This has now been completely reorganized, and I’m more-or-less done with it. The upshot is, we need a mandatory mask law. That’s as much as we can do, at present, to reduce disease transmission in the public spaces. And if we have even a hint that transmission within businesses is an issue, we now need to start thinking about legal restrictions on the non-public areas of businesses, governments, and other organizations. Anywhere unrelated individuals share indoor space. Because any of those places could be vectors for spread of infection. We need to focus on them next. After we get a mandatory mask law in place.
Christopher Hogan, PhD., email@example.com
Virginia’s experience versus Wuhan’s experience.
Virginia, USA. Underlying data are from the Johns Hopkins coronavirus website. The graph is my analysis. And the title is wrong — that’s actually through yesterday 4/10/2020.
Key to actions taken in Virginia:
- A———- Schools closed
- B———- Restaurants, gyms, limited to no more than 10 persons at a time
- C———– Sit-down restaurants, theaters, gyms, beauty salons, etc — closed. All gatherings of more than 10 people prohibited. Adhere to social distancing recommendations “to the extent possible”
- D———– Same as C, plus mandatory social distancing, and (in theory) illegal to be in public places except for outdoor exercise and essential trips. Essential means trips to your job, church, or to essential businesses.
Essential businesses include:
- Grocery stores, pharmacies, dollar stores, and department stores (as long as they sell food or medicine)
- Medical, laboratory, and vision supply retailers;
- Electronic retailers that sell or service cell phones, computers, tablets, and other communications technology;
- Automotive parts, accessories, and tire retailers as well as automotive repair facilities;
- Home improvement, hardware, building material, and building supply retailers;
- Lawn and garden equipment retailers;
- Beer, wine, and liquor stores;
- Retail functions of gas stations and convenience stores;
- Retail located within healthcare facilities;
- Banks and other financial institutions with retail functions;
- Pet and feed stores;
- Printing and office supply stores; and
- Laundromats and dry cleaners.
- And child care facilities (addressed separately).
Retail businesses now fall into three classes. The ones above are categorically open. There’s a separate list of those that are categorically closed (e.g., beauty parlors, recreation facilities, movie theaters). And then everything else is limited to 10 customers at a time. But the Governor’s last order (3/30/2020) may have implicitly closed those, because, in theory, it is now illegal to travel to get to or from those.
Non-retail businesses are not covered by any of this. So in addition to this list, anything that can be done that does not require any retail aspect (literally customers coming into the business) is open. I am uncertain as to whether or not this applies to wholesalers, for example, a wholesale plumbing-supply facility or similar, where customers enter the facility, but … it’s doesn’t meet the plain-language definition of “retail”.
Finally, I scanned the executive orders, and did not find that the Governor had required local governments to close public facilities or not. Ours have.
All of that information on Virginia is from the Governor’s executive orders.
Wuhan, China. Underlying graph taken from: Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in ChinaSummary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention Zunyou Wu, MD, PhD1; Jennifer M. McGoogan, PhD, JAMA. 2020;323(13):1239-1242. doi:10.1001/jama.2020.2648
Daily increase in cases, China:
Blue = onset of symptoms, tan = return of test results (i.e., case officially counted as coronavirus).
Key to actions taken in Wuhan
- A: Wuhan and adjacent cities were locked down.
- Z: Daily increase in known coronavirus cases peaked.
The 12 days is due to the lag between infection and symptoms (about 4 or 5 days), and the lag between onset of symptoms and reporting of the test results (about a week). That’s also why the blue bars have a flat spot at A — those are largely people who were already infected on the lockdown date, but only began to feel the symptoms an average of 5 days later.
And that’s why I said, nearly a month ago, that if closing the schools worked as a way to contain coronavirus, then we’d have expected to see the turnaround around the start of the fourth week in March. Others, at that time, were saying end of March. We are now rapidly approaching the middle of April, and we still have not reached the inflection point — the point where the count of new cases, each day, begins to fall.
So, by the Wuhan metric, Virginia’s steps A, B and C didn’t quite cut it. And, really, in terms of impact on the average person, step D is just step C, rephrased for emphasis. So if C didn’t cut it, I’m skeptical that D will.
The point of this paper is to suggest that our current “soft” lockdown doesn’t appear to be stopping the spread of coronavirus. Not if you look carefully, and compare to an example of lockdown that worked. We’re slowing it, sure. But not stopping it. Or, really, we’re not stopping it fast enough.
