WEBVTT - How Accurate Are Antibody Tests? 0:00:15.396 --> 0:00:23.556 Pushkin from Pushkin Industries. This is Deep Background, the show 0:00:23.596 --> 0:00:26.916 where we explore the stories behind the stories in the news. 0:00:27.436 --> 0:00:32.476 I'm Noah Feldman. As the COVID nineteen pandemic continues, antibody 0:00:32.516 --> 0:00:37.636 tests are gradually becoming increasingly available to ordinary people. You 0:00:37.676 --> 0:00:41.076 can get one online, and some employers are even requiring 0:00:41.396 --> 0:00:43.756 that people take them to go back to their jobs. 0:00:44.756 --> 0:00:48.756 But how accurate are these tests really? And if it 0:00:48.756 --> 0:00:51.036 turns out that you do have antibodies that show you've 0:00:51.036 --> 0:00:54.836 been exposed to coronavirus, what does that actually mean about 0:00:54.876 --> 0:00:58.716 what you should or shouldn't be doing. Next. Here to 0:00:58.756 --> 0:01:03.116 answer these questions is doctor Alex Marson. He's a biologist 0:01:03.196 --> 0:01:06.436 and an infectious disease doctor. He's a tenured professor of 0:01:06.476 --> 0:01:10.836 microbiology and immunology at the University of California in San Francisco, 0:01:11.236 --> 0:01:14.876 where he's the director of the Gladstone Institute of Genomic 0:01:14.916 --> 0:01:19.716 Commune Knowledge. Alex, thank you so much for joining me. Alex, 0:01:19.756 --> 0:01:24.676 your lab ordinarily does high throughput genomic engineering research, which 0:01:24.716 --> 0:01:28.316 is very cutting it stuff. And then when the COVID 0:01:28.396 --> 0:01:32.196 nineteen pandemic started, you went to your lab team and 0:01:32.236 --> 0:01:35.676 you said, okay, we're completely changing course. We're going from 0:01:35.716 --> 0:01:39.036 the highest tech to relatively lower tech, and we're going 0:01:39.076 --> 0:01:41.436 to look at the antibody tests that are out there 0:01:41.436 --> 0:01:44.556 and we're going to see if they work. So, first 0:01:44.596 --> 0:01:46.596 of all, what gave you the idea to do this? 0:01:47.116 --> 0:01:49.996 It's a great question, and looking back, it was really 0:01:50.556 --> 0:01:53.476 it was a confluence of a few things. It was necessity, 0:01:54.276 --> 0:01:58.116 and it was really in many ways motivated more by 0:01:58.116 --> 0:02:00.516 the people in my lab, or equally by the people 0:02:00.516 --> 0:02:03.236 in my lab as it was by me. Our lab 0:02:03.356 --> 0:02:06.356 was shut down. People weren't able to go in and 0:02:06.436 --> 0:02:09.316 work on their normal projects, but there was a real 0:02:09.356 --> 0:02:13.756 feeling in the lab that people had expertise and motivation 0:02:14.396 --> 0:02:16.836 to try to figure out what they could to be 0:02:16.916 --> 0:02:19.716 useful in the midst of this pandemic. It was really 0:02:19.756 --> 0:02:23.516 inspiring for me to see grad students and post docs 0:02:23.516 --> 0:02:27.756 and technicians who were looking for ways to contribute, and 0:02:28.076 --> 0:02:30.796 if anything, I was able to help channel and put 0:02:30.836 --> 0:02:35.236 together collaborations to enable their desire to help out. And 0:02:35.316 --> 0:02:40.476 so we saw a huge flood into the market of 0:02:40.956 --> 0:02:44.316 antibody tests that were becoming available. And actually one of 0:02:44.316 --> 0:02:46.956 the motivations for me was I got a text message 0:02:46.996 --> 0:02:49.236 from a friend of mine not in science here in 0:02:49.276 --> 0:02:51.996 the Bay Area, who showed me that she was testing 0:02:52.036 --> 0:02:55.476 herself for antibodies on one of these home diagnostic kits, 0:02:55.876 --> 0:02:58.996 And so I started wondering, what are the basic test 0:02:59.036 --> 0:03:01.636 performance characteristics of these kits? There were so many that 0:03:01.676 --> 0:03:05.116 were becoming available. We wanted to see could we assign 0:03:05.196 --> 0:03:08.716 some rough numbers to how reliable these tests actually are? 0:03:09.396 --> 0:03:12.276 And so that was really the fundamental goal of this 0:03:12.356 --> 0:03:14.876 study was to say, could we get our hands on 0:03:14.996 --> 0:03:20.116 as many of these test devices as possible and