Transcript
You’re listening to press conference from the Harvard T.H. Chan School of Public Health with Caroline Buckee, associate professor of epidemiology and associate director of the Center for Communicable Disease Dynamics. This call was recorded at 11:30 am Eastern Time on Wednesday, April 1.
Previous press conferences are linked at the bottom of this transcript.
CAROLINE BUCKEE: Good morning everybody, so I’m an infectious disease epidemiologist. By way of background, most of my work is in understanding the transmission of pathogens through populations, so I’m happy to answer questions about any epidemiological transmission questions to do with COVID as well as models, forecasting, and projections, and sort of general epidemiological questions.
The work that I’m doing, in particular, is to try and think through how we can monitor social distancing and, ultimately, think through what we need to do into the future as well. So, I’ll just stop and welcome any questions as they come.
MODERATOR: Alright, if we can take our first question.
Q: Hi, Caroline. Thanks for doing this. I cover the NBA, which is shut down indefinitely since March 11, and they’d like to resume the season at some point. And one of the ideas they’re reportedly throwing around is the idea of creating a bubble, say, in Las Vegas, where they could quarantine the league and it’s [AUDIO OUT] or a hotel compound and then resume the season in that closed community.
Apparently, the Chinese Basketball Association has tried to implement this in some form right now. What are your thoughts on that as a possible solution to resume an NBA season? And how feasible is it to ensure no transmission takes place in a closed-off community? And if that is not realistic, then what options would you recommend to the NBA which has, potentially, billions of dollars at stake?
CAROLINE BUCKEE: So, I think it sounds like potentially a bad idea, especially at this stage of the epidemic, and there are kind of two reasons for that. First of all, I don’t think it’s realistic to completely isolate and quarantine the players. For a start, there are people who will need to clean their rooms, feed them, wash their clothes, you know, janitorial staff and so forth. And those people will not be protected, and they will be interacting with their communities. So, it’s very difficult to truly self-isolate.
The other thing I would say is that nobody is protected from severe disease, from COVID, so purposefully putting people at risk seems foolish. And then the other thing is that’s very important, I think, is that NBA players and the NBA are important role models for a lot of the country. And as people stop playing basketball themselves and parks and courts close around the country, I think it’s important that the NBA sets an example to show people that saving lives is more important than money right now.
Q: Thank you, Caroline.
MODERATOR: Our next question.
Q: Yes, so Dr. Buckee, I’m not sure this question is in your wheelhouse, but I’ll throw it out there. There’s a lot of data on the incubation period of COVID-19, but we haven’t seen much on the duration of the disease. So, from the time symptoms appear, if they do, to when a person is symptom free or however that should be looked at, I’m curious if you know, or if we know anything about the frame, and if that frame is different for people who get a mild case versus a severe case of the disease.
CAROLINE BUCKEE: So, what we know is that the clinical spectrum is quite broad, so it seems as if a significant fraction of people never have symptoms. We still don’t know the extent to which those people are infectious, but it’s becoming apparent that that’s quite a significant fraction of people. And then there are other people that have mild symptoms, as you mentioned, and, for those people, the symptoms seemed to last for, you know, anything from a few days to a week or more.
For severe cases and cases that end up in the hospital, we’ve seen quite a significant time lag between symptoms first arising and the patient weakening and requiring hospital care. So that’s been part of the issue with the time lags around hospitalizations and deaths because, actually, the time from symptom onset to death can take two weeks. So, the short answer is it’s very variable from no symptoms at all to a few days in mild cases all the way through to a gradual deterioration and, ultimately, death over a more extended period of time.
Q: Could you speak just a bit more to severe cases in people who get through it and survive? It sounds to me like that could be two, three weeks, possibly more.
CAROLINE BUCKEE: I don’t have the latest data in front of me, but, certainly, there are there’s evidence that people can be symptomatic for a week or two weeks. I would need to check for you on what the latest data is coming out of the CDC and China and everywhere.
Q: I’ll send a follow up email to –
CAROLINE BUCKEE: Sure.
Q: – follow up on that. Thank you for that.
MODERATOR: Our next question.
