Discussion session 4: Vulnerable populations and Gatekeepers
Jonathan Kimmelman: So, first I want to thank the organizers for inviting me to comment. It's a real honor and privilege especially after such an impressive event. Every talk has been really just incredibly illuminating, thought provoking for me in particular. I spend most of my time thinking compelling at the opposite end of the spectrum in terms of research, I think mostly about early phrase clinical trials, which are the least cluster-like of any realm of research and so for me, this is quite an alien terrain but really quite interesting in terms of thinking through some of the conceptual overlaps. What I want to do after disclosing that provision about my limited expertise here is to try and think a little bit outside the box of the issues that are encountered in a context of cluster trials and to think about what it might mean for the role of gatekeepers and questions that I think are sort of a little bit flapping in the wind of the analysis that at least I've read in the paper. So let me go through these.

First of all, there's an issue of incentives. In most realms of clinical research, these are conducted at academic medical centers; part of the whole enterprise of academic medical centers is the production of evidence. It strikes me that in the context of a lot of cluster trials, these are being conducted in settings that do not have a knowledge production as one of the priority goals and so there's a really important question of how one incentives the kind of behavior and conduct in a non research setting and that obviously has many implications for the validity of a study that's ability to meet its objectives, it's powering, etc. Clearly gatekeepers are a critical role in terms of correcting or aligning the absence of incentives and to me, that raises some interesting questions about what the proper relationship is between research teams and gatekeepers. In the very first talk, someone raised the question about induction inducements of gatekeepers or inducements of research subjects. Given that incentives are not part of their, they are not incentivized to do research, how does one properly get institutions and organizations to align their orientation in a way that is sufficient enough and sustained enough to produce the kind of evidence one needs?

Second issue I want to think a little bit about is the issue of transaction costs. So when we consent a patient, when researchers consent a patient, there's some cost of the actual going through the informed consent and these transaction costs that lead us to worry sometimes about whether consent is so cumbersome that it's perhaps in some circumstances, we might want to wave it. It strikes me that in the context of these cluster studies, you have very high transaction costs for consenting a cluster. It's very expensive to, I would gather, although you can tell me if I'm wrong, to accrue a site to earn their trust to go through the plan with them, sufficient that they're confident to hand over the people in charge, say, for data collection or interventions and that to me raises some interesting questions also about the obligations and duties of gatekeepers if these be the researchers. We don't normally think of subjects of having obligations to researchers but here, sometime that's not addressed in the paper I think and how much can a single paper on the subject address, there's lots more to do. A renewal of a grant perhaps? But here's the most important question, early on, someone made the, described the problem when clusters withdraw consent. Given how power is in these cluster studies and how costly it is to accrue sites, the withdraw the single center can frustrate the powering and thus, the sacrifice or the burdens that other clusters have endured for the cluster of that knowledge, so that raises questions about whether gatekeepers, once they have actually enrolled their site, have in some way, an obligation to maintain their enrollment in some way by virtue of not comprising the commitments that other clusters have made to a research program.

Third issue, let me skip the third in terms of getting on with time, another issue I want to mention really quickly here, is the issue of, the difference between, in regular research non cluster versus cluster. It strikes me that in cluster studies, the idea here, I think, to intervene in a system. There are many different actors in a system. You poke it here and it has effects all over the place and many of the talks have been trying to clarify which points in that system matter morally, who protects what, what modes within that system do we not have to really worry about protecting, etc. But this sort of strikes me as sort of a concern or problem and I sort of raised this early on with the discussion about the impacts on populations that are not human subjects but you poke the system, you get the health care practitioners to apply intervention, it presumably impacts the interest in some way of many different actors in the system and one somehow has to incorporate that into the moral calculus of that study. And the question is, if the IRB's for example are not responsible for protecting those non subjects and you don't have relationships then who is protecting the other recipients of the impact of the intervention within that system?

