Skip to Main Content

Banning gatherings, having people stay at home, closing schools, and other measures to reduce the spread of Covid-19 are working so well that an influential model of the pandemic’s course now projects that the number of U.S deaths from the new coronavirus by early August will be 60,415 rather than the minimum of 100,000 that it forecast last month.

The new, lower estimate is prompting officials to be cautiously optimistic about controlling the outbreak, possibly by early summer. But it comes with a mystery: Which social distancing has reduced transmission the most?

Figuring that out will be crucial as policymakers look for ways to lock in the progress against Covid-19 while eventually re-opening schools and the economy, an imperative that has become even greater in light of the more than 17 million people who have lost their jobs due to the epidemic.


The earlier projection of Covid-19 deaths — 100,000 — from modelers at the Institute for Health Metrics and Evaluation at the University of Washington assumed there would be broad adoption of “social distancing,” which is also how China broke the back of the epidemic that began in Wuhan. But the effect of that social distancing has been greater than the IHME team expected.

“This should be entirely obvious,” epidemiologist Trevor Bedford of UW and the Fred Hutchinson Cancer Research Center tweeted on Wednesday night. “But we now have strong evidence that social distancing results in decreased Covid-19 transmission rates.”


In epidemiologists’ terms, the number of cases caused by each infected person, called the effective reproduction number (Re), has fallen precipitously. If people’s contacts drop by a certain amount, then so does the the reproduction number, said epidemiologist Gerardo Chowell of Georgia State University. The new IHME projection, he said, “looks reasonable.” As additional data become available, the model “can provide a better view of the trajectory of the epidemic in the near future.”

In the U.S., Re has fallen most clearly in Seattle’s King County. There, researchers at the Institute for Disease Modeling calculated, it fell from 2.7 (and possibly as high as 3.5) in late February, when community transmission was accelerating, to roughly 1.4 or even lower on March 18. The county would have had roughly triple the number of cases in late March without social distancing, they found.

In 11 European countries, modelers at Imperial College London estimated in late March, Re has fallen from 3.87 to 1.43, averting 59,000 deaths, and possibly 120,000, just through the end of March.

The new Re in Europe as well as King County, and probably elsewhere, is encouragingly close to 1, the level at which the outbreak will burn itself out.

Like other countries, the U.S. has taken an everything-but-the-kitchen-sink approach to social distancing: close schools and ban many international arrivals and close restaurants and ban church services (and other events that bring people together, even funerals) and order those who can to work from home. As a result, officials can’t tell which measures have been the most effective.

It is also difficult to “decouple” the effect of those government policies from any voluntary measures people took to protect themselves, health economist Rahi Abouk of William Paterson University and computational social scientist Babak Heydari of Northeastern University argue in a paper posted to a social science site on Thursday.

The reason it is so important to determine which interventions lower people’s contact rate the most, Heydari said, is that “policies in many states are still evolving.” In those that have laxer policies but may be about to experience a surge in cases, he said, that understanding “can help them update their policies. [And] it is likely that some states, or the country as a whole, will experience subsequent outbreak waves, after they phase out some of the [current] policies. These results can inform reinstating” social distancing measures.

For instance, projections released this week by researchers at Columbia University identify U.S. counties at risk of seeing their health care systems overwhelmed by Covid-19 cases. The researchers ran models in which they assumed that people’s contacts dropped by 20%, 30%, or 40%. The latter could avert as many as 185,000 deaths in the Northeast and 33,000 deaths in the Midwest. But the model is agnostic on what mix of business closings, work-at-home policies, and other measures achieve any given level of social distancing.

So is the IHME model. That makes it difficult for policymakers to tell whether relaxing one form of social distancing while tightening others might bring the same benefits, at perhaps a lower economic cost, or to know which forms of social distancing can be eased first in an effort to tiptoe toward normalcy, Chowell said.

Results from preliminary studies are starting to provide some answers.

In research posted to the preprint site medRxiv this week, Chowell and his colleagues show that “the pandemic may be readily controllable through a combination of testing, treatment if necessary, and self-isolation after testing positive of symptomatic individuals together with social protection” such as face-mask use and hand-washing.

