Skip to Main Content

In a recent and controversial First Opinion, epidemiologist and statistician John Ioannidis argues that we lack good data on many aspects of the Covid-19 epidemic, and seems to suggest that we should not take drastic actions to curtail the spread of the virus until the data are more certain.

He is absolutely right on the first point. The U.S. has done fewer tests per capita so far than almost any rich country in the world. And many critical details of the epidemiology — including the absolute number of cases, the role of children in transmission, the role of presymptomatic transmission, and the risk of dying from infection with SARS-CoV-2 — remain uncertain.

On the second point, I would say that his article did what contrarian writing should do: started a discussion. We spoke by phone on Tuesday, not long after his article appeared, and found that we had more in common than it appeared when I first read it.


So without trying to characterize Ioannidis’ view, I will state a strongly held view of my own: We know enough to act; indeed, there is an imperative to act strongly and swiftly. It is true that we can’t be sure either how many infections there have been in any population or the risk of needing intensive care or the case fatality rate. These uncertainties are two sides of the same coin. Nonetheless, two things are clear.

First, the number of severe cases — the product of these two unknowns — becomes fearsome in country after country if the infection is allowed to spread. In Italy, coffins of Covid-19 victims are accumulating in churches that have stopped holding funerals. In Wuhan, at the peak of the epidemic there, critical cases were so numerous that, if scaled up to the size of the U.S. population, they would have filled every intensive care bed in this country.


That is what happens when a community waits until crisis hits to try to slow transmission. Intensive care demand lags new infections by about three weeks because it takes that long for a newly infected person to get critically ill. So acting before the crisis hits — as was done in some Chinese cities outside Wuhan, and in some of the small towns in Northern Italy — is essential to prevent a health system overload.

Second, if we don’t apply control measures, the number of cases will keep going up exponentially beyond the already fearsome numbers we have seen. Scientists have estimated that the basic reproductive number of this virus is around 2. That means without control, case numbers will double, then quadruple, then be eight times as big, and so on, doubling with each “generation” of cases.

To stop an epidemic like that permanently, nearly half the population must be immune. While the exact number of people infected in each population is unknown, current estimates are that for every symptomatic case there is about one asymptomatic or very mild case.

In populations with good ascertainment of symptomatic cases, the number of infections is perhaps double what is observed (in the U.S., where testing is limited, true cases are a much higher multiple of reported cases). In well-tested countries, we can be nearly certain that no population has reached anywhere near half of its people infected. That means that when each country lets up on control measures, transmission will increase and the number of cases will grow again.

It is crucial to emphasize that a pandemic like this does not dissipate on its own, as Ioannidis suggested as a possibility. Severe acute respiratory syndrome (SARS) in 2003 was hammered into submission by intense public health measures in many places, which were effective because transmission was mainly from very sick people. Middle East respiratory syndrome (MERS), which emerged in 2012, is a weakly transmissible infection that causes outbreaks in hospitals, but is otherwise much less contagious than Covid-19.

There are two options for Covid-19 at the moment: long-term social distancing or overwhelmed health care systems. That is the depressing conclusion many epidemiologists have been emphasizing for weeks, and which was detailed in an analysis released this week by the Imperial College London.

Ioannidis is right that the prospect of intense social distancing for months or years is one that can hardly be imagined, let alone enacted. The alternative of letting the infection spread uncontrolled is equally unimaginable. We certainly need more data. Even more than that, we need a breakthrough to make effective treatments, vaccines, or other preventive measures available at scale.

Waiting and hoping for a miracle as health systems are overrun by Covid-19 is not an option. For the short term there is no choice but to use the time we are buying with social distancing to mobilize a massive political, economic, and societal effort to find new ways to cope with this virus.

Marc Lipsitch, D.Phil., is professor of epidemiology at the Harvard T.H. Chan School of Public Health and director of Harvard’s Center for Communicable Disease Dynamics.

  • Mr. Lipsitch: my state has no “shelter-in-place” order. I have two teenagers, and it’s business-as-usual for them. Since the supermarket and convenience stores are all that are now open, they hang out there. There’s absolutely nothing I can do about it. They come home and observe no sanitary measures whatsoever. When I tell them there’s an epidemic, they think I’m certifiably insane. How can there be an epidemic when all of their friends are out, having fun, and no one is ill? Makes perfect sense from a teenager’s perspective.

