Just a simple question on psychosomatic conditions. Do all, or almost all conditions have some psychosomatic shadow? Short of losing a limb, is there a possibility of psychosomatic responses when diagnosed with any disease?
And is there any way to actually prevent "catching" psychosomatic conditions?
I wrote up a response to last week’s ACX post on “Raise Your Threshold For Accusing People Of Faking Bisexuality”. I walked through Scott Alexander's statistical argument and wrote up a critique of it: "Is Bisexuality a Political Statement?" https://taboo.substack.com/p/is-bisexuality-a-political-statement
My main concern was that the post failed to take into account the fact that college campuses have a shortage of men comparable to the United Kingdom after the First World War. And liberal women on these campuses are now majority LGBT. This means that on a campus with an average number of male students, bisexual women have a nearly equal number of male/female suitors.
Tangential questions about the bisexuality part. The stereotype is (and has been for awhile) that young bisexual women are likely to identify as, or at least act, exclusively straight later in life, whereas young bisexual men are likely to identify as/act exclusively gay.
Is this true?
If true, why?
Also, if true, is it something biological relating to actual levels of attraction, or a result of societal attitudes/different standards for when you call yourself bisexual?
I have some ideas but not strongly held, so want to see what others say...
This is no surprise. COVID and bisexuality are both very closely tied to vasopressin and renin-angiotensin-aldosterone system.
COVID infects cells via their ACE2 receptor.
Sexuality is understood as driven by oxytocin+vasopressin beyond any other hormones.
Homosexual males have significantly more vasopressin ennervation in their suprachiasmatic nuclei and other key hypothalamic-hippocampal-arousal-adjacent areas.
"In 1990, we described the first brain difference related to sexual orientation in the suprachiasmatic nucleus (SCN)—the brain's “clock”—which in HoM [homosexual males] is twice the size that it is in HeM [heterosexual males] (6). We later induced a similar brain difference in rats by pharmacologically disturbing the interaction between testosterone and the developing brain, using the aromatase inhibitor ATD in the neonatal period (7). This experiment yielded bisexual adult rats that had a larger-than-normal number of vasopressin neurons and total cells in their SCNs."
I don't think 'believe in god' is is quite like being bisexual or thinking you have long COVID. We see belief as much more of a volontary action than these other categories, same with being vegetarian. Also, I think with both bisexuality and long COVID there is an element of anxiety involved (even if being bisexual isn't something you think is bad finding out you aren't what you always thought you are can be scary). OTOH things like ADHD diagnosis is often comforting rather than worrying.
But this would suggest a positive correlation here with some aspect of hypochondria and I vaguely think you said that was ruled out for some reason.
So I agree these points are evidence against my hypothesis and have reduced my probability in it. Just still don't have a better idea.
"JDK writes 'Didn't a pretty well done Norwegian study show that there wasn't even link between "long Covid" and actually having contracted Covid among adolescents?'
I’m pretty skeptical of this. As mentioned above, I think it’s very unlikely Long COVID is 100% psychosomatic. But even 100% psychosomatic conditions obey their own supposed rules; people who had had COVID would be more likely to psych themselves into thinking they had Long COVID than people who didn’t. So the total lack of correlation is surprising on any theory."
I think two separate things are being a bit conflated here.
1) Many of the early, terrible methodology long COVID studies did little/nothing more than ask people who said they had had COVID whether they had X symptoms Y amount of time later. The problems with this are many:
- No control group i.e. what is the base rate of these conditions?
- No controls for confounders and/or self-selection into the sample
Methodology like this is the easiest way to generate claims of 20-30% rates of long COVID.
I guess some researchers eventually got tired of this bad methodology, and attempted to merely check if control groups of people who hadn't had COVID were different in their rate of attesting to symptoms. This is what the Norway study appears to be (from a 5 second glance).
There was a particularly famous UK study - I think in 2020 or 2021 that did just this, and found the same thing - no difference in rates of symptoms in no-Covid controls as Covid treated for children. (I think, but may be misremembering, that they used whether you had ever tested PCR positive). They also found much more muted differences for adults.
This doesn't mean long COVID is necessarily 100% psychosomatic. It implies true long COVID is fairly uncommon (we haven't even controlled for confounders at this point, merely added a control group), and many of these symptoms measures produce a tonne of false positives, because the control group has them too.
Aka: it the treated group has the same rate of the outcome as the control group, it doesn't mean the treated group are all/mostly taking their symptoms. It just means they probably aren't caused by treatment.
2) "But even 100% psychosomatic conditions obey their own supposed rules; people who had had COVID would be more likely to psych themselves into thinking they had Long COVID than people who didn’t."
Suppose the following data generating process:
- Long COVID is rare, say 5%, among those who've had COVID
- Hypochondriacs exist, and will self report having everything. Say they are 2% of the population. Suppose they'll all claim they've had Covid (irrespective of whether they've had positive or negative tests), and will also all claim to have Long COVID.
Then, if you do a study of long COVID at a point in time where seroprevalence is still modest (say most of 2020), e.g. 10%, then:
- ~12% of people will claim to have had COVID, of which ~20% which be people who actually never tested positive but swear they are sure they had it "back in February" etc. (Many such cases).
- ~20% of people who say they've had Covid will claim they have long COVID
- ~70% of those who attest to having long COVID have never tested positive
- There would be very little difference in Pr(Claims long COVID) between "never tested positive" (2%) and tested positive (2.45%). With sampling variation, probably dominated by noise.
For the claim you posit to matter ("more likely to psych themselves into thinking"), you want a situation where the hypochondria acts (at least partially) conditional on actually being infected, rather than on everyone. COVID was probably a perfect storm for this to be less true than usual - it's a disease where the symptoms are vague and common and difficult to discern from other common diseases we all get all the time, severity differs massively, and public attention was orders of magnitude higher than usual. All of this would surely encourage hypochondriacs to think they had been infected. I'd still tend to think there's some truth to your point, but all you need is something like "hypochondriacs take more precautions and are less likely to actually have had Covid" to make the correlation actually go negative.
"There’s no rule saying you can’t detect an effect with a sample size of 254. It depends on the size of the effect you’re trying to detect. If you think groups look different, you do a significance test to see whether the difference you found is significant given the sample size. I did a chi squared test and it was 0.016 for the analysis Coyne is talking about."