Q: Thank you so much for taking my question. We’ve been seeing reports about hospitals centers delaying things like chemotherapy and radiation treatment for cancer patients. Is this something that you would advise them to continue doing? And is there any concern about knock-on effects for health for these people?
CAROLINE BUCKEE: I’m not a clinical doctor, so I can’t comment on specific treatments and whether – the essential nature of particular treatments. What I would say is that there are going to be indirect effects of this pandemic, and those indirect effects are not just related to postponing treatments or, you know, putting off elective surgeries. But also, the shift in priority among many hospitals and health care centers to treat exclusively COVID patients means that many people that need care because they are diabetic or they have some other condition which requires frequent attention, a lot of those people are going to be particularly vulnerable.
And so, when we think through the impact of this pandemic, we need to think not only about deaths attributable to COVID but also any other mortality which might arise as a result of redirected medical care and the redirection of financial and health care resources. It’s a very important issue.
Q: Thank you so much.
MODERATOR: Our next question.
Q: Hi, Dr. Buckee. I have a question about asymptomatic or pre-symptomatic symptoms and their percentage in the population. There was a story about a study in Iceland that showed less than 1% of those who were tested were positive – this is of 17,900 people – but 50% of those were asymptomatic.
And I think there was a Journal of the American Medical Association piece today about a study out of Los Angeles of people who presented with flu-like illness but without COVID risk factors, and it showed about 5% of those were tested positive. I mean, are these studies enough? And at what point do you think we’ll know about the prevalence of this in the population due to asymptomatic or mild symptoms?
CAROLINE BUCKEE: I would say we still don’t have – we still need more information about asymptomatics. As you point out, so there’s been a number of sources of information about this. The cruise liner, we have information about that. There’s a study in China. There’s the Icelandic data set.
And although it seems like a significant fraction of people may have no symptoms, we won’t actually know the precise value until we do widespread serological testing. So that’s tests that look for antibodies that are specific to the virus, so it’s kind of evidence of a past infection. And then we’ll be able to start seeing what the overall size of the epidemic was and how many people were affected and couple that with hospitalization rates and so on. So, we don’t have a definitive answer, but it seems likely that a significant portion of people never have symptoms, which is a different thing than having symptoms but you’re infectious prior to the onset of symptoms, of course.
The other thing that I would say is that, in the Iceland study, for example, that’s really a testament to how much they’re doing tests. So, the fraction of people that are detected really depends on the testing strategy that countries are doing. So yeah, I guess we don’t have a definitive answer yet. It will depend on the demography of the population and the specifics of that setting, but, ultimately, we’re going to need serology to answer the question about asymptomatics.
Q: How important is it to get the serology test before we decide whether and how to ease up on social distancing? Are they related?
CAROLINE BUCKEE: Yes, they’re very intertwined. So, one of the ways that epidemiologists think through, for example, how many people you need to vaccinate to prevent transmission in a community is that we figure out the fraction of people that need to be immune in order to stop transmission from taking off. And that’s related to the reproduction number, so how transmissible a pathogen is.
So, before we start relaxing social distancing, we’re going to want to know that a significant portion of a particular community has antibodies to the virus. And only then can we really start to feel safe with respect to the risk of transmission taking off. There’s one big uncertainty, though, which is that we still don’t have good data on whether we get good immunity to this virus.
For other seasonal coronaviruses, there is some immunity, but it doesn’t seem to last much longer than a year. So many of the models are assuming that this coronavirus will show similar type of immunity, but we still don’t know that yet. So even if we get serological tests, we’ll need to be able to correlate antibody titers, so how much immunity you have, with protection against reinfection. And that’s another piece of data that we’ll need before we feel comfortable relaxing social distancing completely.
Q: If that’s the case that the immunity only lasts a year, does that negate in some way the idea of developing herd immunity and raise the importance of getting a vaccine?
CAROLINE BUCKEE: It doesn’t negate the idea of herd immunity because it is a form of Herd immunity. It’s just going to be waning over time, right. So, it’s herd immunity, but it’s short term. So, it does definitely suggest that we need to – we’re moving fast with vaccine development, and that’s a good thing.
