 Hi everyone, welcome to our digital health speaker series for the month Before we start a few housekeeping issues, so please be aware that our lunch on some webcast and Recorded and will be on the website So this this event is part of a digital health at Harvard Which is co-sponsored by the Petrie from Center for healthful policy by technology by ethics and Berkman Klein Center For Internet and Society, which is right across the street I'm myself is a Shreveena Gadjali. I manage a project called global access in action Which promotes access to medicine and within the Berkman Klein Center And it's really interesting too, and I'm happy to have all of you here today So if you're interested to be part of this conversation on On health tech, please do follow us or email us or grab hold of myself or the teams who are here We would be happy to enroll you on our listserv But also to have you part of a working group Which would be look looking at specific issues over the next year In one last thing we are happy if you tweet it on this specific Luncheon series or if you tweet on and I hope Mark doesn't mind as well If you tag him as well So do make sure that you relay the conversation that's happening today or your own perspective of the discussion So it is my pleasure now to welcome Mark Lipsich from the School of Public Health This is a long outstanding Invitation and I'm very happy that he's finally here So Mark is a professor of epidemiology at the Harvard School of Public Health With a joint apartment at the Department of Immunology and infectious disease and the Department of Epidemiology He also directs the Center for communicable disease dynamics And is this associate director of the interdisciplinary concentration in infectious disease epidemiology His research concerns the effect of naturally acquired host immunity vaccine induced immunity and other public health interventions on Population biology of pathogens and the consequences of changing pathogen population for human health Today specifically he's going to talk about his work on computer stimulation on how to enhance vaccine trials, which I know is of Interest to many of us in the room. So without further delay Mark Welcome Thank You Ash and thanks to all of you for coming in and to Berkman Klein and Petrie Flam for the invitation It's great to be here. It's a very different crowd from people I usually talk to and I will say I am not quite sure what will be of greatest interest to this crowd So a few points of framing One is to say this talk has had various titles. I was told it was too long and so I shortened it This is the shorter title There is the previous one had some more kind of ethics and policy Aspect to it, which I'll talk about a little bit, but I will focus on the on the computational part The other thing to say is that I've constructed the talk To account for this uncertainty to some extent By making it somewhat modular. I have sort of three main sections after the introduction I'm happy to skip the third one. I think it's more technically interesting, but less maybe less conceptually interesting for some of you And so depending on where the discussion goes, please ask questions along the way, make comments along the way and I will adjust as the time permits and the interest permits And the last thing to say is that I have I've put this together at a fairly high level of abstraction There are not a lot of details of the methods. I can provide those if people are interested But but I thought I would go broader rather than deeper mostly because I don't know what will particularly strike a chord But I'm happy to get deeper if someone is interested So with all of that is prelude The the goal that I want to talk about means towards is the goal of providing safe effective vaccines To those who need them and doing so fast I'm going to focus most of the talk on the first two examples on mostly the setting of emergencies where where there's a real source of Urgency and I'll talk about that in a minute So fast is especially important in outbreak diseases And this is kind of the pipeline very simplified of how vaccines Get to people starting from basic research on the mechanisms of the pathogen and its pathogenesis working through development of candidate vaccines testing in animals Then that's all the preclinical phases then the first clinical phases is Tests of the safety of the vaccines of the candidate vaccines and their immunogenicity or ability to elicit A response that might be protective against future infections and then efficacy or effectiveness trials to see whether they actually protect people Against infection and then when those are successful licensure and deployment All of the all of the steps up to Safety and immunogenicity trials can be done in principle before there is any Disease present you can do You can you can test in animals by giving them the disease you can you can test safety and immunogenicity in people without without having the disease present and then The last two phases really require that the disease be present Which is a particular challenge for diseases that are only sporadically present or sporadically detected and so in 2014 with the Outbreak of Ebola in West Africa. We were actually stuck at an earlier phase We we had because of biodefense interest on the shelf several candidate vaccines at different phases of of Preclinical testing up through animal testing but very little human data on these these have not been tested in humans To a large extent even for those things which you could do in principle without the disease being present and For many potential threats were it at even earlier phases for the diseases that we fear might cause outbreaks or larger Epidemics or pandemics were at even earlier phases and even animal testing hasn't happened and Many of