see 0:03:20.196 --> 0:03:23.676 what kind of information they actually give and don't keep 0:03:23.716 --> 0:03:26.996 us in suspensey longer? How were they so? There was 0:03:27.036 --> 0:03:31.796 a range. They're so polite. This is the boring answer 0:03:31.876 --> 0:03:36.716 that which was not totally unexpected. Some were reasonably good 0:03:36.756 --> 0:03:39.516 and some were not going to be very useful in 0:03:39.596 --> 0:03:43.036 this And one of the major determinants is how specific 0:03:43.076 --> 0:03:46.716 they are. And especially in a disease like this, where 0:03:46.756 --> 0:03:49.676 there's in many parts of the world is still relatively 0:03:49.756 --> 0:03:54.596 low prevalence, the chance of misinterpreting these results due to 0:03:54.636 --> 0:03:58.116 false positives is very high, and so one of the 0:03:58.156 --> 0:04:01.396 major things that we wanted to check was are these 0:04:01.396 --> 0:04:04.516 giving us results that are going to be confused by 0:04:04.636 --> 0:04:10.756 false positives, perhaps because they're misinterpreting antibio against other viruses, 0:04:10.756 --> 0:04:15.116 like other common coronaviruses that cause common colds. So we 0:04:15.156 --> 0:04:18.476 wanted to see, in a population where we know there's 0:04:18.556 --> 0:04:22.636 no Saris CoV two infection, how many people in that 0:04:22.796 --> 0:04:27.236 negative population had antibodies on each of these devices, And 0:04:27.276 --> 0:04:29.116 so this was really, in many ways the meat of 0:04:29.116 --> 0:04:32.596 this study. We took one hundred and eight blood specimens 0:04:32.636 --> 0:04:36.196 that had been frozen down well before the pandemic, going 0:04:36.236 --> 0:04:39.276 back to twenty eighteen and before, and we asked how 0:04:39.276 --> 0:04:42.436 many of those blood samples showed evidence of antibodies with 0:04:42.476 --> 0:04:45.836 these commercial tests that were popping up, and we saw 0:04:45.876 --> 0:04:48.476 a range. Many of them found a lot of evidence 0:04:48.516 --> 0:04:51.996 of antibodies against saris CoV two in these samples where 0:04:52.036 --> 0:04:54.756 we know that it shouldn't be there, and so those 0:04:54.796 --> 0:05:00.156 results are really disabling for a proper interpretation of these tests. 0:05:00.476 --> 0:05:02.836 If we want to use these antibody tests to measure 0:05:03.036 --> 0:05:06.116 in a population how many people have actually been infected, 0:05:06.516 --> 0:05:09.596 they would cloud that picture strongly, and many of the 0:05:09.676 --> 0:05:12.356 cases detected would actually be false positives. And if we 0:05:12.396 --> 0:05:15.556 wanted to give an individual patient information. We could give 0:05:15.596 --> 0:05:18.516 them very misleading information if we gave them false positives 0:05:18.716 --> 0:05:20.836 and said that they had been infected with stars Cove 0:05:20.996 --> 0:05:23.516 two when in fact they had it. But I want 0:05:23.516 --> 0:05:26.276 to caveat that there were a handful that seemed reasonable, 0:05:26.396 --> 0:05:30.076 and even one test showed one hundred percent specificity, meaning 0:05:30.116 --> 0:05:32.756 no false positives at all, meaning no false positives in 0:05:32.836 --> 0:05:36.716 the limited number that we tested. I've been thinking about 0:05:36.716 --> 0:05:38.716 the fact that in many ways, the way that people 0:05:38.796 --> 0:05:41.916 responded to our results was a Rorschach test of how 0:05:41.956 --> 0:05:44.276 they wanted to see these results or how they were 0:05:44.276 --> 0:05:48.036 predisposed to see these results. And I've been struck by 0:05:48.196 --> 0:05:50.676 some of the news coverage, which really ranged from saying 0:05:50.876 --> 0:05:56.636 antibody tests show great promise to these are complete hooks. 