And I want to close by asking one last issue that I think is a bit of a contrast here. In a lot of clinical research, at least in drug development, we have regulation, mainly the FDA or Health Canada. And that acts as a very powerful check on the kind of information that is produced in studies. Of course there's limitations with FDA review, etc, but however imperfect that is, it's an important check on the quality of the research. In contrast, in a lot of cluster studies, I take it that these are not regulated in the covenantal sense, there's no sort of government entity necessarily that is acting as a quality check on the quality of the design of a study and the implementation. That means that the burden of maintaining that quality falls entirely on the shoulders of the investigator team as well as on the REB team. But I wonder also whether gatekeepers, in some way, have some role or responsibility in terms of assuring not merely that the resources expending by the institution are well expended and well put in terms of supporting the research but also in terms of reconciling the overarching objectives of the institution or whatever the cluster is with in some sense, the goals of the research. So I want to close by just mentioning a thought experiment. I think it's useful to think through these frameworks and come up with some sort of challenging scenarios that might compel us to sort of revise or rethink or adjust some of the frameworks that we propose. So here's my scenario. Imagine you're a pharmaceutical company and you want to develop a more effective intervention for getting clinicians to switch their prescriptions to your drug, drug A. You know drug A; it's very well established that drug has a comparable safety and risk profile to drug B so the issue of equipoise is not really a major issue. Perhaps the REB has reviewed the protocol and made a decision that the pharmaceutical company’s drug really does have a comparable risk profile. So, arguably, the institute interests are not necessary put in harms way by a hospital deciding they are going to allow their practitioners to be randomized to a detailing intervention and a non-detailing intervention but I would argue that probably academic medical centers would want to think twice, even if there was a net gain to that institution in terms of revenue production, say, for them. But they might want to think twice about allocating their resources accordingly and it seems to be that scenarios like this sort of raise questions about, I guess, how gatekeepers should orient themselves with respect to the research objectives. Ok, I'll stop there.

Jeremy Grimshaw: I really like Jonathan's poking of the loon and trying to work out what happens around here. One of my concerns is as we start to look at the kind of direct and indirect effects and say we need to protect the concerns of individuals who may at second or fourth hand be influenced by something that happened a long way away from them. As a trialist, you start to make it infeasible for me to do the trial and I think we need to find the right balance because these trials when they are well done, can generate important information that will benefit many people but by thinking, by putting forward the kind of protection of any possible person that may be impacted, by the trial as being the thing we have to protect against, we really like, I think there's a real danger that we make these trials completely unfeasible so I'm not sure what the right balance is but I think one of the things we're trying to do is find a way in which we can have that discourse cause at the moment, it becomes very easy to think, “well my third cousin twice removed might actually have a lousy phone call for me cause I've had bad big toe surgery” and therefore, maybe we need to protect that individual and therefore, we can't do the actual research and I think that's fundamentally one of the challenges that the trialists often feel that ethics REB process can be over restrictive and you have to think of 27 different harms that might happen to people, whereas I think the canvas REB's clearly have concern. So that's where I think a lot of what we're talking about it trying to find a practical way forward so both, whenever possible, the rights we have can be done.
Jonathan Kimmelman: I agree. I guess I have two responses. The first response is that I think that we err when we assume that the burdens of protecting the various actors within a system fall entirely on the shoulders of a single node in that system and I think part of what’s very compelling about the framework you have addressed is you have found a way to develop a division of labor within this system of protection, that makes to me, a lot of sense. I think probably is a more, a richer division of labor to be developed, so I guess I would want to think about the different actors and the different responsibilities in the various sort of lines that go between the various actors and that is again, worth many more years of work, that is a complicated issue. The other issue I think I want to address really quickly is what are the criteria? So clearly, you poke a system, you get these ripple effects and the moral gravity of the impact of those ripples attends with the distance. Now, with that metaphor, what do we mean by distance morally speaking? We surely can't mean special distance and I also think we also don't mean by that the proximity of the intervention. If I get you to give a poison to someone else, the effect is intermediary doesn't launder me from the moral harm of encouraging the delivery of a poison to a person so I think one sort of has to think about what are the various moral criteria that we use to describe some kind of moral responsibility in that system?