you are very familiar with this Development that a couple of years ago An organization called sepi coalition for epidemic preparedness innovations was set up with about a billion dollars aspirationally and much of that actually actually pledged From foundations and governments and it was established to do two things one was to move candidate vaccines Against anticipated threats so known potential threats known known pathogens that might be outbreak threats Through testing for through the last stage that you can do before the disease is actually present and then secondly to establish platform technologies to rapidly produce vaccines against unanticipated threats Which is to say viruses or other pathogens that we have never seen or that we imagine like Zika a few until a few years ago to be mild and unimportant causes of Disease that we would never care to to make a vaccine against So the idea is to to move from here Through the efforts of sepi and others but really led by sepi Up to here for as many potential threats as we can think of plus platforms for those that we can't think of and I mentioned that that for Outbreak prone diseases like Ebola in in those three countries there's a sense of urgency and The sense of urgency really kind of has two parts Which all feel very urgent, but they feel urgent for different reasons so This is an epidemic curve of the virus of the cases in two countries two of the three affected countries and It's labeled on the left With the sources of urgency during the the increasing phase of the epidemic so there's near exponential growth of the epidemic there's Feeling that we need countermeasures to mitigate this accelerating epidemic And incidents is typically very patchy in space Which makes it very hard to figure out where if we're going to test some some countermeasure Where are we going to do it and infrastructure is not typically very thick on the ground in places where this might be needed So you have to set up the infrastructure in anticipation of using it, but where to do that is not obvious This Sure the question is how typical is this timescale of about a year from from initiation to peak There's no reason why it should be typical But it's not a bad guess at the timescale so influenza is not new so we anticipated and well It'll be back again next year So in some sense, it's it's not the same the same type of urgency, but influenza Does typically rise over a couple of months and fall over a couple of months pandemics are a little bit different and can be multi-peaked Polio is kind of dribbling along and being pushed towards eradication With periodic upsurges But there's no need for a new vaccine to be tested so for for these kinds of outbreak prone diseases It really depends on the natural history of the disease and on how seasonal the transmission is so there's no real rule about it, but but Epidemics of any size typically take at least several months to get going Just because most things don't transmit ultra fast How much does this depend on communication patterns? For example plague during the Middle Ages would Spread based on communication Patterns and things like that over a period of years. Is this within one geographical area or is this worldwide? So the Again, it's very disease specific most of the diseases that that we can worry about outbreaks of Transmit through close contact so locally People are around people and and you can get a rapid exponential growth at some rate That's not so determined by by global transmission or global movement patterns at the next level up though as you begin to look Outside of a local area then the degree of connectedness matters a lot so so the early part of this could have happened in any century, but the and and the And the connections were not so great in these countries, but But the big worry during this early phase for example was that it would get into Lagos Nigeria because it had crossed the board it did cross the border into Nigeria and was fortunately contained very quickly So had it gotten into a big city Then the potential would have been to infect a large portion of that city, so it I Guess the key point is that most connections between people are local But that that can saturate if if there aren't also longer distance connections And and it depends on how fast it can get out of that area, but these were It most of the parts of most of all three of these countries were affected bordering countries were affected but In particular Nigeria contained it very quickly So there was this feeling of urgency and and at the In 2014 public health experts were saying we may not be able to contain this without a vaccine And so that was and we're projecting very very large numbers And so that was a real incentive to try to get a vaccine going Fortunately they were wrong in about that pessimistic prediction And it did begin to turn over around the end of 2014 Of course, you never know when it's turning over and when it's just a temporary downturn until until it's really turned over So at the time you don't know what to say, but you but at least it was beginning to turn over and then by 2015 it was it was declining fairly dramatically due to basic public health measures changes and burial practices Changes in movement restriction of movement and all sorts of other things to your to your question So then the challenge is if we don't test this vaccine now We may have no cases not enough cases around to test it until the next outbreak and so we'll be in the same position We were in Back at the beginning of 2014, but in the next outbreak, which is whose timing is unpredictable So there the end game was a big rush to try to figure out can we Can we Test a vaccine maybe not for the purposes of ending at this outbreak But for the purposes of having something next time and of course that