[This Transcript Is Unedited]

Department of Health and Human Services

National Committee on Vital and Health Statistics

Ad Hoc Workgroup on Secondary Use of Health Data

June 21, 2007

National Institutes of Health
Natcher Building
Bethesda, MD

Proceedings By:
CASET Associates, Ltd.
10201 Lee Highway, Suite 180
Fairfax, Virginia 22030
(703) 352-0091


  • Call to Order
  • Review Discussion from Full Committee Presentations

P R O C E E D I N G S (3:50 p.m.)

DR. COHN: I want to welcome everyone back. I’m sure you feel like you have never left.

I want to call this meeting to order. This is a meeting of the ad hoc work group on secondary uses of health information of the National Committee on Vital and Health Statistics. The National Committee is a statutory public advisory committee to HHS on national health information policy.

We will do introductions just as we do normally, because this is a different work group. I am Simon Cohn, Associate Executive Director for Health Information Policy for Kaiser Permanente, and chair of the committee and chair of this work group. I want to welcome all of you, and also welcome those listening in on the Internet. As we have been struggling with these microphones today, I remind everybody to speak clearly and into the microphone.

With that, let’s have introductions around the table and around the room. For those on the National Committee, I would ask if there are any conflicts of interest related to any of the issues coming before us today, would you so publicly indicate during your introductions. I want to begin by observing that I have no conflicts of interest on these issues today.


DR. CARR: I am Justine Carr, Beth Israel Deaconess Medical Center, member of the committee, co-vice chair of the work group, and no conflicts.

MR. REYNOLDS: Harry Reynolds, Blue Cross Blue Shield-North Carolina, co-vice chair with Justine, reporting to Simon as the chair. I have no conflicts.

DR. HOUSTON: I’m John Houston, University of Pittsburgh Medical Center. I am a member of the committee. I am sitting in on this meeting. I am sitting beside the person who is the co-chair, and I have no conflicts.

DR. GREEN: I’m Larry Green. I have the same status as John Houston.

DR. SCANLON: Bill Scanlon from Health Policy R&D, member of the big committee and the work group, and no conflicts.

DR. VIGILANTE: Kevin Vigilante, Booz Allen Hamilton, member of the committee. I should point out that it has recently come to my attention that there are a number of Booz Allen staff supporting the AHIC quality work group, and some of them were here today. I am not aware of any conflict that this presents, but I would like to publicly acknowledge this situation.

DR. MC CALL: Carol McCall, member of the committee, also one of the reviewers for the work group on secondary uses. No known conflicts.

DR. TANG: Paul Tang, Palo Alto Medical Foundation, member of the committee, no conflicts.

DR. DEERING: Mary Jo Deering, National Cancer Institute and staff to the NHII work group.

MS. JACKSON: Debbie Jackson, National Center for Health Statistics, committee staff and lead staff for this work group.

MS. AMATAYAKUL: Margaret Amatayakul, and I am the contractor for the work group.

DR. STEINDEL: Steve Steindel, Centers for Disease Control and Prevention, staff to the ad hoc work group and liaison to the full committee.

DR. GREENWOOD: Marjorie Greenberg, National Center for Health Statistics, CDC and Executive Secretary to the committee.

(Whereupon, the remainder of the introductions were performed.)

DR. COHN: I want to begin by thanking all of you for being willing to participate and work on this effort. We have a lot of work to do between now and September. Clearly it is going to be a full summer for all of us.

The purpose of today and tomorrow is to review the charge, and we began talking about that before we broke from the full committee, but also discuss the work plan, get your input and feedback on that. Since we have two sets of hearings coming up within the next six weeks, we do want to get your thoughts about testifiers and topics, and to make sure that we are moving in the right direction, but we also need your expertise to help assist us as we are putting all of this together.

We will be adjourning at 5:00 p.m. today, plus or minus ten minutes probably. Then the plan is to reconvene tomorrow from about 9 a.m. until about noon, when we will adjourn for the day.

Yes, Debbie?

MS. JACKSON: A quick housekeeping note about the transmission over the Internet. We are live today, but we will not be on the Internet tomorrow morning, in case anyone is listening in. Welcome for today.

DR. COHN: Thank you. With that, I am going to turn the session over to Justine and Harry to lead us through the activities and discussion. Thank you.

MR. REYNOLDS: All of you have your handouts. We started as our discussion document; this is the draft background. What we thought we would do first is, if you look at page three, those are the excerpts from the original request from ONC. As you keep those in mind, we are going to want to spend whatever time we have left today, and we will go through them chronologically and we either make it or we don’t, because we have time tomorrow, we will start on page four, which is our scope, just to make sure if there is anything significant that you don’t like about the scope or have questions about the scope or other things that we want to get on the table, we want to do that.

Page five is just related work and collateral documents. We tried to list the kinds of things here. For example, you have heard us mention on some of this the AMIA information and so on. So there is the collateral we are going to be using.

Then we will move to page six, which is the axis of interest. What we really want you to help us with there is, this is what we are asking the people to talk about. In doing so, if you are looking at what we are trying to do, let’s make sure that this brings out from each of the testifiers — again, we don’t have any real pride of authorship or we are not going to protect this, we want to make sure that this really — when somebody walks in here, they are going to want to know what we want to hear. They are going to structure it in a way so that, if they are presenting something, we are going to be able to say, against what we have asked for, where are you. Some of us could start wandering and doing other things, which means we are not going to get to the bottom of what we want to make happen.

Then we have the list. It is a separate document of the testifiers, starting to take a look at the types of people that would be testifying, and make sure that if you have any additions, changes or corrections. We need to continue with the aggressive agenda. We need to make sure we get these people in here and we give ourselves a good solid slate of testifiers to do the job we want to.

So with that, I would like to have everybody turn to page four. I don’t think we need to go through them again, they are up on the wall if there needs to be any discussion. I would reiterate one more time, today and tomorrow we are building the structure of what we are going to do. We are not stamping everything that is up there going yes, that looks good. That is why we have got a working session, we don’t have any testifiers.

Yes, Simon.