Question/Comment from the audience: Thanks, it's Sarah Viehbeck from the Canadian Institutes of Health Research, the Institute of Population and Public Heath and thanks very much for these presentations. Just a couple of questions around gatekeepers. I think you alluded to this in your presentation Angela, but the process by which gatekeepers are actually deselected, and I'm just wondering how we know when we have the right gatekeeper for access to a particular cluster? And you did give a nod to gatekeepers as having a potential role to assist with the dissemination of research results but one thing we haven't heard a lot about today and perhaps you’ll handle this in your deliberations over the next few days, is the feedback process of research results and to whom these results should be fed back to in terms of either the gatekeepers or perhaps the participants in the study themselves. Similarly, when you speak about vulnerable populations and these may or may not apply depending on the nature of the study but the interface between the guidelines that you're developing for cluster randomized trials and other guidelines that exist, for example, the existing guidelines for research with original people and the nature of community engagement that may be required there and similar the forthcoming guidelines for research conducted with children and youth. And so how your guidelines for cluster randomized control trials may interface with other guidelines that exist and may be a particular concern for your area. And then finally, and I'm wondering if you might be able to comment on this but we've talked a bit today about the notion of research subjects and the notion of identifiable data and privacy and I'm wondering how about this applies at the cluster level in terms of animosity at the cluster level and whether gatekeepers might actually have a role to play in agreeing for the cluster itself to be identified or not. Angela White: So thanks very much for those, I'll try, I'm not sure I'll address all of them but for a couple of them, but for the question of identifying gatekeepers, I think that's a really good question and it's, I think, to a large extent, it's going to be somewhat context dependent to figure out whether a gatekeeper actually has the authority or not but one of things we wanted to emphasize in the paper I think is that. So first of all, rarely if ever will a gatekeeper have the authority, legitimate authority to consent on behalf of individuals in cluster trials that are administrating the intervention at the individual level. And at the, with respect to the cluster trials, there, so when I laid out those 4 points, the third and fourth points were meant in a sense to be conditions that so, to the extent where it's unclear whether this falls within the preview of the individuals role and to the extent that it's uncertain whether the constituents within the cluster or the constituency actually do accept this as a role on the part of the person we're considering as gatekeeper. That has the potential to diminish the authority, the legitimate authority of this person to act as a gatekeeper. So, part of the whole, that these things can help to address the question of whether an individual you're considering using as a gatekeeper actually have legitimate authority to serve this role. For the other questions, sorry, I can't remember the other questions you had. Charles Weijer: No those are very helpful questions, thank you. I just wanted to make a bigger picture/comment as someone who’s done research ethics for a long time. Until this project, I actually never encountered the concept of gatekeepers, so I just wanted to identify the uniqueness of gatekeepers in the cluster randomized trial context and it's a very interesting function. I liked very much Jonathan's description of the virtues of a gatekeeper in part being part of a distribution of labor in terms of protecting the various interests at stake. That's a very nice way of putting it. As I read the cluster randomized trials ethics literature, I get the sense that a lot of the imperious for gatekeepers was actually arose out of concerns for the protection of individual interests in cluster randomized trials cause of various difficulties of obtaining informed consent either for randomization or for cluster level interventions. I think the effect of our work has been to diminish the need for gatekeepers to protect individual interests but I think importantly, there are roles for gatekeepers to play to protect cluster level interest, community interest and also, interestingly, organizational interests so when cluster trials take place in hospitals or nursing homes or schools, that there important organizational interests in terms of staff time, resources, impact on other programs and I think that role has not received enough emphasis within the literature, so there's a lot to say but I just wanted to make that big picture comment.

Question/Comment from the audience: It's Kathleen Lohr from RTI. I thought those two sets of comments are ones I'm hoping somebody got done because they were very useful. We've heard several comments throughout the day about some of the language for the projects as a whole and I know I kind of perked up at gatekeeper because in certain circumstances, and they’re more historical than not in the US anyway, gatekeeper is a very negative notion and it generally has the connotation of people who say no. So, I would just encourage you to think that if you're going to define or use the term and you may well want to sit with it, that you've defined it pretty carefully. One of the issues that I didn't understand is why the person had to have legitimate democratic authority? I would have thought, I don't know what democratic is being constricted with and I think I would suggest you may want to lose that term. I think, let me stick with that and if I have some other thoughts I did about language, I may just send it on to Jeremy or everybody. Angela White: Just while it was on the front of my brain, catch it. So I mean, that's a really interesting point. So I had somebody else had raised the question of where the language of gatekeeper came from and I'm not sure if this holds true for others but for me, gatekeepers, I had just automatically associated it with in Canada, we have the role of gatekeepers in sort of general medicine and in that rate, general practitioners served the role of gatekeeper or are thought to serve the role of gatekeepers in sort of, you have to see the general practitioner in order to get anywhere with specialists and move on through the system. And so, I'm not sure that it has that kind of connotation of being necessarily or automatically associated with no, so that's a really interesting point to bring up. Kathleen Lohr: I think you're probably safe with trying to keep the language but just remember that for some people, gatekeepers are going to say no. Their whole predilection is to say no on behalf of jurors of facilities or whatever. So you'll have to kind of work around that. Whether or not you have some kind of other language you want to try to use, not necessarily for this concept of proximity or surrogate decision makers, that kind of thing, because I think particularly for children and perhaps incapacitated people and so forth, there's the gatekeeper for the cluster and then there's a set of people who may have to say yay or nay and they are operating somewhat different kind of context and on behalf of different entities of the facility or the setting or the ward or whatever. Angela White: Thanks very much for that and for your second question, so the reason for inserting the democratic bit into the democratic legitimacy was function I think, my background is in legal philosophy where legitimacy can just refer to the obligatory reinforce of the law or regulations or what not and that may have nothing to do with the justness of the law in democratic terms. And so, it was kind of to highlight that we're not just talking about cases where it is written that a person has the authority to make these sort of decisions and the people are, there are obliged, they are following along with that but there's actually something about this institution that resonates with them enough that they are going allowing with their own accord with the gatekeepers decisions and that sort of thing so that's why having the democratic in there, I don't know whether we want to keep that language or not. Thank you. Charles Weijer: Democratic is good. I just want to expand on that a little bit. For me, one of the really, there are a number of difficult challenges about cluster randomized trials. The interventions at the cluster level oppose all kinds of really difficult problems. The way that the unit of randomization, experimentation and observation can pull apart causes really difficult problems but one of the other things that raises is a core source of challenges, is the fact that cluster trials involve a really heterogeneous set of social groups. And the moral status of those groups is kind of not so worked out, right? There's a literature thinking about communities and the involvement of communities with organized political structures and elected leaders and that sort of thing in research but cluster randomized trials, yes, they involved community sometimes but they also involved sports teams and hospital wards and schools and geographical areas and neighborhoods and trying to think through how to account for group level interests and people who might speak on behalf of those groups in various ways, I think is one of the central challenges of this project. One of the ways in which people may speak on behalf of a group is in some sort of authoritative sense of giving permission for that group to participate in the study and in effect, giving consent. We think that is only really legitimately applies in a narrow set of cases where that individual has legitimate democratic authority to provide consent on behalf of the group, but there are other roles for gatekeepers and really important roles I think in terms of consultation and involvement and consultative process it seems to me, can play a very important role in protecting cluster level interests but don’t require the gatekeeper to have anything like sort of legitimate political authority. That requirement only attaches to basically the function of being able to legitimally give consent on behalf of the cluster or not.

