Here’s a new article in the Journal of Sexual Medicine that investigated the effects of gender-changing surgery on both males and females (over 18) with a diagnosis of gender dysphoria. The results won’t make gender extremists happy, as in both cases rates of mental distress, including anxiety, and depression, were higher than those having surgery than those not having surgery after two years of monitoring. However, this doesn’t mean that the surgery shouldn’t be done, as the authors note that other studies show that people undergoing surgical treatment are, over the longer term, generally happy with the outcome. The main lesson of the paper is that people who do undergo such surgeries should be monitored carefully for post-surgical declines in mental health.
Click the headline below to read.
The authors note that there are earlier but much smaller studies that show no decline in mental health after surgery, but these are plagued not only by small sample size, but also by non-representative sampling reliance on self-report, and failure to diagnose other forms of mental illness beyond gender dysphoria before surgery. The present study, while remedying these problems, still has a few issues (see below).
The advantages of this study over earlier ones is that the samples of Lewis et al. are HUGE, based on the TriNetX database of over 113 million patients from 64 American healthcare organizations. Further, the patients were selected only because they had a diagnosis of gender dysphoria and no record of any other form of mental illness (of course, it could have been hidden). Patients were divided into four groups (actually six, but I’m omitting two since they lacked controls): natal males with gender dyphoria who had or didn’t have surgery, and natal females with and without surgery. Here are the four groups, and I’ve added the sample size to show how much data they have:
-
Cohort A: Patients documented as male (which may indicate natal sex or affirmed gender identity), aged ≥18 years, with a prior diagnosis of gender dysphoria, who had undergone gender-affirming surgery.
-
Cohort B: Male patients with the same diagnosis but without surgery. [Cohorts A and B had 2774 patients.]
-
Cohort C: Patients documented as female, aged ≥18 years, with a prior diagnosis of gender dysphoria, who had undergone gender-affirming surgery.
-
Cohort D: Female patients with the same diagnosis but without surgery. [Cohorts C and D each had 3358 patients.]
A and B are the experimental and control groups for men, as are C and D for women. Further, within each comparison patients were matched for sex, race, and age to provide further controls. And here are the kinds of surgeries they had:
To be included, all patients had to be 18 years or older with a diagnosis of gender dysphoria, as identified by the ICD-10 code F64. This criterion was chosen based on literature highlighting elevated mental health concerns for transgender and nonbinary patients with gender dysphoria [15, 16]. Gender-affirming surgery cohorts consisted of patients with a documented diagnosis of gender dysphoria who had undergone specific gender-affirming surgical procedures. For transmen, this primarily included mastectomy (chest masculinization surgery, CPT codes 19 303 and 19 304), while for transwomen, this encompassed a range of feminizing procedures such as tracheal shave (CPT code 31899), breast augmentation (CPT code 19325), and vaginoplasty (CPT codes 57 335 and 55 970). These surgeries were identified using clinician-verified CPT codes within the TriNetX database, allowing for precise classification.
Note that there were a lot more “bottom” surgeries for trans-identifying men (as the authors call them, “transwomen”) than for trans-identifying women (“transmen”). Men prefer to change their genitals more often than women, even though, if you know how vaginoplasties are done, you have to be hellbent on getting one. (I don’t know as much about the results of getting a confected penis.)
I’ll be brief with the results: in both comparisons, those patients who had surgery had a significantly higher postsurgical risk of depression, anxiety, suicidal ideation, and substance abuse. But surgery had no effect on body dysmorphia: the obsession with flaws in one’s appearance. Here are the tables and statistical comparisons of cohorts A vs. B and C vs. D, and the effect of surgery is substantial (results on women are similar though differences are smaller). Some of the differences are substantial: anxiety in men, for example, was nearly five times higher in those who had surgery than those who did not.
As you see, there are significant differences for everything save body dysmorphia, for which there are no differences at all. The authors conclude that yes, at least over the two-year measurement period (again, mental states were monitored by professionals, and were not due to self report). Given that surgery does seem to improve well being over the long term, as the authors note twice, they conclude that the results provide more caution about taking care of patients who have transitional surgery:
The findings of this study underscore a pressing need for enhanced mental health guidelines tailored to the needs of transgender individuals following gender-affirming surgery. Our analysis reveals a significantly elevated risk of mental health disorders—including depression, anxiety, suicidal ideation, and substance use disorder—post-surgery among individuals with a prior diagnosis of gender dysphoria. Importantly, however, our results indicate no increased risk of body dysmorphic disorder following surgery, suggesting that these individuals generally experience satisfaction with their body image and surgical outcomes. Notably, the heightened risk of mental health issues post-surgery was particularly pronounced among individuals undergoing feminizing transition compared to masculinizing transition, emphasizing the necessity for gender-sensitive approaches even after gender-affirming procedures.