vaccine was tested that was tested Has been used in subsequent subsequently in the Democratic Republic of Congo where there is an ongoing outbreak of Ebola virus disease complicated by conflict and Unsafe conditions for the medical workers as well as the as well as the residents so spurred on by this experience and a little bit of Participation in that Ebola trial as a member of the advisory group to the Ebola vaccine trial We've begun to develop a research agenda that basically Consists of trying to figure out ways of doing new approaches to to these clinical trials of vaccines And understanding the properties of the existing ones as they would be implemented in emergencies Secondly and this was the part that fell out of the title when I shortened it and I'll just mention a little bit is I think there's a big so far has been a big disconnect between the ethical debate and the And the methodological debate which has often pitted methodologists against ethicists as Method as statistician saying you can only get answers from a very pure trial and ethicists saying that's using people as guinea pigs and a lot of not very productive discussion and There are various reasons for that But I think bringing some of these quantitative understandings into the ethical discussion Can only sharpen it and that's something we've begun to do and work that I'll just briefly mention And we've talked about but haven't yet begun WHO has actually begun this in a way To try to compile what we know at the moment into a sort of playbook That allows people to think about trial designs in advance and in a systematic way And ideally that would be dynamic and updatable so as the technology for for By which I mean that the techniques for developing these trials improve. They can be incorporated and Last year with near al who's a bioethicist We had a paper in science that that tried to look in look at some of these issues But but what I want to talk about today is kind of a subset of that which is Which is an idea that a number of people have had and we had a little meeting about it a couple of years ago in Seattle Which is the idea that Simulations computer simulations are part of the of the toolkit for improving our vaccine trials in other words Simulating the trial can be part of the process of planning it and both the design and the analysis of of that trial and In in different audiences I have a long discourse on why that should be necessary because That hasn't been done very much in the past But the summary of that is that with standard clinical trials of non communicable outcomes like heart attacks or or strokes or whatever The theory is pretty well understood for how you design Trial and it's you can write down elegant equations and there's very little need to simulate it similarly for very simple vaccine trials That can often be the case. You can just use basic theory that you learn in in clinical trials Classes and there's not much need to simulate it the real reason to try to to bring computational approaches in is really when the existence of the trial itself is interfering with the disease transmission process Hopefully by reducing it Which is good for the people in the trial and bad for the power of the trial because more cases mean you can study something better So if the existence of the trial itself is changing the process of transmission Then the assumptions that go into standard methods are no longer really valid and sometimes it's useful to to Write down and execute simulations of the trial and I say write down because in my experience doing this so far Actually the the process of just trying to say what are all the steps that will happen in a trial when you have Transmission and where the individual cases are the sources of other cases Just that process of writing down the code or even a sort of box diagram for the code Is as informative and generates probably the more important insights than running the code itself So it's it's really a matter of Formalizing your thought process as much as it is of computing complicated numbers So an example of that is the ring vaccination trial that was used for the Ebola vaccine in Guinea this was the Merck manufactured our VSV vaccine and It's the one on which I served on the technical advisory board and that in thereby Got interested in trying to think about what could go wrong. What are the what are the potential pitfalls of this trial? How can I provide helpful advice some of which was more appreciated than others but But the what was extraordinary about this trial was that the essentially the the technique was invented on the fly Ring vaccination had been used as a way of controlling smallpox and ring vaccination simply means you find cases and Around those cases you vaccinate people and exactly what you mean by around can vary by the situation But you you target your vaccination around cases in order to sort of prevent produce fire breaks around those cases So for known useful vaccines like the smallpox vaccine. This was a well-established approach But it had never been used as a trial method in other words It was it had never been used to try to figure out does this vaccine do anything at all Beneficial or or is it useless? Which is what a phase 3? Efficacy trial is supposed to do So what was extraordinary was that a team of people invented this as a as a means of evaluating vaccines on the fly and Created the trial and ran the trial all during that chaotic period that I showed the curve of a minute ago and the basic design was to find cases to To Then create a ring around each of those cases to find as the contacts of those cases and the contacts of their contacts Then once that ring was identified to randomize the ring to be either To receive either vaccination as soon as possible or vaccination three weeks later and