Discussion on Session 2: Who is a research subject and Informed Consent in CRTs
Kathleen Lohr [discussant]: It's a pleasure to be here. I was tickled to be invited. I think I probably owe that invitation to Jeremy but I'm not sure but in any case, I couldn't possibly turn it down, I'm really pleased to be here. I had read the papers for which Andrew was the first author and you've heard a terrific introduction from him of these first two papers that had been published so what I want to do or what I think I was instructed to do was largely raise some questions, some come from other people or the Wiki pages and some were things that I was uncertain of. I think some have already also been brought up already in our earlier session but I think that's the main object for trying to prepare for tomorrow, some things that need to be clarified as the expert panel goes forward and some things that might get put on a sort of the backburner for the next round of all the work that apparently lies ahead. So I'm going to try and get through mine very quickly and then have further discussion or Q & A, mostly Q I suppose, from both all of you and the folks on the webcast.

These were two papers, one is whole exploration of issues related to defining research subjects and the other was the informed consent and so, I'm just going to run through some of the issues that I saw. It clearly turns this definition that's been proposed on "interests" that might be comprised or that need to be protected in order to say that someone is a research subject and you find what those interests are defined in different ways. One grouping has it as health welfare and privacy. Health seems pretty clear, privacy seems reasonably cleaner. Welfare I'm less certain of. Another grouping has it as physical, psychological or social injury, so it's actually kind of the reverse. It's the harms from which one might need to be protected rather than benefits that need to be secured. I had wondered about economic circumstances and Andrew did indicate that in fact, issues relating to a person's economics’ interests, financial interests, may well be something that might need to be added here, it might not fit into welfare if welfare is considered perhaps to be more well being and I think returning to this perhaps in the discussion tomorrow or at a later date whether there are other such interests that need to be spelled out is worth considering. And the reason I say that is because I wasn't absolutely sure that these interests or potential harms are kind of sufficiently well or commonly understood to cover in part, what might be thought of as secondary effects or injuries. We might worry for instance about information that gets out about genetics that is a risk to employment, if information gets back to your employer. In the US, at least, things relating to whether you're an immigrant could now be a serious threat. So there may be some other broader questions and I've kind of brought up a couple from the US perspective but I want to say, more generally, I think we do need to be considering even more international issues. It’s all well and good to think most of this is done or the trials are run by and perhaps funded by, as you will, top tier countries. But what are some of the additional questions that might arise when the potential research subjects are in fact in developing countries, very poor countries and so forth. And then I also, it's sort of the obverse, if said, are there other interests that need to be considered and I think some of these international ones might be falling in that category, but then, do we need some real about what are all these possible interests and if you did that, would it suddenly get to be too long, too unworkable and are we better with some of these open to interpretation categories. I had somewhat similar to the issue I had raised before. Whether there are still additional harms related to being a potential research subject, if in fact certain kinds of information, again, as with genetics and employment perhaps, but, or insurability but if certain kinds of information about members of the participants in the cluster, whether anything about them might become somehow public knowledge and these stigmatizing kinds of problems of STD's, mental health, a number of other things, does that pose a particular problem that needs to be considered, that would be over and above what you might do for just standard trials or observational studies.

And similarly, somebody I think on a Wiki page, raised the question of whether, if the cluster is made up of, let's say, general practitioner's or teachers or other professionals and it turns out that they do collected on them, sort of relieves that they have not been acting in their patients or students or whatever, best interest or otherwise, are departing from accepted professional standards. What happens to them if that kind of information became public knowledge?