DR. COHN: I wanted just to talk about the tasks for just a minute. I was talking to Margaret about this before. I was looking at them when you were presenting them before. I think we know what we are doing, I think, but I’m not sure this represents it as well as I might like, so some clarity around this I think would be helpful.

For example, and this is how I put things together as I was trying to present this to the full committee both earlier today and yesterday, in item one task — and framework means whatever you want it to mean, but having said that, we were talking about a policy framework. I think we are talking about a conceptual framework around this issue of secondary uses or whatever we want to call this.

I don’t know that it is secondary. It may be uses. We as a committee need to investigate and come up with a framework that is helpful to us as we begin to look at policy and conceptual definitions of how all this comes together.

I tend to think of that as what we are trying to do. I tend to think the elaboration of the types of uses of data almost becomes a taxonomy question. We want to do a taxonomy and a definition, but I just hold out that that overall framework is — once again, I am grasping for words, but when we started talking about this, there was a tension. I think the Office of the National Coordinator is very interested in having us emphasize the issues of quality, quality performance, quality measurement, quality improvement. But I want an overall framework that helps us lay out the issues for all of this, and then we begin to make decisions and get more specifics in terms of quality.

As I said, I’m just not sure that this quite represents that overarching framework piece as an important piece. So when Carol says, why didn’t we do research, it is my view that we are not going to dig deep and try to make determinations on all aspects of research, but where possible we will try to make things that might apply across the entire continuum here the pieces that make sense in terms of policy.

Anyway, I’m not helping because I am not wordsmithing this, but I just want to make sure we are all on the same page here.

DR. GREENWOOD: I really agree with you on that. That is why going above the tasks even, I was a little surprised that the premises both were just focused on the quality use case or agenda.

I think of the premises at a much broader level for starters. That is, I can’t come up with eloquent language, but data are being increasingly being collected, clinical or health care data increasingly are being collected in electronic form for a number of uses, including principally patient care but other uses as well. There are many uses for these data. Yet, in order for this to happen, you need adherence to standards and appropriate privacy and confidentiality protections, whatever.

But it was just, that to me is the high level premise. A second level would be quality assessment, reporting, et cetera, is one of those important uses. But in order to meet any of these, you need the adherence. So I think you need a first high level premise. Then I would agree that the framework picks up on that.

Then in the taxonomy, I would include the different types of uses, rather than just defining things like anonymization, et cetera. It is my understanding Amy is working on such taxonomy, so we are not starting from scratch, and then go on from there. So I guess that just elaborates on what you said.

MR. REYNOLDS: I think you are absolutely right. Again, we are all learning this quickly. If you listen to what John Lunsk said today, I think you start to drill it down. You go all the way up to the NHIN and then you come down to an NHIE, then you come down to an EHR, then you come down to all the different secondary uses, which is the taxonomy that we had from Amy. Then you go into one of those which may be quality. But the point is, it is almost opening umbrellas.

So we put this together. We are growing this body of knowledge. Every time we hear a new presentation, you get a different sense of how this body of knowledge ought to be laid down. So that kind of structure that allows you to make sure that we never lose sight of the overall goal.

As we were talking awhile back, when you capture one piece of data, whether it is the temperature or other things, can it be used for all these use cases and all these secondary uses, or am I going to do it one way for quality and then I am going to collect it 12 other ways for the other ways that I would want to do it.

So we are in total concurrence. As this stuff was prepared we had one set of knowledge, and we are continuing to build on that. So I am in total concurrence with what has been said, and now we have got to figure out how to structure so that we neither stay too high nor dive too deep, so that we actually put together something that is going to be useful to help the quality use case go forward, but not necessarily handicap ourselves and making sure we are keeping an eye on everything.

DR. MC CALL: A couple of comments to what is on page four. The first is, I don’t know if it is in one of these tasks or not, but would be the creation of a set of guiding principles. I think the framework is important, but if you think about the topic, the headline, uses of data or secondary uses, regardless of the taxonomy and the ways we categorize a particular use, it would be wonderful if this activity could put out guiding principles around data stewardship, that anybody who is doing this type of work would be guided by certain of those, regardless of the specifics of the work. So I think that would be good to have.

The second is for this group to develop a point of view, whether or not there will be a scope of work along the lines of what it means to enable uses of data and secondary uses. There is lots that goes in there.

My concern would be that the output becomes another set of work that describes what it takes to enable it. There is a lot of work going on about what it takes to enable uses of data. Maybe the role of this group is to assume that we have a can opener. Assume that it is enabled; now what? Simon is saying no.


DR. MC CALL: Then what I would suggest is, if that not be the case, talk about what it means to enable it, but don’t leave the conversation there.

The second part of the conversation, I believe, must be, once it is in fact being enabled, that there have to be with the definition of stewardship some boundaries and rules of the road. That is where we started to get at the very end of the meeting; will there be entities that will be allowed or disallowed, or will all be allowed but then be required to certify as certain standards.

I think that those are some important frameworks where there are some discussions and decisions somewhere in someone’s mind to be made.

Then the last comment, and I agree with Marjorie; back to my earlier comment about quality, I think quality can be a very important concept to anchor things and pull out of the clouds. However, we have also had conversations at NCVHS that quality also serves a higher purpose around population health.

In that first premise you had up there, it talks about capture and analysis. Yet, the last day and a half we have talked about exchange and aggregation and linking, and other concepts that take a very broad view, and be specific about what it means to enable. It is actually linking.

DR. DEERING: I wanted to pick up on what Marjorie said and Harry, you said it as well.

First and foremost, from a simplicity and efficiency point of view, many of the solutions and recommendations that would be pertinent to quality are going to be generic across all of them. It is not just starting with as you said how the data is going to be collected, but then it is going to go beyond that. It is going to go to all aspects of it, how it is aggregated, if it is aggregated, who can aggregate it, what are the guidelines by which they aggregate it.

So I suspect that — I am just guessing all of a sudden, that maybe this might look surprising like the NHIN report, in that there are a bunch of common things that are really rich and keep and standardized, that peel away some of the simpler aspects that you can see clearly, and then down below is a lot of the complicated stuff, and then we do drill down deep into quality as a use case of it.