to compare the rings that had been Vaccinated immediately to those that had been vaccinated with a delay with the expectation that an effective vaccine would produce lower Incidents in the rings that were immediately vaccinated and higher incidents in the rings that were delayed and So these were two of the two of the papers the first two papers that came out on on that trial And our goal was a little bit to think about that trial But by the time we were we were doing simulations. We were really trying to address a more general question, which is People were saying once this trial proved to be effective in testing the vaccine Or reasonably effective in testing it. It wasn't perfect That this might be the new standard for how we try test vaccines and emergencies And so we wanted to try to ask the question is what what are the conditions under which is this is a good design? And this is the paper that was a part of the doctoral thesis of Matt Hitchings and done in in collaboration with Rebecca Grace of Episantra, which is the Medecincent Frontier Research Center in Paris So So to do this we abstract from the reality and we create a model of the disease dynamics And so I'll tell you a little bit about how we did that. We we assumed that there was a small homogeneous ring of people with of about 50 people typically around a case And we used what's known as an SEIR or susceptible exposed infectious and recovered Compartmental model which is just tracks the stage of illness or infection for for the people in the ring We've allowed people to be vaccinated at a certain time and we also allowed people to be to be hospitalized at a certain time which meant that they were detected with their symptoms and their Symptoms and and thereby isolated and put in the hospital and And that was the basic pieces of the model. This is a graphical picture of it And I guess all pictures are graphical. This is a picture of it and the The when I say that writing down this model was actually the one of the biggest sources of insight what I mean by that is When you write down a model and you have compartments that somebody has to always be in one exactly one of so they're either Susceptible or they're susceptible and vaccinated or they're exposed meaning they've become infected but are not yet sick or they're infectious and infected Infectious and symptomatic or they're hospitalized or they're recovered or dead Which is not shown here the You have to figure out where do people go and how do they how do they contribute to the To the infectious process and in the process of doing that it occurred to me that once somebody's hospitalized They should be isolated and no longer contributing to transmission And in a well-run trial that should certainly be the case But it it then occurred to me and I'll show this as one of the results that that the faster you Detect your cases the better your trial is set up to detect the outcome. It's supposed to detect The fewer cases you're going to get because these people are no longer contributing to transmission So a very well-functioning trial in one sense is going to it risks being an underpowered trial in another sense because Every time you ascertain a case you make them no longer a source of infection And that's an example of the kind of thing that doesn't go into standard power calculations for a trial But could be potentially very important So we simulated this trial and I won't go through all these Bullet points. That's basically says what I just mentioned And we estimated The incidence rates in in each of these rings and then use that to estimate How many rings would we need in order to be able to tell that the vaccine was effective assuming that it was effective with a certain level of effect and So when you run this computer simulation Which is quite straightforward to do what you Find is that the incidence rate in rings that are immediately vaccinated Goes up right after they're vaccinated because the vaccine takes time to to Take effect and also because people have already been infected At the time you vaccinate them some of them and so you but they aren't yet sick And so those people by assumption the vaccine doesn't help with so you start you start with an increase in the in the number of cases per capita, but that starts to decline and it declines especially in the group that got immediately vaccinated because the contacts of the people who were transmitting during that vaccination campaign are now protected after a certain delay and then in the in the Delayed arm this whole process is postponed by about About the by about the amount of time that they're delayed So So the idea is then you create the this trial you run it With the size that your your calculations tell you you need in order to get a high probability of seeing a statistically significant result if there is a real Protective efficacy and by assumption in this in this simulation the effective The vaccine was to be 75% protective 70% protective In our base case which is shown in red and what this shows is that What this left graph shows is that the efficacy that you measure Varies it goes up if you have a more effective vaccine and it goes down if you have a less effective vaccine But it doesn't do that linearly and so if you have a hundred percent effective vaccine you you find about 80% benefit if you have a 50% effective vaccine you find about 60% percent almost 55% effective and the The reasons for that are things that come out of of then analyzing the results You get this attenuation because some people who were protected Get their disease Sorry some people The period of observation that you use to decide whether someone is a case or not has to be defined based on the the period of time before someone gets symptoms and and the time the vaccine takes to to Confer