And it could be that the vaccine may provide similarly – we don’t know the duration of protection from a vaccine either, so it may be something where we need to get vaccinated regularly. It just depends on how our immune response responds to this particular virus.
Q: Great, thank you.
MODERATOR: The next question.
Q: Yes, good. Thank you, thank you very much for taking my question. I’m feeling confused about the timing of the peak of the surge in the hospitals. For example, here, in Massachusetts, we brought in very strict social distancing a good three weeks ago or so, and yet we’re told that the peak surge in the hospitals is expected between April 7 and 17, which seems like it’s a long time after we all separated even though we know that it takes a couple of weeks for the virus to reach its sort of peak effect. Could you explain that timing?
CAROLINE BUCKEE: Yeah, well, first of all, your confusion is warranted because these models are very uncertain. So, one of the ways that the models work is that they assume that epidemic trajectory depends on a couple of factors, how transmissible it is, if you have contact with an infectious person, how long people are infectious, and then, critically, a contact rate, which is the rate at which people in the communities are contacting each other. So, to predict the epidemic peak timing, you throw those parameters into the model with your sort of understanding and assumptions about how it spread and then you kind of figure out what that means for the epidemic trajectory.
The social distancing interventions that we’ve put into place are going to have a severe impact on the contact rate, but it’s very hard – in the model frameworks that you may have seen, it’s very hard to directly link particular social distancing strategies with the contact rate in a mechanistic fashion except to say that it’s probably having a big impact. So, at the moment, I would say the timing of the epidemic peak and the timing of hospitalizations is still very uncertain.
And it’s going to take us probably several more weeks before we can really definitively say that social distancing interventions that we put in place two or three weeks ago are having the effect that we want them to have. So, I would say it’s too early to say and that those types of precise forecasts around when the epidemic peak will occur are very difficult to make. And I think there’s a lot more uncertainty around the timing than people may realize.
Q: So, it’s not even a good sign that, so far, we’re not seeing a huge surge in our hospitals because it’s just too uncertain?
CAROLINE BUCKEE: Well, I mean, it’s always a good sign if there aren’t patients coming in that we can’t manage, but I think it’s just to be cautious in interpreting that as just yet. I think we still need to wait and see.
Q: Thank you.
MODERATOR: The next question.
Q: Thank you so much for taking my call – or my question, and having this call. I kind of had some – I was hoping you could help me understand a little bit more about the transmission modeling in terms of pre-symptomatic and asymptomatic cases. And I know you mentioned before this is the question of what percentage of cases of people who are infected are asymptomatic or – as the WHO has mentioned many times, it seems like a lot of those cases that are initially dubbed asymptomatic, sort of transform into pre-symptomatic conditions later.
But I think the question a lot of people have is, how likely is it that people who don’t know they’re sick at all could pass on the infection? How much of that is happening? And so, there’s been some pre-print studies looking at modeling of this in terms of timing and incubation periods and suggesting sort of all over the map, like 60% have passed on their infection before their symptoms showed up to smaller percentages.
How much can we trust that modeling? How would you look at those different a wide range of possibilities there? And when you do this kind of modeling, is it always certain that someone who’s pre-symptomatic would have the same sort of contagiousness as someone who’s coughing? Which is something that WHO mentioned many times in terms of people who are coughing are much more likely to spread the disease. But if 60% of people might be spreading it before they have symptoms, how do those things fit together?
CAROLINE BUCKEE: Yeah, these are all excellent questions, so the short answer is that we still don’t know. So, in the models themselves, you basically classify people relative to different compartments with respect to the disease. So, they may be in a pre-symptomatic compartment. They may be in an asymptomatic compartment. They may be in a compartment that’s specific for their age group.
And there are lots of ways to split it up. But when you think about somebody’s risk of becoming infected, the way you model it is that you assign a relative infectiousness to each of those groups, and then you have to make an assumption about the contact rate between all of them, which you can imagine becomes quite challenging and uncertain.
