Psychosis History and ADHD Medication: A Swedish Register Study

The relationship between ADHD medication and psychosis is one of the places where worry is talked about most. The question is simple: if someone who has had an episode of psychosis in the past starts a stimulant medication, will the psychosis come back? A new study from Sweden looks at this question with direct data.

All I have is the abstract of the study. Everything written below rests on that abstract; I do not know what is in the full text.

Who the study is from and where

The study was carried out by Patrick Bach, Johan Franck, Jonas Hällgren, Härje Widing, Mika Gissler and Jeanette Westman. It was published in the journal Translational Psychiatry in March 2026 (volume 16, issue 1). By the abstract's own wording, its type is a cohort study: it draws on Sweden's population-based registers, and these registers contain the data of all residents of the country.

The participants are adults aged 18-64. They have two things in common: they have a history of psychosis, and between 2008 and 2021 they started a stimulant (psychostimulant) or atomoxetine. 3,770 people were found who met these criteria.

This post is a Turkish-language summary of the paper published open access in Translational Psychiatry. The paper was published under the Creative Commons Attribution 4.0 (CC BY 4.0) license. The findings belong to the authors of the paper; the account and appraisal here are mine.

What the question was

The first sentence of the abstract names the worry directly: there are concerns that stimulant medications may increase the risk of psychosis, especially in people who already have a history of psychosis. The study asks this: in an adult with a history of psychosis, does the risk of a psychotic event rise after medication is started?

I have to say one detail up front here. The abstract does not state which diagnosis the participants were taking medication for. So I cannot say that all of them have ADHD. Modafinil is also counted among the medication groups. The study looks at adults with a history of psychosis who started medication, not specifically at people with ADHD. For this reason, when applying the finding to people with ADHD, this difference needs to be kept in mind.

How it was done: the same person, two periods

The design of the study is, to my mind, its most interesting side. Instead of comparing people who took the medication with people who did not, the authors compared each person with themselves. In the abstract's term, it is a "within-individual" design: the rate of psychotic events was compared over the 6 months before the medication was started and the 6 months after it was started.

Why does this make sense? Because the person's unchanging characteristics (genetics, long-term makeup, life history) stay the same in both periods. When you compare a person with themselves, the objection that "the people who take the medication are different people to begin with" largely disappears. This is an approach often used in register studies.

A rate ratio (RR) was used as the outcome measure. The rate ratio divides the event rate in the two periods. A value of 1 means "no difference", above 1 means "more frequent after starting", below 1 means "less frequent". The confidence interval (CI) given alongside shows between which limits the true value most likely lies. If the interval includes 1, the possibility that the measured difference is due to chance cannot be ruled out.

What they found

In the six months after the medication was started, the rate of psychotic events came out at 0.95 compared with the six months before starting, with a confidence interval of 0.84 to 1.08. So the point estimate is slightly below 1, but the interval includes 1. The correct reading is this: no measurable increase in psychotic events was seen after the medication was started.

The upper limit of the confidence interval is 1.08. What this means is that with this data we cannot completely rule out the possibility of a small increase of up to about 8 percent. On the other hand, within the interval there is no room for an increase like 1.5 or 2, that is, up to double. So the data leaves a large increase quite unlikely. A small increase it neither confirms nor definitely dismisses.

The authors also tested the robustness of the result in other ways. They looked separately by sex and by age groups. They separated the medications by class: separate calculations were made for (dex)methylphenidate, (lis)dexamfetamine, atomoxetine and (ar)modafinil. They also looked at eight-week intervals. In addition, they examined separately, as an alternative outcome measure, only the psychotic events linked to substance use. According to the abstract, all these analyses gave consistent results.

The authors' concluding sentence is clear: contrary to the concerns, stimulant or atomoxetine treatment was not found to be associated with an increase in the risk of psychotic events in adults with a history of psychosis. They also say the findings may inform doctors who are considering prescribing a stimulant in this group.

What this finding means for me

If you have a history of a psychosis episode and you are considering medication for ADHD, this study says that an increase feeding the fear was not seen in this data. In a six-month window, in 3,770 people, psychotic events did not increase after the medication. To me, this is data that makes you question a sentence that says "absolutely not".

Atomoxetine is a non-stimulant ADHD medication, while modafinil is a medication used for wakefulness. The study gathers these under the same roof too, and when it looked separately by medication group it found similar results. The abstract does not give how many people used which medication, or the confidence interval of each group.

But this does not mean "it is safe for you". The study describes averages. Things like the individual situation, when and how the psychosis was experienced, and whether another medication is being used at the moment are information that the treating doctor knows; the abstract does not know them. Make the medication decision together with them.

If you are someone who uses medication and has no history of psychosis, this study does not say anything directly about you. In a separate post I summarized an opinion paper that looks at the same subject, the relationship of ADHD with psychotic disorders, from a broad frame. There the authors say that the therapeutic use of methylphenidate does not increase the risk of psychotic disorder, and that uncertainty continues for amphetamine. In this study the (lis)dexamfetamine group was also looked at separately and the result came out consistent. They are two independent papers, but they point in the same direction. I read this not as a conclusion but as an agreement.

What it does not say

The first limit: this is not an experiment. The medication was not assigned at random; the registers were looked at. The within-person design balances fixed characteristics but does not balance things that change over time. For example, a person being started on medication could be related to their being in a calm period of the illness at that time. This is a general limit of such designs; from the abstract I cannot know how the authors discuss it.

The second limit: time. The comparison is six months before and six months after the medication. I cannot know the longer term from this abstract.

The third limit: scope. The study looks at people with a history of psychosis. It does not measure, in people without a history, the link between a first psychosis and the medication. The abstract also says nothing about dose, duration of use, stopping, or accompanying medications. How well the medication works is also not this study's question.

The fourth: the abstract does not state which diagnosis the medication was given for, which I said above.

For a wider reading you can look at the ADHD medication side effects post and the ADHD or bipolar post, because questions about psychosis very often come intertwined with questions about mood.

What I take from it

I read this study as good news, but by setting myself a rule: a single register study is not a decision. The good news is this: this study offers no grounds for telling a person with a history of psychosis "no medication". It is natural for doctors to act with caution, but a flat ban should not take the place of caution.

The unknowns also remain: the long term, dose, accompanying medications, the subgroup with ADHD. These are the questions left unasked. If you are going to talk about this with a doctor, I think the best approach is to take this study along and ask, "There is data like this, what do you think in our situation?" The answer will come from the doctor, not from the abstract.