Third set of questions and this has now come up several times and I think we're going to keep returning to it, is exactly what to do about certain so called vulnerable populations and there is the whole issue of children which would go all the way through adolescents, I think and I wanted to sort of PS. that the question about minimal risk, which may pertain for adults may not pertain for children. That the requirements for what is minimal risk may be more rigorous and those of you who are trialists might need to comment on that but I think children are a particular issue. You have cognitively impaired individuals who may be elder or dementia or have other sorts of cognitive impairments of one sort of another. I had put here that persons who are institutionalized can be prisons and jails and so forth but also people who are institutionalized, so to speak, in say nursing homes and so forth, so there's another class of folks to be especially concerned about. You have mentally ill patients, seriously mentally ill patients. Andrew comes from the emergency department and others here, I believe, have been doing work in emergency issues, so that, and people who are unconscious and otherwise in severe pain or what have you may also raise some particular problems. But there may be other sorts of vulnerable or disadvantage populations that we need to take into account and again, as I said, I think some particular thoughts about international work needs to be considered there. The last and I don't, you didn't say this Andrew, I think, in your presentation but somewhere in one of these papers is this issue of whether, let's say you're doing cluster randomized trials and it's patient safety or quality improvement or so forth and there's this issue of so called novel health care delivery and I wasn't so sure I quite understood what the issue was there but I wondered whether people actually share a conceptual notion of what health care delivery is in this circumstance in which perhaps patients are your, sort of the members of the cluster, they are not the physician’s but your patients, but what happens when patients become irrelevant and the care delivery is between two sets of professionals and is there anything going on there that needs to be clarified and then I just wondered what novel meant and whether it has a particular connotation for this application because who gets to say what’s novel versus what standard care if standard care is essentially user care or your placebo armor or what have you and if that's going to vary between lots and lots of settings and professionals and so forth, then is there complication there?

Ok, onto the informed consent questions. Andrew has given us the definition of passive consent and raised the issue of whether it's more or less protective of subjects autonomy and I think that problem needs a little more explanation but the biggest question that I had, and again, I think this came from a Wiki page, is what are the regulatory issues or barriers to anything relating to passive opt-out sorts of options versus either waivers or active. And I think the regularity issues differ across countries and I think that may warrant a little more exploration. Then there’s this question of, suppose you don't get a response? Can people assume that if you didn't get any answer from potential subjects, that is in fact consent to participate or do they have to regard no response in order to be that people don't want to be involved or does it sort of depend? And I also wanted to raise the question of what is intended by the language that says that subjects need to adopt the study aims as their own? I am not convinced that is as clear as it ought to be and is good an expectation on the part of subjects, so I think it needs a little exploration but in any case, if you have something other than active consent procedures and if this goal is important, then do these other approaches other than active consent. In fact, undermine that goal. We heard a lot about, in the early session, about gatekeepers and that's essentially what I was trying to get at here. Who can provide that kind of consent? Is it the same as would be for a regular trial when consent is delegated or is, there are surrogates or do community elders and so forth count as providing third party or gatekeeper consent and I think then the question of timing, which Andrew had already raised, is important.

Two last things which don't pertain too much to the two papers we've heard today, or so far, but I wanted to raise them anyway. Actually, this first one is sort of related. it's when would you be providing information and how much information and are you absolutely obligated to do it or is there just some benefit or advantage to doing it but one of the questions that I had, because it sounded like it was an either or situation with respect to is it, are you an intervention arm or a placebo arm? Or say, usual care arm, but there are lots of trials that have more than two arms and then the question arises, if you have two intervention arms and a usual care arm, do the people who are in intervention A arm and intervention B arm, should they get information for both of those arms. Forget the usual care stuff for a moment, because supposing that you learn that you were in the enhanced intervention B arm, you might be very happy with that. But if you're in the intervention A arm and you learn about B, you may be kind of annoyed that you didn't get included in that one. And the other one is somewhat similar. Exactly how much information should be given and when should t be provided? And then there was this question of selection bias, such that in the context of whether you're doing a passive opt-out kind of thing that the waiver of consent is actually a preferable approach to trying to do a passive opt-out that kind of leaves you not know what people really agree to or not. I think that was that, so, thank you very much.
Andrew McRae: Why don't we talk about what we meant by interests and the term interests sort of comes from is based in fiduciary law and that's where we derive the term from, so that the term interest is derived from fiduciary law where we adopted that particular terminology from and so looking at the things that a physician would need to look out for on behalf of their patients so that a teacher would have to look out for on behalf of their students. And so that's where the interest language came from and we did intend it to be interpreted broadly. I'm not sure whether attempting to come up with an exhausting list of all the possible interests that could be considered is really an appropriate task for the panel and I think as you alluded to, that in different settings and in different countries, the interests that might be at stake might vary and I don't think that we could probably ever hope to identify and categorize all of them. So I think if we came up with sort of basic list of common interests in terms of things like welfare and privacy and social economic interests and that kind of thing, but left some discretion to individual research ethics boards, in particular locations, it may be better suited to identify interests that may be uniquely at stake in that particular setting that may be desirable. That's something that the panel can talk about in the next few days.