  • So… how does the response to Covid-19 relate to H1N1 swine flu? I keep hearing this comparison to SARS and MERS but talk about H1N1 in 2009 is avoided. Why?

    • Nobody wants to talk about 2009 H1N1 because it started out with a huge amount of fear, but by the time it had spread around the world, it statistically didn’t kill any more people than a typical seasonal flu.

      Note that 2009 H1N1 had more things going for it – we had approved vaccines and antivirals (for other H1N1 strains, not that one) at the time. On the other hand, it was more concerning because it tended to hit children the hardest. Early on it had shown quite the ability to overwhelm medical systems when left unchecked, and did so in Mexico. But it just sort of fizzled out as it went pandemic.

      I do worry about overreaction. Enforced social distancing regulations (requirements on companies for both their workers and customers) are one thing. But enforcing mass lockdowns and shutting down the global economy (mass unemployment, mass bankruptcies of companies, huge government indebtedness trying to prop things up, etc) is an entirely different thing. If the disease doesn’t hit hard enough, there’s going to be a huge backlash, and I worry about that. COVID-19 certainly can – like 2009 H1N1 – overwhelm medical systems when no action is taken. But how much action is too much?

  • I don’t have any better answer than anyone else what is the best approach. I believe for a short time (a month maybe?) we need to basically shut down the economy by staying away from work, etc. But the idea being floated by the medical profession we need to shut things down, shelter in place, etc. for 12 to 18 months is crazy. If something can’t happen, it won’t. And that won’t happen.
    Doing something like that would undo the economic progress of the last decade or more. People need to eat, to live, to work. Where will the money come from to fund everyone who loses his or her job, or who’s life savings have been wiped out, whose retirement will now be delayed for many years, or the pension funds that will be wildly undefunded? What would we say to our young people, who will be be looking at 25% unemployment rates, maybe more if you count under-employment? Everyone says the governments of the world can provide fiscal or monetary stimulus. Really, to keep the world economy from collapsing for as long as 12 to 18 months? Where are these governments going to get the money? They won’t able to borrow the money if the real economy has ground to a halt and everyone is hoarding cash. They won’t be able to raise taxes enough to transfer the money from one pot to another because incomes will collapse and wealth is tied up in assets (houses, businesses, stocks and bonds) that will have collapsed in value and be much more illiquid than they are even now.

    • Money is essentially a fiction. It is not a primary concern. The true concern is a drop in the output of goods and services. If people aren’t working, the real issue is not that they aren’t earning money (which isn’t real). It’s that they aren’t producing anything. Things like food and useable energy, which ARE real.

    • Money is an IOU on human labour. Actual goods and services all trace back 100% to human labour. If you shut down the economy, you shut down the supply of goods and services.

      The concept of shutting down the economy for 12-18 months is ludicrous nonsense that anyone who says it should be embarrassed to have the words come out of their mouths. That’s effectively saying, “Let’s all go Mad Max and have a large portion of the planet die.”

      And it’s a nonsensical notion that it’s required to fight the disease, period. China extended its Lunar New Year holiday by a single week, then reopened outside of Wuhan (Wuhan has been steadily opening since then). South Korea has continued operating while eradicating the disease. Taiwan too had early infections, and on and on. Many countries in Asia have successfully dealt with the disease without turning off their economies for a year and a half, by imposing tighter social distancing and sterilizations requirements on employees and customers. This western notion of “It’s either do-nothing or shut-down-the-entire-planet” is unbelievably harmful.

      Even in places that took the do-nothing route, like Italy…. yes, it’s terrible that several hundred people per day have died there. But on a *normal March day* in Italy, a couple *thousand* people die. COVID-19 is not the be-all end-all cause of mortality, *even in countries that do nothing whatsoever until it’s too late*. And everyone has a price on their head – an actuarial cost of life, because in countless fields, the more money you spend, the more lives you save, from all of the various causes that end up killing people (both medical and nonmedical). If you crash the global economy and rob the money supply, forcing a decrease in the price on everyone’s head, you’re going to spike the non-COVID deaths.

      And this is all ignoring the *side consequences* of locking everyone indoors – reduced exercise, increased depression / suicide / substance abuse, likely greater marital conflict / divorce rates, people putting off things that really shouldn’t be put off (critical maintenance, treatment for serious non-COVID medical issues, etc) and so on.

