The Virus Wanes/Our Media Panic

It irritates me that our media are full of what has aptly been described as panic porn with respect to the so-called Covid-19 reopening spikes, particularly given that the virus’s spread has been slowing for months. If you follow along for a bit, it won’t be hard for me to show you that slowdown with readily available data, simple math and a pinch of deductive reasoning, the combination of which elements seems to be beyond the capabilities of the overwhelming majority of journalists and television personalities.

This https://www.worldometers.info/coronavirus/country/us/ website will provide many of the data points. If you open it, below the day’s state-by-state Covid-19 data you’ll find multiple graphs that track the same information, but aggregated nation-wide and showing every day for the last five months. The two graphs that are important for this post are the ones that trace deaths (the last of the graphs) and new infections (a little higher on the web page). If you hit the seven day rolling average buttons underneath those two graphs, you’ll begin to see the numbers as I do.   

Death numbers are hard facts, so the graph that tallies them is an ideal starting point for understanding what’s happening. US deaths attributed to Covid-19, seen on a rolling seven day basis, peaked at 2,238 on April 20th. As of yesterday, July 1st, the rolling average for the last seven days was 560, a decline of 75% from April 20th. The declines in US Covid-19 deaths over that time frame were quite steady.

Another of the relevant data points is the typical amount of time between hospitalization for Covid-19 and recovery or death. This https://www.medrxiv.org/content/10.1101/2020.04.30.20084780v3 website provides the data indicating that the median hospital stay for Covid-19 patients outside of China is 3-9 days. The median stay for Covid-19 ICU patients – presumably those most likely to die – outside of China is 4-11 days.  For simplicity’s sake, in the calculations that follow, I have assumed that the median for all US virus-related hospital stays is seven days (I couldn’t find that specific data point, but if I had assumed hospital stays were a few days shorter or longer it wouldn’t have made a significant difference to my conclusions because the trends are so steady). I will use the assumed seven day hospital stay as a proxy for the lag between when the infection becomes undeniably serious and the ultimate outcome.

Seven days before the April 20th Covid-19 US mortality peak, the number of daily new infections in the US was given as 27,094, of whom, based on the rolling average peak seven days later, something like 2,238, or about 8.3%, would die.

But here’s the thing: since yesterday’s seven day rolling average US Covid19 mortality was 560, if the seven day mortality rate had been constant at 8.3% (the comparable mid-April figure), we would expect to see that there had been 6,776 new cases on June 24th. In fact, there were 38,434 new cases on that date, implying a late June mortality rate relative to newly identified cases as of a week before of about 1.5%, not 8.3%. Something has changed dramatically; maybe several things.

I don’t doubt that doctors and nurses have become better at treating the virus, or that their doing so has improved outcomes – but (in the absence of widely dramatic health care improvements, which would have been widely heralded so of which we surely would have heard) I am quite certain that better treatments do not account for more than a small percentage of the apparent 82% drop (from 8.3% to 1.5%) in virus-related deaths relative to newly identified infections.

A more persuasive explanation is not far to seek. If you look here https://covid19.healthdata.org/united-states-of-america and find the graph labeled Daily Infections and Testing, then press the green button that highlights testing, you’ll see that on April 13th, an estimated 153,402 people were tested, and that on June 24th about 538,795 were. In other words, from April 13th to June 24th, testing increased by 251% while those days’ newly identified cases rose 41% (i.e., from 27,094 on 4/13 to 38,434 on 6/24).

Taken together, these numbers clearly imply that the recent spikes in newly identified cases are a function of increased testing – that is, of better knowledge of the extent of the virus’s spread rather than of any acceleration of it.  Otherwise, we would’ve expected to see the rolling average numbers of daily deaths rising more or less in line with the increases in newly identified cases (after the stipulated weeklong delays). Since, instead, deaths have fallen precipitously while newly identified cases have risen, we can confidently conclude that most of what we’re seeing in the rising numbers of newly identified cases is the identification of cases that would have gone undiscovered given the testing levels back in April.

The increased testing being done today is revealing that the virus is far more widespread, and much less lethal, than had previously been believed.  The increased number of new identified infections is not reflective of a broad, sudden jump in actual nation-wide infection rates (although local jumps certainly happen) – it’s reflective of the 251% increase in testing rates. We have no clarity on how many people have, or have had, the virus; all we have is guesswork and inferences based on positive test rates (which are declining: from 27,094/153,402, or 17.6% in April to 38,434/538,795 or 7.1% today).

Again, the clearest hard fact out there is that deaths have declined steadily from about 2,238 per day in mid April to about 560 a day now – a fact that, in the absence of huge medical advances or a weakening of the virus itself, strongly implies that the spread of the virus has been steadily losing momentum for two and a half months. This is great news – but have you seen it in our media? Didn’t think so – they’re too busy trying to scare you out of your wits.

I think it’s fair to write that we don’t know why the virus’s spread has been declining steadily. Insofar as the decline is acknowledged at all, most politicians would like to attribute it to the lockdowns – but there are (equally) studies suggesting that the lockdowns had little effect. Perhaps it’s the social distancing, or the warmer weather (flu-like viruses tend to subside in summertime). Or maybe the virus itself is weakening, as happens with many viruses, and as has been argued in this case by prominent Italian doctors. We don’t know. But we do know that its spread is slowing.

That news that the virus’s spread has been slowing for months should not be taken as meaning that the virus isn’t still spreading at all – and of course it’s still all-too-lethal for the roughly 560 people nationwide who are dying from it each day at present. Clearly lots and lots of people have it – testers are discovering more all the time – and many people are still getting it, though apparently (based on the inference drawn from the declining deaths) only about one quarter as many as were getting it a few months ago.

The other good news is that even the 1.5% death rate seemingly implied by the comparison of the current seven day rolling average deaths with a week ago’s newly identified infections number must be overstated because of the large number of asymptomatic cases that still go untested; current estimates are that the infection fatality rate (as contrasted with the case fatality rate, which references only known cases) is far below 1%.

My takeaways from these numbers are that 1) the rising number of identified infections is overwhelmingly a function of increased testing rather than loosening restrictions, 2) so many people have the virus – many without knowing it – that stopping community spread via lockdowns is a practical impossibility, 3) the virus is on the wane as evidenced by the steady but nonetheless dramatic declines in Covid-19 deaths over the last two and a half months, 4) the death rates relative to the number of infected people are much, much lower than we had been led to believe they would be (and Covid-19 deaths are overwhelmingly concentrated among the very elderly and those with well-defined vulnerabilities, though I haven’t run through those numbers here) and, finally, 5) the media prefer to make scary predictions than to report on known facts that implicitly contradict the premises behind those predictions – particularly, I have to believe, if they think that engendering panic will harm the political fortunes of a president they despise.  

Scary stories can be fun, but not when they’re passed off as real news and used to frighten people into making bad decisions.

M.H. Johnston    

6 comments to The Virus Wanes/Our Media Panic

Leave a reply

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>