Matthew Christiansen: “What Counts as a Coronavirus Death?”

YouTuber Matthew Christiansen took a deep dive last week into the media’s astonishing dishonesty around the reporting of  Coronavirus deaths. We’ve included a complete transcript of his findings with relevant links and screenshots provided.

In the effort to understand the lethality of Coronavirus, until recently, I thought it was strictly a question of the denominator in that ratio. That is to say, we can be pretty confident about who’s died, of course. Its just a question of how many people have actually had this thing? And without the ability to test everyone en masse, that denominator is really anyone’s guess.

In other words, I had assumed the mortality rate was simply a known quantity in deaths, over an unknown quantity in cases. I had assumed that numerator of Coronavirus deaths was a pretty straightforward one: if someone dies from Coronavirus, then that’s a Coronavirus death. Pretty hard to miss that, count it and carry on.

But what if it’s not that simple? What if the numerator is, in fact, just as murky as the denominator? How useful would the stats be in such a scenario? What if we’re actually in a worst-case scenario, statistically, where we’re both underestimating the number of cases and overestimating the number of deaths? If that were happening, we’d be getting a massively inflated presentation of just how lethal this thing actually is. And I stress that I don’t know that that’s the case, just that the circumstances are present for that to be true.

And I give a shout out to Dave Cullen and his Computing Forever channel for putting together a great, thorough video on this topic that inspired me to look into this issue for myself.

Why is it that the effects of Coronavirus vary so widely from country to country, or even state to state here in the US? Take you pick from the grab bag of possible variables; population age, population density; population health condition, government regulatory response. These are all plausible explanatory factors for why Italy has been absolutely ravaged but Germany has seen nowhere near that total of death, for example.

But thinking of these factors may already be one step too far ahead.

Before we ask, “What explains the measurements we’re seeing?” we need to ask, “Are we even measuring the same thing?” That is the reason we so readily dismiss China’s numbers, after all. We don’t think they’re measuring correctly. We don’t think they’re even measuring at all. I’m not saying Italy is lying like China, of course. I’m saying that even among countries we do trust as doing a good faith job in reporting their data, there’s enough variability in what actually counts as a Coronavirus death, that we need to be careful in comparing data from different places, and in concluding that the mortality rates we see are a final, objective, undebatable truth.

I mentioned the concept of a Coronavirus death seemed like a pretty straightforward one to me. A Coronavirus death is a person who died from Coronavirus. But not everyone conceptualizes it that way.

The Telegraph reports this week that part of the reason the Italian death toll is so high is because they’re not counting deaths from Coronavirus – they’re counting deaths with Coronavirus. A professor and scientific advisor to the Italian government says, “The way in which we code deaths in our country is very generous in the sense that all people who die in hospitals with the Coronavirus are deemed to be dying of the Coronavirus.”

01 Telegraph02 telegraph 2

In other words let’s say you have Diabetes and COPD, you’re already hospitalized for those conditions, you have a lethal stroke in the hospital, you die, and you also test positive for Coronavirus incidentally. In such a case, that counts as a Coronavirus death, even though attributing the death to Coronavirus is pretty tenuous.

Now maybe you think I’m exaggerating. Maybe you think that’s a ridiculous scenario that’s so rare it wouldn’t be statistically significant anyway. Well, it’s not that ridiculous of a scenario when you consider that Italy says 99% of it’s Coronavirus death cases had “other illnesses present” and nearly half of the cases had at least 3 prior illnesses present.

03 bloomberg

And CDC figures released this week show a similar situation in the US, at least preliminarily. Analysis of early Coronavirus deaths in this country with complete information reported to the CDC, shows 94% of those deaths had at least one underlying condition. Having preexisting conditions or complications doesn’t necessarily mean that Coronavirus wasn’t the primary factor in causing or accelerating these deaths, of course. I’m not saying that underlying conditions invalidate these deaths. I’m saying that these deaths aren’t always as simple as clearly Coronavirus-caused, and we shouldn’t treat them that way.

And there’s reason to believe that this complicated and inconsistent measurement internationally is mirroring itself right here in the US. Just like we see different countries adopting different measurement criteria, so too are the states. The total numbers you see in the most widely-cited Coronavirus case count sites, are the aggregation of numbers reported by the state health departments each day. The CDC does have guidelines that say a Coronavirus death should be counted if a person died as a result of Coronavirus. But as this county coroner in a Philadelphia Inquierer story on the issue says, that is a matter of interpretation.

08 phily inquierer

People are trying to standardize something that has no precedent. Anecdotally, we are seeing these interpretations get stretched as far as they possibly can be across the country.

Now, whether these reaches are due to an abundance of caution or whether there’s a political motive in play –whatever the reason – there are plenty of cases that make me wonder just how reliable those death stats on the news actually are. There are repeated efforts to quickly assign Coronavirus as responsible for deaths before the facts are even established.

In Illinois the headline last weekend reads, “Cook County Infant Dies From Coronavirus, The First Such Case Of Infant Mortality In The US.

09 cook county

Again, note the headline language: “Dies from Coronavirus,” even though the details strongly suggest that is not the case here. The infant had a bowel blockage and organ failure, and was hospitalized for those conditions for a month prior. And in fact, the cause of death is not even yet established. And investigation is underway. Despite that lack of clarity the title labels Coronavirus as definitively responsible. And even though the cause of death is not yet established, this infant’s case has been immediately counted among Illinois’ Coronavirus death statistics.

