Interdisciplinary Quantitative Research on COVID-19

May 22, 2020

Information

Saad Omer, MBBS, MPH, PhD Director, Yale Institute for Global Health

ID5231

To CiteDCA Citation Guide

00:00I would now like to introduce our
00:03next Speaker, Doctor Saad. Omer.
00:05Doctor Omer has conducted studies
00:07in the United States, Guatemala,
00:10Kenya, Uganda at the opia,
00:12India, Pakistan, Bangladesh,
00:13South Africa and Australia.
00:15Doctor Omers research profile
00:16includes Epidemiology,
00:17a respiratory viruses such as influenza,
00:20RSV, and more recently COVID-19.
00:22He is a director of the Yale Institute
00:25for global health and associate Dean
00:28of Global Health Research Doctor Omer.
00:30Thank you for being here.
00:33It's my pleasure to present
00:35and I will share my slides.
00:42I'm assuming everyone can see my slides.
00:45Correct Fatma Yes Yeah.
00:49So it's my pleasure to present
00:52some of our work in on this
00:55topic on the topic of COVID-19.
00:57Broadly, I will focus on our group in
01:00both Yell Institute for Global Health
01:03and my more immediate research group
01:06has been focusing on a bunch of things,
01:09but this presentation will focus
01:11on somewhat interdisciplinary,
01:12quantitative research on code 19.
01:15But before I start,
01:17I just want to step back and talk about,
01:20you know, when there is an
01:23outbreak when there is a so called
01:25once in a century of event.
01:28Educational institutions and
01:29research institutions should
01:30think about their role actively.
01:32We're not Jackson Labs.
01:34We're not scripts who have their
01:36own position in the ecosystem
01:38of discovery and deployment.
01:40We are not the CDC.
01:42We're not a health Department.
01:44We are an educational institution
01:47and educational institutions have
01:49to be responsive both in terms
01:51of their research and some of
01:54the other activities to emerging.
01:56Issue so one way of looking at what
01:59educational entities do is the Inter.
02:01Logically to say that the appropriate
02:03Ness of our actions and our responses
02:05to public health emergencies are
02:07judge irrespective of their impact.
02:09I don't think that's that's
02:11an important thing.
02:12Part of a University,
02:13so you know we should ask why someone
02:16is looking at zebrafish biology?
02:18Because it serves as well that if you
02:20don't ask those questions earlier on,
02:23we benefit from that kind of free
02:25thinking and. Go taking where going.
02:28Whether our curiosity takes us,
02:30but in a pandemic we have to
02:34be decidedly consequentialist.
02:35We have to say Watt,
02:37the time we are spending the things
02:40we're doing and the research
02:43questions we're going after.
02:45How does that contribute to
02:47our response to this?
02:49This event I would say of
02:51unimaginable proportions.
02:52It is not unimaginable. So a lot of people.
02:55Part of it talked about.
02:57It worked on it.
02:59So I will go through a series of questions,
03:03so I'll take the so called Socratic
03:05approach to discussing what
03:07we have been doing.
03:08So the first question is what was
03:10the whole country in the world shut
03:13down and the immediate question
03:15was even in March when this was
03:17the started in early March in most
03:20a lot of places,
03:21or actually mid March 2nd week of March,
03:24is when there was imposition
03:26of these lockdowns or social
03:28distancing measures and then bye.
03:30Date March the question was was this?
03:32What are these policies working?
03:34So we did two projects around there.
03:37The first of all we contacted the CEO
03:40of one of the apps that is out there
03:43and the reason why we went after this app.
03:47It's called the mobility index.
03:49It's called the city mapper app because
03:51it combines both the IT combines
03:53the the the trips taken through
03:55public transportation and private
03:57transportation and trip planning etc.
04:00So they have.
04:01Calibrated this mobility
04:02index of 41243 cities.
04:03Depending on the type of
04:05data that you're looking at,
04:07and then we got those data.
04:09The other one was look at cell phone
04:12data that tells us with some minute
04:14detail where people are spending
04:16time and there are more advanced
04:19users of data that we're working on.
04:21So the second part of the work,
04:24the cellphone based data, came out
04:26of collaborations with Eli furniture.
04:28I'll pause here for a second, so
04:30the approach the consequentialist approach.
04:32Tells us that you don't start
04:34with the discipline.
04:35You start with a question and work backwards,
04:38and so I hope these snapshots and
04:40examples will convey that if you
04:43start with a question you don't
04:44stay within your own research group,
04:47certainly,
04:47but also not in a single school or
04:50two schools or even an institution.