      People swing from one extreme to the other, and it’s frustrating to no end. NO, “doing nothing” is not acceptable. NO, “shutting everything down” is not acceptable either. IF (and only if) a local situation has been scientifically shown to have gone too far already, a BRIEF shutdown is acceptable; beyond that, the only acceptable solution is for life to go on, but under much stricter disease control measures. You simply cannot shut down the planet for long periods of time.

    • In terms of “nonsenses”, another one is the concept that hospital capacity will remain constant over the next year and a half. In what world can any person with a mental age over 4 actually believe this? China built new ICU wards in *weeks*. The global production rate of personal protective gear and ventilators is growing *exponentially*. Whatever target level of hospital capacity we’re going to need to reach ultimately, it’s *not* going to take a year and a half to reach it.

      What’s also not going to happen over the next year and a half:

      * Zero progress in antivirals (indeed, huge progress has *already happened*)
      * A normal 1 1/2 year vaccine development and testing schedule
      * No pre-production and pre-distribution of candidate vaccines while testing is ongoing

      Lastly, even the notion that this would take a year or more to reach herd immunity is a highly shaky concept. It’s becoming increasingly clear from the Iceland deCODE study and the Italian study in Vó (that is, random sampling / whole population studies) that the disease is already much more widespread than previously assumed, but 50-75% of cases have no symptoms at all, and for most people who do have symptoms, they’re so mild that they never suspect anything. That the disease is widespread, but the IFR is low. This matters because it means that in order to overload a medical system, a large percentage of the population must already be infected. A recent paper for example suggests that Wuhan was about 19% of the population infected at its peak. Given this, it’s hard to see how – even with curve flattening – it could possibly take a year or more to achieve herd immunity, like is assumed based on the notion that the disease *isn’t* widespread but has a high IFR.

  • Seems to me we are debating from different starting points…
    (1) yes, it’s understood that the overall fatality rate is unknown, quite possibly under 1%, and more measurement in wider society is required. No dispute there.
    (2) but, the immediate risk, *independent of the mortality rate debate*, is overwhelming of ICU beds and ventilators which, if it happens, leads to many additional fatalities
    (3) whilst nothing is certain, given the experience of modern, well-respected healthcare systems like Italy, this is a real and imminent risk for the US as well (and again, doesn’t rest on any assumption about overall fatality rates).
    (4) Now you could argue a more refined model might show the ICU shortage risk is somehow lower for US, but I have not seen any such model or argument (and I don’t believe Ioannidis’s article provided one), and meanwhile multiple countries are experiencing the same result (at different scales, quite possibly depending on whether and how early they implemented social distancing)
    (5) so the case for action now is based on the precautionary principle. If we are right about this risk, we may well already be close to the last opportunity to head-off an overwhelming of ICU beds (with many consequent excess fatalities). If somehow the risk doesn’t materialise in the US that should become apparent soon (through good measurement and stats) and social distancing measures can be relaxed quickly (and geographically selectively as we understand things better).

    • It seems the experience of Northwestern Italy is the driving force behind everything we are doing.

      I know the Italian healthcare system is good, but how prepared were the hospitals in Milan given the disproportionate amount of seniors living there?

      We haven’t seen ICU collapse outside of Lombardy and Wuhan, yet we had cases working through retirement communities outside Seattle before they were in Lombardy. So why haven’t Seattle area doctors been forced to ration ventilators like Italian doctors have?

      Our whole response is based off of outliers and bad data, it seems.

    • Of course Italy will have differences to US. But the argument isn’t just: US will have a problem because Italy did. It’s that Italy (and other places) have graphically demonstrated two things at least:
      – far more people than was widely understood end up requiring ICU ventilation when suffering covid-19 disease, including younger people
      – ICU capacity can be rapidly overwhelmed and hard to scale quickly
      Now, even if you believe US is different, and in particular is better placed, the behaviour of exponential growth curves might mean that the differences merely buy you a few extra days before being overwhelmed.

      None of this is certain of course but, as said above, it’s the precautionary principle that drives you to action now. Better to take short term action now to reduce very material risk of catastrophic outcome, when that action can easily be relaxed later. The alternative of waiting means that, if the risk materialises, it will be too late to recover (remember, the real infection rate is highly likely to be an order of magnitude greater than what we measure today).