Question/Comment from the webcast: The first comes from Laura-Lee Balkwill from the Secretary for Research Ethics here in Ottawa just reminding us that Chapter 9 of the new Tri-Council Policy has a lot to say about looking at sort of the role for community decision makers and the relationship between community decision makers and the needs of the members of those communities. The second is from Matthew Liang in the United States who does again cluster trials in the VA system in the area of mental health and just asking whether we would consider listing or thinking of acute mental illness as a separate cause for vulnerability separate from being an incapable adult cause quite often, people who are mentally ill may have the capacity to make decisions but may at the same time be vulnerable for other reasons. So we may want to consider adding that to our list of reasons for being vulnerable.[Panel nods affirmation]

Question/Comment from the audience: Annette Brown from 3iE. So I'm going to bring up some of the development issues again, because when I hear gatekeeper and I think about the context, I sort of cringe. So, first of all, thinking about legitimate democratic authority, whether or not you include the word democratic, even trying to define legitimate authority in the development, obviously, is very tricky. You can spend as much as your paper trying to defend your choice of this gatekeeper as meeting that definition as you could spend doing anything else. I also think that approaching them for any kind of consent as you suggested, could raise other ethical issues because for example, most villager’s leaders in a lot of low income countries, consent is going to be granted based on whether you pay the appropriate research registration fee directly to that person. And so, you have issues around what are you willing to do in order to get consent from this gatekeeper. My even greater concern though is this crossover between these kinds of authorities and your vulnerable populations. A lot of times these populations are vulnerable because of these authorities. And so by involving them, whether through consent or not, you’re giving them information that could jeopardize the interest of the people who are supposed to be benefited by the research so if I think about for example, female family planning interventions, the authority, the village elder, the mayor may say, oh of course, but now that person knows this is going on, I need to tell other people, the men in my community, to make sure they are not partaking of these services. Or female genital mutilation would be another one where by involving these people, you could actually be hurting the interest of your vulnerable population that you are trying to benefit so I don’t have any answers but just that the whole concept of introducing this gatekeeper raises a lot of concerns, I think, in the development context. Jonathan Kimmelman: So this was actually another issue I had that I wanted to raise that I omitted but the question of involvement of gatekeepers who don't have the moral standing to stand in as gatekeepers and I think initial, our impulse to say would be they have no bearing, they have no role whatsoever in initiation or conduct or cluster studies but it strikes me that part of the concern in some of the settings where these studies are undertaken, especially in conditions where you have extremes of inequality between the recipients of the intervention say and the researchers is that one needs to establish buy in from the community and I think that the paper has discussed the role of gatekeepers in terms of establishing that buy in but that may mean that one may be involving gatekeepers who don't have any moral standing, there may be in fact a moral obligation in some ways to be engaging gatekeepers who don't have any moral standing in a sense that number one, that the entire research project may not be launchable unless they are involved. But secondary, if you screw up in that setting, you potentially make it harder for the next researcher who comes along to do that investigation and that seems to me that one wants to be very careful about one dynamic one establishes in that setting and that may mean engaging actors who in a way, are sort of morally quite, not necessarily clearly linked to the actors, to me, it seems like an interesting question to be thinking about here. Angela White: Yeah, I think that's really helpful and that is part of what I was thinking about with the third case study that I introduced, so the idea there was in part that, I mean, people may have legitimate concerns about involving these community leaders as gatekeepers who have the ability to give access to this population and I think that's entirely reasonable. But it may also serve, one of the points I was trying to make was, it could also serve a really important role in terms of promoting the values that are underlying this research and the sort of attitudes that seem to be kind of coming in with the research that could be beneficial to the research subject participants. The other thing I wanted to say in response to your comment that was really helpful to kind of bring attention to the challenge of even talking about democratic legitimacy in some of these settings is that I think that's certainly a good point to keep in mind but, our strategy hasn't been to kind of introduce these roles and endorse them for gatekeepers but rather to look at the roles that researchers are giving to gatekeepers and try and say something about when it's legitimate to that and when it's not and so hopefully, says something about look, yeah, if you haven't met these sort of condition, as I've called them, these success conditions for democratic legitimacy in these settings that you may well be very wary of giving these people the authority the permission to conduct these trials in these situations.