Question/Comment from the audience: Paul Hebert (unidentified): So Martin, this is payback for putting me off last time, so I'm going to purposely be quite provocative so I warn you ahead of time. So from where I sit, being a researcher, having done a lot of research in very vulnerable populations and cardiac arrest and other in the critical care setting. What I mostly commonly see, when I review clusters and I did at CMAJ, is one, is our clusters designs are appropriate for the question, which you guys didn't not get at, so the issue for me is, would you be better served with an individual based randomized study versus a cluster and the reason that's important is that's where the ethical issues arise? If you choose the wrong design, you're in trouble. You'll see where I'm going with this. If a true cluster is where, if the cluster designs appropriate for the question, then by default, where do you get where I sit, how do you get a collective consent? Because the issue is a collective one, not an individual one from where I sit. If it's a true cluster design, you're trying to affect a population’s based cluster effect, I don't understand why you need an individual based consent under most conditions so the default to me ought to be the other way around, I think. So waived or some variation of that. I guess my other comment would be how much information does one provide the community and what type of community engagement if what I said, number one and two is true, what information do you provide the community ahead of time? Not after the fact, that just ruins the design, causes contamination, selection bias and all kinds of corruption of the appropriate of the science, so I would disagree with that. So the idea then is, what do you provide the community ahead of time prior to randomization, what level of community engagement would be essential and appropriate prior to randomization and institution of the design? And then finally, we have opt-out clauses in Canada, I mean, Susan's code of conduct, our new Tri-Council code, talks about quality insurance versus research. Quality insurance must be done if we were to improve our quality of care. Anything we do in that regard, you would say, oh, it’s sucker search. Well, it's not. It's quality insurance and we need to decide ahead of time what we're going to do in that regard and whether we're going to use these appropriate designs more and more. So from where I sit, I would suggest that the opt-out clauses ought to be used and we need to basically think of it much more differently than I'm hearing right now. So individual ethics are probably less appropriate than collective ethics and there are ways to get at that. Otherwise using wrong design. Andrew McRae: I think hearing a number of things from you and I completely agree that at some point, probably near the beginning of this discussion document, we should probably undertake some discussion of when is the cluster design appropriate and when is it not and that's probably something that many of us agree on and so, as the panel goes forward, I think that's something we should probably think about putting near the top of the document. As far as, I think a couple of the concerns that you raised are going to be addressed or at least touched upon in the discussion about how gatekeepers ought to make decisions about enrolling communities in or clusters in cluster randomized trials this afternoon and sort of what criteria they used to decide whether or not it's appropriate to enroll a particular community or cluster in a cluster randomized design. As far as when do you need to get or when is it appropriate to seek individual informed consent, I think what you need to keep in mind is that, yes, the interventions may be happening at a cluster level but in a number of cluster randomized, in a large proportion to cluster randomized trials. The community is just a unit of randomization and you're actually the level of intervention is actually on the individual community members and so, the, if you hadn't been using a cluster randomized design, you would be seeking consent from those individual subjects. You're just using the cluster randomized design for whatever legitimate reason it might be, whether it's an avoidance of contamination or the fact that it's a community level intervention but you're actually doing something to those individual subjects in a way that meaningfully threatens their welfare, privacy and interest and so there are individual interests at stake in a lot of cluster randomized trials, not all of them, you're right. But there are certainly individual interests at stake in many cluster randomized trials and I think to adopt a stance where you only need to get permission at the community level, I think undermines the autonomy of those individual cluster members.