MR. REYNOLDS: Simon, you had made another comment, whether we call it secondary uses. I thought Paul had made an excellent recommendation to us that you have data for the actual care of the patient, and then you have other uses as a thought process. So rather than just say secondary, maybe that is the way to approach it.

DR. COHN: I am going to comment, since you are asking me. I want to get additional testimony to find out all the different ways people think about all this stuff. This is not just framework, this is actually definitional, and how you define the building blocks sometimes moves you into either natural or unnatural circumstances.

Once again, I just think we need to make sure this is the way we want to define things, because at the end of the day we are talking about uses of clinical data. The question is how we chop them up and what is helpful to us to begin to make recommendations and put things together.

MR. REYNOLDS: Kevin, you had a question?

DR. VIGILANTE: No, I just want to affirm my concurrence with the current direction we are going. I have a hunch that research will be a special dimension tightly coupled to quality, if the premise is that quality is based on evidence and that it is methodologically sound research that produces the evidence from which we develop quality measures.

So going up the food chain, research is indeed a secondary use of data that contributes to the quality of care. Depending on how far up the food chain you want to go, it falls in a special way within this context.

DR. HOUSTON: To dovetail on what Simon and Paul have said, I think it is important that sometimes it might be easier to think about secondary uses are not. My fear in listening to all this is that when you try to define what it is, there are so many things that it could be that you either miss something inadvertently or you just miss things.

I think it would be helpful to describe what a primary use would be, or what would not be a secondary use, either because it is something that is expressly permitted. Then it also might be helpful to make yet another category of things that aren’t permitted uses, but also aren’t secondary use.

I think there is going to be a group of individuals that look at what is happening with the NHIN as being an opportunity for a land grab. They can get at data that they never had access to before, maybe thought as a way to get at data that maybe they weren’t permitted to get access to before. So I think there is a value in trying to segment and define this in a number of different ways.

DR. SCANLON: I am very sympathetic to starting with a very broad focus. It is like Mary Jo says; we maybe are back to the NHIN. We are not just focused on improving health, but we are also improving health care, we are interested in improving the health care system.

If we only focus on quality, we are jumping in the middle. One of the things that we don’t recognize today is that as we are desperate to try and do something about quality, having ignored it for so many years, that what we are doing is so limited. Part of our limitations is what Kevin mentioned about research.

We don’t have — and this is a non-clinician speaking — we don’t have the evidence base to say here is how we should measure quality in great detail. There is a huge issue in terms of who should get what, and we don’t have that to be able to do it. We need to develop that kind of information.

One of the things that is happening now is that — and this relates to the quality group hearing on Tuesday — we are trying to develop some of that information, living in a world short of the electronic health record. We are trying to use administrative data to improve our knowledge through CMS’ coverage with evidence initiatives.

So I would start broad. We may end up after we have played out a framework, then talking about the quality application in the context of what we know about quality today, but at the same time caveating that that is very inadequate compared to what we need to know in the future, and how there needs to be another application that helps us develop the knowledge base for the future.

DR. TANG: I hope I am making a constructive comment. We have so little time and so much to do, it seems like we want to be — as with this meeting that just recently ended, we want to be very productive with our time, and spend the least amount of time on marginal gains.

I think a lot of marginal activities is when we spend a lot of time on definitions, where it would be easier to draw a bright line even in an arbitrary way. As an example, secondary use has no definition that we can all agree on. One of the things that Harry mentioned was, there is use to take care of individual patients for their individual benefit, and then there is everything else. If we could just use that bright line, one will end the, what is a secondary use and not, get rid of that.

The other is, if we could stick with things that have a precise definition, research has a federally defined definition, it has a definition, generalizable use. So that might be a kind of use that we could cordon off and look at it.

Quality I think has a fairly crisp definition. Then if we could cordon off all things — Amy called it commercial uses, all things that generate money that aren’t for the direct care of patients. I am just trying to find something, draw some kind of bright line, then let’s share that with our testifiers so that we don’t spend any time arguing about definitions that do not carry the ball forward. I’m sorry if I am pleading on that.

That is part of a framework that I would find useful as we carry on our work for clarifying who we need to hear from. One of the things that you wanted to know, Simon, is what is the scope of other uses that are out there that could be helpful. To the extent that we could compartmentalize it, then we can start figuring it out.

I do think AMIA Kostrick was fairly helpful. If it is not for the benefit of the patient, then we have higher hurdles for them to go past. They need the consent of the patient. We can decide this later; we will get a report from AMIA. It just seems like we want to find ways of compartmentalizing as much as possible to get through the setting here.

DR. GREENWOOD: I am sympathetic with the case that Paul is making, but it makes me very uncomfortable to take that approach, because I think it would come back and haunt us in various ways.

First of all, I really do not like the term secondary uses or secondary users. I think we are talking about uses and users, and they are not always aligned. But to the extent that we can talk about uses and users and say which uses and which users we are talking about, we are much better off.

There is a hierarchical implication of the word secondary, and maybe there should be in some cases. But that can create problems in developing policy. How do you define what is for the benefit of the patient? The patient might actually benefit more from some research that is being done on their disease than on any treatment that they can be given at that moment. You could make that case.

There are elements of data, and I have been thinking about this a lot, that you can collect for business associate population health or public health purposes that maybe there is some value in knowing it about the individual patient, but there is much more value if you are getting it in a standardized way about a group of patients.

Something like race and ethnicity, you might really feel for treating that individual patient, and they don’t have anything that is genetic, but isn’t that useful, but you are collecting it in the patient encounter because you have a commitment to send that data to the state public health department.

So it breaks down. As soon as you take that, I think it breaks down. I won’t go on. I could go on on that, but I won’t. But I caution you on that type of a distinction.

