Wielding a Red Pen: Correcting a Fear-mongering Ebola Piece with Facts

If you catch me on Twitter, or read the fantastic Red Ink, you might have seen my corrections and edits to the first page of a genuinely awful, fear-mongering piece on Ebola that was inexplicably published by Pacific Standard.Per policy, I won’t drive traffic to horrible pieces. You can find it on your own relatively easily. You might have also realized why:

  1. I was forbidden from grading in red ink when I TA’d (“did you dip that in red ink?”);
  2. I was consistently voted most likely to become a doctor or teacher in those elementary school “most likely” contests.

Sorry about that. Well, at least the second one; handwriting has never been my strong suit. Due to said possibly challenging handwriting, I figured I would go ahead and expand on my comments here.Okay, most of this is taken from a Facebook rant the other day that accompanied a snapshot of the edits I did. I’m not sure if this counts as self-plagiarism or self-citing.

By What Definition of Worse?

Red Ink was kind enough to host the full view of this; please give the clicks thru for their page counts. Thanks!

Red Ink was kind enough to host the full view of this; please give the clicks thru for their page counts. Thanks!

This piece starts outAuthors don’t always set their own titles, but beyond that, as I explained this morning on Twitter, I feel that an editor’s job is to fact-check, demand accuracy, and kill projects that are inaccurate, get the facts wrong, etc. Since it’s not possible to look at a finished piece and know which errors are there because of the author and which are editorial, I am keeping my criticisms vague and at the publication as a whole. with the title “The Scariest Virus: Ebola is Back, and It’s Worse than Ever.” It’s always nice when you don’t have to read much before you start reaching for your red pen.

In this case, the big question is: by what definition of worse? As I noted, is it the one where worse means “less deaths than normal for the Zaire strain of Ebola?” A the time the article was published (August 2, 2014 at 5:26pm), the Case Fatality Rate was approximately 55 percent.Once again, I am relying on Dr. Ian Mackay’s excellent blog for these numbers and other general information. All mistakes are my own. Unless they’re things he told me, in which case I’ll point that out when you correct me and let you fight it out with him. …my money’s on him, though. This is significantly less than the typical CFR for the Zaire subtype of Ebola, a fact pretty easily checked by looking at the Centers for Disease Control and Prevention Ebola outbreak table, which charts Ebola outbreaks by years, country, Ebola subtype, reported number of human cases, the reported number of deaths among cases and the associated case fatality rate, and outbreak specific details. And a quick glance at the chart tells us that the average CFR for the Zaire subtypeThere is some question as to whether the current outbreak is the Zaire subtype–it has both been identified as a new subtype and the existing Zaire subtype–but until more conclusive papers are published, I’m staying with Zaire, as apparently are most people. (listed in the chart as “Ebola virus) is 78.48%.

Now, I’m not the best at maths,Confession: I’m so badly not the best that I consulted with my husband just to make sure I was doing this simple concept of figuring out CFR correctly. Spoiler: I am. Now. but even I can see that 55%Or even the roughly 56.7% that the CFR is today, according to the CDC. is not worse than 78.48%.

Well, how about if you’re just talking about Ebola in general, regardless of subtype?All Filoviridae, which includes both Marburg subtypes, will run about 66%. This does not include cuevaviruses, which have not yet been formally recognized as part of the Filoviridae family. It wouldn’t matter if it did, though, since as far as we know, the cuevaviruses only affect bats.

Even if you do the math to include all cases, including the laboratory acquired infections, Tai Forest, and Reston,The Reston subtype, likely make famous Richard Preston’s book The Hot Zone, does not appear to be symptomatic in humans. There’s only been one known Tai Forest subtype infection, and while the researcher developed dengue-like symptoms, she survived; we don’t really have any idea of how dangerous it might be for humans. It appears to decimate chimpanzee populations, though. And the LAIs are questionable to include only because needle-sticks have the potential to deliver a more concentrated viral load than you’d receive “in the wild.” the CFR is 62.92%. In fact, the only way you can make a case for this being “worse” than something that’s already happened is if you only look at the cumulative CFR for the Sudan subtype, which is 46.97%. But making that decision would be cherry-picking and utilizing what is seemingly simply wrong information.

How Does This Math Even Work?
Skipping my more editorially-focused complaints, such as a reliance on The Hot Zone (see my final paragraph) brings us to a really strange claim about Yellow fever, Philadelphia, and Ebola. I’m actually going to quote it, because, as I noted, I’m honestly not sure how this math works.

Yellow fever may have wiped out more than 10 percent of Philadelphia’s population in 1793, but that stunning death toll is nothing compared to the devastation that Marburg and Ebola wreak.

Really.

The 1793 Philly yellow fever epidemic killed 5,000 people in four months. So far, in 38 years, not including current deaths, Ebola and Marburg (all subtypes) have killed around 2933 people. Including the current outbreak, approximately 4700 people have died.Per ProMED Mail. That’s total. Ever. It’s a lot, but do you know how many people Yellow fever kills a year?

30,000.

A year.

In 2014.Figures from WHO.

There’s definitely a stunning death toll and incredible devastation being wreaked, but it doesn’t involve a filovirus.

Maybe you want to give Pacific Standard and their author the benefit of the doubt,and argue that certainly when it comes to numbers infected and people who die, it’s the filoviruses that should scare the pants off us, because they’re fatal 40-90% of the time, and Yellow fever is only fatal 3% of the time (or 20% in the toxic cases).

Well, no.