0:05:56.716 --> 0:06:00.116 And my true interpretation is actually somewhere in between that 0:06:00.356 --> 0:06:02.796 in some ways we got out in front of ourselves 0:06:03.116 --> 0:06:07.316 where in response to a pandemic, many many suppliers started 0:06:07.396 --> 0:06:11.396 racing into this market, and it wasn't a total shock 0:06:11.476 --> 0:06:14.756 to me that we saw arrange. I think what was 0:06:14.756 --> 0:06:17.556 a bit surprising was that these tests were becoming available 0:06:17.636 --> 0:06:22.396 to individuals in some cases before this basic information was available, 0:06:22.716 --> 0:06:24.316 and so we felt like we were just building in 0:06:24.356 --> 0:06:27.836 that gap. One thing I've been sensing a lot of 0:06:27.916 --> 0:06:30.956 recently is that as more and more states begin gradual 0:06:31.076 --> 0:06:36.036 kinds of opening, lots of people are now saying, gee, 0:06:36.076 --> 0:06:39.036 you know I was sick in March at one point, 0:06:39.556 --> 0:06:41.436 or I had a long lunch with someone whom I 0:06:41.596 --> 0:06:44.436 found out later turns out to have had it. Maybe 0:06:44.436 --> 0:06:48.396 I should look into having an antibody test now. And 0:06:48.956 --> 0:06:51.236 one question that they asked me, probably because they have 0:06:51.356 --> 0:06:56.076 heard me talking about your results, is how reliable are 0:06:56.156 --> 0:06:59.356 these tests now? I don't know what to tell them, 0:06:59.396 --> 0:07:02.316 So I'm asking you, what would you say under these circumstances. 0:07:02.356 --> 0:07:04.836 I have companies on the whole taken the lessons of 0:07:04.876 --> 0:07:07.636 your research and figured out so that more of the 0:07:07.636 --> 0:07:09.556 tests that are available now are like the better ones 0:07:09.596 --> 0:07:13.036 that you saw. So there were probably at least one 0:07:13.116 --> 0:07:16.076 hundred different companies that are offering different tests, and these 0:07:16.116 --> 0:07:19.476 will pop up in different in different settings. We tested 0:07:19.916 --> 0:07:24.556 sampling of these. We tested about twelve of these different tests, 0:07:25.236 --> 0:07:28.796 and we have a website available so that people could 0:07:28.956 --> 0:07:33.196 compare our results to whatever test information may be becoming available. 0:07:33.276 --> 0:07:34.996 How would people find that website? What should they be 0:07:34.996 --> 0:07:39.956 looking for the COVID Testing Project dot org. Now, keep 0:07:39.956 --> 0:07:41.876 in mind it's a preprint, it has not yet been 0:07:41.916 --> 0:07:44.516 peer reviewed, and it's a small sample, and so it's 0:07:44.516 --> 0:07:47.716 not intended to guide any kinds of clinical interpretation, but 0:07:47.796 --> 0:07:51.716 it provides some basic information. We've been in conversation with 0:07:52.076 --> 0:07:54.716 a larger testing effort that has now come up as 0:07:54.756 --> 0:07:58.276 part of a governmental effort. The National Cancer Institute is 0:07:58.316 --> 0:08:02.076 now doing a large test in concert with the FDA, 0:08:02.156 --> 0:08:05.796 where FDA is going to be assessing tests for antibodies 0:08:05.796 --> 0:08:09.556 going forward with the National Cancer Institute. And there's another 0:08:09.596 --> 0:08:12.276 website on the National Cancer as website and on the 0:08:12.276 --> 0:08:15.676 FDA website that you can look and find information about 0:08:15.836 --> 0:08:19.356 commercial tests that are now undergoing evaluation by the FDA. 0:08:19.476 --> 0:08:21.836 So the short answer is, let the buyer beware. Right, 0:08:21.996 --> 0:08:24.356 you may not even be able to find reliable information 0:08:24.396 --> 0:08:27.676 online about whether a given test you're taking is reliable 0:08:27.756 --> 0:08:30.796 or not. That's true. I think we're starting to see 0:08:30.836 --> 0:08:34.196 now some of the larger commercial vendors that have traditionally 0:08:34.196 --> 0:08:37.796 been major suppliers of lab diagnostics entering into this field 0:08:37.836 --> 0:08:40.476 and playing a bigger role. That these large vendors get 0:08:40.476 --> 0:08:43.516 into it. There's a hope that there's more consistency and 0:08:43.876 --> 0:08:48.156 more quantitative information on the levels of antibodies and also 0:08:48.236 --> 0:08:52.596 on the test performance characteristics. With the rapid diagnostics, not 0:08:52.676 --> 0:08:56.556 only is there variation among the different vendors, but there's 0:08:56.596 --> 0:09:00.076 some anecdotal reports that even within something bearing a label 0:09:00.116 --> 0:09:03.956 of one particular vendor, there may be batch variation, and 0:09:03.996 --> 0:09:07.316 so the picture is even more clouded with these rapid diagnostics. 