DR. TANG: I think you may have misinterpreted what I said. One, I am not making any judgments on the other uses besides the today use in patient care, so it does not imply that there aren’t other uses of a patient’s data on today’s care. It is just a way of saying, the patient walks in thinking that I am going to share information with you so you can do something for me today, and there are other uses of that same data which still can accrue benefits to that individual patient. I am just trying to draw a bright line anywhere I can.

DR. STEINDEL: Harry, I was going to pass until this conversation just started.

MR. REYNOLDS: Do you want your time or not?

DR. STEINDEL: I want my time. I thought Paul articulated the situation fairly well. Basically there is uses of data, period. As Marjorie pointed out, secondary uses of data has a connotation to it. It also has a lot of baggage to it.

We have also heard some other baggage terms around the room. John said permitted. That is a loaded term, too. It comes from the HIPAA world. I think we need to be careful about using a lot of these overloaded terms, because what we are trying to do, and I think Carol articulated it fairly well and it was agreed to, is, develop some type of — Simon called it a framework, but as we have discussed it, it is really much broader than a framework, but something that we are going to hang the policies and directions of the uses of health care data within the construct of the NHIN.

Now, Paul has done a little bit of differentiation, and I agree with his differentiation. There is health data that is used for the care of the patient at that particular moment. It is basically laying on of the hands that the provider does. Then there is all the other uses of the data that comes from that practical use.

What Paul is suggesting, and I agree with him, is that that is a very good demarcation of where we are not going to be articulating things, and that is the care of the patient, and where we are going to articulate things, and that is the other uses of the data.

MR. REYNOLDS: Justine, you had a comment?

DR. CARR: I’ll pass. No, I forgot, I’m not going to pass, sorry. I wanted to — not everybody has seen the AMIA framework. The two-day meeting, we talked about four quadrants, quality, public health, research and other, which included commercial. Then for each of those things we talked about characteristics that would frame the decision about use of that data. Those were accountability, transparency, patient consent, cost of re-identification and oversight, regulation and law.

So if you wanted to use data to look at your practice of patients in house getting diabetes care, that is in the quality domain and it is covered by HIPAA. So the issue of patient consent would be low because it is already covered by HIPAA. If you are Google and you want to take stuff and put it on the Web, that needs oversight and also needs a high level of patient consent. We talked a little bit about this last night. So I think that kind of look at the area and then say where are the checks and balances I think is helpful.

Harry also added to that list of accountability, transparency, consent, cost of re-identification, oversight, regulatory and then data standards. But I’m not sure about that, actually, whether or not you have standards. I think these others are more in terms of, if you have high oversight and it is a covered use, then the consent is lowered.

MR. REYNOLDS: I’ll make my rebuttal later. Simon, you’re next.

DR. COHN: I want to pull us out of the weeds for just a minute here, because I think we can immediately get in and start arguing things which I think we really want to get testimony on. I don’t think we need to decide a final definition of secondary versus whatever versus whatever, because I think we can make good progress getting information regardless.

Paul, I know you were responding to my comments about frameworks and definition. All I was saying — and I don’t know which allows us to make better progress, but I see there are two frameworks out there, the framework such as you were describing, and then believe it or not, the HIPAA definitional pieces that we ought to review and reflect on to see if that helps us. I think that is something we ought to get testimony on.

Having said that, I don’t think there is anything about that that prevents us from beginning to think about the quadrants of research or quality of public health and the other bag, which is something that AMIA began to develop very well.

I think it is more as we get more information reflecting on where the cleavages help us figure out policy and all of this, which is to my view once again something I think we may be smarter at after we have heard from AMIA, after we have heard maybe a briefing reminding us of how HIPAA cleaves things and chunks things up, and where the pros and cons of that framework are.

So anyway, I just bring it up as one of the things we are going to have to decide on.

DR. HOUSTON: In looking at the quadrants that Justine just described, that concerns me a little bit, for fear that something doesn’t fit nicely or easily into a quadrant. My thought is that there is another way to look at this also.

By the way, before I get to that, I think it is not treatment that we need to think about in terms of what is a primary use, because I think all along, payment is also a primary use that needs to be considered as such.

But back to my point, rather than using quadrants, there is a sliding scale of, I am going to say user consent, that maybe is another way to look at this. There are some things that don’t require any user consent that by their nature are permitted. I know permitted use is a loaded concept, but there is a scale that goes from no consent required to some level of user consent or a proxy for user consent such as through an IRB, all the way to the other end of the spectrum of uses that are never acceptable. As a society we say those uses are never acceptable. I think you can almost tag the different types of secondary uses somewhere along that continuum.

I think it might be meaningful to think of that concept of where does it fall in terms of how much consent or proxy consent or some concept like that, where does the use actually fall.

MR. REYNOLDS: Paul, I’ll let you comment and then I have a comment. Are you you or are you Carol?

DR. TANG: I was instructed to pass on comments from Carol. One comment she wanted to make is that a definition for other uses is anything for which any use of data that is not for the purpose it was gathered. So it is the same thing as when you walk into a doctor, and you are giving information. I’m just repeating her.

The other point she wanted to make is a request to build the problem statement, in other words, for each of the testifiers to state what the purpose of them gathering the data.

DR. GREEN: Is it too late to rename this?

MR. REYNOLDS: We want your input.

DR. GREEN: What I hear, there is so much harmony in the conversation about what I hear in everyone’s comments. What we really want to do is to figure out how to optimize the value of health data that is collected in routine practice. We have got a list of what we think we might want to use it for from public health to commercial reasons, from research to quality improvement. But it is not so much about secondary use versus primary use, is it? Isn’t it about getting to value, getting value out of the data that we are collecting? Isn’t the idea behind the notion of secondary use saying, we can get more out of the data that are collected to take care of patients, but we need to do it in a correct way and a right way.

So anyway, for the good of the order, I think the people you invite to testify could benefit from being urged to testify about what they know about what it is going to take to get more value out of the health care data machines that are running in the country and that are going to be enriched by electronic health records and all this sort of stuff.

So for your rejection, I think you should start talking about the NCVHS and the ad hoc work group on optimizing the value of health data collected in routine care.