I think this is a bit misleading. Statistical significance testing *doesn't* mean you don't have to worry about small sample sizes, for two reasons:
1) Small sample sizes usually have low power, so statistically significant results tend to be overestimates. Andrew Gelman calls this the "statistical significance filter".
2) To use a statistical significance test, you need to assume that the statistic you're testing comes from a certain distribution (i.e. normal, chi squared, etc. ) You can either do this by making unrealistic assumptions about the world ("alll my variables just happen to be normally distributed"), or you can rely on theorems that say things like "as N gets big this approaches a normal distribution", which is what people implicitly do in practice but relies on having a large enough sample size for the tests to give the right answer!
I haven't thought about it enough in this case (and probably won't) to know if having numbers like 1, 17, and 18 in some of the bins is too small for these tests to be reasonable, so I won't take a side on that. But I do think the "I did a hypothesis test so I don't have to worry about small sample size" assumption is common practice but not good statistical practice.
I'm a rheumatologist who sees a lot of fibro, EDS, and long Covid. Rheumatologists actually sometimes joke that if a bi or genderqueer woman with brightly dyed hair walks into your clinic with joint pain, there is a 99% chance she/they has hypermobile EDS and fibromyalgia. It's not to denigrate these people or not to take them seriously (their suffering is real regardless of being "psychosomatic" or not, and besides we gotta catch that 1% who has RA or lupus and would actually benefit from immune suppression), but there's something about that personality type/set of life experiences that just always produces this same cluster of symptoms.
As someone who's actually got boots on the grounds with long Covid and other conditions in that cluster, you got it exactly right with this statement: "But I think the strong version of this is that straights have some fatigue, ignore it, and it goes away, whereas bisexuals have some fatigue and focus on it in a way that makes it worse and turns it into a trapped prior. This is how I think of chronic pain and several other psychosomatic illnesses." That's why these patients NEVER get better with immune suppression, or blood thinners, or hormone adjustments, or whatever other biological interventions, but they sometimes/often do get better with SSRI/SNRI and pain psychology therapy.
Scott, I think there's some sort of copy-and-paste error: Your quote of Toggle's comment seems to contain a copy of Mike's later comment. (I didn't check to figure out the correct original comments.)
> When I’ve looked into depression biomarkers, it’s been very hard to distinguish them from general bad health markers, and Long COVID would be especially hard since you would have to distinguish them from previously-had-severe-COVID markers.
[posting this to provide more information re biomarker studies and to point at a promising biomarker finding which seems remarkably feasible to independently replicate: about $1200 for Angiopoietin-1 + P-selectin ELISAs, and access to a laboratory with a plate reader. Not arguing anything re psychosomaticity here. Thanks for featuring my comment and thanks for clarifying what you meant re psychosomaticity]
Several of the long COVID biomarker studies do compare against "people who recovered fine from mild COVID" (most people these days!) and/or "people who were in the ICU for COVID" groups and nevertheless effectively classify. Re specificity, indeed, some of the biomarker findings are not very diagnostically useful on their own. This study https://www.medrxiv.org/content/10.1101/2022.08.09.22278592v1 finds cortisol levels (halved in long covid patients (without significant changes in ACTH) compared to controls, even a year after acute infection) the most predictive single factor, but cortisol is...involved in a lot of things, so it's not very specific.
Re the Angiopoietin-1 + P-selectin (proteins involved in vascular/endothelial/platelet function) finding; there is essentially *no* overlap between long covid patients and any of the controls (healthy controls, people with mild covid, people with severe covid). The variance for the control groups is an order of magnitude smaller than the difference between the long covid group and any of the controls. Purely from eyeballing the figures, the average levels in long covid seem to be an order of magnitude higher than those in any of the other groups: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9549814/figure/Fig3/
This points to "somehow actively involved in the disease process" more than "biomarker for bad health".
Some chronic conditions do elevate Ang-1 and P-sel, but much less than was observed in this study. https://pubmed.ncbi.nlm.nih.gov/30047017/ finds *at most* a 30% elevation in P-selectin in people with T2DM with high vascular risk factors. Not an order of magnitude increase.
Between the remarkable reported specificity/sensitivity, samples collected by simple blood test, and affordable immunoassays available for the relevant biomolecules, this seems like the most promising biomarker finding for LC I've seen so far. If I had access to a laboratory, I would try to replicate these findings (assuming I could find some people with long COVID who'd be up to participate; I don't have it, I'm just interested in it).
"a study showing that bisexuals had more cancer, asthma, and heart disease than straight (or gay) people"
Yes, but the interesting claim about long covid is that gays and straights are similar, while bisexuals are outliers. This study makes bisexuals look like gays. That's not a surprising result. Lots of the arguments in this post assume that bisexuals are like gays. Maybe those arguments explain these results, but they don't explain the long covid results, where gays are like straights.
In 5/6 comparisons (3 diseases x 2 sexes) bisexuals have more disease prevalence than gays and straights, but in only two cases does it match the long covid pattern of gays and straights being similar. In three cases gays are in the middle, twice halfway, once close to bisexuals. In the final case, bisexuals are like straights and gays are the outlier. The cases that match long covid in bisexuals being the outlier are heart disease in both sexes. The cases where gays are halfway in between are asthma in men and cancer in women. The case where bisexuals are in the middle but close to lesbians is asthma in women. The case where bisexuals are like straights is cancer in men.
"I find libertarians and Marxists, who I would expect to be less interested in the right-wing project of minimizing COVID than conservatives, sort of interesting."
Libertarians in America are for the most part aligned with the right, many with the extreme right. Even those who are not, in my anecdotal experience, were very anti-COVID restrictions (for the obvious reasons one would expect libertarians to be against extreme restrictions on individual liberty), and also COVID skeptics, as a way to justify the resistance to even mild COVID restrictions. For these reasons I'd by and large expect libertarians to be similar to conservatives on questions relating to COVID.
Marxists, on the other hand, are just weirdos. There is no one political camp in America where self-identified Marxists can find a home, and in my (also anecdotal) experience a lot of people who identify as Marxist don't really mean much specific by it. I've known folks with a wide range of political orientations (including a small number who would probably be called right wing, though this has been rare) who call themselves Marxists. So yeah, I'm not sure what this means.