protection, but that varies from person to person so you misclassify some people And that attenuates your efficacy on the on the other side You're not just measuring the effect of the vaccine on the person who gets it But also the effect of the vaccine in all the that person's contacts in the ring on them And so you get some of the indirect benefits measured in this in this way When when you set the vaccine efficacy to 70% as it happens what comes out is exactly 70% on average with some variation between between runs but but Looking at other parameter values shows you that that's kind of a coincidence that where these two effects just happened to balance and Then you can also use this for trying to figure out how many sample how many rings you need to sample as You would expect the more effective the vaccine is the easier it is to see that effect and the fewer rings you need to do it and then you can ask about some of the other parameters and I mentioned this probability of detection So a well-run trial will have a very high probability that if someone is a case they'll be detected and detected fast and a well-functioning public health system will Isolate that person as quickly as possible so when you have a When you have a daily probability of detection and these numbers are kind of Just guessed at as being reasonable because because those were not reported in the trial But if you if it takes on average five days, so that's 20% chance per day of being The cases are detected then you get this estimate of efficacy. That's about 70% if you if you detect less well You're actually able to see more Efficacy because you see more in cases and those cases are better prevented by the vaccine Due to its indirect effects and if you do very good case detection it actually makes the vaccine look a little bit worse So it's a case where one aspect of good trial running leads to attenuation of the of the effect of the observed effect and The daily probability of detection also has strange effects on the sample size that's required Again, because if you shut down transmission by detecting case as well Then you need to have a lot more rings to get the same number of cases to see an effect so these are the kinds of trade-offs and we can play these games with with other parameters and maybe since this is Maybe I'll just Skip through some of that because Talking more than I anticipated on each slide so I'll summarize the the findings from this that The sample size and the estimate are sensitive to various parameters of how you run the trial of how the rings are set up and of of the proportion of infections that come from outside the ring so the Premise of designing a trial like this is that most infections come from within the ring and that's why Setting up your trial around a case is actually an effective thing to do but But if you deviate from that it actually becomes easier to do the trial in some sense because you're not interfering with the transmission by doing the vaccination itself so there's actually more More transmission to stop So So the purpose of doing all of this is Is to really ask the question alright is this the best trial design for all settings probably not It it's clearly a good trial to design in the setting where Where transmission is local and very focused And also declining so that you really need to focus your your efforts right around cases because otherwise You'll you'll need to vaccinate way too many people but So those those are the settings in which a design like this could be Could be well suited It doesn't measure the quantity that it purports to measure Not withstanding the way it was described in the publications it measures a composite of the efficacy on the person who was being vaccinated and the and the indirect protection of of their contacts And Efficient case detection turns out to be a double-edged sword. It's obviously good for the person for the individuals It's also good for knowing that you have Cases which is important for the trial, but it also shut on the other hand it shuts down some of the transmission And so the idea of having a Simulation machine like this, which is just a short bit of our code actually is not not very complicated Is that if you have a new disease and you want to run a trial like this? in a different population with different sizes of rings or different sizes of Or different types of natural history you can very lightly modify it and And ask these same questions and try to see what would the what would the predictions be? So it's a design tool second example I want to talk about is is work That that Rebecca Khan now a PhD student did as part of her master's thesis and just was published And the question there is what if you have asymptomatic infections in a vaccine trial? And you might think well who cares if you have an asymptomatic infection, maybe we don't care about preventing it Maybe it's not that interesting But in fact it turns out that almost all of the big big Infections on the sepi list including Zika. Well Zika is not on their list, but but Nipah Lassa Arguably Ebola And and a number of others have a proportion of infections that are Asymptomatic or subclinical so you would be very hard-pressed to detect them without very intensive Surveillance and even with very intensive surveillance Probably you would miss a number of infections without unless you tested people for antibodies after the fact But these infections can be important for at least two reasons one is they can serve as sources of transmission to others and Second is that they can have Sequelae such as gambare syndrome for Zika so you can have a Zika infection that does nothing that you can detect at the Moment, but that leaves you with a neurological syndrome that afterwards some time afterwards And so preventing that ace previously