Science is Failing Its Communications Test
Why science is losing the argument about COVID-19
Politico reports this morning that Morgan Stanley economists are now predicting a 30.1% annualized decline in US GDP for Q2 — which would be the worst quarterly performance in 74 years. That’s the good news. The bad news: lots of other economists think that prediction might be optimistic. The really bad news: those in power in the United States seem ready to risk the full wrath of a pandemic instead of the economic depression fighting that pandemic might bring on (even though their actions might well give us both).
The U.S. chapter of the COVID-19 crisis was always at some point going to be about epidemiology vs. politics and economics and social psychology. Epidemiology was at some point going to be forced to present some serious modeling about how its recommended public health measures were going to be less harmful to public health (and all that supports it) than would be throwing tens of millions out of work, shuttering hundreds of thousands of businesses and tanking the economy for months or longer.
Epidemiology, we might say, was going to have to do more than just be science; it was going to have to be persistently convincing against powerful competing interests and dynamics.
An old friend, Dave Lowenstein, sent a thoughtful response to my essay to you last Thursday. Dave essentially asked two questions:
- Why aren’t we isolating high-risk groups as the core of our strategy against the spread of the virus, since the data shows COVID-19 hits those groups disproportionately the hardest (and their removal from harm’s way would considerably ease the strain on our health care system)?
- Why aren’t the costs of public health interventions (such as lockdowns) being weighed against the public health costs of their damage to the economy?
I began a quick email back, and then realized my answers to Dave’s questions were unsatisfying — well, we have to isolate everyone because that’s the only way to stop the spread and protect the health care system…but if the only ones who get hit hard are the high-risk groups, then wouldn’t isolating them save the health care system aaaaggh? Only a day later, after considerable reading, did I feel as if I had enough information to craft an intelligent response, to move from “everyone needs to isolate now” as a mantra to that phrase as informed conclusion. (I’ve published our full exchange below; I encourage all of you to dive into it, especially for Dave’s careful formulations and pushback.)
The problem: Dave’s questions are ones many thinking people are posing, and science either hasn’t answered them or doesn’t have the means to answer them. There was virtually no content addressing them — certainly no journalism. Perhaps Morgan Stanley or Moody’s or Goldman Sachs is doing this modeling, but they’re not sharing it with all of us.
Which creates an even bigger problem. Quoting Dave:
Relative to the global population, I’m pretty well educated and informed, and have a pretty good lay-person ability to synthesize and understand data, and to assess relative risk based on that data. I’m also not anti-government or anti-authority, i.e., I’m not particularly resistant to having people in authority suggest or tell me what to do, as long as I have a baseline level of trust in their motives and commitment to making decisions driven by facts, data, and qualified subject matter experts. I’m not certain if this makes me more or less receptive to effective messaging, but right now I’m not convinced we’re on the right track (with making fact-based and data-driven decisions that reflect a broad perspective of societal factors and expert inputs), and I’m remarkably unmoved by the messaging that’s coming from our leadership (not just at the Federal level).
He’s not the only one, as we’re now seeing. It’s easy to now shout “follow the science” and tweet #notdying4WallStreet to the like-minded. It’s going to be much, much harder to motivate universal public commitment to a prolonged war against an invisible enemy, especially in the face of deepening and widening impoverishment, especially when we’ve been repeatedly told it won’t kill the vast majority of us. And if we fail, it won’t be simply a failure of leadership. It will be a failure of science communications — or, to be more precise, of scientists to recognize that making their case is about convincing us it’s also the best-case scenario.
Science communications has to be far better than “we’re scientists, trust us” — especially when the stakes are this high.
After you’ve read our exchange, let me know your thoughts through LinkedIn.
‘To better balance the costs vs. benefits of these incredibly powerful societal concussions, shouldn’t we be focused on social distancing and quarantining of only high-risk populations and those who must engage with them intimately?’
