COVID-19 Webinar Series Special Issue – From Data to Decisions: The Role of COVID-19 Projections

April 2, 2020

COVID-19 Webinar Series Overview

This COVID-19 Webinar Series synthesizes the information in the headlines to provide cohesive insight into the status of the response and remaining gaps in the system that must be addressed to limit the severity of the COVID-19 outbreak in the United States.

Special Issue – From Data to Decisions: The Role of COVID-19 Projections

As the nation responds to the COVID-19 outbreak, leaders are using modeling and projections to make evidence-based decisions about balancing population safety, health system capacity, and the economy. During this pre-recorded interview, we spoke with Dr. Rebecca Katz, who explored how decision-makers can use these tools to weigh policy tradeoffs. She discussed the range of existing models, as well as the key inputs and assumptions that drive projections.


  • Rebecca Katz, Ph.D., MPH, Professor and Director, Center for Global Health Science and Security, Georgetown University; Collaborator, CoVidActNow
  • Sarah J. Dash, MPH, President and CEO, Alliance for Health Policy (moderator)

The Alliance for Health Policy gratefully acknowledges the support of the National Institute of Health Care Management (NIHCM) and The Commonwealth Fund for this event.


(This is an unedited transcript. For accurate quotes and presentations, please refer to the full-event video.)   0:09 Hello and welcome to a special edition recording in our covid-19 webinar series. 0:16 I am Sarah – president and CEO of the Alliance for Health policy for listeners new to the alliance. We are a non-partisan resource for the health policy Community dedicated to advancing knowledge and understanding of Health policy issues. We launched this webinar series to provide insight into the status of the covid-19 response and shed light on remaining gaps in the system that must be addressed to limit the severity of the pandemic in the United States. 0:43 You can find recordings of all the webinars in the series and additional resources on our website. 0:51 The Alliance for Health policy gratefully acknowledges the National Institute for Healthcare Management foundation and the Commonwealth fund for supporting our covid-19 webinar series, you can find all of the materials that accompany this recording on our website including a list of additional resources and expertise. 1:07 Models and projections can be useful tools to help policy makers make evidence-based decisions and leaders all over the country have been making indeed some very tough decisions over the last few weeks today. I am so pleased to be joined by dr. Rebecca cats to explore how decision-makers can use these tools as they weigh policies to slow the spread of covid-19 and prepare the health system to respond. Dr. Katz is professor and director of the center for Global Health Science and security at Georgetown University. 1:37 She’s also a contributor to the covid act now project a tool that helps policymakers evaluate the effects of strategies reduce the spread of covid-19 Rebecca. Welcome. Thank you for joining us today. Thanks for having me. 1:54 Great, so I just want to ask you. I mean, you’ve been spending your whole career essentially preparing for a pandemic learning about global Health security. Tell us a little bit about how you’re working now in the response to covid-19. 2:10 yeah, so I have to say that we spent the better part of the last 20 years working on issues related to Global Health security and pandemic preparedness, and it’s been me the entire core of our teaching research and service focused on how do we prevent detect and respond to Public Health emergencies, and I think it is fair to say that, you know, we didn’t have a A terribly large community, but pretty much everyone within the community. 3:08 The 19 happening and because of that we spent a lot of time thinking about it trying to build up an Evidence base and and trying to better prepare different corners of the world to be able to have sufficient capacity to be able to respond. So when the virus did emerge in in December, we all kind of went into action so starting with and it’s been a crazy. 3:38 Cross a pretty broad spectrum of activities thinking about at a very local level and it’s gone from from local to Global AS Global health does but also standing working in the details of operations and response to research to Consulting diving. We put together a working with colleagues at NCI and Center for Global development volatility analytic. 4:08 We launched a Frontline guide for for local officials in a website called covid local dot-org and that was based off of what we’ve been doing for years with with Municipal level leaders in particular the global parliament of Mayors. 4:26 So we’ve been trying to support local officials and then we’ve also been involved in in modeling and in thinking and we had actually started To do we you know, we were actually calling it a consumer reports of all the models that are out there and in the process of trying to put this together we were going very deep and and kind of playing with the math behind a lot of the different models. And in that process the came across the covid act now team and we saw a group that had not had great vision and reach and and reach out to them. 5:08 To see if we could help them advance so model. 