Thursday, April 16, 2020

Ready to reopen?

Are we ready to reopen? No but not for the usual reason.

Once we have testing we can reopen, says conventional wisdom. I doubt that. Who is going to give these tests, and what are they going to do with the information? 

Maybe maybe maybe if we had a free test, with instant results, that every American (and person flying in on a plane) could take every day; if almost all Americans were actually willing to take said test; and if people were universally willing to quarantine themselves upon its results, the test might help. But even that's a pipe dream. 

We won't have a vaccine, applied to the entire world's population, for a long time. "Herd immunity" seems unlikely. It's not even clear that exposure to COVID-19 confers immunity. (I've been looking for any study of how often people who had it once get it again. Let me know if you see one.) The whole point has been to bend the curve so that the vast majority don't get it. 

So, we will have through summer and fall, a pretty susceptible population and a virus ready to break out any chance it gets to. 

The point of "testing" and "reopening" is to have a system whereby public health interventions take the place of draconian economic shutdown to keep the reproduction rate under one. 

Public health does not mean just lots of tests. It means using the tests to identify small outbreaks and keep them from getting big. And that requires a tooled up, effective, nimble, local, public health bureaucracy. And a bureaucracy that steps on a lot of toes. 

Suppose your neighbor gets one of these tests and is positive. What gets done about that? First of all, under HIPPAA, his or her test result is nobody else's business, not even local government. Who forces him or her to self-quarantine? Who forces them to get the test if they know forced quarantine is coming? 

In Asia, apps tell you who got tested and the result. You can judge if you had contact. Or state  surveillance tells you that tracking your cell phone and theirs the authorities know you had dinner together last week and you're being isolated now. We, properly, have big laws against all of this! Are we willing to do this? How fast? Will we faintly have the capacity to do it? In a month? 

Maybe after weeks of wrangling, local public health officials (which local public health officials?) can use the information to decree there has been a positive test on your block and impose a quarantine on the whole block. The heck with that, say you, proud American. Like me, you've been strenuously distancing for a month, so you know you haven't got it from the neighbor, you're off to the newly opened park for a jog. Obviously, such edicts will have to be enforced, against a restless and resentful populace. Can you really see cops cordoning off apartment houses, blocks, towns, controlling who goes in and out? Do we even have cops to do it? And the method to figure out who goes where?

"Testing" and "tracing" are popular. Do most Americans (and our 10 million undocumented residents!) really want to tell someone from the government every person they've met in the last two weeks? Knowing the government is likely to quarantine them and ask more questions?  Or let them track your phone? Or carry your phone the minute you know they're tracking it? 

Birthday parties are just as bad as bars. Are we going to allow health authorities to monitor our cellphones and bust up birthday parties? 

A key part of public health is to isolate known areas. Are Americans going to put up with travel bans?  Are there any public authorities with the competence to put in place data-driven nuanced travel bans? Again, for the vast majority of people, the travel ban will be a senseless annoyance. 

We will need a robust public health response, to keep a small number of cases from ballooning, and allow the economy to open. We will need the response we should have had in January. "Testing" is one of many inputs to that response. But "testing" is not the response itself. An effective public health response needs a detailed, competent bureaucracy, temporary relief from thousands of privacy regulations -- and swift assurance that those privacies are reinstated when it's over -- and enforcement in order to something useful with the tests. I doubt Americans will put up with the enforcement. I doubt our government has the capacity to put them to that test. I hope I'm wrong as the alternative is waves of lockdown.  


  1. Prof. There is excitement out of the UChicago about Gilead drug Trials. How don these therapies change the equation?

    1. DNewman: I'm not anyone special, but in my opinion, an actual treatment is almost as good as a vaccine. As long as it, too, is widely available. You go out, you come in contact with your neighbor who is positive. The gods of random numbers stack up against you and you get sick (you've infected 2-3 others in the meantime). But you can take the drugs, isolate for a couple of weeks, and recover, as opposed to "go to the ICU and die". Seems like a VERY different equation, to me. Of course, we still have to deal with the exponentially increasing exposure (the 2-3 you infected, which turns into 4-9 they infected, and then 8-27 etc.) But if only 20% need drugs, it might be doable.

    2. The article states, "The whole point has been to bend the curve so that the vast majority don't get it." That's not quite correct. The curve is bent so that the hospitals don't face a melt down situation. There is good news from the Stanford study in Santa Clara which finds that the viral incidence is up to 80 times greater than has been assumed. This means, if you do the calculation, that the novel coronavirus is not more deadly than the flu - it only works faster resulting in hospital melt down in highly interactive places like NYC.
      For most of the US the moral is that getting back to normal ASAP is the right thing to do.

