Some of the criticisms here can be answered by the paper and the code. To handle the "green jellybean" problem of multiple testing, they apply the Bonferonni correction. (Actually, the code uses the Simes variant of Bonferonni, so I think the article wording is not precise but it's not a major difference.)
I believe the code has a flaw in that it does not apply a multiple testing correction to the confidence intervals in the diagrams. (See lines 3199-3217). This makes the diagram inaccurate but does not alter the number of symptoms that would meet the bar of corrected-p < 0.05.
The deeper challenge is of course that it is a retrospective study with synthetic controls, so it will not be a gold standard. I think the best criticism is the one they note themselves:
"Conversely, with the evolving awareness of long COVID, it is possible that patients with a history of COVID-19 may have been more likely than those without to access primary care and alert clinicians of their symptoms, which could potentially lead to an inflation of the observed effect sizes. This is potentially supported by the increased aHRs observed for symptoms such as cough, sneezing, fever and allergies among patients who were infected during the second surge of the pandemic, compared to those infected during the first surge"
How likely are you to go to the doctor if you symptom is reduced libido and you have no recent major medical issues? But if you have diverse symptoms AND recently recovered from COVID, you may be much more likely to seek a medical opinion.
My comments above are methodological and not at all meant to deny that people do suffer long after acute COVID. Some of the ratios (like anosmia being 6x more likely in COVID group) are pretty large and that to me would continue to merit further investigation.
Thanks for that. It's a shame no one posted the original nature link previously.
> How likely are you to go to the doctor if you symptom is reduced libido and you have no recent major medical issues?
What's your estimation on how likely this is? If someone had just recovered from the worst pandemic in a century, would they self-assess as having no recent major medical issues? It's an interesting question.
The word "vaccine" appears nowhere in the linked article.
As for your off topic post, if you were concerned with the data I would think your first step would be to verify if the data was real. Your second step would be to understand the context in which the data was collected and reported. Your third step would probably be to attempt to establish a baseline of how many reports we'd expect if the vaccine were a placebo.
Even one of those anti-vaccer web sites has a disclaimer that says "Reports are not proof of causality." Which seems to hint that you misunderstand what the data is reporting.
Interesting, looks like it has enough information to track individual people's symptoms over time. Perfect. And there are a lot of check-in timepoints per registrant (144 mil. time points, 9 mil. registrants)
Really? HN comments are full of tangents that only partly relate to an original post's topic, and it makes them all the more interesting in general. The comment above is talking about COVID vaccines with interesting links and a decent argument, in a post about long covid and its effects. There's plenty of conversation-worthy correlation between the two.
I'm on HN for a decade at least, it is always either clear how a comment relates to the subject or it explicitly declares it's off-topic which is fine as well
Yes, there are some studies indicating that vaccines may cause long covid. This of course raises the interesting question about where vaccine reactions end and long covid begins.
Further, the vaccine is associated with lengthened menstrual cycle times in women. Some have conjectured this is due to the Lipid Nano-particles from the vaccine coalescing into the ovaries and to a lesser extent in the testes (as happened in Pfizer mice trial data released earlier this year and was opposite of their public statements about those particles staying in the muscle tissue around the injection site). Combining the two raises the question about if that is a spurious correlation and the actual cause is the spike protein itself.
> Yes, there are some studies indicating that vaccines may cause long covid
You linked one study to be precise, which is about "rare cases". So going off that into vaccine alarmism and calling this vaccine rollout only better than a rollout that paralysed half a million children (and so I assume worse than a rollout that may have created AIDS?) is a bit disingenuous.
> Further, the vaccine is associated with lengthened menstrual cycle times in women.
We have basically three causal options on vaccine reactions assuming you believe the CDC and doctors who admitted patients to the hospital.
1. The vaccine mRNA code is buggy and has side effects beyond the spike protein it creates.
2. The vaccine lipid nano-particles are causing side effects in addition to the spike protein.
3. We don't fully understand the effects of the spike protein and most or all of the vaccine side effects are actually associated with the virus spike.
If it's case 1 or 2, then the vaccine is simply bad. If it's case 3, then vaccine side effects are actually analogous to long covid.
The third case is the least "alarmist" take of the three. In the third case, the only issue with the article in question is quantity and not counting enough of these side effects in their long covid analysis.
Everything you stated is based on emotional stances and conclusions. Medicine doesn't work that way. This was a pretty good study with a very large sample size.
The cohort included 486 149 people with confirmed SARS-CoV-2 infection who were not admitted to hospital, matched with a control group of 1.9 million people with no recorded evidence of coronavirus infection.
> People who tested positive for the virus reported at least one of 62 symptoms more frequently 12 weeks after initial infection with SARS-CoV-2 than those who had not contracted the virus.
It was also a self reported. Self reporting for something like this is probably going to be useless.
