A riddle wrapped in a mystery inside an enigma

Winston Churchill’s famous characterization of the Soviet mind may well be attributed to the nature of the COVID-19 disease. It seems that the more we learn, in some respects the less we understand of the disease. Every expert has a different opinion of policy in regard to the degree of lockdown necessary, and the extent of the spread under different policies. The uneven spread, infectiousness and lethality of the disease has puzzled most of the professionals in the field.

What is laughable is to hear politicians pronounce that they will be “following the science” in implementing policy. The truth is they can find among scientists support for about every position on the spectrum of the degree of lockdown necessary! Unfortunately, in today’s highly charged political climate, it appears that we have politicized even this horrendous disease.

The Numbers

In last week’s post we posted a graph comparing the the infection rate of COVID-19 to the mortality rate in the United States, with a lag in mortality by 12 days from the infection date (the estimated time between infection and death).

As will be seen on the updated graph, shown above, we adjusted the scale of infections, causing the peak of mortality and infections to coincide on April 21.

This week, we have added to the graph a line representing the relative number of tests conducted per day.

The most important conclusion is that the mortality rate continues in its secular decline, and at the same steady rate it has experienced since the peak in late April (the slight bump after May 31 is, we believe, due to reporting anomalies around the Labor Day weekend)

It can be also readily seen that the reported infection rate has move steadily upward against the mortality rate, correlating closely to the rise in the testing rate. This correlation has remained steady since the peak in Mid-April.

The conclusion to be drawn is that the rising number of reported infections is due to significant additional testing.

Even with this rising disparity between infection and mortality, we are probably not even beginning to count the total number of actual infections. We can deduce this from the IFR (Infection Fatality Rate) that the current numbers reflect. The IFR of COVID-19 is currently being debated, but there is little disagreement that it is around 1%. (for example: https://www.medrxiv.org/content/10.1101/2020.05.11.20098780v1)

However, if we use the statistics that are being reported to us, then we get an IFR that reached as high as 6% in mid-May, and around 5.3% currently. From this we can infer that the true gap between the rate of infections and the rate of deaths are dramatically greater than that indicated on the graph. Only widespread antibody testing will reveal the breadth of infection that has actually occurred, but we should expect the gap in the graph to continue widening as we detect greater and greater numbers of those infected, compared to those dying, from the disease.

The question the remains: how do we accurately determine whether we are reducing, rather than increasing the death rate, and whether our opening of the economy is placing us in danger?

Consider the Major States

One method is to closely examine the impact of lifting the lockdowns on the most populous states that are reporting rising infection rates.

If we examine the www.rt.live site, we can see that among the largest states a relatively even distribution of statistical experiences in managing the pandemic.

This slide represent R_t in each of the indicated states. To quote rt.live, “These are up-to-date values for R_t, a key measure of how fast the virus is growing. It’s the average number of people who become infected by an infectious person. If R_t is above 1.0, the virus will spread quickly. When R_t is below 1.0, the virus will stop spreading”. The site publishes the continuing change in the R_t from day to day, providing a graphic view of the changing dynamic of the pandemic. Of course, a lot depends on the quality of the methodology, and we are not commenting on its accuracy: instead we are relying on a relative accuracy between the states we examine, and to find those at the highest risk. So we look for the states that lie above the line of R_t=1 in the diagram above. For this reason we selected California, Texas and Florida. These are the largest of the at-risk states, and together comprise more than a quarter of the country’s population, span all of the time zones, and represent a very wide range of demographics. Also, we see a good spread of risk between the lowest risk state – California – the median state – Texas – and the highest risk state – Florida.

California

The 7-day moving average of daily new cases continues to climb from its high water mark of last week to a new high this week, reaching above 4,000 cases for the first time. Let’s see the impact on Mortality.

Like last week, the 7-day moving average appears largely unchanged, with no indication of the significant and consistent climb in the infection rate. This indicates that the relaxation of the lockdown that started on May 25th has – as yet – little impact on the spread of the disease.

Texas

Texas has shown a more dramatic rise in infections than has California, and it is a fact that Texas began the relaxation on April 29th, a month earlier: this is could be an important factor, indicating the potential danger in the relaxation. Let’s view the mortality rate:

The mortality rate does not, however, bear any fruit to this argument. If anything, it is slightly lower than it was when the lockdown was eased. By May 7th, a week after the end of lockdown, the death rate peaked at 38: it is now at or around 30, but has been as low as 20. There was a spike after June 16 that bears watching which will have to be reviewed in the coming weeks.

Florida

Florida ended its shelter on May 4, and the infection rate remained constant for a month, suddenly beginning a very significant spike in early June, from an average below 1,000 diagnosed infections a day to 4,000 diagnosed infection a day by June 20. This spike exceeds that of other states by a wide margin, and brings into question whether such a disparity could be explained by wider testing. Perhaps an examination of the mortality rate will shed some light.

Again, the mortality rate does not seem to reflect the alarming rise in infections. In fact, this graph seemed so counterintuitive compared to the infection rate, that we went back to review the actual numbers, rather the moving average. Deaths for the first three weeks of June were 244, 237, and 219 respectively, showing a DECLINE in the weekly rate! This figure should have been increasing dramatically to reflect the dramatic and persistent increase in the infection rate beginning in the first week of June. However, we need to reserve judgement until the forthcoming weeks, as the deaths could still show up in late June.

Conclusion

We have not yet seen evidence in the global numbers or in the individual states numbers to confirm that there is a second wave, or even an increase in the actual number of infections. Our thesis remains intact: the rising infection rate is a function of the increase in testing, and the very large number of still undetected infections.

Comparison

We end each week with a running comparison between the pandemic experience in the U.S. and in a collection of comparable countries in Europe representing the size and development stage of the US.