Kathleen Lohr: One quick question on the quality improvement or patient safety aspects. If in fact you were doing something that was sort of clearly a single facility equality improvement intervention confided to that and never to be published, although I know people don't necessarily use that as the criteria. And perhaps where some of that kind of information is protected because peer review type of information, I think at least in the US, is often held to be non-disclosable in legal situations, then I think you’re making a case that a lot of this might not be so pertinent but if in fact you were trying a major patient safety, and I regard that as equivalent to a lot of quality improvement efforts, and they were being done in multiple sites, different kinds of heterogeneous populations and a variety of other kinds of differences, you might well expect that information, especially if you had multiple arms or types of interventions and then some usual care kinds of efforts. Where the idea really is to inform public policy or to inform professional associations, those that are concerned with quality improvement, then I think these issues will arise more and do deserve, whether it's randomized trials or cluster randomized trials, they do deserve considerably more thought about how to go about getting informed consent. I may not have quite understood your point on the last but I think there are different levels of patient safety and quality improvement activities that do essentially fall under the category of research rather than purely quality efforts internally to a facility or organization.

Question/Comment from the audience:
Mohammed Ansari, OHRI: My question is, I have two questions. One is, is there a difference between legal and ethical issues when it comes to cluster randomized trials? Should we be worried about legal issues as well as, for example, right to descent? Not only a right to consent. The other one is you give an example of bed nets and control being patients were not considered part of research, so I was thinking, let’s say if I were that villager who was not getting a bed net, what I would think is, and let's say the other villagers in proximity to mine and I would think that most of the mosquitoes would eventually migrate to my village and the bloodthirsty one and that could actually impact my outcomes? Thank you. Andrew McRae: So, I think what we're attempting to do is identify a minimum set of ethical requirements and I think it's beyond the scope of what we're attempting to do to try to address all the potential legal issues that may vary from jurisdiction to jurisdiction and even between provinces and Canada. I think we'd have a difficult time addressing the legal issues that may arise so what we're attempting to do is to identify sort of a minimum set of research ethics regulations that may be useful to research ethics boards and regulators and investigators and if legislators want to come on board and adopt what we're proposing then great, but I don't think that we're hoping to address the particular legal differences from jurisdiction to jurisdiction. As far as whether controlled communities, whether citizens in controlled communities count as being research subjects or not, and not being an expert of migratory patterns of mosquitoes, what we're trying to get at is that if the study team is not actively doing anything to a particular group, including and collective indefinable information or doing anything that may otherwise be thought of as potentially affecting their physical well being or what have you, that's what we're using for whether somebody is a research subject or not. And so, to a certain extent, is up to the discretion of the particular investor and research ethics board that is reviewing the cases as to whether people ought to be considered research subjects or not. But the point that we're trying to raise is that the key in, we're trying to clearly lay out who is the research subject to ensure that they are adequately protected and also in laying out who isn't a research subject. We're just trying to ensure that the whole process isn't over regulated and so, I think, probably require some local discussion as to whether considering the control community research subjects would be appropriate or maybe over regulation and in the absence of any definite proof that the mosquitoes are all going to massively flock to the control communities. You can probably say that usual practice and not deviating from sort of usual public health policy probably counts as not intervention at all and in the absence of any other reasons to consider those control member research subjects, then you would have to make a really strong argument that you should consider them to be research subjects and in the absence of sort of a compelling evidence that there's going to be a mass mosquito migration, yeah. Kathleen Lohr: This is sort of a quick response, I think I take a slightly different view than you've heard from Andrew because, but it may fall under the category of are there secondary or effects or potential harms that might be visited on those controlled; now you used the example of would mosquitoes migrate? I had assumed that bed nets either killed them or rendered them sterile or something and it was kind of immaterial but I can think of lots of situations in which people in control communities might either learn about something and decide they want to go partake in whatever is going on in the neighboring village or where certain kinds of things might move human beings. You can imagine some type of cluster trial that tries to deal with drug sales. And if you intervene in certain neighborhoods and you get drug dealers off those particular corners, they are going to go somewhere. I mean, if in fact they go to a controlled place, you have invoked or imposed a harm on them, so it seems to me that you couldn't possibly lay out every possibility for harms in that direction but it may be as part of your guidelines as people who want to try to do these sort of trials to think it through and raise with their IRB's any potential harms of that sort of that might be visited even in controlled communities and groups.

