You’re listening to a press conference from the Harvard T.H. Chan School of Public Health with Michael Mina, assistant professor of epidemiology. This call was recorded at 11:30 am Eastern Time on Friday, April 24.
Previous press conferences are linked at the bottom of this transcript.
Transcript
MICHAEL MINA: For those of you who I haven’t met yet, I’m an assistant professor of epidemiology and also of immunology and infectious diseases at the School of Public Health at Harvard. I’m also one of the medical directors at Brigham and Women’s Hospital where I help to oversee molecular biology diagnostics and, more recently, some of the PCR-based molecular biology diagnostics at the Broad Institute.
And I have been very active in both monitoring and modeling the epidemic and developing a lot of testing strategies for the epidemic as well and generally I’m very familiar with, in particular, the conversation surrounding serology and how these new tests are used. And I imagine that that’s what a lot of people may have questions about at this point, but feel free to ask anything. And if I don’t have an answer for you, I won’t answer, but I will let you know that I don’t know. So, I’ll take any questions.
MODERATOR: Alright, looks like our first question.
Q: Hi. Thank you, Dr. Mina. I wanted to ask about community serology surveys and what they can tell us about the coronavirus’ spread and what the limitations of the surveys are. A local university is releasing findings down here in Miami today; they have a two-week serological survey of about 1400 people in Miami Dade, which is a county of about 2.7 million people. I know that similar studies have been done in California and elsewhere. So, I was curious, you know, what you could tell us about, again, what the surveys will tell us about the spread and what the limitations are?
MICHAEL MINA: Sure. So, it’s funny you asked. I just got a whole shipment of – I heard my doorbell ring a moment ago, and I got a shipment of these serological assays to trial that just got shipped, but I won’t go answer the door right now. I’ll let them sit there.
So, the assays and these tests that are being performed will – if done properly, they can really show us a tremendous amount, not just about the prevalence of the pathogen, which is something we still don’t really know at this point in time. And what I mean by that is we don’t know how many people have been infected. We can take point prevalences – we can capture a cross section right now and say, this is how many people are infected with the viral. But serology really allows us to say how many people have been infected essentially up to about 10 days ago. And that’s because it’s about a 10-day span before some of these antibodies actually form and develop to a point where they are detectable.
And the reason that that’s so important is not just to understand what the epidemic has done so far, and how many people it has impacted and infected, but it tells us a tremendous amount about the actual biology of the pathogen. And that includes everything from the severity of the pathogen to the actual infection mortality rate associated with it. It will tell us about how many people who became infected were asymptomatic or sufficiently mildly symptomatic that they did not require medical attention. And so, it actually gives us an extraordinary window into both the epidemiology and the biology of this virus and it will be
one of the main features of one of the main processes that we deploy to plan a path forward for getting out of sort of the shutdown and restriction mode that we’re in at the moment.
And so that’s really why there’s been so much interest in in getting serological evaluations performed. It’s the huge missing piece from this epidemic and certainly I come from it from despite helping to lead a lot of efforts to get testing for virus more widely available, I would say that that in that position, I’ve been able to get a very clear firsthand look of just the extreme magnitude of how deficient our testing has been, which I think actually is more broad than many recognize.
I was recently speaking, for example, with a network of hospitals that cover about a million people, who pretty much have no access to testing still. And so, when I start to think about just how poor testing access has been, I think serological studies that are well designed and well performed with well performing assays are going to just really fill in a huge knowledge gap that currently exists.
Q: In terms of limitations?
MICHAEL MINA: Well the limitations, as I think people have probably heard about, fall into the category of non-split – well, there’s really two. There’s issues with the test and then issues with the design of the surveys. So, the tests, especially all of these point of care instruments that have been coming from China and Korea and some other places, a lot of them have flooded the markets and there’s been really savvy brochures to get CEOs of companies to want to buy into these products and they’re even giving them away. You know, I’ve had people write and say we want to give you 500,000 tests and, in my opinion, anytime somebody offers you something like that, you probably should say no if it’s for free. And because it’s probably a problem.
And so, we’ve seen that. And the problem is primarily if you have a test that even is just – so 99% or 98.5% sounds really good if I tell you that that’s the specificity. But if I then tell you that the prevalence of an infection in the population might only be one or 2%, then if you go and do serological survey and you end up getting 2% of the population to be positive, but it turns out that one and a half percent of those are false positives, because they all say is not as accurate as you would hope, then that really changes your evaluation from what you can determine from that study. Is it that .5% of the population was actually infected or 2%? And that’s actually a pretty big difference when it comes to trying to understand the epidemiology of this epidemic. And so that’s one of the major limitations.