  2. John - I don't disagree with your comments. But lockdown or no lockdown, at some point everyone will be exposed to this - you well-understand this "geometric thing". Testing or no testing, all you need is one person who sneaks in, has a bad test result...

    Unfortunately it's hard to get actual facts. The numbers of cases, hospitalizations, etc. in NYC were dropping very quickly. Yet they just decided to add people who had died, seemed as if it might be CV, but weren't tested, to the list. Here are the links

    Also, while I don't mean to minimize the pain and tragedy of any of this, according to NYC data, of the 6200 people who'd so far passed away, less than 150 did not have an pre-existing condition.

    I think the bottom line is that the economic damage from keeping the economy shut down is way worse than allowing our citizens to understand the facts and decide for themselves. Anyone who feels concerned for any reason should qualify for government programs. But if we delay restarting things, we may not be able to.

    1. I agree as well. The goal at this point should be to delay what infections we can at relatively low cost. That means eliminating most govt controls and letting people make their own judgments about how to do that. The delay gives us a shot a getting lucky that we find therapies or vaccines that work, but that is not assured.

      I think this approach will dictated by some new knowledge we have gained over the past month:

      (1) Lockdowns (at least in the West) don't appear that effective. At best they push R down slightly below 1 and come at great cost. That is the experience of Europe. The great cost is evident in the drastic decline in US employment and the growing political pressure to lift them.

      (2) Distancing doesn't require lockdowns. Data pre-lockdown in most areas show that people were already doing significant distancing at the expense of things like conventions and restaurant reservations. This made the marginal effect of lockdowns small, and means that lifting them won't restore normal levels of economic activity.

      (3) SARS CoV2 is significantly more infectious than we thought. This makes attempts to stamp out the virus likely to fail even when done well (see Singapore). Even were one successful, you would have to isolate your country from the rest of the world (as China is now trying to do) which imposes significant costs itself.

      (4) SARS CoV2 is significantly less deadly than we thought-- especially to healthy young people. The IFR is probably between 0.3% and 0.6% overall and much lower for the healthy young. This means it makes sense for many to run the risk of infection and build toward herd immunity. It also means it makes more sense to concentrate efforts on protecting the vulnerable (even if that is a lot of people).

      (5) The dynamics of building toward herd immunity mean that as more people recover from infection and become immune, R falls because there a fewer people to infect. As the number of recovered people increases, the safer it is for the vulnerable. This is a reason to flatten the curve of infection (ideally with people that recover) rather than to bend it down.

      I think things will be better the sooner that government imposed lockdowns end. As in other cases where you eliminate harsh regulation, people will adapt in innovative ways to these new circumstances in ways that make their lives better.

      BUT, I think that absent finding a great vaccine or great therapies, the facts outlined above mean that you probably have 100,000-500,000 Americans dying and significant losses to the welfare of the living as they adapt to a distanced world.

      I wouldn't end lockdowns because I think that will make things great and return to normal. I would end lcokdowns because they make a bad situation worse.

    2. DWAnderson, It is so great to see that at least one person has the knowledge and ability to express it with regard to the current situation. If only you were advising the President and also editing the MSM so that the news would be as intelligent. This comment is not sarcasm.

  3. "Bend the curve so that the vast majority don't get it."

    Are you nuts?!?

    "Flattening the curve" (a meaningless catchphrase if there ever was one) was about reducing the peak number of cases and related demands on hospitals and infrastructure. It was NEVER about preventing the vast majority of people from infection.

    Nor is such prevention desirable for a virus that is less dangerous to the general population than the seasonal flu (because the flu, after all, kills kids).

    Furthermore, as an economist you should know better than to attribute the lower-than-expected death rate to the imposition of tyrannical social controls. It's not just that correlation does not equal causation; it's also that the restrictions (that force the vast majority of people into identical - and identically unhealthy - patterns and routines) may themselves be a significant contributor to the death rate.

    1. Totally agree. There are two monsters that are destroying us. Monster #1 is the Virus. Monster #2 is Quarantine & shutdown. People are losing jobs, savings, homes, health insurance plus having life saving surgeries postponed. We need to attack BOTH monsters.

  4. "Jeong Eun-kyeong, director of the KCDC, said the virus may have been 'reactivated' rather than the patients being re-infected." See:

  5. There are two kinds of testing. One is to see if someone HAS the virus. That test eventually turns negative. The second, antibody testing, tells you if someone HAD the virus.

