DR. SHAPIRO: Well, thank you very much. I propose that we take a break until 3:45, that would give us about a ten or fifteen minute break, and then we will return to our agenda. Our next agenda item is the human biological materials staff draft. I'd now like to discuss the July 10th memo. Some chapter material put in the appendix and so on. But the report, the draft that you have from the staff in front of you, attempts to cover all aspects of our discussion, save the really critical aspect, namely, the recommendations themselves. That is, what are we to make of this field currently described in chapters 1 through 4, which deals of course with this—I think in a pretty comprehensive way—with the general environment within which we're going to try to make some recommendations.

The memo then goes on to think about possible conclusions and recommendations. And that's where I think we should focus our discussion here today. We really would appreciate comments from the Commissioners also on the first four chapters—I'm not going to deal with those right now. But when I say we really would appreciate it, I really mean it. We are not receiving a lot of feedback from the Commissioners that are detailed and in fact helpful to us in moving this forward. And that's an absolutely essential aspect of trying to get us on schedule here. So, while I'm not going to pause for it today, if you have marked-up copies of the first four chapters, if you have comments, perspectives, whatever way you want to describe it, please give them to myself or to Eric or any members of staff and they will receive very careful attention. But it is really quite important to get that feedback if we're going to move this forward in a timely basis.

My current anticipation is that I would like, in fact, to issue this report this calendar year, which means that we should have a complete staff draft by September, which is at our next meeting. And hopefully that will be sufficiently close as a result of our discussion there to be able to move toward a public comment period and so on and so forth, much as we're doing with the capacity draft right now. And so that's at least the objective. And I don't see any reason why not, right now why we shouldn't be able to achieve that, pending some fundamental disagreements which may be generated by our discussion and need to be ameliorated or . . . What we do not, incidentally, in any of these drafts have to reach is a complete consensus on all the issues. It's not even clear that's a good idea, but we'll just see how far we get.

Now, with respect to this particular July 10th memo, which I think we could use at least as a way to begin our discussion, under possible conclusions and recommendations the first one dealt with there deals with what I think is a very important decision for us to make in this area. Namely, whether we think that, by and large, we can work within the existing structure and simply by clarifying what are things that are truly not very clear under current regulations, to clarify them and therefore improve the structure and level of protections that are involved in the use of human biological materials. And in some sense this is that point of view at least, that's summarized on top of page 3 of this memo, which you can read for yourself. It's the first complete paragraph on the top of page 3. That's not to say that no regulations have to be changed or that we might not say that within this overall structure that provides on the whole pretty good protections if appropriately implemented and understood. It may be that there are regulations we do want to change, but that would be very different from saying the structure is just not adequate, we just can't deal with it, we have to start constructing a new structure. Just to tell you where I'm leaning on this and just as a way of getting this discussion started is, I think, the existing structure is adequate for us to deal and improve the situation, given my own particular views on what has to be done here. But that's an issue which is open for discussion. Others may have quite different views of that, and I'd be interested and perhaps we can start that way. Let's see how the Commissioners feel about that. Larry?

DR. MIIKE: It's not specifically on the guidelines whether it’s the same process to begin with. In the taking of clinical samples . . . In the informed consent process for clinical samples, the issue about use of samples for research is so buried in the informed consent process that I don't see how that's any kind of consent. So I think that we have to address explicitly, especially in a research setting since there are going to be research subjects, is a very different issue. But just the fact that it's buried in there. And I think a simple solution is to just separate it out and make people sign twice or separate it out, put it at the very end. They can . . . Some means has got to improve that process.

DR. SHAPIRO: I think that issue, that is, the quality of the informed consent process and what we might recommend to make that process more real, so to speak, more genuine—put it that way—is an important issue which we certainly need to discuss. And I think in my view of things it qualifies as being within the current structure but something that needs specific improvement. And perhaps that recommendation or another might be suitable. But I think that's an important issue. I agree. Bernie?

DR. LO: How can you sort of conceptualize this? As I read this through, it struck me that I thought you were trying to kind of get us to see the big picture and are things basically going pretty well versus a complete overhaul. I was finding it very hard to sort of backtrack from the much more specific topics we had been talking about with our grids and such. But I'm just wondering—do we have a sense of closure or agreement on those issues? I think if we do, then we can easily go back and say, "Well, are we basically just saying the general framework is there and we just need to clarify, give guidance, so forth?" I just find it hard to get to your bottom line, Roman I versus Roman II versus Roman III, without having some sense of whether we agree on the much more sort of detailed, specific issues that we've been discussing at other meetings and in other drafts.

DR. SHAPIRO: I don't know if my response will be adequate. First of all, I'll certainly give you my judgment of this. I'm not the author of this draft, so I don't want to speak for that. Unfortunately as you know, for health reasons, Kathi can't be with us today who's supposed to lead this discussion. My own view is that the basic structure of the protections that are currently involved is quite adequate to deal with all the issues that were raised in those, as you say, more detailed discussions. As we go through particular recommendations, we'll get to those and deal with them one by one. I don't think that nothing has to be changed or nothing has to be added and no clarifications have to be given. But I am myself, at least at the moment, convinced that taking the existing structure and simply clarifying and perhaps adding a few regulations as necessary. Maybe something about the informed consent process—there's a whole series of things we'd have to go through here—would in fact meet that objective.

MS. CHARO: Let me also try to respond to this. The way I read the big picture of the current scene, current situation is as follows: We define the people who are considered to be the subject of research very broadly so that it includes people who are obviously personally identifiable to the researchers and people who are not as obviously identifiable but could be with a lot of work. And then we take a fairly protective stance and we say for this very large universe of people, their tissues can be studied only if they give consent. And then we make one fairly big exception to that: we have a provision for waiving that consent requirement if two key conditions are met—it's minimal risk, whatever the heck that means, which is a good area for us to work on, and it's not practicable to get their consent, whatever the heck that means. Another good area for us to work on. So that the current situation tries to create a balance between protectiveness of the fairly broad universe of people with some kind of large-scale exception carved out to permit research to go forward. And I've been struggling all along to figure out whether I think that that existing system, with some content better developed for those key terms, provides the appropriate balance between protectiveness and fostering ease of research. And for the moment, what I have not seen is good evidence that the current system can't work well and serve all of our various purposes if we focus on giving some content to these terms. I agree with you, though, that it would be nice to get a clear consensus around the table about that very, very basic issue.

MR. CAPRON: This is just a question for information. What is the relationship between the document with the recommendations in it dated the 10th and the staff discussion of protection of groups from harm and when do you plan to take up these different aspects?

DR. MESLIN: The former is the cover which we used as a model in the past to help the Commission focus its attention. The latter is a summary of a longer piece of writing that staff was asked to do following the Cleveland meeting on the subject of community and harm. Andy Siegel from our staff took the lead in preparing some of that material. And since it had been sort of buried in the longer text, we wanted to pull it out since it had been an issue that the Commission wanted us to focus on.

MR. CAPRON: What I really meant was for the purpose of this meeting, I understood that the chairman was asking that we really go through the cover memo as it identifies conclusions and recommendations and sign off on them or correct them or whatever we need to do so that we come out of the meeting with something that approaches chapter 5 of the report. And I wondered if, since it's not unrelated to some of the things in there, we were going to pause during that and discuss the group harm issue or if you thought, because that's really part of the text for the earlier chapters, you were at some point during today or at a future meeting, going to do that. That's all I was asking.

DR. SHAPIRO: I want to get to group harm today or tomorrow morning.That issue. Community and group issues is something that we've been talking about. We need to focus some attention on it either of these days, so I do intend and hope we'll get to it.

DR. MESLIN: Allen, just a reminder on page 4 of the cover memo, we mentioned the memo, so it could have been picked up at that point.

MR. CAPRON: That's what I wanted to know. Are you going to do it at that point?

DR. SHAPIRO: Yes. Larry.