The other one is who you’re sampling and what are the bias is inherent in this in the population. So, there’s always going to be bias and there’s no perfectly representative population but there are ways to try to improve the representation of a population you’re studying. So, if we only study hospital employees and patients for x and for an extreme example. Well, we know that will probably get a huge number of very, very high prevalence. Those individuals are no longer representative of the population at large, because this virus has transmitted through healthcare systems. For example, if we only go into nursing homes and try to use nursing home residents as our representative population of the whole population, that would obviously be wrong. And so that’s what we’re finding and that happens. That’s also at a city level. If you only go into a hotspot, if you only go into New York City and extrapolate from there really, you can only extrapolate about New York City and you don’t want to necessarily make very large judgments about a whole population. And sometimes it’s interesting to do that. And I’ve certainly done that, you know, as thought-provoking issues, because I think that the testing has I think that many more people in this country have been affected them testing so far would suggest.
But when it really comes down to doing the science, it can become an extraordinarily large limitation to a study if you don’t properly design the study to capture a population that is representative, or have the appropriate statistical considerations to account for areas where bias might exist. Those are the biggest limitations.
Q: Thank you.
MODERATOR: Next question.
Q: Hi, Doctor. So, my question is among the different serological surveys around the world, which has published the referee doubt in the last few weeks. So, which one or ones do you think is more viable are authentic? Because according to the current data or do you think herd immunity is a stage we can get to within a short time, or we still have a long way to go as different predictions your lead to totally different pass on reopening the calm?
MICHAEL MINA: So, I would say none have been great so far. And in large part, that’s because a lot of the surveys that have been done have used these fairly unreliable point of care instruments, because they’re easily accessible at this point because everyone has them. I would say that the Netherlands did a study, and it wasn’t really officially reported yet, so it’s very hard to make too much sense of it, but what I can say is that RIVM, which is the Institute in the Netherlands that performs the assay, they’re one of the world leaders in development of these tests. They use very sophisticated types of tests and they’re very accurate usually. And so, I trust that when they say that they have a positive that they probably have done their due diligence to actually know that that’s a positive and vice versa.
And so, they had a serological survey of blood donors and they found that amongst people who were blood donors whose blood was collected in the first week of April, they did find about a 3% positive rate amongst those individuals. And so that I think was a – well, you know, my guess is that it was a well-done study, but I can’t be sure because it hasn’t actually been published that I know of at this point.
The other studies that we’re seeing in the United States, I think, we have to look at them and start taking them, you know. We can look at them in aggregate and start to infer that maybe we can glean something from them, but we also have to be very mindful of the fact that they’re using these tests that can be inaccurate, and so we don’t know yet how inaccurate they are. So, of course, Santa Clara is now famous, you know, for potentially having this very biased and inaccurate study. We don’t know if it’s going to turn out to be incorrect. But we also don’t know if it’s correct. And so, we have to just be a little bit mindful of those things, and I would say that we’re all waiting to see what will be the first really good studies that come out and we’re certainly developing some of them here at Harvard. I know that Mike Bush out in Seattle is developing some nice ones. I’m guessing New York Blood Center and a lot of different institutions are starting to develop serological surveys, as well.
And so that’s something that, we will see a lot of these come out over the next month. Some of them will be high quality, some of them will be low. But what we can say is we can start aggregating them once we get a larger sample size. There was the New England Journal paper, which wasn’t a serological survey, of of pregnant women in New York that I think did give us a glimpse at least of what could be happening in New York, at least within a population or a region of New York City, and that was about 15% or 30% of women were virus positive. Now the interesting thing there is being virus positive suggests that if you actually did a cumulative incidence studying with serology probably it wouldn’t be 15% but it might be 30% or more, you know. And now it’s hard, necessarily without knowing the full dynamics of the epidemic to know how the cross section of viral positivity translates to cumulative incidents from serological data.
But that is a pretty telling investigation, at the very least, to tell us just how transmissible this virus is. And I think we’ve, of course, seen it on Diamond Princess, and in many other settings. But now we’re seeing it in a whole region of a city like New York, where we’re seeing that the virus is probably just transmitting at a very extraordinary fast rate in those settings.
Q: So, it’s still too early for us to predict when they’ll achieve the herd immunity right now, right?
MICHAEL MINA: Yeah, because we still don’t really know if it’s – we still don’t know I would say, if it’s 1%, 2%,10%, or 20%. And, you know, we’ve done a pretty good job, all things considered, with social distancing and that the benefits there, the more successful we’ve been with that aspect of this epidemic, the less herd immunity we will have coming out of this phase of it, which is, you know, a catch 22 a bit and we don’t know. What we can say is we won’t have herd immunity though coming out of this. I mean, I don’t think anyone believes that there will be herd immunity in May or June or July, you know, unless there’s a massive outbreak again and we don’t social distance.
Q: Thank you.
MODERATOR: Okay, next question. What do you think about the study the White House highlighted at yesterday’s briefing about UV light and humidity killing the virus and President Trump’s suggestion that bleach and UV could be used as treatments?
MICHAEL MINA: I didn’t hear him say that – I’ve been busy. Did you say that he said bleach should be used as a treatment?