5:10 So in addition to all the other things that we’ve been doing also now working with the covid a finality to to try to support the the model that they’ve built and the Outreach that they have for decision makers to help them kind of think about what they need to be doing and when Great. Thanks. So tell us a little bit more about covid act now. And for those who are maybe less familiar and then I want to ask you about the range of models that are out there to and what are some of the things that you as a data scientist, you know, really look for in those kinds of models, but what is covid act now and what is it was that you thought was unique about it? 5:56 So it’s you know, I keep saying that one of the one of the things that’s going to happen what we are already seeing with this pandemic is that every smart person in the world from every discipline is solely focused on on the pandemic, which means that we are we are seeing incredible Innovation and incredible thinking going into a space. That was yeah. 6:23 It’s a pretty small community before and with Added resources. So one of the things that excited about this group is it was it’s a it’s an all-volunteer organization primarily made of people from the tech industry, but also now statisticians and medical professionals epidemiologist and folks really coming from every corner of the country who were who were just committed to act and think about how to put their expertise into this space. So this is why they have come up with it there. 6:56 a new model is released yesterday that includes actually county-level data were available and the what the model showed and again we have been trying to to assist in in bringing in a little bit more traditional epidemiology and and kind of more formal disease modeling into this group of True Tech experts and incredible thinkers, but the That is what well the policy point is that the need to take action now and this has been a critical message for over a month and and now that most but not all states have issued shelter-in-place notices. The tools is also recognizes the importance of continued action and the need to plan accordingly and I have to say that there I think we are start there are a lot of models out there, but we’re also starting. 7:56 A lot of people and all of the models there’s some discrepancy in some of the in some of the factors that are used or how much weight to put on one thing over another but it’s all directionally the same every model out there points to the importance of of aggressive action early and doing so saves lives. 8:22 Thanks Rebecca. That’s very helpful. So so in terms of just looking at the modeling that the data is to affect, you know, some some of the decisions that are being made. I mean, obviously the actions that are have been taken in the last month have been unprecedented in American history. There are obviously still a lot of state and local leaders kind of grappling with what to do. Can you talk a little bit more? 8:51 Specifically about the terminology like what you know, you talked about shelter in place. We’ve heard stay at home. We’ve heard lockdown, you know, what are sort of at some of the underlying assumptions about the actual decisions what those entail and then how that effects what some of the projections might look like. 9:16 Yeah, there’s we can talk about like what is shelter-in-place mean versus you know, the the different the different assumptions that people are using what is the Wuhan style lockdown look like there’s there continues to be even in states are using the same language there continues to be a lot of variation. So I think one of the things that we all we do know. 9:45 Is that the the physical distancing measures close the spread of the virus and that this is critical for being able to buy time to build up the supply chains necessary to get the personal protective equipment to our Healthcare Providers to assess where there’s going to be times. 10:04 So that hospitals will become overloaded so they can do sufficient planning to ensure that there’s additional bed capacity that we Served on our medical workers that all the how we determine how the timing will be very much depends on how these physical distancing measures are actually implemented. So when you talk about like a Wuhan style containment that is that is where you treat everyone as infected, there’s Force community-wide quarantine. There’s full shutdown a business. 10:45 Is closed borders active monitoring full population mandatory testing very aggressive actions and in note that you know, what what is important thing to remember? 10:58 Excuse me, when we when we talk about Wuhan is to remember that that this just came at a tremendous cost to to Human Rights and to civil liberties and that again people were actually like forced and literally force and you know, Field into their apartment and that’s not necessarily so very effective for stopping spread of virus, but not necessarily the type of thing. We actually want to see in an American context. So instead we look at these issues with shelter in place, right? So when you say that these are these are Vaughn’s usually voluntary or almost like voluntold shelter, please community life on quarantine particularly for high risk groups. 11:41 It’s shutting down on essential business has its closing schools it standing Event with people outside of your family the passive monitoring its public. Advocacy. It’s hand hygiene possibly restricted travel possibly closed borders, but I have to say that as I talk to talk to colleagues and also talk to my students because I continue to teach online and my students are now spread all over the country and all over the world. 