Question/Comment from the audience: Abha Saxena: My question is in fact, my comments are related to the role that you describing to gatekeepers beyond the permission to conduct the research and something you said, the gatekeepers might identify individual who meet eligibility criteria and that actually goes a little bit against the relationship say the physician gatekeepers have with patients in terms of privacy and confidentiality. So that relationship, it makes it important for them to maintain the confidentiality of the information they have and they cannot or should not be giving that information to researchers to help them to indefinite so I think you need to rethink that again and make sure that this is sort of balanced. In the same way, when we talk about hospital administrators being gatekeepers, as you mentioned and the thing is, these people perhaps also have a conflict of interest and that needs to be kept in mind. For example, if you're doing a patient research study and the hospital administrator has an interest in knowing who is the person who is making, who has a high patient safety research adverse rate, and that's not what you are trying to do. You're trying to identify the systems level, so they have a conflict of interest and that needs to be kept in mind and I think your whole analysis needs a little bit of a tweaking here and there to make sure that the ethical principles and other things are not conflicting with the gatekeeper roles that you are trying to provide to them. And finally, you talked about vulnerability and other issues which I think, when you write them down or when you sort of put them as guidelines that are existing that you talk of these issues so just to make sure there's consistently between existing guidelines and these guidelines.

Question/Comment from the audience: Diana Elbourne: I'm a bit jetlagged and I may have missed this in an odd senior moment today but I think I've been awake the whole time. I'm Diana Elbourne, I'm from London, England and that's why I'm jetlagged. When Monica was talking very early on about doing interviews with trialists who were taking and done trials and Chairs and everything, I may have missed you saying, did you do any interviews with people who had taken part in trials as participants, research participants? Because one of the issues that we always find in trials that we do is that it's quite important to do almost an audit of people who are taking part in trials to find out how it was for them. And that is a way of trying to find out whether the process of approaching them and the gatekeeper role, if there has been one, has felt ok for them. Both in terms of that trial and in trying to learn from that trial for future trials. Now you may have discussed this already and if you have, I apologize. Monica Taljaard: I think the short answer is just stand by because that is actually planned but we haven't yet started the process. We are planning to do focus groups with trial participants at both the cluster level, the individual level and also with gatekeepers.