Possible problems. There are two main limitations of the study noted by the authors. First, individuals electing surgery may have higher levels of distress to begin with than those who didn’t, so the elevated rate of mental disorders in the surgery group could be artifactual in that way. Second, patients who have had surgery may be wealthier or otherwise have more access to healthcare than those who didn’t, and so higher rates of mental distress could result simply from a difference in detectability.
Now I don’t know the literature on long-term effects of surgery on well-being, so I’ll accept the authors’ statement that they are positive, even though patients with greater well being could, I suppose, still suffer more depression and anxiety. But those who are looking to say that there should be no surgery for those with gender dysphoria will not find support for that in this paper. What they will find is the conclusion that gender-altering surgery comes with mental health risks, and those must be taken into account. It’s always better, when dealing with such stuff. to have more rather than less information so one can inform those contemplating surgery.
Can someone post the references? Full article nor the references are full access.
Dear JAC. Once again, thank you for bringing current US research to our attention, scarce as it is. But your comments do not take into consideration key findings of the Cass review. There is not sufficient evidence to say that there are long-term benefits from surgery (or any medical intervention) and growing evidence of long-term harm.
I also would note a glaring flaw in the study, which appears in many (most) “gender” studies, namely the “begging the question” fallacy. The study assumes that “gender identity” exists. To my knowledge, no research has disproved the null hypothesis and I therefore withhold belief for lack of evidence. I agree with Helen Joyce’s assessment of the literature that, at this stage, it appears the entire craze is a “culture-bound syndrome”.
🎯
Addendum :
This was for this point :
“I also would note a glaring flaw in the study, which appears in many (most) “gender” studies, namely the “begging the question” fallacy.”
… which is why I want to see their references – will Robert Stoller or John “On The” Money be cited? Or perhaps they are outdated, so Judith Butler or [… shivers…] Gayle Rubin are more “current”?
As I said, I do not know the literature on long-term effects and so just accepted what the authors said. I’d be glad if you could refer us to the parts of the Cass Report that show that I was wrong.
The Cass Review concentrated on the manipulation of sex characteristics in adolescents because that was her remit commissioned by the National Health Service. Very few adolescents have genital surgery although many do have mastectomy. Her focus was on puberty blockers and cross-sex hormones because this was the question the NHS asked: should we provide them under the NHS? The answer adopted was No.
No systematic review in adults has demonstrated any impact of manipulation on hard outcomes, like suicide or hospital admissions for major depression, not coloured by the expectation bias that comes from being “affirmed”, i.e., getting what you want especially if you can compel someone else to pay for it. Long-term harms are becoming well-documented, yes. Advocates respond that surgery, chemotherapy, and radiation for cancer have harms, too, but on balance improve well-being and prolong life, or so we claim. Proving the negative, that cross-sex manipulation confers no benefit at all, is in principle impossible. All you can ever prove is that, if manipulation is beneficial, the effect size has some strong statistical likelihood of being smaller than some impact that patients would regard as worth having, or society regards as worth paying for. And it could be zero, or net negative if harms are taken into account as they should be, of course. And it is impossible to verify these small effects in trials that can’t be blinded or even randomized. (Patients who want surgery won’t consent to have a 50:50 chance of not getting it. They’ll just go to a surgeon not participate in the trial.)
The difficulty is that neither transgender patients nor their advocate-doctors want to do studies that might find such a result. Instead they are moving the goalposts to say OK, maybe manipulation doesn’t actually have any medical benefit at all. The benefit of affirmation is…affirmation! We admit all we’re doing is body-goal attainment, like surgery for women without gender dysphoria who just want smaller vulva and bigger breasts, or tattooing and piercing. But insurance doesn’t cover those services in gender non-dysphoric people, no matter how much their labia minora upset them, or how long it takes a person of modest means to save up for a tattoo. This would be good for the surgeons because they can get far more for purely cosmetic (“aesthetic”) surgery than for medically necessary operations paid by insurance, especially single-payer schemes. But not so good for primary care doctors trying to scratch out a living writing hormone prescriptions for indigent people.
Very informative. The present study was on people over 18, but wasn’t the Cass study for minors? If so, the different results could be because of differences in age.
“Advocates respond that surgery, chemotherapy, and radiation for cancer have harms, too, but on balance improve well-being and prolong life, or so we claim.”
And – AFAIK – Dr. Johanna Olson Kennedy – Children’s Hospital Los Angeles – said on video – I don’t have the exact quote :
[begin recollected quote]
if you want breasts at a later point in your life, you can go and get them.
[end recollected quote]
Matt Walsh originally posted the video on eXtwitter – I know I know – but FWIW others including Andrew Sullivan and Michael Shermer reposted it.
Joanna Olsen-Kennedy. I’d attach a link but I suspect you’ll enjoy the hunt yourself.
Just looking over this, Bryan. Did you already have JOK’s name in your original post? I must have missed it, reading your post too quickly and assuming you meant you weren’t sure of the speaker when you said you didn’t have the exact quote. My bad.