Question/Comment from the webcast: So we have quite a lot of traffic on the webcast and I've grouped things into 4 or 5 categories of questions. So we'd like to come back a couple of times if we can. The first information is Laura-Lee Balkwill from the Research in Ethics in Ottawa, notes from TCPS 2, which is a tri council policy in Canada, does offer definition of research participants which I think the research team needs to look at. Then there's a set of comments which is really about the lenses that I think we've applied and the focus as Andrew spoke on jeopardy so again, Laura-Lee said, “Why has the sole emphasis on our analysis been on jeopardy, which studies that stand to confer benefits not be subject to our scrutiny?”, and then Elaine Collier, whose with an actual stand for resources at the NIH and the US, also I think when Andrew was talking about patients interest being protected by the practitioner in the normal ethical code of practice, said well who decided when interest are in jeopardy, the view of the practitioner might not be the same view of that as their patient. So those are two comments that Andrew would be good to respond to if he can. When you frame the discussion around the research subject, you particularly frame this term in the jeopardy of interest and basically I think the question is, why did you largely focus on jeopardy with studies that stand to refer benefits not be subject to our scrutiny under the scuttle? Andrew McRae: I think that arises from sort of, that may arise from sort of the historical lens of bioethics in the sense that the field arose largely in response to insistence and studies in which subjects or individuals interests were unreasonably jeopardized and so that may be sort of the historical lenses, but I think one of the things in trying to define a research subject, sort of the conceptual approach to it was trying to identify what's the difference between a patient in a physician’s practice and being a subject in a clinical trial and the essence there is that even though in a clinical trial, you are accruing benefits, you are also putting your interests in jeopardy for the purpose of developing general knowledge. And so that's in spite of an independent of any kind of benefits that might accrue to you as a result of participation in the clinical trial and so, that's sort of the approach that we used. So the essence of being a research subject is independent of any benefits that may accrue to you is to have your interests jeopardized as a result of research participation and that's something that, as far as consent goes, is something that people need to take on willingly. I'm sure everybody will gladly sign up for the benefits associated in being a clinical trial but it's the downsides that causes the problems.

Question/Comment from the webcast: Yeah, I think when we talked about the indo, it may be sort of being indirect effects on patients and say quality improvement and knowledge translation trials that target the physicians or the nurses, the assumption was that the normal protections within the health care professionals of the patients relationships, would actually protect the patient from indirect problems and I think the question that Elaine was sort of saying is, well, is that always going to be the case? Can we be confident that the view of the practitioner is the same as the patient and that they can fulfill that role of protecting the interest of the patient within the professional ethics code? Andrew McRae: I guess what we're trying to get at is sort of the role for the physician or nurse or health care professional whose intervened on the setting of cluster trials, if they were to undertake their normal practice, then what they are doing to sort of summarize there is to they're acting generally in the patient’s interest. That's sort of what they do on a day to day basis in terms of that's sort of the essence of their professional activity and the fact that they are being intervened on in an attempt to improve practice doesn’t change that. I don't think that we're expecting that they are consciously taking on the role of a protector of patient’s interests, but what they are doing in their day to day job, independent of what they are intervened on in the setting of a cluster randomized trial, is that they are acting in their patient's interest. Whether or not they take on some sort of protector role relates to studies in which the physician might actually be a gatekeeper in allowing access to cluster's of patients and that’s a separate issue that will hopefully be touched upon this afternoon.

Question/Comment from the audience: Ray Saginur: Ray Saginur, Ottawa. I'd like to ask a question about waiver of consent. Getting consent Andrew, is a hassle. It takes all sorts of time; it costs me as an investigator all sorts of money. Its expedient for me to say, it's just not feasible or if, I won't get consent of everybody and it might introduce a bias, who knows? And it's expedient for the investigator to suggest that, to commute that, to pause that. The question is, if the REB, sometimes it's easy for the REB to agree and say it's clearly not feasible or there's clearly an issue of a bias, what sort of test might the REB impose to access whether indeed feasibility or bias or issues in terms of waiver or consent? Andrew McRae: What tests does your REB use? Ray Saginur: Well, this arises actually from time to time and what I do is I say, show me the money. Go with, we won't waive consent yet, try not to scuttle the process of consent and show me that there is an issue. Andrew McRae: I think that's exactly right, that the burden of proof is on the investigator if they are making that claim, they need to back it up with some stand of evidence.