DR. MIIKE: Just picking up on—and you tell me if I'm interpreting it correctly. We spent an inordinate amount of time in our subcommittee meetings going over a grid about previously collected samples, future collected samples, etc. I look at that as still valid. This is the overarching thing and we can use that grid to say from the sample point of view, how are we treating these and how do these things apply? So I don't see this as replacing that. I still see that as part of the discussion, although I don't see it in the current draft.

MR. CAPRON: I thought we dropped it after the meeting in Tysons Corner.

DR. MIIKE: What I'm saying though is that it's all well and good to talk in these grand, lofty terms, but researchers and everybody else are going to know what was meant in terms of a specific biological sample, and I think we need that.

MS. CHARO: I'd like to emphasize that the researchers currently organize their work around the existing regulations, which are not focused on the status of the sample. It's focused on whether or not an individual is identifiable. So the issue is not whether or not a sample is identifiable; it is whether an individual is identifiable by virtue of research being done. And to that extent, the grid is not in fact the way researchers would figure out how to apply some set of rules. They're going to be looking to the regs. Now, again, we could change how we approach this, but my personal preference would be not to go back to the boxes because I found that in the end it did not add to clarity and it did not dovetail well with the existing regulatory system, which, unless we're going to scuttle it totally, we must work with and we must insert our work into.

DR. LO: I'd like to express a concern I have about trying to decide, "Can we make it work by coming to agreement on what's minimal risk and what's identifiable?" I mean, I've looked through the briefing book. The folks at Mayo Clinic have spent a lot of time thinking about it and say categorically they don't think genetic research per se is minimal risk for a whole lot of reasons. Then you're bound and determined today saying that if he doesn't see any reason why genetic research by and large can't be minimal risk, I think we're going to play this nod-and-wink game where you have the regulatory grid laid out. The researchers, because of these wonderful diagrams that staff has drawn, now understand what the regs say and they know what they want to get. They want to get to do research on samples without going back to patients for their consent finally, because it's just a lot of hassle to respect people's autonomy. They're going to figure out a way of saying "This is really minimal risk. It's impractical to do it. Therefore, by algorithm, table 3, I don't have to go back." And I'm not sure we're going to come up with a convincing way of saying, "Yes, this is minimal risk in these circumstances but it's not here." I think there's a lot of confusion out there about minimal risk and part of it is it's a very tough thing to define. I'm not sure we've done it in the other report either. We've struggled with that. I think it's the same thing. When I have a specific protocol, can I reliably say that most of these are going to go through a deliberative process and come up with a reasonable principled decision why it is or is not minimal risk? I have real concerns about that.

DR. SHAPIRO: And the implication of that is what, Bernie?

DR. LO: Well, I would like to go back to what are the issues that investigators are bringing to our views that may raise issues as to whether you should get consent from individuals in certain types of studies or not and give them some guidance on prototypical issues that come up. And we've had a lot of discussions here about the paradigm that I think David, you and Steve laid out for us where you have 100,000 samples that someone collected for some other reason; you go through to get a candidate gene, or several candidate genes; and you get down to a thousand and you want to go back and recontact. It's a real-life dilemma. I'd like us to try and figure out do we agree that you can go back and recontact those folks without their having consent in the first place? Is that minimal risk? Those are the kinds of issues that IRBs are facing, and I think if we're going to provide guidance on that level, I'd like to see us provide some guidance.

MS. CHARO: You know, Bernie, I don't think what you're asking for is at all inconsistent with going through what has been laid out as a potential way of answering, and here's why. The boxes aren't going to answer your questions. The boxes were just a way of explaining 1,000 different versions of traceability of a sample back to individual. They were a useful tool for figuring out how it is that individuals can become identifiable, but they don't tell us anything about the risk levels. So if we're working within the universe of individuals that we understand to be identifiable—and that actually—there are a few bumps and grinds here, but mostly that's manageable under the current regs. We're going to be spending all of our time on exactly what you're talking about, which is how does one provide content so that we can achieve a goal. Now I don't know whether your predilection is to be more, less, or equally protective of people's privacy than is currently the case. I don't know what your goal is when you discuss whether or not we should be making it easier or harder for these researchers to recontact folks, but that's exactly what we'll discuss if we were to discuss the meaning of minimal risk or the meaning of practicable to get consent again. So I think all of your concerns would come out.

MS. KRAMER: I'm wondering if, as a matter of process, if we go through these points and we bring the document to a sufficient point that we feel that we can sign off on it and put it out for public comment, if we might not then invite the appropriate persons or organizations to take a look at it and say, "Looking this over, can you come up with an example where the researchers or the IRB will not have sufficient guidance?" as we're putting this out there. For instance, to take the example you just raised, Bernie. Would there be sufficient clarity in the document as we put it out there to guide them as they need it? And try to approach it from that way.

DR. SHAPIRO: Let me just remind everyone in that respect that what the plan is here—I don't think we're going to reach today or tomorrow a sufficient degree of closure on this to send it out for public comment. My objective is to try to get enough sense of where the Commission would like to go so that we can produce, so to speak, a final staff draft for us in September, which would then go out for comment if we were satisfied with it. That's just the logistical arrangements. Now, I think it's useful to get comments from others at any time, so I think that's a very useful suggestion. David, you had . . .?

DR. COX: I very much like the idea of following through on discussing these recommendations. I, number one, think that there's a lot of—my foundation concern is Bernie's, that basically people want to do something and they'll find a way to do it. In order to obviate that, I believe that the best thing to do is to clear ambiguity in the regs. If we can clear up the ambiguity in the regs, it closes loopholes and it allows us to say what our major principles are. Right now by reading through this I think that it leaves very vague what our major principles are as a Commission. But I think that we can define those if we go through these different points. But as a starting place, to say what is ambiguous or what are ambiguities in the regs right now is a really safe place to start because that's something we have to do no matter what. And in doing that I believe that it will help us define our principles. And as we define those principles, it will close loopholes. So as I think about all the things I'd like to see get done, I think that this is a very efficient way of doing it.

MR. CAPRON: In an attempt to move us along, on page 3, Guidance 1 suggests that HHS and the other participants in the Common Rules need to describe the types of research that are included in the category of research that uses identifiable information. I have two comments on this. The first is what this says to me is that our contribution on this issue is going to be commentary which would be useful to that process but we don't feel that we can come up with refined enough language to do that job. I guess that may be the case, but I would hope that our commentary would push that process almost to the brink, even if we don't want to take the time to write the regulatory language. I mean, it would be strange if we spent three years talking about this and cannot be pretty clear about that identifiability.

The second question is what about this aspect of risk? Does it come in at this point or not? And I'm not sure I understood the response to Bernie on this. I may not have. But Allen Buchanan's argument was, it seems to me, that one has a lesser interest in being identified if the harm that may come from being identified is going to be very, very small. And therefore, that the same kind of consideration—nonidentifiability—which allows an exemption from the consent requirement, might be applied even with identifiability with small harm. I'm not in favor of adding that aspect as sort of a modifier to the identifiability. I think that invites the kind of complexity that Bernie was complaining about. It's worthwhile talking about it, I guess, in the report, as we now do a little bit, but it seems to me that in the end we should not be encouraging HHS to modify identifiability that way.

So I'm basically in favor of Guidance 1 as it's written here. The only part of it—I would like our commentary to push along the idea of identifiability in a very commonsense way. It seems to me that identifiability means that, through facts about you or your name or your social security number or date of birth, you can be linked with the information that's been developed by the researcher. And I think we are right to try to point out that now it is incorrectly thought by people that if you have a code you're not identifiable. And we should be saying that that's wrong. I don't understand the last paragraph of this section, which says, "Finally in developing this guidance, consideration needs to be given to the aspects of protocol design that are intended to slow or prevent the flow of intermediate or even final findings back to the tissue donors and treating physicians." New? I mean, what does that say to me? Yes, if you put up rules and say identifiability means you need consent and you say, "Well, wait a second. We're putting in this kind of coding so we can get back." Well, that's a good reason for having coding, but in our rationale it's a good reason also for requiring consent. So I . . .

DR. SHAPIRO: First of all, on that latter sentence I don't know what it means either. I haven't the slightest idea what it says. I tried to figure it out the last day or so. I don't know what it means, so I don't know the answer to that.

