Prevalence and predictors of long COVID among non-hospitalised adolescents and young adults: a prospective controlled cohort study, 2022, Wyller et al

That typo is not Dolphin's fault, it's in the original. It's hard to work out if it was meant to be 'in conclusion, our study aims were met' or 'in collusion, our study aims were met'.

This was the study aim, to reach this conclusion. We have seen at least one paper co-authored by Erin Cvejic that tortured the data to produce a similar answer - and I mean seriously tortured it, and then when it still didn't admit to what the authors wanted, they went ahead and put words in its mouth.

This alone suggests that the way they implemented the definition of Long Covid was a problem. They seem to have magicked away the whole idea of Long Covid. Nothing to see here folks, just a bunch of young people making a big deal about nothing, move along.

 

First point is the selection. Look at the huge drops along the way, increasing selection bias.

Then, the 404 Covid-19 positive group reduced to 382 at the 6 month mark. The 105 Covid-19 negative group reduced to 85 at the 6 month mark.

Second point is the criteria. Check out what they are saying is the WHO definition of Long Covid. If that is correct, then the definition is meaningless. Remember that the relatively small number of controls were tested for Covid-19 for some reason, perhaps it was because they had another medical cause for their symptoms. Also, 21% of the Covid-19 positive group had co-morbidities, while 35% of the Covid-19 negative group had co-morbidities.

Edit to add - the supplementary material confirms that everyone was tested for Covid-19 for a reason, because of symptoms consistent with Covid-19 or exposure. So, some of the Covid-19 negative people may have had another infection challenge, or some other symptom-causing condition.

 

Okay, so that's a lovely thing, isn't it.

Loneliness is made to be quite a big thing- it gets mentioned in the abstract.
If we go to the actual finding, here it is. The relative risk of having "WHO defined Long Covid" (i.e. the authors tell us that that is at least one symptom persisting after Covid-19) is 1.00 if you aren't lonely and 1.00 to 1.02 if you are lonely. The gap at the right is because loneliness didn't turn out to be relevant to the risk of Post-Infective Fatigue Syndrome, the other definition they looked at.



So, the risk of getting "WHO defined Long Covid" was about 48%. The risk of getting WHO defined Long Covid if you were lonely was 48% to 48.96% - according to the study with the relatively small number of controls. And this is a result worth presenting? Of course if someone is lonely, they might make a little more of a symptom, because they have to cope on their own. I'm surprised there isn't a bigger result. But to claim loneliness is predictive of Long Covid from this is silly.



Here's another result. 'Negative life events prior to last year' didn't turn out to be relevant at all to whether someone had WHO-defined Long Covid. But look, a higher score on the 'negative life events prior to last year' was protective against post-infective fatigue syndrome!! 60 young people were assessed as meeting the criteria for PIFS, out of a total of 464. That's about 13%. But the young people who had experienced negative life events prior to last year had a lower rate of around 11.6%. So, should we be ensuring that all young people have negative life events, so that they are protected from PIFS? But wait? How does this fit with the idea that all people with PIFS have suffered child abuse, or was it sexual abuse, or sexual harassment... And where are the figures in this study for 'Negative life events in the last year' which surely must have been measured - could it be that they had no effect on PIFS or WHO-defined Long Covid at all?

Could it be that a lot of this is just psychosocial researchers grasping at the noise in the data that suits them and ignoring the noise in the data that doesn't suit them?

That is from the discussion, and I don't disagree with it if the WHO definition is as these authors have presented it. But, my, there's a lot of very nasty stuff in that discussion. The things that we have come to expect like, and I'm of course paraphrasing, 'we aren't saying these young people don't have a real problem, it's just that we can fix their problems with behavioural approaches'.

Lots more that could be said.

 

@dave30th - there's happy hunting ground here.

 

I went to see what the prevalence of post-exertional malaise was, and found the Supplementary Table S11. These are the 'Fatigue and post-exertional malaise' symptoms that were assessed. PEM isn't even there. There are actually quite big differences in the percentage of people reporting fatigue depending on Covid-19 status. I'm not surprised to see no difference for unrefreshing sleep - it's such a vague term, it should be got rid of in ME/CFS criteria I think.

They did assess for PEM using the De Paul Symptom Questionnaire at baseline. It was not very predictive - there was a risk score 1.01 for WHO defined Long Covid and 1.04 for PIFS (P values indicated significance). (I would like to point out that, although not very good, it was a better predictor of PIFS than loneliness was of WHO defined Long Covid - and yet it wasn't mentioned in the abstract.)

Risk ratios (95% confidence intervals) for WHO-defined Long Covid and PIFS

That one is interesting.