Question/Comment from the audience:Sarah Edwards, also jetlagged from London, UK. I've got a couple of comments really. The first is a purely terminological one. You might find that the term guardian better fits your purposes and indeed it's the term that we used 12 years ago in the paper we published in the British Medical Journal, so that might better suit your purposes. I don't think on reflection it suited our purposes as we had then. I think the term gatekeeper was more what we had in mind. Following on from that, the more sustainable point, is about the role that the individual has in order there are in a position to give consent to trial enrollment and I'm wondering how that squares with what Andrew was talking about earlier when he was saying it wasn't necessary to reveal all the details about all the various trial arms to the individuals so on what basis can they give consent to randomization if they are not told if there is a possible control? I just saw a sharp intake of breathe over there and I was expecting a big reaction but we'll probably wait for that. And the final quick comment is to say that I'm very thrilled the idea that you'll be interviewing members of the clusters and indeed, that's something that one of my PhD's students is doing out in Mumbai with Dave Alren, so, yes, maybe we'll have to collaborate more on that. Charles Weijer: Fantastic, I would like to talk to you more about that. It's just to clarify what we were saying with regard to sort of the tailoring of the consent process in certain cases. We were only there referring to instances of informed consent that was occurring after the randomization of clusters so of course, if individuals are being approached for informed consent prior to randomization of the clusters then there would naturally need to be a full disclosure of what's going on in all of the arms that they might end up allocated to. The only issue after individuals have been randomized that really, the choice that was before them was determined that either the decision is to carry on with the treatment to which their cluster has be allocated or not participate in the study and thus we thought the consent might reasonably be relatively focused on the choice in fact before the individual. Sarah Edwards: So they are giving consent to the intervention rather than to trial enrollment and to having been randomized? Charles Weijer: So, they are giving consent to trial enrollment and to...Sarah Edwards: But I thought that had been determined. Jeremy Grimshaw: So what we are talking about now is the gatekeepers...Sarah Edwards: No, in an individual cluster trial where I take it, you're trying to downplay the authority of the gatekeeper in saying it's the individual whose going to be receiving the intervention should be in a position to consent or refuse. Now it's exactly to what they're consenting or refusing that I'm interested in. Is it just the intervention or is it something more which is what I took from your presentation, why inform the randomization and trial enrollment otherwise? Jeremy Grimshaw: Well, it's like the intervention and data collection around that. So if you like, they've been allocated already to a cluster so that intervention choice is fixed and what they can do is say I can refuse that intervention, I can also refuse data collection aspects of the invention, of the trial. Sarah Edwards: So why tell them they've been randomized? Charles Weijer: Well, just general, general requirements to be honest about how it is that we've come to approach someone, that it's happening at a certain point further down the road in the process than in standard randomized controlled trials. I think in the literature, there's been this assumption that randomization, because people generally provide consent prior to randomization and standard patient randomized controlled trials, there's this assumption that one of the things that people need to consent for is the randomization. I think we are actually challenging that and we are saying that in our paper at least, we are suggesting that as long as individuals, if it's impossible to approach individuals prior to randomization and they are approached as soon as possible and prior to any study interventions for consent that the moral requirements of informed consent can be fully fulfilled. There's no need to appeal to a gatekeeper to give permission for randomization. That's the point that we're making. It might be one that people may disagree with. Angela White: So one way that you might understand it if I understand it right is if the sense in which consent to randomization is by continuing to participant is by agreeing to participate in the trial after they've been informed that they've been randomized to a particular arm. Sarah Edwards: So why stop there and not tell them what the other arms are? Randomization implies that it's something or something else and you're not telling them what the something else is. Angela White: So Andrew and I have had discussions about this and we've disagreed about how much information there's an obligation to tell research subject participants about to respect about what’s going on in the other arms that I try to be more that there should be more disclosure and that sort of thing but I think it's something that, so one way you might address it is there's not a need to tell them at that point about what's going on, detailed information about what's going on in the other arms because as they've said, it's already been decided. So in a way, this is your choice, you either participate in this intervention arm or you don't participate in the trial. I think personally, at some point along the day for the sake of legitimacy of institutions, somebody should, there should be a mechanism in place to let subjects know that this choice has been made and what's gone on.
Question/Comment from the audience: Hi, Stuart Nicholls, University of Ottawa. I just wondered if it would be useful to draw on the discussion of literature and responsibility to sort of pick up on the points that were sort of coming up. So particularly, we might draw responsibilities based on role and that might be the issue of gatekeepers and build on that. But also some of the points that Jonathan made, so we're talking about sort of poking and sort of downstream effects. We talk about cause of responsibilities but also distinguish that through the lens of forcibility and things like moral responsibility of downstream effects and we also have the social science around sort of how you link the outcomes to the intervention. But I think we also have the opportunity to sort of split retrospective and prospective responsibilities in terms of assigning praise or blame and also prospectively so maybe looking at the work of Claude McCloud and looking at strategy allocation but also principled allocation as well and sort of maybe engaging those who don’t have the moral authority but have a strategic role and maybe that language might be useful here.

Martin Eccles: We are in to the final round up. The first thing to do is to once again say thank you to you all both in the room and on the webcast for the degree of engagement and thought that you have contributed to the meeting today. I think it would be safe to say that this has exceeded even our wildest expectations about what would happen today so on behalf of all of the team, thank you very much indeed. What I wanted to do was to reflect back to you what I think I have heard to do a number of things. One is to begin to shape what I take forward into a similar role over the next day and a half with the panel members who you have seen but the other is to give you the opportunity to say that I am fundamentally misrepresentation what you think you have told us. I also, if there are any things, big issues, which are specific to cluster randomized trials, because one of the issues that's emerged is there are big issues out there but a number of the observations that have been offered have been discussed in terms that they are not specific to cluster randomized trials so we are specific to cluster randomized trials so if there's anything you think we've missed, this is also your chance to put it back in play.