To date, there is no compelling discriminant validity evidence for a distinct "Long COVID" syndrome. It is generalized post-viral symptomatology until empirical evidence clearly indicates otherwise. Nonetheless, the common cause herein is negative emotionality (NE) /affective dysregulation.
Is there data comparing the prevalence of individual long-covid symptoms? Some seem more 'psychosomatic' than others. 'Fatigue' is pretty vague and subjective, but a friend of my dad's has lost all sense of taste and smell since getting COVID in 2020. He's lost a significant amount of weight since he's no longer been able to taste food, it seems very likely that losing your sense of taste/smell is much less likely to be psychosomatic than fatigue or 'brain fog'.
> If this were true, the bisexuality effect would be stronger for milder cases of Long COVID
I don't think this is right. I'd expect a personality trait that pushes people to round off "maybe long COVID" to "long COVID" would also make them round "maybe severe case" to "severe case", so you'd get the same bias in severe/mild case as you do in long COVID.
Is «could and wanted to expend all the effort to get formally diagnosed» a significant factor in «diagnosed by MD»? Because all the healthcare access discussions seemed to imply there is some barrier.
Reporting thresholds and slack to compensate for mild damage are surely always relevant: I (male, straight) had some lingering effects after a definite viral cold likely to have been CoViD that might be too weak to even call symptoms and they are perceptible in very specific situations that can arise or not…
"Also, if sexual contact caused immune problems down the line, this would be a big deal and we would already know."
Do we not already have pretty good hints in this direction? This seems obvious enough that I wonder if I've misunderstood you somehow.
I mean, ignoring the low-hanging fruit of HIV/AIDs, most of the pathogens which maintain persistent infections do so by tamping down the immune system in one way or another. The herpatic viruses are notorious for this. They often interfere with iNOS or deplete arginine which has a similarly depressive effect on macrophages.
• Reactivations of EBV, CMV and HHV-6 are frequent in severe COVID-19.
• EBV reactivation is associated with longer ICU length-of-stay.
• EBV reactivation occurs early after ICU admission.
CMV reactivation occurs later after ICU admission and may require anti-CMV treatment.
" cytomegalovirus (CMV, HHV-5, a β-herpesvirus) imposes a surprisingly profound impact. The majority of the world’s population is CMV+, and the virus goes through 3 distinct infection phases en route to establishing lifelong détente with its host. Immune control of CMV in each phase recruits unique arms of host defense, and in turn the virus employs multiple immune modulatory strategies that help facilitate the establishment of lifelong persistence....This strategy of hijacking host IL-10 immunosuppression to promote viral persistence is similarly employed by MCMV "
To be clear, I'm not saying that this is relevant to the immediate discussion. I just wanted to address the point that STDs can impact immune responses.
Man, the entirety of section 5 should be its own essay emphasizing the perspective that basically anything can be psychosomatic. The heart attack and chest pain statistics are a significant update for me, and while I intuitively understood that some illnesses are heavily mixed - I knew this is very true for well-known-frequently-psychosomatic stuff like LC and back pain; incidentally, just the other day I had to wonder if a headache I was having was a recent stress or dehydration - I still didn't realize it was such a common pattern, and based on the previous discussions I'm guessing most people didn't.
I think part of the problem is that it is down to self-identification, so you will have real bisexuals and the social contagion bisexuals and the real Long Covid or post-viral syndrome sufferers and the 'maybe I have it! everyone else on Tiktok does!' lot.
The audience here will tend to skew towards "yes indeed real bisexual" along with a lot of other minority positions on things, but in the general population?
How do you define bisexual?
And again, I do incline to the view that people who are more on the left/liberal/progressive side of the fence will more easily identify as A, B, or C than people more on the right/conservative side, as well as older versus younger. If you're in a place and/or time where it's socially acceptable to be some variety of queer, then you're more likely to identify as that than if it's a place where you're in the closet unless you really, strongly, definitely are gay, lesbian or bi.
So how do you define bisexual? It's not "have you had sex with people of both sexes?" because according to this randomly selected site:
"People use a few common labels to identify their sexuality. Your sexuality isn’t defined by who you have sex with – it’s about how you feel and how you choose to identify yourself. The important thing is that you choose what label feels comfortable, or you choose no label at all. You might find, like many others have, that the label you choose changes over time."
So if you're young and you go "I think Barbie is cute", then you may go "Oh wow, am I bi/lesbian?" and the answer seems to be "If you think you are, yes! So long as you find that description comfortable, then you are, and when it's not, then you're not".
So, TMI time again. In my attempts to figure out what the heck was wrong with me over the years of my youth and adulthood, one of the considerations was "well, do you prefer girls then?" And considering it as a hypothetical, the idea of having sex with a woman didn't make me go "Ugh, no!!!" (Sex with a real person in reality being not at all the same notion because no thank you to anyone of any sex or gender).
I don't consider myself bisexual. *But* were I a young adult *now* and a good bit more left-leaning in social attitudes, I well might identify as such. "But you've never had sex with/dated a same-gender person!" Yeah, shut up bigot, not necessary for self-identification.
Same with long covid. I've had covid, and since then I do feel that I have a perceptible change in taste. I have bouts of fatigue, breathlessness, 'brain fog', many of the very vague and multiply-applicable symptoms.
But I can also put these down to getting older and existing health conditions.
So, do I have long covid? I say no. But again, a slight difference in inclination, and I could say "yes".
When answering the survey, I said no to both the bisexuaity and long covid parts. But rotate me 45 degrees, and I'd be answering yes. For the same physical entity with the same physiological symptoms.
Now you tell me: am I bi or not? long covid or not?
Is there research on if life or death social situations cause psychosomatic symptoms?
Like consider rooms clapping for Stalin or the purge by Saddam Hussein; I dont know how you'd ethically test such a situation if say >50% of the room had migraines, but would it be the most unusual thing in the world?
If you social group is burning witches maybe its for the best for your nervous system if you join in to the hysterical dancing, illnesses and people do just have a shut down switch when they believe there is lethal danger and their tribe is angry with them.