MODERATOR: Um, so this is Nicole talking. From what I understand, there was – I think it was kind of an off the cuff remark about injecting disinfectant and that it could clear the infection out of the lungs, but I cannot give you 100%. I don’t have the news briefing in front of me.
MICHAEL MINA: I think I will decline to answer that, unless somebody else can sort of give me a more specific quote about that, if anyone wants me to comment on that and wants to give that. But I can say, I mean we do, you know, the question about UV light – UV should you know does kill viruses if in high enough power and close enough range, things like that. And this is part of, you know, what is the viability of a virus in warmth, in the sun when it’s desecrating and droplets are evaporating and those types of things. And I would say that at this point really good studies have not yet been performed to understand all those features, but the that data is coming. And we still don’t know though if it’s going to necessarily slow transmission in the summer or not. We have an extraordinary number of susceptible people in our population and the force of infection might be so great because of that that any benefit that might usually be incurred from summer months with normal viruses just might not be felt with this virus because of the extraordinary number of susceptibles.
MODERATOR: Next question.
Q: I heard you say earlier you still think that quite a few more people have been infected than we know, but the serosurveys that have come out so far really haven’t born that out. I mean, with the exception of the data out of New York City and State yesterday and the small study out of Chelsea, north of Boston, you know, most of the serosurveys are clustering around 3, 5, 9 percent, you know, single digits. Do you think that that’s just down to the methodology? I mean why do you continue to believe that there are a lot more people?
MICHAEL MINA: No, that’s what I think. I mean, I would say that that’s more people than our PCR-based testing has resulted in and I would say that if we start to assume that there’s maybe 2, 3, 4, 5% of the population positive, I think that that’s actually quite a lot more people than our estimates from PCR base positivity, even when accounting for maybe 50%, you know, loss of sensitivity in terms of who’s actually getting tested.
So, I would say that I’m agreeing with those kinds of numbers. I don’t think it’s going to be much higher than that but when we look at sort of eight or 900 – I’m not sure what the current number is, 800,000 maybe for the states – you know, that’s 0.2% or something. And so, I think going up to 3% I would say is actually very much in line with what I would think. I think that probably the testing has failed to capture at a low – at a minimum, it’s tenfold higher. I don’t think we’ve caught through our PCR based surveillance and testing, that we’ve caught one out of 10 people with this infection. I think it’s probably been worse than that. So, I think that I’m thinking it’s at least an order of magnitude more than the number of people we’ve actually identified as confirmed positives through PCR.
Q: Can I follow up, then? On the flip side, how do we make sense of 30% in Chelsea, or one in five in New York City, 14% in the state? I mean, are those – nobody else is seeing that kind of density of infection. Do you question those results?
MICHAEL MINA: No, I don’t. I think that, you know, when we look at the transmissibility of the virus, it should scale with population density to a certain extent. The R0 value that we assume or that we calculate, which has been largely said to be two to three, but R0 is not just a function of the virus itself. It’s a function of the environment the virus is passing through. So, we certainly would anticipate, for example, that the R0 in New York City, or Chelsea,
or other very urban and densely populated areas will actually be a higher R0 than the R0 that we calculate for somewhere in the Midwest, for example, in a low-density place, or in Western Mass.
And so, I think that they are very congruent with each other. I absolutely believe the data that’s coming from Chelsea. I know that we have nursing homes that have had, you know, 50% positivity in this country. And that’s probably very common. We’ve had – clearly that this virus has spread very widely through certain populations. And then in other more rural or less urban populations, I would say that it’s spread more in line with what an R0 of 2 might look like. And so, I think that they’re actually – that, you know, it’s not that we’re not going to have homogeneous spread in any sense. This is not a virus that has homogeneous spread. This is a virus that has clusters of really, really high infection rates. And then, and then there will be areas where it’s just not so much.
Q: Thanks.
MICHAEL MINA: Sorry, I have – I don’t know if people can see me, but I’m laughing, I’m sorry, because somebody did just post the quote that Trump said. I’ll just read it. He said, ‘supposing we hit the body with a tremendous whether it’s ultraviolet or just a very powerful light. And I think you said that hasn’t been checked because of the testing’ – I don’t even know what he said there, and he said, he goes on ‘and then I said, supposing you brought the light inside the body which you can do either through the skin or some other way I think you said ‘you’re going to test that too’.
For whoever asked about this, I mean, we’re not going to be using UV light to kill virus inside of the body. This isn’t – I don’t know what he’s referring to, but I don’t think that that’s a modality that we are going to be looking at for treatment of people and it seems insane. So, go on. Sorry.
MODERATOR: No problem. Alright, the next question. Go ahead, you’re unmuted.
Q: Hi, Dr. Thanks for doing this. I was wondering if you could talk about some of the supplies for the technical material needed to produce testing kits, particularly reagents, whether there is enough global supply to meet the demand that we would need over the next month, and how an alternate to meet that demand, maybe based on conversations you’ve had with clinics or your insights into the laboratory environment?
