The mysterious paper around the Flu and Covid link

I noted in the previous post that I’m not looking at deciding (yet) if the Tegnell/Swedish approach worked. I’m still working around how this is being reported and what is pushing people to argue about this as if it’s of utmost importance.

However, there has been a very odd non peer-reviewed paper by 3 economists (first publication for one of them) at the centre of some renewed controversy. The paper is here as yet unpublished but has been referenced by Tegnell and caused an annoyed response from his Norwegian counterpart. Tegnell when cornered about it said that we should be asking the authors questions. Since it’s being quoted by the architect of Sweden’s response (though yet still unpublished and not peer reviewed) and used by amongst others Ivor Cummins (ketodiet author), it seems worth casting a look at it and seeing what the fuss is all about.

It offers 16 possible explanations for Sweden’s high death rate from Covid-19 – the first is the light lockdown, the rest range from interesting to stretching the imagination a bit – but it’s always worth thinking outside the box at times.

Here’s a few excerpts from it that got me wondering if it was actually a serious paper at all:

We suggest that in a complete causal account, the lighter-lockdown counts for less than, oh, say, 25 percent

p.2

To address the question of what some other causal factors might be, we pondered Nordic
experience, focused on Sweden, reading widely but also talking to friends and fellow
researchers, bona fide creatures of Nordic affairs.

p.2

So a fairly casual approach to writing which would not be acceptable in anywhere I submit papers to (maybe economics papers take a more casual chatty style than medicine?)

Likewise, The Lancet recently published a paper finding no effect of lockdowns

p.3

So that’s this paper. And what they saying here is not actually true – there was a significant effect of early border closure and hard lockdowns (as opposed to partial or curfew) on patient recovery and on the amount of covid cases. Mortality was associated with lower income countries but not directly to the lockdown. So what they are saying is that there was no statistically significant link between lockdown and mortality found in that paper – they noted that this was likely due to their data being limited to the 1st of May 2020 (which seems like a lifetime ago!)

Our decisions can alter the future, but not the past.

p.10

Again, very odd style making me wonder whether they really think if they are seriously trying to get this published. Very Star Wars-lite.

February skiing, vacation, and travel

p.12

This is where things get a bit weird(er) – they talk about the holidays which may have caused Sweden to have more cases due to less downhill slopes so they went to Italy (a hotspot). Fair enough but I guess the question was didn’t the Norwegians/Danes/Finns go there?

Here they have to call Wikipedia and Quora in as references to answer this one. See this screenshot:

So they are using medals to demonstrate how much a country likes a certain type of skiing then how likely they will go to the Italian alps. It’s a bit tenuous to hang it on that but there could be something there but this is just isn’t enough to get the data to demonstrate anything at all (except that France, Italy and Norway are better than Sweden at skiing at the Olympics).

They then note the prevalence was higher in migrant communities which is larger in Sweden than in the other Nordic countries which is fair enough (though some of the links to evidence didn’t work). But, as we know from the Lancet article they quoted earlier, an early lockdown and closure of frontiers did cause a lower prevalence of the disease so it’s hard to say that the high death rate is due to the larger immigrant population.

Age care homes

They report that 70% of the deaths were in care homes in Sweden but that’s in May. By July it was 47% which actually makes a difference to the rest of their argument in that they are trying to argue that the high death rate was due having a very high amount of Swedish deaths from Covid in care homes and that Sweden had a different organization of care homes and a higher percentage of the population.

I think we can accept that the care homes were very poorly managed in most countries but I don’t really get how an early lockdown which we know would have reduced the rate of community covid-19 (as per the Lancet article) would not have improved the outcome of the care home residents. Unless I’m missing something…

“Dry Tinder” (p.18)

This is the area that has been most quoted but also where things getting much more controversial than they already were.

Their theory goes thus – there was an unusually low amount of deaths the previous year in Sweden, therefore there were more vulnerable people around to die. They argue this was not the case in Denmark/Finland/Norway thus they have that very low death rate and none of this is related to Sweden not having a lockdown.

It’s not a completely crazy idea (Tegnell himself was interested in it though obviously he’d probably be happy to find another reason for the deaths than his soft lockdown) but the issue is all the data they are referring to is second hand – they are basically grabbing nice graphs from Twitter users and pasting them in as evidence with no analysis, just eyeballing. Weirdly, the first graph refers to someone who refers back to them – so it’s a bit a self-fulfilling evidence chain (and also a bit chicken or egg) – “I use you to back up my argument, whilst you use this very same argument that I make here to back up the initial argument”.

Maybe the issue with this isn’t that clear to casual readers but it’s a huge problem. The data they refer to is not one they have personally analysed and I found it actually quite hard to extrapolate the same data from the sources of the mostly anonymous Twitterers.

We don’t have any of the analysis that was in the Lancet article they were using to incorrectly demonstrate their point earlier. Figure 7 is (we are told) uses mortality data to demonstrate the difference between Sweden and the Nordic neighbours to show they are very different but as far as I can see, there is little difference between Norway and Sweden (bar the huge Covid peak that Sweden had in deaths). As there is no analysis, we are just left to be led into that opinion.

Look – there may be an effect there. The way to find this out is to do proper analysis like in that Lancet article, not just post a load of Twitter users cool looking graphs.

Conclusion

First up, this is not a serious paper. It screams from everywhere as a bit of a collage of ideas that really haven’t been fully fleshed out let alone seriously tested. To call it a paper is a bit of a stetch as it will never get published in it’s current state. If this gets published in a serious journal in its current state, then we have witnessed the death of peer review.

Secondly, that it gets quoted seriously by Tegnell is in itself concerning and makes me seriously question if he had actually read it himself or was just told about it. Public adulation on Twitter until further notice is no replacement for serious scientific inquiry. If there really is a serious scientific basis for explaining the extra deaths, I’m happy to see it. This just isn’t it…

Finally, this seems to be another milestone in the poor quality of evidence that the lockdown sceptics seem to be using. I’ve been over how not believing in Virii will make you a “reliable source” on Twitter and the poor reporting of the knock-on deaths from lockdown. The change now is that we have moved beyond just fringe Twitterers peddling their version of science and moved to a more concerning situation where international health figures are starting to quote this non-science. Lord help us…