04:52Broadly speaking,
04:53you work with people who have the
04:55interest and something to offer
04:57and combine forces in producing
04:59some of these outcomes.
05:00So the first thing was
05:02actually more immediate.
05:03So this was blood by amine Malik.
05:05One of my post docs and we did a very
05:07quick analysis and we did statistical
05:09analysis as well as sort of.
05:11This is graphic representation
05:13to be went through a few models
05:15and we found that in April.
05:16So this is by cities and if you look
05:19at the trends that red dots are the
05:21ones where there wasn't an imposition.
05:24Of these orders,
05:25so the initial very early inside.
05:27So this was on March 28th that we did
05:30the first analysis and we found that
05:33there was actual anticipatory this
05:35decrease in mobility approximately
05:37by three point 4% based on the
05:40depending on the baseline compared
05:42to the baseline for a day.
05:45Even before these orders were
05:47implemented and it was almost
05:48universal across continents etc.
05:50With these data were available,
05:52however, there was an effect.
05:55Overall effect of mobility decrease
05:57after using this measure after social
06:01distancing policies were implemented.
06:04Then with this cell phone based
06:06data that I talked about,
06:08we were able to look at time spent at
06:11home in the US and we try it stress
06:14testing these data in terms of these
06:17trends which were apparent earlier on.
06:20But just what the interesting part was,
06:23whether we looked at New York or
06:26Poker Tellow, Idaho or Arkansas,
06:28or Atlanta or Vermont,
06:30this these are the counties we had
06:33some pretty interesting trends,
06:34so they worked at first of all the
06:37change in behavior is the increase
06:40in time spent in home preceded some
06:43of these policy actions and this
06:45dark line anchors this work on the
06:48policy action and the imposition.
06:51Uh,
06:51of these lockdowns and social
06:53distancing measures.
06:54But there was also an effect and
06:57impact of the number of cases
07:00in the community,
07:02but there was differential
07:04that was the difference between
07:06Metropolitan areas in rural areas.
07:08The rural areas where going for
07:11the local cases and Metropolitan
07:13areas where tracking the added
07:16effect was tracking and
07:18there was a bit of a crowding out effect.
07:21That we're tracking national cases.
07:25And so that's an interesting observation.
07:27And that's rational,
07:28and that that works through.
07:30The phenomenon of availability heuristic
07:33in psychology that we just the probability
07:36of events by how vividly they are
07:39covered and so availability heuristic
07:41was working and it's rational to say
07:44that Metropolitan areas were actually
07:46linked more than non Metropolitan area.
07:49So they were tracking national cases whereas.
07:52Rural areas and on Metropolitan areas
07:55were tracking local cases and that has
07:57implications on when we open up and long
08:00term strategies for social distancing.
08:03The second thing was that a lot of us
08:06working with populations and governments
08:08and policymakers outside the US as well
08:12as inside the US an the testing in the US
08:15has gone up substantially and will go up.
08:19This testing is load University
08:21but in low income countries.
08:23That those limitations are even more stark,
08:25and therefore the testing capacity
08:27is a little bit more long term,
08:29so you know you can always use a
08:31little bit of interdisciplinary
08:32work where you could look at,
08:34you know what what is happening and
08:36what are some of the approaches.
08:38But before I get to that,
08:40so you know, as I've been saying,
08:42there's a cottage industry of plans
08:44to open up America in the world.
08:46Ours was shared initially in late March,
08:48it came out as part of a peace in JAMA where,
08:52you know, we focused on.
08:53Extensive testing and contact tracing.
08:56Reduced reduction of household transmission.
08:58An additional treatment options will
09:00be said is going to help US Open up
09:04America safely, but now you know,
09:06you know this is.
09:08You know several program plans have since
09:11come out for the Rockefeller plan, the.
09:14Eh, I plan.
09:16This one is published, I think April 4th.
09:20So and so forth.
09:22So there are multiple plans
09:23that are out there.
09:24One thing that is shared is
09:26testing and contact tracing.
09:27And so how do we optimize it in one way?
09:30And I'm sure at Kaplan was done brilliant
09:33work on this in the more sort of.
09:35Local and domestic settings will
09:37talk about that in more detail.
09:39One strategy that we figured
09:41out would be too for government,
09:44specially in low income countries,
09:46is pool testing and so this was
09:48a little bit of massive working
09:51with an economist at the awesome
09:54charger is the director of.
09:56Center for International Development
09:57at Harvard and for Han is from magic
10:01is a economist based out of rice and
10:03the three of us combined forces and
10:06looked at said that there is lab based
10:09information that says that you cannot
10:11pull up to beyond a certain level,
10:13which is entirely appropriate.
