C.E. Sum looked at CNN this morning to see if there was anything special going on in the US. A massive blast in Beiruth. Trump is talking to a FOX journalist who caught him off his feet on how well the federation is doing in terms of control over COVID-19 in the US. Trump out of his depth. A festive sight for those who, with me, fear him and his antics. And with regard to the more than one hundred and fifty thousand COVID-19 deaths in the US, he got no further than “it is what it is” and that having control is tantamount to doing your best.
C.E. feels a problem coming. It is good for the world if Trump falls and keeps dropping just before the elections. But maybe it is also good to look at those numbers again. After all, it would be disastrous if he managed to get out of these troubles. And those numbers are considerably more difficult to read than we think at first glance. Let’s look at a picture of US numbers shot on July 31.
The left panel has two lines of interest. They are both black. They are based on records of COVID-19 infections and COVID-19 deaths. The number of infections increases more or less linearly between week 14 and week 31. That is a sign of control in that the exponential line observed up to week 14 has been turned into a straight line. But a line that has increased from about 280,000 to around 4.7 million in those 16 weeks. In the same period, the death toll rose from 2,000 to 156,000. In Fig. 3 C.E. gives gives the exact numbers in a table.
The right panel shows the numbers of new infections and new deaths (red) per week. C.E. has scaled the numbers of contaminations to show both lines in a single graph. To achieve this, the numbers of infections were divided by 14. C.E. expects that these figures should show how well and how bad measures against the expansion of COVID-19 work. And such makes it possible to compare policy measures. At least that is the expectation. C.E. shows in Fig. 2 the numbers for the Netherlands.
At first glance it is clear: the Netherlands has considerably better control than the US. But is that real? How good are those numbers actually? And why is the difference in d-rate2s (the numbers of COVID-19 harmonized for deaths / population sizes) so small (0.0353 – 0.0477)? This needs to be looked at more closely. And it must be investigated further how the interpretation of the data is influenced by the numbers and the focus of tests. And is the uptake of Dutch contaminations since week 29 really exponential?