Hacker News Reader

Comment by bill_joy_fanboy | Hacker News Reader

One way to cut the nonsense is to look at things that can't be fudged.

Suicide is an example. You can't "overdiagnose" suicide unless you outright lie that it happened. The rate for suicide was something like 10/100K in 2001 and is hovering at 14/100K now. This is a significant increase.

If more people are killing themselves, it's almost certain that there are far more people are also miserable or struggling than there used to be.

This seems like pretty direct reasoning, but I suppose it could be flawed.

View on HN

Yep, suicide rates increased by 20-30% across all races, ages, and genders over roughly a decade between 2006 and 2018 https://www.cdc.gov/nchs/products/databriefs/db464.htm

> Suicide is an example. You can't "overdiagnose" suicide unless you outright lie that it happened.

People have been outright lying about suicides for a long time.

If it's counted as an accident rather than a suicide then the numbers are still high.

If it's counted as a suicide rather than an accident then maybe. I would then ask who is lying.

If it's faking deaths then that's the coroner system which you would have to question.

Or we have gotten better about recording suicide today and they were under-reported years ago.

Do you mean in regards to under-reporting to save face?

"Died cleaning his gun" etc, yeah.

You absolutely see this in insular communities - lots of attempts to hide abuses or family politics that led to a suicide, by hiding it behind accidental death of some kind.

Yes. Also if you put yourself in the shoes of a coroner/investigator, how do you determine if it's a suicide vs an accident? And would you get the context you need from the family/friends/colleagues of the deceased, keeping in mind they might not tell you, or lie because they don't want their loved one to be remembered as someone who lost their battle with mental illness?

It's kind of undeniable that there's a rising mental illness crisis in the US and undoubtedly more suicides, but keep in mind that suicide, like all symptoms of mental illness, is hard to track accurately. Especially as the stigma towards mental illness is starting to cede.

> Suicide is an example. You can't "overdiagnose" suicide unless you outright lie that it happened.

But you can do that. Throughout much of the 20th century (in the United States, at least), the cops might be willing to just call it an accident (he was cleaning his gun) or an unsolved homicide. Ostensibly to spare the family shame or to allow the family to bury them in the cemetery of their choice.

Furthermore, there seem to be factors that cause upswings and downswings in the suicide rate. There were alot more people jumping out of skyscraper windows in 1929 and 1930 than there were in 1925 and 1926, for instance (though we might expect the opportunity to have been roughly the same). Now, that doesn't directly rebut your argument here. But it would mean you'd need to show that if there is an upswing today that it has lasted longer or otherwise been worse than similar ones in the past, or that it doesn't seem to have the same sorts of causes.

If possible, it might actually be better to pick some other measurable event/behavior than suicide. Psychotic breaks, schizo episodes, or voluntary commitment.

NB, this the fundamental reasoning behind Emile Durkheim's seminal and foundational work on sociology, Suicide: A Study in Sociology (Fr.: Le Suicide: Étude de sociologie, 1897).

The reasoning being that it's very difficult to hide bodies, and that statistics on suicide tend to point to some very deep-rooted pathology within a society.

Yes, it's possible to fudge those statistics somewhat (the distinction between "accidental" and "intentional" deaths being salient), but far less so than other indicia of psychological trauma, particularly at population levels (the focus of sociology).

<https://en.wikipedia.org/wiki/Suicide_(Durkheim_book)>

I think you are probably correct that suicide is a decent proxy measurement. It is possible that it could be skewed by changes in how likely people are to honestly report suicide; as opposed to calling it an accident.

In which case the skew is positive.

If suicide is directly correlated to mental illness, then what of euthanasia and assisted suicide? Anyone who applies to end their own life would be incompetent to make that choice, due to mental illness.

Those modalities should be discernable when looking at data by age (assisted death tends to occur at older ages) and, with a slightly greater data burden, comorbidities.

When looking at data, an otherwise healthy 20 year old, and an 80-something with multiple chronic and/or terminal Dx's, would stand out.

The challenge in looking at population-level data is that the Dx may well not be available, though other indicia might suggest this (e.g., residence is an assisted-living or long-term care facility). In practice, studies are frequently limited to available data, and proxies are frequently used. These may not be appropriate for specific individuals, but at scale, the law of large numbers and general correlation are your friends.

Like you said, I think looking at younger ages is sufficient.

In 1999, teen (15-19) suicide was at around 8/100k, and it's now around 11/100k. It's lower in absolute numbers than in the general population, but there's still a disturbing uptick.

https://www.charliehealth.com/research/the-us-teen-suicide-r...

It would be interesting to know how many people committing suicides were on SSRIs or antipsychotics, etc. at the time. These drugs are known to induce suicidal ideations (and ideations of violence, aggression, and homicide) and have carried black-box warnings. Sometimes, the cure is worse than the disease. Oh sorry, not a cure.

At the same time it would be interesting to know how many had these ideas before SSRI. If you've been feeling suicidal for a long time, and SSRIs haven't improved your condition, you might feel hopeless enough to commit the act - without SSRIs being any kind of cause.

The fact of having a prescription for psychotropic drugs is a knowable thing, albeit jealously guarded by HIPAA and lawsuit-averse manufacturers. Perhaps it is less knowable whether the victims were taking those drugs and had "therapeutic" levels in their systems at time of death.

But it would be extremely unknowable whether or not they had experienced suicidal ideations, and how long they had had them. Because the most you will get is a note in the medical chart, based on a verbal claim by an unreliable narrator.

Then it seems like a pretty terrible idea to just assume they increase the likelihood of social ideation more than they prevent, no?

Tell that to the pharmaceutical manufacturers who ran clinical trials, received guidance from the FDA, and have been required to black-box-label the drugs according to the suicidal ideations which were scientifically proven to increase due to taking the drugs. I don't know where you got "assume"?

Reply on news.ycombinator.com