10:15But can we do a bit of math to figure out
10:18who to test in low resource settings?
10:21So this is one example of that,
10:24and we found that even at.
10:26The at lower prevalence levels
10:28we can find a lot
10:30of efficiencies by pool testing.
10:32Even a full size is as low as 10,
10:36and as as soon as you increase them hire.
10:39That would be of pretty
10:41high level of efficiency.
10:43We also prioritize is based on
10:45some network modeling approach,
10:47which I'm not showing here in the interest
10:50of time that for low resource settings,
10:53looking at figuring out who has
10:55a who's a high degree node.
10:57In a network and prioritizing them actually
11:00gets you a lot of bang for your Buck,
11:03especially when the boxer shorts.
11:06And then what do you so?
11:09A lot of these interventions are as much
11:12about human behavior as much as they
11:14are about the behavior of the virus itself.
11:17So very earlier on, we focused on
11:20what does the US population think.
11:22So we did a population representative survey.
11:25Did an online survey online
11:27servers have gotten better.
11:29We paid a lot of attention
11:31of representative samples.
11:32So there. We not be.
11:35It won't be able to stratify them
11:36by a lot of characteristics,
11:39but overall it was fairly representative.
11:41We did robustness checks and
11:42this was done on February 8th,
11:452020,
11:45so before a lot of the features
11:47before doctor found,
11:48she became a household name and
11:50that would be relevant in the next
11:53slide we asked a bunch of questions,
11:55but I'm showing a bunch of.
11:57Pieces of data there on February 8th,
12:00they told us when we asked them
12:01who should be the the in charge of
12:04America is COVID-19 outbreak response
12:06and people said the CDC director,
12:08the director of NIH and then followed
12:10by the president and the Congress was,
12:12I think 1.5% what they were saying.
12:15They did not know the name of the
12:17direction of the NIH or the director of CDC,
12:20but where they were telling us
12:22that it should be a scientist.
12:24LED response.
12:25The other thing we asked them,
12:29whose advice would you?
12:31Trust the most and we found that it
12:33was that the top Dewar Healthcare
12:36Professional which is common to all
12:38sorts of other phenomenon phenomenon.
12:40A lot of our study, for example vaccines,
12:43but CDC ranks fairly low in terms
12:45of vaccine information source,
12:46but CDC and NIH were right up there and
12:49we have repeated these surveys, etc.
12:51We are analyzing it for this as well,
12:54but going back to some of the
12:56other questions.
12:57So how can we control the COVID-19 outbreak?
12:59Sorry for the typo in confined spaces and
13:02the reason why we are thinking about that.
13:04Is a lot of these micro level and
13:08a lot of the policy decisions as we
13:12open up our micro level decisions,
13:16companies,
13:16dorms,
13:17residential colleges in a certain
13:19300 over University for example,
13:22and cruise ships.
13:23So the first thing we started
13:26attacking with a postdoc amine
13:29and Sam Jenness who is a close
13:32collaborator based at Emory.
13:34Collaborating on some of the other
13:36stuff we said look the best model.
13:39The best source of data we have right now is
13:42from Diamond Princess Anne.
13:44If we could because there was a lot of
13:47reporting over reporting because there's
13:49a lot of other cases were not that high.
13:53We know a lot of details,
13:55so we sort of through with with
13:57another graduate students, etc.
13:59Katy Villa brand.
14:00We really did God really detailed data and
14:03we figured out that there were two networks.
14:06Uh, that were a small world networks
14:09that were existing side by side,
14:12the staff and the passengers and we said OK,
14:16but they are interacting so
14:18they have their own dynamics,
14:21their own degree of mixing etc and
14:24they are interacting with each other.
14:27And again from the beginning.
14:29The idea was that we work it out for
14:32Diamond Princess and then that serves as
14:35a template for other confined populations.
14:39So these are initial results.
14:41So we did we doing model,
14:43collaboration etc.
14:44And so with the initial thing
14:46is the first thing we evaluated
14:49and these are work in progress.
14:52That network, log downtime etc is.
14:54The earlier use stop the interaction
14:57between the two networks,
14:58the better control that you
15:01have on the outbreak.
15:03The other thing is,
15:04you know we are in a situation
15:06where going forward we're going
15:08to have a lot of decisions we made
15:12will be made on imperfect data and
15:14a lot of decisions will be made
15:17based on the idea that we have to
15:19have a grasp on reality for both
15:22symptomatic San asymptomatic.
15:24So how can we innovate in the
15:26context of imperfect data?