But the two other issues you raised, however, are very important, and I want to at least give you what my sense of that is and Bernie has something he wants to add, too. One, I think we should push identifiability as far as we can. Whether we can complete it in the right regulatory language I don't know, but you just can't wave a flag here and say here's something you salute and tell us how to do it. I think that's not adequate. I completely agree with that.

Second, and I think equally important, issue that Alex raised and which I very much agree with what I think is your perspective on that; namely, that we don't want to open the possibility here by just dealing with identifiability by making this more complex. It's either identifiable or it's not. And when it is, you fall under a certain regulation; when you don't, when it's not it's not. And so I think both of those issues that Alex highlighted there I think are important ones in this area. And if Commissioners disagree with that or that general kind of thought, now's a good time to engage in this discussion. Yes, Bette?

MS. KRAMER: No. Just as a point of information. Do I hear you recommending then that we say that information that is coded, or samples—I'm sorry—which are coded, are identifiable?

DR. SHAPIRO: That's my view. Absolutely.

MS. KRAMER: Okay. So it's on the table that you're proposing that this is what we say: "Until such time as there is an encryption scheme that does all these . . . gets to them."

DR. SHAPIRO: We'll get to that later. I've also tried to do some investigation in that area, but we'll come to that later. Bernie, I'm sorry. I don't want to talk so much.

DR. LO: I just want to say that Alex, as sort of clarified by Harold, goes a long way to addressing my concerns. Here's an issue, guys; figure it out. Here's another issue. I'd like to push one step further. Codes don't make you—if you're coded, you're still identifiable. Even if you strip what you think are all the identifiers—this goes back to, I guess, David, your comment from Framingham that even in large data bases, just by knowing things like the zip code, the number of children, you can often identify people. And that may be of particular concern in genetic research when you have knowledge of a pedigree in the family. So that I'd like to suggest that we err on the side if you're not sure, treat it as if it's identifiable rather than I think what a lot of scientific may want to do, saying, "Well, it's not iron-clad nonidentifiable but it's good enough that we can consider it nonidentifiable." I think, if anything, there's going to be an increased ability, either using computers or more sophisticated dramatic analyses, to make things more identifiable than they were deemed to be.

MS. CHARO: Now I'm keeping in mind that the reason we're talking about this is because as soon as we call somebody identifiable, this whole panoply of protections is going to come into place. And my question is going to be: Do we want to make this a definition in which the key is whether somebody is identifiable to 100 percent certainty? Or is it more probabilistic than that? There was an example of somebody who was put into the staff draft of somebody who gives the samples of a Family Jones that are made up of people who have the gene for a particular form of dwarfism, to a researcher with nothing saying Jones. The only thing that's on there is family physician, father, mother, daughter, son, maternal aunt, etc. And at this point, in fact because the collection knows that it's Family Jones and the researcher knows that it's mother-daughter, there's a high degree of identifiability but it's not 100 percent because if there were several daughters and you've got several samples that are all listed daughter A, daughter B, daughter C, you can't link a particular sample to a particular daughter. But you can link them to two people who might be daughter A for a particular sample. So I want to understand how far you want to push this. If you want a bright line, you might want to make it 100 percent reliable determination of who the individual is. If you want to make it fuzzier, then we need to discuss exactly how we're going to make that fuzzy line.

DR. COX: I'd like to be as specific as possible. Coded means identifiable. Having said that, there are other situations where one needs to be sensitive because, even when you think you're not being identifiable, you might be. But we can't deal with all the situations, and we're just asking the people to be thoughtful . . . That's the place of most wishy-washiness on the part of researchers right now. If this Commission says coded means identifiable, that's what it means. In fact, that's what we saw at the regs meeting. So let's say it. With respect to these other issues, it's going to be extremely difficult to ever be 100 percent or not 100 percent, but we just have a discussion about it so people are sensitive to it.

MS. CHARO: If I may . . . David, being sensitive to it is a wonderful position to take as far as what the ethics of the researcher ought to be, but please keep in mind that on an implementation level the significance is that you do or do not have to go through anything having to do with the IRBs. If you say it's unidentifiable and you want people to be sensitive, you're saying, "Please be sensitive while you don't go to an IRB. Please be sensitive while you don't go through any of these protections." And if that's what you're advocating, that's fine; but I want to make sure that's really what you meant.

DR. COX: And that's what I meant, because unless we can define the specific point of what we want people to do or IRBs to do, we are fooling ourselves to think that it's going to get done. So if we can't be specific enough to say these are the things, that's why I bring up this point about coded. It's clear what a code is. If somebody holds a code, then we say that this is identifiable. But on the other hand, for some of these other issues, it's not so quick. I'd be happy if we could come up with clear points where my four-year-old could implement them. I think that's the level that we have to beat, the clarity that we need here.

DR. SHAPIRO: Okay. There's a number of people who want to speak, including myself, so . . .

MS. BACKLAR: I just want to start. When we are talking about coded, we mean the researcher may know the code or the repository may know the code, anywhere that that code is. Yes? Is that what we're meaning?

DR. COX: That's what I mean.

MS. BACKLAR: Okay. That's all. I just wanted that point made clear.

DR. BRITO: I share with David the fact that we had to say coded is identifiable. But I also understand Alta's concerns. And I'm not sure if that's not addressed, that concern's not addressed in the third recommendation on page 6, or the suggestion for the third recommendation on page 6: "Where research is conducted on kindred, special attention must be paid to the extent of identifiability of family members." And maybe here we can just use stronger language and recommend that where's there a possibility of identifiability of family members or other affected individuals, there may be a requirement for it to go through the IRB process. That way, when we talk about coding as identifiability, that's 100 percent all or none, and then when we get to the recommendations, specific recommendations, this might take care of Alta's concern.

DR. SHAPIRO: Alta, did I understand your concern? That example that we, the following type—I may have misinterpreted what you said, that should we consider identifiability to be someone who's saying I'm 50 percent sure it's you. I'm just asking if that's the question.

MS. CHARO: My concern was raised when Alex and Bernie began saying that they wanted an expansive definition of identifiability. Now, saying that coded samples are identifiable is not expansive; it's reiterating what's currently the case. So I began to hear that they wanted something even more expansive and I got concerned that at that point they were going to go past the boundaries of certain identifiability and get into this "I'm 50 percent sure it's your stuff," where it gets very tricky.

DR. SHAPIRO: Let Bernie respond first. I have a view, but let Bernie . . .

DR. LO: I must say I find these little charts extremely helpful. I only wish they were sort of color-coded rather than gray. But the way I read it, it's "will information be recorded by the investigator in such a way that it cannot be linked to the subject?" You have to say "yes" to be exempt. If you say, "no," even though there's a pretty high chance, a pretty good chance it won't be, then you ask, "Is it more than minimal risk?" and then you get expedited IRB review. I guess one of my concerns, Alta, is that I'd like to keep the IRB involved in the questionable tough cases if only to look over and say, "Yes. We agree with you that it's identifiable or not, of minimal risk or not." What I'm concerned about is if you have a complete bypass of the IRB, if you just say it's not identifiable or using the more regulatory language, there's a tremendous incentive to answer that question a certain way. And if the expedited review is really expedited review but it's an outside person looking over the investigator's shoulder and saying, "Yes, we've had a lot of experience dealing with these genetic issues and we think it meets the not-linked-to-the-subject and it doesn't involve greater than minimal risk. Who's your expediter?"

MR. CAPRON: I also need a little clarification of that. If researcher A goes to repository B and says, "Do you have any families of dwarf samples, tissue samples from a family of dwarfs?" They say, "Yes, we do. We've got a mother, father, maternal aunt . . ." "Could you give those to me and label them as to which is the mother, which is the father, and which is the aunt, and so forth?" Other than that, this could be a family collected in 1900 or 1998. It could be from Europe; it could be South America; it could be the United States; it could be anywhere. And he doesn't know anything. And that sample isn't coded; it's an anonymous sample. The fact that the person, if he knew the family, could say, "Oh, you're 3-1/2 feet tall and 6 feet tall, you must be the dwarf here." I don't get the identifiability.