asymptomatic case would have been important so we would like to measure efficacy against that and Standard approaches Don't work, and I think the easiest way To see this is with a diagram So in a vaccine trial you're studying people some of them are at risk Say you have whatever that number of circles that is about ten people in the vaccine arm and ten people in the control arm and Proportion of them get infected, but a smaller proportion in the vaccine arm because you have an effective vaccine So if that's true, then the people at risk will be the ones in pink They're fewer people at risk in the control arm because they've already been infected But what you observe if there are if some of those are asymptomatic is that a few of those people have symptomatic infections and What you think if you don't know that there are asymptomatic infections is that there are a lot more people at risk And especially if you compare Here the the people you think are at risk in the control arm To the ones who are actually at risk you've gotten it very wrong Whereas in the vaccine arm you've gotten it less wrong because most people haven't gotten infected your vaccine protected most of them see you've You've overestimated the number of people at risk, especially in the in the arm that got control and when you do that You get an estimate of how well the vaccine did that is biased towards being ineffective Biases your estimate towards the estimate that the vaccine didn't do anything because you think there are all these people at risk And most of them aren't getting infected yet anymore in the control arm because they were already infected. You just didn't see And this is a pretty well-known problem But what hadn't been done was to try to figure out how to solve it So one way you could solve it would be you just test people Every week To detect asymptomatic infections and you could do that by testing for the virus if you did it very Intensively or by testing for antibodies to the virus which persists so you can be a little bit more You don't have to do it as frequently But these are challenging because they're expensive the serology the antibody tests are not very good, especially for something like Zika It's often hard to tell whether they had the vaccine or they or the infection. There are a bunch of challenges so What we wondered was can we test can we figure out a way to test people efficiently? For example only test people in some of the of the Study and use what you learn about them to inform estimates for the other people And you do that by a technique called imputation which I won't dwell on So again, what we did to test these ideas was to simulate a trial We simulated with various levels of transmission various trial lengths These are the base case numbers, but we tried various variations on these and And and what we did to just Focus on was initially to look at a vaccine that was 60% effective against all infections and People who were infected were equally likely to be symptomatic with they were vaccinated or not But they were just less likely to become infected And so these are the results the results are that if you could have perfect knowledge So test people essentially continuously have a have a port in their arm And as soon as the virus got to them you would know it Then you get the right answer, which is which we've labeled impractical You can with low effort the Labels on here refer to the level of effort with low effort if you just Note the symptomatic people And do various analyses that that you could do with just the symptomatic infections You get an underestimate you think the vaccine is about forty two forty five percent effective, but it's actually sixty percent effective and then there are various Ways if you test people at the end of the trial Either at the end of the trial or or three times and Beginning middle and end You can with Considerable effort Meaning you're testing every individual in the trial you can get essentially the right answer But what was the main contribution of this this work was to show that if you test Only ten percent of the people in the trial and then you use those people to the results from those people to figure out What is the ratio of symptomatic to asymptomatic cases and extrapolate that through imputation to the other people in the trial you can get You can get the right answer with essentially far fewer resources than it would take to test everyone in the trial So that's a shorter example But but here the idea is really it doesn't so much matter What kind of trial you're doing the point is that there are ways of economizing on on the tests which are expensive to do accurately And you can you can nonetheless get the right answer To figure out if your vaccine is working So in the last couple of minutes before opening up for discussion, I just want to mention couple of sort of ongoing areas and Some questions that we're thinking about and maybe these will spark some discussion So I mentioned I think there's a connection between these questions of quantitative issue of quantitating vaccine help vaccine trials work and and research ethics and We've begun to explore those mostly we being mostly near Al and I and Rebecca Khan the student who's who's become interested in the ethical issues as well and a net read another bioethicist And we've been trying to think about questions like are randomized trials needed in emergencies Are they ethical even if you want to use placebo and can we use them to maximize public health benefit? in case the Vaccine proves effective and this is their various thoughts ideas about this another interesting Project that I was not involved in but now collaborate with some of these people was a Study by Guy Harling and colleagues that tried to figure out Designs that were where that was an explicit goal