MICHAEL MINA: Sorry. Did you mean for PCR, or serology, or both?
Q: For I’m sorry – for diagnostics mostly. For PCR.
MICHAEL MINA: Okay. So, for the PCR diagnostics, I think that we’re starting to see companies get into more of an equilibrium with this epidemic. In many ways, at least in urban areas, I would say testing is not exactly the limitation at the moment, and we’re actually seeing just getting the supplies to do the testing, like the swabs and tubes. This has been ongoing for a while now and it’s still ongoing. One of the most difficult pieces in this puzzle has been these 20 cent pieces of plastic that we just don’t have a ton of, and now it’s become really a tube limitation, so we’re doing everything from – you know, now we’re trying to validate if saliva alone can work instead of putting the swab that has to go in this tube of viral transport media.
So, we’re trying to get creative and see if we can use fewer materials that are not available. In terms of the actual supply chain for the PCR tests, I think that we’re starting to see Roche, and Abbott, and Palogic, Cepheid – they are starting to get their assays more available, but they’re still certainly limited. We have our assays that we use in the hospital. We are essentially rationing which ones we use at any given time. We have enough redundancy to be able to perform all the tests that we need at this point, and a lot of that’s because we’ve created this whole other effort at the Broad Institute that uses a whole different supply chain away from these big manufacturers, like Roche and Palogic. So, we have that, not as our backup but we’re putting a lot of samples into that supply chain and that assay and we’re essentially rationing our cepheid cartridges, for example, for just the most emerging cases that we need to have a result back in 45 minutes instead of, you know, six hours.
And so, we’re still seeing that there’s difficulty and this is in Boston, you know, and this is – our hospitals are generally some of, sort of, the priority list hospitals for a lot of these companies because we buy a lot from them and we, you know, push technology first, a lot of times. So, I don’t actually know enough, you know, from a first-hand experience to know what’s going on in the enormous number of small towns and cities that we have across the country. But I can’t imagine that they’re in a good spot with regard to testing.
Quest and LabCorp have certainly picked up their ability to perform a lot of these tests and they’ve gotten their turnaround time back to a somewhat reasonable turnaround time. It was up at like seven days at one point, which makes the test practically useless. It’s gotten back down to a day or two, which has been good. But clearly, there are still – you know, whether that’s some bit of reagent shortage, or whether that’s accessioning and just the workflow, you know, this is an unprecedented number of people wanting a specific test.
So, it’s unclear to me, really, what that horizon looks like in the long run, but I do know that they’ve been working and now it’s coming into fruition, where these companies have sort of co-opted other aspects of their supply chain and manufacturing lines to produce an increased number of reagents for this assay. I know Hologic, for example, is probably going to be applying for an EUA for a whole different assay that’s not a real time PCR, but that falls under their isothermal amplification. So, they’ll have two totally different assays which use different reagents for the same instrument. So that is a signal to me that, you know just scaling up the one assay was probably not feasible and now, for them they’re actually pulling from two different supply chains for their own manufacturing. So, I think this is clearly still an issue. It’s getting better, but I don’t anticipate that, you know, in the next month, we’re going to see that the country is flush with test capacity.
Q: Thank you. Just quickly, is the picture significantly different on the serology side for serology testing?
MICHAEL MINA: Well serology is, you know, it’s going to be the same thing, I think, to be honest. We have DiaSorin, and Abbott, and Roche coming, you know, to the front of the line to get their assays developed. We see, of course, this huge plethora of all these point of care tests which frankly shouldn’t be used. I think, you know, they’re unreliable and until there’s a few of them that have been really vetted well and quality control is very high, I would say, you know, we have to be focused on more central lab diagnostics for serology. And I do think that we’re going to see some bit of – we will see that serology also runs into the same bottlenecks as PCR. My anticipation is that serology and antibody testing is going to play an even more crucial role in the long run for this virus.
And part of that is because testing for virus in a nasal sample when you’re doing population studies is a little bit inefficient. It’s rare that you’ll happen to find the person with the virus in their nose for that five- or eight-day period of time when you’re doing population surveillance. But antibodies, on the other hand, will be very efficient to tell you who has been exposed already in the past. So, I anticipate that serology is going to ramp up and it may dwarf the amount of tests that were required and the demand for testing from the PCR side. And so, we are trying to develop very high throughput serological testing, for example, here. And I think we’re going to see different laboratories trying to do the same.
MODERATOR: Alright. Next question.
Q: Actually, my questions were really following up on what you were just saying. I know there are various estimates of how many tests we need to be doing per week to get to the point where we can reopen the economy and, obviously, we know there are so many challenges as you’ve been discussing. Is there anything else that we as a nation can be doing to expand testing and get to that goal of more safely reopening the economy?
MICHAEL MINA: Well, we need to do the studies first and I think a lot of people have heard at this point that there are questions about what exactly it will mean to be antibody positive.