MS. CHARO: No, no. It's that the collection knows that it's Family Jones. It might be from 1905; it might be from 1995; but they know it's Family Jones.

MR. CAPRON: But they give it over to the researcher with no code on it. They're simply saying . . .

MS. CHARO: Each sample is labeled by the position within the kindred.

MR. CAPRON: But that's not coded.

MS. CHARO: I agree with you that it's not coded, Alex. My point simply was that we have a difference of opinion here, a real difference of opinion between David and Bernie about whether or not how we want to be treating this in terms of protectiveness. It's not coded. And as we currently understand these things, it's not necessarily going to be subject to the regulations. But in fact, by putting together just the name of the family. I don't even meet them. The collection can say it's Family Jones, and I know this sample . . . I have a sample in front of me that's labeled mother, and between my information that this sample has this particular gene in it that I'm looking at in my lab and it's labeled mother and the collection's knowledge that this is Family Jones, we've now in fact identified this individual. Okay. That's identifiable under the regs. If I'm looking at daughter and there's more than one daughter, there may be a question. I know this sample belongs to a daughter in the family, but the best we can get is a guess that it's one of two people in the family. We don't know which one. I'm understanding Bernie as saying that he wants an IRB to have an opportunity to think about whether that meets the definition of identifiability. I'm hearing David say, no, he just wants the researchers to be sensitive to that. And all I'm saying is that if we're going to want to send it to IRBs, we're going to have to come up with some very good language to explain to researchers when they have to go to the IRB for that extra look. In no case was I talking about meeting up personally with Mr. or Mrs. Jones.

MR. CAPRON: Okay. If I can pursue this one step further. I really thought you were using Jones to mean a name that they just make up a name.

MS. CHARO: No. A Family Jones.

MR. CAPRON: This is Family Jones, so the researcher, the person who's conducting the research, is told this is the McGillicuddy family, to use a more particular . . .

MS. CHARO: They're not told. But the collection knows that it comes from the McGillicuddy family.

MR. CAPRON: So are we now saying that the only thing that is not identifiable are anonymous samples? Because I had not understood that.

MR. SHAPIRO: I'm sorry. I didn't hear that last one.

MR. CAPRON: Are we now saying that the only thing that counts as nonidentifiable are anonymous samples? In past discussions of this—we've had so many discussions that we've brought different things, I suppose—I had thought, to go back to those much-feared charts, I had thought that at some point we were distinguishing between those things which are provided with a code, meaning I write down that this is from Mrs. Jones and I put a code number 1 and then I give it to the person with a number on it and they can come back to me and I can figure out who goes with that. And that we were calling identifiable, potentially identifiable. Are we saying the only ones are those in which the researcher, when he pulls them off the shelf, doesn't even see a name on it? Even if he sends it over and doesn't put a code on it. Excuse me, the repository. Thank you for your question.

MS. CHARO: I think, again, that we will confuse ourselves if we focus on the samples and on the details of how the samples are being managed and the details of the coding schemes. If we focus on the individuals, I think we're going to be on stronger ground. The question is always whether or not the individual is identifiable, not whether the sample has a code. That's just one way of making an individual identifiable.

Now, there are these weird gray cases in which an individual is identifiable without a code. That's no problem. The regs say that they're a human subject; the panoply of protections comes into place. Great. Then we have the second set of weird cases where an individual is not completely identifiable, but we can move it down to a level of certainty where we can look at two or three specific people in the world and say, "We know it's one of you." The question is do we want that to be subject to this panoply of protections. That's what I'm just asking. We're going to try to suss out what identifiability means. I'm only asking is it enough to say we know that you're one of one or two or three. Or does it have to be "We know you're the one?"

MR. SHAPIRO: If I could interrupt here, maybe I could say something that'll be helpful. I don't see how we could get into a situation. What we have to decide is whether the probability of 1/3 versus 1/30 versus 1/100 versus 1/10,000,000 is going to be an issue in deciding whether or not the panoply of protections come in or doesn't come in. So my own view is that what we have to focus on—I want to come back—let me focus on the individual first. It is whether it's identifiable or not. And just how it's identifiable is not a critical issue, but identifiable in my view means identifiable. I think it's 50 percent sure it's you or I'm 1/1,000 percent sure it's you. And it seems to me that can work in a fairly straightforward way. A much harder one, which gets into the next issue, is does identifiable refer only to individuals? Then you can say, well, does it refer to family, does it refer to group, some other kind of group or something. That to me is another kind of issue which we need to address. But for the moment, if we stick to the individuals, it seems to be that the best way to go is to say—I guess I'm just reiterating what a lot of people have said already— that it's either identifiable or it's not. And identifiable means "could be identified." And as technologies get better and you can identify with different kinds of information, then we have to adapt to that as it goes along, things we could not identify 20 years ago, we can now identify. And so something would have to change. And this will be a moving target as we go along. And it seems to me—but we'll get to the nonindividual case, that is, family or group, community or whatever it is in a few moments. Let's not try to redo that.

MR. CAPRON: May I ask Alta—do I take it then that the distinctive thing you were saying there—I agree with Bernie that we ought to have the presumption in favor of potentially identifiable, which means some probability you can do it. After all, that's the whole possibility that lies behind this. You have a coded sample, and it's unlikely that the code is going to be broken. But it's potentially breakable. That's why we say we should call it identifiable, and you should go through this review process to see where it comes out. And if it were potentially identifiable with a very high-risk, high-discriminatory factor subject being studied, you'd say, "Whoa. You're going to need consent to do that because potentially if that ever came out, that person would be very badly harmed." If it's something of low-risk or of minimal risk, then you may be able to not have all of those attached, but you still have the initial IRB review. You're still subject to the protection.

The question that your example raised was in making it members of a family, you really have already identified the collection of samples you have, and you don't know if it's sister A or sister B. You've got the 50 percent probability. You're potentially right if you took a guess; you could be wrong. But you know that it's either, to use your name, one of the Jones girls here that you're looking at in this sample. If a researcher says, "I want to study XYZ gene in people who have had thyroid cancer," and he goes to his colleague who runs the lab and says, "Can you give me 26 samples of people from whom you extracted thyroids because they had thyroid cancer?" And he says, "Sure." I say, "I don't care if they're male or female. Just give me 26 samples." And he goes into his lab, and as he pulls the samples, he remembers his involvement with each of those patients and he remembers Mrs. Jones and Mrs. Smith and Mr. Johnson, and so forth, one after another. And he has all those. And he sends them off but they're just labeled A to Z and he doesn't keep any record as to whether Jones is A or whatever, so they're not coded. He just separated them into 26 categories, A to Z. But he remembered these 26 people. We don't call that identifiable because when the research is done, he won't know who you're talking about. So what was distinctive in your example was the fact that it was a family and we knew that much of the identity of the group of people. Is that the point of the example, because otherwise the result would be that, if you don't have your repository only having anonymous samples, you'll always have "identifiability" because the researcher may know the people from whom they came.

MS. CHARO: Yes. No. Of course. I mean, the example was constructed to meet Bernie's concerns and to test out where we . . . I don't disagree with Harold's conclusions about we ought to come out. It's actually my preference as well, but Bernie seemed to be advocating for a broader definition of identifiability, and I wanted to put an example on the table.

MR. CAPRON: The presumption in favor of review rather than . . . In other words, if there's the potential of identifiability, we ought to have a review process. That's all.

DR. SHAPIRO: Could I just ask Bernie a specific question? You said something, Bernie, I wasn't sure that I fully understood it. And it had to do an aspect of IRB review. Did you say that you wanted decisions with whether something was identifiable or not? A decision of deciding whether something's to be subject to IRB review . . . I think that's the . . .

DR. LO: Where there's some legitimate question, I'd rather have a really expedited IRB review where someone in the IRB looks at it and say, "Yes, based on . . . "

DR. SHAPIRO: You don't want to leave that decision to just the investigator. Okay. I just wanted to make sure I understood you.