0:09:07.636 --> 0:09:10.316 So the one takeaway there would be, if you really 0:09:10.316 --> 0:09:12.876 really feel like you have to get the test, send 0:09:12.916 --> 0:09:16.036 it away. Don't do one of these rapid diagnostic tests 0:09:16.036 --> 0:09:18.276 at home. If it looks like a pregnancy test, probablesion 0:09:18.356 --> 0:09:21.076 rely on it. So we've been thinking a lot about 0:09:21.116 --> 0:09:24.076 this going forward, about what are the possible ways that 0:09:24.116 --> 0:09:28.716 you could have an efficient testing algorithm. So if someone 0:09:28.796 --> 0:09:32.076 does a home pregnancy test, what's the first thing they 0:09:32.076 --> 0:09:34.396 do with that information. If it's a positive, well, they 0:09:34.396 --> 0:09:36.436 go to their doctor and they get a lab based test. 0:09:36.956 --> 0:09:39.836 And so maybe there's some opportunity to do something like 0:09:39.876 --> 0:09:43.356 that where there's a more complex algorithm that could be devised, 0:09:43.636 --> 0:09:46.956 where there's multiple tests that are used for confirmatory testing, 0:09:47.196 --> 0:09:50.236 where it's either a combination of home diagnostics or a 0:09:50.236 --> 0:09:54.036 combination of home diagnostics and lab based to expand the 0:09:54.076 --> 0:09:59.516 testing infrastructure without sacrificing sensitivity or specificity. That requires some 0:09:59.596 --> 0:10:02.276 more thought about exactly how that algorithm is designed, but 0:10:02.316 --> 0:10:05.276 I think these types of test performance numbers are the 0:10:05.316 --> 0:10:07.756 basic building block that you would use to design an 0:10:07.756 --> 0:10:12.036 algorithm like that. Going forward, let's talk about what someone 0:10:12.036 --> 0:10:14.876 could actually do if they did one of these tests 0:10:14.876 --> 0:10:18.396 it was reasonably reliable and they got a positive. Given 0:10:18.436 --> 0:10:21.556 the relative uncertainty that's out there, what would it mean 0:10:21.596 --> 0:10:25.076 for someone who said, well, I've tested positive. If anything. 0:10:25.636 --> 0:10:29.756 I think that there's really two measurements of what it means. 0:10:30.276 --> 0:10:33.156 One is how likely is it that it's actually giving 0:10:33.156 --> 0:10:36.956 reliable information about whether or not you've been infected? And 0:10:37.076 --> 0:10:40.076 the other implicit question, which I think is what people 0:10:40.156 --> 0:10:42.756 really care about, is what information is it giving you 0:10:42.796 --> 0:10:45.716 about how likely you are to get reinfected in the future. 0:10:46.636 --> 0:10:49.876 And so let me tackle both of those. The first 0:10:50.076 --> 0:10:52.516 is we are getting to a point where some of 0:10:52.516 --> 0:10:56.996 these better tests, especially the lab based diagnostics, are starting 0:10:56.996 --> 0:11:00.916 to give reasonably reliable information about whether or not there 0:11:00.956 --> 0:11:05.836 are in fact antibodies present in an individual's blood. The 0:11:05.916 --> 0:11:08.876 second piece is much more complicated, What do we actually 0:11:09.196 --> 0:11:13.436 hell someone who has a positive antibody test. I think 0:11:13.996 --> 0:11:16.876 no matter how many times we say in the news 0:11:17.316 --> 0:11:20.356 that we can't yet tell someone if they have a 0:11:20.396 --> 0:11:24.196 positive antibody test, they're actually protected from future infection. People 0:11:24.276 --> 0:11:28.716 have such a strong intuition and desire for antibodies to 0:11:28.876 --> 0:11:33.156 mean immunity that I'm concerned that there will be an 0:11:33.196 --> 0:11:37.076 implicit behavioral message that people are safe from prior infection 0:11:37.396 --> 0:11:40.956 and should take on risks and go into the community 0:11:41.076 --> 0:11:44.076 and do things that they wouldn't otherwise do without real 0:11:44.116 --> 0:11:47.076 science to back up to that behavior. So I think 0:11:47.196 --> 0:11:51.156 right now we're starting the next round as a community 0:11:51.476 --> 0:11:55.076 of scientific