    • > If somehow the risk doesn’t materialise in the US that should become apparent soon (through good measurement and stats) and social distancing measures can be relaxed quickly (and geographically selectively as we understand things better).

      That would be a mistake. If what we are doing is working, then it will appear that we are doing nothing at all. The ICU’s won’t be pushed to capacity. Speaking as someone with a Biological Science background, we have zero innate immunity for this virus. It has never been seen by humanity before. Everyone is susceptible. And even without knowing that, just observing the way it quickly spread throughout the world should be enough to show that it will do the exact same thing here, and is already well underway in doing so. If our social distancing efforts were early enough and earnest enough and we happened to avoid hospitals approaching critical capacity, it may appear to some that we are doing nothing at all. And relaxing the social distancing would be a mistake if we wanted to slow the spread of this virus.

    • @Lawrence: I understand your points, and the risk of it appearing like nothing is happening.
      But much of this thread is about whether we should be taking action now, or measuring more first and then taking action. My point is that if we act now and measure more (eg, percentage of infected requiring ICU in different demographics) then we will be in a position to relax social isolation measures *if* the evidence supports it. And there is a decent chance it will, at least in places – the much cited Imperial College research (the basis of UK gov policy) for example, models a repeating sequence of social isolation periods punctuated by periods of relaxation to keep infection levels low enough to avoid overwhelming ICUs, whilst still allowing as much social interaction as possible.

  • For those who are dissing Ioannidis’s article, his whole point is that right now we are making decisions based on unknowable and potentially false data, decisions that may be — and the emphasis is on “may be” — more draconian than necessary.

    His whole point was that we need a true statistically valid assessment of this epidemic. That will require a properly sized and stratified random testing to determine (a) what percentage of the population is infected by the virus, by age cohort and underlying conditions; (b) what percentage of the population, by the same stratification is NOT yet infected but does not have antibodies; and (c) those who HAVE the antibodies and either (i) have been infected and presumably are immune or (ii) haven’t been infected or are naturally resistant.

    And, given the fairly well understood transmission vectors, we can assess how those numbers will “scale up” to the population as a whole.

    Until we do this, we are making decisions based effectively on unreliable and potentially misleading anecdotal evidence; are we Italy, are we South Korea, or are we China? The fact is we are NONE of those countries — we have different demographies, different racial makeup, different population densities. To look at those countries and say, “Oh, my, we need to shut down the economy!” is insane.

    Look, we may have to continue to do just that. But until we get a true statistical handle on what’s happening we are flying blind. And I fear we WON’T do the kind of testing we SHOULD do, because all the tests are going to go those who we suspect are infected, thus providing useless data on which to make sensible decisions.

    • Karl K, yes, indeed, we are flying blind. It is much better to stop the plane before we crash than to keep on flying blind until we do. Sad but true: the US has wasted the time it had to get to this very point with the possibility of taking sensible decisions.

    • > His whole point was that we need a true statistically valid assessment of this epidemic.

      I’m sure COVID-19 will wait its turn until we have submitted our assessments

  • Sir, your argument to act swiftly is supported by existing data. The average number of monthly deaths in the 70+ cohort in Lombardia, Italy was 6973 in 2018. Since 21th February 2020 the mortality of COVID-19 in the region is 2168. Almost all of them were over 70. That virus is proven to cause virtually 30 per cent of the monthly average loss in the most vulnerable group.
    Professor Ioannidis was right, we do not have the adequate size of data to be 95 per cent certain in our decision (and we will never have). Hence, I believe that the risk of Type I error of his argument is too high to gamble.

    • Mortality isn’t the whole story. Those who are hospitalized but survive still take up a hospital bed, and possibly a respirator or ICU bed, for weeks. These far outnumber the dead. And in Lombardy they’ve filled the hospitals to the extent that mortality rates for non-coronavirus causes are likely rising significantly.

  • “ In Italy, coffins of Covid-19 victims are accumulating in churches that have stopped holding funerals. In Wuhan, at the peak of the epidemic there, critical cases were so numerous that, if scaled up to the size of the U.S. population, they would have filled every intensive care bed in this country.”

    Couple of big straw mans.