March 20, 2020 4:23p
Hi Bob -
Good edition as always. I believe that I have a pretty thorough understanding of these issues, and was communicating them to my own networks several weeks ago. But your discussion doesn’t address a key epidemiological question that I’ve had for some time, that I haven’t seen addressed in any of the discussions I’ve read. Specifically, to better balance the costs vs. benefits of these incredibly powerful societal concussions, shouldn’t we (starting with governments, but then extending to organizations and individuals) be focused on social distancing and quarantining of only high-risk populations and those who must engage with them intimately? My assumptions and logic are as follows:
- This is a fairly infectious virus that is likely to eventually infect a majority or a very substantial minority of global populations, at least until a vaccine or drug-based cure is available (way out). So all social distancing and more aggressive lockdown activities are only serving to “flatten the curve”, and will not substantively affect the long-term cumulative infection rate.
- ICU hospitalization and mortality rates for the highest-risk population segments (e.g., over age 60; underlying immune/pulmonary health issues; etc.) are at least an order of magnitude greater than for the lowest risk population segments, and the majority of the population falls into the lower-risk categories. The latest report from the CDC (3/18) has the mortality rate at 10–27% for ages 85+, and a fraction of 1% for those under age 54.
- Due to the low risk of CV-driven ICU hospitalization and mortality for the majority of the population, even a rapid spread of the virus among the low-risk majority will likely result in a peak of hospitalizations that does not come close to exceeding system capacity (unless accompanied by a parallel rapid spread among the high-risk groups).
- Isolating high-risk populations would be very challenging, but surely not more challenging than the societal and economic costs we are currently faced with (and will face when integrated out over time). For a fraction of the amount of money the US government is going to spend on broad-based support to individuals and businesses as a result of the pending recession or depression, never mind the cost to the government in lost revenue from depressed economic growth, we could have crafted strategies delivering necessary goods, services, care and comfort to those at high-risk, while protecting those on the front lines of doing so.
- The massive scope of the global lockdown is/will cause trillions of dollars of damage to regional and global economies, and there are huge associated human costs. How many people who live paycheck-to-paycheck will end up on the streets and die prematurely due to exposure/illness/crime? How many people who got health insurance through their employment will lose it when they lose their jobs, and die prematurely due to illness/disease? How many incremental deaths will occur due to increased crime? How many excess suicides will occur due to loss of businesses and financial stability? How much will the global standard-of-living decline? [We know that lifespan obviously correlates strongly with SoL]. It doesn’t seem implausible that excess deaths (among people otherwise at lower risk for CV death) due to the CV-instigated economic catastrophe will outpace excess CV deaths that would occur among the same lower-risk groups if we were to focus social distancing & quarantining only on the higher-risk groups.
- So shouldn’t we have at least considered an immediate and rigorous combination of voluntary and enforced social distancing, lockdown and quarantining of just the high-risk sectors of the population, with a focus of support on those who engage with and care for them, instead of doing it haltingly (and then rapidly once it was too late) but indiscriminately to all of society?
‘I’ve certainly seen variations of your question/proposal — as someone put it crassly, “why are we crashing the world economy for elderly Italian smokers?”’
March 21, 2020 5:16p
Thanks for reading and sending these thoughts. Interesting questions and they forced me to do a bit of additional reading, for which I’m grateful.
So let’s pause first to acknowledge how we got into this fix: the US is, by increase in the number of reported cases since the 100th reported case, the second worst in terms of coronavirus response, behind only China. As Brad DeLong points out, China got better, and so could we.
I’ve certainly seen variations of your question/proposal — as someone put it crassly, “why are we crashing the world economy for elderly Italian smokers?” Tyler Cowen has just posted about it as one of the three ideas people are writing him the most about, asking: “Given how many older people now work (and vote), and how many employees in nursing homes are young, I’ve yet to see a good version of this plan, but if you favor it please do try to write one up.”
