical research not to a group of historians, but to a group called Public Responsibility in Medicine and Research, a body dominated by professionals involved in biomedical research.

2. To understand the ethics of historical research, he sought an analogy in medical research: drawing blood.

In both these actions, he shows his deep grounding in biomedical research and his discomfort and unfamiliarity with other fields.

Dr. Cohen himself continues:

It is not the methodology that determines whether an activity is human subjects research, but whether it meets the regulatory definition of research - a systematic investigation designed to develop or contribute to generalizable knowledge. . . . The problem is that the regulations don't define "systematic investigation" or "generalizable knowledge" and they don't say who is to make that determination.

It is unfortunately true that the regulations, 45 CFR 46, fail to define "generalizable knowledge." But Cohen could have considered the two places where the regulations use the term, other than to define what is research. Section 46.406 twice refers to "generalizable knowledge about the subjects' disorder or condition." In a medical setting this makes sense; it distinguishes between knowledge about an individual's prognosis and knowledge about the disease itself. But the distinction is meaningless when disease is not the issue.

Thomas Kuhn writes, "in science . . . novelty emerges only with difficulty, manifested by resistance, against a background provided by expectation." (The Structure of Scientific Revolutions, 2d edition, p. 64). Both Carome and Cohen expect to see research that fits into their expectations for medical research. Faced with cases that don't fit, they try to shoehorn them into their preconceived models, ignoring contrary evidence. We need a revolution.


Jeffrey Cohen said...

I finally got around to reading this and I want to make a couple of comments. First of all, I want to take issue with your categorizing me as someone who "...expect[s] to see research that fits into their expectations for medical research." As you know, I am not a physician. I have a Ph.D. in Psychology and spent twenty years working with an IRB that primarily reviewed social and behavioral research. I certainly don't expect everything to fit the medical model. Second, you quickly dismiss Mike Carome's example of blood draws because it comes from the field of medicine. Can social scientists only understand examples from their own fields? The example of blood draws is very apt. Even in medical research, blood draws are "generally not human subjects research". As I said, it is not the fact that you are drawing blood from someone that makes it human subjects research, but the purpose for which you do it what is the deciding factor. People draw blood for many purposes, if it is part of a systematic investigation that is designed to contribute to generalizable knowledge, only then it is human subjects research. The analogy to oral history is obvious. If I have to spell it out for you, I will. People interview people for many reasons. If they are doing it in a systematic way to draw conclusions beyond the subjects being interviewed, then it may be human subjects research. Finally, your characterization of "generalizable knowledge" in the context of disease is not applicable to the way it is used in the regulations. Generalizable knowledge, in the context used in the regulations, is information that is gathered to draw general conclusions beyond the subjects from which the data is gathered. This has nothing to do with disease. Finally, even though PRIM&R's name refers to medicine, that is a carry over from it formation over 30 years ago. Since then PRIM&R has grown into an organization that includes all aspects of human subjects research, including the social sciences. In fact, Mike's talk was in the social/behavioral track of the conference. Critics love to throw around the term "biomedical" as if that somehow attack the validity of the argument. There are many of us who fully understand non-biomedical research, even qualitative research, and still believe that the regulations should apply to all human subjects research.

Zachary M. Schrag said...

I refer Dr. Cohen to the findings of the National Bioethics Advisory Commission in its 2001 report, Ethical and Policy Issues in Research Involving Human Participants:

"Because the National Commission focused its attention on medicine, it used a framework of 'the individual versus the group.' The basic element of this distinction was that the practice of medicine referred to activities designed solely to enhance the well-being of the specific patient, while research, often using many individuals, was designed to develop or contribute to generalizable knowledge.

"Of great interest to the National Commission were ways in which to classify innovative therapies or practices (London and Klerman 1979; Robertson 1979; Sabiston 1979). It was widely recognized that a common practice in medicine involved physicians trying therapies or administering drugs in a manner that differed from generally accepted practice standards. Such innovative practices were considered not to be research because they are generally not carried out in a systematic fashion and they do not generate generalizable knowledge. In addition, there is no built-in conflict of interest between the physician and the patient, because it is presumed that the sole interest of the physician is to provide potential benefit to the patient. However, the National Commission recommended that innovative practices be studied under a research protocol as soon as it became appropriate to study them systematically (National Commission 1979). NBAC agrees with the views and recommendations of the National Commission regarding the handling of innovative therapies.

"Missing from the National Commission’s deliberations, however, were activities outside clinical or behavioral settings. For example, public health and certain types of research in the humanities and social sciences (e.g., oral history) were only minimally addressed in the National Commission’s reports. Thus, a definition of research emerged that differentiated clinical practice from clinical research, but did little to help differentiate practice from research activities in other areas."

There is more work to be done on the history of the National Commission. But the available scholarship contradicts Cohen's assertion that "Generalizable knowledge . . . has nothing to do with disease."

Jeffrey Cohen said...

Whether the National Commission adequately considered non-biomedical research in its deliberations is a matter of historical interest, but not directly relevant to understanding the regulations. The regulations were not written by the National Commission, but by individuals within the then Department of Health and Human Services. The regulations, as they are written, do not relate "generalizable knowledge" to disease. When those regulations were written, in the late 1970s, they were always intended to cover non-biomedical research. I was at the PRIM&R meeting in the fall of 1979 when officials from the Office for the Protection from Research Risks (OPRR, the predecessor to OHRP) discussed how the "new" regulations would apply to the social and behavioral sciences. At that meeting they discussed how they were building into the regulations adequate flexibility for IRBs to effecitively review social and behavioral research. The subsequent regulations had that flexibility built in and it works well. The interpretation of "generalizable knowledge" that I described in my comment works to help us differentiate between research that needs IRB review and that which does not.

Zachary M. Schrag said...

This discussion continues at Jeffrey Cohen on Generalizable Knowledge