"It's possible that both trans-ness and bisexuality are the result of a sort of anatomical and neuro-hormonal "chimerism" that isn't present in either fully straight or fully gay cis people. I don't mean they're the result of literal genetic chimerism (which is far rarer than bisexuality or gender dysphoria), but rather, that they're caused by some sort of mismatch involving the body's neurological and hormonal transmitters and receptors."
This doesn't seem right. Wouldn't this mismatch she speaks of be more prevalent in homosexuals than in bisexuals? We've all encountered effeminate gay men and butch lesbians, have we not? They would figure to have the highest degree of this mismatch, I would imagine, but that's not where the incidence of long Covid seems to be the highest, right?
"Christians and Republicans had no more Long COVID than people who said no religion or no political party, but polyamorous people and rationalists did."
During the pandemic, I noticed that the SF tech scene (i.e., polyamorous rationalists) went deeper into social distancing than any other demographic on the planet. People I knew who were actually immunodeficient did not isolate nearly as much as these guys. Social distance became a moral imperative and covid became The Worst Thing That Could Ever Happen To You.
"I think the worst-case scenario is that, since Long COVID is in the news, extremely sympathetic, and has maximally vague symptoms, its psychosomatic shadow could be much bigger than normal"
Especially if you were so traumatized by a positive covid test result, or maybe even just something that you suspect was covid but couldn't get a test for (remember that other infectious diseases are also still a thing), that your recovery was followed by a laser focus for anything that might be long covid. Because of course you must be injured! If you isolated like a champ only to discover that covid was a nothing burger for you, how foolish would that feel?
Coyne’s arguments convinced me that your survey effect probably isn’t real. There is huge difference between replicating an established result and establishing a new result. One of those is significantly harder than the other.
And if you (formally or informally) considered more than 3 hypotheses when looking at the data then a p-value of 0.016 is not significant after adjusting for multiple comparisons.
Quick note: Coyne expects confidence intervals from Scott, but I don't think it's possible to construct those given the type of analysis he used. I say this because I attempted to replicate Scott's numbers and provide these confidence intervals.
I'm assuming Scott used a chi squared test for independence. This is well and good for seeing whether long covid in bisexuals or not follow the same distribution. In tests like these we get a p-value and then see how small or large it is. Then we can reject the null or not.
This is not the time to construct confidence intervals. My understanding is we do that when we're estimating the parameters of a distribution. (i.e. what's the average height of Uzbeks?). Scott is not estimating any quantities here, so whatever confidence intervals Coyne has in mind don't apply here.
I also don't think we can construct a confidence interval around the p-value. Mathematically, it's just a single observation from a chi squared distribution.
As a digression, you can also see Scott as already having computed a confidence interval of some sort when he conducted the chi squared test. All a statistical test is is using the null hypothesis to say your test statistic will follow a certain sort of distribution (in this case, a chi squared one). You then say if the null is true, there's a .95 chance my statistic will lie in this interval. That interval looks a lot like a confidence interval.
Note I don't think all of Long Covid is psychosomatic, I discuss biological contributors in the 1st post. In my own case I suspect either Epstein-Barr reactivation or persistent virus for several months.
I explained this debate to my bisexual friend and she found the causality obvious: “bisexuals reproducibly sit in weird contorted positions which constantly make us hyperaware of our lung capacity.”
My thoughts on this are not fully formed, but I think you're missing out on the fact that sexual attraction is IN ITSELF subject to suggestion. When someone expresses attraction to you, that emotional state can be contagious-- people are turned on just by being desired. So bisexuals can be genuinely attracted to both sexes, while at the same time being more prone to social contagion that would lead them to delusionally believe they have a disease.
"Hundreds of biomedical findings have been documented, with many patients experiencing dozens of symptoms across multiple organ systems7 (Fig. 1). Long COVID encompasses multiple adverse outcomes, with common new-onset conditions including cardiovascular, thrombotic and cerebrovascular disease8, type 2 diabetes9, myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS)10,11 and dysautonomia, especially postural orthostatic tachycardia syndrome (POTS)12 (Fig. 2). Symptoms can last for years13, and particularly in cases of new-onset ME/CFS and dysautonomia are expected to be lifelong14. With significant proportions of individuals with long COVID unable to return to work7, the scale of newly disabled individuals is contributing to labour shortages15. There are currently no validated effective treatments."
Self-identified by bisexuals are more likely to be left of center, and left of center after more likely to report long covid as long covid? 🤔
Just a simple question on psychosomatic conditions. Do all, or almost all conditions have some psychosomatic shadow? Short of losing a limb, is there a possibility of psychosomatic responses when diagnosed with any disease?
And is there any way to actually prevent "catching" psychosomatic conditions?
If you're going to include the Kinsey Scale on the next ACX survey, please make sure to include the "X" (asexual/nonsexual) option!
I wrote up a response to last week’s ACX post on “Raise Your Threshold For Accusing People Of Faking Bisexuality”. I walked through Scott Alexander's statistical argument and wrote up a critique of it: "Is Bisexuality a Political Statement?" https://taboo.substack.com/p/is-bisexuality-a-political-statement
My main concern was that the post failed to take into account the fact that college campuses have a shortage of men comparable to the United Kingdom after the First World War. And liberal women on these campuses are now majority LGBT. This means that on a campus with an average number of male students, bisexual women have a nearly equal number of male/female suitors.
Tangential questions about the bisexuality part. The stereotype is (and has been for awhile) that young bisexual women are likely to identify as, or at least act, exclusively straight later in life, whereas young bisexual men are likely to identify as/act exclusively gay.
Is this true?
If true, why?
Also, if true, is it something biological relating to actual levels of attraction, or a result of societal attitudes/different standards for when you call yourself bisexual?
I have some ideas but not strongly held, so want to see what others say...
This is no surprise. COVID and bisexuality are both very closely tied to vasopressin and renin-angiotensin-aldosterone system.
COVID infects cells via their ACE2 receptor.
Sexuality is understood as driven by oxytocin+vasopressin beyond any other hormones.
Homosexual males have significantly more vasopressin ennervation in their suprachiasmatic nuclei and other key hypothalamic-hippocampal-arousal-adjacent areas.
