My PhD student Emma Zohner has updated our web application to study testing
, incidence and mortality by state for the USA or by country across the entire world. This app is built around the daily data posted for USA by http://covidtracking.com and internationally by https://covid.ourworldindata.org
The user can choose any number of states in USA, or number of countries in world, and it will plot the following information in 4 figures:
1. Number of tests conducted
2. Number of positive tests (i.e. official incidence)
3. Number of hospitalizations (the least accurate data)
4. Number of deaths.
These numbers can be plotted either (1) cumulatively or (2) incrementally, i.e. listing amount for each day, or (3) per capita, with cumulative numbers per million residents.
They can be plotted on (1) raw or (2) log_10 scale.
Additionally, if you choose "other rates" you get:
1. Proportion of tests that are positive
2. Number of hospitalized patients per positive test
3. Number of deaths per positive test.
Again, they can be on raw or log scale.
The international part was just added. Here are some interesting observations:
1. USA has now tested about 2.3% of the population, which is in the middle of testing rates, notably above the UK (1.6%), France (1.1%) and South Korea (1.3%) and behind Germany (3.3%), Italy (3.8%), and Spain (2.9%). Note that no country has done anywhere close to universal testing -- some high ones include Denmark (4.7%) and Iceland (15.1%).
2. When you look at incidence per capita, you see that almost all countries have flattened pretty well with 3 notable exceptions: USA, UK, and Sweden. These are growing linearly still, but clearly not exponentially as all countries have "flattened the log curve".
3. The deaths per positive test, or offical-case-death rate is an interesting number. France is very high with 19.1%, and UK, Italy, Spain and Sweden next between 12% and 15%, and USA quite low at 5.9% with Germany 4.2% also very low.
4. Sweden is an interesting case study. They are about the only country to never Institute lockdowns. If you compare with their close neighbors geographically and culturally, Norway, Denmark, and Finland, you can see the differences in this policy. The per capita incidence and death rates are considerably higher, with incidence (Sweden 0.23% incidence, Denmark 0.17%, Norway 0.15%, and Finland 0.10%) showing roughly 50% relative increase to its neighbors. We see the neighbors have leveled off, while Sweden is still growing roughly linearly. Per capita covid death rates are also higher (Sweden 283 per million and Denmark 87, Norway 39, and Finland 34 per million) . The deaths per positive case is much higher (Sweden 12.2% vs. 5.1% Denmark, 2.6% Norway and 4.6% Finland), indicating perhaps more vulnerable populations are getting infected in Sweden, and may be partly explained by Sweden's documented problem with long term care facilities that their officials now insist has been corrected. Interesting that the incidence curve for Sweden is not close to exponential, and their healthcare system has not been overwhelmed.
Update: We have added 3-day and 7-day moving average options in addition to raw data; this is to account for the the fact that some municipalities report numbers unevenly -- e.g. not reporting much on weekend and then reporting a much larger number on Monday -- to smooth out these spikes and potentially better show the trend.
Anyway, feel free to play around with the app and have fun!