In fact what might be more interesting is using this as an opportunity to study med student syndrome.
EDIT: Looking at the quote again, less troubling than the self reporting part is the 62 symptoms, I mean how many of you know how much hair loss you've had over the past year vs 12 weeks. They did have a considerable control group, but I also know there are plenty of ways to massage data.
But seriously 62 different possible symptoms? This seems like a very wide list of possible symptoms cast over a very wide net of people.
Still, it's an observational study with no reasonable, plausible mechanism of action presented. Worst of all, the study participant were in no way blinded: they knew, from testing, that they contracted a virus that was the constant focus of fear and uncertainty and pushed by media, quite literally 24/7, for months.
> Still, it's an observational study with no reasonable, plausible mechanism of action presented.
Anything that supports the narrative is absolutely fine. Doesn't matter if the methodology is crap, the data is crap... none of it matters. Pointing any flaw out makes you a horrible person.
Now if you publish any kind of research that goes against the narrative suddenly every single flaw, no matter how irrelevant, comes into play. I don't think you can ever publish research that goes against the narrative and not have it somehow "discredited" by "experts".
I trust absolutely none of the research that has been conducted over the last 2.5 years. It's all garbage. Too much emotion and incentive is involved in making things follow the narrative.
And for people that downvote this... ever been on the other side? Ever been a critic of our covid policies? You get yelled at, called absolutely horrible things, and wished horrible death upon you. Your career can fall apart, relationships can dissolve, friends will stop talking to you. It's miserable. Yet, here we are... on the other side. It's not like we chose this... The science, data and morals just happen to be on our side (according to us (which history will almost certainly support)).
> I trust absolutely none of the research that has been conducted over the last 2.5 years. It's all garbage. Too much emotion and incentive is involved in making things follow the narrative.
Glad I am not the only one with this level of skepticism. I say that implicating both sides of the debate. Every study I read seems filled with overt examples of confirmation bias.
UPDATE: I've read up a bit more about Long COVID. I believe a relative minority of people who suffered from COVID experience lingering health issues. I think Long COVID is currently ill-defined. The respiratory and cardiopulmonary symptoms make perfect sense given what is known about COVID. Other symptoms like mental fogginess make sense as secondary symptoms from respiratory and cardiopulmonary complications.
I still don't find this particular study convincing in its claims of hair loss and sexual dysfunction being mechanically related to COVID aside from the general stress of it.
As a layperson, I have some sort of cognitive dissonance calling it "Long COVID" as if it's a separate but related disease, or a continuation of the primary infection. People's bodies suffer long-lasting damage while successfully fighting off all sorts of infections. But I don't recall it being framed this way.
ORIGINAL: Like most people here in the comments, I'm coming at this report with a healthy amount of skepticism.
The relatively paltry numbers and self-reporting remind me of the folks in cancer clusters in Londonderry Township, Pennsylvania, who completely believe they're victims of the Three Mile Island accident even though it's scientifically impossible for that to be true.
I believe the symptoms are real. But about a million things can cause hair loss and sexual dysfunction, including stress. Perhaps the stress of getting sick during a pandemic when people are dying and you know you're infected with the same illness? The fear of the unknown? Seeing people survive with complications and worrying that the worst of your particular case is yet to come? The guilt from wondering why you were pardoned when so many others weren't so lucky?
I'm pretty sure there's historical precedent for psychological symptoms made physical among the survivors of pandemics throughout history.
That's not to say that there are no precedents of lingering disorders, for example encephalitis lethargica in the wake of the 1918 flu pandemic. But my layperson's application of Occam's razor is telling me this is more Morgellons than encephalitis lethargica given the data currently available.
I would be more than happy to be proven wrong about so-called "long COVID" if it meant there could be a more effective treatment than the complicated world of psychiatric healthcare. If there could be a pill or a shot that could relieve these symptoms, I'd be delighted to eat crow.
"A total of 486,149 non-hospitalized individuals had a coded record of SARS-CoV-2 infection [...]. From the pool of patients with no recorded evidence of SARS-CoV-2 infection, 1,944,580 individuals were propensity score-matched to patients infected with SARS-CoV-2."
Put another way, according to Johns Hopkins, there have been a total of 24,079,325 reported cases of infection in the UK. That means a smidge more than 8% of infected people are suffering from prolonged symptoms as they are defined in this study.
That is statistically significant, but not exactly overwhelming by my personal yardstick of whether or not I'd gamble money on it. Not exactly scientific, I know.
I may have misunderstood one or both of your numerical dismissals, then.
As I read it, your first post suggested this was a blip, similar to some subset of the 8000 people in a village that thought they had increased cancer risk from a fission power plant accident.
I wasn't sure what you meant by 'paltry numbers', given the study was across 2.5m patients.