Question/Comment from the webcast: Opting out for cluster interventions. One is more of a comment from Bruno Giraudeau from France who notes that the very nature of participant consent must be discussed with the nature in relation with the nature of intervention since for some of them, there is no opt-out option for fluoridation of tap water. So I think that was a comment in the first session that we delayed to this session and then, I think Laura-Lee comes back with a more specific issue or comment. It's not always the case that communities have no choice but to opt into an intervention put forward by the authorities. For example, in Winnipeg, there are neighborhoods that refuse to be sprayed with mosquito pesticides due to health concerns. In trials, studying these types of community wide interventions, how closely are researchers expected to investigate the rights of communities or segments of communities to opt-out? Andrew McRae: So I hope we've addressed Bruno's point that the feasibility of consent quite often does turn on the nature of the intervention that's being evaluated and as far as how much we want to give individual communities the ability to opt-out. I think again, that's something that I hope will come up in this afternoon's discussion and later on in the next couple of days in terms of who has legitimate authority to actually make decisions on part of communities or even communities within clusters as the case of the Winnipeg study may be.

Question/Comment from the audience:
Annette Brown: My name is Annette Brown and I'm with the International Initiative for Impact Evaluation. So to give you an idea where I'm coming from, we fund evaluation of development programs in low and middle income countries, so first I want to thank Kathleen for bringing up the point that a lot of issues in the international sphere, there are a lot more issues, probably that have been coming up in the papers and the discussions and I'm not going to raise all of them here but I will raise one now, which is not unrelated to the last point and that is I encourage you to move forward to talk about the cases where there's a separation between the designer and implementer of the intervention and the research team. So the majority of the evaluations that we fund even those that are experimental design evaluations are of programs that have been designed and are being implemented by somebody completely different. So two concrete examples, one is not one that we fund, but region government in China says, we want to try one egg a day in schools in rural communities to see if that improves health and educational outcomes in children. And we understand our CC's so we will randomize across schools as the researchers tell us. We're going to go find researchers, we just want them to tell us, does this improve outcomes or not? But the design of the program and the program is all the government. A second example would be a combination prevention programs for HIV incidents in developing countries as well. These designs are coming from a large array of international donors. The implementation is sometimes government in these countries and the researchers are being brought in often to help with match pair randomization but to do the research, not to talk about the design of the intervention and so, the question is, where does the responsibility lie on part of the researchers in all of these questions in opt-out and informed consent and all of these things? Kathleen Lohr: I would suggest that there are types of studies typically under the rubric of community based participatory research that would invoke from the very beginning some elements of, and it may be needs assessment starting as early as that but certainly including elements of the design of studies in which you bring in the stakeholders from the community. They may be people who are sort of in fraternal organizations, community organization, perhaps religious organizations and involve them from the beginning and that may be a partial answer to what happens when researchers parachute in with study designs in their back pockets and don't really try so much to get at all kinds of issues at the community level, community broadly defined. So I would encourage perhaps some thought to whether these principles you all are trying to develop can be somehow expanded to include the principles of community based participatory research, which is a fairly well known, I think, approach, at least in health research if not drug trials and whatnot, per say. Andrew McRae: I think your comments about specific issues in developing countries that echoed the issues that Kathleen raised about vulnerable populations are well taken. I think one thing the panel is probably going to have to consider in the next couple of days is the issues that have just been raised aren't necessary specific to cluster randomized trials. Tthey would arise anytime an investigator parachutes in with whether it's an individually randomized trial, whether it's a perspective study or what have you, and so we'll need to consider how much detail we want to expend on issues that aren't necessarily unique to cluster randomized trials but are nonetheless note worthy a mention.