12:15 Is that how this is a force at within different communities is very varies. Greatly there. It is. Very uneven across the country. So you may walk into into into a grocery store in one part of the country and it is it’s almost a token right there. You’re handed gloves your total. 13:14 All what that spread might look like and what projected hospitalizations might look like. So it’s a little tricky which means that pretty much all all models are wrong some level right? We know that so instead this is how do we how we think about this? How do we how do we try to get grandma’s? 13:32 So the new model in the least is is at the county level or County data is available because there is such incredible variation and in the The Hope is that as we get more actual data as opposed to just you know, projected data that will be able to to to to make these these models bit more precise and a bit more informative for local and state and hopefully National level decision makers. Thanks Rebecca, and it’s interesting. 14:06 I mean what instruct me throughout this this whole this whole experience has been just the the intersection of the Keishon the data the science and really almost kind of the anthropology of all of this. 14:20 I mean even in my own neighborhood and you talk about getting really granular it’s like, you know, and I’m in Maryland, but you know people were pretty actively social distancing but I was still seeing kind of, you know kids playing basketball still seeing some of the you know, little kids using the playground equipment and then in the last few days, once the governor issued a even stronger warning or shot, you know order that kind of thoughts that just kind of the the communication and the spread of information and then what people choose to do with that has been fascinating to me. I just wonder if you can kind of comment on that that piece of it as well two things certainly are changing so quickly on the ground and the interpretation of it. Like how can the data science kind of help to inform people’s daily choices? 15:15 I think that the challenge is trying to capture those daily choices at the same time of trying to inform, you know, there’s an entire there’s entire discipline around risk communication crisis communication particularly and and in adopting that two different Health seeking behaviors. We know that from other areas of Health Systems research. 15:37 We see it in in outbreaks and but we have not been good about I think the globe Health Community is great about studying how to do this in other communities but has have not been great about how to translate it to an American context. And so that’s been it’s been a bit of a challenge to see that and and see it emerge over time and the messaging shift and I think it’s still even even once the the messaging gets stronger and more aggressive. 16:07 I also live in Maryland and I’ve seen some changes but even yesterday before still You know kids are still playing together and and folks are defining their their community and who they are interacting with in a I think a way that is much broader than I personally would have liked so so so it does it does between be a challenge and it’s one of the reasons why you can look at what a policy is, right? You can say, okay. 16:37 This is what the guidance is from this Governor, but but then actually trying to ground truth that is the challenge and it One of the and and even in this challenge, you know, you get back to the data science of it all a lot of the the modeling the thinking is trying to look at documented cases, but we the the testing remains so uneven that that’s a really hard. Those are those are really hard numbers to to examine and to think that they might be predictive which is one of the reasons we’ve been trying to shift much more to thinking about hospitalization, right? 17:14 Wait, and actually in unfortunately number of deaths as well because those are those are more tangible numbers than and then looking at you know, who’s who’s actually getting tested how where there’s a backlog and what parts of the country where there continues to be a really strict triage for for who can even access the test how fast you can get turn around and until we see more uniformity there and until we actually have more faith. 17:44 In the end the data that we are receiving as being representative population. It is going to continue to be a bit tricky. 17:54 That’s really interesting. So I want to ask you about a couple of the other because you said you had looked at a bunch of different models. And there’s there’s a model The Institute for international Health metrics and evaluation. I hme model of University of Washington, which is one of the models that the other model has been widely referenced and I’m just curious. You know, how how does that differ if at all or is there a different Focus? Can you speak to that? 18:21 Absolutely. So, you know as I mention there’s a bunch of models that are out there. The ihe model at the University of Washington is is certainly one of the more reference ones these days that team and I hve is has a tremendous amount of expertise. 