DR. LO: Well, I think the expedited review should really be not very onerous. So that from an investigator's point of view, I don't want there to be a huge incentive to have it exempt. And I think that may require some changes in the way IRBs are run.

DR. SHAPIRO: Okay. I just wanted to understand that. Thank you very much. Bette?

MS. KRAMER: Just to address some of Alex's last remarks. Alex, I'd be very surprised if any repository would ever send out samples without knowing the individuals or without making some record of the cases from they sent it. So it seems to be that if they're identified at the repository, they're going to be at least coded and therefore identifiable.

MR. CAPRON: If that's an accurate description of what they do, then the answer is they're all going to come under that. And that only exception would be those which start off being anonymous. And they say, "We sent you sample number 471, and we don't know who 471 is. But if they did it the other way and just said, "We just sent them over. Here's the list, but we didn't write down A next to so-and-so, so we can't tell you." When you come back to us and say "We want more of A," we say we don't know which one it was. We'll send you the 26 samples afresh, but we can't tell you which is which.

DR. SHAPIRO: Let me suggest that we move on here. I think I have a pretty good . . .

DR. LO: If I could make one quick corollary to just sort of flesh things out. I think, if people take this bypass where they say it's nonidentifiable and therefore exempt from 45 C.F.R., the implication would be that you then can't later go back and say, "We want to contact the subset because they're genetically interesting." So with David Cox's example, you either say, "It's really unidentifiable" and live with it or you say "It is identifiable" but I think it's minimal risk and I'll go to the IRB.

MR. CAPRON: And just to be clear, on Table X. On the right-hand column where the question is, "Will information recorded by the investigator in such a way that it cannot be linked to the subject?" And then it's yes and then your example, or no. We almost want to drop a footnote to the "no" and say, "And that includes maybe or almost certainly not." In other words, if there's any doubt, have that issue reviewed. And then underneath know there's the—or maybe we do need the maybe category. There should be IRB review of the question of nonidentifiability. That's what we're saying, isn't it Bernie? I'm trying to identify places where we're in effect suggesting to this HHS process that we say should be operating what they should do and one thing they should clarify the regulations to say "It's the responsibility of the IRB to confirm that it's nonidentifiable" or that that determination has been made correctly.

MS. CHARO: Alex—and I guess also for Bernie—just by way of information, you're discussing these in a way that makes me infer that you believe this is not already the case. But the way it works now is there's an initial judgment that has to be made by an investigator as to whether or not her work constitutes research and whether or not there's a human subject involved. This is the inevitable first point of self-regulation that cannot be evaded under the current scheme. These investigators then voluntarily present themselves to the IRB. They're already supposed to in cases of ambiguity consult with the IRB chair or administrator for an informal ruling as to whether or not it needs to go through the IRB. And when it's genuinely ambiguous, that person often will say, "Let's go to the committee and have everybody talk about." So I have objection to what you're suggesting, but I wouldn't want to put it on the table as if differs somehow from what's currently going on. What you're perhaps suggesting is that there ought to be strong language urging people to take it more seriously, that there ought to be examples made of what constitutes an ambiguous circumstance that people should recognize as such, but we already do exactly this.

MR. CAPRON: Well, you're describing the University of Wisconsin at Madison. We do not know what goes on in the other several thousand IRBs and the Inspector General's report would make me think that they don't all look like that. The alternative here would be to say that that issue ought to be an issue that gets at least administrative review in each and every case.

MS. CHARO: It can't be. Somebody's got to alert the IRB to the fact that there's a research project out there. That has got to be the investigator. There's nobody else who can do that.

MR. CAPRON: Excuse me. You know this, I'm sure, from submitting things for funding to HHS and NIH. When you sign off on that, you say you've gone through the regulatory process. If you can now check yourself as exempt because it's not identifiable, that's different than saying that the IRB or the IRB chair or the IRB administrator has checked you off as exempt. Now we could say that a better protection would be that before you can submit, you have to have their signature saying, I looked at the protocol and this is truly one of those nonidentifiable situations. It doesn't have to be only the research.

DR. MESLIN: Perhaps as a point of clarification, as a former Project Officer at NIH, it might be helpful to raise the following suggestions. That if the issue is at what point ought sensitivity be raised as to whether or not this would go somewhere, it could be helpful in our guidance were we to make precisely those points. There are a number of places where individuals might need to pay special attention to the fact that this kind of issue would benefit from IRB assessment. There are more than just investigators. When you go to Project Officers, oftentimes you haven't gone through an IRB. And your approval for funding will be dependent upon an IRB approval. So it's not either one or the other. There are several instances in which that alert should be provided. I think that's why we are opposing that the Commission consider the guidance that is developed indicate those areas of ambiguity. They says nothing about whether the Commission ought to find and formally find that coded equals identifiable. These are not mutually exclusive suggestions.

DR. MURRAY: The one question I had about this section, page 3, the first full paragraph. It looks very reassuring about the existing regulations provide adequate protection, etc., and I think I agree with Harold that that is true insofar as it pertains to research which would come under IRB review anyway. What about research that takes place purely in the private sector, no government funding, that might otherwise escape regulation generally? I guess my point is that I think the regulatory regime is generally satisfactory, the framework is satisfactory for research that is currently covered by the framework, but there is this world of potential research which currently lies outside the framework. And I don't think we should reassure people that all research is covered if it isn't.

DR. SHAPIRO: I take that point, Tom. That's correct.

MS. KRAMER: On page 2, that bottom paragraph, the 4th line from the bottom. It says "except when carried out by private investigators, etc." I made a footnote, a note that it seems to me that it would help this if we fleshed out something about the scope of that. I mean, there's . . .

DR. SHAPIRO: Okay. That sentence I think should read "investigators and organizations not currently covered" because you can be covered if you're part of an organization that's covered even if you yourself don't have government funds and so on. But I accept your clarification. I think that would be really helpful. David, I'm sorry. I have your name on my list. I apologize.

DR. COX: Arturo first I think.

DR. BRITO: It's very frustrating. When I originally read through these guidance—I want to stick to what the guidance principles that were recommended were—on page 5, and I'm going to go back to Guidance 2 with this, and after hearing what Alta was saying, I think this might help a little bit. One of the problems I had with this is that minimal psychosocial risk. I think that any time there is a potential for a group, someone outside the individual, group, community, kindred, somebody in the family, etc. to be identified, then you are no longer "minimal risk." Psychosocial risk for the most part is not the minimal part; it's greater than minimal risk. So I haven't had enough time to think about the language, but if we go back to the second Guidance 2, and in here, somehow put in here, as part of our guidance make it very clear, that when there's minimal risk to an individual, but the information or the protocol might implicate another individual or group and there no longer is minimal risk. And then we go to Table X and say that any research that implicates another individual is no longer minimal risk, therefore would require full IRB review. I don't know how that relates to all the other conversations that went on afterwards. But I was hearing more about implementation problems than actual wording. What I'm getting at here, what Alta's talking about, when you can identify other family members, or a group of people, then you pose a problem where it's no longer just a minimal risk. It's greater than minimal risk; and I think that would take care of it itself worded correctly by having that required to have IRB review.

DR. SHAPIRO: Okay. That's a helpful point. But let's come back to that in just a moment, Arturo, when we get Item 2 and we can bring in five more issues like that. But do you have something regarding the first item?

DR. COX: I do indeed.

DR. SHAPIRO: I do want to move on to this area.

DR. COX: Yes. And I hope this will allow us to move on. The vast majority of the samples and the situations that we're talking about in terms of people as individuals are going to be coded individuals. The vast, vast majority. Right now, that there's ambiguity, genuine ambiguity on the part of most researchers in terms of whether those individuals represent something that should go to the IRB or not, because they say to themselves—in some Machiavellian, but in most cases totally clueless—if I don't know who that person is, then they're not identifiable. Well, it turns out that those researchers were wrong, but they don't know that they're wrong because nobody ever told them they were wrong. So we need to tell them. So I would just like to emphasize that point.

DR. SHAPIRO: Well, I don't know if we're doing something monumental, but I think we'll do that.

DR. COX: Because it's not clear to me, Harold, that we are going to be able to do that, but if we do that...

