The methods section is completely and utterly absent of methods. They show where they got their data from, and say that rates of psychiatric encounters were calculated from these data. This is a problem, because in their results, they include p values which indicate some kind of statistical test (I would assume a t-test judging from the kind of data they are using) was performed, but they do not say what.
Oh, look at you! Knowing statistics, p-values, and trying to guess where they arrived at that p-value from because p-values can be manipulated a million and one ways to arrive at the conclusion you want. I have the answer for you: those are not p-values. 'P' here is shorthand for 'proportion,' not probability. Or they are assuming probability means proportion, even worse.
Take a gander at the chart:
Regarding vaginoplasty, they cite n=869, with 193 experiencing an overall psychiatric encounter (193/869=22.2%).
Regarding Suicide Attempt, they are dividing 38/869 = 4.4%.
Overall risk before is 74/869 = 12.8%, and presented-before is 38/(74+38) = 33.9%.
I can't figure out where they're getting the 'did not present before" thing from. No matter in what convoluted way I divide these numbers, I cannot make 81/869 into 11.1.
That's it. That's all. Oh, and if you think that's a pretty big tell, well check out my next point:
It's an interesting result essentially, but needs to be verified with properly conducted studies.
It's barely a study. Here's their methodology:
All persons undergoing feminizing (vaginoplasty) and masculinizing (metoidioplasty/ phalloplasty) genital gender affirm-ing surgery were identified in California from 2012-2018 from the Office of Statewide Health Planning and Development datasets using appropriate ICD-9/10 codes. All emergency room and inpatient psychiatric encounters were identified (noting those involving suicide attempts). Rates of these encounters were calculated before and after gender affirming surgery (using date of first operation if staged). On average, patients had a balanced two years of observation before and after their surgery.
This is closer to a pull of an SQL database. It probably was. The study did not have contact with individuals nor poll them - it just looked through records in a California database. Which is to say - the researchers have no idea if these psychiatric encounters have anything to do with the gaping axewounds and the rotdogs. They did not have contact with any of the people being observed here. So we have to assume that California's results are accurate, well-kept, and we can't really extrapolate anything from them because there's no... control group or proportioning or anything. They just scanned the database and shat out some numbers. There's a reason this one never went anywhere.