but we we actually took a more simple view and just pointed out that When you do an individually randomized trial which some people have ethical concerns about One of the benefits of doing that is that it happens faster You get the answer faster than you would in a Cluster randomized trial such as a step wedge trial which is one where you roll it out roll out the vaccine to different communities in some randomized order and Compare the incidence in vaccinated communities to ones that have not yet gotten the vaccine Some people prefer this this upper one because it's it gives everybody the vaccine so I Don't know how to classify this as ethics or just sort of common sense or quantitative or what but but what? We pointed out was that if you're really anxious to give everyone the vaccine as soon as possible because you imagine that it's likely to to be beneficial Then doing an individually randomized trial and then immediately giving the vaccine to To those in the control arm Still allows you to vaccinate everyone and in particular the least advantaged if you have equity Considerations least advantages Who get it last get it earlier in this setting than they would in the setting of this? What some people consider to be an ethically preferable? Study design so we we think that by being a little bit creative in terms of the Way you set up the trial you can address many of the ethical concerns Another sort of broad policy question, which I would love to think harder about and just have not really had time to focus on But maybe we'll be in the interests of some of the people here Excuse me is are there circumstances where we would forgo randomized trials to test whether a vaccine works? Because rolling out a vaccine is so urgent And I think we start with a strong presumption in Favor of requiring that we test vaccines as we do for all ordinary vaccines before we roll them out There are economic reasons to do that Nobody wants to make a vaccine that they in large quantity that they can't be sure is actually any good Their regulatory reasons the FDA exists to make sure this kind of thing happens Among other reasons their ethical reasons giving people an unproven vaccine is is arguably not a good thing And there are historic reasons in that there are examples of really promising looking vaccines that actually did harm When they were used in efficacy trials those were not licensed. Those are not the vaccines that we use now Those are vaccines that were appropriately stopped by the regulatory or the clinical trials process But you could imagine a scenario of a very very fast moving very deadly disease where It was so dire that some kind of Observational rollout rather than a trial could be desirable And I think it would be valuable to have those conversations and discussions before Such a thing happens. So if people are interested in that I'd be happy to discuss further And then the last area that I'll mention that we are working on that's that's now Rebecca cons main Research area is to try to incorporate pathogen sequence data into vaccine trials Where you can estimate who affected whom who infected whom and thereby enhance the inference of what the vaccine is done So most of the time when you do a trial you know if someone was vaccinated and whether they got infected But you don't know if they infected anybody else or who infected the people that did get infected With pathogen DNA or RNA sequences you can estimate make estimates of who infected whom and therefore you can tell how infectious someone was as well as whether they were as well as how how susceptible they were and tell what the vaccine did to that so What we're doing now with a with a grant from maybe a grant It started in April and we haven't gotten the money yet So hopefully a grant from the from the UK government is to try to figure out how you can enhance trials in that way So just to close I think simulations can aid in the design and analysis of trials for infectious diseases We think it's really valuable to do this work and studying these properties of these designs before The emergency comes As I briefly talked about I think the ethics are really entangled with the methodology and we should try to discuss both of these during peacetime between epidemics I Think ring vaccination is a promising design for the end of an epidemic, but with some caveats that I discussed and We think we found a way to study efficacy against asymptomatic infections in a way that is economically feasible Or more feasible and so I will skip the last example as I Promised he didn't get a choice. I just skipped it And just end with a list of collaborators and thank you for your attention and take questions burning question or can I Have used my moderator prerogative to to start the discussion Early on you were talking about Hospitals as isolation mechanisms How realistic is that and how necessary is that to your analysis? If hospitals weren't Yeah, so if they if they weren't You could imagine two thing two possibilities one is they don't change your risk of infecting others Which would be kind of odd because it's a big behavioral change and the others that they increase it And so for example in the the outbreaks of MERS in both the Middle East and And especially in South Korea have been largely focused in hospitals So the hospital is itself the venue for transmission So and so that has actually encouraged us to think about are there Trial designs for something like a MERS vaccine that would use that fact that it seems like hospitals are Are the settings where the transmission is most likely so it's not that we're positing this as a universal fact It was a reasonably Reasonable approximation of a fact at that stage of the Ebola outbreak Not at all stages because in places that used that had funeral rituals that involved