Post #703: Virginia, please publish redacted case reports and crowd-source a key analysis.

Source:  Analysis of data from the Virginia Department of Health.  Areas of Virginia that entered Phase I of re-opening are in red.  Areas that did not — but will start tomorrow — are in Blue (NoVA, Richmond City, and Accomack County).

Aside:  I finally found out why rural Accomack County is in the late-reopening group.  They had more than 500 workers test positive in chicken processing plans owned by Perdue and Tysons, per this news story.  So they are the Virginia chapter of the national story on meat-processing facilities, and contributed to the recent small uptick in cases in Virginia.


Rule-outs

If re-opening has had any impact on new COVID-19 cases, I sure can’t see it.  Not in Virginia, above, where we have a nice, clean comparison of areas, and a single start date (5/15/2020) for a uniform re-opening policy (our Phase I re-opening).

Not yet, anyway.  But we’re reaching the point where we can no longer say that it’s too soon to tell.  The basic arithmetic says that there should be somewhere around a 10-12 day lag between a change in infection rate, and the point at which that change appears in the official numbers.  That’s the lag between infection and onset of symptoms (tightly clustered around 4-5 days), between onset of symptoms to seeking testing (typically reported at 3-4 days, but who really knows), and then lag between testing and reporting of results.

Aside:  I note that, in Virginia, the state guarantees overnight turnaround of results for tests submitted before noon, in the State’s lab.  (But there’s no way to be sure about the lag in private labs.)  Here’s what the Commonwealth says about testing in state labs:

DCLS performs same day testing for all samples received by noon each day. ... Positive and negative results are faxed to hospitals within 24 hours of test completion ...

My conclusion is that if there was any material uptick in infection rates in the early-reopening areas (6/15/2020), that should have started easing into the data above (ending 6/27/2020).  But just barely.

I feel fairly comfortable presenting those results, even with the scant timing, because a similar analysis of national re-openings gives the same message.  That’s a far sloppier analysis, but contrasts those states that the New York Times classified as “early opening”, versus others, which I described in detail in earlier posts (e.g., Post #694).  As with the analysis above, the actual counts of cases are on two different trend lines.  But the key point is that, in terms of percent increase cases per 10,000 population, there appears to be no difference, on average between states that re-opened early and those that did not.  (Though, at some point, as they all re-open, this analysis will eventually become meaningless, because we lose the “control group” of late re-openers.)

Source:  Analysis of Hopkins daily case counts by state, and NY Times division of states into early and late-opening states circa mid-May.  See Post #694 for links to details.

And so, if the question is, where are the new infections coming from, we can begin to look at these analyses as a kind of rule-out.  In a sense, we can now start to rule out those activities newly allowed under re-opening as significant sources of infection.  More generally,either:

  1. They are being done so cautiously, on average, that they really do have little impact, or
  2. It’s still too soon to tell, or
  3. They are only being undertaken by the low-risk populations, so infections, if any, would not likely appear in official statistics, or
  4. Those activities never were the problem in the first place.

I think we can just plain ignore a final option 5, which is that these activities are just exactly offsetting some other natural reduction that would otherwise occur, absent re-opening.  (I.e., even though things aren’t worse, they’d have gotten better, faster, absent re-opening.  That would require way too much coincidence, both nationally and here in Virginia).


The key question:  what, if anything, is risky for the average mask-wearing citizen?

So, that’s progress, I think.  But it still leaves one burning question:  Where are all the new infections coming from, then?  I mean, it’s great that we seem to be able to rule out haircuts and outdoor dining and so on, as currently practiced, as a major public health hazard.  (Let me emphasize, that’s not sarcasm.  Those are important facts.)  Those are activities permitted under Phase I, but not under “Phase 0” (what we currently have in Northern Virginia).

All that the contrast between early-reopening and late-reopening areas tells us is that our new infections are coming from something that we were already doing.  (Duh.)

Arguably the most annoying aspect of “re-opening” is that nobody seems to be able to tell you how the disease continues to spread.  Nobody can give you the odds, so that you can make an informed choice about risks you’ll take, and risks you’ll avoid.  Whatever you choose to do or not, you do it more-or-less at random, or at least, based on some non-quantitative judgements.

Let’s review.

Initially, the CDC said that the disease was primarily spread by droplets emitted when symptomatic (clearly sick) individuals sneezed or coughed.  Those “droplets” (particles five microns or larger) rapidly fall out of the air, and are the basis for the six-foot social distancing rule.  And so, the CDC’s advice was, in effect, stay out of sneeze range and you’ll be OK.

That may have once been true, and might be true within the home, but this has not been true in public for months now.  I noted in my early shopping reports that nobody was appearing sick, in public, and that there was strong social discipline enforcing that (e.g., Post #565).  Coughing, at the grocery store, was simply not done.