This post is now saying that (extrapolating) the 2 million people in the UK with long COVID are not compelling to you, either because you believe there's no connection to COVID, or because that's not many people -- I'm not sure which?
What percent of your total net worth would you gamble on a 92% chance in your favor btw? Genuinely kinda curious because to me 8% is an incredibly high chance for something that would negatively affect you for the rest of your life.
I wish there was a way to measure how many people would self-report these symptoms without having had COVID first. Unfortunately, those numbers aren't being collected.
> I wish there was a way to measure how many people would self-report these symptoms without having had COVID first. Unfortunately, those numbers aren't being collected.
There is a way, those data are being collected, the authors cover this in their study - you may care to read it:
This feels like another of those cases where some random guy on the internet intimates that with just a few moments' consideration they've uncovered a fatal flaw in a huge study (500k actual, 2m control) conducted and reviewed by experts.
Specifically to your implication - if you're suggesting the chronic stress came from having COVID, then it's a valid side-effect of the disease. OTOH if it's a background effect of the pandemic, then (naively) I'd suggest that with those cohort numbers it's naturally controlled for, as both the 500k and the 2m demographics both lived / are living through the same pandemic.
EDIT: if you're concerned with the tone of my first paragraph, please click on parent's profile to understand my phrasing choice.
Really? You don't think that living in a pandemic and also getting the disease associated with that pandemic wouldn't be an inherently more stressful situation than not getting the disease?
It seems like a potential confounder to me.
> EDIT: if you're concerned with the tone of my first paragraph, please click on parent's profile to understand my phrasing choice.
Fair - but remember that your negative tone is read by everyone, not just GP.
> Really? You don't think that living in a pandemic and also getting the disease associated with that pandemic wouldn't be an inherently more stressful situation than not getting the disease?
Well, two things. First, I think I addressed that in the first point of my second paragraph. Second, I don't believe it matters what I think, as I'm not an expert, or even terribly well informed, on this subject.
It's possible that those 37 qualified & experienced professionals have missed some basic aspects of epidemiology, statistics, etc that are obvious to an untrained amateur who hasn't looked at the source data -- but that supposition shouldn't be the starting position of a casual reader, and it's that exhausting phenomenon I rail against.
HN readership includes some breathtakingly smart people - but the rest of us don't acquire that via forum-osmosis.
It's annoying to come into a discussion where people are trying to figure out object-level stuff and appeal to authority. You're also overstating the case for doing so. The actual results in this paper are easy to understand and don't involve any complicated epidemiology or statistics, so being an "untrained amateur" isn't much of a barrier to thinking about them.
I agree that the thing you were originally replying to was not a very well-thought-out criticism, and your original comment was useful and didn't require a followup sermon about humility.
Some examples of the processes followed by the authors - there's some terms in there I don't understand, but perhaps they aren't complicated to everyone else.
"Structured data on diagnoses, symptoms and referrals are recorded using SNOMED CT coding terminology. Selection of SNOMED CT codes for data extraction was conducted by a team of clinical researchers using an inhouse developed software platform called Code Builder, with systematic searching of existing code lists, reference to the SNOMED CT terminology browser and through clinical knowledge and discussion. Data extraction was performed using the data extraction for epidemiological research (DExTER) tool for automated clinical epidemiological studies.
"To control for confounding, each patient infected with SARS-CoV-2 was propensity score-matched with up to four patients with no recorded evidence of SARS-CoV-2 infection using a logistic regression model including the covariates listed in the covariates section below and a caliper width of 0.2. The SMD between patients infected with SARS-CoV-2 and patients with no recorded evidence of infection was reported for each variable before and after matching, and a variable with SMD > 0.1 after matching was considered to indicate imbalance in baseline characteristics. Kernel density plots were drawn for the two groups before and after matching to check the distribution of propensity scores."
Compare against dismissive comments proposing this is all coincidental, or the data set / study is is useless because the 2-page summary article mentions 'self-reporting' (it feels this phrase has been grossly misunderstood in this thread), and some casual claims that all long COVID is psychosomatic, etc.
EDIT: I don't think it's an 'appeal to authority' when you defer to expertise of the authors who've detailed their data & processes -- that's more a logical fallacy when you claim an opinion is valid because of who the author(s) is/are.
I know what SNOMED is, and I don’t have to click the link to know that this report is based off some garbage NHS data or something. There’s a lot of politics involved here.
The entire 2.5 years has been nothing but appeals to authority. You drill anybody who supports the mitigations and the argument almost always distills down to an appeal to authority. It isn't science or data, it's blind trust in some authority.
On am aside based on my understanding of academia I could believe that many of them are qualified. I find it less likely that all 37 reviewed the paper, data and methodologies with throughness.
> I find it less likely that all 37 reviewed the paper, data and methodologies with throughness.