If you have antibodies, that is a good indicator that you were exposed and infected. And that’s one thing. And that’s oftentimes what a point of care type of device can provide for you is to just know how have you been infected. We know that if people have been infected and have recovered, this virus will probably not be extraordinary in terms of the immunological response it elicits. And by that I mean, we know a lot about respiratory viruses. We know that generally when somebody gets a respiratory virus, that they develop an antibody response and an immunological response and that generally tends to offer them some protection for a period of time. Even without the studies have been done yet, we don’t anticipate that this virus will be any different.
In many ways, the only extraordinary thing about this virus may well be the sheer number of susceptible people that it impacts and the severity that it seems to have a higher age classes, but in terms of the immune response, the kinetics of the virus looks similar to other viruses. And so why I’m saying all of that is because on the one hand, having a binary antibody test yes or no, positive/negative could be beneficial and scaling those point of care tests up could be useful. But what’s really going to be useful is understanding are there thresholds of protection. Do you have to have above a certain antibody level to essentially correlate with protection from subsequent infection?
And that’s something we don’t know yet. We don’t know what that level is. And there’s studies going on to understand it, but ultimately what needs to happen is that the antibody tests that are being developed need to be quantitative. They need to tell people how much antibody do you have. Are you protected or did you just have sort of a brief exposure, for example, and end up not developing a very robust response, or are you elderly and you had a very bad infection and you actually still didn’t develop a particularly good response.
And so, we don’t know a lot of those questions. We actually have really no good understanding with this virus what the correlates of protection will be. So, to really be able to do our due diligence and use serological data to help us get people back to work, we need to develop quantitative testing that can be performed at high throughput. And those will generally not be these point of care devices, and we need to make that available to the population. And so how to do that is a difficult task. You know, you can’t have phlebotomists go to everybody’s house to draw their blood so that they can know if they can go into work. So, we have to come up with new ways to test antibodies. That could be from saliva, that could be from a finger prick on a dried blood spot card – there’s ways to do it more simply.
I don’t know if people can see me. This is a little device that we use – can you people see me or no? I don’t know if I’m on video or not.
MODERATOR: Yes, you are. You’re on video.
MICHAEL MINA: So, these are like little devices that I use in the research space and it’s a little micro needle device. It’s a patch, it’s a plastic patch, but it sits on your arm and you can push a button and it will collect 100 microliters of blood, which is more than enough to do serology on some of the new tests that are available. And so, there’s new technology in the pipeline to really make getting people tested more widely available, but it’s going to be a while until that’s there. And we can’t – our system, our testing system for clinical tests is based on hospital laboratories for the most part. And that means it’s not – we don’t have a system that’s ever been built for surveillance, for wide scale population surveillance, or wide scale testing for people who aren’t presenting to the hospital or the clinic.
So that means that we have just a – the demand is just so much larger than our system was built for. Most of the testing facilities are in hospital labs, and then you have Quest and LabCorp, and even those were developed with clinical indications in mind. So, the idea that we have testing being requested from every employee across the country is an extraordinary ask, and it’s going to require new testing facilities, I think. And we’re actually building one now specifically for this reason, or we’re in the early stages of building it. And you know, I think that these are the types of things that are going to need to happen moving forward.
MODERATOR: Thank you. Next question. He would like to know – he’s asking for your reaction to the fact that Georgia has begun to open up and other locations are also considering easing up restrictions among the current pandemic. Are you worried, and if so, what is your biggest concern?
MICHAEL MINA: Absolutely, I’m worried. There’s no question that that should be front and center on anyone’s mind opening this up. It is a serious concern. We don’t know how quickly this virus is going to bounce back. We know a few things and one of those things that we know is we haven’t built up herd immunity, at least not – we’ve built a very partial herd immunity, maybe, you know, very partial. And we really don’t know enough about the virus to know if we start opening up hair salons and restaurants and, you know, all the things that we’re normally used to in our society, what will happen when people sort of start mingling again. And I think there’s a lot of fear that if that happens without the appropriate surveillance efforts in place that we could find ourselves in a situation again like we just found ourselves in at the beginning of two months ago.
So, I’m very nervous that the virus will bounce back. I’m also nervous that may be the virus won’t bounce back because of the summer and people will become very comfortable and then it will sneak up in the fall and lead again to massive outbreaks. So, I think those are the things that I’m nervous about.
I also am very cognizant of the fact that we need to have a serious discussion about economics and the toll social distancing on individual life livelihoods that go far beyond the scope of infectious disease epidemiology. So far, the conversations have been driven by the epidemiology and risks associated with the healthcare system and associated with the virus itself, and that’s essential, that’s absolutely essential, because we can’t function as a society of our hospitals are completely overrun. It’s just not a place we want to be. But we also need to be extremely aware that there are millions and millions and millions of Americans who live paycheck to paycheck and are currently out of paychecks for weeks, months now.
