Thread by @tomaspueyo

Is Sweden a recent coronavirus success? I get asked that a lot.

Here's my take on it, and thoughts on how to science along the way.

Esp relevant since Anders Tegnell, Sweden's chief epi, will speak in SF today, where I live. What should ppl ask him?

Thread [1/

First, let's look at cases and deaths. Clearly, both are down, and deaths faster than cases. Awesome!

We can see it in the Case Fatality Rate, which at ~2.5% is only slightly higher than in countries like South Korea.
[2/

And positivity (share of tests that turn out positive) is ~4%, which means the country has a good grasp of the situation. Good job!
[3/

The question becomes: Why?
Why have cases gone down so much?
Why has the case fatality rate gone down?

The answer is important to then answer the more important question: What will happen this fall?

[4/

If the drops are due to good mgmt, cases and deaths won't go up anymore. If they're due to luck, cases and deaths will go back up.

As you can imagine, the defenders of the Swedish strategy will say it's good mgmt.
[5/

A good example of that position is in this article. Summary:

"The fact that cases are down is most likely explained by the fact that Sweden achieved herd immunity, which is what we tried all along. We succeeded."

[6/
sebastianrushworth.com/2020/08/04/how…

The only problem with that is... That's not what data says.

Swedish officials projected that 26% of the pop would be infected by May 1st. Instead, the rate was just…5.4%, half of these likely false positives.
[7/
medium.com/@tomaspueyo/co…

You'd imagine if antibodies had gone up substantially since, they would have published that data. But they haven't published seroprevalence since (AFAIK), even as they've kept testing for it...

So how can you state that herd immunity has been achieved?
[8/

They're using the new insight that maybe immunity comes not from antibodies, but through T-cells. The idea is that the brunt of the immune response is through T-cells, for which we can't measure immunity well yet, but somehow antibodies are not usually developed
[9/

The author proposes 1 hypothesis and immediately says it’s the right one.
Specifically, it would require:
1. T-cell-based immunity is the one making all the work, despite the fact that we can’t know this today
2. Immunity is invisible through antibodies

[10/

Although there is support for an important role of T-cells in the immune response, there isn't much around the lack of antibody presence in immune ppl. Some heavy-hit coronavirus areas have reached antibody prevalences of ~60%.

Could be. But unlikely to be THE cause.
[11/

This reminds me of the reason why medicine believed for thousands of years, since the times of Hippocrates and Galen, that illnesses were an imbalance in humors, which drove common treatments like bleeding and not cleaning pus.

How could they reach that conclusion?
[12/

They thought the way to achieve the truth in medicine was observation ➡️ interpretation ➡️ treatment.

The observe the facts, they come up with the interpretation that best fits the facts, and they then treat accordingly.

The pbm is that process is missing crucial steps.
[13/

Observation ➡️ hypothesis ➡️ test hypothesis ➡️ if proven, conceive treatment ➡️ test treatment ➡️ validate and use treatment.

But this is not what the MD who authored that article does. He looks at evidence and comes up with his best interpretation of the evidence.
[14/

Surprisingly, that conclusion fits his preconceived idea: Sweden must have the right strategy, therefore any data should be seen under that light.

This is confirmation bias.

It is the main reason why you test your hypotheses.
[15/

How should we approach this instead?

What the data is telling you is:
1. Because the Case Fatality Rate (CFR) is ~2.5%, it is now as expected, thanks to proper testing.

[16/

2. R, the reproduction number (number of new infections caused by one infected person), is probably hovering around 1: cases are stable and aren't going up or down.

R was below 1 in June-July, so cases went down too, but it's now above 1 again.
[17/

So the more relevant questions become:
What was R0 in Sweden (at the beginning, without measures)?
Why did it go down below 1 in June-July?
Where is it going next?

Here's my take on these:
[18/

For a variety of reasons Sweden has a lower-than-average R0: less gregarious society, natural social distancing, 50% ppl live alone at home, healthier-than-average, low # of incoming travelers from infected countries...

[19/

Add to that:
- Many Swedes still working from home
- Many traveling abroad right now
- Lots of outdoors activity (park visits are at all time highs, historic heatwave)
- Fewer businesses prone to super spreader events

electroverse.net/europe-battles…

[20/

- Some ppl already immune (probably 10% at least, which would reduce the normal R0 by 17%)
- Some T-cell immunity
- Some physical distancing (masks, etc)
- Lower # of infected foreigners traveling in
- Likely more reasons
and you can easily have gone below R=1.

[21/

When ppl jump from the lower case data to concluding the Swedish strategy was a success, they're basically saying that, out of all these factors, the one doing all the work is T-cell immunity.

[22/

If they're right, cases will keep going down and R will stay below 1.

If it's more some combination of the factors above, R will switch back above 1 and cases will go back up during the fall.

(R is already above 1 btw)

[23/

So if you get to ask Mr Tegnell questions, ask:
1. Why has Sweden stopped publishing data on antibody prevalence?
2. What level of immunity do you think the country has reached by now?
3. If he answers high, what evidence does he have beyond cases going down?
[24/

4. Will there be a 2nd wave?
5. How many deaths can Sweden still expect from the virus this year?
[25/26]

I'm writing a post about Europe, expanding on this.
I also have another big article dropping this WE.
Sign up if you don't want to miss them:
mailchi.mp/b44a5f38b4f7/c…
[26/26]