The doubling time for cases, at the peak of spread was about three days. We had 25%/day growth in cases, which translates to doubling the cases every three days. Right now our case growth in Virginia is down to about 10%/day, which means cases double every week. But it doesn’t seem to be budging any lower than that in Fairfax County. And if it’s going to get materially lower than that in Virginia, it’s going to take quite some time to get there.
What’s the problem? If we continue at that rate, it’s just a matter of time before Virginia hospitals run out of ventilators.
Today, per the Virginia Hospital and Healthcare Association, we currently have 283 ventilator-dependent COVID-19 cases in Virginia. There remain about 2150 ventilators on-hand in Virginia hospitals and not currently in use (same source). Doing some crude arithmetic, at the current one-week doubling time, we’ve got about three weeks to get this under better control.
OK, I guess I had better say this, even though I would rather not. That three week figure? That’s at the end of the pipeline, where the pipeline starts with infection, and ends with people on a respirator. That pipeline is arguably 10 days long — median five days from infection to symptoms, and then (purely a guess) another five for the typical case that will need a ventilator to get on the ventilator. So if the SHTF day appears to be three calendar weeks from now, at current growth rates, that means we have about a week and a half to make some material reduction in the spread of the infection in the population. Or hope that such a reduction simply appears spontaneously.
It’s not really as doom-and-gloomy as that, because physicians in New York have figured out expedient ways to extend the respirator supply, such as splitting one respirator between two patients. That’s truly inferior medical care. And with a one-week doubling time? That buys you one more week, if the growth rate remains unchecked. It would be better if we could avoid that.
The Virginia experience: Case growth rates in Virginia and Fairfax County
Wuhan managed to reach the inflection point — the point where the daily increase in new cases begins to decline — after just 12 days of lockdown.
By contrast, if I now look at case growth rates in Virginia and Fairfax — defined as each day’s new cases, divided by total cases so far — for Virginia, that’s been slowly declining. But it is currently about 10%/day. For Fairfax, in particular, that shows no trend toward falling lower than 10%/day.
Either way, there’s no indication that case growth is stopping any time soon, from the measures taken so far.
A daily 10% growth rate means that cases double every week. This is better than it was at the outset, where the doubling time was about every three days, for the US, for Virginia, and for Fairfax County. (Above, where the solid line is at 25%, that’s a three-day doubling time.)
And if they double ever week, three doublings ought to get us near the point where every free ventilator in the Commonwealth is occupied. Per the rough calculation in the first part of this paper.
So it’s not as if the current attempt has done nothing. Plausibly, it took us from having cases double every three days, to having cases now double every seven days. That is, I guess, what flattening the curve is all about.
But it may not yet be good enough to avoid a crisis here in the Commonwealth.
What else can we reasonably do, short of a full lockdown? Mandatory mask ordinance.
Right now, whether or not we can end the current shutdown in any reasonable length of time hinges on whether or not masking up in public, on top of step D, does stop it. That’s the only material change since Point D. And if it doesn’t? Well, you tell me what’s next. Either we do things that are even more invasive. Or we wait this out for a very long time.
The rest of this material was moved to Post #622.
My main point is that we may have to go full Wuhan on this. Total lockdown. Or we may have to live with this for a vastly longer period than Wuhan did. And if those are the realistic options, we need to starting thinking of the economic consequences accordingly. We may be living in a fool’s paradise if we are expecting Wuhan-like results without paying a Wuhan-like price for it.
So at the minimum, we have to take the tools we have in place now, and make them work as well as they can. The CDC finally got around to recommending masks in public. It’s time to make that mandatory.
Background: Key Questions to Ask.
The Town of Vienna, VA is going to start public hearings on the tax and water/sewer rates Monday 4/13/2020.
That got me thinking about whether the Town’s budgetary response to COVID-19 — which I characterize as “business as usual” — is warranted. Near as I can tell, they seem to be working under the assumption that all will be well. That if there is a revenue shortfall, they can cover it out of reserves. And so, they haven’t even bothered to postpone discretionary items. For example, it’s full-speed-ahead on the quarter-million-dollar contract to rewrite all the zoning laws in Vienna.
They aren’t alone in taking that approach. I’d say Fairfax County is doing roughly the same. Pay freeze, hiring freeze, and that’s about it. No plan for any major retrenchment, as of yet.
But to know whether “business as usual” is a prudent budget strategy or not, you need some guess as to how this pandemic is going to play out. Locally, at least. If not nationally. I think that any reasonable economic projection has to start with that. (Or, turning that on its head, “business as usual” implies that you’re assuming a relatively modest course of the pandemic, at least locally.)