DR. SCANLON: Just to react, not that that is not a nobel goal, I think that there is an issue that we are trying to draw a boundary. There is a role for electronic health records that can improve care in the context of an individual physician’s office, can improve care in the context of the physicians that treat somebody in the hospital to treat somebody. Then there is the role for this information beyond the treatments.

I think in order to keep our task manageable, we are going to be always beyond the treatments themselves. We are going to be into something — and I will avoid the word secondary, but it kind of fits — we are into other. That is the only reason I think it is the issue of the manageability of the task, not that your point isn’t value that we want IT for all these purposes.

DR. GREEN: So use Justine’s four quadrants to optimizing for.

MR. REYNOLDS: AMIA had a little bigger list. Those were the quadrants, but they had a bigger list.

Margaret, if you would put that chart up that we did. I only want to use it for thinking. I scribbled this down, but it is not for prime time by any way, shape or form. But as we are thinking about this, you hear everybody talking about drilling down and so on, and forget the words, look at the concept. You could change those words to what are the overall things we have to consider, what happens when you transmit data, where is it captured at sites and what are the use cases. You can change those words any way you want to, but we are looking at layers.

So whatever we say is good for all these uses goes at the top. That is part of anything we are going to be doing, whether it is data standards, whether it is privacy. The NHIN obviously — if this stuff is going to be moving throughout this country for many different reasons, there will be some overarching policies and all these other things that will have to be in place. If your institution is doing quality with another institution and you are passing it back and forth, there has got to be some overarching thing, not every single use case having its own definition of how we do everything. Otherwise it is not going to get — so I am trying to take it up there.

So there are certain things that will play in there. Our privacy letter for example is being used in a number of the RFIs that are being sent out, because if you are thinking about what you are doing, here is how you think about it. Your next level down is, where are the types of places and what are the types of functions that are going to be out there that are going to group this data up and/or group people up and/or group the institutions up, and what might they be doing. That is, using John’s term only for discussion, the NHIE’s, or it could be data stores, it could be some research group, or it could be whatever it is.

The next one down is the actual EHRs that people may be collecting. If you think of adoption and you think of having the data available to do any of these uses that we talk about, individual doctors’ offices are going to be collecting it and putting it somewhere, large institutions are going to be collecting it and putting it somewhere and so on.

Then you go down to the next one, which is, what are the uses. That is that bottom list. As we get to adjudicating this, what will be the differences at the bottom level of that, that would have us tell ONC or tell anybody what are the specific things for other uses that are going to be different than everything else that is going to be done in this whole process that we are dealing with.

Now, this hasn’t been vetted in any way by you and by me, but if you listen to this discussion, we have some discussion about going down to that bottom line and coming up with the things you have to do. Somebody said earlier, but aren’t those general things that we have to do anyhow.

I’m not anxious to debate this, I’m throwing it out just like everybody is throwing their comment out. I am throwing it out as a thought process.

DR. TANG: I’m still trying to simplify it so that we can be finished in the summer.

MR. REYNOLDS: But I wasn’t try to complicate it.

DR. TANG: Well, for me it got complicated, because all I care is about the use, how do you intend to use it, and how can we be respectful of where it came from, the patient.

DR. VIGILANTE: I am very sympathetic to the way Paul is going here. I think what is our charge today and tomorrow morning is to organize our approach to this and to be able to get direction to the testifiers so that we can be informed by their wisdom and their practical experience.

So we don’t want to over manage it or micro manage it. We want to have them — but we do want to have them in appropriate channels of thought so that they can inform us regarding the scope. I see what Paul is saying here as giving them some swim lanes in which to think about certain categories of data. But I would present it to them as a hypothesis.

We are going in thinking there is quality, there is research, there is public health. I would even say there is commercial, and then there is maybe other besides that. Think about these categories in the context of what you do and what you would like to do. If there are other categories that we have missed, please suggest them. But this is a way to organize your thinking as a testifier.

I think overthinking this and going beyond that is more than we should be doing.

DR. SCANLON: Let me defend Harry’s characterization.

MR. REYNOLDS: Let me defend it. I’ll defend it.

DR. SCANLON: No, because it helped me a bit. It doesn’t have to become too complicated, but for me it raises an important issue which is, if I have got this set of uses at the bottom, what do they imply about what comes before. Paul has raised this in terms of, here is a use, but I want to respect the patient’s privacy.

But we have talked in other meetings about the fact that the privacy issue is not absolute. There is the social value of something, and there are tradeoffs involved. I think that when we identify the uses that we want to put the data to, they raise questions for what would be required in order to be able to accomplish that use, and in some respects at what cost.

If there is this issue that we are going to be seriously compromised by some set of standards in terms of achieving a use, is there a way that those standards should be modified so that the use can be achieved.

So for me, this helps because it tells me where we are going. It is more than a specification of the uses, it is a specification of the conditions under which the use can actually be realized.

MR. REYNOLDS: Let me just make one other comment on this. Many of us sit on committees. This is an ad hoc subcommittee, but the subjects that — a lot of us have been here all week, and we have dealt with about any subject that you could put on any charts you want to put it on related to this.

So yes, we are going to look at this through the filter of other uses, but it will immediately show back up as subjects in the other things that we are already doing in one way or the other. As we sit in privacy and confidentiality, we have to think of privacy and confidentiality on whatever picture you draw, whether you like mine or not.

So if we get goofy over privacy just for quality or just for other uses, and we don’t consider it in thes other things, then I don’t have a problem with that, this subcommittee will have done what it did, but then we will have actions about this subcommittee in our other groups taking it up to see if we can endorse it from our other views because, does it play across the whole picture or does it just play for the other use.

That is why the thinking. It is bouncing it back and forth, it is not staying high. It is going where you want to go. But if you don’t use that as a filter, then we may come out too myopic in what we want to get done. That is where I was heading.

Other comments?

DR. GREENWOOD: I wasn’t at the AMIA meeting, I was out of the country, but I wouldn’t have been invited anyway, probably. I have just seen this for the first time today, and it is interesting to me.