10 infant

Similarly in Connecticut, on Wednesday the Governor announced a pediatric fatality linked to Coronavirus. His words: linked to. Not “caused by,” not necessarily even “diagnosed,” at least according to this tweet. Though later reporting did confirm a positive test result, just “linked to”.

12 linked to

The child was brought unresponsive to the hospital days ago and could not be revived. Was Coronavirus the cause of that unresponsiveness? Unclear.

Just like the case in Illinois, the cause of death has not yet been established. It doesn’t matter though. The Governor is counting it as a Coronavirus-caused death, and the state health department is counting it too.

In Easton, Pennsylvania, a 61 year-old man suffered a fatal head injury after a fall in his home. He died at the hospital, but incidentally tested positive there for Coronavirus. While the cause of death is listed as the head injury he suffered, Coronavirus is listed as a “contributing factor”. Despite being secondary, Coronavirus gets the headline, and yes, this case counted in Pennsylvania’s Coronavirus death statistics.

16 - 61 215 61 yo number two

In New Orleans a 17 year-old boy was originally reported as a Coronavirus death on March 22nd. The Orleans Parish Coroner’s Office cited the boy’s positive test for Coronavirus in making that determination, though as of this reporting, the official cause of death was still under investigation. And the boy’s father said that he was surprised at the suggestion that his son died of Coronavirus, saying no, it was heart failure.

17 New Orleans

We don’t have much more information, but dad disputes the supposed certainty that his son died of Coronavirus doesn’t matter. The state of Louisiana’s practice is to include anyone who tested positive and died in their tally of Coronavirus deaths. This reporting notes that it is possible that people on the list died as a result of another condition they had, in addition to Coronavirus.

19 heart failure

In Southern California another 17 year-old boy was originally counted as a Coronavirus death. He died in March in Lancaster outside of Los Angeles. The story was first sensationalized as possibly the first Coronavirus death of a minor in the country. That is, until they actually bothered to look at the specific facts of the case, discovering there were “extenuating circumstances that pointed to an alternative diagnosis as well”.

21 lancaster22 extenuating circumstances

23 mayor doesn't doubtAnd so his case was removed from the official count despite the mayor saying he doesn’t doubt that the teen died from Coronavirus. Okay. Well, the medical professionals do doubt, Mr. Mayor. So what matters more? The truth or the politics?

There are other faulty cases that may be as simple as clerical error. Who knows. But again, these are mistakes that matter.

In Florida, a Pasco County man says reports of his death are greatly exaggerated. The State Department of Health reported his death on March 20th. That was news to him since he was sitting at home recovering. The Department of Health blamed an error in the reporting system.

24 pasco

Another death case in Leon County just disappeared on March 22nd. A 48 year-old Georgia woman was reported as the county’s first Coronavirus death and was originally included in the State Department of Health’s map, the scrubbed, then remained on the case list but was missing from the death list. The state is saying the confusion happened because she wasn’t a Florida resident. She was from Georgia so she belongs in Georgia’s count, not Florida’s.

25 leon

Fine. Fair enough. I know this is a rapidly developing thing so I’m not going to take every little mistake as evidence of some big conspiracy or political opportunism or whatever else. But another factor that bugs me is even where states are trying to do the right thing and be as precise as possible, we are still taking their worst-case data instead of their precise and confirmed data.

In Alabama, for example, state officials keep two different counts on Coronavirus deaths: reported deaths and confirmed deaths.

Reported deaths are people who died and had a positive Coronavirus test result. Confirmed deaths are those whose cases have actually been evaluated by a physician and infectious disease outbreak staff, confirming that the death is primarily attributable to Coronavirus. As of Thursday morning, according to the state’s website, Alabama had 32 reported deaths (as in possible Coronavirus deaths pending review) and 17 confirmed deaths. But if you went to the popular aggregation sites at the same time, those sites are using the count for the possible cases pending review, not the confirmed cases.

And I would bet that use of the higher unconfirmed number for the purpose of counting the national total isn’t limited to just Alabama.

Now for those of you already typing your angry Email to me, before you smash caps lock and type away and hit send, understand exactly what I mean here. Do these questionable cases mean that Coronavirus is no issue at all–nothing to see here–no problem–move along? No. Of course not. I am not a medical expert and I would never encourage taking medical advise from me. Just because these cases are sketchy, does it mean that all cases are sketchy? No. It doesn’t.

It just means that the obviously sketchy cases are numerous enough to warrant at least a little skepticism about just how valid those big scary numbers you see flashing on the news actually are. Particularly when those numbers often come with some sort of solution that you’re just supposed to accept automatically.

“Ooh, the numbers are big, so you’d better surrender your rights.”

“Ooh, the numbers are big, so you’d better accept big daddy government as your lord and savior.”

“Ooh, the numbers are big so you better live in fear and let us make decisions for you.”

And these numbers may be big in the end. I don’t know. We’ll find out. But let’s take a moment to evaluate them for ourselves before we automatically grant that as true. And while we’re at it, let’s take another moment to wonder why there are so many forces intent on making these numbers as big as they possibly can. Is the truth what they’re after, or is it something else? You’re not a bad person for asking.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s