Peeling that back one more layer, the obvious first question is: Is the current shutdown working? And just to cut to the chase, I think I’m going to have to say no, not really. Halfway measures locally and in the Commonwealth appear to be producing halfway containment, so far. Slowed the growth, but not stopped the growth.
And so, as I see it, everything right now is riding on the effects of masking up when in public. Because as far as I can tell from the data, shown above, everything done up to that point has merely reduced the exponent on exponential case growth. We haven’t contained it up to that point. All we have done is managed to slow the growth some. And the 4/3/2020 change in CDC guidance, to include wearing a mask in public, has resulted in the only material change in prevention behavior in the past few weeks.
Which is why I am calling for the Town to issue a mandatory public mask ordinance. This is both because I think that’s a proper public health measure, and because I’m going to have a supply of cheap single-use masks that could be used to “prompt” the people who don’t yet get it.
In the following sections of this paper, I’m going to ask a few more questions that I think are vital to understanding this, but for which there is absolutely zero public information. We need to start asking those questions, because if public masking doesn’t work, this lockdown is going to have to get tighter, or last a long time. Worse, lockdown is a blunt instrument, and it is not purely driven by public health concerns, but by economic concerns as well. So not only are we unsure where to target any further restrictions, in an information vacuum, they’ll get targeted based on economic criteria instead of public health criteria.
If the answer to “is this working” is “no”, the next obvious question to ask is: Why are people continuing to catch COVID-19? And is there anything more we can do about that?
Is this all within-family spread of already infected individuals (so-called secondary attack). Is this people picking it up at work? Are they picking it up as they shop? Is it a smattering of everything? Or, worst case, is it literally true that nobody can tell where people are picking this up now? If true, we’ll have to go from “shut it down” to “shut it all down” if we’re going to get to the other side of this in any reasonable amount of time.
Three key questions.
Will masking up in public change this materially? Because that’s about all we have left, short of a Wuhan-style shutdown. By eye, at least in the Town of Vienna, 85%-90% of people appear to be taking that seriously when in the public spaces. That was my best guess, from my trip to the grocery store yesterday.
Maybe next time I’ll stand outside and literally count faces.
But for businesses, governments, and other organization, near as I can tell, there is no law requiring mask use in non-public business spaces. The Governor’s last order applied to anyone in a space that is open to the public. So, you can (e.g.) still have a staff meeting, talking at one another, without masks. I’m also pretty sure that even the social distancing requirements are not mandatory within non-public business, government, and private spaces.
And so, a sub-question is, what fraction of all time spent indoors, around non-family members, remains non-masked time? Are there any places where individuals (other than family members) are sharing indoor spaces and not wearing masks? Particularly, where they do a lot of talking? Because if so, per my prior posts, that’s a transmission risk, no matter what the CDC does or does not say about it.
And as a second sub-question, is the current spread of disease primarily among the non-compliant? Are we looking at people who (e.g.) didn’t keep their social distancing, weren’t washing their hands, and so on? Can we simply explain it by sloppiness? If so, presumably, at some point we will exhaust our supply of stupid people, they’ll all have gotten it, the new infection rate will drop, and we can get on with our lives. By contrast, if these new infections are among careful and compliant individuals, then all bets are off.
Second, how much more will we have to do, if masking up in public doesn’t work? I’m not even going to answer that one, but instead will use it to emphasize that a mandatory mask law, broader than the current mandatory social distancing law, will be a heck of a lot less intrusive and destructive than the alternatives that may become necessary.
And, finally, how does this virus continue to spread? Because if we could do the epidemiology right now, to figure out typical situations that are leading to virus spread now, we could target interventions to that. For sure, the whole CDC story about symptomatic individuals coughing and sneezing in public is just baloney, at this point. It has to be pre-symptomatic individuals, and it has to be some route other than coughing and sneezing in public, because nobody is doing that now, and there is extreme social reinforcement of not appearing sick in public.
But if it’s not that, then what is it? My fear is that epidemiologists are interpreting absence of evidence for evidence of absence, for anything other than droplet transmission. I.e., maybe they think it’s short-range, close-contact “droplet” spread because that’s the only thing they can actually trace. Maybe their traditional methods can only find that type of transmission.
And so, I’m afraid that what’s being reported out is that they have identified cases where it was spread by close personal contact. And what is NOT being reported out is that they are also finding cases where close personal contact cannot plausibly explain it.
