“Reverse-carpetbagging,” disease ecology and COVID-19
When others transformed into SARS-CoV-2 experts overnight, we did the opposite: utilized data from the pandemic to explore broad questions about how indirect transmission influences the shape of outbreaks
Ogbunugafor, C.B., Miller-Dickson, M.D., Meszaros, V.A. Gomez, L.M., Murillo, A.L., and Scarpino, S.V. Variation in microparasite free-living survival and indirect transmission can modulate the intensity of emerging outbreaks. Sci Rep 10, 20786 (2020). Available here (open-access)
On: “Reverse Carpetbagging”
Early in the COVID-19 pandemic, the debates about what constitutes expertise moved swiftly to DEFCON 1. The nature of the emergency? Too many voices, too many opinions, so many signals that we didn’t know who to believe, who to follow, or listen to.
The cynics among us (me, for example) talked about the problem of COVID-19 “carpetbagging.”
The term “carpetbagging” is used most often in politics, describing someone who seeks election in a geographical area where they have no connection. We can extend it to any setting where outsiders enter and pretend as if they are a “local,” know what they are talking about, are qualified to speak on local matters.
The carpetbagging conversation in science relates to debates about expertise, who should and shouldn’t speak on a given topic, etc. This is a messy topic, and beyond the scope of anything that I care to speak about now (more than I already have, that is).
The reason that I bring this up?
Well, early in the COVID-19 pandemic, I was afraid to comment on the COVID-19 pandemic (as in, offer original insight), both because of fear of being a carpetbagger, and because of good-old-fashioned imposter syndrome (What could I really offer? Me? No way).
And so my early work on SARS-CoV-2 lived in science communication and outreach, trying to explain some basic properties of the virus to the public (these efforts led by Prof. Pleuni Pennings, in collaboration with Senay Yitbarek). Scientific communication isn’t any less smart than original science, and if my contributions had ended there, I would have been satisfied (I mean that).
But as the pandemic progressed through the spring, I started to feel an itch.
Specifically, questions arose and persisted about the ways that SARS-CoV-2 was being transmitted from person to person…or from object to person.
SARS-CoV-2 and surface transmission
The notion of “fomite” mediated transmission, where a pathogen can be transmitted when a host interacts with a physical surface or object, gained traction as a real plausible way that SARS-CoV-2 was being transmitted. The potential reality of this had real-world consequences: if indirect transmission through objects and surfaces was a major route, then it could have profound implications for how we controlled the spread of SARS-CoV-2.
For example: should we be deep cleaning our office spaces? Spraying down our packages?
The data that had fueled some of this speculation came from a published study in the New England Journal of Medicine in April, 2020.
The study compared the survival of SARS-CoV-2 and SARS-CoV-1 on different physical surfaces, like copper (~4 hour), cardboard (~24 hours), and steel, and plastic (on the order of days). It was a very cool study.
After I read it, I was like:
“Aight, I got an idea. Lemme get to scribbling.”
This is because I have, for many years, been interested in questions of how long viruses survive outside of hosts, and their potential for indirect transmission through environmental surfaces or reservoirs.
I have done experimental work on the topic, and most recently, explored it using computational and mathematical approaches.
In 2019, My group came up with a mathematical framework known as the W.A.I.T. (waterborne, abiotic, and other indirect transmitted) model for studying diseases that are transmitted using a route that is not only person-to-person, but also indirectly from environmental reservoir to person. The “WAIT” nickname is an appeal to the notion that pathogens that are spread through surfaces must “sit and wait” in the world, and infect hosts as they become available (Early in 2019, I wrote a little blog about it).
With this framework in mind, I considered the new data on SARS-CoV-2 survival on surfaces, and asked a few questions:
“If SARS-CoV-2 can be transmitted via environmental reservoir/surfaces, and we have data on how long it survives on different surfaces, then why don’t we use the W.A.I.T. framework to ask how the physical composition of the world would influence the spread of disease?”
“That is — what if touch surfaces in the world were composed of substances of a single type….as in, all copper (“copper world”) or all plastic (“plastic world”)…how would that influence an epidemic?”
I sketched some notes and a basic model.
I first ran the question and model past Samuel V. Scarpino, one of the best mathematical epidemiologists in the world, who also happens to be a friend and collaborator. He loved the question, gave excellent feedback, and signed on as a collaborator. But we needed more help to survey available data, and learn about the natural history of SARS-CoV-2.