You are (as I understand you) advocating a mandatory, rigorous and indefinite (until we find a vaccine) lockdown for the high-risk US population. That population is at least 68.7 million people — those who were age 60 or older in the US as of 2016. It I include other cohorts not in that age group who are at risk, let’s call it 75 million, somewhere between one in every four and one in every five Americans. So, a lot; making a lock down difficult, tricky and very expensive to pull off. (If you take the $1 trillion stimulus package and distribute it evenly across the cohort, it yields $13.5k per person. Not nearly enough to do the job.)
You acknowledge that lockdowns have huge knock-on costs. They also prompt evasion. They seldom work, and not for long. The Imperial College report acknowledged same, which is why the authors came up with their weird recommendation for (I think) eight punctuated social distancings after each new outbreak.
I’m not sure what you mean by “an immediate and rigorous voluntary and mandatory social distancing” strategy for the low-risk population, but would guess that your “social distancing” doesn’t economically kneecap many sectors and would thus make slowing spread harder. The Imperial College authors’ definition of what’s needed — what they called “suppression” — included comprehensive, population-wide social distancing; school closures; and case isolation as the only strategy that could get spread below R=1 and a) allow for comprehensive testing, which facilitates much more effective treatment and allocation of resources, and b) rapidly reduce case incidence while also keeping caseload manageable until the autumn for current critical care bed capacity. All models are wrong, some are useful; but all of them say anything short of suppression really isn’t putting the brakes on spread and caseload in a significant way, as every country that has dealt with this before us has found, not to mention in past flu pandemics. Also: Can suppression bridge us to comprehensive testing that then facilitates a reinflation of the workforce and the labor market with those who test negative? (Alex Tabarrok says yes, and that it’s key to a quickish economic recovery.)
Hospital bed capacity and equipment supply are, obviously, some of the biggest issues around health care system stress and fatality. (I have seen one estimate that said the fatality rate starts doubling as soon as ICU capacity is exceeded.) You say that there is a “low risk of CV-driven ICU hospitalization and mortality for the majority of the population” and thus, presumably, once we’ve locked down the old cohort, we can worry a lot less about the capacity of the health care system to deal with COVID-19. You’re correct about mortality rates by age cohort, but I’d disagree with you about ICU demand, based on the latest CDC numbers. For hospitalizations and ICU hospitalization, the latest CDC numbers say: 38% of US hospitalizations for COVID-19 have been for ages 20–54. For ICU US admittances related to COVID-19, it’s 7% above 85 years of age, 46% 65–84, 36% 45–64 and 12% 20–44. Small sample size, yes; but we have to make decisions now based on the numbers we have and that’s almost 50% of ICU admittances under age 65 — hardly low-risk for that cohort.
Let’s assume for a moment we can lock down everyone age 60 and over. Here’s the Imperial College table for hospitalization, critical care and fatality rates by age cohort for infected cases:
That doesn’t look like much for, say, even the 50–59 cohort: about one in every symptomatic cases requires critical care. There are 43 million people in that cohort. If we assume a 50% population-wide infection rate, 1% of 21.5 million people yields 215,000 critical care cases in that cohort over the course of the infection. Not all at once, of course — but you get the picture. Just because people aren’t dying as much doesn’t mean the hospitals don’t get creamed.
ICYMI the Imperial College chart on critical care bed capacity — as best I can tell, the orange line is closest to your scenario:
Back to that cohort of 75 million or so: let’s triple or quadruple or quintuple that $1 trillion stimulus. You’re still talking about creating a system of support, not just paying people to stay home. Balance that with the many, many plans economists are putting out there (here’s Tyler’s) for fiscal interventions and other tactics to keep workers paid and give businesses bridge loans. I’m not an economist; but I’d argue you would have to factor those into your cost-benefit analysis, not just assume the government won’t intervene or will but not enough to let the bottom drop out. (Then again, with this government, I’m not sure why I would have confidence in a sudden switch to repeated bold, smart, decisive action.)