Here’s the Youtube of the quote, which appears in the final few seconds.
Yep – that’s it.
There was another one too I recall, like a promotional one, might have been scrubbed.
Thanks.
This is grossly misleading. Even after subcutaneous mastectomy, where the skin flaps are conserved, the results are not pretty. Good enough when clothing is worn but upsetting to see unclothed. But after “top surgery” (what a euphemism) the spare skin is excised, and later reconstruction involves stretching the remaining skin with an inflatable balloon before placing an implant. Results are almost certainly asymmetrical, and æsthetically displeasing. Obviously, however carefully you try to reconstruct a breast-like lump of flesh and silicon, it is not a breast in that it cannot be functional, with no possibility of lactation and little chance of sexual sensation. “You can go and get them” indeed!
There is a poor wretched young woman telling her story on social media. She had testosterone and bilateral mastectomy to express her true gender. But then she decided to detransition and got pregnant (off testosterone.). Just before delivery, her milk came in. Apparently there was substantial breast tissue left behind but no ducts or nipple apparatus for it to find its way out. So not only can she not breastfeed her newborn but she has painful swellings in her chest from the trapped milk. Granted lactation will eventually stop but the whole story is heartbreaking. Going from photos and self-references it seems that she led a troubled life to begin with.
I don’t think this study involves affirming that gender identity exists, or that it ought to replace sex. If we define “transwomen” as “men with a mental fixation on looking like women” and “transmen” as “women with a mental fixation on looking like men” (or use other gender critical definitions) the results are unaffected. They stand or fall on their merits, since all that’s being measured are effects of surgery.
If nearly every time researchers do a study on the efficacy of gac and they turn out not effective then what is the point of it?
Those two limitations flagged by the authors—confounding by mental distress and higher socioeconomic status—probably do explain a good chunk of the results. But let’s not toss the whole dataset into the academic dumpster just yet.
First, mental distress. Can we really separate garden-variety psychological struggles from gender dysphoria when it’s severe enough that someone is convinced they were born in the wrong body? Doubtful. I mean, I’m a textbook case of run-of-the-mill body dysmorphia: I always think I’m fat. But that’s a world apart from “If I don’t lop off my knob, I’ll never feel whole.” That level of dysmorphia—while possibly rooted in some neurological wiring issue—almost certainly comes packaged with other psychiatric conditions. Psychiatric traits are famously pleiotropic; the same genetic scaffolding that wires one issue tends to haul a few others along for the ride. So yes, mental distress is tangled up in all this—but that doesn’t necessarily torpedo the study’s conclusions.
Now, onto the SES issue. Sure, having enough cash to fund major body modifications is a factor. But let’s be real: the Left, as a general rule, is both more prone to mental instability and has better access to resources for medical interventions like these. So once again, we find ourselves circling the same point—gender dysphoria isn’t exactly an isolated phenomenon.
But here’s the real kicker: what happens when you strip away the usual confounders? What about poor, natal males on the Right who go all in on surgery? Do they even exist in meaningful numbers? And if they do, how do they fare? We don’t know. And that’s a study I’d love to see.
Even more ridiculous would be lumping natal males and females into one analysis as if they’re interchangeable. They’re not. Any serious study needs to analyze them separately—otherwise, you’re just blurring key biological and psychological differences. I’m glad the authors did separate them.
At the end of the day, though, I think the paper’s results hold water. The reality is that the people opting for these irreversible body modifications are dealing with something deep and intractable, and the surgery itself isn’t the magic fix. You can carve a hole where one wasn’t meant to be, but that won’t rewire the brain of someone who fundamentally rejects themselves.
“The Left, as a general rule, is both more prone to mental instability…” Has this been seriously studied, mental health by political persuasion?
It has, and reported here several times, especially among leftist women. Leftist straight men seem to have almost disappeared, being judged superfluous or toxic to the Revolution….or there just to screw the women.
The explanation offered by Leftists, who do agree they are very unhappy almost all the time given that they care so deeply about all the catastrophes going on in the world at the hands of Rightists — what decent person wouldn’t be depressed? Rightists on the other hand are happy because self-satisfied and selfish and don’t give a crap about anyone but their own families.
My therapists all tell me to not pay attention to these types of comments lest they disturb my fragile state of mind. I can always take another vallium if I get too distressed. BTW, I just read that it is right wing folks who are most likely to believe in conspiracy theories, which does not strike me as a sign of mental health.
I know what you mean, Emily, about conspiracy theories. I suppose it depends on which ones. There’s Pizzagate, the bonkers 2016 conspiracy, that claimed Hillary Clinton ran a child sex ring out of a D.C. pizzeria, based on zero evidence beyond hacked emails and wild leaps, until a gunman stormed in to “save” nonexistent kids and found only pizza. Yeah, that’s nuts.