You know there’s a lot of people who actually can’t purchase food at the moment, and these checks that are coming in, you know, are not necessarily sufficient to abrogate that. And so, there’s this whole economic facet to this epidemic, which is unprecedented as well. And we’ve so far been tackling this through a public health lens, having to do with infectious diseases, but I think we have to tackle this through a public health lens dealing with the economics of poverty and people who aren’t even normally impoverished now becoming impoverished.
And so, all of that does have to be taken into consideration. I’m not advocating for opening up businesses again at the moment. I think we should be waiting until infections are even more well controlled and surveillance efforts are put in place. But I do think that I can at least see that side of the coin and understand, and I recognize acutely that this is a serious problem that, you know – I think the biggest fear of those of us who have advocated for social distancing is that those policies will themselves become more detrimental to society then had the virus just blown through the population. You know, that would be the worst-case outcome. So, we don’t want to see that.
MODERATOR: Thank you. Next question.
Q: Hey, can you guys hear me?
MICHAEL MINA: Yes.
Q: I have a question about smart watches and fitness wearables and I came in late, so sorry if you answered it or sorry if you don’t know, but I’m writing about smart watches and fitness trackers. Do you think they’re efficient way to potentially tracks symptoms of COVID-19? I know there are a lot of studies going on right now with them, especially with Fitbit and I was wondering, kind of, what you think as an epidemiologist are both the pros and the cons for using them to track health issues in general, right? Not necessarily COVID, although obviously that’s the main thing with using that technology.
MICHAEL MINA: I think they’re incredible devices. If networked appropriately, I think they can provide the type of data that is needed to be able to use it for tracking COVID and other infectious diseases and health issues. I think what’s needed first is to really aggregate that data and do supervised learning algorithm. So essentially, take all of this data and then say, okay, we know that these people got sick and these people didn’t, or we know there was an outbreak here and there wasn’t an outbreak here. And can we glean from the patterns that are inherent in the data coming from these watches and smart devices something that gives us an accurate signal of who might be infected.
And these kinds of issues have been – this type of use of medical and healthcare data has been really pushed for, you know, years, a decade now, to try to use AI and machine learning to start gleaning new insights from otherwise seemingly mundane data. So, things like subtle differences in patterns maybe of breathing or of heart rate, you know, while somebody’s sleeping relative to their baseline might be an indicator, for example, of being sick and things along those lines. But we first need to gain the data, evaluate it probably using artificial intelligence neural networks and those types of things, and create algorithms that can quickly identify somebody based on some of these patterns that these devices, pick up to say, hey, this is your probability of being sick today, you might want to get tested for example. So, I think that there are great inventions and I’m excited to see where that data goes and how it ends up becoming used.
Q: Are there any, you know, should we look at the data with any – I mean, I guess everyone I’ve talked to everyone said we have to look at all kind of medical data with a grain of salt regardless. But do you think that there’s anything specifically that stands out to you with smart watches and fitness wearables where you’re like, okay, maybe there was a spike in heart rate because you were watching, you know, a scary movie, or maybe your temperature went up because you were sitting in the sun all day at the park? Are there any kind of glaring things we should look at? You know, don’t rely on it too much because there could be something else there. Do you think for the most part it’s pretty okay to look at this, to use the data in the preventive virus type of way?
MICHAEL MINA: Well, I would say that I don’t think it’s okay right now to use it. I think we need, again, to figure that out with by collecting the data first, comparing it to actual known, sort of, gold standard results, and then deploying AI to figure out how well it’s working
and what the algorithms are. But it’s one of the – so your example was, you know, is this spike happening because somebody is just more active today. So, we would want to see trends. We wouldn’t want to say, oh, this person just got infected, you know, and now
for the last 10 minutes has had a different heart rate and make any assumptions based on that. But that’s why, for example, I said, maybe when people are sleeping you can follow their patterns, because that’s something that people tend to do relatively stably over time. Those values that when somebody sleeping tend to be consistent.
But there’s other features that can happen with the AI and artificial intelligence approaches to looking at this sort of longitudinal data across people that can be very, very accurate at times, but we don’t know. It could be that we end up going down this road, and these companies deploy all of this, all these algorithms and actually don’t find it accurate and predictive model.
Q: But it’s something to look at something at least look into right now because of how intimately they track your body?
MICHAEL MINA: Absolutely.
Q: Okay, cool. Thank you.
MODERATOR: Next question.
Q: The question I wanted to ask relates to testing for acute disease. You know, with the scarcity of tests, I’m wondering why really severe cases are being tested in hospital. If they look like a duck, and walk like a duck, and quack like a duck, why can’t you just assume they’re a duck? Wouldn’t it be more useful to be using the available tests to find mild cases in the community so you could put them into isolation and quarantine their contacts?
MICHAEL MINA: Absolutely. I think that there would be a potential role for that. You might hit up against a lot of angry physicians and by that, you know, being a doctor is – I think a lot of physicians, they want the result, you know, even if it’s just this confirmatory idea, that they actually get a true laboratory result. And this is why now laboratory medicine has only increased in its utility in the healthcare system over the years where, you know, estimates are that 80% of all medical decisions are based on a laboratory or radiological data.