15:28So there are two lines of
15:30inquiry that we started.
15:32They are existing surveillance systems too.
15:34All to our surprise that CDC frankly wasn't
15:37utilizing it too to its full extent.
15:39When is influenza like in Leicester
15:41pin system and we went for flu
15:43negative influence?
15:44A negative I'll I there has been
15:46a change in healthcare behavior
15:47in terms of seeking care for
15:50respiratory illness,
15:51which is can be accounted for
15:52and we depend through that.
15:54And the other thing is sewage.
15:56Essentially sludge based surveillance,
15:57and for that again work.
15:59Starting with the problem and
16:01working backwards.
16:01So this this sort of these
16:03ideas came out with.
16:05Discussions with at Kaplan and then
16:08subsequently we reached out to Jordan
16:11patio what it came out of was that a lot
16:14of us were observing that very early around.
16:18There was evidence that at least viral
16:21name was being excreted in feces.
16:23Then the question was if it's
16:26being excreted in the feces,
16:28is it detectable in sewage or sludge,
16:31etc?
16:32Because there other attractive
16:33part interactive part was.
16:35That the world does this
16:37kind of surveillance,
16:38including and low income countries for polio.
16:41So there is this muscle memory
16:43and the ability to do this
16:45in this kind of a situation.
16:48So the first thing is that
16:50these data for flu influenza.
16:52I'll I influence,
16:53sorry flu negative influenza like illness.
16:55So we created standardized
16:57these scores and we found that
16:59this can be a leading indicator even
17:02adjusting for imperfections and testing
17:04Ware and sort of honing in places
17:06where we had some confidence. Uh.
17:10Can be a leading indicator for monitoring.
17:15Kobe 19 deaths and incidents.
17:16These are the data this should come
17:18out in a bio archive fairly quickly,
17:20and what we found a few things
17:22in from the sewage and such data.
17:24And then there was a.
17:26I would be remiss if I don't recognize the,
17:30uh, the dedication of not just Jordan
17:32Pescia from the school of engineering,
17:34but the medical students who
17:36collected specimens on a daily basis,
17:39and they're pretty trying circumstances.
17:40Obviously each is approved,
17:42etc from sewage medical students
17:44and graduate students as well
17:46as native grew by in his lab.
17:48Being a collaborator,
17:49the AG station folks being apart of it,
17:52this is a true collaborative project
17:54and what we found was that first of all.
17:58That the we can identify this RNA
18:02from sewage sludge that tracks well
18:06with the size of the with the.
18:11The trends in the outbreak and
18:13can serve as a leading indicator
18:15for cases by 5 to 7 days.
18:17There is other preliminary work
18:19that says that you can quantify
18:21the size of the outbreak,
18:23but I am a little bit more sceptical
18:26of that part right now because we need
18:28to calibrate that a little bit more.
18:31But as those indicators,
18:33so this has resulted already in the city,
18:35contacted us and Jordan will be providing
18:38weekly information to the city of New Haven.
18:41But also I've been working with
18:43the World Bank to expand this.
18:46Uh, and piloted in at least two countries.
18:50Countries in large Metropolitan areas,
18:51as well as smaller areas.
18:53So then what is the acceptance of vaccine?
18:56We're talking about vaccines,
18:57but vaccines if not taken,
18:59are pointless,
19:00and so we looked at that and we
19:02predicted that there is a lot of
19:05variability and we had sort of a
19:07predictive model based on demographics.
19:09So we calibrated this and said,
19:12can we create a risk map based
19:14on directly asked questions?
19:15But calibrating it again based on
19:18demographics to identify higher
19:19risk areas and that's.
19:20What some of this work?
19:22This is fairly recent,
19:23like a couple of days old,
19:24and this will be coming out fairly soon.
19:27The other thing is we don't want
19:29to be in a situation in vaccine
19:31distribution where governor's are
19:33fighting over vaccine like they
19:35did on some occasions on PE.
19:37So what we did was I worked with
19:39one of my keyboard mantis,
19:41who happens to be at the is a geography
19:44Republican geographer at UNC.
19:45Although matter and we looked at the
19:48what are the risk groups that we know?
19:50What are some of the other population
19:53characters characteristics we know
19:55and we came up with this map for.
19:57Distributing in placing vaccines
19:58and we have 5 version of it.
20:00Based on the policy goals.
20:02So if you have mortality reduction
20:04is your policy goals,
20:05this is very you situation,
20:07distribution centers etc and distribution
20:09based on the underlying and if you have
20:12it for transmission and so and so forth.