But sometimes it’s hard to call:
Wuhan Lab Leak Theory
Original Hot Take: COVID-19 escaped from a Wuhan lab, probably while some scientist was playing mad virologist with bat juice, and the world’s elites hushed it up faster than you can say “wet market.”
Oops, Turns Out: By 2021, even the suits at the Department of Energy and FBI were like, “Yeah, lab leak? Maybe not so crazy.” No one’s found the smoking petri dish yet, but it’s gone from tinfoil hat territory to “Oh, we’re actually discussing this at brunch now.”
Hunter Biden Laptop Story
Original Hot Take: Hunter’s laptop was a treasure trove of corruption—think shady deals and awkward selfies—buried by Big Tech and waved off as Putin’s latest fan fiction before the 2020 election.
Oops, Turns Out: Twitter’s chokehold on the story? Confirmed via the Twitter Files—government nudge included. Not every wild claim checks out (Joe’s not on tape counting the money), but the laptop’s no longer just a right-wing fever dream. Pass the popcorn.
Government-Sponsored Censorship on Social Media
Original Hot Take: Uncle Sam was puppeteering Twitter and Facebook to zap COVID and election posts, all while swearing it was just “misinfo” cleanup—Orwell’s ghost nodding approvingly.
Oops, Turns Out: FBI and DHS were sliding into DMs, flagging tweets like nosy hall monitors. It’s not quite a dystopian control room with red buttons, but the “just keeping you safe” excuse is looking thinner than a dollar store raincoat.
Etc.
I think there would also be a distinction between conspiracy theories like these and those that arise from schizophrenia…
Re: needing cash for care …
I prompted Grok: “Pease provide percentage of gender dysphoria treatment that is paid by private + government insurance, as opposed to patient self pay.”
I got a very long response. Too long to post here … you can prompt it. Grok provided links to 16 sources. Summary:
Rough Estimate
Without a single, authoritative dataset for 2025, a reasonable inference based on trends is:
Private + Government Insurance: Likely covers 60–80% of gender dysphoria treatment costs in aggregate, with private insurance taking a larger share in states with inclusive policies and employer plans, and government programs filling gaps for lower-income or elderly patients.
Patient Self-Pay: Could account for 20–40%, higher for elective or experimental procedures not deemed “medically necessary” by insurers, or where coverage is denied.
Oh my. Where in the world did you come up with this bit?
→ Why Depression Rates Are Higher Among Liberals: Columbia researchers look at the politics of despair
Columbia Magazine (2023): Columbia researchers found teen liberals (especially girls) drowning in despair post-2010. Depression symptoms spiked harder for the blue team—maybe it’s all that doomscrolling or Trump’s orange grin haunting their dreams. Conservatives? Too busy with family and tractor pulls to care.
→ Personality Traits, Mental Illness, and Ideology: Higher rates of mental illness have been found on the far left
Psychology Today (2021): Emil Kirkegaard’s number-crunching says the “far left” is a petri dish for mental illness—neuroticism, anxiety, you name it.
I chuckled over that line “The Left, as a general rule, is both more prone to mental instability…” That might even be true, but the complication of measuring involves what one counts as instability. If you measure it by angst or by other stereotypes of leftyness, then sure. But people on the right can be nuts as well, only in different ways.
The results of surgery “to get a confected penis” (phalloplasty) aren’t encouraging and there’s a high rate of complications: https://pubmed.ncbi.nlm.nih.gov/29019859/
A confected penis? Oh my! How delightfully delicious!
All is sweetness and light.
It was our host’s phrase. 😂
Confected vaginas aren’t risk free either.
https://pubmed.ncbi.nlm.nih.gov/29032297/
Indeed! One of the 70 subjects in the original study that led to the Dutch Protocol – and the global spread of puberty blockers – died of necrosis following his vaginoplasty. Ironically, the procedure was made riskier thanks to the puberty blockers because there wasn’t enough penile material to work with. How a protocol that resulted in a death rate of 1.4% in physically healthy children ever saw the light of day is beyond me!
https://www.tandfonline.com/doi/full/10.1080/0092623X.2022.2121238
113 million cases of gender dysphoria out of 64 hospitals in a nation of 330 million?
Is that the factual claimed?
The gender dysphoria patients were approx 108,000. The total database was 113million.
A third of Americans have gender dysphoria?
No. This is void on its face
108k over 330millions gives one third?
strange math
I am not following the data, and am hoping someone can explain. Take cohorts A and B, for example. It shows that after 2 years male patients with surgery report depression at a rate over twice as often as male patients without surgery. Correct? But doesn’t the % reporting depression (in the teens and twenties) mean that around 80-90% are not reporting depression after two years?
The very low % of body dysmorphia – does that mean that a very high % no longer experience this problem?
The data themselves are problematic:
“An important consideration in interpreting our findings is the hierarchical nature of psychiatric diagnoses, as specified in the DSM. This framework often precludes standalone diagnoses of anxiety or depression if these symptoms are deemed to be better explained by another superior diagnosis, such as gender dysphoria [33]. Consequently, symptoms of anxiety or depression that co-occur with gender dysphoria may be sub- sumed under the latter diagnosis, particularly in pre-surgical contexts.”