    The sensitivity (the ability to detect infection) and specificity (the rate of false positives) is currently unknown and since different test kits are used, the water is pretty muddy.

    In other coronavirus variants the antibody levels drop off to zero after a few years. Oh happy day. Did you know that there are on average 7 cases of plague in the US every year and about 3/4 is bubonic?

    Contact tracing is definitely a HIPAA problem, but also a feasibility problem. Numerous companies - including the usual gang of privacy violators like Google and Facebook- are working on contact tracing but there are some technical problems, such as reporting that someone who's positive was 5 feet away from you, but not the fact that they were in the car next to you at a red light.

    I like the way Jim Rickards put it when he said it was the result of two complex systems colliding. I was just getting used to the futility of understanding one complex system.

    Meanwhile I went to the UPS store to get a document notarized and they would only accept cash. It was $6 and I only had a twenty so now I'm trying to find out how to sterilize cash. Yes, I've been reduced to money laundering.

    At least I'm learning a new job skill that could lead to a cushy job at certain banks.

    1. so the notary trusted you enough to sign their name to prove honesty, but not enough to cash your check? ironic.

  6. The lockdowns might save an unknown number of lives or might not save any lives other than for a few months.

    The lockdowns are economic suicide.

    Yes we are talking about lives, but is there not anything like a cost-benefit analysis here?

    The majority of people exposed to C19, including nearly all children, are asymptomatic and immune. There are some cases of relapse, as there are with cases of ordinary flu.

    I am baffled by America's macroeconomic community. You can get 1000 economists to sign a public letter declaring that President Trump's trade tariffs, on a single nation, could well trigger a global Great Depression.

    But when our government implements a depression... the US macroeconomic community goes mute. Even after initial estimates, horrific in their C19 death tolls, fail to materialize.

  7. Again, why not seriously consider risk-based approaches for achieving herd immunity? Vaccines and herd immunity are the same biologically. So if you are really skeptical about herd immunity, then this never ends.

  8. How about we trust people to do the right thing?
    People are usually pretty good at knowing what's best for them - vulnerable people will keep isolating no matter what the government says, and there's no point in keeping inside people with much lower risk.
    The authorities increased hospital capacity and equipment availability; maybe they should also 'ask' companies to let employees work from home for as long as they need to be comfortable. Other than that, further government action most likely won't affect the long-term health outcomes, and most certainly will damage the economic and social ones.

    1. People are good at knowing what is best for them. They aren't nearly as good at acting in the best interest of the community. There is an enormous positive externality to staying home right now, and it isn't folded into the price of doing so.

      So you for it by regulation, and subsidize it with massive government spending. It is a reasonable short-term approach.

    2. Dear Anonymous,
      If you actually follow the science of epidemics, epidemiology, you will find that what the government has done is not a reasonable short term approach. To make a long story short, the lock down should only be applied locally where there is a very great risk of hospital over crowding. The lock down does not in fact save lives for the most part except in that situation. Given that it does not save lives, it has no benefit whatsoever except with regard to the hospitals. For no benefit at all, millions are out of work and trillions of dollars have been wasted.

  9. I am surprised to find myself defending the lockdowns, but here goes:

    1. The lockdowns were intended as much to preserve the hospital system as to save lives. We were looking at the Italian hospital collapse as what was coming for us.

    2. Unemployment was going to be a problem with or without a legal lockdown. The restaurant, bar, hotel and casino industry employ over 15 million persons, and contraction was well under way by the time that lockdowns were imposed.

    1. Indeed - Days before the announced lockdown in Illinois, restaurants were empty on a Friday evening, just out of fear. That's may change somewhat, but not entirely.

    2. Bob,
      You are correct on point 1. except that the lockdowns are only needed locally where there is a risk to the ability of hospitals to care effectively for the sick. With regard to your point 2, although restaurants and bars may have been affected anyway what about all of the other business which could have gone on relatively normally including: all other retail, medical, office work, banking, etc.etc.etc.

  10. People have fought against pubic health actions to protect against infectious disease since John Snow took away the handle of the Broad street pump in London in 1854 to stop a Cholera epidemic. People didn't like it or understand then and they are almost as stupid or ignorant today. That is why we still have local public health laws today to compel them to do so. Many of the comments here are testament to that. Yes sometimes there are reason we needs laws and a nanny state approach - to get people to wear seat belts and helmet laws and people still reject them despite clear logic and effectiveness. Is it perfect? No, but to some extend they are necessary. Lives are at stake and sometimes we have to weigh the importance of the outcomes. Sometimes we must wait to figure it out with better information. Luckily we can vote to change our politicians and laws - but that comes later. Hopefully we will have the wisdom to decide later what mistakes were made and correct them. Until then we must work to figure it all out and that takes time and the right people rather than a rush to judgement. Sorry, just the facts.