18:38 It’s a large group and they’re the ones that produce the global burden of disease study as a year and when they when they start to build out their model, they we we’ve had the fortune of actually being able to get on the phone with them and then talk through a little bit of assumptions and that the I’ll see you the the it pains almost I would say actually probably a to Rosie of the picture because we know that the actions in in China are not the same as the ones that are here in member is this is basically holding up people’s front doors arresting people who try to leave their homes and basically the equivalent of a martial law and that’s not something we actually want to see here, but that’s that’s the type of measures that they built. 19:50 It into their model right now, but I will say that in speaking for their team their model is evolving and as they add data and Analysis from around the world. We actually we expect their model to to evolve and we also actually kind of see a lot of our models will start to converge in in ways that they haven’t to date. 20:15 Right, so I want to I want to get into a couple of other questions and kind of go back to something you mentioned which is to you know, you said kind of just some extent all the models are you know, no model is perfect, you know, but they’re all of the ones you’re saying or kind of directionally correct or in this in going the same general direction and you’re starting to see some convergence. I mean, so, you know to that end. Can you just kind of get into that piece of it a little bit more? 20:43 Or I mean again, like what are what policy decisions are those models pointing to? 20:52 Well, that’s probably the easiest question you’ve asked me and in this is because the it’s the the models are are supporting what we what we know and see in the rest of the public health and epidemiology Community. 21:06 Is that because we know that it saves lives and now I can point you to some of the the some of the analyses that have been done. 21:50 Because what we know is that if you if you put physical distancing measures in place, it slows the spread of the disease and it’s if it’s slowing the spread of disease. It means that we are we are saving hot and the what and and swelling is spread by time and what we need to buy time for right now and it’s the same language that people been talking about for weeks about this idea of flattening the curve but what that is really getting at is the fact that our Hospital Systems right now. 22:20 Now just do not have sufficient supplies Staffing or or beds to be able to manage an onslaught of cases. We have we have Frontline healthcare providers who don’t have Vision access to personal protective equipment. There aren’t enough beds. There aren’t enough emulators, but that these are these are problems that are solvable through Logistics, right? 22:48 So if you buy yourself enough time if You’re organized enough then you can you can get your supply chain going you can figure out how to how to manufacture or get sufficient PPE to all the Frontline healthcare workers who acquire it that you can try to address the challenges are being posed right now about but with the lack of ventilators that you can look and do the nails you can figure out how many beds do I have how many ICU beds do I have? Can I look at the projections of what’s about to happen and think about what are my alternatives? 23:20 So whether that’s what what type of surge capacity is required and we’re how how does one do that? So it’s the there’s a moves right now to you know, we’re already seeing kind of the you know, it’s as you and all your listeners know that any given day most of our hospital beds that you know, the number of folks who are currently occupying hospital beds and the requirements and the the other surgeries and all the other Healthcare problems that are happening in the country. 23:51 How do we ensure that that that population those populations are cared for efficiently because we’ve certainly seen another outbreaks around the world that it’s often the let’s say if we take West Africa ebola. It’s the the non Ebola deaths that outweighed that that outnumber the Ebola death because there was there was the inability to you know, do get a C-section or to be treated for a road a road accident. 24:17 So ensuring that we have all of those all of of the the our systems in place so we can identify. Okay, which hospitals are going to be required just for covid how and then and Warehouse where we going to like kind of reorganize for our ICU beds where we going to send other people who you know broke a leg and and what capacity do we have for that? And how is that? How do we mix the the and what what I think the logistics aside it’s an issue right now figuring out how we combine local knowledge because you know all public health is low. 24:50 All with state and federal assistance. So as the only Corps of Engineers goes around and figures out where should they be? How can they be a system? Where can they be building kind of and and expanding a bed capacity coordinating very closely with local officials who know their know their population know their resources? No, what no what may work and what may not work for further localities and then to also be pairing that with what do we think about surge capacity? 25:19 So You know how many Community have to research medical professionals whether that, you know, the early graduating and and advancement of Advanced Medical students and nursing students whether that is bringing back people out of retirement, which is often high risk groups. So you won’t have to be quite careful about that whether that is looking to volunteer cores. 