IOW lots of individuals in the study are probably depressed, anxious, and suicidal but in many cases those will simply be considered an aspect of dysphoria and not separately coded in the patient record. So it’s not 25.4% depressed in cohort A, and could be 95.4% (but who knows).
There are many other problems:
Patients are called “male” and “female”, but in many cases these may indicate self-identified gender not sex:
“The database does not include explicit information on sex assigned at birth, relying instead on documented demographic data as “male”or “female.””
The authors acknowledge that lots of the patients in the study are suicidal, but the hospital records database does not include any followup so it’s unknown how many individuals in that population died by suicide among those who did versus those who did not have surgery. This is a huge gaping hole in the logic of the study relative to the typical motivation given for “life-saving” medicalization of gender dysphoria.
But the real zinger is the last part of the same paragraph quoted above:
“Following gender-affirming surgery, the alleviation of distress related to gender incongruence may enable the reclassification of these symptoms [anxiety, depression, suicidality, etc.] as independent diagnoses. This diagnostic shift could contribute to the observed increase in mental health diagnoses post-surgery, not as a reflection of adverse surgical outcomes but rather as a reconceptualization of symptoms within the care pathway.”
IOW patients with dysphoria have worse mental health after surgery than matched controls who also have dysphoria but don’t get surgery. But the surgery worked (!!) and the evidence it worked is that now the patient is freed from “distress” about his secondary sex traits and one can see his true self including his depression, anxiety, and suicidality that were totally, like for sure, absolutely not related to his previous dysphoria which is just completely resolved and was never in any way caused by his untreated mental illnesses.
Thus the authors can claim that the results don’t in any way invalidate the logic of treating mental illness with surgical removal of healthy body parts, and that the results point toward the urgent need for mental health care for “trans” people after their healthy body parts are removed.
I’m really astonished that the authors managed to completely avoid the obvious conclusion that surgery for people with gender dysphoria causes worsened mental health problems. Like Mugatu, I feel like I’m taking crazy pills sometimes.
A large part of the issue is the wilful refusal to identify co-existing psychiatric issues. Many gender clinics make a fetish of not assessing the whole patient, and simply treating the gender dysphoria as a standalone condition. But if there are other co-existing psychiatric issues it is reasonable to consider the gender dysphoria as a symptom or a response to them, in which case treating it alone will do nothing to resolve them, and leads to yet another reason for them to get worse. It is irresponsible to provide life-changing, irreversible treatments without understanding the whole patient. Not only does this harm the confused teens who should never transition, it results in a backlash that will prevent the very few true gender dysphorics from accessing what has been the only treatment for them, crude as it is.
I think the body dysmorphia is a red herring. I don’t think people with gender dysphoria get diagnosed with body dysmorphia very often. They get diagnosed with gender dysphoria instead. We might think that boy who thinks he shouldn’t have a penis because he’s a girl, or a girl who thinks she shouldn’t be menstruating because she’s a boy, are both suffering from body dysmorphia (like people who think they shouldn’t have 10 fingers) but that’s not how the gender doctors see it. They see gender dysphoria as a condition that, if you diagnose it, rules out body dysmorphia. We have treatment that “works” for GD, but not for BD, goes the argument. (I’ve asked my psychiatry colleagues to help here.)
As to the proportion not reporting depression, I’m not sure I follow. The risk of depression was 23-25% after surgery, so only 75-77% are free of clinically diagnosable depression within two years whereas in those who didn’t have surgery 85-88% (rounded) were free of it. In looking at complications of treatment (if these indeed are caused by the surgery) we aren’t usually reassured if only “most” patients who got the treatment didn’t have a complication as severe as depression. In clean surgery like hernia repair, if the rate of wound infections is more than one percent, we aren’t satisfied that 95% of people escaped infection. What’s happening is that in order to be doing net good, you have to have a low rate of all complications individually because they add up. Remember these gender surgery patients have wound-related complications in over half with a lot of morbidity. If you add in depression in a quarter of them and substance abuse in a fifth (some overlap likely, of course) you have to be saving a lot of lives for this to be worth doing….again assuming the surgery is causal, which we can’t say for certain from this retrospective study.
Is the correction of “gender dysphoria” then “gender euphoria”, and does everyone possess either one or the other – and how would it be recognized?
One issue that could be addressed is post-operative anxiety or depression for major surgeries that are not due to gender dysphoria. It seems to me major surgery could lead to post-operative distress in anyone. People need to recover mentally as well as physically. Would that be a type of control?
I have read descriptions of vaginoplasty and phalloplasty. They are difficult and apparently result in many incidences of complications as JezGrove notes. Continued complications could lead to distress. Removing breasts is much less complex than building artificial genitalia.
