In a new study, researchers found SSRIs increased the risk of suicide attempts threefold for those younger than 18 and up to twofold for those aged 18-24. They had no preventative effect at any age, even for those at high risk of suicide.
“The present study finds similar results to prior observational research—that is, consistent evidence of an increased risk of suicidality during treatment with SSRIs in children and adolescents,” the researchers write.
The study was led by Tyra Lagerberg at the Karolinska Institutet and was published in Neuropsychopharmacology.
The data came from a Swedish registry, including 162,267 people who received a diagnosis of depression. Of these, 52,917 people began treatment with an SSRI antidepressant within 28 days, while 109,350 did not. The outcome of interest was suicide attempts as recorded in the registry (thus, attempts serious enough to result in an encounter with the medical profession). The researchers analyzed data at a three-month follow-up and at one year (the results were similar at both time points).
The researchers included both the intention-to-treat analysis and the per-protocol analysis, two different ways of handling the outcome data. The gold standard is to include both so that the reader can see all the data.
In this case, both the intention-to-treat analysis and the per-protocol analysis showed that SSRI use was associated with increased suicide attempts in those under 25. The only difference was the magnitude of the effect.
In the intention-to-treat analysis, 6-17-year-olds were 2.9 times more likely to attempt suicide on SSRIs. In the per-protocol analysis, they were 3.34 times more likely.
Those 18-24 were 1.59 times more likely to attempt suicide in the intention-to-treat analysis and 2.01 times as likely in the per-protocol analysis.
Those over 25, however, were not more likely to attempt suicide if they received an antidepressant, based on either analysis.
However, this null finding also means that antidepressants did not have a preventative effect—they did not reduce the likelihood of suicide for those over 25, either.
The main limitation of this study is that the researchers couldn’t account for baseline severity. That is, people who had more severe depression—and thus, increased suicide risk—may have been more likely to take an antidepressant. This is especially true since the researchers note that psychotherapy is the first-line treatment for mild-to-moderate depression in Sweden.
However, the researchers accounted for suicide risk in another way: they were able to account for those who had a history of suicidal behavior (which is a good proxy for those at the highest risk of future suicide attempts).
Those who had previous attempts were more likely to attempt again—whether or not they took antidepressants. That is, antidepressants seemed to neither worsen the problem nor help at all.
“Those with a history of suicidal behavior (N = 4221) showed greater absolute risks among both initiators and non-initiators. We found no evidence of a difference between initiators and non-initiators in this group in terms of suicidal behavior risk,” the researchers write.
Thus, probably the best way to interpret the results of the study is as follows:
For those younger than 25, with no history of suicidal behavior, taking an antidepressant makes you up to three times more likely to attempt suicide, with that likelihood decreasing as you age. For those 25 or older, taking an antidepressant doesn’t help reduce suicide. Similarly, for those at high risk for suicide, taking an antidepressant doesn’t reduce the risk.
The finding here is consistent with previous studies, which have repeatedly shown that antidepressants increase suicide risk, particularly for children and adolescents, with some studies finding more than doubling the risk of suicide and at least one analysis finding a sixfold increase.
Studies have also found that antidepressant drugs worsen outcomes in the long term, even after controlling for the baseline level of depression severity.
And one study found that those with more severe depression, those with comorbid anxiety, and those who were suicidal were the least likely to benefit from antidepressant drugs.
****
Lagerberg, T., Matthews, A. A., Zhu, N., Fazel, S., Carrero, J. J., & Chang, Z. (2023). Effect of selective serotonin reuptake inhibitor treatment following diagnosis of depression on suicidal behaviour risk: A target trial emulation. Neuropsychopharmacology. Published online July 28, 2023. https://doi.org/10.1038/s41386-023-01676-3 (Link)
[Duplicate Comment]
Report comment
Could it be possible the non difference in risk with previous attempts actually exists, but that difference is statistically not significant? Given perhaps the sparcity of data points?
Could it be, given this study reports lower rates than other studies in the under 25, under 18 and over 25, like in some of those over 4 times increased risk, is due because actually the risk of self harm when using SSRIs is ACTUALLY HIGHER in people WITHOUT depression, whatever that is, than in people WITH depression, whatever that is?
Suggested to me because people with previous attempts, yeah, maybe caused by previous medication, even tardive, do point to something that could be depression, and maybe non depressed people who take SSRIs do have a higher risk than the patients with depression. Which one could asume is higher than the rest of the population.