And yet the disease continued to spread.  So the CDC’s description was wrong, in the sense of being incomplete.  That’s a way the disease can be transmitted, but not the way.  And the CDC has only slowly and partially changed its advice to reflect the reality of how this is is being transmitted now.

Circa 4/3/2020, the CDC changed it’s guidance on mask use, and in so doing, began to acknowledge, in a back-handed way, the potential for aerosol (i.e. airborne) spread of disease.  I noted this in Post #603, Radical Change in CDC Guidance.  Aerosols are much finer particles (less than five microns) that can remain suspended in the air and travel considerable distances.  See the front page of this website for links to a good explanation.   You emit considerable amounts of aerosols merely by talking — no coughing or sneezing required.  And the six-foot social distancing rule is helpful, but by no means foolproof, against short-range aerosol spread of disease.

Even now, the CDC cannot quite bring itself to use the words “aerosol” or “airborne”  in the context of COVID-19.  My recollection is that the original release of guidance to churches specifically mentioned risk of aerosols from singing.  Now, even though this risk appears to be common knowledge among major denominations (Post #682), that has been scrubbed from the CDC guidance.  The CDC retails an emphasis on ventilation and outdoor air (which only makes sense in the context of aerosols), but still will not use the word “aerosol” in its official guidance.

Some narrowly-defined places of high risk of disease transmission are now widely acknowledged.  Those include meat processing plants, for reasons that are attributed to crowding of workers together.  These include congregate living facilities, including jails and nursing homes, where the number of persons in close proximity is an invitation for spread of disease.  And, of course, hospitals, where infection of health care workers contributes significantly to the overall upward trend in total cases.

One final place of high risk is in the home itself.  Living with an infected individual puts you at high risk of infection. E.g., the majority of infections in the later stages of the Wuhan epidemic were “secondary” infections, that is, infections of family members.

But these well-identified routes of disease transmission aren’t really relevant to the average citizen.  To the average person, living in a home or apartment, with no infected family members.  What we would all like to know, then, for that person — what I would term the typical NoVA resident — is where are the remaining high-risk situations?  Where are the hotspots of infection, for the average resident?  Are there any?  Are there any that can be identified?  Is there any way to put numbers on those identifiable risks?

This is a question where, as far as I can tell, epidemiology has totally and utterly failed us.  I have yet to see a single educated guess as to places you ought to avoid, other than generic advice that outdoors is safer than indoors.  There’s simply no hard data, from any source that I can find.

So I have to guess.

The first guess is that community transmission today (not in the home, congregate living facility, or health care facility) is occurring through short-range aerosol transmission by asymptomatic individuals.  That is, people are likely being infected merely by being in the same room as an asymptomatic-but-infected individual who is talking, shouting, or singing.  (As illustrated on the front page of this website.)

Beyond that, we really need more guidance, but I don’t see it forthcoming.  For such cases, is that mainly occurring at work?  While shopping?  In church?  Where, exactly, is there risk to a reasonably prudent mask-wearing citizen?

Surely the US as a whole must have hundreds of thousands of case reports from contact tracing.  Surely the Commonwealth must have hundreds, even is it gears up to do vastly more. 

Surely, within all of that, there must be some modest clue as to what is and is not dangerous to the prudent-mask-wearing-citizen.  Either within the current body of case reports on contact tracing, or within the much large body of reports that will soon be generated as the Virginia Department of Health fills these positions.

Maybe those have been so focused on “outbreaks” so far as to be useless for this task.  That is, if all the Commonwealth had time for was major outbreaks (e.g., in jails) then their case reports simply would not address situations relevant to the average citizen.

Maybe the source of infection is unknown in the vast majority of the remainder, so that the case reports provide no information on likely infection sources in the community.

But if not, maybe it’s time to start trying to answer this key question, at least in some crude terms.  Either by employees of the state, or by crowd-sourcing redacted case reports.

The first step in such as task is a “nosology”, which in this case is some way of classifying the new infections as to how they occurred.  After eliminating cases not relevant to the average citizen (noted above), devise some sort of grouping that seems to make sense, and seems to capture a large fraction of cases.

Then, matching that grouping, you need some crude estimate of rates of exposure.  E.g., if churches occur frequently, then you need some estimate of the church-going population of Virginia in order to assess the true risk.

Absent this sort of exercise, all we will see is the spectacular cases of disease spread, as reported in the news.  The church service that results in multiple new cases.  The factory where half the workers get sick.  And so on.  The outliers, in other words.

That sort of information is pretty much useless for making an informed decision.  All it does is spread FUD.  But it’s going to be out there, regardless.  The Commonwealth needs to produce something more systematic, if possible, to guide the citizens on what is and is not risky  behavior.  Or, barring that — if they can’t or won’t do that — put the raw data in the public domain and allow that task to be crowd-sourced.  It’s too important a question to leave it unanswered.  We need a better answer than we can get just by reading the newspapers.