Forecasting Long-term Wuhan Coronavirus (nCoV-2019) Mortality Rates

What's at stake

Over the last few days, I’ve received some requests to expand the long-term forecast for infections to include mortality. And, to be honest, this one was really hard for me to do.

It was straightforward to put together a simple forecast. But thinking about what it represents has been really hard. While I don’t think this is the place to write extensively about what even one death means, or potentially hundreds or worse, I did think it warranted a mention given just how serious this situation is and should continue to be taken.

That side of the story now being said, let’s first look at the long-term forecast that was updated in yesterday’s post:

This is a very long-term forecast for a situation that is very uncertain. It has been fairly accurate the first few days but that isn’t too difficult to estimate. The value in this forecast is more the shape of the curve versus the final ending number, as it follows a shape one might expect from the nature of a virus spreading.

In this forecast, the 90 thousand tally for infections is, in some ways, arbitrary. However, having some target, even if too high or too low, gives a benchmark as we continue to watch the Wuhan Coronavirus situation develop.

Let’s next consider a new forecast for the mortality rate:

As I have written before, it’s important to be humble when making a forecast, as the future is impossible to precisely predict. A forecast is a tool. And, when something as precise as a mortality rate has to be estimated, various scenarios can be useful as opposed to one single forecast. So, that’s what I chose to do here.

I decided to consider a scenario where the mortality rate stays flat at 2%, which is fairly close to the current mortality rate as of writing. I then considered a scenario (think of it like a worst case scenario) where the mortality rate starts to climb, as it did with SARS, and eventually converges to be equal to the SARS mortality rate. Then, I did an average of the two scenarios in the middle.

I didn’t consider a scenario lower than 2% because mortality is really a measure of the downside. Everyone will be happy if the mortality rate is less than the 2% rate! That is what we are hoping for, less deaths. Anything that turns out better than the current observed rate will be welcome.

We then come to the final step of the forecast, which is multiplying the forecast of infections with the forecasted scenarios of mortality:

I think it goes without saying that this prediction is incredibly subjective. Any time a forecast is made that is one thing. But this is a forecast times another forecast. The uncertainty is much greater and the outcome much more sensitive on the assumptions. Still, it provides a useful tool as we track the outbreak.

In the case that the mortality rate stays flat and deaths top out around 90 thousand, the Wuhan Coronavirus will claim twice as many victims as SARS. However, if the disease were to end up as deadly as SARS, it would mean ten times as many victims as SARS. That seems very unlikely. The death rate is uncertain but I haven’t heard anyone claim that the Wuhan Coronavirus is anywhere near as lethal as SARS was.

Remember, this forecast is a function of the total infections and the mortality rate. The two variables could be decreased (increased) and have a drastic decrease (increase) on the outcome of deaths. There is also a mix in between if the input functions move in opposite directions. So, a lot of variability and something to watch as things progress.

As a final point, one could also just make a direct forecast for deaths in the same way I originally forecasted infections. The issue with that approach is that the mortality rate would still exist as a function of those two forecasts and then have its own level of uncertainty. There is no way around it. It’s just a matter of what angle the forecaster wants to take to tackle the problem.

In short, let’s continue to hope that the outcome of total infections and mortality is much much lower than any of these forecasts above. I will continue to be watching closely and writing as developments warrant new discussion.


Enjoy the article? If you haven’t already, subscribe for free below and never miss a post. No spam. Cancel whenever!


Related content:

Pandemic? It’s still early…

Forecasting Coronavirus Cases

Forecasting Long-term Wuhan Coronavirus Cases

Helpful Coronavirus Resources

Five Useful Coronavirus Charts