- JoNova - https://www.joannenova.com.au -

Two Californians doctors say the mortality rate is like the flu, but it was sampling bias instead

A few people are asking about the video the Youtube removed which is now on Bitchute. (h/t AC Osborne)  The cack hand of Youtube strikes again with unnecessary censorship. But even if the docs were wrong, it’s better to discuss why than try to disappear them — we can all learn something. Plus the censorship gives them a de facto kind of hero status among some, but for the wrong reasons.

The Doctors main point is a sampling bias error

Doctors Dan Erickson and Artin Massihi of Accelerated Urgent Care are ER doctors, they look convincing, and speak well, but their thesis rests entirely on an error. They take test results from a high risk group of people, and extrapolate the results to the whole state. What they have is non-randomized data, and they assume it represents millions of people who didn’t come out for testing. This is not a small or incidental point. It’s the foundation of nearly everything they say.

They repeat this same mistake over and over, and say “we are following The Science”. They come up with some just-so rationalizations, some truisms that aren’t true, and inject some superfluous vocabulary in there so people know they have studied virology once. (Yes DNA does mean Deoxyribnucleic acid, but these viruses are RNA ones, and none of that jargon needed to be there.) They have virtually no data that supports any of their other claims, and they simply ignore, or aren’t aware of the properly randomized studies that don’t agree with them. In Austria only 0.3% had a current infection at that time, so probably only 1% of the total Austrian population have immunity (if it exists).

UPDATE: How Sampling Bias makes the mortality rate appear too low: If people who have a disease are more likely to go out and get tested, then the sample they provide will overestimate the true rate of infection, which leads to an underestimate of the mortality rate.

People without Covid are less likely to want to get tested — they who have carefully self isolated are unlikely to rush down to the testing stations where they might catch coronavirus. And if they think they are negative they won’t want to pay for a test either.

Imagine 1% of the population is infected with disease X but a company sells tests for X, and it attracts 100 people who have had some kind of cold or flu and who think they had X. In the end 20% of them turn out to have X, the rest had a normal cold. The researchers might claim 20% of the whole state have X (which would be wrong). The real rate of infection is still 1%. And if the mortality rate was 5% (of those infected), after the number of infections is accidentally bumped up from 1 to 20% that makes the death rate appear to fall from 5% to 0.25%.

Nothing can rescue them from the sin of extrapolating from small unrepresentative data

Essentially they say their tests of 5,000 people in Curran Country found 6.5% were positive “of the population” they say (erroneously). This is “widespread”. They point at testing throughout California which returns a 12% positive and calculate that people have a 0.03% chance of dying in California. Which is misleading and wrong. Careless.

33,865 Covid cases out of 280.000 tested. That’s 12% of Californians infected. We have 39 million people. That’s 4.7 million cases.  We’ve seen 1,227 deaths, that means you have a 0.03 chance of dying….

At one point, Erickson goes on to say almost 96% of people recover without any significant sequelae, as if this is good news — but if 4% do not recover isn’t that a problem? (By his own bucket maths that’s up to 1.5 million Californians that face a threat they might not completely recover from, and who might carry long term health issues — assuming they don’t die.)

He repeats like a mantra: millions of cases, small amount of death.

At the core of their message, the two docs have a message that the high positive test rates in NY, Spain and their own work support their theory that millions are infected, and therefore the death rate is tiny, so we can drop the lockdown. But a high positive test rate mostly  means a state isn’t doing enough testing. It is not a good proxy for the prevalence which needs random tests. ( See The Atlantic on high positive test rates. ) Obviously current testing is missing all the people who have isolated at home, or had no symptoms at all and are less likely to front up to a testing clinic and pay money to find out they probably don’t have coronavirus.

Doctors Dan Erickson and Artin Massihi of Accelerated Urgent Care are ER doctors, and they run a business doing these tests. So we’d think they would be aware of sampling bias.  And since their business would profit from lifting the lockdown, we’d think they’d be very careful about putting out reasonable caveats when they spoke in a press briefing, yet they didn’t — even when reporters asked them why their message conflicted with so many other doctors.

Possibly they didn’t realize they were jumping into a global fire. After the video went viral, and Elon Musk raved on twitter to his 33 million followers,  many other doctors came out and explained why it was wrong.