  11. Isn't it somehow possible to get people to realize that quarantining 2% of the population is less abusive of human rights than quarantining 80%? It's simply crazy how people accept our current massive lockdown, but object to the age-old idea of quarantining just sick people and those exposed to them.

  12. You make a valid point - a "post-epidemic" scenario that prevents a major resurgence of CV-19 active cases, will require a level of self-constrained human/social behavior that Americans are certainly not well known to exercise.

  13. I keep coming back to Romer’s simulations. As I understand it, his exercise assumed frequent random testing (7% of the population per day), isolation of those who tested positive but no contact tracing. He ran a horserace between that policy and one of random isolation. (What he calls random isolation is, I think, more or less what we’re doing now.) He showed that for the random isolation policy to achieve the same path of infection we would need to isolate about 50% of the population. This was an uncalibrated simulation and so we can’t trust the precise results but it does suggest two things, both encouraging.

    1) Isolation without contact tracing can still work. Big if true since isolation is easier to enforce and less intrusive than isolation with contact tracing.
    2) Romer showed that in simulations with a very high false negative rate you could still get R0 below 1. The key is lots of tests, not accurate tests. That’s great because it suggests that even badly enforced isolation can still help. Under Romer’s simulations a false negative is equivalent to a non-compliant true positive (in both cases someone is walking around spreading the virus.)

  14. "We, properly, have big laws against all of this!"

    The USA having to remain closed-for-business while the rest of the world gets to enjoy life (basically) will force a lot of people to reevaluate that judgment.

  15. There are two problems with the current approach. The first is the logical fallacy that a positive test for the virus means the patient is positive for the virus, the patient is carrying live virus or infective virus and any symptoms are due to the virus. The virus is tested for by PCR. PCR detects live virus, dead virus and viral fragments. Patients may shed non-infective virus for days after clearing the live virus.. This problem is known in all areas of infectious disease when the pathogen is established by PCR, Therefore a positive test by PCR devoid of the clinical context is useless information. (Distinguishing live from dead virus is possible but requires extensive high level laboratory evaluation that is not readily available).

    We run into a similar problem with any laboratory testing, including alcohol breathalyzers. The validity of a positive test (a true positive) must be put into the context of a positive test (devoid of the disease; a false positive).

    As an example:
    Assume a test with a 5% false positive rate. What does that mean to the individual. If the prevalence of the disease is 1/1000 then a 5% false positive rate means that there are 50 patients identified with the disease who do not have it.

    This fact is used by lawyers to have the results of a breathalyzer excluded if the arresting officer cannot show good reason why he/she believes the patient had been drinking. The first question asked by a police officer: Have you been drinking?

    The second problem is an ethical/moral issues:
    If we begin from these premises:
    All lives are precious.
    All lives are equal.
    Then any action to preserve life A must be balanced against the likelihood of eliminating life B. Otherwise, we are valuing A more than B.


    Interesting idea

  17. I think the talking head's endless, misinformed and politicized coverage has confused the public at large. People simply don't understand that mitigation's goal is to slow the spread, not avoid the spread. The goal of mitigation was to avoid overwhelming hospitals due to a rapid spread, however most of the population will need to get infected and develop immunity in order to get past this pandemic. The only way to really avoid everyone from eventually "catching" this thing is vaccination, but that's unlikely to happen anytime soon. Forget all these notions about some vast, capable bureaucracy that will keep us from getting sick. Trying to stop this virus is like stopping the wind. We need to unwind mitigation in tranches, and people will simply get sick in as orderly a fashion as possible, hoping not to outstrip the healthcare system's ability to care for them. We can get through this, but not around it.

  18. The wonderful Dr Michael Osterholm of the U of Minnesota reminds us that there will be new waves of the virus in future months. We are not going to do another lockdown when we are still counting up the unemployed from this lockdown.

    Americans are used to spending infinite amounts of money to prevent anyone from dying. In normal times, this leads to $400,000 health insurance claims from a long ICU stay, a transplant, or a specialty drug. Analysts like me are troubled by these claims, but if we question them in public we are vilified as heartless Darwinists.

    The argument is going to get tougher in the near future. We will have to face the fact that saving every save-able life might cost us our job, not just a little higher insurance premium.

    Americans are used to feel that every death is an outrage. Our national skin might have to get a little thicker.