So, I have serious questions about a number of your parameters: that we could lock down that 75 million, enforce a kind of social distancing that delayed spread significantly but didn’t cripple economic activity, and assume that the resultant caseload wouldn’t also overwhelm the health care system. I also question whether there are only two choices — in essence, solve for now, or solve for the future. For instance, here’s what Aaron Carroll (one of my favorites) and Ashish Jha wrote in The Atlantic Thursday in favor of a short suppression:
Because of this, some are now declaring that we might be on lockdown for the next 18 months. They see no alternative. If we go back to normal, they argue, the virus will run unchecked and tear through Americans in the fall and winter, infecting 40 to 70 percent of us, killing millions and sending tens of millions to the hospital. To prevent that, they suggest we keep the world shut down, which would destroy the economy and the fabric of society.
But all of that assumes that we can’t change, that the only two choices are millions of deaths or a wrecked society.
That’s not true. We can create a third path. We can decide to meet this challenge head-on. It is absolutely within our capacity to do so. We could develop tests that are fast, reliable, and ubiquitous. If we screen everyone, and do so regularly, we can let most people return to a more normal life. We can reopen schools and places where people gather. If we can be assured that the people who congregate aren’t infectious, they can socialize.
We can build health-care facilities that do rapid screening and care for people who are infected, apart from those who are not. This will prevent transmission from one sick person to another in hospitals and other health-care facilities. We can even commit to housing infected people apart from their healthy family members, to prevent transmission in households.
These steps alone still won’t be enough.
We will need to massively strengthen our medical infrastructure. We will need to build ventilators and add hospital beds. We will need to train and redistribute physicians, nurses, and respiratory therapists to where they are most needed. We will need to focus our factories on turning out the protective equipment — masks, gloves, gowns, and so forth — to ensure we keep our health-care workforce safe. And, most importantly, we need to pour vast sums of intellectual and financial resources into developing a vaccine that would finally bring this nightmare to a close. An effective vaccine would end the pandemic and protect billions of people around the world.
All of the difficult actions we are taking now to flatten the curve aren’t just intended to slow the rate of infection to levels the health-care system can manage. They’re also meant to buy us time. They give us the space to create what we need to make a real difference.
However, if there’s a true comprehensive lockdown and they try to extend it past a few weeks, I agree: forget it. Our societies won’t stand for it, especially with how poorly this threat has been explained.
Thanks, stay healthy
P.S.: ICYMI: Your argument was essential made in tomorrow’s NYT: https://www.nytimes.com/2020/03/20/opinion/coronavirus-pandemic-social-distancing.html
There have been many scientific as well as political criticisms of the now-abandoned UK herd immunity approach — here is one that has gotten a lot of attention: https://unherd.com/2020/03/the-scientific-case-against-herd-immunity/
‘I do expect two things from those in leadership positions who are faced with managing our responses to such national and global crises’
March 22, 2020 11:54a
Hi Bob -
Thanks for the really thoughtful reply, and for pulling together some of the data and ideas that have been published on the subject. You’ve been poking in some places that I haven’t and some of those have produced compelling analyses.
You’re correct that David Katz’ Times Op-Ed piece (which I had not seen until you flagged it) does make much of my point, and I’m glad to see it appear (however belatedly).
In terms of your rebuttal, the key assumption you make for which I’m not convinced (from the data I’ve seen) regards the likely rates of mortality and severe (requiring ICU) morbidity among populations younger than 60. While I fully acknowledge that I haven’t studied the available data carefully, everything I’ve read has indicated that the vast majority of severe cases and deaths among younger people have involved underlying or complicating health conditions. The hospitalization data you present from the Imperial College doesn’t disambiguate or differentiate between those with underlying issues and otherwise healthy individuals, and therefore (I think) results in an exaggeration of the scope of the challenge in implementing the strategy I was trial-ballooning.
I was not advocating the UK wait-for-herd-immunity approach, and I definitely agree with Alex Tabarrok that comprehensive and ongoing (repeated) testing would be needed, to make the “isolate the at-risk” strategy work.
