Looking for the next wave in the US

There has been lots of talk about whether there will be a new wave of infections and deaths in the US as the economy is re-opened.

These charts look to find which states may be experiencing 2nd and subsequent waves of infections and deaths.

Confirmed CasesĀ 


The filter to find these states is as follows :

  1. Identify a first peak by identifying a drop of at least 25% after the peak and at least 10 cases for deaths and 100 cases for confirmed infections.
  2. Identify a second peak which is at least 33% above the value at the minimum after the first peak.
  3. Further peaks (ie. third, fourth etc) are similarly identified.
  4. Only display regions where the last peak has occurred in the last 14 days and also is bigger than the first peak.

Country Testing Performance

This post ranks countries and regions by the ratio of deaths to confirmed cases. This is a good indicator of testing efficiency. Assuming that the ultimate death rates are similar across countries and regions then a higher ratio at this point in time indicates that the number of confirmed cases is too low compared to the actual number of infections in the community.

The Diamond Princess is used as a benchmark as it is a population that was 100% tested and hence the ultimate death rate on the cruise ship is a good indication of what will be the case in the broader community.

The ratio calculation uses the confirmed cases from 10 days prior given that there is a lag between a case being confirmed and deaths.

The ratios for all countries with 500 hundred current confirmed cases and over 100 confirmed cases 10 days ago are included.

The OECD rankings are presented below to compare countries with broadly similar health systems.

The testing performance of different US states is compared below.

Country Specific Growth Rates

I have put together some code to analyse the Johns Hopkins Covid-19 global stats that are updated regularly.

Use the drop down boxes below to pick analysis on either Confirmed Cases or Deaths and a country to see the results.



The code looks at a country to see if there has been a change in the exponential rate of growth of cases. The series starts from the point where 100 cases were recorded for confirmed cases and 10 cases for deaths.

If y is the cumulative number of confirmed cases or deaths then exponential growth with look something like the following after t days.

y = K^t

If exponential growth is constant the K will not change over time. Note that exponential growth implies that K>1. If K changes over a period of time then we can see if the country has changed to a new level of exponential growth.