Question/Comment from the webcast: We have two comments which are around cluster trials, knowledge translation QI interventions. The first is from Deborah Cook from McMaster University and she says, "Some QI communities and clinician scientists distinguish between a) cluster randomized trials testing intervention that is a strategy. For example, education order feedback versus education loan, where the aim is to try and an effective intervention into practice. From b) cluster randomized control trials, testing an intervention with an unknown effective on outcome. eg, evaluating whether school sex education decreases pregnancy rates and her question is, how will you distinguish consent issues for these two situations in the guidelines? Andrew McRae: So, in one, in the first example, the intervention would be addressed to health care practitioners and the hope that the intervention may result in the uptake of the practice downstream beneficial effects for patients and so, the way I'd approach it is the way that we'd sort of, a step by step approach. First, who are the research subjects? And so the target to the intervention would be the health care practitioners and then as far as the patients go, what, the approach that we've taken is that the patients would not necessary be subject unless they are interacting with the research team in some other way to obtain the outcome either through interviews or identify using identifiable information and so, you would need to get the consent of the health care practitioners but unless the patients are research subjects, then there's no need to seek consent. Whereas, the school based study of sort of novel or implementing sex ed curricula, so the intervention would be targeted at the students and so the students would be research subjects and as far as the teachers go, who would be doing the teaching, the question actually being done to the teachers to get that novel curriculum off the ground, are they’ve been subjected to some kind of experimental education intervention, yes or no. And so, the students would be the research subjects and the teachers may be depending on the type of intervention understudy and so, if there aren't any compelling reasons to waive consent then informed consent from the students would be required and as far as the teachers go, it would depend if they were research subjects or not.

Question/Comment from the audience: Stuart Nicholls: Stuart Nicholls, University of Ottawa. Most of the discussion, I think seems to draw informed consent from issues of autonomy, protecting autonomy. And, but I just wanted to clarify the language you used around autonomy because you talked about autonomist individuals which seemed to relate to competency of individuals but also autonomist choices, which seemed to relate to deliberate decisions and of course, autonomist individuals may not make autonomist choices so I just wanted to clarify that but also, what was it that was intrinsic about autonomist decisions that made it worthy of respect and protection? Andrew McRae: The points well taken that we probably need to be clear about we need to be respecting autonomist choices by autonomist individuals.

Question/Comment from the webcast: Tom McDonald from Dundee in the UK and I think this was a researcher trying to do this: “Should we compare a new guideline policies with previous guideline therapies when the evidence of the new is better than the old is largely opinion based? And delaying the new randomize allows a cluster randomized evaluation that would be a benefit for everyone as we will actually discover what is better. How can this be unethical?” Andrew McRae: So I gather at what the question gets at is what degree of evidence is sufficiently compelling to make the community decide that new guideline is better than the old guideline and what degree of evidence is supporting the new guideline and if it's largely conscious or opinion based, then there may actually be a need for a cluster randomized trial to establish which guideline approach or which clinical approach is superior so that by the time that guideline three comes around, it can actually be based on good evidence. Kathleen Lohr: Quick couple of responses. One is that one assumes that new guidelines will have been based by now on authorities systematic reviews or comparative effective reviews in which the grade for the strength of evidence, if not high, at least I moderate. Perhaps you could get away with low but one hope that in this day and age, you don't get guidelines that are simply good old boys sitting around the table and so, I'm not sure I quite understood the premise or at least maybe I disagreed with it. The other issue is whether some of this also needs to be thought of in terms of delayed start trials in which eventually, somebody’s going to get, or cross over designs, or somebody's, these clusters are going to eventually get what it was that the first people got and whether there are any special situations that ya'll need to be thinking of for that.

Question/Comment from the audience: Dean Fergusson, Ottawa. I was struggling with your definition of research subject. I would think that anybody that contributes to your results, that contributes data actively or passively, would be considered a research subject, not dependant on what they know, what they get, but they contribute to their results and if we don't call them research subjects, what do we call them? Andrew McRae: I guess the, I think what we wanted to be clear about is that, so say hypothetically, you were evaluating outcomes based purely on administrative data sets and not collecting data from, you’re not collecting identifiable information, you're not collecting blood samples or any other, doing any other kinds of physical examinations but if you have an administrative data set that gives gross infection rates, do you need to consider every single patients that passed through the doors of the hospital wards, who has their data collected in an administrative data set, do you need to be considering them research subjects? And are their interests really at stake when you use that administrative data and we concluded that no, there really aren't any key welfare interests or privacy interests or what have you at stake and so we don't need to consider them research subjects and what do you consider them? You can consider them patients but I'm not sure that research subject is the right label for them either. Dean Fergusson: I don't think it's a question about what's at risk or what's a stake cause they contribute to your study or question, so that's your sample frame. And I would think they would be considered research subjects for consent, that's a whole different issue, right? But I think it's an important dissention, otherwise we can't go anywhere.

