Increased Cases or Tests?

(Right click to expand images in a new window)

I have heard people trying to make the argument that the increasing number of cases in the US is due to more testing. While that certainly plays a role, I don’t believe that what we are seeing occur supports that notion.

I wanted to try to figure out if there was a way to answer that question. I will admit that I may be making some type of error in my thoughts and certainly invite another interpretation. I’ll try to explain my conclusions as clearly as possible.

This graph above is two graphs superimposed on each other and offset so as not to obscure each other. The blue one is related to cases, the red one is related to deaths. In both, these represent the slopes of the data.

There is a fainter line for both colors with dots on it. These are the true slopes for those three prior weeks. The dark solid smoothed lines are the averaged seven day trend lines for those slopes. These flatten out the weekly cycles of disease that have emerged.

The blue data is on the left axis, the red on the right. The colored horizontal lines represent zero for each respective color.’

The way to interpret these is what I have described before for the slope graphs. When the line is above zero, cases are increasing. When below, they are decreasing. The distance vertically the the zero line from a given point is the speed at which cases are increasing or decreasing. You can see that around the first week of April, we were experiencing exponential growth in the US. It should also be noted that deaths have consistently lagged about a week behind rises and falls in cases.

For the two week period March 15th to March 29th, to April 1st, we went from 2859 total cases in the US to 121,786, an increase of 43x. I was shocked when I saw that figure and thought I must have calculated wrong. I got the same answer. One way to validate it would be to compare the increase in total deaths, but one week later for each measuring point. On March 22nd, there were 453 total deaths, and two weeks later, there were 10,849, and increase of 24x. Either multiplication factor is frightening. That’s how exponential growth works and how something that doesn’t seem like much of a problem right now can explode in a matter of days

One interpretation has to do with the bump in cases related to deaths from the end of April to mid May. My assumption had been that had represented more testing. However, as I was searching for data to support or reject that hypothesis, I found something much more illustrative that the increase in numbers is NOT due to increased positive tests, but is due to true illness.

Chart from the Johns Hopkins COVID-19 Testing Insights Initiative depicting daily total tests and daily positive tests using COVID Tracking Project data. This visualization is not a dynamic representation of case data and will not update automatically

This is the part I hope I’m interpreting right. If the percentage of positive cases from tests is going down (as represented by the black line on the above graph), then we should be seeing a downward trend in cases and deaths. However, the trend of cases in the upper graph has been going up since about May 20th and among deaths since the beginning of June. That indicates to me that we are experiencing a resurgence in cases.


Leave a Reply