My search for coronavirus artifacts began in the vast sedimentary sludge of late Holocene reporting from national health centers around the world. (That is, not quite a week ago.) I noted, for example, how the very high death rate from the virus in Italy contrasted with the very low death rate in Germany. Specifically, 8% of infected Italians were dying, while only 0.3% of infected Germans were dying. That is a ratio of more than 25 to one, and I concluded it could not possibly be accounted for by the various explanations offered—the smoking rate in Italy, an older population, less frequent hand washing, closer personal space etc. There had to be something else at work.
On the morning of March 19th I wrote to family and friends: “I have a suspicion that the data on the coronavirus epidemic is being misapplied. Looking at the exponential growth in the number of new coronavirus cases, it strikes me that this could simply be an artifact reflecting the exponential growth in the number of people being tested.” (An “artifact,” in this sense, is a misleading or confusing creation or application of data.) I explained that as the number of people tested increases, the number of new cases will fall off because we will reach a point of diminishing returns. “It will be like picking cherries,” I wrote, “slow at first until you bring out the ladder, and then increasing rapidly until only the most difficult fruit to access is left on the tree. The more rapidly the tests are being deployed, the more dramatic this data artifact will be.”
Continuing, “If this explanation is correct, we should be able to see it in the shape of the curve. First, we should see a much flatter curve for the number of deaths than for the number of new cases. (Providing an identical scale is employed on both x and y axes.) That is because the number of deaths is not dramatically impacted by the number of tests administered.
The second implication is that as test use is expanded the tests will be given to a higher proportion of asymptomatic people more remote from the centers of infection, of whom more and more will test negative.” Moreover, I wrote, “As test use expands, the number of infected individuals with serious symptoms who have not been tested will fall to near zero. Because most of the new cases identified at that point will be mildly infected or asymptomatic individuals who never did (or will) become seriously ill (approaching 80% of all cases), this will load the data bank in favor of survivors. That is, it will lower the death rate even more dramatically than it will lower the infection rate. (With expanded testing), watch for a falling death rate to precede a falling infection rate.”
On the morning of March 20th I followed up. “I looked at the data from Italy. Aside from the (demographic and health factors noted above), I came upon a couple of other explanations for the high death rate. First is a relative lack of testing which has resulted in Italy’s failure to identify mild and asymptomatic cases. It looks to me like the extent of their epidemic was so much larger than imagined by the time they identified it that their medical system was quickly overwhelmed. Consequently, they were forced to triage and to concentrate their efforts on those already seriously ill. As a result, testing of the wider population was haphazard and minimal, and I expect they have continued to vastly underestimate their total number of cases.
This would of course make their death rate appear artificially high. On top of that, (it has been reported that) a very large number of their deaths befell ‘pre-morbid’ individuals—people who were already seen as terminal in the near term when they contracted the disease. All of these deaths, regardless of the actual cause…, were recorded by Italy’s doctors as coronavirus deaths.”
“By contrast, Germany has reported an extremely low death rate. It is true that the coronavirus outbreak in Germany is of much shorter duration, and… that the average time from infection to death is 17days. In addition, the infection in Germany has involved a lower aged segment of the population. But I think the real explanation for the low death rate in Germany vs. the high rate in Italy can be better explained by the statistical artifact I have described. Quoting the Financial Times: ‘Another factor that may help explain the variance is the unusually high number of tests being carried out in Germany. According to Lothar Wieler, the president of the Robert Koch Institute, German laboratories are now carrying out about 160,000 coronavirus tests every week—more than some European countries have carried out in total since the crisis started.’” https://www.ft.com/content/c0755b30-69bb-11ea-800d-da70cff6e4d3 This is also roughly equivalent to the total number of tests performed so far in the U.S.
“I don’t have data on the total number of tests for each country, but from this report it seems likely that Germany has tested 10 or even 20 times as many people as Italy has tested. That would of course result in a reported death rate in Germany lower by approximately that same factor.”
At this point in the conversation it occurred to me that those who have survived coronavirus infection may not be identified by testing. Obviously, this would lower the reported rate of infection still further, thus simultaneously raising the reported death rate. I found the answer in USA Today reader questions. As I quoted from the answers provided, “’While antibody tests can identify coronavirus survivors, that type of test is not being used here.’ If we are not using it, I doubt that Italy is. As you can see from the article, (https://www.usatoday.com/story/news/health/2020/03/12/coronavirus-reader-questions-answered-symptoms-testing-elderly/5012457002/ ) in most of the countries where the virus appeared first many if not most of the earliest cases have already survived the virus and yet when finally tested will test negative. As time passes this will become a more and more significant cohort, falsely lowering the infection rate and thus upwardly distorting the death rate.”