Question/Comment from the webcast: Ok, this is from John Hermos; whose from Boston and I think this is trying to clarify some of the principles outlined by Andrew. “If the study intervention is deemed no more than minimal risk and/or consent is unfeasible, to what extent does having to collect private information say through perspective medical record review preclude receiving a waiver of consent?” Andrew McRae: So, the question sort of turns around what degree of risk is involved in perceptively collecting identifiable private information from health records and there is some literature and I think some regulation that clearly specifies that type of data collection falls under the minimal risk category. Certainly it may have some implications as Kathy touched upon on these adverse consequences of inverse data release but that sort of historically, and in regulation, falls under the minimal risk category and so, the same for clustered randomized trials. Just as well for perspective observation studies and that kind of thing that consent wouldn't necessary be required if the only subject intervention is the use of identifiable private information.

Question/Comment from the audience: Jonathan Kimmelman: I want to make sure I understand, so part of the purpose of defining the research subject I take it, is to finding the purview of finding the interest you ought to be protecting. So am I understanding you correctly that if the REB makes a determination that a certain population are not research subjects then it is not within their moral ambient to be protecting their interests, is that an accurate understanding? Andrew McRae: Yup. Jonathan Kimmelman: Ok, so then, if that's the case, I guess I would sort of raise as a question whether or not that's really, from a regulatory standpoint, maybe that makes sense, the common is quite straight forward about that, for example, but does it really make sense that REB have no obligations or duties to protect or look after the interest of entities who are not research subjects, but who are nevertheless affected by the limitation of research? So for example, the question, your point earlier about secondary effects, there was a very famous housing voucher study in 1970's where property values of all sorts of people who were not in the study were impacted by that. So surely, someone who is responsible for making risk benefits assessments has to be responsible for thinking though the implementation of that research so if it's not the REB, then who should that be? Andrew McRae: In addition to considering the interest of individual research subjects, it certainly is fair game for the research ethics board to take into account the interest of community, so there's a history and bioethics literature on the interest of communities and so that may apply or may to a certain extent answer your question not completely. And so, but I think there is some, it's an unresolved question in terms of are there other interests that aren't germinated to research subjects and germane to the preview of research ethics boards. I think that's something we can talk about further.

Question/Comment from the audience: Diana Elbourne: The question really was about ethicists. My prejudice about ethicists is its all principle and not thinking about practicalities and feasibility. I'm obvious charactering here. So it's quite reassuring to find ethical framework that takes into account feasibility whether something is doable or not, so ... if that's the case for cluster randomized trials, shouldn't it be so more individually randomized trials, where it's often not so much feasible but very difficult to get what one would think of as being properly informed consent? You can get a boxiest[s1] but you may not get really informed consent and we're talking here again about ethical guidelines for cluster randomized trials and we're taking feasibility into account, which I think we should, but I just wondered how that fits with an ethical framework. And we also talk about developing countries as sort of more of a vulnerable population and that's often, the two are often put together, but in terms of consent issues and trials that I've been involved in, quite often, the villagers in villages are very active participants in the whole process of understanding about a trial, learning about it, giving consent for the village before randomization in a way that I haven’t met in the same sort of way in many developed countries, so there's two issues around there. But stand up for emphasize if they are going to be talking about feasibility. It's great. Andrew McRae: In that case, I hope you won't be terribly disappointed with the final product. You raised the point of you are glad we're taking into account feasibility and I think the thing that we want to be clear on is just as current research ethics guidelines are not inspirational documents and things that you should strive to, but things that you should actually do in practice are research ethic guidelines for cluster randomized trials is going to be much the same. We're not going to set a list of things that people should aspire to, but we are going to set a list of minimal requirements for the ethical conduct of cluster randomized trials and yes, we are going to take into account the restrains raised by feasibility because of large cluster sizes and the types of interventions involved. But what I don't think we're going to be doing is giving people a free pass just because things are difficult. So I hope you won't be too disappointed with the final product.