That said, I’m obviously not a scientist, an epidemiologist, a behavioral psychologist or an economist (among the many types of subject-matter experts who should be contributing to our official government response), and don’t presume to make arguments on behalf of or superior to those of experts in the relevant fields. But I do expect two things from those in leadership positions who are faced with managing our responses to such national and global crises:
- That they will make fact-based and data-driven decisions that reflect a broad perspective of societal factors and expert inputs. In a crisis this massive and complex there are many such inputs required but they *must* include short- and long-term cost-benefit analyses of the type I highlighted below [e.g., excess deaths and diminished standard of living not just in the immediate aftermath but downstream and systemically].
- That they will clearly communicate both the strategy and the underlying logic in a way that galvanizes the majority of the public to act in their own and the larger public interest, and provides a cognitive pathway through the wilderness that they can follow despite their fear and minimal knowledge of the issues.
While you certainly address the 2nd point (it goes to the core mission of your newsletter itself), my take is that your assessment elides the part of my first point that demands an attempt at a quantitative analyses of the options and strategies to both deliver the best possible outcome from among some very unpleasant options, but also to allow the resulting messaging in item 2 to be honest and compelling and likely to result in buy-in from the population at large.
In the end though, I’m not sure that you and I are so far apart. I completely agree with you on the Carroll/Jha Atlantic piece, which I had not seen (thanks!), and that’s exactly the kind of logic I wish we were seeing from our leadership. Excerpting another very brief paragraph from that article, that quite effectively captures a big part of my suggestion: “To achieve this, we need to test many, many people, even those without symptoms. Testing will allow us to isolate the infected so they can’t infect others. We need to be vigilant, and willing to quarantine people with absolute diligence.”
‘Many, but not all are convinced by the messaging; and the durability of the messaging will erode as pain and panic increase’
March 23, 2020 10:23a
Uncertainty is worth considering when judging policy responses, or lack thereof. I would guess, because of the untrustworthiness of the data out of China, that governments are favoring general vs. cohort lockdowns because there is still a lot of uncertainty about how this virus hits various populations. I have not seen enough to agree with you that we’re only talking about people with underlying complications — and anecdotally, the news and Twitter is now full of people in their 40s who are being hospitalized for this.
On Sunday, Marginal Revolution’s Alex Tabarrok posted “The Internal Contradictions of Segregating the Elderly.” I am not sure about the numbers underlying his 330,000 deaths point, but his point that “the more young people get COVID-19 the less realistic protecting a subset of the population becomes” seems true to me, especially given how common the young caring for the elderly is in our health system. Megan McArdle gives her own spin to the problem on Twitter.
Your point #1 highlights a clear division that is about to surface publicly, I predict: that between epidemiology and economics. The epidemiologists have the upper hand in national responses right now, but the economists are about to assert themselves and will do so increasingly as quarantines continue. It seems unlikely — based on those national responses — that anyone has developed robust modeling of the economic cost/benefits of a range of public health interventions/policies, with or without economic interventions.
Your point #2 highlights the same — many, but not all are convinced by the messaging; and the durability of the messaging will erode as pain and panic increase. A combined public health/economic strategy would provide that playbook. I’m not sure it is possible, but it sure is needed, if only as political communications.
Let’s keep passing on relevant pieces to each other as we see them. Thanks again, and stay healthy.
‘We need at least a robust analysis of a more targeted approach that focuses on more testing and then isolation of the highest-risk populations who may in fact turn out to be a much smaller segment than our current strategy assumes’
March 23, 2020 11: 20a
Hi Bob -
Recognizing that we need to be circumspect about the quality of the data coming out of China (I absolutely do, for several reasons), I thought it was interesting to note that on NPR Morning Edition today they referenced a new report from China out today that the mortality rate for all known infected patients (including the oldest) was well under 1%, for those without an underlying health issue. I tried searching online today for the data in support of this report and couldn’t find it, but will be very interested to see if that report holds up. Even if it does, more detail will be necessary, as there’s a big difference between mortality of 0.1% and 0.8%, even though both are “well under 1%”. But either way, if accurate, and the latest analysis from Goldman that GDP will decline by 24% in the next quarter (April to June) is also deemed credible, the combination of these inputs reinforces my view that we need at least a robust analysis of a more targeted approach that focuses on more testing and then isolation of the highest-risk populations who may in fact turn out to be a much smaller segment than our current strategy assumes.