The 95% confidence intervals are very large and so are the P values, so there's almost certainly nothing to it, but the data suggest that having one or more vaccinations was not protective for PIFS. If the risk of PIFS at 6 months was say 13%, then the risk of having PIFS having had one or more vaccinations was 33%. It's an indication that there is so much noise in this data that there are few worthwhile conclusions to draw from it, but there is no hint there that vaccinations are preventing the development of PIFS.

Edit: The supplementary table says, with respect to the vaccination data,

It's possible that the data is confounded by the collection time being at the 6 month followup. It's not clear if the responses included vaccinations received post-baseline.

 

Oh, joy, there is more to come. And they did fMRIs and a qualitative studying hope.

And this is yet to come too:

and this:

 

What's the criteria used for post-infection fatigue syndrome?

Fukuda, K. et al. The chronic fatigue syndrome: a comprehensive approach to its denition and study. International Chronic Fatigue Syndrome Study Group. Ann. Intern. Med. 121, 953–959 (1994).

So, yeah.

The authors say that they are willing to make the data available to researchers. I'm not sure yet, but PEM data may have been collected at the 6 month mark, so an analysis could be made of people meeting a modern ME/CFS criteria. If anyone is up for it, I'd love to be involved.

 

Par excellence.

 

And half of all young people who didn't get infected with sars cov-2 (notwithstanding @SNTGatchaman's good point) also developed "Long Covid".

Partly these authors are using that to make the point that Long Covid isn't that bad and isn't immunological, which is rubbish. But they are also making the point that the WHO Long Covid definition is overstating the prevalence of Long Covid, which is right because someone with a single symptom of altered taste, or headaches qualifies.

This is the WHO definition:
https://www.who.int/publications/i/...-19_condition-Clinical_case_definition-2021.1

 

They aren't just claiming these type of numbers for Long Covid:

9. Hanevik, K. et al. Irritable bowel syndrome and chronic fatigue 6 years after giardia infection: a controlled prospective cohort study. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am. 59, 1394–1400 (2014).

10. Hickie, I. et al. Post-infective and chronic fatigue syndromes precipitated by viral and non-viral pathogens: prospective cohort study. BMJ 333, 575 (2006).

11.Pedersen, M. et al. Predictors of chronic fatigue in adolescents six months after acute Epstein-Barr virus infection: A prospective cohort study. Brain. Behav. Immun. 75, 94–100 (2019).

I haven't looked at those references, but I find it hard to believe that half of everyone who gets e.g. glandular fever has a 'post-infective fatigue state' 6 months after. There does seem to be quite a bit of evidence for around 10% to have something like ME/CFS 6 months after the top of infections that seem to be ME/CFS triggers though.

 

Indeed. As I've commented before, my first pre-ME/CFS Myers-Briggs rating was a gratifying thing with wisdom and other good stuff. My second go, post-ME/CFS, turned out to be something like a cheerleader. Make of that what you will.

Assessments of anxiety and depression came from HADS surveys, which we know overstate mental health issues for people with physical illness.

Here's the huge number of measures that potentially could contribute to a suggestion of emotional maladjustment. So many opportunities for cherry picking - it would be interesting to go through each and work out what they latched on to and what they didn't.

Re the unreliability of these, for example, the NEO Five-Factor Inventory-30 (NEO-FFI-30) which was used to assess neuroticism has been found to be influenced, specifically the neuroticism and extraversion measures, by brief priming for happy or sad emotions at time of answering the survey. More on that here:
Personality Trait measures
If watching a 10 minute emotion inducing film and a bit of thinking sad thoughts can influence personality as measured by a survey, then what would having a debilitating illness do?

The baseline measures in this Wyller study were taken between 10 and 28 days after the Covid-19 test. I think people with lingering symptoms, especially something like changed taste or smell or cognition issues, would already be feeling worried after a few weeks of that. And, for sure, the selection bias would have contributed to people who were most concerned about their ongoing symptoms and wanting some answers being over-represented in the study.

 

I've just looked at the detail of how the WHO definition of Long Covid and Fukuda for PIFS were operationalised (in the supplementary material). There's devil in the detail, but they were a bit better than I expected.

There was good screening to eliminate other obvious causes for symptoms. The Fukuda operationalism actually required an aspect of PEM to be at least occasionally present (as measured using the De Paul questionnaire) - it was not optional.

I think anyone thinking of formally commenting on this study needs to get their head around how the criteria were operationalised. And that could take some detailed poking around. For example, from memory, sleep problems turned out to be protective for PIFS, but the researchers eliminated people reporting the highest level of sleep problems from the PIFS category on the grounds that they had a primary sleep disorder.

 

Nice work, @Hutan.