Tuesday, April 21, 2020

COVID, and the Graphic Representations of Death, and Transmission

Let's do two curves today: the first is from the Journal of the New York Times, and shows what is likely to be a more accurate death toll from the COVID outbreak thus far. Counting, it turns out, is a hard business in medicine. What is often the "official" count of COVID mortality in any given country or smaller municipality is a count that a government officer tallies by receiving reports from professionals--usually but not always physicians--who note in a form that so-and-so died, and had a positive COVID test. But that excludes people who never got a COVID test, either because they died outside a hospital, or because testing was woefully inadequate and there simply weren't enough tests to go around, which might resemble a large, wealthy country located in North America.

Thus, a simple but different way of sizing up the impact of COVID is simply to compare all deaths to the averages in previous years (which are, generally speaking, roughly stable from year to year in most places in the US where there hasn't been mass influx or exodus). If the needle moves above the historical average, then that is a very strong indicator that COVID deaths are being missed. In the immortal words of Rasheed Wallace, the ball don't lie. Anyway, the article looks at 2020 compared to previous years, and each figure, from various parts of the world, all show the impact of the COVID outbreak in stark terms. Because the graphics at NYT can't be pulled and reproduced, a similarish graphic can be found in this piece from Reason:
The upper, flat, gray line, hovering around the number 2000, is the average number of deaths over the preceding seven days in New York City during the years 2015-2019--that is, it's an average of averages. The colored numbers toward the bottom are the seven-day averages of deaths from influenza in New York in any given year from the same period. Where this departs from the NYT piece is that these data only look at official COVID and flu deaths and compares them to the overall total. For 2020, you would also need to add in everyone else who has died, which now includes a sizable chunk of people, and the red spike would climb even higher. (It's also worth keeping in mind, in an apples-to-apples comparison, that the same issue applies to deaths from influenza, which is at least as hard to diagnose as COVID, so the total deaths from flu in any given year are also higher than the official number. Epidemiologists have various tricks to tease out a reasonably accurate picture of flu mortality in any given year. But as noted before, a very useful way to get a sense of how lethal an epidemic is in a given year, whether from flu or COVID, is to just look at the total number of deaths and compare it to recent years, and if the official number just from the infection exceeds the total mortality in previous years, it's a sure bet that the epidemic is especially severe. The ball don't lie.) The Times piece includes the observation that "the totals include deaths from Covid-19 as well as those from other causes, likely including people who could not be treated as hospitals became overwhelmed," and link to an article explaining this heartbreaking phenomenon. During the Ebola outbreak, for instance, pregnant women were unable to receive adequate prenatal care due to hospitals and clinics being closed, and a small percentage of women died from complications of routine problems that would have been handled had such facilities been open. Thus, Ebola killed those who never became infected with the virus, and as the COVID outbreak proceeds, there will be similar effects on overall mortality. Curve #2 is here:

It is from this paper, entitled "Temporal dynamics in viral shedding and transmissibility of COVID-19." The key here is that the zero of the x-axis is not at the left corner where it is normally found, but rather can be found four ticks in. The gist of this figure--which, to be clear, is like so many other things in COVIDland merely a model based on relatively limited data--is that people with COVID may actually be in their most infectious phase before they even know they might be infected. And that is one very serious problem as we collectively contemplate, at the local, national, and global levels, about how to resume economic and social activity that resembles something approaching normal. "Significant presymptomatic transmission would probably reduce the effectiveness of control measures that are initiated by symptom onset, such as isolation, contact tracing and enhanced hygiene or use of face masks for symptomatic persons," the authors state in the characteristic understatement of scientific communication (that's my emphasis; journals don't allow for italics). But make no mistake, this makes ongoing policy discussions, which have already become weaponized, not only among the violent nitwits and infants of the MAGA coalition in the US, but in various other locales as well, a good deal more challenging. The net result will almost certainly be even more graves, and those who will soon reside there.



  1. Billy, the web counters right now (4-24-2020 at 11 am EDT) are reporting about 50.4k deaths directly from covid. We know that's an undercount, but any guess on exactly how under it is? Or is this a "they'll figure it out in 5 years when all the dust settles" type situation?

  2. We "know" that's an undercount. What you "know" ain't exactly "so". Compare 2019 deaths from 2020 deaths for NY. What we actually know is that there is a high degree of uncertainty. Sloppy thinking.

    People keep saying that covid is coming to my area of flyover country, but our new infections have been flat for a month and our random swab sample showed almost 4% infections. I figure we're at 16% who have or have had covid. Herd immunity by end of June.