The thing with mortality is that people talk about mortality rates without also discussing morbidity rates. That is, yes, when 1700 people are infected and 900 people die, there’s a pretty high mortality rate that’s eye-opening and definitely a thing to worry about if you’re near the outbreak. But is 60% fatality all that “scary” when the infection rate is 10 people? In general, people view that differently.

This, as my husband likes to point out, is why flu scares the pants off people in the know. The 1918 H1N1 Spanish Flu, for example, only has a case fatality rate of 2.5%… but it infected half of the world. In that case, the morbidity is off-the-charts high, but mortality is really low-when you can call 50 million deaths low. You need both numbers in order to truly understand the “scary” level of a disease.For Yellow fever, we’re looking at something like 200,000 infections a year, and 30,000 deaths. That gives it a CFR of around 3%.

Facts: They Mean Something
The last bit of the red ink spilled on this front page that I’ll address hereI addressed the “we don’t know where it comes from!” issue, which is contradicted in the subsequent paragraph, in this post. was correcting simple figures and facts that are easy to verify, so it’s hard to understand how they made it into the story at all, and make me wonder about everything I’ve ever read at Pacific Standard.That’s how it works. Tank your reputation for facts, have everything looked at with suspicion. What makes this particularly painful is that the data listed doesn’t match the cited source; either the author or her editor didn’t take the time to compare the two. It’s incredibly sloppy.

Let me show you what I mean. Here’s the paragraph; text in red is wrong.

Two hundred and eighty of the 318 people who contracted the virus died during the first known Ebola outbreak in what’s now the Democratic Republic of the Congo in 1976. That’s an 88 percent fatality rate. The same year, a less virulent strain appeared in Sudan: 284 infected, 151 dead. It was 20 years before the next outbreak: In 1995, Ebola infected 315 and killed 250—a 77 percent fatality rate. From 1995 to 2012, the Centers for Disease Control and Prevention recorded a series of distinct Ebola epidemics in Gabon and the D.R.C. claiming 621 lives, their fatality rates roughly between 50 and 90 percent.

If you are looking at the CDC outbreak table that is actually cited, you’ll find that after the initial 1976 dual outbreaks in Zaire (now the Democratic Republic of Congo) and Sudan,Where the names of each subtype come from. there was a death from the Zaire subtype in Zaire in 1977 and a small outbreak of the Sudan subtype in 1979. Even excluding the Ebola Reston fiascos of the late 1980s and 1990s (although if you’re using The Hot Zone as your model, I don’t know why you’d do that), the next Ebola outbreak in humans is in 1994,There was also an outbreak in chimpanzees in Côte du Ivorie in 1994; a researcher was also infected, although she recovered. This strain is now known as the Tai Forest subtype. in Gabon. Fifty-two people were infected and 31 died.

There were a series of six distinct Ebola (Zaire subtype) infections in Gabon and the D.R.C. between 1995 and 2012; maths from the CDC website suggests that 773 were infected and 571 died (CFR: 73.86%). However, beyond the basic maths problem is a selection problem. Between 1994 and 2012, the CDC records 17If you look at the CDC chart, the October 2001-March 2002 outbreaks in RoC and Gabon are actually the same, as it’s crossing the border between the two countries. So while they are listed separately, it’s the same outbreak. distinct series of outbreaks in the African countries of: Gabon; the Democratic Republic of Congo; South Africa; Uganda; the Republic of Congo; and Sudan.There was also, as noted above, the Tai Forest outbreak, infecting one human.

The accurate version of this paragraph would read:

Two hundred and eighty of the 318 people who contracted the virus died during the first known Ebola outbreak in what’s now the Democratic Republic of the Congo in 1976. That’s an 88 percent fatality rate. The same year, a less virulent strain appeared in Sudan: 284 infected, 151 dead. It was three years before the next outbreak: In 1979, Sudan ebolavirus infected 34 and killed 22. The next large outbreak was in 1994, when Zaire ebolavirus infected 52 and killed 31—a 60 percent fatality rate. From 1994 to 2012, the Centers for Disease Control and Prevention recorded a series of distinct Ebola epidemics in Gabon; the Democratic Republic of Congo; South Africa; Uganda; the Republic of Congo; and Sudan, claiming 1092 lives, their fatality rates roughly between 41 and 89 percent.My changes are in purple.

Illustrating the Point. Literally.
Sadly, I could have gone on correcting this article; it does not get appreciably better after the first page. But hopefully a single-page correction illustrates the point.

What is that point? At it’s most simple, it’s that you do actually have a responsibility to facts when you’re writing. If a writer’s editor accepts sloppy and loose-with-facts stories, well, okay, but that editor’d best fix that sloppy and loose-with-facts story before hitting publish. There’s probably also a point in there about who you hire when you want a science-based story written, about the need for thoroughness over speed,This, for example, took me nearly five hours to write because I fact-checked my numbers–and then ran them by someone else. and about fear-mongering for page clicks and points, rather than a responsible dissemination of data. But I’m already at nearly 2100 words.

I’ve been asked, for the last week or so, if I would write a post about why The Hot Zone is a bad reference for journalists to make, as this Pacific Standard article does. This will probably be my only post even related to that topic, because the reality of the situation is, The Hot Zone is between 20 and 27 years old. Yes, it claims to be based in reality and truthA claim some people challenge., but if you want to accept at face value that it is, then you’re using a resource that’s decades old and significantly out of date. Citing a paper that old, when there are literally libraries worth of newer information, wouldn’t be allowed in homework assignment, let alone an academic paper, and shouldn’t be allowed in anything referring to itself as journalism.

It’s almost as sloppy as not bothering to fact-check your work.