fact finding to start saying how can we 0:11:55.116 --> 0:11:58.236 advise people about risk of future infection if they do 0:11:58.276 --> 0:12:01.556 test positive for antibodies. There's really a big range of 0:12:01.596 --> 0:12:05.236 what infectious disease doctors can come to expect from what 0:12:05.436 --> 0:12:09.396 antibodies and prior infection mean for the prospects of symptoms 0:12:09.436 --> 0:12:14.076 and contagion on reinfection, there's a few lines of evidence 0:12:14.156 --> 0:12:17.436 that we as scientists are really looking for that will 0:12:17.596 --> 0:12:19.876 firmly tell us that we can give the recommendation to 0:12:19.916 --> 0:12:23.636 someone that they will be protected. The first are starting 0:12:23.676 --> 0:12:26.676 to emerge now, and these are animal studies. There have 0:12:26.756 --> 0:12:30.196 now been a few animal studies, including one published just 0:12:30.316 --> 0:12:33.916 recently in Science by Dan Baruk's group that looked at 0:12:33.956 --> 0:12:39.476 Reese's macaques infected with SARS CoV two and then reinfected 0:12:40.356 --> 0:12:45.996 upon infection, The monkeys developed signs of immunity and reinfection 0:12:46.116 --> 0:12:48.716 was far less severe. There might be small amounts of 0:12:48.796 --> 0:12:51.796 virus that actually infected the monkeys, but they seem to 0:12:51.796 --> 0:12:54.636 clear it relatively and quickly and didn't have signs of 0:12:54.676 --> 0:12:58.156 infection similar to the first infection, So that's highly promising. 0:12:58.436 --> 0:13:00.676 How that will translate to humans remains to be seen, 0:13:00.996 --> 0:13:03.756 and there's two levels of questions. One is will people 0:13:03.916 --> 0:13:06.596 clear the virus and not have severe symptoms and will 0:13:06.636 --> 0:13:09.596 they be contagious because that's also something people care about. 0:13:09.796 --> 0:13:11.636 Can you go back to work and not worry about 0:13:11.636 --> 0:13:14.396 spreading it to more vulnerable people that you also come 0:13:14.396 --> 0:13:17.516 in contact with, and that will require time. We'll be 0:13:17.636 --> 0:13:29.396 right back in the monkey studies. Were they able to 0:13:29.436 --> 0:13:31.916 determine whether they were infectious to others or were they 0:13:31.996 --> 0:13:34.076 only able to determine how much they showed symptoms or 0:13:34.076 --> 0:13:36.436 clear the virus. It looked like they were really only 0:13:36.436 --> 0:13:40.396 able to see evidence of symptoms and quick viral clearance 0:13:40.436 --> 0:13:43.676 and really pretty limited levels of virus infection, if any. 0:13:43.756 --> 0:13:45.836 Because that seems hugely significant, right, I mean, if it 0:13:45.876 --> 0:13:47.836 were to turn out that what was observed in monkeys 0:13:47.916 --> 0:13:50.756 was also replicated in humans, so that you would get 0:13:50.756 --> 0:13:54.516 a much milder case the next time around, that would 0:13:54.556 --> 0:13:56.836 be very reassuring to individuals. But if you had a 0:13:56.876 --> 0:13:59.676 mild case and we're still infectious to others, we don't 0:13:59.676 --> 0:14:02.476 want you going back out into public at that point. 0:14:02.996 --> 0:14:04.676 And we definitely don't want you walking around and thinking 0:14:04.716 --> 0:14:07.396 that you're effectively immune because you could be as much 0:14:07.436 --> 0:14:09.636 of a spreader as the next person. Right, So that 0:14:09.996 --> 0:14:12.436 these data are incredibly important and to my knowledge or 0:14:12.476 --> 0:14:15.876 not yet available for this particular virus. So the upshot 0:14:15.876 --> 0:14:17.356 of that for an ordinary person, just to bring it 0:14:17.396 --> 0:14:20.076 back to the you know, our hypothetical person who's thinking, So, 0:14:20.116 --> 0:14:22.636 now this person takes a test, it's a send away 0:14:22.676 --> 0:14:26.756 test from a reliable deliverer. Having done due diligence, our 0:14:26.836 --> 0:14:30.276 person now thinks that he's been exposed and has antibodies, 0:14:30.636 --> 0:14:33.276 and he says, Okay, now I'm going to go out 0:14:33.316 --> 0:14:36.076 and go about my business and interact with people, and 0:14:36.316 --> 0:14:39.236 the takeaway for that person is not so fast. Yeah, 0:14:39.436 --> 0:14:41.796 you know this was brought home poignantly to me recently. 