When the basic reproduction number is as high as 2.4, which King County and some areas in the U.S. are likely well below, 65% effective social distancing (that is, people have only 35% of their previous number of close contacts) brings Re below 1, the magic “this is ending” number. But social distancing that’s only 20% effective (people have 80% of their usual close contacts) also brings Re below 1 if 75% of infected and symptomatic people are identified and isolated within 12 hours of symptom onset.

“Sustained universal mask-wearing and frequent hand-washing have the potential to dramatically reduce the transmission rate of the virus after the first wave,” Chowell said. They “have the potential to help us safely return to work and school,” though he’s in favor of “remote learning and telework as much as possible.”

  • Yankee2,

    Look at the recent California numbers and trend. Whatever Governor Newsome enacted in February is working nicely here in the most populous sate in the US. He also resolved PPE supply issues. Additionally, there are several mobile testing stations (one in my neighborhood) up and running for a few weeks now.

    Gavin for President in 2024!

  • Yes, it would be nice to know which strategies are most effective, but that doesn’t mean we should give up on any of them, provided they all have positive effects. Until this virus is solidly under control (far from easy), we should keep ALL of the methods in place.

  • As a layman, it looks to me like we need to find or invent protective gear that brings the chance of infection down to some very low level, even if everyone went back to work, produce enough of it for everyone, distribute while training everyone how to use it, mandate proper use, and then, finally, we could safely go back to work.
    It is not clear to me that N95 masks are sufficient. First off, the masks are rated to remove particles much larger than the virus alone and only remove 95% of those – and, from what I can gather, in New York and hard hit areas, the medical folks who had and used N95s often got infected anyway.
    It seems like some system combining a helmet which makes a nearly airtight fit over the head with a backpack filter system, battery powered, which pumps extremely filtered air into the helmet would work. This is just a guess, I am pretty sure very few people will have the privilege of using or owning such a thing, most of us will be making due with the N95s.

    Obviously, it would cost at least many hundreds, maybe a few thousand for such a device – and the ongoing bill to keep them all running – but that has to be a very small expense compared to the current cost of closing the economy.

    • Peoole infected not becauae of N95 masks inefficiency, but bwcause thwy don’t know how to use them properly. N99 masks filter out more, but it is very hard to breathe in them. Closed air systwms are for labs, not for everyday use. Do you see yourself going shopping with 40lbs of respirator gear on your back?

    • I very much doubt it would have to weight 40 lbs. I have a (granted it is AC not DC) CPAP machine which weighs maybe 6 lbs at most – a battery and a DC motor, and a bunch of filtering media, and I think the entire device would be well under 20 lbs – and YES, I see myself going shopping with a 20 lb weight on my back if it means I CAN go shopping very safely. For that matter, most men 45 to 65 (in the moderately high risk category ) can carry 40# on their backs for half and hour without too much strain- a PROPER backpack, that is – you need to know how to properly load and attach your backpack but that is not really too much, on level ground, walking through a store – could I do a strenuous physical job with that?
      I am not trying to be argumentative, but can you prove the doctors and nurses getting sick do not know how to use the N95 masks? I know there is a right way to use them but I do not assume they are perfect if used correctly.
      I would lastly point out, If you are right that N95 are being used wrongly and that is leading to infections, that is not an argument AGAINST developing a wearable helmet/backpack device, that is an argument FOR such a device, if it can be made more foolproof.
      Lastly I would point out

  • Keep on your best assessment on how to identify the most and best control measures to control covid-19 pandemic.
    Good works with great results better endings.

  • We don’t know if it was social distancing, regardless of type. California implemented social distancing only 2 days before NY, yet cases (13%) and deaths (12%) are a fraction of NY’s. California was site of highest international US travel w China during Dec/ January. Perhaps some degree of herd immunity was developed due to high degree of interaction of Californians with these travelers. Only antibody testing will tell.

    • Dr. Edward Geehr I presume?