DR. SHAPIRO: We're going to do. What the world will do is yet another matter but I think I don't really hear any dissent amongst us on that issue. I think we're really completely agreed on that. And I think we've heard enough and had some very interesting useful suggestions regarding this first item, which will enable us to articulate it more carefully in a more helpful way, and I want to go on to what is called here Guidance 2. Do you have something before we do that?

DR. MESLIN: The only thing I was going to ask is it helpful if we were to add a finding, an actual statement that preceded the guidance of the kind you've just described, such as the Commission finds that coded equals identifiable—don't take that language identically. Because from what Alex had said earlier, we don't want to simply say there ought to be a guidance, and from what Bernie has said, state the finding, the finding within the contexts of the regs which we feel to be appropriate. That allows us to say there is inadequate information of the kind that David Cox's investigators ought to have. I raise that now because for each of these guidances, you may want to have statements of finding of that kind. The Commission finds that...and as a result we recommend the following.

DR. LO: I think we really don't want to do that because that will be a sort of richness and the tale of the text that will make it much more useful. And since this is the commentary to the principles that Al Jonsen was talking about earlier, showing what we mean in various ...

DR. COX: What are the frontiers, we want to concentrate on the frontiers.

DR. LO: And sort of underlining what the issues are, so not only do I know what I need to do but what the implications are for other things I might be able to do.

DR. SHAPIRO: Alex, last comment on this because I want to ....

DR. CAPRON: It seems to me that with Guidance 2, we have three situations. The easy situation is one in which the sample is identifiable and therefore has to go through an IRB process. And what we're saying is the IRB ought to consider the possibility of group harms and comes to the conclusion that the group harms have been resolved in a way that's satisfactory to allow the consent of the individual. The hard cases are the opposite sides of that. On the one hand, Arturo, we get to the situation right now there is no process for IRB review of something where the samples are not identifiable. They are anonymous samples that came from an identifiable group. That is to say they were samples collected during Tay-Sachs screening that you now want to use for BRCA-1 screening, but they were just numbers and you don't know who the numbers refer to. The repository doesn't know so you don't have a problem with their passing them on, there's no coding to people. The question is, the present regulations wouldn't provide to the IRB to even see that, so the first question is should we say that if there's any potential that samples could yield results which would reflect upon a group or community that they have to have IRB review. I can understand an argument for that but it really would require a big change. And then we have the opposite side where you have the IRB look at this and say wait a second, this is going to harm the XYZ group a lot if it's done this way and the results come out. But all the people who are members of that group whose samples are going to be used are eager to have the research done. And the researcher's a member of the group, too, and they all want to have it done and they're all ready to go. Do we stop them because of that? Do we have any authority to say that this research should not be done because one of the harms that would be weighed on the scale is a harm to a group of people who are not research subjects. And that's...on that latter issues, I think that the staff memo suggests no, in the end, you really can't.

DR. BRITO: Isn't that when you get community consultation? Isn't that ... but that's my point. Should the IRB be determining whether or not they think it's harmful to the group. Should it be somebody....

MR. CAPRON: Well, but even for them to say you have to have community consultation, you go through community consultation for two purposes. One is to get advice from the community about how to do the research in a way that minimizes the harm. But if in the end the community says it would be harmful to have it done but the researchers think that there'll be great value and the subjects are all willing to have their samples used, does community consultation mean well, nice to talk to you, thank you very much. Or does it mean the community can go back to the IRB and say we went through the consultation and we still don't want it done and they're not willing to change it sufficiently to make us agree to it. Does that...and I'm not prepared in that situation to say the research should not go forward. Mostly because I don't think that the process of community consultation is going to be precise enough to end up with answers of what the community is and how it's going to be.

DR. BRITO: Right, so I have that fear also, and anxiety, about the community representative. But then what you're saying is you're leaving it up to the scientists to decide what research should be done that may identify community and up to an IRB that has no community representative.

MR. CAPRON: It may or may not, that's true, there's no requirement because in most situations there are a thousand communities, as it were. You can break up communities in so many different ways. So I would be inclined to think that it might well be worthwhile to add another question to this flowchart, and the question would be will the research produce results that will identify characteristics of a community. And if the answer to that is yes, even though it's not identifiable to the individual it ought to have IRB review just so that they get a chance to see that the risks have been minimized at least. Because in that case you're not dealing with the consent. All those people whose samples you used might be equally alarmed if they knew that there samples were going to produce information that's harmful to their group but they don't have to be asked because we don't know who they are. You see, you don't have the argument that you have in the other dilemma that in that dilemma, at least you're operating on the informed consent of individuals who say, we think this information is worth getting even if it comes with a price. There you don't know that, you have no such guarantee. And the only substitute for that would be some kind of a—which we talked about before—going back to that community and saying well what do you think of this, we're not asking for your individual consent because we don't know if it's your sample. So you go back to the temple where the Tay-Sachs screening was done and say what do you people who happen to be here today think of this research using your former congregants' information from 25 years ago. I mean that would be the only way to go about that.

DR. SHAPIRO: David, then Alta.

DR. COX: So I come down on this issue number two the same way I came down on issue number one, guidance. Right now, if you have a group identified, and in large part what we're talking about here are clearly defined racial groups, there could be many other groups but right now, at least in terms of genetic work, people are identified by their skin colors in terms of group.

MR. CAPRON: Broadly, you mean population groups.

DR. COX: Population groups. But let's....

MR. CAPRON: Racial may be the wrong word.

DR. COX: Yes, but that's how it's designated.

MR. CAPRON: Colloquially, right?

DR. COX: Now, and heretofore, there's never been any consideration of this. If you don't identify the individual, no big deal. Well, sometimes it is a big deal, sometimes it isn't a big deal. But the IRB isn't necessarily a representative or any better than anybody else, but at least it's a body that's sitting there looking at it and trying to deliberate on it to have the researcher think about it in terms of whether there's risk involved or not. So it's being considered. That's certainly a better situation than what's going on right now. I quite agree that going and getting community consultation is not practical, but on the other hand not having somebody even consider this if the group's identified strikes me as irresponsible. So, just to simply say if the group's identified, the IRB looks at it is a simple, straightforward recommendation. And I support it.

MR. CAPRON: If you're the IRB, the question is what do you do when you look at it.

DR. COX: Well you know what, Alex, it's situational. And since it is situational, I go with Bernie on this. We have some examples of what we would do in different situation, what we would recommend as guidelines. But there are not going to be clear-cut, push button A and you get answer C on this. We're not ready for that, and we don't know what those answers are. But this is one of those frontiers and the frontier where the IRB tries to make some statements. If it looks at it and it says we don't know, we don't see any risks here, if no member of the IRB sees any problems, then you go forward. But not to have it reviewed by the IRB if the group's identified, I think, is a mistake.

DR. SHAPIRO: It seems to me, and this is in response to your question, as Alex says, what do they do besides lose a night's sleep, and it seems to me I thought you really mentioned what they should do before. They ought to look at this and say if there is a potential harm, the design is such that it minimizes that.

MR. CAPRON: Yes, exactly. But I guess my example, I was being rhetorically but not entirely about going back to the temple. Suppose, take the most extreme example. You have the research is going to be carried on at the Mother of Perpetual Sorrows Hospital, a Catholic hospital with a Catholic IRB and so forth. And it's going to involve a Jewish community, Jewish samples. And the Catholics look at it and they don't see any problems. It turns out that there's some particular sensitivity that has to do with cultural or ethnic differences that relates to...they don't have access to that. It's not that they're not being conscientious. They look at it and the basic parts of good design are there, and the researcher is sensitive to not do broad generalizations, but he's out to look if there's a Jewish gene is the way he's looking at it, that's it. And if you went back to the group that was in some ways involved, they might think of something that you didn't think of. So have you added anything by going to the IRB? Well, in nonextreme cases, you probably have, I mean, if you don't have this chasm that I described. A conscientious IRB might be able to suggest ways of modifying the risks or tweaking it or doing this or that, or do you really have to do it this way or that? And the researcher might say no, I don't really, and they could get around that problem. But sometimes if you don't take the next step, then in the end you don't have any sense that the people involved gave consent. When we talk about an identifiable sample, the people involved are the people from whom the sample came. And my view is that that person can override the group interest if they're not being persuaded by.