20:14The other thing is how do people
20:16mix during during an outside?
20:18Pandemics has been a question,
20:20so I've had a narrow one and a CDC grant
20:22funded on that area before the pandemic,
20:25but we repurposed it and so there are
20:27three 4 size the first original one with
20:30through STD CDC grant is workplace,
20:32the second is home,
20:34particularly low income countries,
20:35and then we have a supplement which is
20:37likely to get funded for health care
20:40workers that we're tracking them using
20:42RF ID technology, and the idea is.
20:44Uh, do, and with consent this has been
20:47there and we have an anthropologist in
20:49the team who has done the heavy lifting
20:52on acceptability and all sorts of stuff.
20:55That is that people are guessing
20:57now about the acceptability of other
21:00kinds of contact tracing and these
21:02kinds of these are the Maps that
21:04that we will be creating.
21:06This is from a previous study of Contacts
21:09between different kinds of people,
21:11not just 8 bass mixing.
21:13How can be facilitated vaccine development?
21:16There's all sorts of studies that
21:18are going on leveraging our health
21:20worker cohort that several people came
21:22together and establish to evaluate
21:25cordless of protection and severity.
21:27Advanced phase getting ready for
21:29advanced phase clinical trials
21:30and focusing on vaccine access.
21:32The last thing is that University
21:34of a unique responsibility to make
21:37sure lessons learned during this
21:39pandemic are remembered for centuries.
21:41So the first thing is public scholarship.
21:44This was. This is an op Ed.
21:46I wrote in the New York Times that came
21:49out right after the when the count was,
21:51I think one or two in the US and the title
21:54was is America ready for another outbreak?
21:57And the answer was no,
21:59but you can do a few things and I
22:01warned about not consequences of
22:03scientists not taking the lead.
22:05No false assurances,
22:06and it was very clear that if
22:08you give false assurances,
22:09there's a trust diversion it trust
22:11and then compliance will be low.
22:13And then scientific and public
22:15misinformation.
22:15Was the thread that a lot of us saw coming.
22:18And as you know, in the public discourse,
22:21a lot of that is happening yesterday.
22:23I think what the day before yesterday.
22:25There's another op Ed,
22:27that that was published,
22:28and I wrote highlighting.
22:29Look what is going to happen
22:31day after tomorrow.
22:32This is not the last big outbreak
22:34on unfortunately not even the
22:36last pandemic that will happen
22:37in this country in this world.
22:39And so the idea is to focus on
22:41CDC reform based on lessons now.
22:43And and and because CDC remains our
22:46best defense against Japan damage,
22:48but as an educational institution,
22:50it's our responsibility to make
22:52sure that we pass on this knowledge
22:55to our students an minties.
22:57So being Brown restart to a few of us to
23:01say that, can we create
23:03options for students elective,
23:05so we paired with myself,
23:07Doctor Sheila she noise from adults,
23:09ID and three students.
23:11Student volunteers came together
23:12with with having a course that.
23:15Worked on the higher terms of in
23:17the blooms taxonomy so that their
23:19education is not compromised in
23:21the middle of a pandemic and we
23:23were very encouraged in our initial
23:25evaluation approximately 86% says
23:27that the course achieved its stated
23:29aims and similar 87% of students
23:31said that it would recommend this
23:33course through the classmates
23:34because we obviously measured
23:36and then you know the efficacy.
23:38But the best way to pay back to
23:41this privilege of being part
23:42of this outbreak is to pay it.
23:45Forward and and I've had the privilege
23:47and a lot of us have had the privilege
23:49to invest in the next generation
23:51of interdisciplinary scientist.
23:53So they range from left to right
23:54of faculty member who's about to
23:56join my existing entities about to
23:58join Yale as an associate professor
24:00in Zaidi to medical students,
24:02post graduate students,
24:03and all of them,
24:04and so these are the ones only the
24:06select ones that have directly
24:08contributed to the work I have shown.
24:10And that's the most incredibly gratifying
24:12part of a lot of this work, so I'll.
24:16I said something.
24:18In a seminar that you know
24:19found that on it was surprising,
24:22unfortunately highlighted on the
24:23years full of Medicine website,
24:25the root page and and I'll repeat it,
24:28universities have a privileged
24:29position in civilization.
24:30We're guardians of knowledge
24:32for future generations.
24:33We are what we are learning now
24:35through science, through experience.
24:37Through policy we will pass on to our
24:39students are manatees and future scientists,
24:42so I'll end with the expression.
24:44No pressure, no pressure.
24:45My academic friends.
24:46This is a responsibility.
24:51Thank you very much.