For example--
https://www.pnas.org/doi/10.1073/pnas.0805542105
"In 1990, we described the first brain difference related to sexual orientation in the suprachiasmatic nucleus (SCN)—the brain's “clock”—which in HoM [homosexual males] is twice the size that it is in HeM [heterosexual males] (6). We later induced a similar brain difference in rats by pharmacologically disturbing the interaction between testosterone and the developing brain, using the aromatase inhibitor ATD in the neonatal period (7). This experiment yielded bisexual adult rats that had a larger-than-normal number of vasopressin neurons and total cells in their SCNs."
I don't think 'believe in god' is is quite like being bisexual or thinking you have long COVID. We see belief as much more of a volontary action than these other categories, same with being vegetarian. Also, I think with both bisexuality and long COVID there is an element of anxiety involved (even if being bisexual isn't something you think is bad finding out you aren't what you always thought you are can be scary). OTOH things like ADHD diagnosis is often comforting rather than worrying.
But this would suggest a positive correlation here with some aspect of hypochondria and I vaguely think you said that was ruled out for some reason.
So I agree these points are evidence against my hypothesis and have reduced my probability in it. Just still don't have a better idea.
"JDK writes 'Didn't a pretty well done Norwegian study show that there wasn't even link between "long Covid" and actually having contracted Covid among adolescents?'
I’m pretty skeptical of this. As mentioned above, I think it’s very unlikely Long COVID is 100% psychosomatic. But even 100% psychosomatic conditions obey their own supposed rules; people who had had COVID would be more likely to psych themselves into thinking they had Long COVID than people who didn’t. So the total lack of correlation is surprising on any theory."
I think two separate things are being a bit conflated here.
1) Many of the early, terrible methodology long COVID studies did little/nothing more than ask people who said they had had COVID whether they had X symptoms Y amount of time later. The problems with this are many:
- No control group i.e. what is the base rate of these conditions?
- No controls for confounders and/or self-selection into the sample
Methodology like this is the easiest way to generate claims of 20-30% rates of long COVID.
I guess some researchers eventually got tired of this bad methodology, and attempted to merely check if control groups of people who hadn't had COVID were different in their rate of attesting to symptoms. This is what the Norway study appears to be (from a 5 second glance).
There was a particularly famous UK study - I think in 2020 or 2021 that did just this, and found the same thing - no difference in rates of symptoms in no-Covid controls as Covid treated for children. (I think, but may be misremembering, that they used whether you had ever tested PCR positive). They also found much more muted differences for adults.
This doesn't mean long COVID is necessarily 100% psychosomatic. It implies true long COVID is fairly uncommon (we haven't even controlled for confounders at this point, merely added a control group), and many of these symptoms measures produce a tonne of false positives, because the control group has them too.
Aka: it the treated group has the same rate of the outcome as the control group, it doesn't mean the treated group are all/mostly taking their symptoms. It just means they probably aren't caused by treatment.
2) "But even 100% psychosomatic conditions obey their own supposed rules; people who had had COVID would be more likely to psych themselves into thinking they had Long COVID than people who didn’t."
Suppose the following data generating process:
- Long COVID is rare, say 5%, among those who've had COVID
- Hypochondriacs exist, and will self report having everything. Say they are 2% of the population. Suppose they'll all claim they've had Covid (irrespective of whether they've had positive or negative tests), and will also all claim to have Long COVID.
Then, if you do a study of long COVID at a point in time where seroprevalence is still modest (say most of 2020), e.g. 10%, then:
- ~12% of people will claim to have had COVID, of which ~20% which be people who actually never tested positive but swear they are sure they had it "back in February" etc. (Many such cases).
- ~20% of people who say they've had Covid will claim they have long COVID
- ~70% of those who attest to having long COVID have never tested positive
- There would be very little difference in Pr(Claims long COVID) between "never tested positive" (2%) and tested positive (2.45%). With sampling variation, probably dominated by noise.
For the claim you posit to matter ("more likely to psych themselves into thinking"), you want a situation where the hypochondria acts (at least partially) conditional on actually being infected, rather than on everyone. COVID was probably a perfect storm for this to be less true than usual - it's a disease where the symptoms are vague and common and difficult to discern from other common diseases we all get all the time, severity differs massively, and public attention was orders of magnitude higher than usual. All of this would surely encourage hypochondriacs to think they had been infected. I'd still tend to think there's some truth to your point, but all you need is something like "hypochondriacs take more precautions and are less likely to actually have had Covid" to make the correlation actually go negative.
"There’s no rule saying you can’t detect an effect with a sample size of 254. It depends on the size of the effect you’re trying to detect. If you think groups look different, you do a significance test to see whether the difference you found is significant given the sample size. I did a chi squared test and it was 0.016 for the analysis Coyne is talking about."
I think this is a bit misleading. Statistical significance testing *doesn't* mean you don't have to worry about small sample sizes, for two reasons:
1) Small sample sizes usually have low power, so statistically significant results tend to be overestimates. Andrew Gelman calls this the "statistical significance filter".
2) To use a statistical significance test, you need to assume that the statistic you're testing comes from a certain distribution (i.e. normal, chi squared, etc. ) You can either do this by making unrealistic assumptions about the world ("alll my variables just happen to be normally distributed"), or you can rely on theorems that say things like "as N gets big this approaches a normal distribution", which is what people implicitly do in practice but relies on having a large enough sample size for the tests to give the right answer!
I haven't thought about it enough in this case (and probably won't) to know if having numbers like 1, 17, and 18 in some of the bins is too small for these tests to be reasonable, so I won't take a side on that. But I do think the "I did a hypothesis test so I don't have to worry about small sample size" assumption is common practice but not good statistical practice.
I'm a rheumatologist who sees a lot of fibro, EDS, and long Covid. Rheumatologists actually sometimes joke that if a bi or genderqueer woman with brightly dyed hair walks into your clinic with joint pain, there is a 99% chance she/they has hypermobile EDS and fibromyalgia. It's not to denigrate these people or not to take them seriously (their suffering is real regardless of being "psychosomatic" or not, and besides we gotta catch that 1% who has RA or lupus and would actually benefit from immune suppression), but there's something about that personality type/set of life experiences that just always produces this same cluster of symptoms.
As someone who's actually got boots on the grounds with long Covid and other conditions in that cluster, you got it exactly right with this statement: "But I think the strong version of this is that straights have some fatigue, ignore it, and it goes away, whereas bisexuals have some fatigue and focus on it in a way that makes it worse and turns it into a trapped prior. This is how I think of chronic pain and several other psychosomatic illnesses." That's why these patients NEVER get better with immune suppression, or blood thinners, or hormone adjustments, or whatever other biological interventions, but they sometimes/often do get better with SSRI/SNRI and pain psychology therapy.