Here’s one:

Two California doctors go viral with dubious COVID test conclusions

By Calmatters, Mercury News

The doctors should never have assumed that the patients they tested — who came for walk-in COVID-19 tests or who sought urgent care for symptoms they experienced in the middle of a pandemic — are representative of the general population, said Dr. Carl Bergstrom, a University of Washington biologist who specializes in infectious disease modeling. He likened their extrapolations to “estimating the average height of Americans from the players on an NBA court.” And most credible studies of COVID-19 death rates in reality are far higher than the ones the doctors presented.

“They’ve used methods that are ludicrous to get results that are completely implausible,” Bergstrom said.

Just a bit of junk science: ….”we’ve never seen a situation where you quarantine the healthy”.

There were lots of other signs they hadn’t thought about their theory too hard. Right at the start Erickson declares “Typically you quarantine the sick, and we’ve never seen a situation where you quarantine the healthy”. Yes, well, that’s except for all of history. The use of quarantine goes back to biblical times, and often meant isolating people who might have a disease whether or not they were sick. That was the point, we quarantine healthy pets that fly overseas, not because we know they are sick, but because even asymptomatic dogs can carry nasty diseases. Obviously, if we knew who was healthy and who wasn’t, we wouldn’t be quarantining the healthy ones. See, for example, American Samoa where they escaped the Spanish Flu by insisting all boats wait for 5 days offshore. Quarantine has to work this way if the disease spreads asymptomatically.

Erickson et al have some fair criticisms of lockdown, and the hypocritical, contradictory ways governments have manhandled the situation. But plenty of other more careful commenters are saying that already.  Let’s have that conversation. But don’t mix up the basic success of isolation measures with the other debate on which kinds of isolation are the most cost effective, with the other debate about how pandemic preparedness plans were a disaster, we didn’t have the PPE (personal protective equipment) and we had to close off elective surgery because all the mask and drug factories had moved to China.

It’s not the flu

There’s no equivalence comparing simple death numbers after 2 months to a whole year of flu statistics, yet they do it:

Drs. Dan Erickson and Artin Massihi, who co-own Accelerated Urgent Care facilities in Kern County, said, “The death rate of the coronavirus is similar in prevalence to the flu,” according to ABC 23.

“If you study the numbers in 2017 and 2018, we had 50 to 60 million with the flu,” Erickson said. “And we had a similar death rate in the deaths the United States were 43,545 — similar to the flu of 2017-2018. We always have between 37,000 and 60,000 deaths in the United States, every single year. No pandemic talk. No shelter-in-place. No shutting down businesses.”

In 8 weeks, half of which was under the most severe isolation measures USA has ever seen, coronavirus has already killed as many people as the worst flu season ever. Imagine what it might do, if there were no efforts to slow it down?

When reporters ask why their view is so different to almost all the other experts they vaguely mention “theories and models”, and “how they are working with patients” not being desk-bound. This was when a good science team would have talked about the contradictory studies and said something that gave an insight in to the inadequacy of the other tests. Instead we get nothing but motherhood cliches, and sometimes one cliche bites the next one. Like when they discuss New York, they explain they are right because they “don’t have to live in Manhattan to look at the data”. So sometimes being a desk jockey is OK after-all. 

A few days later, Doc Erickson has backed down somewhat when asked again on Fox News. Dr Erickson now claims he was just was answering questions of raw data which is quite obviously not the case. He now agrees it wasn’t randomized and claims they weren’t making a biostatistical comment, when they were. He got it wrong. His mortality calculations in the first video were his main point, repeated throughout. They were the reasoning behind everything he recommended, but they were junk. He had claimed to calculate the prevalence, and he knew he was extrapolating. Erickson is just shifting the goal posts. After the stinging criticisms, he’s been pushed back to just saying the banal truth that this is a balancing act between conflicting interests, something we all knew.

Erickson and Massihi probably never intended to go viral, they didn’t think it through, just popped it out there, but it fell in an incubating vat and grew into a gargoyle. We all want Coronavirus to  get back in its box. But being skeptics we seek out the opposing view.


Things worth knowing about Coronavirus:

6.6 out of 10 based on 61 ratings