  19. Part I: I think we're finally still starting to see what a geologist, or ex semiconductor reliability guy like me would call preliminary "faulty" data, i.e., the 1st real data on critical "missing" COVID-19 parameters that the models have used to date (being a little blunt here) "wild guesses" for in the models, i.e., what are the real population infection (disease penetration), mortality, and transmission rates.

    Stanford's Dr. Bendavid's preliminary results this week on Silicon Valley infection rates via off the shelf commercial anti-body tests indicated local infection rates (in our one of the oldest US COVID clusters) might be as high as 2-4% of overall population, or 60-80,000 more than have locally (Santa Clara County) tested positive (just under 2000).

    Yesterday's WSJ OpEd by local PhD molecular biophysicist fund manager Aaron Bogan referenced Dr. Bendavid's preliminary data plus similar numbers from studies across Europe from Italy, Germany, Denmark, Iceland showing "true" infection rates 10-30X the "tested" infection rate. Even with huge error bars (alpha, beta, false risks on the Premier Biotech test used by Dr. Bendavid, or selection bias in those that wanted to take it, e.g., folks that think they had it might preferentially select to volunteer initially), these results are phenomenal in filling in the huge unknowns we've had all along on rate of true rates of "asymptomatics", true mortality rates, etc. that we've had no "evidenced" based clinical results for earlier.

    Even a 4:1 ration of undetected to detected (as a limited Chinese sample study indicated some weeks back) drives the true mortality rates below 1%, and 20X to 50X or more really makes the true mortality rates much more like those of say, annual influenza infestations,, ~0.15% range, where annual mortality ranges 30k-70k/yr; e.g. 2017-2018 was particularly bad, perhaps as many as 80k.

  20. Part II:

    Even a 4:1 ration of undetected to detected (as a limited Chinese sample study indicated some weeks back) drives the true mortality rates below 1%, and 20X to 50X or more really makes the true mortality rates much more like those of say, annual influenza infestations, ~0.15% range, where annual mortality ranges 30k-70k/yr; e.g. 2017-2018 was particularly bad, perhaps as many as 80k.

    This really makes me think we should start going after the idea of how fat and how long are the tails as John Cochran talked about a few weeks back. That it, once we know overall mortality rates are close to flu, the concerns are really those at risk of the severe cases, the elderly, the chronic diseased sub-populations (asthma, Diabetes II, CHF, COPID, obesity etc.)

    The next most important numbers then become disease sub category outcomes, i.e., "how many infections transition to moderate or severe disease", how many moderate cases transition to ICU, and then % to intubation care, how many intubated cases actually survive (pretty low number on the last one currently). Can we change the rates of these outcome categories of "the tail" with increasing, improving therapeutics?

    For example, anti-virals such as remdesivir are much more likely much sooner than full vaccines to be able to change the therapeutic statistics of "immune system pulmonary cascade response to local virus presence", i.e., transition point to moderate and severe and life threatening disease.

    While 2-4% population disease penetration rates really aren't yet getting to significant levels of herd immunity, there's another value for the previously infected. Now we can identify & use large populations of local "previously infected" folks who have had the disease with high counts of IgG antibodies, i.e., enabling recovered patient plasma therapeutics for hospitalized patients which might really start to impact "outcomes" rates on the tail as well.

    Focusing on the tails and going after them allows us to add more safety error margin to opening up strategies and tactics without risking or creating a new "super bloom" of disease, or risk overwhelming local health care resources.

    Besides these tactics to go after the most significant negative outcomes rates of the infected, obviously we still need to transition to effective but less severe forms of social distancing and transmission mitigation. I.e,., shaking hands is out, masks are sort of ubiquitous, & all similar related extensions, but we do open up again and get to 90-95% no longer "shuttered at home" sooner and with less risk, from current 90% shuttered at home rates, and yes it will all be different, maybe sooner than currently is thought by many.

    Also throw in full contact tracing, very large escalation and significant sub-population sample or full antibody testing, etc. And yes, finally, use the mobile GPS apps automated contact tracing to ID possible contacts for folks (like South Korea and others continue to do already for many weeks).

    In true libertarian sense, it may make perfect sense to give up privacy concerns short term to conquer COVID transmissions, then later scale back as we get closer to normal again. Even in the short term, there are already well know ways to anonymize large data sets, e.g. for HIPAA and other concerns (my start up does this), ahead of when we start to scale back.

    Appreciate this whole discussion and line of discourse the last several weeks, i.e., it is not economy vs life or death, black or white, but rather, how will we open this up again (which has finally gotten a lot of traction the last couple of weeks).



Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.