Scott, I think there's some sort of copy-and-paste error: Your quote of Toggle's comment seems to contain a copy of Mike's later comment. (I didn't check to figure out the correct original comments.)
> When I’ve looked into depression biomarkers, it’s been very hard to distinguish them from general bad health markers, and Long COVID would be especially hard since you would have to distinguish them from previously-had-severe-COVID markers.
[posting this to provide more information re biomarker studies and to point at a promising biomarker finding which seems remarkably feasible to independently replicate: about $1200 for Angiopoietin-1 + P-selectin ELISAs, and access to a laboratory with a plate reader. Not arguing anything re psychosomaticity here. Thanks for featuring my comment and thanks for clarifying what you meant re psychosomaticity]
Several of the long COVID biomarker studies do compare against "people who recovered fine from mild COVID" (most people these days!) and/or "people who were in the ICU for COVID" groups and nevertheless effectively classify. Re specificity, indeed, some of the biomarker findings are not very diagnostically useful on their own. This study https://www.medrxiv.org/content/10.1101/2022.08.09.22278592v1 finds cortisol levels (halved in long covid patients (without significant changes in ACTH) compared to controls, even a year after acute infection) the most predictive single factor, but cortisol is...involved in a lot of things, so it's not very specific.
However other long COVID biomarker findings are unambiguously COVID-related: persistent circulating SARS-CoV-2 Spike in blood months after acute illness, persistent SARS-CoV-2 nucleocapsid + RNA in gut months after acute illness: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9057012/ https://academic.oup.com/cid/article/76/3/e487/6686531
Re the Angiopoietin-1 + P-selectin (proteins involved in vascular/endothelial/platelet function) finding; there is essentially *no* overlap between long covid patients and any of the controls (healthy controls, people with mild covid, people with severe covid). The variance for the control groups is an order of magnitude smaller than the difference between the long covid group and any of the controls. Purely from eyeballing the figures, the average levels in long covid seem to be an order of magnitude higher than those in any of the other groups: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9549814/figure/Fig3/
This points to "somehow actively involved in the disease process" more than "biomarker for bad health".
Some chronic conditions do elevate Ang-1 and P-sel, but much less than was observed in this study. https://pubmed.ncbi.nlm.nih.gov/30047017/ finds *at most* a 30% elevation in P-selectin in people with T2DM with high vascular risk factors. Not an order of magnitude increase.
To put it another way, the *chronic* P-selectin elevations found in long COVID patients seem (if I've done my math right) to be commensurate with *acute* P-selectin elevations in conditions like acute myocardial infarction (https://pubmed.ncbi.nlm.nih.gov/9672405/) or acute ischemic stroke (https://www.ahajournals.org/doi/10.1161/01.STR.28.11.2214).
Between the remarkable reported specificity/sensitivity, samples collected by simple blood test, and affordable immunoassays available for the relevant biomolecules, this seems like the most promising biomarker finding for LC I've seen so far. If I had access to a laboratory, I would try to replicate these findings (assuming I could find some people with long COVID who'd be up to participate; I don't have it, I'm just interested in it).
The authors use a "Endothelial Injury Marker 12-Plex Human ProcartaPlex™ Panel, EPX120-15849-901" (which costs $2600 and requires a Luminex xMAP fluorescent magnetic bead system) which let them assay 12 endothelial-related proteins per sample at once. There are much less expensive commercially-available immunoassay kits for Angiopoietin-1 and P-selectin, which only require a plate reader: $1200 gets you 96 ELISA tests for Angiopoietin-1 (https://www.thermofisher.com/elisa/product/Angiopoietin-1-Human-ELISA-Kit/EHANGPT1) along with 96 ELISA tests for P-selectin (https://www.thermofisher.com/elisa/product/P-Selectin-Soluble-CD62-Human-ELISA-Kit/BMS219-4).
This study (https://molmed.biomedcentral.com/articles/10.1186/s10020-023-00610-z) by the same group finds even *more* proteins in the blood with excellent specificity/sensitivity for long COVID, and again, the levels in long covid patients are completely disjoint with the "5th percentile to 95th percentile protein expression range of healthy control subjects": https://molmed.biomedcentral.com/articles/10.1186/s10020-023-00610-z/figures/2
"a study showing that bisexuals had more cancer, asthma, and heart disease than straight (or gay) people"
Yes, but the interesting claim about long covid is that gays and straights are similar, while bisexuals are outliers. This study makes bisexuals look like gays. That's not a surprising result. Lots of the arguments in this post assume that bisexuals are like gays. Maybe those arguments explain these results, but they don't explain the long covid results, where gays are like straights.
In 5/6 comparisons (3 diseases x 2 sexes) bisexuals have more disease prevalence than gays and straights, but in only two cases does it match the long covid pattern of gays and straights being similar. In three cases gays are in the middle, twice halfway, once close to bisexuals. In the final case, bisexuals are like straights and gays are the outlier. The cases that match long covid in bisexuals being the outlier are heart disease in both sexes. The cases where gays are halfway in between are asthma in men and cancer in women. The case where bisexuals are in the middle but close to lesbians is asthma in women. The case where bisexuals are like straights is cancer in men.
"I find libertarians and Marxists, who I would expect to be less interested in the right-wing project of minimizing COVID than conservatives, sort of interesting."
Libertarians in America are for the most part aligned with the right, many with the extreme right. Even those who are not, in my anecdotal experience, were very anti-COVID restrictions (for the obvious reasons one would expect libertarians to be against extreme restrictions on individual liberty), and also COVID skeptics, as a way to justify the resistance to even mild COVID restrictions. For these reasons I'd by and large expect libertarians to be similar to conservatives on questions relating to COVID.
Marxists, on the other hand, are just weirdos. There is no one political camp in America where self-identified Marxists can find a home, and in my (also anecdotal) experience a lot of people who identify as Marxist don't really mean much specific by it. I've known folks with a wide range of political orientations (including a small number who would probably be called right wing, though this has been rare) who call themselves Marxists. So yeah, I'm not sure what this means.