One other thought/comment on the #2 (messaging) issue: Relative to the global population, I’m pretty well educated and informed, and have a pretty good lay-person ability to synthesize and understand data, and to assess relative risk based on that data. I’m also not anti-government or anti-authority, i.e., I’m not particularly resistant to having people in authority suggest or tell me what to do, as long as I have a baseline level of trust in their motives and commitment to #1 (making decisions driven by facts, data, and qualified subject matter experts). I’m not certain if this makes me more or less receptive to effective messaging, but right now I’m not convinced we’re on the right track with #1, and I’m remarkably unmoved by the messaging that’s coming from our leadership (not just at the Federal level).
‘At some point, I have to ask: why did these Asian countries start to act this way with the data they had, even if those data were showing a preponderance of deaths in older cohorts?’
March 23, 2020 1:08p
JHU’s global cases map (which is pretty much gold standard right now) says China’s death rate is about 0.4%.
There might be a “natural” fatality rate for this virus, all things being equal. But all things are never equal. Given the variations among demographics, socio-economics, culture, public health, health care capacity and public health responses, fatality rates are going to vary. We have a good sense now of the steps China, Singapore, HK, S. Korea, Taiwan took to put down spread. All different, but all extraordinary — and all involved either aggressive containment strategies or, eventually, population-wide isolation. Spread has been confined (thus far) in all those societies to less than 1% of the population.
At some point, I have to ask: why did these Asian countries start to act this way with the data they had, even if those data were showing a preponderance of deaths in older cohorts?
Part of it has to be their experience with SARS, MERS, other flus. General quarantine is what you do when something like this happens. You stop spread, test, isolate the sick, stop giving the virus a chance to advance.
Part of it has to be that it’s next to impossible — especially in these cultures — to quarantine a age-based cohort. Families live together. The old are cared for by the young. Nursing homes are not a thing, generally.
Again, I’d argue that our high-risk cohorts are entangled with our low-risk cohorts in support systems that would be next to impossible to replicate on the fly, especially by a government that struggles to produce and distribute a surge in basic medical supplies, much less surge testing. Caring for my mother, for instance, would have been impossible over the last three years of her life without a phalanx of aides in low-risk cohorts. And we simply don’t have the testing in place to ensure those people are virus-free right now, or whomever a new program might replace them with.
Europe. Italy, Spain, France were all late in responding and have and are responding quite differently from Asia. Those death rates are going to be much higher. Italy is below 1%, but only a small fraction of their cases have resolved one way or another. We shall see.
But: mortality from the disease isn’t the only thing to look at, as we’ve been discussing. If your health care system goes on tilt, fatalities immediately skyrocket. Are the numbers of recoveries reported by other countries for younger cohorts happening in systems that surged capacity or practiced radical containment?
I was looking for the study you heard on NPR today and came across this quote from our surgeon general:
Adams also noted that COVID-19 seems to be acting differently in the U.S. and other countries than it did in China, where it was detected in December. In an important shift, the virus seems to be affecting young people at higher rates.
In New York, the epicenter of the disease in the U.S., Gov. Andrew Cuomo recently said that people who are between the ages of 18 and 49 represent 53% of the state’s cases.
“It’s important for young people to know, you can get this disease, you can be hospitalized from this disease, you can die from this disease — but most importantly, you can spread it to your loved ones,” Adams said. “So we need you to really lean in.”
Agree completely with you that the communication of strategy has been near-woeful on all fronts and has and will undermine all these efforts.
2.5 million research papers are published every year — so how are research-driven organizations breaking through the noise? Find out: subscribe to my Tuesday and Friday emails on research communications and authority content.