Don't forget we live in a highly politicized environment where going against the narrative gets you banned, censored, mocked, and dethroned. There is very strong incentives to publish things that support the narrative. Doesn't even matter if the methods used to arrive at the conclusions make any sense. As long as it supports the narrative, it's good stuff. To do otherwise can cost your career.
A woman I know in her early thirties experienced severe, sudden hair loss in the two weeks she had covid (the original strain). She'd never had hair loss before. It did grow back but it scared her half to death. She's had covid twice now, but not long covid.
Tack it onto the list of bizarre symptoms I guess.
One of my more bizarre ones was Reynaud’s-like symptoms in my feet during the winter. Scared me half to death thinking I was going to lose a toe or something and had to run my feet under a hot bath several times to bring feeling back into my toes.
Another one was odd susceptibility to bruising in my feet. I wore some tight shoes one time, the same shoes I had always worn before COVID (some running shoes) and the resultant bruising/swelling made me think I had deep vein thrombosis.
In both cases, after my long COVID symptoms mostly went away around 2-6 months after I got COVID, I’ve never had anything like it happen again before or since.
> Electronic health records of 2.4 million people in the UK from January 2020 to April 2021
I don't know when the UK rolled vaccines out en masse, but I'm guessing that most of these people were not vaccinated when they had COVID.
Also, if they're looking at records only through April, that means they would have had to have had COVID weeks/months before then, for the follow-up symptoms to be recorded by April 2021. That would make it even more likely that the people were not vaccinated before infection.
My understanding is that people who were vaccinated before infection are at somewhat lower risk of long COVID, so it'd be great to see a large study like this of vaccinated people.
Indeed. Unsurprisingly, they did make some notes about this in the study:
"We were unable to estimate the effect of vaccination and infection year on long COVID symptoms in our study due to the very short follow-up period among those vaccinated and infected in the year 2022 (median 8 (IQR 4–14) and 12 (7–16) days, respectively) compared to those unvaccinated and infected in the year 2021 (33 (16–77) and 64 (31–90) days, respectively). Furthermore, the majority (81%) of patients vaccinated before infection in our cohort were infected with SARS-CoV-2 within 2 weeks of vaccination, which would be before acquiring immunity from vaccination, thus restricting the validity of our data to assess the effects of vaccination on long COVID."
> ... it'd be great to see a large study like this of vaccinated people.
I, too, am very interested in further research in this area, because it will strongly inform my actions now and for the next couple of years. As noted above though, it's frustrating, but it sounds like it's way too early.
"Another study pointed out that the results of most studies regarding the association of sexual dysfunction and marijuana usage are inconclusive and contradictory."
From your link... not exactly a good example. So how about this: ED _is_ a known side effect of smoking *tobacco*, but that doesn't seem to discourage many people.
If you’re implying that they must have been mostly vaccinated because most people were vaccinated then you would have to also assume that most of the control group was vaccinated as well.
Don't worry, it's totally a bastion of morals and ethics. Nothing better than a subreddit full of people celebrating people dying. I mean, they died because they had different values than the poster! The horror!!
Same here, lots of vaccinated people complaining and falling ill. Do not believe some of the 'official' versions of un-vaccinated people falling more ill ( in one version - people who had just one shot were consider un-vaccinated because they did not have the full dose of 2 shots). When the 'data' contradicts anecdotes, believe the anecdote. The emperor has no clothes, etc...
Their code is public: https://github.com/AnuSub/Stata-and-R-codes/
Some of the criticisms here can be answered by the paper and the code. To handle the "green jellybean" problem of multiple testing, they apply the Bonferonni correction. (Actually, the code uses the Simes variant of Bonferonni, so I think the article wording is not precise but it's not a major difference.)
I believe the code has a flaw in that it does not apply a multiple testing correction to the confidence intervals in the diagrams. (See lines 3199-3217). This makes the diagram inaccurate but does not alter the number of symptoms that would meet the bar of corrected-p < 0.05.
The deeper challenge is of course that it is a retrospective study with synthetic controls, so it will not be a gold standard. I think the best criticism is the one they note themselves:
"Conversely, with the evolving awareness of long COVID, it is possible that patients with a history of COVID-19 may have been more likely than those without to access primary care and alert clinicians of their symptoms, which could potentially lead to an inflation of the observed effect sizes. This is potentially supported by the increased aHRs observed for symptoms such as cough, sneezing, fever and allergies among patients who were infected during the second surge of the pandemic, compared to those infected during the first surge"
How likely are you to go to the doctor if you symptom is reduced libido and you have no recent major medical issues? But if you have diverse symptoms AND recently recovered from COVID, you may be much more likely to seek a medical opinion.
My comments above are methodological and not at all meant to deny that people do suffer long after acute COVID. Some of the ratios (like anosmia being 6x more likely in COVID group) are pretty large and that to me would continue to merit further investigation.