And so, given that there’s quite a broad range of estimates of the asymptomatic category – so that’s one source of uncertainty. We don’t know what fraction that is. But then the other uncertainty, as you point out, is that we don’t know the relative infectiousness of these different groups. We don’t know whether asymptomatic people are just as infectious as symptomatic-infected people, probably not because of this issue of coughing and so on.
Although if they’re infected for longer, you might assume in your model that they’re more likely to be walking around and making contact with people as opposed to sick people who are self-isolating. So, all of these nuances have to become assumptions in your model. And that’s where a lot of the uncertainties in these projections come from.
And just to highlight one particularly important issue right now, we know that children tend to have – often have no symptoms at all or very mild symptoms, but we really don’t know the extent to which they’re contributing to transmission overall and how infectious they are. So that means that school closures, for example, it’s not clear how effective those are going to be and whether, potentially, we could relax social distancing around school closures if it becomes apparent that asymptomatic people don’t transmit very much.
At the moment, we still don’t know that. So, if you add up all those different sources of uncertainty, what models usually do to address that is that they run the model across all the possible ranges that we think are reasonable and then come out with a range of outcomes. And that’s why you see a lot of very broad kind of estimates and trajectories about what might happen. It’s reflecting those multiple steps in which we have uncertainty.
So, infectiousness is something that is going to be important to measure moving forward. And a lot of the work that’s been done so far, we have to kind of infer a lot of these parameters because we’re not getting direct measurements of them. The data is not there.
So again, it means that what we’re doing to put this into the model is saying, OK, we don’t know, but it could be across these range of parameters. And that’s true for a bunch of different parts of the model. And then so, overall, your estimate is going to be quite uncertain and cover a range of different outcomes. Does that make sense?
Q: Absolutely, yeah. Thank you so much. And just a quick follow up, I think this question, it seems more pressing now with all of the discussion about whether or not people in the public, if they think they’re healthy, should wear a mask. Could you speak to that and how that may, good or bad, could influence disease spread?
CAROLINE BUCKEE: Yeah, so I think, until very recently, there’s been the feeling that people shouldn’t be wearing masks. And I think part of that is the concern that the public is buying up masks that we need to make sure are reserved for health workers. We’re seeing a global shortage of PPE, personal protective equipment, and we’ve seen, in our hospitals in the US, that health care workers on the frontlines are having to reuse their masks and so on. And so, part of the concern around public messaging is to ensure that our frontline health workers have the protective equipment that they need going into work with COVID patients.
Now, we’re seeing the discourse change a little bit to suggest that maybe masks, as long as they’re not limiting masks for our health care workers, then potentially the public might benefit from wearing them, if not to protect yourself, to protect others if you are asymptomatic. And to be honest, I haven’t seen data on this. And I don’t think that there are yet big randomized controlled trials that actually show strong evidence of protection from mask wearing. But that’s not to say that they won’t help. I just I can’t comment on the efficacy of masks for asymptomatic people right now because I haven’t seen much data on it.
Q: Great, well, thank you so much. This has been super helpful.
CAROLINE BUCKEE: Sure.
MODERATOR: Next question – he asked me to relate it to you, Dr. Buckee. We found out today that the Florida Department of Health is working with a Harvard-led consortium using Facebook data to evaluate the effectiveness of social distancing measures using cell phone location data.
The group is called the COVID-19 Mobility Data Network. We’d like to get your thoughts on what we can learn from this and, given that we are one of the few states – Florida is one of the few states that doesn’t have a statewide stay-at-home order, what this could mean in that context.
CAROLINE BUCKEE: Sure, I’m happy to answer this. I’m actually leading this effort at Harvard. And just to give you a bit of background, so this is a network of researchers from a bunch of different research organizations. We have been working with existing data-use agreements and within privacy frameworks and ethical frameworks that have been kind of thoroughly vetted by the university, and we’re working with Facebook to use aggregated mobility data.
So, this is not identifiable at the individual level. And what we’re doing is taking information about general mobility patterns, for example, on a county level, how much are people moving around within the county, how far are they going on average, these kinds of metrics. And we’re working with state and local public health policymakers to give them situation reports and update them and help them to make decisions about public messaging around social distancing and figure out what’s going on with social distancing in their regions.