And “samples” of people with previous attemps actually are enriched, have a higher frequency of people, with this still hypothetical thing called depression. Just so happens SSRIs add no or little benefit to self harm in them. But do increase the risk of self harm in people without depression.
And that could explain why some relatively skeptic clinicians still advocate for SSRIs in some patients. Some, very few, do have real depression, do look better and do feel better. Just no or very small benefit regarding self harm, and it does put people without real depression at way higher risk of self harm. And probably without benefit in them, but with worsening and longer lasting now medication induced depression. Similar to lithium, no suicide risk mitigation, anecdotal benefit in some with real, whatever that is, bipolarity. And worsening outcomes in those without real bipolarity.
Tutuous and confusing to my mind, but could explain at least the clinical, “anecdotal” observations of benefit, and the depression diagnosis and self harm pandemic, not just epidemic, following just the newspapers’ reports…
Like overdiagnoses and overtreatments do actually make more difficult, even impossible to pinpoint depression and bipolarity as actual real diseases, perhaps hypothetically absent self harm attempts and previous medication. Asuming of course they exist as such. Me being hypothetical, not advocating either way, diagnosis or treatment wise…
Report comment
Tutuous=Tortuous
Report comment
I hate to admit and say it, but I feel compelled to sumarize by MY madness: Maybe there is a biological (uugh!) depression, just so happens most, your garden variety depression diagnosis, is not biologically based, reflects maybe no vulnerabilty AND is more of a social, cultural, situational and transitory thing.
Just, as I remember reading in the 90s, is rare, like 1 in a 1000 rare. Lifetime, yearly, or in the clinical office I don’t remember. Maybe I am missremebering the number.
And, uggh again, maybe there is biologically based bipolarity, just so happens is RARER than biological depression. And again, the diagnosis of bipolarity, now, in practice, is more of a social, cultural and situational thing. Whether it’s transitory or not, the garden variety one, I don’t know. The biological one, uggh, might be/is cyclical, and therefore recurrent.
So at those or similar numbers: how large would have to be the sample size of patients with “biological” depression, that you have a P < 0.05 to find the "biological" reason, not only the mechanism of this "biological" depression with an actual biological set of tests?
Obvioulsy, I am assuming that history already proved that if biological depression is real, the clinical method won't enrich a population enough to even with large correlation genetics find its cause? Too much damn noise…
I wrote it, I loath done that, but I have to try to be honest…
And mad as hater…
Report comment
The key question would be whether one could distinguish in some objective way between “biological” and “situational depression.” Of course, no one has come close to doing that, and they don’t even bother trying any longer. It’s just “let’s try these drugs out and see what happens.” No pretension in reality of any kind of scientific diagnostic process. But as long as the big money keeps coming in, no one has an incentive to look deeper.
Report comment
The method IS the problem, not the sample size. No method no larger sample size will do, I guess zero times infinity is stil zero…
Report comment
Quite so. When the experimenters have no objective means to separate those “in” the group from those “not in” the group, the experiment is DOA. Which applies to pretty much all “mental health” studies, except perhaps those measuring adverse effects of the “treatments!”
Report comment
Hence the “need” for new, more medication, now “this” new disease is very obvious, AT, after treatment, and different from the garden variety…
But of course! related to the old one, somehow!?, I mean, beyond the treatment.
Like that kind of “clinical” situation might be numbing on reason, no offensive intention, just contortive.
Report comment
Even more, there is no way to distinguish between ‘endogenous depression’ as a kind of ‘mental illness;’ learned helplessness from developmental effects, which would be environmental causes but during critical periods in childhood development; purely situational depression as in no history of serious childhood adversity but some severe environmental stressors account for the depression alone; or depression as a symptom of some other physical illness like hypothyroidism. Or, what is probably more often the case, depression as a result of all the above, except the first one, which I doubt is even a coherent entity, since really it reduces to a special case of the last one.
Report comment
Or the old: “just happens”. Normality in many, a lot? or most? psychiatric symptoms has, as far as I know, not being studied in “normal”, i.e. not diseased. population. Like: How many people feel hopelessness once a week, twice, etc, at this age, time of year, etc. Even that “experiment” requires knowing which variables are relevant to design it.
Let alone, in the past year, two years, etc.
Let alone how many have hopesslness with some other symptoms, etc.
Let alone if the are actually independent, not just not correlated. If there is something that makes their together appearance more often that one would expect by chance, causally, not just in the numbers.
Just a bunch of wizards agreeing that A+B+C in my expert opinion is depression and that’s it…
Even the methodology of “knowing” someone “has” hopelessness is doubtful. Asking just questions that fill a questionaire “proves” this person is feeling hopeless? ieven if it were, is it at ALL related do ANYTHING clincal?