washing the body and touching the body You know the hospital was not the end of the story. There was still an opportunity for transmission afterwards So it's not we don't mean that as a as a sort of Assumed fact about everything. It's just that in this case it probably was and If the trial was working well, it was Talking about the Ebola right now. We are starting to test the multi drugs in DRC And I was wondering whether how does the ring vaccination apply in that particular context Given what we know now in DRC And the fact that you mentioned that thinks as well I think it's a good case study. Is this applicable? Yeah, I Probably not in the sense that The drugs at least for the moment the purpose of the drugs is not to affect transmission, but to affect Clinical course so you can't really give them until you have a case and then you give them to the case not to the contacts You could imagine testing drugs prophylactically, but but probably way down the line Probably not anytime soon I Do think that the ethical questions that come up are Very closely related Especially the question of whether to use placebo or some some similar something similar to placebo and There's a kind of simmering debate about that in the literature And whether back so we've really limited our focus to vaccines because I think the The issues are more clear-cut in vaccines and it's better to solve easy problems first and then hard ones but But I think some of what's been written recently for example by Alex London on the ethics of unproven interventions, I think applies equally well to Vaccine I mean to drugs to treatments And that there's a good case to be made that that some Some standard of care plus placebo or standard of care That does not include the experimental Drugs is what you should use as a comparator But that would be Controversial with many people although I think that argument is starting to hold some sway A few a few weeks ago Harvard had this special week on contagion where they were talking about outbreaks of new diseases how realistic is How applicable is this kind of study to planning for something like that? I think many of the things they were talking about were things like Ebola and Zika It was more focused on flu, but but I think The only way that any disease gets to be global is that we fail to contain it locally no diseases sort of pop up in many places at once Spontaneously because they're transmitted by infectious agents and I've tried to make the case In a number of ways that Dealing with something like Ebola and Ebola outbreak or Zika outbreak is at the small level is the way to prevent large-scale catastrophes it is true that The chance that Ebola becomes global without Barring some biological change in the virus is not very great because because we can handle it in resource rich places and Or or barring a massive attack in which it was distributed to lots of people simultaneously But a trickle of cases is something we can handle if Chris Christie and his friends would get out of the way But but putting aside Politics we can handle a trickle of cases, which is how it would begin in in rich countries the respiratory transmitted viruses like influenza and Some of the coronaviruses are of greater concern because they are It's much harder to prevent transmission and so but those two start small and get bigger so I think I Think finding ways to contain smallish outbreaks is by by vaccines or other other things Other measures is the way to is a way to prevent large-scale outbreaks Influenza is a special case in that it will travel very very fast and will be in many places at once But but with that exception there really aren't a lot of cases where Where you get beyond this phase very quickly Does that answer your question? One of the worries that some people have been speculating about lately is some kind of bioterrorism Manufactured synthetic new disease where some of the characteristics of the Warburg type diseases are combined with some of the transmission capabilities of the respiratory diseases and Have the ability to react quickly to that It's also communication patterns a lot of The isolation mechanisms in traditional societies for example in Africa and the villages would just block the pathways to isolate themselves In epidemics in the past whereas if something materialized let's say during some kind of Emergency in a large city it would be much more difficult to contain quickly. Yeah Well, I think that's where this this issue that I raised toward the end on Under what circumstances would we sort of dispense with the formalities of testing? Something properly and roll out a countermeasure. Of course, we have to have the countermeasure to roll out. I Think those sort of nightmare scenarios whose probability. I don't know how to estimate any better than than anyone else Except to say that we don't know of such in the public domain. There is no evidence that such viruses exist. I Think those nightmare scenarios are the ones where you have to think about what you would do In terms of trying to roll out and test a countermeasure simultaneously, you still look like I'm not answering your question These are probably questions that he looks at regularly and also How does it work if we have to test something and how do you protect patients as well? So the setting up of a compensation fund is something that's being actively discussed at the World Health Organization And how do we handle it if there is a state of emergency? And we do have to test or to use a vaccine without having gone through the full clinical trial So if you're interested, we can chat later on one last question Well, thank you very much Mark. That was a very Interesting we hope that the ring vaccination trial gets more conclusive as progress and you get your NIH grant soon Thank you everyone