Is COVID-19 getting better or worse?

Notes

Is COVID-19 getting better or worse in your part of the world?

It’s difficult to exaggerate how important it is to know whether the COVID-19 situation is getting better or worse. If it’s getting better – if there are fewer deaths each week – then the measures we’re taking are sufficient to keep the virus in check, for now. If it’s getting worse – if there are more deaths each week – then we urgently need to take more stringent measures to contain the virus.

This chart aims to make it clear. Select your country or region from the drop-down. If the line for your country is in the red zone, things are getting worse (towards the top of the red zone, a lot worse: ten times more deaths week-on-week). If it’s in the green zone, things are getting better.

This crucial information is impossible to glean from most COVID-19 charts.

Take this chart from the COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU):

Or this tiny chart from the Government of Canada’s Coronavirus disease (COVID-19) Outbreak update:

Like most COVID-19 charts, whether from governments, universities or newspapers, these soar alarmingly up and to the right.

There’s a reason for this. Whether their publishers make it explicit or not, these charts show cumulative numbers of cases or deaths, which will always be up and to the right.

To understand how horribly misleading this is, imagine I ran a business with declining sales: suppose I made $950,000 sales in January, $900,000 in February, $850,000 in March, and so on down. Clearly, I’ll be out of business pretty quickly, as my sales chart shows:

However, if I wanted to change the poor impression given by this sales chart, perhaps to impress potential investors, I could simply change it to show cumulative sales:

This is a much more impressive chart: up and to the right! I’d be sure to win over those investors! The trouble is, it’s fraudulent. By showing cumulative sales, I’m counting each month’s sales over and over again: in February, I’m including the sales for both January and February; in March, I’m counting the sales for January and February and March. It’s up and to the right because cumulative charts are always up and to the right: even when my sales have fallen to zero and my business is bankrupt, it’ll still show impressively high sales.

It’s the same with the COVID-19 charts: even when the last case of COVID-19 is a distant memory, the charts from Johns Hopkins University and the Government of Canada will still show alarmingly high numbers of cases. Like my up-and-to-the-right sales chart, these COVID-19 charts are deeply misleading: when COVID-19 is over, the charts should show the number of cases falling to zero, not remaining forever at a high plateau.

There are other problems with these COVID-19 charts. Take this chart of daily death tolls from the Financial Times, whose egregious cumulative COVID-19 chart went viral earlier in the pandemic:

This non-cumulative chart is much better than the FT’s earlier efforts (though the fact that I’m still talking about these sensationalist earlier efforts is perhaps an indication of why newspapers are so irresponsible in their portrayal of data; universities and governments, too, enjoy the attention such sensationalism brings).

However, it’s still very difficult to tell from this chart the extent to which COVID-19 is getting better or worse (unless you’re really good at assessing the gradient by eyeballing a curve). Further, by focusing on absolute numbers, charts like this one make COVID-19 look far worse in larger countries like the US than in smaller countries like Italy, when the reality is the reverse: Italy has had far more deaths in proportion to its population than the US.

By plotting not the numbers of COVID-19 deaths but how much better or worse those numbers are week-over-week, I hope to give a more immediate impression of how the disease is progressing over time.

If it hasn’t already, I hope that your country dips deep into the green zone of my chart soon.