So, I fully agree that we have now probably learned enough about this virus that at least in some hospitals, the clinical acumen is sufficiently high at the moment to probably be able to diagnose individuals without a viral PCR test. And I think that that would be something that some people would feel more comfortable than others with doing, but the culture of medicine in a hospital setting is just so far removed from that very sensible thought. The culture of medicine is, you know, if you don’t have the diagnosis sitting there on data, then it’s not, then it’s not a diagnosis, it’s a presumed possible maybe diagnosis. So, it’s very sensible, but it’s really far away from it.
And I also would say that the other reason is the actual number of patients being diagnosed in the hospital setting maybe isn’t, you know, relative to what you might want to do to take those tests and bring them out into the community. The numbers are on different scales. And so, there’s also this idea that, well, you know, if you’re just using it to diagnose the severe cases, who are clearly severe enough that you could know it anyway, it’s actually not a huge number of people compared to what you would really need to deploy in the community. So, there might be that psychology as well, where the trade off in terms of tests available just isn’t a huge trade off.
Q: It’s not as if, if you test somebody who seems to have ARDS and whether it’s SARS CoV-2 ARDS or something else, I mean, you’re going to treat them the same way, right?
MICHAEL MINA: You will. You know, the PPE that you use will be different in this moment in time, so the personal protective equipment which has become as critically limiting as the testing has been right now. For somebody with COVID, the interactions with those individuals is in full gowns and masks and, sort of, facials and all of that. And so, that’s actually also part of it. A lot of these tests are being used to rule people out as much as they’re being used to rule people in. But I do agree with you that a lot of times, the supportive treatment until we have actual treatment, it’s, you know, not necessarily a different approach. Once we have therapies, then there will be a difference in sort of how we approach the patient, but at the moment most of it is supportive care, unless somebody is on a trial.
Q: Thanks.
MODERATOR: Next question thing good Greenland from slate.
Q: I’m actually writing a story about the growing use of drones for COVID-19 detection. Some companies are advocating and claiming that they can use drones to detect fevers and even heart and respiratory rates in the sky. I’ve been looking into this. I’m a drone specialist myself and there’s essentially no scientific evidence or validation of the accuracy of this whatsoever, and there’s been very little real-world testing. And my question is, I am interested in your opinion on the validity of this technique for aerial screen of COVID-19 testing, but I’m also more broadly interested in the question of what is the danger of using untested technologies that lack real world validation or scientific evidence by authorities? The police agencies are largely using these for the purposes of determining which individuals may be more at risk of COVID-19. What are some of the dangers of these false positives in your opinion?
MICHAEL MINA: I had no idea about that, to be honest. I have thoughts about it, but I did not know that authorities were using drones to try to get at this. And, you know, without having given it a lot of thought, I would say that there’s a huge danger. I mean there’s dangers even, you know, having nothing to do with biology but having to do with ethics. To find that somebody is positive, you know, becomes a massive disruption to that person’s life at the moment, which, you know, from a public health perspective that that’s one thing, but to diagnose somebody or even partially diagnose somebody unwillingly, if that’s what’s happening to me feels wholly unethical at the moment, at least in our society and the type of medicine that we practice.
I don’t really think that it would be accurate, either. I can’t imagine – I might be wrong. Maybe there is enough data to actually create algorithms to do it, like I was talking about before with the Fitbit kind of idea. But I don’t believe that there is enough data or enough evidence or enough science to really suggest that we could accurately diagnose somebody with a drone.
But I think the ethics of it are first and foremost in my mind. I think the other dangers are, what do you do if you think somebody is positive. What do you do with that information. And that’s whether it’s from a drone or from a more traditional surveillance effort using swabs even, you know. If you think somebody positive, what do you do. Do you force that person to stay at home for the next two weeks? Do you take away their liberties and rights? You know, what do you do with that person? We don’t have a society that has really grappled with that in any that’s been at scale. So, it makes me particularly nervous.
And then the other thing is just the idea of using a test that might not be proven and that might not be accurate has big public health implications. If for every person you think is positive, if you’re totally nonspecific and you’re calling every other person positive, then that ends up becoming a huge drain on the resources. It ends up pulling personal protective equipment away from the hospitals where it might be most needed. And at the same time if you feel more confident that you should, that you’re actually capturing people who are infected, then maybe you end up letting – you stopped doing other surveillance efforts for the use of some invalidated tool and you end up getting outbreaks because you were unable to detect them as well. So, I think, you know, a lot of the problems biologically and public health-wise are similar to using unvalidated testing in the first place. And so, where the drones come in, I think a) I really doubt it and b) I’m very concerned about the ethics.
Q: Great, thank you.
MODERATOR: Next question.