Across the areas that we've discussed, the first thing to say is that I haven't detected a sense from either the room or the webcast that there's a feeling that we've got this fundamentally wrong across the board. Starting with the issues about who’s a research subject, there seem to be a general principle agreement. There were some very good and clear areas where there were suggestions we should be reworking some of the rethinking. The issue for instance of the potential for a non consented control group to be potentially harmed by an intervention and therefore being a requirement that brought them into play as research subject and raised the issue about their need for consent and how that should then be handled. That was one example of how we need to continue to development the argument around those issues. I don't think I heard any disagreement with the basic idea that research subjects should provide informed consent. What I think I did hear was a number of suggestions about the fact that once got into the area thinking about the criteria for requesting and granting waivers of consent that became new and we might need to do some further thinking about trying to come up with examples where we thought it might break some of the arguments that we had made.

We also from the area around equipoise and component analysis, the message I took from that was that a method component analysis seemed to be attractive and to offer a number of advantages of a structure to think about some of the issues. We got very clear messages about some of the terms we used within that and indeed, the issue of improving some of the terms that we might use has applied across many of the areas including, not least, the ones we just discussed. So, there were issues about the terminology we might use within that. I think we have to work it exactly how we deal with the arguments of the issues around equipoise and the end degree to which one can potentially take that if one began to insinuate it to apply to pretty much anybody and so I'm confident that Charles is going to be able to fix that tomorrow.

There was also the challenge that we should think about whether or not non clinical studies offered us the same examples as clinical studies and it would be really good to get some examples of non clinical studies that aren't going to work because then we can work them through and see what they do to the principles so that would be incredibly helpful for us to move that area forward and there was the area that Charles identified in the presentation as one where we're just not sure at the moment and this was around interventions with health care professionals in areas of knowledge translation where we just got to keep thrashing our way at the arguments and we don't think we're there yet and unfortunately, none of you gave us an answer to our problem so we're going to keep going with that.

And then the last two presentations, the issues of vulnerable population and what was termed in the presentation gatekeepers, though again, there were suggestions about terminology and what words we should be using. There was considerable discussion around issues of legitimacy, what constituents legitimacy and what the range of roles were for these people. Whatever it is we end up terming them and to whom they are responsible in terms of whatever role it is that they are occupying and hopefully what we will be able to do is come up with some principle frame work by which that can be thought through. Again, I don't think we will necessarily take it to the nth degree of every possible context but it would be really good to have problematic examples that can inform the way we think those sorts of things through.

Vulnerable populations, finally, a clear agreement that vulnerable populations exist. The thing that I was left uncertain about is that vulnerable populations are already fair and square addressed by existing ethics guidance and clearly anything that we would take forward would be done within the context of existing guidance. And one of the things I think we need to work on over the next day and a half is how within cluster trials separate from issues about gate keeping, the issues are different. So one of the things your comments and questions did was begin to at least in part inside my head, crystallize those issues that we still need to work on. You gave us a range of issues to think about around gatekeepers, we need to play those into the conversations tomorrow. That is my sort of overall take of what I think I heard, so first question is, have I just got that wrong? Or do you think that is a reasonable overview of what has been an incredibly wide range and amazingly interesting set of conversations? Have I got it wrong? No, ok. Thank you. Is there anything we've missed? You saw that from the outset, we've structured this around the 6 issues we identified through the processes we undertook right at the start of the project. Remembering that we're interested in issues that are specific to cluster randomized trials, I haven’t heard anything major as a big issue that I think we've missed. There are some issues about nuisances of what we're talking about so one of the examples would be the impact of economic considerations right the way across the piece. But is there anything big you think we've missed? If there's anybody still on the webcast, there are 2 comments coming to the microphone.

Question/Comment from the audience: Abha Saxena from Geneva. I'm not sure that we specifically discussed issues related to confidentiality of clusters and confidentiality about clusters, information that you find out about clusters, all about individuals within clusters which is beyond the confidentiality issues that you talk about in individual trials. I think that is something to be discussed there; I don't think we have discussed it. Martin Eccles: What we need to do is to make sure we keep these issues. It may be that we may not be able to completely resolve them but absolutely we want to know what they are.

Nicholas Birkett: Nick from the University of Ottawa, I'm not sure I'll be presumptuous enough to say you've missed this but as we've been listening to things all day, it seems like we're trying very hard to put participants in cluster trials into a dichotomy of research subject or not research subject. Given that cluster trials are inherently multi level with multiple types of participation, I'm wondering whether it would be some advantage to thinking about that in a more broader sense as a tracheotomy or multi level classification with consequential differential ethical obligations from different, to the participants based on the level they're in.