Think of it like a police investigation. If they only report the cause when they “solve” a case, and the only cause of infection that they can trace is close personal contact, then … what they are going to report is that this is all being spread by close personal contact. And they’re not even going to mention the cases that they haven’t solved, because they don’t know how those folks got it.
Now lest you think I am crazy, have a look at the table below. I can’t tell whether this just reflects an agency overwhelmed by case load (so “under investigation” largely means “haven’t gotten to it yet”), or whether this really means that they have looked a a bunch of those “under investigation” cases and in fact have found many for which close personal contact is not a reasonable explanation.
But note the numbers, and in particular, note the categories. They don’t even have a category for anything BUT transmission via close contact, for domestic cases of the disease. And note that about 96% of cases are “under investigation”. By eye, that almost looks like, when they identify close contact, they close a case, right? And so, as far as they know, all of this, that they know about, is transmitted by close contact? Please tell me that’s not what’s going on.
Source: CDC, at this link
So the huge uncertainty here is whether that last category is a lot of unexamined cases, or a lot of unexplained cases? That’s a pretty crucial distinction, and if that contains a large number of unexplained cases, that has huge implications for the next steps we need to take. Because, for starters, it means that droplet transmission from being in close contact with somebody probably isn’t the main transmission route.
So I now need to find some state agency that has a version of that table where they have, in fact, split out that last category into two pieces: We haven’t looked yet, versus we looked and close personal contact did not explain the infection.
Searching for some clues about community spread cases as a fraction of all cases.
(And the very first thing I find is that Virginia is using Tableau to put their data on the web. Tableau is the Roach Motel of data — data go it, but they never comes out. Ah ,they have a separate data download section outside of Tableau.)
The second thing I note is that the Virginia Department of Health data appear more updated than the Hopkins data. More or less, they are a day ahead on reporting, and the dates used to tag the data are off, by one day, relative to the Hopkins data.
The first odd fact is that if I calculate known infection rate per 100,000, what I get is not at all what I expected to see. We’ve all seen the data showing that death rates are highly skewed toward the elderly. But, aside from the oldest old, the infection rate is not. Presumably, below, these are all cases who were ill enough to seek medical attention, or were known to have been exposed by (say) a family member. So this reflects some combination of exposure rate, propensity to get infected if exposed, and then, propensity to develop some severity of symptoms once infected.
Source: Case counts are from the Virginia Department of Health. Population counts are Census 2018 civilian population projections by state and age.
The interesting thing to me is how flat that looks for adults, putting aside the oldest old (80+). All things considered, you’re almost as likely to get a reportable case of this if you’re a 30-something as if you’re a 70-something. I’m sure that (e.g.) hospitalization rates and death rates don’t look like this. Those are strongly skewed toward the elderly. But reportable infection rates are not. Unfortunately, I could not find Virginia numbers for either of those, by age.
And, again, putting aside the oldest old, the peak for adults is in the working-age population (50-59), not the retirees (70-79).
The second clue is that what I am looking for, initially, is the fraction of cases attributed to “community spread”. “Community spread means there is at least one case where we don’t know how someone contracted COVID-19. They didn’t have contact with someone who has tested positive or traveled to highly impacted area.” (Source)
But nobody tracks the fraction of cases that are thought to be community spread (i.e., where there’s no obvious source for the infection). The only thing you can find is news reports when each state or area got that dreaded first community-spread case. Like so:
Mid-March, “several” out of 17 total cases in Madison, WI were investigated and found to be community spread cases. (Source).
North Carolina, one case out of the first 100 was a community spread case. (Source).
Four out of the first 64 US cases were known community spread cases (Source).
North Dakota, 2 out of the first 7 were community spread (source).
One out of the first five in Long Beach, CA (source).
Several out of the first 45 cases in Virginia (source).
Four out of 13 in Denver (source).
One in the first 20 in LA (source).
One in the first three in Pasadena, CA (source)
I don’t think I can draw any conclusions from this. When the statistic is based on finding the first community-spread case, what you see is then subject to the randomness of the order in which the cases were discovered. From this sort of information, I don’t think you can even conclude whether or not community spread is (or was) common.
Time to hazard a guess: Why is the infection rate so high in the working-age population? Maybe they’re picking it up at work.
Let me repeat that graph of disease prevalence per 100,000, above. As noted, that combines a) amount of exposure, b) likelihood on being infected when exposed, and c) likelihood of having a case that is severe enough to warrant testing.
Here’s the funny thing. I’m certain that the prevalence of severe cases, upon infection, ramps up steeply with age. All the numbers indicating severity — hospitalizations per infected person, ICU stays per infected person, and deaths per infected person — all ramp up steeply.