I presented the model to my research group (the OGPlexus-GEEQS Lab). This was during the spring of 2020, and we were not only social distancing, we were all between institutions (a few of us were moving to a new university). And so this was our first scientific excursion in a fully digital world: no in-person meetings, no pizza discussions.
We had to do all of the science of this project (100% of it) virtually.
We met, we argued (over Slack and video conference), discussed, laughed, argued some more. Always towards building and learning. And even when we fought about equations or logic, we laughed it off, and supported each other.
And everyone crushed it.
Me and Samuel Scarpino played quarterback/coach. We drew up the plays, got the ball into the hands our “playmakers.”
Victor Meszaros and Miles Miller-Dickson, both research assistants and physicists-by-training, have been a tremendous two-headed mathematical monster in the group for years. Their mastery is almost old news at this point (each has published several magnificent papers), but should not be taken for granted: they did the bulk of the hacking and slashing and coding. But elegantly and artfully.
Professor Anarina Murillo is an outstanding mathematical modeler and statistician. I’ve wanted to work with her since I met her last year, and this project ended up the perfect opportunity to collaborate. She is the rare applied mathematician who is great at both dynamical systems modeling and biostatistics. She is skilled at fitting mathematical models to real-world data sets, which was an very important skill set in this project (and mostly foreign to me: I’ve been mostly a theoretical epidemiologist, where I construct models based on real-world mechanics, but have never really attempted to explain any particular outbreak with a model. Dr. Murillo taught me how this works).
And Lourdes Gomez is the individual whose efforts might have impressed me most: she was a first year graduate student, was neither trained in infectious disease nor in mathematical modeling. But she attacked this topic with no fear. I told her that, should she choose to participate, that the expectations would be very high (I am always honest about these things). She “looked me in the eye” (virtually), and gave me her version of “put me in the lineup, coach, I got this.” Gomez absolutely devoured literature, taught herself methods and analysis tools, and was critical to our success. So thorough was her involvement that she eventually co-lead a companion study (with Victor Meszaros) that was also published. You are reading that correctly: by the fall of her second year in the PhD program, Gomez already has two peer-reviewed publications on SARS-CoV-2, disease ecology and epidemiological modeling. These are fields that she learned in our lab, over the course of 6 months or so.
Together we built a mathematical model and designed a set of analyses to explore questions about SARS-CoV-2 transmission in settings composed of different high-touch physical surfaces. Our model was based on the W.A.I.T. framework, and on our latest understanding of SARS-CoV-2 transmission dynamics from the literature.
We asked the question of how free-living survival on different surfaces would influence the COVID-19 epidemic. We created theoretical epidemics based on empirical data, and hypothesized that differential viral free-living survival on physical surfaces would have real-world consequences for how severe a given epidemic is.
Our work used empirical data on SARS-CoV-2 survival and other aspects of the virus natural history (e.g. incubation period, etc) and demonstrated that an outbreak in a “Plastic World” would be much worse than an outbreak in a “Copper World.”
The implications are rather striking — the composition of physical surfaces can profoundly influence the way an epidemic looks. For example, even though the “Copper World” and “Plastic World” hypothetical epidemics are caused by the same virus, they look like essentially different epidemics: just check their R0 values — around 2.4 for “Copper World,” and well above 3 for “Plastic World.”
This approach was, at its core, a theoretical exercise: we were asking the question of how indirect infection happens when high-touch surfaces are composed of different materials. But given that we were using data from SARS-CoV-2, we have to ask: is surface transmission really a thing? As in, do our results translate to the actual COVID-19 pandemic?
The answer is “probably not,” but a complicated “probably not.” As in, surface transmission probably doesn’t play out exactly as it does in our model, where it can be transmitted from surface to person with relative ease.
There remains a gap in our understanding: we know that virus is viable on surfaces for some time, but we still aren’t sure if these viruses are truly infectious. That is, we aren’t sure how easy it is for a virus to move from these surfaces into the body orifices and cavities (e.g. mouth, eyes, etc) in a manner that facilitates a successful transmission event.
Here’s the kicker, though: our mathematical model — containing indirect transmission through the environment — seems to fit real-world data on SARS-CoV-2 better than a model without this route.
What does this mean?
Well, it probably does not mean that our model captures how the world works exactly. One thought that we are considering, that is the object of current study, is that our model might fit real world data better because our indirect transmission feature might capture a lot of less-understood, complex phenomenon that simpler models miss.