DR. SHAPIRO: I agree with that.

MR. CAPRON: If you don't have that person exercising that judgment, who him or herself might have some sensitivity to the thing that is of concern to his or her group. Do you try to seek a surrogate for it? Sometimes you might have a surrogate. It might be a temple. If it's not, if the samples are just random Jewish women who came in for this procedure and you now want to test their samples for some other thing, do you know who to go to? Do you go to Hadassah? Do you go to the AJC? I mean who do you go to in that case? It is a dilemma, what's your community.

DR. COX: Alex, in advice, simple advice to the researcher, if the group is identified, it's the onus on the researcher to say I've identified this group and I've thought of the fact that it's identified, I've thought about these issues and this is why, this is how I feel as a researcher about that. The IRB looks at those statements, it doesn't come blank where the researcher says it's up to you IRB, I don't take any responsibility for this. So you have a discussion about this where the onus is on the researcher. And that's not the situation right now, let's make that the situation.

MR. CAPRON: So you're saying that's a step forward, even if it's not perfect.

DR. COX: That's correct.

DR. SHAPIRO: And I think the problems raised, myself, identify a group and then are just too overwhelm any benefits you get, as far as I.... Excuse me, Trish and Bernie.

MS. BACKLAR: One of the things that's interesting in listening to this, I'm harking back to our other report and how we dealt with this, these people that have mental disorders. And how we say to the IRBs, or we're going to say to the IRBs, that you have to have a record. When you're looking at this kind of issue, you should invite somebody from this group, either an advocate or a member of this population, relative of this population, to sit on the IRB and think through the issues at that if you're addressing this kind of thing. And wondering if we might not look at this in the same way and consider making the same kind of suggestion, depending on the populations being looked at. Because one of the issues that's so interesting to me when we look at the communities in these populations, they're not just ethnic, they're not just religious, they also maybe diseased populations. And more often than not, that is what we are going to be because we're going to find these things everywhere. And this might be a simpler way of dealing with it. Taking that model, that template that we're already using, with a diseased population, with a variety of diseases within that particular population.

DR. SHAPIRO: Bernie, then Alta. Alta, I missed you before, I apologize.

DR. LO: First I'd like to say this, the dialogue we've been having, needs to be in this chapter because I think it makes it much richer and much more helpful. So I hope the recording's working. I just want to point out a problem that I'm going to keep trying to flag which is brute, putting on the shoulders of the IRB a lot of important new things. And there's a lot of criticism now that IRBs are having trouble, at least some IRBs, are having trouble doing what they're really supposed to do. I think at some point we need to try and address that, lest what we suggest not ring true. I have a question with regard to the points Alex is raising about, and Arturo raised before, sort of taking into account community harm. Are we saying that we don't want to put this as a regulatory proposal for regulatory change, but we want to put it forth as kind of best practice to be adopted on a voluntary basis by the most conscientious lawful IRBs? Because this is a departure, it seems to me, from current regulatory structure. And I think it's an issue I'd raised before. Are we saying it's going to be such a hassle to change the regs that let's not try and push for regs? Or are we really saying that it's a little too early yet but we ought to try it? IRBs like Mayo Clinic ought to be commended for adopting what I think is essentially Alex's approach, but to really be inspirational rather than regulatory. I think there's a good argument to be made that there are enough regulations as there is. And finally, I'd like to introduce the concept of best practices, that there are a lot...it struck me as I've been thinking about what we've done—you know it's taken a lot of time—but we've heard some thoughtful people, I think, really try to grapple with what's going on. So some of the folks at Heart/Lung, some of the folks at Mayo, the breast cancer group, are really trying to push this forward. And where we can highlight a best practice example without saying everybody has to do it, I think it will give some incentive for other folks to do that. So you know, it seems to me what, what's his name, Vogelstein at Hopkins has done, is really a real-life example of the kind of thing that Alex was talking about. I think what's interesting about that example is sort of a converse to what Allen Buchanan argued in what's now chapter 3, that because we can't predict the future and things that seem real bad now might not be so bad later, it works the other way. The things that didn't seem so bad starts to get really bad. Where people say I didn't mind when I just had Tay-Sachs but now I've got this gene and that gene, and my kids are asking me are we all going to die prematurely. So that I think because it's unknown, the best investigators, the best IRBs are going to want to have a very low threshold for going to the community and at least talking to them. I mean, again, David's principle that let's get the bulk of what's going on. Right now investigators aren't thinking about it and aren't making any effort to talk to anybody who's related. These questions, while we don't know what's the exact representative to speak with legitimacy, talk to anybody and you'll probably get some ideas you've never heard about. It's a dialogue. When this gentleman from Illinois, I was going to say, they never even talked to us, they never came to explain. I mean that's part of what's going on. I think, again, to go back to the AIDS examples, you know it's that dialogue you start and if you really are there and you listen, you go back and think and you lose sleep, but you change what you do over a period of time and it's not a matter of saying who has absolute political power.

DR. MESLIN: Bernie, we may come to this later, but we have two proposed recommendations on page 7, Recommendations 7 and 8, which may get to what you're describing. One is this language of recommending that the scientific community should agree on a set of standard practices. That could be a way of encouraging them to describe what already may be best practices. And recommendation 8 is a way of us referring both the consent document documents that we've received and heard from the others on, the NHLBI, the National Action Plan on Breast Cancer. And that's what was meant by that language. We would attempt to reference them, not to endorse them as models, but to encourage people to go and see who has already made great strides. So the language is there. It could certainly be developed as a textual form if you thought that was appropriate.

DR. SHAPIRO: The...Bernie, I want to ask you. You said two things today which—you said more than two things, but two things are on my mind right now—that stand a little bit, perhaps I don't fully understand, which is most likely the case. You've just said that one of the strategies we could take is to be inspirational and inspire people to do better by best examples, and I think we certainly should do some of that. On the other hand, both you and David have said over and over again that people will do whatever they can get away with, and that leads me to think that inspiration is not, may not be so helpful in this community. But can you...both of you raised this issue and I'd be interested in your observations. I believe in inspiration for the most part.

DR. LO: I think we have a spectrum of things. But one thing is very clear in saying you guys have been interpreting the regs wrong, this is what we think they mean, this is what the implications are, this is how you interpret these slippery terms. And we get 80 percent, 85 percent of the cases just because everyone thought you didn't even have to bother, coded meant unidentifiable. Then as we get to the really tougher cases where we're not wise enough, not experienced enough, or the field hasn't developed enough, rather than trying to be proscriptive I think we should just say keep working on this, keep thinking, keep talking, here's some good examples of people justified to do that. So that it's sort of a multipronged approach where we can be fairly clear and sort of definite about what we think should happen, I should say that. And when we're less certain, we should say keep trying to do what the best people are doing and we'll get to it eventually.

MR. CAPRON: If the requirement were a well-informed review, and then the examples would be well how do you go about getting that well-informed review, you might follow Trish's example sometimes. And if you have representatives of the community, bring them in to sit with the committee, the IRB. Other times it might be telling the researcher he or she has to go out to certain identifiable organizations. And it could be any number of things, these best practices.

MS. CHARO: Two things, as a detail, we can certainly suggest to HHS that they provide the resources for OPRR to actually follow up on those IRBs that attempt to implement any of these in order to learn from those experiences in preparation, perhaps, for regulatory change in the future. I don't think OPRR currently has the resources to do that, and so we should accompany that. More substantially, just so we understand where we're coming out here, I understand that nobody here is suggesting that any form of community consultation gives any member of the so-called community a veto power. But I haven't heard us decide whether or not an IRB should ever in its own judgment decide to deny approval of a protocol because they feel that it is too psychosocially or politically damaging to a larger community, although it otherwise meets all of the necessary requirements for approval. And that's a hard...I just don't think I've heard anybody say anything about it. And so my question is what do you want an IRB to do when they've concluded that a particular piece of research is psychosocially, and I really mean socially, damaging. And I'm thinking about so many, it's not just race and it's not just disease, thinking about Simon LeVay and Dean Hamer's research on homosexuality, which engendered such controversy within the gay community on both sides of whether or not this was helpful or harmful to that community's effort to gain political equality. And I'm really interested in hearing people discuss how they think an IRB should react to something like this.