So, the types of insights are really quite simple at this point. It’s how much of the message is working. Are there places where we need to do more? Can we understand why some places are unable to respond to social distancing messages? We want to make sure that we’re not targeting or punitive in targeting social distancing efforts but, rather, ensure that the right people get the messages they need to protect them as best we can.
So that is ongoing. And we’re hoping that it will provide some sort of actionable information for policymakers as we’re moving into these social distancing interventions. Ultimately, we will want to be able to couple this analysis with their data on, say, hospitalization rate and overall cohort transmission because, very soon –
Well, it doesn’t feel like soon. But several months down the road, we’re going to have to relax social distancing, and we need to be able to know what the impact of different interventions and different public messaging is on human mobility and, ultimately, on COVID transmission so that we can have data to back up policies when we start to relax them. And so, we’re working to try to make sure that we have that evidence in hand and that we’re providing support, analytic support, locally to decision makers.
MODERATOR: He would like to know how long do you think it will take to get the useful data out of this effort.
CAROLINE BUCKEE: So, we’re already producing data, and we have partnerships with local and state governments that we are working with to provide them information on what’s happening in their state.
MODERATOR: OK, thank you. Next question.
Q: Dr. Buckee, I have two questions now, following on some of the previous comments. I’ll ask them one at a time first.
I understand we don’t know how many asymptomatic cases there are with the percentages. But given what is known, in your mind, how unusual is it that we have so many asymptomatic cases compared to other communicable diseases?
CAROLINE BUCKEE: It’s not very surprising. So, for many pathogens, there is a significant asymptomatic reservoir. So, whether that be dengue, there’s lots of asymptomatic people. Malaria, there are lots of asymptomatic people. Those are the diseases I work on a lot. But similarly, with streptococcus pneumoniae, which causes strep, lots of people are carriers and show no symptoms at all, similarly, with Neisseria meningitis, which can cause meningitis. So, a lot of different pathogens have a significant asymptomatic reservoir.
For an emerging pandemic like this, the issue with asymptomatics is that, A, it means that you can’t rely on your surveillance systems to passively catch cases because they show up in health clinics because you don’t know how many people have the disease if they don’t have symptoms. B, it makes it much harder to contain for obvious reasons. But the upside is that, if you do have a significant fraction of asymptomatic people and they are gaining protective immunity, then, hopefully, the overall impact of the epidemic is lower than if you didn’t have any asymptomatics.
Q: Fascinating. My second question, do you have any guess – and I know this is a tough one – any guess on what the multiplier might be right now, meaning the number of likely actual cases right now in the United States or, you know, just today or tomorrow, given that it’s a multiplier, given the number of diagnosed cases?
CAROLINE BUCKEE: So, the reason that that’s such a hard question is not just because we don’t have enough tests. It’s that we have a heterogeneous testing. Not just like – the tests aren’t everywhere, and there are different numbers in different places, but also the criteria for testing varies a lot, too. So, who is getting tested is different from place to place.
But I think the multiplier is going to be on the order of – I don’t know, ooh. I don’t know if I should try and guess on this kind of press conference, but I wouldn’t be surprised if we had a 5 to 10 times more cases than we’re seeing in some places, especially in areas where there’s very limited testing. We just don’t know how much transmission is happening.
Q: OK, yeah, and we had seen the 5 to 10 the figure from an early study of cases out of China, I believe it was. I’ll include that in my follow-up email in case you want to be a little more specific. I don’t –
CAROLINE BUCKEE: OK.
Q: – want to put you on that, given the hesitation in your voice there.
CAROLINE BUCKEE: Yeah, it’s really hard to make those estimates just because of this heterogeneous testing strategy and capacity.
Q: Yeah, OK, thank you.
This concludes the April 1 press conference.
Bill Hanage, associate professor of epidemiology (March 31, 2020)
Howard Koh, Harvey V. Fineberg Professor of Public Health Leadership (March 30, 2020)
Yonatan Grad, Melvin J. and Geraldine L. Glimcher Assistant Professor of Immunology and Infectious Diseases (March 27, 2020)