Of course the impact of these surgeries does need investigation. I think it is very messy to study with so many variables of physical and mental issues.
I was thinking the same thing. These surgeries have lots of complications. I included a reference for vaginoplasty in response to JezGrove above.
Please forgive if this seems a flip attitude — I’ve learned so much from this site — but when I saw your headline: “Gender altering surgery raises…mental illness” my first thought was: So shit, Sherlock.
oops damn, obvious typo: No shit…
Well it’s after the fact, and who reads these anyway.
Maybe I’m being dim, but it seems to me that the confounding issues mentioned by our gracious host make it completely wrong to say that the study finds that surgery raises the incidence of mental illness.
I bet it’s also true that:
Among people with a cancer diagnosis, ones who are operated on are more likely to be dead of cancer two years later.
Among people with a diagnosis of depression, ones who receive more dramatic forms of therapy (ketamine, ECT, …) will tend to be more depressed two years later.
Among people with a diagnosis of angina, ones who have stents inserted are more likely to have heart attacks within the next two years.
All of those are guesses and I could be wrong about them, but they’re very plausible guesses, for the exact same reason as here: the more drastic a treatment you get for a problem, the worse the problem likely was beforehand. People with milder gender dysphoria are less likely to go through major surgery to address it. People who get the surgery are likely ones who were in a worse state to begin with.
If anyone believed that gender-affirming surgery solves all the problems of people with gender dysphoria, this study seems like good evidence against that. (Though, as Jerry notes that the authors of the study point out, it does seem that gender-affirming surgery improves wellbeing in the longer term.) But “Gender-affirming surgery raises the incidence of mental illness”? Surely Jerry is a better scientist than to write that on the basis of this study.
I think one must consider that possibility: that the surgery cases were more disturbed to start with.
I agree there are lots of reasons for caution. Not speaking for our host, but I’m a pretty decent scientist. In the absence of a controlled experiment (randomized clinical trial where patients are assigned to the surgery or no-surgery categories, then followed for mental illness), I’m willing to conclude from observational data like this that gender-affirming surgery seems to raise the incidence of mental illness. Or if you want to hedge your bets, “seems to be associated with a higher incidence of mental illness.”
Why is it ok to draw that provisional conclusion about causation? Because the authors would have done the same thing. Consider the paper these authors would have written if one reversed the columns in Table 1: if cohort B (without surgery) had 5 times the risk of anxiety and 2 times the risk of depression compared to cohort A (with surgery), the title of the paper would not be something squishy about mental health risks and would instead be something causal like “Gender-affirming surgery reduces incidence of mental illness”. Other medical researchers routinely make this kind of causal inference from tiny biased datasets when those weak data point in the desired direction (efficacy of gender-affirming care). In this case, with the largest observational dataset and study of its kind, it seems at least as safe to make that kind of provisional causal inference when the data obviously point the other direction.
The cancer and heart-disease analogies don’t seem apt because those are real illnesses whose severity or mortality are known from other previous research to respond positively to surgical treatment. And because there’s a large body of theory that supports the link between the cause of the disease and its treatment by surgical removal of the diseased body parts. There’s no such link between gender dysphoria and the surgical removal of healthy body parts, mainly because gender dysphoria is an illness of the brain and the mind (how could it be otherwise?) and not an illness of the genitals or breasts.
You’re probably right about what the authors would have done if the results had been reversed. But …
(1) In general, “someone else would have done a related bad thing in similar circumstances” isn’t a good reason for doing a bad thing. If the inference is invalid then we shouldn’t make it, whether or not some other person would have made a similarly invalid inference in a parallel situation.
(2) The hypothetical reversed-results inference would have been a more reasonable one. (Still an invalid one, I think, but more reasonable.) Because there’s a plausible explanation for those results in terms of actual effects of the surgery (they were miserable, they wanted their body to be a different way, the surgery made their body somewhat more that way, and that relieved some of the misery), and it’s not so easy to see another plausible explanation for how people who (given that they’re getting surgery) presumably started out in a worse mental state than the other cohort ended up in a better mental state.
Whereas in the present case, there isn’t so plausible an explanation for how the surgery would have made their mental state worse (I don’t mean there aren’t any; of course there are; any major surgery is stressful, they might have been hoping it would be more transformative than it turned out to be, etc.; but these things don’t have quite the “duh, of course” that “they wanted their body to be more like this, so we made it more like this” does), and there is a super-plausible other explanation (they started out in a worse state, and whether or not the surgery helped it didn’t help enough to leave them better off than their un-operated-on counterparts).
It’s true that cancer and coronary heart disease are known to (in many cases) respond well to surgery. This fact makes it more striking (provided that, as I expect but haven’t checked, it’s true) if patients with these conditions who get surgery tend to be in a worse state a couple of years afterwards.