25:50 Whether that is, you know scope looking at scope of practice. I mean, there’s a lot of different things that can be examined. 25:57 But all of that takes a little bit of time and preparedness and and that’s this is what the physical distance thing that hers is advised us that time to get those those prophecies those systems and and those physical spaces ready so that when additional cases were not we’re not having to to resort to the same type of Crisis standards of care and horrific decisions that we’ve seen overwhelm Healthcare Systems in other parts of the world. 26:30 Great. So so when when you talk about the variability around the country and and and just you know, we’re a big country. There’s a lot of unevenness of all sorts of things including a lot of variables you mentioned and then sort of differences around the country in terms of the response so far. What does that tell you about both, you know going into you know a surge. 26:54 Like how do we deal with that variability and then, you know a lot of Are now asking already turned to ask well, how long is this going to last? I mean when can we safely come out of this physical distancing? When can we safely start going back to work and obviously for good reason how you know, how can the data in the modeling kind of help us to think through some of those questions as it relates to, you know, again, we’re free country with open borders between states. I mean, how do we deal with that? 27:30 Well, I mean what you’re asking is the question. I think the entire Community is focused on at the moment and trying to figure out how how we how we scoped this. What is the what is the long term plan look like and what we are calling in the reverse triggers, right? 27:46 What are the triggers for for removing some of the physical distance things require restrictions and and there is a tremendous amount of Of modeling and thinking about different ways that that happens. There’s there’s stuff some people calling it like the damn some of the lights which model like how do you turn on turn off? What are the triggers that you’re going to be using but there’s still everybody wants the crystal ball answer right? What day do we go back to normal? 28:18 And I don’t nobody can give that at this point because we still have to see how how this plays out making sure that all of the other factors are Are put in place for being able to handle the cases as they go forward. We still don’t we’re still looking at different types of medical countermeasures. So, you know, the the projections change dramatically if in two months or three months, there is we have sufficient clinical data to say that X treatment will actually prevent death right? I mean that that would be a game-changer that if we can better understand. 28:59 How much natural immunity is concerned after you have the disease and how long that immunity might last and then if we if we know that and we pair it with with broad serologic testing and we can identify who has antibodies and what kind of natural immunity there might be. Then we can start thinking about putting those folks on the front line or getting them back to work. 29:22 So there’s a lot of a lot of different factors that are to the models and all these things are being considered as we think about how I don’t I’m not going to call it normal because I think it’s going to be what a new normal looks like there’s going to be a period of time that’s not going to look like what we expect but or what we are used to but will be almost like a transition period as we as we move out of this stage again, I you know, I’m very nervous about telling you, you know on June 11th X is going to happen because there’s just too many doctors but But I think there’s a lot of people staring at all those factors right now and trying to figure out you know, can we tell me put parameters on them? Can we stay, you know, if your if your hospitalization rate Falls below 50% or you know trying to actually play with what those parameters might look like to then help inform decision-making. Yeah. 30:19 And is it fair to say that the faster we all act now the the the shorter the period of pain will be I mean is that What the data tells us or is that is that not that’s a good I know that’s exactly that’s exactly what what we’re being told is that if you act the longer you wait, so the more difficult it will be to to kind of do a reset and to get into a place where we can start thinking about reopening up that we certain parts of society, which is again, it’s important with acting now and this is the one thing that all the models are saying that you have to take aggressive. 30:59 agent immediately Well Rebecca, thank you for joining us. It’s been a really great conversation and I’m sure we could talk for much longer than this but we are out of time and want to respect the rest of your days. I’m sure you are going to be busy all day with with these kinds of questions and looking at the numbers Etc. So I just want to thank you for joining us at the Alliance for Health policy and for for all of your work. 31:30 Work to help keep us safe. 31:34 Well, thank you for having me and also deep appreciation to your community and for everything that the the entire Community is doing to fight this pandemic and and strengthen our system. 31:48 Thanks. Take care.