Covid 19: What about the data?

Does the data really tell a story?

The world went into the Covid-19 pandemic with a general understanding of what we thought we knew about most flu type viruses. These flu viruses have been studied extensively over the last 50–75 years and some of the things we understood included:

• The flu is usually not transmitted through the air, but through physical contact with the virus, typically on surfaces. Washing hands reduces the chance of spreading the virus.
• Sunlight is very destructive to the virus. There have been various tests done to prove this.
• People in close quarters are more likely to spread viruses. Homes and office buildings are worse than parks. This is related to the first point.

This article looks into the very limited history of what we do know (very little) and what we cannot conclude. Let's jump into data.

New York

On March 20, 2020, the governor of New York gave the order, effective March 22, to shut down all non-essential businesses and people to stay in place. The graph below shows what happened following the order. You can see more details from the source I used here.

It appears new cases peaked between the 5th and the 11th of April. About two weeks from the shutdown order. This appears logical. If people had been spreading the disease more freely until the shutdown and Covid19 has about a two week incubation period, then the maximum sick people (ignoring the asymptomatic, since we have no way of measuring those with this data) should be detected about two weeks later. The graph shows this to be the case.

It is important to clearly state something about the data you see above. It is scientifically suspect to draw conclusions from this data, as the acquisition of the data is not controlled. In this case, this is primarily information from people that show “covid” symptoms. These show up at a hospital or doctor's office and are prescribed a test. It excludes most of the population who does have any recognizable covid symptoms. It appears more people without symptoms are being tested as tests have become more readily available.

The next problem begins when one compares the daily new cases to the daily death numbers [below]. These peaked at about the same time. One would expect there to be a relationship between cases and deaths. If there were none, we would not be that concerned about people getting infected. That herd immunity thing. Assuming there is a relationship, then the people dying from the diseases should be dying days, weeks, or even months after being identified by the disease. The peak of deaths should be a few weeks, following the peak of the cases. For New York, this peak happened simultaneously. Later for Georgia data, we see a picture closer to the one we expect.

California: lockdowns in other states

In New York, we saw a really big peak and then a fade that represented a bell curve. New cases peaked at over 10,000 a day. California on the other hand followed a completely different trend [below].

I would like to point out some significant differences. California also had an initial “new case” peak around early April. Their daily death peak did not occur until the end of April. The data is more aligned with what we think makes sense, people fall in and then some of them die a little later. This data does not have a bell curve like the New York data. The peak occurs and then other peaks occur. I added a 7-day moving average to demonstrate this. Whereas the New York deaths peaked at a 1,000 deaths a day and then came down to less than a hundred, California (a bigger state) peaked at a 100, but then just stayed there, since the end of April. Meanwhile, the daily new cases have continued to climb, but the death rates have come down. Defying logic!

Post lockdown increase in cases?

Recently we have heard of states that have removed their lockdown and seen significant increases in cases. Georgia [graph below] ended its lockdown on April 24th. The graph below shows the increase in new cases per day about three weeks later, by the middle of May, with the increase in cases per day being higher than past peaks. Some claim it is due to the increased testing while others believe it is because of the greater personal contact. It is nearly impossible to determine the root cause.

However, one would expect an increase in deaths either at the time of the increase in cases or two weeks later. We do not see that happen — yet.

Iowa — the no lockdown story

Iowa did not have the traditional lockdown, like most other states. There were businesses shuttered. I was unable to use the same source of data and had to use the state website instead. What is interesting with the data below is that it follows a path one might expect with an infection running through the population. The number of cases peaked in early May and the number of deaths peaked in late May. Unlike New York and California, the Iowa population is much smaller and more rural.

Sweden — no lockdown

Yet, there is urban Sweden. Again no lockdown and instead the government recommended social distancing. Here are their numbers. They have not followed the bell curve (like California) for the new daily cases. Their deaths per day however, do follow the bell curve.

No Conclusion

The important aspect of this data is in hindsight there is no better-recommended course in handling this pandemic. Lockdown curves and no lockdown curves and even opening curves did not have much of an impact on the death rates. New York with the most severe lockdown had the highest deaths and there is no explanation for that.

When we first started down this course in Feb, all the experts wanted to flatten the curve to limit the impact on the health systems. The US health system did a fine job of handling the challenge, even in New York City, the worst impacted city in the US. If that is still the intent, we should not be concerned about future increases in infection. Then again, what will we see in the news?