    1. In Italy about on average 250,000 to 300,000 people die during WINTER season. A lot of them are 70+ years old. That’s a lot of funerals to handle in a quarter of a year. Did you hear about funeral crisis before? No. Because there was no lockdown. Now you impose lockdown, create panic and produce a funeral crisis for deaths comparable to the average daily rate without covid and then use the same fact to argue that we should have lock-down.

    2. Wuhan scales to the size of US is not US. US is a large country, population density is different, age distribution is different, lifestyle is different, civic sense is different (all of which are factors in disease transmission). It is an interesting high school homework problem to calculate probabilities, but hardly worth musing about when serious public policy decisions are in hand.

    • Wait, do you really think that they can’t handle funerals because of the lockdown? Did you miss the part where doctors from Bergamo or other Lombardy hospitals plead for help, saying their ICUs are full and they can’t take in more patients? Did you miss the videos of people being literally left to die either at home or in hospitals because there’s nobody available to help them or they don’t have the space?

      What about the part where they are actually building hospitals? Is that another “bad flu season” phenomenon?

      What an ignorant troll.

    • People havent stop dying from orher deseases, so Covid19 doubled the deaths in Italy. Thats the problem with intensive care too.

  • Thank you for providing this necessary rejoinder to Ioannidis’ argument. I am absolutely appalled that a man of his expertise failed to understand the possible mass deaths of a “herd immunity” non-strategy for approaching coronavirus.

    • He didn’t fail to understand the possibility. He just thinks there are better ways to move forward given the information (and that economy collapsing efforts will likely kill more than the virus).

  • But are we burning our house down to kill a big spider? Because while we aren’t overwhelming medical resources, we are destroying the economy, creating mass unemployment, shuttering small businesses for good, hurting our children’s education and socialization, and likely will see a big spike in suicides and eventually violent crime. Is the cure worse than the disease?

    • No the cure is not worse than the disease; the disease involves people suffocating in droves. As for the fear that shutting down businesses temporarily, that takes no account of the fact that businesses all over Europe shut down every summer for 4-6 weeks for vacation. Entire countries are largely denuded of their populations as they flock to sunnier climes. Somehow miraculously people go back to work afterwards, nobody is bankrupted, and everyone takes it in stride.

    • this is indeed a much more serious question than almost anyone (except the stock market) seems to be considering.

      unlike the first response here, that notes that some businesses take vacations in the summer in Europe, this is a massive, worldwide, UNPLANNED cessation of business.

      entire industries are going to dissolve, starting with travel, restaurants, bars, and many others. the repercussions are going to be extreme. small businesses are likely going to be hit much harder than larger ones.

      there will be many many deaths, depression, and illness based on this economic damage. it is unprecedented.

      i usually don’t like “cost-benefit” analyses, but right now the whole discussion of the virus seems to focus on the costs of mass infection, which are very real. but there is almost no discussion of the cost of mass economic shutdown, which is also very real.

    • The coronavirus is raising a lot of important questions about priorities, which I hope continue after the virus has abated. It’s worth pointing out that global emissions have shrunk dramatically since this all began. People talk about the Economy like it’s some giant living in the hillside that we have to appease lest it destroy us all… but now there are other more powerful giants looming just beyond the hillside – Climate Change and its cousin Pandemic who are even more capricious and deadly than the Economy. It seems increasingly unlikely that we will be able to satisfy the hunger of the Economy and these other giants at the same time. We are entering a time of hard choices. Sorry if this comment is too psychedelic I’ve been quarantined for 10 days

    • We can easily quarantine 5% of the highest risk population to their homes and let everyone else keep on going with their lives and economy stays up, and the hospitals won’t be overwhelmed because 80% of the ICU cases will be captured in the quarantine.

      What we don’t need to do is shut down the economy and quarantine everybody.

      There’s smart and then there is excessive. Shut down the economy and the poorest will not survive. This is certainly burning down the house to kill a big spider.

  • There are 164 drug overdose deaths each day in the US, 131 suicides. With no end in sight. Do we do anything about this?

    • yeah we actually do quite a bit, same goes for heart disease, etc. without our efforts, the rates would be much higher. want to see what happens when a population does little, look at TB and HIV.

    • To the best of my knowledge, there is no risk of drug overdose deaths or suicides doubling in number every three days.

    • “To the best of my knowledge, there is no risk of drug overdose deaths or suicides doubling in number every three days.”

      Shut down the economy for months, and it will be worse than that.

Comments are closed.