To date, there is no compelling discriminant validity evidence for a distinct "Long COVID" syndrome. It is generalized post-viral symptomatology until empirical evidence clearly indicates otherwise. Nonetheless, the common cause herein is negative emotionality (NE) /affective dysregulation.
Is there data comparing the prevalence of individual long-covid symptoms? Some seem more 'psychosomatic' than others. 'Fatigue' is pretty vague and subjective, but a friend of my dad's has lost all sense of taste and smell since getting COVID in 2020. He's lost a significant amount of weight since he's no longer been able to taste food, it seems very likely that losing your sense of taste/smell is much less likely to be psychosomatic than fatigue or 'brain fog'.
> If this were true, the bisexuality effect would be stronger for milder cases of Long COVID
I don't think this is right. I'd expect a personality trait that pushes people to round off "maybe long COVID" to "long COVID" would also make them round "maybe severe case" to "severe case", so you'd get the same bias in severe/mild case as you do in long COVID.
Is «could and wanted to expend all the effort to get formally diagnosed» a significant factor in «diagnosed by MD»? Because all the healthcare access discussions seemed to imply there is some barrier.
Reporting thresholds and slack to compensate for mild damage are surely always relevant: I (male, straight) had some lingering effects after a definite viral cold likely to have been CoViD that might be too weak to even call symptoms and they are perceptible in very specific situations that can arise or not…
"Also, if sexual contact caused immune problems down the line, this would be a big deal and we would already know."
Do we not already have pretty good hints in this direction? This seems obvious enough that I wonder if I've misunderstood you somehow.
I mean, ignoring the low-hanging fruit of HIV/AIDs, most of the pathogens which maintain persistent infections do so by tamping down the immune system in one way or another. The herpatic viruses are notorious for this. They often interfere with iNOS or deplete arginine which has a similarly depressive effect on macrophages.
• Reactivations of EBV, CMV and HHV-6 are frequent in severe COVID-19.
• EBV reactivation is associated with longer ICU length-of-stay.
• EBV reactivation occurs early after ICU admission.
CMV reactivation occurs later after ICU admission and may require anti-CMV treatment.
https://www.sciencedirect.com/science/article/pii/S2666991921000051
discussion of CMV and autoimmune disorders
https://www.sciencedirect.com/science/article/abs/pii/S0882401018304789
Subclinical CMV viremia is associated with increased nosocomial infections and prolonged hospitalization in patients with systemic autoimmune diseases
https://www.sciencedirect.com/science/article/abs/pii/S1386653221001165
" cytomegalovirus (CMV, HHV-5, a β-herpesvirus) imposes a surprisingly profound impact. The majority of the world’s population is CMV+, and the virus goes through 3 distinct infection phases en route to establishing lifelong détente with its host. Immune control of CMV in each phase recruits unique arms of host defense, and in turn the virus employs multiple immune modulatory strategies that help facilitate the establishment of lifelong persistence....This strategy of hijacking host IL-10 immunosuppression to promote viral persistence is similarly employed by MCMV "
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5992493/
To be clear, I'm not saying that this is relevant to the immediate discussion. I just wanted to address the point that STDs can impact immune responses.
Man, the entirety of section 5 should be its own essay emphasizing the perspective that basically anything can be psychosomatic. The heart attack and chest pain statistics are a significant update for me, and while I intuitively understood that some illnesses are heavily mixed - I knew this is very true for well-known-frequently-psychosomatic stuff like LC and back pain; incidentally, just the other day I had to wonder if a headache I was having was a recent stress or dehydration - I still didn't realize it was such a common pattern, and based on the previous discussions I'm guessing most people didn't.
I think part of the problem is that it is down to self-identification, so you will have real bisexuals and the social contagion bisexuals and the real Long Covid or post-viral syndrome sufferers and the 'maybe I have it! everyone else on Tiktok does!' lot.
The audience here will tend to skew towards "yes indeed real bisexual" along with a lot of other minority positions on things, but in the general population?
How do you define bisexual?
And again, I do incline to the view that people who are more on the left/liberal/progressive side of the fence will more easily identify as A, B, or C than people more on the right/conservative side, as well as older versus younger. If you're in a place and/or time where it's socially acceptable to be some variety of queer, then you're more likely to identify as that than if it's a place where you're in the closet unless you really, strongly, definitely are gay, lesbian or bi.
So how do you define bisexual? It's not "have you had sex with people of both sexes?" because according to this randomly selected site:
"People use a few common labels to identify their sexuality. Your sexuality isn’t defined by who you have sex with – it’s about how you feel and how you choose to identify yourself. The important thing is that you choose what label feels comfortable, or you choose no label at all. You might find, like many others have, that the label you choose changes over time."
So if you're young and you go "I think Barbie is cute", then you may go "Oh wow, am I bi/lesbian?" and the answer seems to be "If you think you are, yes! So long as you find that description comfortable, then you are, and when it's not, then you're not".
So, TMI time again. In my attempts to figure out what the heck was wrong with me over the years of my youth and adulthood, one of the considerations was "well, do you prefer girls then?" And considering it as a hypothetical, the idea of having sex with a woman didn't make me go "Ugh, no!!!" (Sex with a real person in reality being not at all the same notion because no thank you to anyone of any sex or gender).
I don't consider myself bisexual. *But* were I a young adult *now* and a good bit more left-leaning in social attitudes, I well might identify as such. "But you've never had sex with/dated a same-gender person!" Yeah, shut up bigot, not necessary for self-identification.
Same with long covid. I've had covid, and since then I do feel that I have a perceptible change in taste. I have bouts of fatigue, breathlessness, 'brain fog', many of the very vague and multiply-applicable symptoms.
But I can also put these down to getting older and existing health conditions.
So, do I have long covid? I say no. But again, a slight difference in inclination, and I could say "yes".
When answering the survey, I said no to both the bisexuaity and long covid parts. But rotate me 45 degrees, and I'd be answering yes. For the same physical entity with the same physiological symptoms.
Now you tell me: am I bi or not? long covid or not?
Is there research on if life or death social situations cause psychosomatic symptoms?
Like consider rooms clapping for Stalin or the purge by Saddam Hussein; I dont know how you'd ethically test such a situation if say >50% of the room had migraines, but would it be the most unusual thing in the world?