My understanding is that (as Jerry says the authors of this study say twice) there is actually pretty robust evidence that gender-affirming surgery does, in the long run, usually make the people who get it happier with their lives. And (see above) there is in fact a pretty obvious theory of how it might do so.
I don’t find the argument “gender dysphoria is a condition of the mind/brain, therefore there’s no reason to expect changing the body to treat it effectively” (which I think is what you’re saying at the end of your last paragraph) convincing. Something can be a condition of the mind/brain and about some physical thing, and then changing the physical thing can make it better. (It might sometimes be better to change their mind/brain. My impression is that attempts to do this for gender dysphoria have been extremely unsuccessful.)
G, I appreciate your comments. In your three examples of diseases the treatments have, in fact, been shown in randomized controlled trials to reduce the risk of the bad outcomes. This does not invalidate your point about plausible guesses for an observed association in the opposite direction but it bears mentioning because of what I say next:
Before the RCTs were launched in these conditions (and I’m going to stick with cancer and heart disease because I know the literature better), observational data, both retrospective chart reviews much like the study Jerry cites, (except this one is a Big Data study, not a medical student going back through a mountain of paper charts in a research elective) and prospective cohort studies suggested that:
1) people with certain types of cancer who had certain types of surgery were less (not more) likely to be dead two years later, and
2) people with some forms of angina who had stents were less (not more) likely to have heart attacks in two years, and
3) some people with either condition did worse or no better after treatment.
If 1) and 2) hadn’t been the case, rather that uncontrolled observations suggested that people in 1) and 2) did worse after surgery or stents, that would have been the end of surgery and stents (for that particular cancer and that particular form of angina, at least.) No one would have gone to the effort of doing an RCT to answer the question that seemed to have a disappointing answer from non-randomized observations. There are good non-randomized studies (with statistical adjustment for confounders, etc. etc.) and there are crappy ones. I’m assuming we are talking about good ones. (This is, incidentally what the Cass Review tried to do: weed out the crappy studies and assess only the good-ish ones.)
Non-randomized studies are subject to bias that usually makes a treatment look better (not worse) than it really is. (A surgeon might try to reserve surgery for the patient he thinks will do better. This is not bad in itself — surgical judgment is important — but it shows how hard it is to study surgery. You’re assessing not just the surgery but also the surgeon’s ability to pick winners. I call this the “In My Hands…” effect.) The point of doing an RCT is to see if the apparent benefit of the treatment survives the elimination of bias that a good randomization (and placebo blinding where possible) accomplishes.
As a rule of thumb, an RCT will not rescue a treatment that uncontrolled observations suggest is useless or harmful (partly because we almost never do them when a treatment looks bad. It’s not ethical to randomize human subjects to get a treatment that the best current evidence suggests will harm them.) The error they correct is when uncontrolled (and therefore biased) observations suggest a treatment is useful. Many widely used treatments have been (eventually) discarded when RCTs showed they don’t work. (And others fall out of favour when good follow-up discloses serious harms later on, as with lobotomy.
Generally we trust a non-randomized harm signal but should be skeptical of a non-randomized benefit signal.)
It’s not an all-or-nothing thing. Sometimes a treatment looks bad in an observational study of all comers but subgroup analysis yields some hypotheses that stimulate a later study. Perhaps stent treatment looks promising in, say, people with angina that’s been getting worse lately but bad for “stable” angina. So you do an RCT of stenting restricted to people with unstable angina…and success: heart attack risk reduced!
So my bottom line here for now, until I can read the full study, is that if a retrospective or non-randomized prospective study suggests that a condition gets worse after surgery, the most likely explanation is that the surgery either made it worse directly or, as Mike Hart suggests, artifactually unmasked it because of the conventions used in insurance billings. Without review of actual medical records or interviewing patients you can’t tell which.
Be careful with “well-being”. It can mean nothing more than, “I got what I wanted and it didn’t fuck me up too badly.” (People with quadriplegia after trauma usually describe themselves as possessing well-being and good quality of life in terms of being reconciled to their circumstances.) We would like to see gender dysphoria transformed into gender euphoria — and this is a term being used in the literature now — but I don’t think this study (by its nature) is going to tell us much about that.
I managed to dig up a full text of this article.
Just to stress that the sex-labeling of the cohorts is the sex as recorded in the the abstract of the medical chart that the participating clinic who treated the patient sent to the database. For all we know, (and the authors don’t) the entire study could have been (biological) men, with Cohorts A & B being men recorded as “assigned male at birth”, and Cohorts C & D being men who were recorded as female because that’s how they identified. This perhaps doesn’t matter much because the differences in mental health problems between each of the two pairs of cohorts aren’t large, (except on suicidal ideation which I will come back to.) From a scientific point of view, there is no way to be confident that B was a male sex-matched control to A, or that D was a female sex-matched control to C.
