March 11, 2020 | Joe Brew
COVID-19 Data Explorer

A tool for comparing COVID-19's epidemic curve across countries


(Don't feel like reading an article and want to go straight to the data? Here is the link to Databrew's COVID-19 data explorer.)

With COVID-19 having spread to more than 100 countries, the world is quickly getting a crash course in epidemiology. Specialist concepts like "R0", "case fatality rate", and "epidemic curve" are now common parlance on news television. And though a plethora of academic research is being produced on the topic, few tools are available for the public to explore and understand how the epidemic is behaving differently in different countries, in real-time.

That's why we built the COVID-19 data explorer, a simple web application for comparing the outbreak across countries. It's based on data compiled by Johns Hopkins University from the WHO, the American CDC, the Chinese NHC, the European CDC, and the South Korean DXY, as well as news and other reports. You can see the app HERE

The COVID-19 data explorer's main feature is the "Day-zero comparison plot". This is a tool for comparing the epidemic curves across countries. The tool allows you to view growth in cases in one country against another (or multiple others), while "adjusting" where the x-axis should start (ie, what should be considered day zero). For example, one can compare growth between Spain and Italy with day zero set to 1 (ie, the day of the first case), or one can consider that day 0 should be set to a higher number such as 150, which could be considered a "critical mass" of infected individuals, after which it's clear from the chart that the epidemic behaves very similarly.

Let's do a walk-through. The below chart shows cumulative cases of COVID-19 in Italy (black) and Spain (red) since their respective "day zeroes" (ie, the day at which each country had its first case). Clearly, the epidemic took off more quickly in Italy than in Spain.

Visualizing exponential growth is better done on a logarithmic scale. So, we'll adjust the chart input controls to set the y-axis to logarithmic.

Having made that change, similarities between the curves are now more apparent.

A classic "epidemic curve" shows day-specific (rather than cumulative) cases. To see the chart in classic form, we can uncheck the "Cumulative cases?" box...

...which yields the below chart.

It's clear in the above that the slope of the curves for Italy and Spain are similar. The main difference is that Italy seems to have started the exponential growth in daily cases prior to Spain. This is where the "day zero" adjustment functionality comes in. To go from 1 case to 100 cases varies radically across countries; but once a critical mass of infected people exists (say, for example, 150 cases), growth thereafter seems fairly uniform. Let's set that "day zero" value to 150.

Having made this "adjustment", the epidemic in Italy no longer looks so different from the epidemic in Spain. In fact, they look strikingly similar. It took very different amounts of time for each country to get from 0 to 150 cases, but it took exactly 6 days to go from the critical mass of 150 to 1000 cumulative cases. And on Spain's 7th day (after the 150 case "day zero" adjustment), yesterday, Spain reached a cumulative number of cases of 1695; on Italy's 7th day (after it reached the 150 case watermark), it was at 1694 cases. Nearly identical.

We can compare other countries too, of course. Few are as similar to one another as Spain and Italy in terms of the epidemic curve, but the similarities (at least once 150 cumulative case "critical mass" is passed) appear to outweigh the differences.

Explore the data yourself on Databrew's COVID-19 data explorer.

Back to Blogs →