If you social group is burning witches maybe its for the best for your nervous system if you join in to the hysterical dancing, illnesses and people do just have a shut down switch when they believe there is lethal danger and their tribe is angry with them.
"It's possible that both trans-ness and bisexuality are the result of a sort of anatomical and neuro-hormonal "chimerism" that isn't present in either fully straight or fully gay cis people. I don't mean they're the result of literal genetic chimerism (which is far rarer than bisexuality or gender dysphoria), but rather, that they're caused by some sort of mismatch involving the body's neurological and hormonal transmitters and receptors."
This doesn't seem right. Wouldn't this mismatch she speaks of be more prevalent in homosexuals than in bisexuals? We've all encountered effeminate gay men and butch lesbians, have we not? They would figure to have the highest degree of this mismatch, I would imagine, but that's not where the incidence of long Covid seems to be the highest, right?
The link under "psychosomatic blindness" goes to http://file///C:/Users/Scott%20Alexander/Desktop/admin,+9729-34293-1-CE.pdf, which non-you people cannot access.
@Scott:
FYI, the first link in this sentence:
> Psychosomatic blindness has fallen out of style these days, but used to be quite popular - the British commander in the Revolutionary War had it.
Apparently links to a pdf file on your desktop, rather than to an actual web link.
Whenever you talk about psychosomatic illness, people read a whole lot of things into it that you didn't say.
To call something psychosomatic does NOT mean:
—that it isn't real
—that it can't cause real pain, distress, and disability
—that the person suffering is lying
—that the person shouldn't get treatment
—that the person doesn't deserve sympathy
It's saying: these symptoms do not have a physical cause.
"Christians and Republicans had no more Long COVID than people who said no religion or no political party, but polyamorous people and rationalists did."
During the pandemic, I noticed that the SF tech scene (i.e., polyamorous rationalists) went deeper into social distancing than any other demographic on the planet. People I knew who were actually immunodeficient did not isolate nearly as much as these guys. Social distance became a moral imperative and covid became The Worst Thing That Could Ever Happen To You.
"I think the worst-case scenario is that, since Long COVID is in the news, extremely sympathetic, and has maximally vague symptoms, its psychosomatic shadow could be much bigger than normal"
Especially if you were so traumatized by a positive covid test result, or maybe even just something that you suspect was covid but couldn't get a test for (remember that other infectious diseases are also still a thing), that your recovery was followed by a laser focus for anything that might be long covid. Because of course you must be injured! If you isolated like a champ only to discover that covid was a nothing burger for you, how foolish would that feel?
Long covid as biological sunk cost fallacy.
Coyne’s arguments convinced me that your survey effect probably isn’t real. There is huge difference between replicating an established result and establishing a new result. One of those is significantly harder than the other.
And if you (formally or informally) considered more than 3 hypotheses when looking at the data then a p-value of 0.016 is not significant after adjusting for multiple comparisons.
Quick note: Coyne expects confidence intervals from Scott, but I don't think it's possible to construct those given the type of analysis he used. I say this because I attempted to replicate Scott's numbers and provide these confidence intervals.
I'm assuming Scott used a chi squared test for independence. This is well and good for seeing whether long covid in bisexuals or not follow the same distribution. In tests like these we get a p-value and then see how small or large it is. Then we can reject the null or not.
This is not the time to construct confidence intervals. My understanding is we do that when we're estimating the parameters of a distribution. (i.e. what's the average height of Uzbeks?). Scott is not estimating any quantities here, so whatever confidence intervals Coyne has in mind don't apply here.
I also don't think we can construct a confidence interval around the p-value. Mathematically, it's just a single observation from a chi squared distribution.
As a digression, you can also see Scott as already having computed a confidence interval of some sort when he conducted the chi squared test. All a statistical test is is using the null hypothesis to say your test statistic will follow a certain sort of distribution (in this case, a chi squared one). You then say if the null is true, there's a .95 chance my statistic will lie in this interval. That interval looks a lot like a confidence interval.
Is "PHQ-1" a typo for "PHQ-9", or is it a tongue-in-cheek way of saying "just directly asking if someone is depressed"?
I'm late to the party, but see my three posts on Long Covid psychosomatic effects:
"Psychosomatic contributors to Long Covid suffering"
https://moreisdifferent.substack.com/p/psychosomatic-contributors-to-long
"The "false fatigue alarm" theory for Long Covid fatigue"
https://moreisdifferent.substack.com/p/the-false-fatigue-alarm-theory-for
"How to treat Long Covid as a brain-based (psychosomatic) illness"
https://moreisdifferent.substack.com/p/how-to-treat-long-covid-as-a-brain
Note I don't think all of Long Covid is psychosomatic, I discuss biological contributors in the 1st post. In my own case I suspect either Epstein-Barr reactivation or persistent virus for several months.
I explained this debate to my bisexual friend and she found the causality obvious: “bisexuals reproducibly sit in weird contorted positions which constantly make us hyperaware of our lung capacity.”
Sorry if this is a hopelessly stupid question. Is Scott’s use of the term “weird” in this piece
My thoughts on this are not fully formed, but I think you're missing out on the fact that sexual attraction is IN ITSELF subject to suggestion. When someone expresses attraction to you, that emotional state can be contagious-- people are turned on just by being desired. So bisexuals can be genuinely attracted to both sexes, while at the same time being more prone to social contagion that would lead them to delusionally believe they have a disease.
Is there a reason you're ignoring the absolute mountain of evidence of pathophysiology in long covid? This Nature review has the goods.
https://www.nature.com/articles/s41579-022-00846-2
"Hundreds of biomedical findings have been documented, with many patients experiencing dozens of symptoms across multiple organ systems7 (Fig. 1). Long COVID encompasses multiple adverse outcomes, with common new-onset conditions including cardiovascular, thrombotic and cerebrovascular disease8, type 2 diabetes9, myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS)10,11 and dysautonomia, especially postural orthostatic tachycardia syndrome (POTS)12 (Fig. 2). Symptoms can last for years13, and particularly in cases of new-onset ME/CFS and dysautonomia are expected to be lifelong14. With significant proportions of individuals with long COVID unable to return to work7, the scale of newly disabled individuals is contributing to labour shortages15. There are currently no validated effective treatments."