Explain the math of Ebola to me

[elfkin477]

This is probably not really a math problem, ftr.

I’m thinking specifically of this article, but I’ve seen many others with similar numbers.

 5,100 **dead**

• 14,413 additional infected
  = 70% mortality rate

How does that figure?
5100 out of 19513 is 26.14% isn’t it?

So what other factors are they using as a metric to claim almost three times as high a percentage as the mortality rate?

[Nametag]

IANAEpidemiologist, but I expect that many of the additional infected are from later cohorts than the dead, and many of those will die in the near future.

[Iggy]

You have to exclude those currently infected when calculating a death rate as you don’t yet know how many will live and how many will die.

[DSeid]

An article explaining it.

Using data supplied to the WHO, the researchers came up with the 70 percent estimate using a subset of the number of Ebola cases as of Sept. 14 in four main Ebola-affected countries — Guinea, Liberia, Nigeria and Sierra Leone — with a known clinical outcome.

The “known clinical outcome” is key. When they calculated the estimate from the ratio of all reported deaths to all reported cases, the death rate was lower than that of past Ebola outbreaks and varied widely among the West African countries hardest hit by the virus.

When they used only 46 percent of the cases that had a definitive recorded clinical outcome, they got a higher estimate — 70.8 percent — that was consistent among Guinea, Liberia and Sierra Leone.

[elfkin477]

@Iggy

You have to exclude those currently infected when calculating a death rate as you don’t yet know how many will live and how many will die.

Too bad the articles don’t break out how many were infected and survived versus who is currently infected and might yet die of it.

[Magiver]

@elfkin477

Too bad the articles don’t break out how many were infected and survived versus who is currently infected and might yet die of it.

The mortality rate changes as medical help is applied. So it would have a mortality rate of 70% if left untreated and a lower number based on whatever treatment was available.

[Ravenman]

The key fact here is that the numbers in the OP are “reported” deaths and infections. People may report deaths but not infections, other may be reported as infected but then the disappear and the death may not be reported.

CDC attempted to model the actual number of infections, and once corrected for underreporting, the numbers look pretty dismal. Link. For example, Liberia was reporting a cumulative number of roughly 7,000 cases since the outbreak began, but the estimate of the true number of cases was three times that.