0:14:41.796 --> 0:14:44.396 I had a conversation with a close family friend who 0:14:44.476 --> 0:14:47.116 called me up and said, should I take an antibody test? 0:14:47.156 --> 0:14:49.676 I really want to be able to see my grandson? 0:14:50.476 --> 0:14:53.076 And I said, look, I would love to be able 0:14:53.116 --> 0:14:55.436 to tell you that a positive antibody test could safely 0:14:55.476 --> 0:14:58.276 mean this, that you could you would be protected, but 0:14:58.316 --> 0:15:01.596 we don't yet have that information. I so deeply understand 0:15:01.636 --> 0:15:05.516 the yearning to have that level of security, but trained 0:15:05.516 --> 0:15:08.316 as a doctor, I don't feel that I yet have 0:15:08.436 --> 0:15:11.436 enough information and to say that the test result would 0:15:11.436 --> 0:15:13.876 allow me to actually recommend that you'd be safe to 0:15:13.916 --> 0:15:16.516 go and change your behavior in any way that you 0:15:16.516 --> 0:15:19.356 wouldn't otherwise. And although it's hard to put numbers on 0:15:19.356 --> 0:15:21.796 these things, how confident would you have to be? I mean, 0:15:21.796 --> 0:15:24.436 I understand that as a physician, you want to be cautious, right. 0:15:24.436 --> 0:15:26.076 You don't want to say to somebody you know what, 0:15:26.156 --> 0:15:28.116 you'll be fine and then have it turn out to 0:15:28.156 --> 0:15:30.036 be the case that they're in some small tale of 0:15:30.756 --> 0:15:32.996 the data where they actually were still able to get 0:15:32.996 --> 0:15:35.516 it again or to give it again. But how confident 0:15:35.556 --> 0:15:38.916 would you have to be to say to somebody, yeah, 0:15:38.916 --> 0:15:41.236 you know what on the whole once you've had the 0:15:41.276 --> 0:15:44.076 positive antibody test, this is basically almost certainly going to 0:15:44.076 --> 0:15:46.316 be all right. Though I'm making you no promises. I 0:15:46.356 --> 0:15:48.796 would like to see the human data. So I just 0:15:48.836 --> 0:15:51.836 told you in detail about the experiment in the monkey model. 0:15:52.236 --> 0:15:55.796 I think that we really need some basic information from humans. Now. 0:15:55.876 --> 0:15:59.316 Some people have been advocating strongly for actually doing what 0:15:59.356 --> 0:16:00.836 I just told you was done to the monkeys, to 0:16:00.876 --> 0:16:02.836 actually doing that in humans, and there's been a group 0:16:02.876 --> 0:16:06.356 of scientists that have signed letters talking about advocating for 0:16:06.516 --> 0:16:09.836 human trials of actually active infection. Now, this has been 0:16:09.876 --> 0:16:13.196 done for other coronaviruses in the past, the kind that 0:16:13.276 --> 0:16:16.316 caused common colds, and for something that causes mild symptoms 0:16:16.316 --> 0:16:18.996 like a common cold, that may be an acceptable risk. 0:16:19.356 --> 0:16:21.676 The question is would it be an acceptable risk here, 0:16:21.836 --> 0:16:26.076 perhaps in a young, healthy individual. I would advocate that 0:16:26.076 --> 0:16:29.716 that's not necessary and perhaps not ethical. In this case. 0:16:30.076 --> 0:16:32.796 I think that there's still high enough rates of transmission 0:16:33.076 --> 0:16:35.796 that well designed studies and high risk individuals should be 0:16:35.876 --> 0:16:38.236 able to give us this information, not quite as rapidly, 0:16:38.476 --> 0:16:41.756 but rapidly enough that we can interpret them. So I 0:16:41.796 --> 0:16:46.476 think what we really need is a carefully designed epidemiological 0:16:46.476 --> 0:16:50.756 study that aggregates all the data from everyone who's had 0:16:50.756 --> 0:16:54.956 antibody testing and watches them carefully over time, especially in 0:16:54.956 --> 0:16:57.556 situations where they'd be high risk if they're healthcare workers 0:16:57.556 --> 0:17:00.956 and high incidence regions, and ask the question very carefully 0:17:01.076 --> 0:17:04.796 and numerically, what degree of protection do we actually see 0:17:04.836 --> 0:17:07.796 in the people who