Seems to me not hopeless but arbitrary, capricious or irrational, etc. 🙂
Not denying it feels bad, never had it, but I can imagine.
Report comment
A least unpleasant thought: What would be the NEED to search for biological depression, when the symptoms, the effects, AND the relatively safe and effective treatments, as psychotherapy, familial and social interventions actually could work both for the biological depression AND the garden variety socially constructed depression?. Diagnosis or otherwise…
“missa thinks looketh too much like liveth in this society”…
So, asked like that, biological depression could be a mirage, an illusion, by reductio ad absurdum, by irrelevancy when it comes to managing it, and maybe, therefore, diagnosing it at least specifically.
Ironically, coming kind of full circle: given the symptoms the needs point to the non pharmacological treatment, they are basic human needs and wants. Just for the symptoms for EVERYONE, it sounds after all as human suffering in simpler or simplest form, somehow.
And sadly if biological depression were real and rare, even if not orphan disease very rare, comercially unviable for specifically targeted pharmaceutical research even if chronic.
But on happier thoughts, addreassable with non pharmacological interventions, just for the needs, wants and symptoms, for everyone. A fairer, kinder, more humane society for EVERYONE. More on some, maybe less on others. Marxian phrasing comes to my mind.
And sadly most of the self harm not caused by medication would be, per hypothesis not caused by biological depression, even if in some rare or unfrequent people it could be biological. I’ll leave it at that…
Report comment
Yay, Peter – this the-drugs-cause-the-suicides point cannot be emphasized enough.
Report comment
Anyone using the 2023 results to argue risk will be confronted with their opposite conclusion the year before. How can we reconcile the new evidence with the conclusions of Lagerberg et al., 2022, “Selective serotonin reuptake inhibitors and suicidal behaviour: a population-based cohort study”, Neuropsychopharmacology 47:817–823, which concludes “The results do not suggest that SSRI-treatment increases the risk for suicidal behaviour in either youths or adults; rather, it may reduce the risk.” In 2022, they said “clinicians should be reassured that SSRI initiation is not associated with an increased risk of suicidal behaviour”. In 2023, they say, based on that different study, “Our finding of an increased suicidal behaviour risk among individuals under age 25 reflects evidence from RCTs.” In the 2023 paper, they cite their 2022 paper [22], and seem to explain the discrepancy based on methods: “A recent study on the impact of SSRI initiation in a Swedish register setting, utilising a within-individual design, found an elevated risk of suicidal behaviour in the first year of SSRI treatment as compared to the month a year prior to SSRI initiation across age groups, but a reduced risk when comparing the month immediately after to immediately before initiation [22]. While that paper accounted for all time-invariant confounding within-individuals, it could neither control for time-varying confounding by the course of the disorder indicating an individual for treatment, nor for the impact of contact with the healthcare service, which receipt of an SSRI is a proxy for. By comparison, while the current study is subject to between-individual confounding, it minimises other common sources of bias, such as reverse causation bias [14] by emulating a target trial.” Just saying the new study is better will not be convincing to opponents.
Report comment
I will suggest here that per the scientific method, we assume something is NOT effective until it is proven to be effective. The “null hypothesis” in this case is that antidepressants do not reduce suicidal thoughts or behavior. If such controversy and conflicting studies continue to exist 30-40 years after ADs were introduced, I think we can conclude at this point that they DO NOT reduce suicidal thoughts and behavior.
Of course, in the “antidepressants cause suicidal thoughts and behaviors,” the “null hypothesis is that they don’t. However, the fact there is a Black Box warning on all drugs impacting serotonin suggests there is enough evidence to concern people about their safety, especially in the young. Since there is almost zero evidence of any AD effectiveness in youth, and literally none in young children, even a chance of increasing suicide rates ought to be enough to eliminate these completely from the psychiatrist’s toolbox, at least for children.
At a minimum, the evidence to date is more than sufficient to conclude that antidepressants do not decrease the suicide rate for any subgroup of identified “depressed” people. The fact that certain “researchers” and “thought leaders” choose to cherrypick the studies they like should not divert us from this conclusion. Any decent review of the literature shows no positive effect, and many show possible increases in suicide rates, even in studies involving people who were screened for suicidality before starting. It’s not really that much in dispute, except for people who want to believe otherwise.
Report comment
What about actual completed suicide? was there any difference? I take it not statistically significant if any.
Report comment