I’m willing to be that the propensity to be infected, upon exposure, is at best level with age. Arguably, younger people with better immune systems and generally better physical condition ought to be able to shrug this off. So that, too, might plausibly rise with age. But, conservative, let’s assume it’s uniform for all ages.
And so, back-solving, the easiest way I can think of, to explain this graph, is if people are getting exposed to coronavirus at work. That is, a much higher exposure rate for the working-age population is more-than-offsetting the much higher propensity to get a bad case of this, if infected, among the retiree population.
That’s conditional on what I think I know about the other two pieces. If exposure were equal, the working-age population should have vastly lower rates of illness severe enough to warrant testing. Something is pushing up the infection rate in the working age population.
(Oh, and nothing suggests that this might be due to higher exposure to children. In the WHO report on the China experience, they could not find a single instance where children were the vector of infection for a family. Plus, the infection rate in kids is miniscule.)
So I’m guessing it’s work. We all either have to shop, or get our food delivered. Nobody’s going to church. Mask recommendations only apply in public spaces. Nobody goes out to a public space if they don’t have to. I hope. So that’s all more-or-less a wash.
But if you have to work, you have to work. And if exposure at work is the last untouched disease transmission channel, we need to focus on that. Because at this point, assuming a mandatory mask ordinance, that’s the last area we haven’t done anything with.
And I keep coming back to Boris Johnson, the Prime Minister of Great Britain. Nobody let a symptomatic individual anywhere near Boris Johnson in the past month. But I’ll bet he’s been in a lot of meetings, with a lot of people, in stuffy little rooms, over the past month. If aerosol spread is more common than has been thought, and there are a lot of pre-symptomatic individuals, spending a day in a small, poorly-ventilated office, with a pre-symptomatic individual seems similar to spending an hour or two in a hospital room, with a person who has symptoms (and so likely sheds virus at a higher rate). And for the hour in the hospital room, an N95 respirator is mandatory. But for working in an indoor setting, it may not be.
Does anyone have a clue what fraction of people in Fairfax County are still “going to work”, meaning, physically showing up at some workplace? And if so, are any businesses not making their employees wear masks while at work?
Five days ago, the DoD decided that anyone on DoD property has to wear a mask. (But this seems to apply only in cases where 6′ social distance can’t be maintained.) Presumably that means that, to that point, there were DoD employees working in the same space without masks.
So here’s the pitch. I think, with the addition of a mandatory mask law, we’ll have done as much as we can do to protect people from infection when they are in a public place. Near as I can tell, if we do that, the only other step would be outright shutdown and lockdown.
But I’m far from sure that we’ve done everything we can for workplaces. And if we could get any hint that transmission in workplaces remains a significant channel of disease spread, we’d have the rationale for (e.g.) imposing mandatory mask use on the non-public areas of businesses, governments, and other organizations where people are still “coming in to work” in person.
The groundswell for mandatory mask laws for essential businesses.
(Moved to a separate post).
Finally, this all links back to the WHO report on China. The WHO report said that almost all transmission was due to close proximity to symptomatic individuals. As I said in an earlier post, I am coming to suspect that the disease transmission section of the WHO report on China may have strongly reflected the fact that the Chinese population routinely masks up during epidemics. And that transmission routes in a largely non-masked US population might therefore be quite different. Masks not only stop large droplets, my assertion is that they reduce spread of aerosol-sized (5 micron or smaller) droplets by reducing the velocity of your breath as you breathe or talk. (Those small droplets have far too small a Reynolds number to act like projectiles — they are more-or-less stuck to the air mass that they are in.)
With a mask, you don’t spray somebody with aerosol just by talking to them. Without a mask, you do.
And the original pictures for “flattening the curve” don’t really tell the correct story about the long end-game of this. Those flatten-the-curve graphs may reflect what has gone on in Wuhan. But with our far-less-restrictive changes, they grossly misrepresent what we’re looking at, for the intermediate term.
Original graph source: NPR. Red X is not part of original graph.
February 24, 2020 Characteristics of and Important Lessons From the Coronavirus Disease 2019 (COVID-19) Outbreak in ChinaSummary of a Report of 72 314 Cases From the Chinese Center for Disease Control and Prevention Zunyou Wu, MD, PhD1; Jennifer M. McGoogan, PhD1 Author Affiliations Article Information JAMA. 2020;323(13):1239-1242. doi:10.1001/jama.2020.2648