have antibodies? Do they get infected 0:17:07.796 --> 0:17:11.356 and do they spread to their context? Are we moving 0:17:11.356 --> 0:17:12.956 in a direction where we're going to have to rely 0:17:13.076 --> 0:17:15.636 on two kinds of tests simultaneously, where we're going to 0:17:15.716 --> 0:17:19.636 have to rely both on swab testing of whether people 0:17:19.636 --> 0:17:21.916 have the virus in real time and also on an 0:17:21.956 --> 0:17:24.876 antibody test. Or are we heading for a world where 0:17:24.916 --> 0:17:28.756 one of these will predominate over the other. I strongly 0:17:29.196 --> 0:17:32.076 believe that we need both, and we need to be 0:17:32.196 --> 0:17:35.836 very clear about what information we'll get from each of 0:17:35.836 --> 0:17:39.636 those types of tests. The virus testing is the gold 0:17:39.716 --> 0:17:43.916 standard for seeing who is infected with this virus SARS 0:17:43.956 --> 0:17:48.356 CoV two, and that is what happens when people get 0:17:48.356 --> 0:17:51.836 the nose or throat swab. Increasingly this is moving to 0:17:51.916 --> 0:17:54.636 saliva testing, which I think is very promising and perhaps 0:17:54.676 --> 0:17:57.556 more scalable. Those are looking for the presence of the 0:17:57.636 --> 0:18:03.076 virus itself inside of an individual. After about a week 0:18:03.596 --> 0:18:05.436 on average, we start to see the evidence of the 0:18:05.516 --> 0:18:09.316 virus itself will wane, and by about two weeks or 0:18:09.356 --> 0:18:11.956 three weeks with some variability, we'll see that people who 0:18:11.996 --> 0:18:16.396 have been infected will start to produce antibodies against the 0:18:16.396 --> 0:18:18.756 stars covi two and that's where we'll be able to 0:18:18.756 --> 0:18:21.556 detect the presence of antibodies, and those will stay up 0:18:21.556 --> 0:18:24.116 for some period of time, although the exact period of 0:18:24.116 --> 0:18:26.316 how long they'll be detectable we still don't know. We 0:18:26.396 --> 0:18:29.716 have to trace that out farther. But by looking at 0:18:29.756 --> 0:18:33.236 both of those, we'll get information both about early infection 0:18:33.556 --> 0:18:37.796 and later information about who had been infected in the past. 0:18:38.036 --> 0:18:40.996 And if we want to accurately put together measurements of 0:18:41.516 --> 0:18:46.476 prevalence across a population, transmission dynamics, mortality, we really need 0:18:46.516 --> 0:18:50.676 information from both of those, and likewise, for back to work, 0:18:50.916 --> 0:18:53.956 we also need information from both of those. We've talked 0:18:53.996 --> 0:18:58.036 a lot now about will the antibody testing help us 0:18:58.036 --> 0:19:00.156 determine who's safe to go back to work, and we've 0:19:00.156 --> 0:19:02.556 talked about the pieces of knowledge that we still need 0:19:02.756 --> 0:19:05.996 about immunity which will start to guide whether antibody tests 0:19:06.036 --> 0:19:07.916 can tell us whether some people are safe to go 0:19:07.956 --> 0:19:12.116 back to work. But virus testing itself is also important 0:19:12.156 --> 0:19:14.236 for knowing who it goes back to work. One of 0:19:14.276 --> 0:19:16.956 the key things that we know about transmission of this 0:19:17.076 --> 0:19:21.556 virus is that it often occurs from asymptomatic individuals before 0:19:21.796 --> 0:19:23.916 or they even know that they have symptoms, or even 0:19:23.916 --> 0:19:25.756 some people who may never go on to have symptoms. 0:19:26.036 --> 0:19:27.836 And the only way that we could really tell whether 0:19:28.076 --> 0:19:31.076 people are infectious is if we screen for the presence 0:19:31.116 --> 0:19:34.596 of the virus itself. Now this may sound unusual, but 0:19:34.636 --> 0:19:38.676 there was an article in stat News recently about a 0:19:38.716 --> 0:19:42.316 model for this from the adult film industry. It's almost 0:19:42.356 --> 0:19:44.996 too good, yes go on. In response to the HIV 0:19:45.116 --> 0:19:49.476 epidemic in a group of very high risk workers, there 0:19:49.516 --> 0:19:51.916 had to be a model for how to figure out 0:19:51.956 --> 0:19:54.876 how to pe get people back to work with