I had a question. It was triggered by what you said about respecting the person from whom the data came. In defense of your what you might call simplistic dichotomy here, we wouldn’t have this data if somebody hadn’t appeared to provide it. We don’t have another way to get this data for the most part, not in an electronic health record. You could have other kind of data going into it, but we wouldn’t have it if the patient weren’t there, and we do have to remember that.

Yet, as Bill said, we have all agreed that privacy is not absolute in the sense that it isn’t the highest goal. But there is in the care process the strong mandate, do no harm. Some releases of data can potentially be very harmful to people, whereas others have so little risk of harming them, even though you may have affected their autonomy by not asking them.

I am wondering why risk or potential harm or whatever was not one of the lines here, and if that came up.

DR. CARR: I would say yes, it did come up in the discussion. I think the thing that came up, the opposite of that, is do good. Think about the potential good in the face of somebody finding out Harry’s blood pressure on one office visit. I think somebody gave that example and said, it would be breaking confidentiality but there would be no harm done. Then you weigh that against what is the good that comes out of having these data together, usable and leveraged for quality, not just of the individual, but of groups.

I think I heard that there are tradeoffs. There is no easy answer. But if there has to be tradeoffs, do them toward the greater good, and try to avoid the serious harm, but to be guided by the good.

MR. REYNOLDS: Let me jump in for a second. We have put on the table a number of inputs for us to consider. That four quadrant thing has created the discussion that we were looking for. If it needs to be a ten quadrant thing for us, if it needs to be whatever it is, I think it was a good starting point. You can start from nothing, or you can start from something that people have spent a lot of good time on and have some good thinking on. It may be over inclusive or not inclusive enough for what we are trying to get to. So I don’t think anybody is endorsing it, Marjorie.

No, your comment is perfect, because risk is a good way to look at it. So I put that out only to create the discussion. We could have come in here and said, let’s everybody start from zero. There has been some collateral put together, there have been some significant efforts gone on, so that is where we are using the basis.

So back to our earlier discussion, no chart that has gone up is endorsed, no document that has gone up is endorsed, and no mention of anything is endorsed. It is collateral into this committee to decide what it wants to pick up, use or not use, use as a filter, use as a fact, use as whatever it is. So that is where we are headed.

So rather than debating whether somebody said this or said that, adding to what we want it to be and creating what we want it to be so we adjudicate what we want to do, that is why this session today and tomorrow morning is not a bunch of testifiers, because we have got to figure out what we want to do.

So this is exactly the debate we had. So keep your energy up on the debate, keep focusing on what we want to do. Keep being willing to agree or disagree in a professional way, because in the end we are going to have people coming to see us. When they finish seeing us, we are going to have to do something. I want to be sure that when we do it, it is what we all agreed up front we were trying to get to, and we adjust as we go along to make sure we have got it covered. That is where I am going.

So we are not endorsing a single solitary thing in putting these things on the table.

MS. BICKFORD: Let me toss out the question, are you really talking about the transformation of data into information, knowledge and wisdom? Is that really what you are talking about?

MR. REYNOLDS: I think yes, that would be a statement, but I don’t think it is the whole picture. Once you do that, then it gets moved around, then people touch it and they do certain things. So yes, for the data itself, I think that is probably true.

DR. DEERING: A couple of things. First of all, just to emphasize that whatever we recommend, I don’t think we do ONC a service by recommending a separate siloed fix. Whatever policy recommendations we are going to make, certainly there would have to be a small carveout for any one use case, but you don’t want to go modifying HIPAA solely for one purpose. So I think that just underscores what you want to do there.

I liked your hierarchical approach there too in terms of buckets. I wanted to just hold out something I don’t have copies of here, but I can share with you tomorrow, which is a way that our big group is beginning to look at things. We are trying to also classify things by four types of sensitivities, one of which is intellectual property, one of which is data sensitivity for privacy, one is IRB and another is sponsor restriction. So those are different buckets that you would use.

But then, within each of those buckets we have a high, medium and low sensitivity. So what we are doing is taking this concept from AMIA, high to low, but then bucketing it according to these other types of sensitivities. What we are hoping to come up with but we have many more months of work before we finish it off, is looking across all of these, anywhere where it is a low, anywhere on this where it would have been low, can you make some policy recommendations, bucket them all in there.

We are calling it the easy pass system. There is little sensitivity here, it is an easy pass, go with it. We only are thinking right now about a three-tiered structure, and the high end is where there are obviously lots of sensitivities.

Whatever purpose it is in any of those categories at all, it is all high, or at least, there are some key ones that are high. Those are always going to have to have one-off, bilateral types of stuff.

Then the real value added though is in the first tier, because you are just peeling away the low-hanging fruit. People can stop worrying about it. But in the middle, if you can clarify bunches of approaches and categories of the way that some of these difficulties congregate or group, and then present menus of choices, that is what we are doing.

So all I’m saying is that since there seems to be some resonance with this, if you think also of adding another layer of that and saying for the low what do we do, for the medium what do we do, for the high what do we do, that might be a useful construct, too.

DR. CARR: I really like that very much. It was about where I was coming to, and also building on what Paul had talked about.

As we look at the uses, what is the use, what is the risk, and borrowing from you, what is the type of sensitivity, what is the sensitivity. So use, risk, sensitivity, benefit and protection. If we looked at things with those questions just to organize it, because I think we can’t just be focusing on risk, we have to think about the benefit. Then with the risk, it doesn’t mean we don’t take risk, it means we put protection in. But if we had some set of questions like that and then applied it to these areas, I think that would help begin to build a body.

I like your idea about low risk, easy pass, if we can get consensus.

DR. DEERING: Just one other thing that I am remembering from the past hearings. Getting to the issue of what exactly do you instruct your speakers to do, what I am hearing that I think we did before, at a minimum they are not to talk about exactly how they do it, at a minimum. What they are supposed to do is take some broad constructs that we give them and say how do you fit in there.

MS. AMATAYAKUL: I just wanted to ask Justine, you talked about use risk, sensitivity and benefit and protection and how they apply to these areas. What are these areas?