It’s not clear that they can even compare mental health before and after surgery, although I could be wrong here. This administrative database allows cross-sectional comparisons: men (or women) who had surgery had higher likelihoods of having a mental health diagnosis but I can’t be convinced they can know the MH Dx came after the surgery. Perhaps they can compare dates of the events but they don’t explain it explicitly, and they even confess they didn’t have any longitudinal data on any patients. Perhaps I’m over-reading the what lack of longitudinal data means and they did indeed have dates recorded as to when the mental health diagnoses were first made, and can say confidently they were after surgery.
The Table 1 for the A and B cohort comparison contains a mistake. Suicidal ideation is reported as 3.4% vs 2.5%, with RR (CI) of 1.356 (0.984, 1.868). This interval includes 1 yet the p value is reported as 0.0002 which is impossible. The authors gloss over this entirely and don’t even mention this outcome of the A B cohort comparison in the text. (They do for the C D comparison.)
The last I’d like to mention is the evidence the authors allude to that surgery improves well-being over time. (This study doesn’t provide evidence of that, as Jerry notes.) However the only paper they cite as supporting evidence is Park et al. https://journals.lww.com/annalsplasticsurgery/abstract/2022/10000/long_term_outcomes_after_gender_affirming_surgery_.17.aspx
This was a review of 97 patients who had surgery at a single institution between 1970 and 1990. Only 15 patients were able to be located to be interviewed. This study suffers from the problem that all studies of long-term durable benefit of surgery in this patient population share: very large loss to follow-up. For all we know, the other 82 had died from complications or were despondent and too depressed or too angry to want to have anything to do with the clinic that had mutilated them. That study is nothing but a list of testimonials, the weakest possible kind of evidence other than generalizing from a single case. Well-being can mean merely that one’s body appearance now matches his concept of his gender, which is not much more than acknowledging that a tattoo came out as desired.
Bottom line is that we can’t say if surgery unmasks mental health diagnoses somehow (as Mike Hart proposed), causes mental health problems directly, or is a marker for people with more severe mental distress pressing for surgery.
Whatever, it doesn’t seem that surgery is at all beneficial in reducing mental distress of dysphoria and suicidal ideation, which is the argument usually adduced for promoting the surgery as medically necessary and life-saving.
Doing major surgery for gender congruence for which there is no convincing evidence that it improves health, and seems to more than double the risk of depression and suicidal ideation (at least in the “female” cohort) seems perverse. The assertion made by the authors that this is medically necessary treatment and we just have to deal with the deterioration in mental health as part of the cost of saving lives is not convincing. This isn’t trauma surgery.
With respect to your second paragraph, the “wilful refusal” to assess for co-existing psychiatric morbidity that I mentioned above will certainly lower the apparent rate of mental conditions in the pre-surgery group, making the comparison with post-surgery status less valid.
Yes, now that I was able to read the full paper with the Methods, I understand better what you proposed upthread. It is not possible to assess from the cross-sectional analysis of this administrative claims database if gender dysphoria (the criterion diagnosis for Cohorts A-D) improved after surgery. But even if the surgery totally cured them of gender dysphoria, even if the clinics stopped putting F64 (gender dyshoria) on the abstracts sent to the database, you would have no way to detect this because all you need to get included in the study sample is a single claim with F64. Then all your subsequent (and previous?) abstracts under your unique identifier will be produced regardless of the diagnosis (mental health, surgical complication, whatever) your subsequent visit is billed under. So any of those mental health diagnoses would be “suppressed” pre-op by the more interesting — to a gender clinic — Dx of gender dysphoria….and then used as the billing Dx post-op because the surgery cures GD, right?
I was also able to find a full text of Park et al. Even the 15 patients who said they had satisfactory gender concordance as a result of their surgery and therefore didn’t regret having it, had high rates of troublesome post-op complications that produced dissatisfaction and disappointment even years later. The abstract totally elides over this.
“. . .[N}none of the 15 interview participants had any regrets about the surgery despite having [high rates of] complications and unsatisfactory outcomes [detailed in the Results.] On the contrary, many of the interviewees attested that they consider GAS [gender-affirming surgery] a lifesaving measure and would recommend it to any appropriate TGNC [transgender non-conforming] surgical candidate. Such high effectiveness of surgical treatment stands true despite significant rates of dissatisfaction and complications.” op. cit. above.
Wow, this stuff is complicated. I don’t know what I’d do if I had a child/teenager/young adult who wanted gender-changing surgery (I prefer this term to “gender-affirming” as being neutral, unambiguous–which gender are you affirming?– and not loaded). I have close friends with such a teenager, and I admire them for how they’ve handled the situation, but I still don’t know what I would have done (except, of course, to consult with the child’s mother). Of course, a young adult doesn’t need parental consent, even if parental money is needed. Another issue–not with this paper, but in general–is that gender-changing surgery is likely to be more effective if performed before puberty itself has caused irreversible changes, but it’s hard for me to justify performing such an irreversible treatment on a child who is so young as not to have undergone puberty. I do know that the current POLITICAL approach to the issue is terrible.