relative safety, 0:19:55.716 --> 0:19:59.876 because most HIV testing that done is actually based on 0:19:59.996 --> 0:20:04.396 antibodies against HIV, but those come up relatively late. Again, 0:20:04.436 --> 0:20:06.916 they don't come up in the earliest acute phases of 0:20:06.916 --> 0:20:11.076 infection for HIV, they wouldn't come up for the earliest 0:20:11.076 --> 0:20:15.116 phases of SARS CoV two. So that means that there 0:20:15.116 --> 0:20:17.636 would be an individuals who could be acutely infected with 0:20:17.756 --> 0:20:21.076 HIV who would be infectious before they would be detectable 0:20:21.116 --> 0:20:23.636 with an antibody. And so the adult film industry has 0:20:23.676 --> 0:20:28.316 come up with a very aggressive testing strategy that depends 0:20:28.356 --> 0:20:31.476 not on the antibody tests used for most individuals, but 0:20:31.596 --> 0:20:35.316 on actual viral testing that would sensitively detect the presence 0:20:35.316 --> 0:20:38.716 of virus at earlier time points. And individuals have to 0:20:38.756 --> 0:20:41.756 get tested at very regular intervals to be cleared to 0:20:41.836 --> 0:20:43.876 go back to work. On the set of an adult 0:20:43.916 --> 0:20:46.676 film industry, I think it's every fourteen days they have 0:20:46.716 --> 0:20:48.956 to get clear to go back to work. And there's 0:20:48.956 --> 0:20:53.836 a whole infrastructure set up with thought about privacy and 0:20:54.116 --> 0:20:56.356 also coming up with a plan for how individuals would 0:20:56.396 --> 0:20:59.636 get treated if they were to test positive, and also 0:20:59.676 --> 0:21:03.116 how contact tracing would be done to identify people that 0:21:03.196 --> 0:21:06.436 have been exposed. And so all this has been actually 0:21:06.636 --> 0:21:09.516 carefully thought about in one industry where there's a high 0:21:09.556 --> 0:21:13.356 risk of viral infection. Now, as a result of this pandemic, 0:21:14.196 --> 0:21:16.636 all industries are at high risk of infection as people 0:21:16.676 --> 0:21:18.556 go back to work, and I think we have to 0:21:18.556 --> 0:21:21.196 give some similar thought. I don't mean to over extend 0:21:21.196 --> 0:21:24.676 this analogy. HIV is a chronic virus, sarrys. CoV two 0:21:24.796 --> 0:21:27.516 is not is something that will get cleared, and so 0:21:27.596 --> 0:21:30.356 the dynamics are quite different. But I think we have 0:21:30.436 --> 0:21:34.596 to think about how can we detect early cases, asymptomatic 0:21:34.596 --> 0:21:37.396 cases and have infrastructure in place to make sure that 0:21:37.476 --> 0:21:41.076 positive cases get treated and traced, to limit the spread 0:21:41.076 --> 0:21:44.676 of infection and to limit mortality and morbidity. The idea 0:21:44.716 --> 0:21:46.996 that we could borrow a protocol from the adult film 0:21:47.036 --> 0:21:53.236 industry is its delicious and it would certainly be fascinating 0:21:53.276 --> 0:21:56.276 if that ended up being being applied more broadly. Can 0:21:56.276 --> 0:21:58.476 I see a totally outside the box quirky question that 0:21:58.516 --> 0:22:01.276 I've noticed in talking to different people, I've noticed that 0:22:01.476 --> 0:22:05.996 serious scientists all say stars CoV two, and they almost 0:22:06.036 --> 0:22:10.476 never say COVID nineteen. Why it seems to me like 0:22:10.516 --> 0:22:15.316 a little insider outsider code. I've actually realized this recently 0:22:15.716 --> 0:22:18.676 talking to some people that this is confusing. I think 0:22:18.716 --> 0:22:21.356 the simplest way to explain this is by analogy to 0:22:21.676 --> 0:22:25.716 HIV and AIDS. AIDS is the disease caused by HIV 0:22:26.236 --> 0:22:29.276 and STARS. CoV two is the virus that causes COVID 0:22:29.436 --> 0:22:32.276 nineteen the disease, And so when we're really talking about 0:22:32.316 --> 0:22:36.716 the mechanics of infection, scientists will gravitate to talking about 0:22:36.836 --> 0:22:39.596 the name of the virus rather than the syndrome that 0:22:39.596 --> 0:22:42.756 it causes, since there's a wide range of outcomes of 0:22:42.796 --> 0:22:45.236 what the virus may actually do, and so COVID nineteen 0:22:45.316 --> 0:22:47.156