The Shape of Things to Come

I had a very interesting conversation yesterday about how governments were planning for the future and what the loading would be on the health services once the level of new cases goes down. Of particular interest is the rate at which cases decline and whether what remains in due course is a very low number that approaches zero or whether a more significant background trace remains.

So following on from this idea I've found a few examples where new cases have reached a peak or plateau and there are clear signs of a decreasing level of new cases. I've used these as analogues to determine if any common themes occur.

Here's the chart showing the seven countries with declining rates of new cases. The data has a small amount of smoothing applied to make the trends easier to see, the data is then indexed to 100% at the average number of new cases in the period around the plateau and the last day of the plateau for each country is set to day P0, PP1 is the first day post-plateau etc. This way we can observe the relative rates of decline across multiple countries at multiple time periods on the same chart.

There are some interesting shapes appearing post plateau, but clearly there is some variation in the degree to which new cases slow down, but it does suggest that daily cases could fall significantly 7 days after the plateau has finished.

It's worth commenting at this point about South Korea where the level of new cases did not so much plateau as have a sharp peak of 1062 on 29th February and a sharp decline thereafter. We have more data for this country post peak than almost any other and there are still new cases being reported now - the number held to about 100 per day for about a month, but has fallen to about half this level in recent days. This might provide some insight as to possible run-out scenarios elsewhere, but as an n=1 sample it may not be too representative give all the other potential factors in play.

To get a better understanding of the data I've shunted a few countries backwards in time to see how the lines compare, in the chart above, Australia, Israel, New Zealand and Korea have similar shapes, so this allows me to look at the data in an overlapped fashion. This provides the chart below:

Clearly this demonstrates the similarities in the trajectories for these four countries with a 60% reduction in cases in just 4 days, whilst Spain, Italy and Iran seem to be on a very different curve with 40% reduction in new cases in about a week. So potentially, we have two segments emerging.

Interestingly, the former group of countries have relatively low numbers of daily cases at peak (80-1000), whereas the latter countries were had many more daily cases at peak, (3000-8000), which might be a factor in determining which curves we start to see once other countries have passed through the plateau in new cases.

It will be interesting to see how these curves develop in the coming weeks and how other countries map onto this chart in due course.