MR. CAPRON: What's your answer?

MS. CHARO: I'm not sure yet.

DR. SHAPIRO: Could I ask a clarifying question? Are you thinking of cases, Alta, where people, the subjects, have individually given consent already, or are you thinking of cases where they're....

MS. CHARO: No, I'm limiting myself to the moment to Alex's example in which we're working with genuinely anonymous samples. You give Simon LeVay a bunch of genuinely anonymous slices to use. So there are no individuals who can stand as the decisionmakers. And we're suddenly asking the PIs to go to the IRB for the very first time for this kind of discussion. And the IRB looks at it and says, you know, there ain't no good going to come from this. This is only going to pour fuel on the fire, or whatever.

MR. CAPRON: And the traditional view is that you can't stop it for that reason.

MS. CHARO: Absolutely, the traditional view is that that is irrelevant. I can imagine for a beginning, that faced with this kind of political consequence, the IRB could have a much more rigorous insistence on scientific validity of the experiment itself to justify the psychosocial harm.

DR. COX: What a concept.

MS. CHARO: So the bad sociobiological research wouldn't automatically get approved because it doesn't hurt anybody because we'd actually be paying attention to whom it hurts.

MR. CAPRON: But if it's good design, and "bad policy," bad social policies, that's hard.

MS. CHARO: That's the hard case, that's the hard case. And I understand everybody's concerns. Every time I see another study published about why women are congenitally able or not able to do certain things, I feel implicated now because everybody wonders now if I'm able to see a triangle versus a square. You know, so I understand this and at the same time I'm very nervous about censoring science.

DR. COX: But Alex, 99 percent of studies that are bad social policy are even worse science, right? And so if you use that as the criterion, you're in great shape, right? I am very hesitant, though, if you have good science for social or cultural reasons to suppress that. You know Galileo would roll over on his grave. So if you just use the quality science criterion, you will solve 99 percent of these cases.

MR. CAPRON: May I ask just a ...if there were no samples in tissue banks that satisfied the need of the researcher, and the researcher went out to the group in question and said, I need to collect samples, and they said what's the research and he described it, and all of them said well, we're not going to give samples for that reason, would that change your view? They couldn't do it, then, in other words, but now we're saying would it change your view if that hypothetical were applied to anonymous samples which conveniently are there and you don't have to ask anybody about. That's what...in the back of my mind, that's what worries me.

DR. COX: But Alex, I have other...there are other social controls for that. And one of the best examples is in this situation when there was an NIH grant for criminality. Now, I actually...and it got taken away right for that conference. Now that was a conference, and because of that social pressure, that grant, it was basically withdrawn. Ultimately it got re-funded but there are lots of other social controls over things that are going to be good science. So even if it's good science, if it really, you know, gets up people's noses, they're not quiet about it. So I think the IRB is sitting there in the context of looking at the science of it.

MR. CAPRON: You say if it's a political objection it should be handled in the political process.

DR. COX: Exactly, that's what I'm saying.

DR. SHAPIRO: I would ...that's my own view, also. Even though it's not an easy question, that's what I would....

DR. COX: But see, the reason why I'm so comfortable with this is that almost all of these, if we just used the scientific criteria which IRBs are supposed to be about, that these things would go down the drain. The point is is that the scientific quality isn't paid enough attention to. And so the researcher needs to show that, and that's why you have experts on the IRBs that are supposed to judge the time.

DR. MESLIN: I don't know if it helps, but in the report that we'll be discussing tomorrow, we have an extension section on the ethics of study design and it may be useful for Commissioners to review that before coming back to this draft tomorrow to see whether an equivalent section might be inserted into this draft. I see no reason why questions about the selection of subjects which we raise on the report on subjects with mental disorders that may affect decisionmaking capacity could not equally address the issue of why one selects groups or individuals or kindreds. Because the issues of scientific design which David mentioned are already discussed in another commission draft.

DR. COX: And how can you justify it? Because Al Jonsen said it, how many people...I mean the scientists, they may be pretty smart in science, but you know they're busy. They need this to be simple. They don't want to have to read the Belmont Report to figure out what is it they're going to do. So we need clear guidelines on this.

MR. CAPRON: All eight pages of it.

DR. SHAPIRO: Alta, then I want to move on.

MS. CHARO: Trying in my head to kind of summarize how this would work, let me see if I understand this. For research using both samples that are anonymous, meaning unidentifiable, and research that involves samples that are being used with the explicit permission of the sample sources, in both individuals, it encompasses a concern about the community. And second, and contrary to current practice, when you're working without single, explicit human subjects (i.e, you've got anonymous samples and no identifiable person), that the investigators will be asked to go to the IRB anyway, if their research is about examining a salient characteristic of an identifiable group in order to allow the IRB to do that risk-benefit balancing. And it's in that risk-benefit balancing that you'd get the review of the scientific value of the research which is tied up with the integrity of the study it's on.

MR. CAPRON: But they would never disapprove something purely because they object to the potential findings, if the findings are legitimate conclusions from the science as designed.

MS. CHARO: So when you're doing a risk-benefit balancing, any valid scientific finding is considered a benefit that outweighs the risk of social harms.

MR. CAPRON: No, not really.

MS. CHARO: Any valid scientific finding.

MR. CAPRON: Oh I see what you're saying. You're saying overall in terms of there are limits.

MS. CHARO: Right. I mean it could be a completely trivial scientific finding that has tremendously controversial political repercussions. I'm actually, despite this, extreme and on rare case willing to go along with this, but I want to make sure we're really clear what we're saying here. There's a risk-benefit balance in which any iota of scientifically-valid result trumps social harm.

DR. SHAPIRO: Well I like...let me talk to that. I usually phrase this in a somewhat different way and it may not get around the issue you raised, and I know where I'll come down from it if it doesn't get around it. When I try to write or think about this, I think of scientifically valid and scientifically important issues which the IRB has to deal with. Just because it passes a statistical significance test doesn't mean anything to me. That's important in some context, but it has to also has to be scientifically important, that is it answers or is on the way to answering something of some considerable importance to somebody. And if it satisfies that, in my view, this may not be shared by other Commissioners, that's right. This is not a group that should be...this...the IRB should not be speculating and governing the scientific agenda by their particular views of what causes harm.

MR. CAPRON: I think Alta should be writing headlines for the Washington Post because you've taken a recommendation which basically said most of the bad stuff is going to fall out on science crowds, said well, if it's good science, any little tiny finding will justify doing terrible harm to a group, says National Commission.

DR. COX: Harold, can I comment on that, because the standard peer review practice in reviewing grants isn't just if it's scientifically sound. It's is it important? Because that's how these things get sifted. So scientific review, peer review, it has a series of values to it that the IRB is looking at too in terms of the scientific merit of it. And it's not simply validity. Validity's one component of that. But I really agree with the way you put it, Harold, with those two components.

DR. SHAPIRO: We are unfortunately running out of time this evening. I'm going to...and so we do have to adjourn. We're supposed to adjourn at 5, it's now 5:30. And we have other commitments. Let me suggest that we come back and spend some more time on this tomorrow morning. I don't think we've allocated enough time to this, we'll have to somehow spend...now what is everybody's schedule tomorrow morning? Is it possible for us to start earlier or is that, because we've published the agenda, impossible? We can put out a new agenda which gives more time to this discussion. Why don't we... I don't know how successful we'll be, but why don't we try to start at 7:30 tomorrow morning. And somebody can get double caffeine coffee or something somewhere. And we'll try to do these other things as well but I think there's still quite a few things and we're getting to a sense where we can help ourselves here and I want to continue on. So let's adjourn here, let's reassemble tomorrow as close to 7:30 as our ... as we can.