Mark Gibson from Portland, Oregon with Milbank Memorial Fund. So this will be quick. All of you are, the folks who are involved in this have done a great job in sort of teeing this all up and I think there's obviously been a lot of serious thought that goes into it. The one thing I hope we don't miss and I say this because I work with policy makers all the time at a state level is that I think with cluster randomization, you come closer to political considerations in the general sense that you do with any other kind of research that's randomized on an individual basis for example and it's clinical in nature. And so, while I was encouraged to hear that there was going to be a focus group with gatekeepers, I think you need a deeper conversation with people who understand issues of governess who can inform your thinking about some of these very tricky areas that will be, that you'll end up needing to address one way or another and it would be better to get that ahead of time then to have it come around and slap you at the end when you think you've got a neat package that comes with a bow and it's all done after that.

Sarah Edwards: How to deal with trials created in inequalities which might be larger in scope and in magnitude than you'd find in a conventional trial and maybe more difficult to digress after the trial's ended.

Hi, I'm Vivian Welch, University of Ottawa. I think I'm actually just echoing these last two comments is we work on health equity and I wonder, what if there is no gatekeeper and you've designed a trial which recruits communities through a gatekeeper, you may be further disadvantaging those communities that don't have such a thing.

Kathleen Lohr: Wanted to go back to one of the early sessions. I'm Kathy Lohr from RTI and it has to do with your long term agenda perhaps of returning back to some of the statistical issues and whether they're, at least one person today said this, but whether there are some interactions between problems with the statistical design and what happens with respect to patrician and what sort of ethical burdens does that perhaps put on, or losses, to people who still stay in the study and so forth, I don't think you can deal with that for your guidelines coming up but struck me as something to perhaps return to. The other issue was, I don't know that we resolved as well as we should and this maybe something for tomorrow and the next day, is who gets what kind of information, either after this randomization or at the close of a study because that moves off into feedback and dissemination and I think perhaps some guidance on what it is that researchers owe back to the cluster people and then in some fashion, back to everybody who did "participate" in the sense of consent to data collection and in fact, to serve as your subjects, even if they're data collection was through administrative data. There is some set of pieces of information that I think it's worth considering what exactly to do at the close of a trial that may in part get to what Mark was saying, that there's bigger audiences out there for your trial then just the professionals who did it or the clinicians who were involved and the professionals and so forth. There's a much bigger community if, particularly if you're spending tax dollars that deserve to know something about the findings pro or con.

Annette Brown, 3iE. So, I understand you don't want to take on all of the ethics of all RCT's and you need some boundaries. But I also would encourage you not to assume that everybody who comes to see RCT's comes from doing RCT's. I think a lot of researchers do cluster randomized trials and before they were doing cluster randomized designs, something entirely different. And so, there are not going to have all of those ethically standards from RCT's in the back of our head when they look at this additional set of standards. So I think it can be useful at least for the really important assumptions that you're bringing forward to make sure to bring them forward so that people know where you're starting from in case they are coming from the other side of research.

Sharon Freitag, from the Canadian Association of Research Ethics Boards. I think that's a great comment. What I was going to say was I thought this today, I know you've talked to some of the Chairs but you really, the Chair is just one voice on a board and sometimes does not share by the rest of the board and if there's ten members, there's 11 opinions and I think we have to remember that and I think, it's matter of fact. It's interesting because I actually had a consult last week on a cluster randomized trial at my own institution and possibly it might be something we can talk about because they're actually very interested as to even how to present it to our board and this is sort of been a real life example and something that probably we're going to struggle with so I think it's really important that I think you've got to talk about more than just the Chairs, just an education process with REB's and I think it could be something that we can all learn from, so thank you. Martin Eccles: You meaning that after we've produced whatever it is that we've produced, there's an issue about how that is then taken forward and the context of how it's used be potential audiences? Sharon Freitag: Well I think before you even get to that step, I think even some sort of test cases and this could be a real test case, how even a board looks at it because it's going to be an education to the board as to how they view it and you've got to look at the TCPS and see how you work with that and are you satisfying the waiver of consent and there may be some questions that come up on that issue alone and I could just seem my board dealing with that so I'm thinking from a broader sense and I think Doctor McRae, I think you've probably already thought about some of these things but I think going forward, I think we could actually learn from it and thank you for the invitation to have us here.

Martin Eccles: Absolutely amazing day. I just found it incredibly stimulating. It's incredibly useful in terms of input to where we go next so just a final thank you to you all.