So while indirect transmission might not the major route in the real transmission, we might have stumbled onto a model that better captures the complexity of the SARS-CoV-2 transmission in the world. This is one of the understated potential strengths of mathematical models: modeling is just as prone to “eureka” and chance successes as experimental work. If the model is smart and well-engineered, it can tell you things about a world — even if it isn’t the one that you originally set out to understand.
Skepticism of indirect transmission in COVID-19
We had published a preprint on this work fairly in the pandemic (May, 2020, barely a month after the New England Journal of Medicine paper).
When we sent the paper to journals, we were met with some interesting responses from editors: “This looks really cool….but we can’t review this.”
I get it. It seemed pretty clear that journals wanted no parts of the potential controversy involved with our study. Their argument seemed to be that we (the field, not only us) didn’t know enough about surface transmission, were reluctant to publish a result that might be cause a controversy.
I empathize. 2020 was such a wild year for science that I can understand the reluctance to publish some work that might be controversial.
But the “we’re scared to review” take might even be cowardly: getting things wrong is a part of science. And the purpose of our work was to explore the concept theoretically, using empirical data and mathematical methods. We made no concrete public health claims about COVID-19 in any particular setting. (for example, we did not suggest that copper surfaces were a potential solution to the problem of disease transmission)
Since we posted our preprint, the overall consensus seems to be converging on the idea that surface transmission (“fomites”) is not a serious route of disease transmission.
One article, published in The Lancet: Infectious Diseases, stated it plainly:
“In my opinion, the chance of transmission through inanimate surfaces is very small, and only in instances where an infected person coughs or sneezes on the surface, and someone else touches that surface soon after the cough or sneeze (within 1–2 h). I do not disagree with erring on the side of caution, but this can go to extremes not justified by the data.”
Still, some debate remains about the likelihood of indirect transmission via surfaces and fomites. There have been instances when people have speculated on SARS-CoV-2 traveling on frozen packages. But these cases are few and far between, and there is still no real “smoking gun” for a widespread outbreak event where surface/fomite transmission has been the suspected culprit.
The outcome of these public health debates are of no direct consequence for our work, because we never made any specific claim about how SARS-CoV-2 was playing out in any particular place. Again: we only used the data to explore the broader question about surface composition and epidemic severity.
But like we discussed earlier: our model might be meaningful for other reasons: maybe we’ve stumbled on something interesting about the heterogeneity in disease transmission across settings.
What is almost certainly true is that different places may have different styles of epidemic, defined by different parameters that underlie transmission events. Our model has at least helped us (as in, the research team), and hopefully can help others, reconsider how this heterogeneity plays out from setting to setting.
So in the end, we started with data from the pandemic, and built a model to explore how surface composition influences the outbreak dynamics.
Through our research, we were successful (I hope) in using a real-world pandemic to address basic questions in the ecology of infectious disease. Instead of “carpetbagging,” where one can over-apply a skill set to a problem out of their lane, we “reverse carpetbagged” — used the topic that everyone else was studying to address questions of broad relevance.
A criticism: one might suggest that its unethical to use a real-world problem to ask theoretical questions.
But is it?
Well, of course not.
Part of the problem with the COVID-19 pandemic...or rather...one of the causes of the pandemic (in some ways) is our collective naivete about how emerging infectious diseases occur.
SARS-CoV-2 is not the last of the emerging infectious diseases. Hell, it may not be the last of the emerging coronaviruses within the next decade. And so, while we need to understand how to treat, manage, and properly disseminate a vaccine for SARS-CoV-2, the pandemic has taught us something else: our basic understanding of pandemic science has long way to go.
The data on SARS-CoV-2 is some of the best in existence. Never before have we collected more data on a live pandemic as it unfolded. Within this data are the means to address questions that might be relevant for other pandemics, past, present and future.
As I remain,
Big. Data. Kane.
November 28, 2020
There is a companion paper (see below) on this topic that we also published in the fall of 2020 (fittingly, the companion paper actually hit print before the paper it was a companion to….LOL. There’s a chance we’ll write a short piece on that as well. More soon!)
Gomez, Lourdes M., et al. “The Epidemiological Signature of Pathogen Populations That Vary in the Relationship between Free-Living Parasite Survival and Virulence.” Viruses 12.9 (2020): 1055.Available here (open-access)