      “Perhaps some degree of herd immunity was developed due to high degree of interaction of Californians with these travelers.” Seriously?

      Are you an infectious disease immunologist? To develop sufficient antibody in merely two months? Wow Californians must have super-duper immune response better and stronger than any other population in the World!! News to me.

    • NY numbers are skewed by NYC, which is heavily reliant on the subway system. Having lived in NYC for a number of years before moving, I am sure that played a huge role in transmission when compared to California.

    • Hey John Chu, You don’t have to be so sarcastic to Edward. These forums sometimes get so nasty for no reason.

    • The Bay Area issued a shelter in place order at least a week before NY, also well before Los Angeles. With exponential growth, a week is a long time and can have a huge impact.

    • And John, about transit and cars, transit probably also made some of the difference between CA and NY, but SF also uses a fair amount of transit. And the difference between SF and LA certainly can’t be explained by public transit.

  • I am not sure that your convulsions match the modeling. The IHME model assumes full social distancing from the outset. Not only are we not social distancing consistently across the country NYC and NJ didn’t close public parks until this week. The R assumptions are based on and assumed R to start that has not been validated.

    The initial forecast of 100-200,000 now reduced to 60,000 cannot reasonably be justified by “social distancing” which didn’t change. In reality the 60,000 number falls well within the confidence level of the model. The upper range was the 95th percentile which is a “will not exceed number”

    The real issue with all of these models is that they are invalidated, have huge margin of errors and are being touted to us as being able to “predict the number” and then policy makers are pulling a fast one by saying the recent activity is driving the numbers lower when they know that is simply a bastardization of the data.

    The IMHE state model show the exact same reduction curve regardless of state social distancing rules. That on its face shows that they are all brand crude assumptions, not legitimate epi models based on solid data and controlled.

    That doesn’t mean social distancing is not working but the Experts are hiding behind these very gross numbers that were obviously on the worst case end and now claiming credit when they actually fall closer to the 50th percentile which is essentially the toss of the coin number.

    The Justification for the economic destruction …. which has been modeled and controlled … might be no where with the long term damage … I have no idea.. but we are making enormous long term policy decisions that we are abounding a calamity that may or may not have been justified.

    We should not be making massive 5 trillion dollar decisions justified by scientists with no policy backgrounds and immune from any blowback.

    This is not the flu … don’t get me wrong … but it is not Ebola either. We cannot as a society abdicate policy matters to such a narrow minded group.

    As an analogy we did not let ministers decide that the D day invasion was too great a price to pay because of the immediate loss of life. Someone with a big policy picture had to make the cost benefit analysis and then face the bitter pill of the decision.

    The NIH, CDC and modelers are driven by a singular mission and Thad good . They are not policy makers and models should not make policy and the reverse engineer victory.

    • In the words of Dr. Sanjay Gupta, “All model are wrong, but some are useful.”

      Input data is grossly insufficient and determining the relative importance of contributing variables is a work in progress. Models have short-term value and are viewed with the assumption (and hope) that they will soon be replaced with a better model.

      With each day that passes we are better informed, but we are still far from definitive solutions.

  • Its too early to designate social distancing a success. This virus will ultimately infect 2/3 of the populace. Social distancing is just delaying this inevitable result. Hopefully existing or future medicines will cure the sickness, or vaccinate the masses. But that’s not going to happen in less than a year.

    • Gary Hudson,

      Are you an epidemiologist or public heath expert? 2/3 of the US population is what? >200 million? Really? Inevitable? In how many years? I will keep track and see when would it hit 100 million. I may die of old age before then!

  • Ann,

    Johns Hopkins says 184 countries and regions. Where did you get the number 200? Is your source more credible? Please provide evidence!

    • First chance to respond to you directly John Chu. Yes, you presumed correctly about my degree. And spent several years working for Dept of Public Health in SF. Your confidence in epidemiologists is admirable, particularly considering their COVID-19 models have missed actual by orders of magnitude.
      So when we do test for antibodies in CA in the next month or so, what is your estimate of prevalence? Certainly, you have an opinion. Under 10%?

Comments are closed.