Q: Hey, thanks. So, to follow up with the question about like a state like Georgia opening up perhaps too early. I guess. I wonder if one of the concerns seemingly would be if cases there start to increase, as they might presumably, if, you know, someone travels from Atlanta to Seattle or Boston and reseeds transmission there. And so, I guess I wonder how likely that might be, and what states that are still in these protective modes can do to try to be on the lookout for that type of thing?
MICHAEL MINA: It’s a really great question. It’s something we grapple with both at the national level and also at the local level. If a company chooses to go back to work, what are the ramifications? What’s the role of that company if they end up seeding infections to the wider public? So, I’ll speak to the state in a moment, but I’ll give an example.
If universities in Boston decide to open up, or universities anywhere, these are places that are generally filled with young people. So maybe the decision could be made that says young people can go back into the university setting and maybe faculty stay at home, for example. Now if that happens, and young people start transmitting the infection amongst themselves, but largely asymptomatically, for example, or with mild symptoms, then if you don’t look deep enough at the question you might say, okay, great, you know, we’re building a herd immunity. But then those young people will inevitably start seeding outbreaks to the wider community and can put older individuals at risk or just others – anyone else can be at risk.
So, I think that these are very, very difficult questions. This is this, you know, this age-old struggle between personal and public health, the role of the physician versus the role of the public health practitioner has always been a tension in medical and public health world. And then this idea that you bring up about states is kind of like a bigger version of that, where you might have Georgia decide to open up and they take that risk, Georgia takes this risk upon its constituents. These are elected officials, so, arguably, you know, they are trying to speak for the populace, so that’s something they do. If they end up causing major outbreaks that then continue to seed the rest of the country with cases and make life more difficult for states that are trying to mitigate transmission, I think it becomes a real problem.
And that’s where I don’t think we have good guidance for that type of issue from the federal government. There certainly hasn’t been. And I think that – I don’t know what the answer ethically is. You know, should people in certain states that are opening back up not be allowed to travel? What I would say is the good thing is that even if people travel, if they go to a different state that is still practicing social distancing, then ideally, that person wouldn’t necessarily transmit to other people because those other people are themselves protecting themselves from becoming infected. So, in that sense, there shouldn’t be necessarily a huge issue if those people end up traveling to places that are doing their due diligence and staying indoors. But it’s a hard question.
Q: Thank you very much.
MODERATOR: Next question.
Q: Well, I wanted to ask in the context of the ethics issue on drones, whether in a more likely scenario where you be using infrared devices at the workplace, whether you don’t have those same issues. And also, privacy issues seems like an important one, too, because if you’re going to be checking people at the workplace and you find someone positive, you need to share that information with all the people who know that person at the workplace. So, privacy seems like a tricky thing to deal with here.
MICHAEL MINA: Absolutely. And so that’s something else that we have been certainly talking a lot about. You know, both with privacy and just with serving people in general in the workplaces when we bring people back to work, whether we’re using some technology to test people or we’re actually testing them swabs and PCR, for example, what you do with those answers. Is this like a reportable disease that you have to report? Will somebody have to, you know, tell all of their peers that they’ve interacted with that they all may have been exposed, and then all of those people go home for four, or five, or six days or something until they get a PCR result back that says they’re negative? It’s unclear.
And I think these are absolutely crucial questions. You know maybe inside of a workplace, some of these issues are maybe better or more tenable to deal with than in the wider public because employers have more latitude over their employees, but I absolutely agree. I think there’s huge privacy issues here.
MODERATOR: Okay. Next question. So, this is the quote from Trump. He says, “I see that disinfectant knocks it out in a minute, one minute. Is there a way that we can do something like that, by injection inside or almost a cleaning? As you see it gets in the lungs and does a tremendous number. It would be interesting to check that.” And Trump did not specify the type of disinfectant.
MICHAEL MINA: (laughing) I can’t listen to it. I think that it’s just ridiculous. We don’t talk about therapeutics as disinfectants – that’s just not a thing. You know, if he wants to call an antiviral a disinfectant, well it’s just not a disinfectant. We’re not going to be pouring ethanol – well, maybe some people will be pouring ethanol – but we’re not going to be pouring bleach and these things into us. It will it will harm us. These are caustic agents and we can’t put caustic agents into us. If we could, then we wouldn’t worry about infectious diseases. We would just, you know, pour bleach into our bodies, but we would die in the process. It’s just, you know – it’s just really sad that our president is just so naïve to these fundamental aspects of human biology that he would think about you know saying something like that. It just doesn’t make sense to me.
We’re not pouring disinfectants into our lungs or our veins. And, you know, if we come up with antivirals and monoclonal antibodies, I guess semantically, you could think about calling them a disinfectant, but I don’t even want to – I mean you just wouldn’t it’s not it’s not medically relevant. You know, if you’re doing it on the skin, that’s one thing and that’s truly, you know you can disinfect a surface, but otherwise, you know, it’s a nonsensical statement.
This concludes the April 24 press conference.
Bill Hanage, associate professor of epidemiology (April 22, 2020)
Michael Mina, assistant professor of epidemiology (April 17, 2020)