DR. CARR: I think that we need to think a little bit more about that. But if we start with what AMIA had, they had quality, they had research, they had public health, and they had what they called commercial, but we found that commercial cut across many things, so other.

DR. TANG: I think what Mary Jo said was complementary to our previous discussion. I am going to try to formulate a question. It is not a headline, it is a question. This would be the question that I would imagine ONC would ask and CVHS. If we agree that that is the question being asked, then we would turn around and ask the same question to the testifiers.

The question might be worded something as follows. What policies, enablers or restrictions would facilitate the effective other use, non-patient care use, of personal health data while protecting patient privacy? I would view — in simplistic terms, I would view the question being asked of us is to elucidate what policies, enablers or restrictions would facilitate the effective other use of personal health data while protecting patient privacy.

So it is somewhat agnostic in terms of why it is used or what is the benefit. It is just saying what would make it more effective. A facilitator might mean — one example of a facilitator is, let’s use data standards. That is a facilitator. A restriction might mean you are going to need patient consent. And a policy might mean — a policy is about accountability.

I’m not getting a whole lot of resonance in the facial features of the group.

MR. REYNOLDS: Yes, but finish your point.

DR. TANG: From a list of questions to ask people, I think that is a very good question to ask, to allow them to focus. I would also like to ask the corollary question, instead of enable, restrict. You had restrictions in there. I would like to facilitate their use. I would like to see the corollary question that also asks the same type of things that would prevent their use. I want to know it from both points of view. So what policies, enablers or restrictions, so you could flip it both ways.

So if I were not allowed to — well, at any rate, I think we agree.

DR. CARR: Maybe it is impact. You said what policies, enablers, restrictions would impact the effective use.

DR. TANG: Would either facilitate or impede.

MR. LANDAU: I am Morris Lendel with ONC. Rather than taking words out of your mouth, we cannot try to rephrase what ONC may want to ask of this group.

I’ll put it in broader terms, broader context. That is, what makes electronic health information different from the status quo, what rules of the road would you consider that makes it different than the status quo. I am putting it in very broad context.

I am with the Office of Policy and Research, Jodie Daniels’ shop. She is on maternity leave. I am just putting it in a different context than what you are asking. I am talking about uses and restrictions, but that is the overall umbrella that we are thinking about regarding policy decisions.

DR. COHN: Morris, thank you very much for piping in here. Paul, I have got to say, as you were talking, I kept saying, the answer is HIPAA. I think Morris is maybe bringing up what we really need to hopefully be providing guidance around. What do you think?

MR. REYNOLDS: Morris, will you make sure that Margaret gets your words, please?

DR. COHN: Morris, would you do us a favor? You are actually staff to this activity, so why don’t you sit at the table with us?

As I was listening to the question that you were bringing up, I kept going, in about 90 percent of the cases, the answer is HIPAA to your issues, because there is a broad expanse of law that already exists about enablers, restrictions or whatever. The question was to my view what is different in this new environment, which is why I liked the reframing of that question.

DR. TANG: I guess I did not see that. I think the reason we are here is because most of these entities are not covered by HIPAA. If they were, we wouldn’t be posing this. So in some sense, yes, I am asking to fill out the HIPAA questionnaire, but for folks who aren’t covered by HIPAA.

DR. COHN: Maybe we are talking about the same thing then, because I don’t think we need to be necessarily talking to absolutely everybody about garden variety uses of data that they would already be saying HIPAA already covers that.

DR. TANG: The reason for the bright line that I was trying to propose is, let’s dispense with that. We don’t have to rehash what it is that you are allowed to do with HIPAA.

DR. COHN: I just want to get a little clarification. I sat through the AMIA meeting. I was there with the definitional discussions and everything like that. There are many secondary uses of data that are quite well covered with HIPAA, that are handled by covered entities, et cetera.

Yet, the way the definitional work was, those were all considered in the context. So I am just wondering, was it your intent to open up discussions on every single secondary use of data? Now I am listening to you, and you are talking about what is different now, what are these new unforeseen uses by unperceived holders.

DR. TANG: If we go back to this, if you are making an other use, in other words, not to individual patient care, then you need to discuss it. If you happen to be a covered entity making that use, then you might already be covered by HIPAA and we can dispense with a lot of these other things.

MR. REYNOLDS: I was going to play off Mary Jo’s. If you think about it, HIPAA defined treatment, payment and health care operations. If you have implemented HIPAA, then there is your philosophically easy pass using Mary Jo’s comment. If you are using it for any of those and you are HIPAA compliant, and you are willing to hold yourself accountable to that, then you may have done so. So that is using that as an example.

DR. COHN: And as I said, maybe the issue is also that there were uses since HIPAA was developed that maybe were not contemplated, which is the edge areas. This was a useful interchange.

MR. REYNOLDS: Any other comments or questions on this? I think at least that gives us — again, this is why we don’t have testifiers. We want to kick this around. Now we have got to come up with something that maybe we could work with Margaret and come out with something tomorrow that tries to put this together.

DR. GREENWOOD: This is interesting, this last discussion. It is very useful. But is there a significant question that needs to be explored as to whether all these quality activities are in fact part of health care operations or is the issue more that as Paul said, a lot of these entities aren’t covered by HIPAA?

I have looked at the quality use case, but I can’t remember, I haven’t looked at it real recently. Is this more about doing the quality assurance reporting and all of that outside of the covered entity, or is that part of it?

DR. HOUSTON: I think the assumption is that we would be talking about if it is was stuff that HIPAA permits a covered entity to do. I think in large measure, there may be some discussion, but I think the issue is that all of this falls outside of the concept of permitted use.

So to your point, I think the whole concept of health care operations is — I always felt was not a secondary use, but rather was almost a primary use.

DR. CARR: Remember, one of the groups that spoke, Lou Diamond from MedStat, who is a vendor, commercial, for profit. What he does is take your data, aggregate it and give it back to you. Is that covered?

MR. REYNOLDS: He is a business associate.