Later that day I found a Science magazine article confirming that the tests thus far employed fail not only to identify those recovered, but “even to identify many of those mildly infected at the time of the test.” https://www.sciencemag.org/news/2020/03/new-blood-tests-antibodies-could-show-true-scale-coronavirus-pandemic “The type of antibody testing required to identify such cases is only now being rolled out, and data from that limited testing is not yet available. As previously noted, this deficiency in the testing will necessarily result in the reporting of artificially high death rates.”
As it turns out, per a World Health Organization recommendation Italy is testing only those with obvious indications of illness. As I wrote, “I think Italy’s compound failures—both to act quickly and to test beyond the obviously sick—has put them in a helpless position. The numbers are expanding so quickly that they cannot even effectively quarantine all of them. And their medical system is so overwhelmed that they are unable to treat most of the seriously ill, let alone attempt to identify the other carriers. In metaphorical terms, they failed to put out the spot fires, so now they will just have to wait until the inferno burns up its fuel.”
Later that same day I reported “some data for Italy that seems to confirm the statistical artifact I have proposed. The first Italian coronavirus death was in the city of Vo, population about 3,000. The Italians quickly locked down that city and tested everyone. They found roughly 100 people infected, and immediately quarantined all of them. As a result, Vo was virus free in two weeks.”
“Unfortunately, Italy did not apply this lesson to other regions. According to the Guardian, only obviously symptomatic subjects needing medical care have since been tested in Italy. Quoting from the article, ‘In Italy, we have struggled with a rampant rise of mortality (the number of casualties divided by the number of infected people), which has reached an apparent value of 8%–far higher that the mortality rate in China…. This high rate is misleading, though…. Only the obviously symptomatic—those needing medical care—have been tested for the virus and thus counted as medical cases…. The consequence has been that people who haven’t asked for medical attention have only been tested occasionally in Italy. Nonetheless, asymptomatic or quasi-symptomatic subjects represent a good 70% of all virus infected people and, still worse, an unknown yet impossible to ignore portion of them can transmit the virus to others…. If the fact that only those presenting with the virus were being tested was accounted for, the mortality percentage would fall to ‘normal’ levels.’” https://www.theguardian.com/commentisfree/2020/mar/20/eradicated-coronavirus-mass-testing-covid-19-italy-vo In other words, the actual death rate in Italy is comparable to that of other countries.
As I summed this all up, “Germany, which seems to have the highest level of testing, has the lowest death rate. Italy, which tests only obviously infected individuals, has the highest death rate. This all but confirms that the reported death rates are artifacts of the (variable levels) of testing.” It also demonstrates that accurate and early testing is a prerequisite to effective social isolation, which is essential to keeping the death rate down.
I next noted that the Los Angeles Times of March 20 has reported that L.A. County health officials are advising doctors to cease testing patients unless “a positive result could change how they would be treated….” https://www.latimes.com/california/story/2020-03-20/coronavirus-county-doctors-containment-testing That guidance, sent by the Los Angeles Department of Public Health, the Times reported, “was prompted by a crush of patients and shortage of tests, and could make it difficult to ever know precisely how many people in L.A. County contracted the virus.” I think we can safely apply this final caveat to nearly all the data being received from all around the world.
Testing across the U.S. is now being limited in the same way, not only due to a shortage of tests, but also to a shortage of personal protective equipment which must be changed after each test. https://townhall.com/tipsheet/bethbaumann/2020/03/21/dr-fauci-nonsymptomatic-people-need-to-refrain-from-testing-to-protect-health-care-workers-n2565429
There are many lessons in all this. First, a more subtle message: When you see hysteria and hand-wringing by the press over the various charts and graphs they present for coronavirus, or for that matter for anything else they are reporting, bear in mind that most of them haven’t the faintest idea how to interpret that data and will almost invariably draw alarmingly false conclusions from it. That is what sells papers.
The more obvious message may also be more important. We need to bring home the manufacture of all health and national security items, and in particular the manufacture of our own medical supplies. Right now we lack not only the capacity to test for coronavirus and the drugs to treat it, but also the medical masks and gowns needed to protect our healthcare professionals from infection when they do test.
If there is one concern I would take away from the “artifact” I have identified, it is that we in the U.S., as official policy, are still only testing those with obvious symptoms who have either travelled to an infected region or had known contact with an infected individual. I understand the reasons for this. But that we have allowed ourselves to fall into such a predicament is inexcusable. It is also a formula for creating here the death rate we are currently seeing in Italy—and that is no artifact.