MR. LANDAU: There are a couple of points. The first is, I want to remind the group, I don’t know if you are familiar with the CPS work group to the AHIC, but their recommendation was that anyone who directly participates in a health information exchange, they would be subject to the same standard. Everybody who directly participates would be to the same standard as a covered entity.

That doesn’t mean definitionally that they are a covered entity. I am going on a higher plane for something for you all to think about. The notion was that everyone who participates in a health information exchange would be subject to the same rules of the road, regardless of whether or not they were a quote-unquote covered entity as defined under HIPAA. Or at least, they should be held to the same standard.

The other thought is, simply because something is not a covered entity, they could be a business associate, which under the current construct, they are just subject to the contract provisions. So if there is a breach and someone’s data is misused or not disclosed. The only duty placed upon them is that they cure the breach, as opposed to a covered entity, who is subject to higher penalties.

So I just want to throw that in the mix in your thought process. The other thought is that some folks, as you mentioned, Justine, are not a covered entity and are not business associates. It depends upon the functions that they are doing. So they may be outside the construct of HIPAA. What you are talking about are possibly three different constructs, three different entities, that may be using secondary use and subject to three different standards.

It seems to me that as you all are developing your policies, that may be something in the mix, because it doesn’t seem equitable or fair that entity A is using a secondary use but subject to three different constructs. So I just wanted to add that in.

DR. COHN: I am very much agreeing with what you are saying. I think I am going back to an old slide that was up there, which talked about — there is the piece that you are describing, which is the NCVHS. We now have two letters out describing much the same thing. So there is a commonality of interest and policy around all of that, which may again be presented in this thing.

I was going back to something that Margaret had written up there about a question that had to do with the status quo. I was just observing, I am wondering if the issue — and maybe this is a John Houston question also — if whether or not the issue is with electronic data use as the status quo, we are talking about non-contemplated by HIPAA, or things that are borderline.

It is like, there is the HIPAA world, there are things that we have been able to stretch, and then there is the non-contemplated. It is like there is gray and outside, and we need to somehow get our arms around both, which are not all secondary uses of data but certainly some subsets.

MR. REYNOLDS: Much better said. When I drew the hierarchy, those things go up to the top and then if you fall into some of those, you don’t even go down to the bottom and have to worry about it.

DR. STEINDEL: Simon, thank you for that clarification. That was going to be one of my comments. Without the clarification, thought that was a very naive statement, but with the clarification I think it makes a lot of sense.

What I originally wanted to comment on was what Marjorie was noting a little bit earlier about quality, and are we talking about internal use and is that covered under HIPAA, or are we talking about some other use.

I think that is a very important question. I think it is a very important question with respect to the ONC use case. I think it is something that needs to be decided. But I don’t think that should be the objective of this ad hoc work group. I don’t think we should delve into that question.

MR. REYNOLDS: Say that a little more succinctly.

DR. STEINDEL: The specificity of what is going on within the ONC use case, that is going to be looked at by parties outside of this work group. I think we should be looking at a higher plane. I think we may receive some testimony that talks in that area, but I think what we should be using is that testimony to make higher level recommendations, not recommendations to the Secretary about, it is the opinion of the NCVHS that with respect to the ONC use case, this means.

MR. REYNOLDS: The only way I would personally temper that is, since they do want to move forward with that after you went to your higher plane in your discussion, I would want to lay it over top of that to say, are they going to have to create another subcommittee to move forward, or did we give them enough to move forward.

I didn’t say go down there. I’m saying overlay that as a filter. Otherwise we hand it to them and they turn around and go, now I am going to get a committee to think about it.

DR. STEINDEL: I think we will be addressing a lot of the needs of the use case without going down to that level.

MR. REYNOLDS: I didn’t say go down. I am saying, use it as a filter, use it as a consideration. That was where we used our rhetoric all along. We are going to discuss this as we would normally with something that comes to NCVHS, but we will make sure — you used the word aspect or filter or whatever we want to have it be, that we at least consider that, so that we haven’t left something completely unattended, that they can go forward.

It is five. What I thought is — John has a comment. Kevin, you go ahead and then John.

DR. VIGILANTE: Just so we don’t lose it for tomorrow, I liked Justine’s classifications. These are just things to hold onto either from the point of view of giving direction to testifiers, or just questions that we want to have testifiers testify, depending on which way they are going. We may ask them to address these issues such as the ones you have listed before, Justine. Just to capture that stuff is worthwhile at this stage.

MR. REYNOLDS: Again, one of our goals is also to build collateral so that we can use it even in the questions. That is what this whole debate we are trying to set up here is; what do we want to have in our own hands and everyone is going, I didn’t think about that, let me ask that question, or they didn’t go there so let’s take them there. I think that is great.

Simon, what I would like to recommend is that we go ahead and adjourn. I think we have had a good debate.

DR. HOUSTON: Can I ask one question?

MR. REYNOLDS: I’m sorry.

DR. HOUSTON: To answer the one point about the CPS work group, the recommendation regarding the fact that all entities that contribute to the NHIN be covered entities or have some formal status in that regard, since I am on the CPS, let me say, I don’t think it was intended for this purpose, but rather it was intended to insure that there was some standard that we could apply all entities to so that some entities couldn’t go out there and freelance and do things that were so beyond the pale, and we couldn’t have any reach to go get them when they did something inappropriate. So I just wanted to make sure of that clarification.

MR. REYNOLDS: Okay. What I would like to recommend, and please, Simon or Justine, jump in, obviously we have had a good discussion on this. Margaret has captured some notes. We could spend a minute restating what we did here, then go into the axis of interest and make sure, now that we have all discussed this, what do we want to ask these people, because I think we added some new things today.

For example, I know on the axis of risk now, I would want to ask them, state whether or not you are discussing things that would normally fit into covered entities. As I have listened to it today, I have got some things I would want to change on it. Then we go into the testifiers. So by the end of the morning tomorrow, we will have made sure we have at least adjudicated those three items, and we will have a plan to go forward with. Does that make sense?

So with that, we stand adjourned. Thanks.

(Whereupon, the meeting recessed at 5:10 p.m., to reconvene Friday, June 22, 2007 at 9:00 a.m.)