Skip to content
The Neurodiversity Directory

“High
adherence”
to ADHD
medication
in Sweden —
after 62% of
treatment starts
were excluded

A Takeda-funded study of Sweden’s prescription register reports high adherence to ADHD stimulants. The figure describes treatment episodes that lasted at least 180 days — and 62.3% of the treatment starts recorded from 2005 to 2021 were excluded before it was calculated.

A nationwide register, a headline of 78.2%, and the threshold the headline does not mention

A study published online in the Journal of Attention Disorders on 21 January 2026 set out to measure how consistently people in Sweden take the stimulant medication prescribed for ADHD. MaiBritt Giacobini and colleagues used the Swedish Prescribed Drug Register, which collects prescription data from all pharmaceutical retailers, whether the treatment was prescribed in the public or private sector. Between 2015 and 2020 they counted 158,413 treatment episodes in 128,366 patients, on methylphenidate, lisdexamfetamine, dexamfetamine or amphetamine. The paper puts estimated ADHD prevalence in Sweden in 2021 at 3.2% of children and adolescents and 1.4% of adults; the Directory’s neurodiversity statistics page gathers figures from other sources.

The abstract is clear: “Overall treatment adherence was high (78.2%)”. Adherence here means the proportion of days covered, worked out from how many pills were collected and how often, assuming one pill a day of the main stimulant. The 78.2% is the average across episodes.

An average isn’t a threshold, though. The paper itself describes 80% days covered as “a commonly used threshold for defining good adherence” (p. 3), and its Table 2 shows that 88,560 of the 158,413 episodes reached it: 55.9%. The body puts it plainly: “more than half of the patients were adherent to ADHD medications at least 80% of the time” (p. 6). So “more than half” and “high” are both in the paper, and only one of them made it into the abstract’s opening finding.

Figure 1: the 397,445 treatment starts that never reached the denominator

The more important number is in the flow chart on page 4. The register held 8,064,047 stimulant dispensations from 1 July 2005 to 31 October 2021. From those the authors built 637,971 treatment initiations. The first exclusion removed 397,445 of them, all for one stated reason: “Episode duration less than 180 days”. That left 240,526 episodes. A further 82,113 were dropped for overlap and study period: 46,995 had ended before 1 January 2015, 20,390 fell under 180 days once overlap or the 2015 start was applied, 14,639 overlapped completely or began after 1 January 2021, and 89 lacked a birthdate. The remaining 158,413 make up the final sample.

From treatment starts to the final sampleTreatment episodes at each step of the study’s Figure 1, Swedish Prescribed Drug Register, July 2005 to October 2021
StepEpisodes
Treatment initiations637,971
Excluded: episode duration less than 180 days397,445
Episodes of 180 days or longer240,526
Excluded: overlap and study period82,113
Final episodes, 2015–2020158,413

Source: Giacobini et al. (2026), Figure 1. The 62.3% share is the Directory’s own calculation.

On the Directory’s own arithmetic, not the paper’s, dividing 397,445 by 637,971 gives 62.3% of treatment starts excluded at the first step because the episode lasted less than 180 days. The final sample is 24.8% of all initiations.

Three qualifications go with it. The initiations span July 2005 to October 2021, so 62.3% is not a rate for 2015–2020. They are treatment starts, not people: an episode is tied to one drug, so someone who switched or restarted could appear more than once. And the bar was higher than 180 days: the methods treat the first 90 days as dose titration and count the 180-day minimum after it. Someone who collected prescriptions for eight months and then stopped was outside the study.

The authors saw it, and called it persistence

None of this is hidden: “The reason for the apparently high rate of adherence shown by our study could be due to the stringent inclusion criteria used, which selected only those patients who had received more than 180 days of treatment” (p. 6). The limitations section adds that the exclusion “may have removed a large proportion of patients with low medication adherence” (p. 8).

The authors also argue the other side, and it deserves full strength. Adherence is whether you take the medication while you’re on it; persistence is whether you stay on it. The authors say the design ends up “distinguishing treatment persistence from treatment adherence” (p. 6), and that the excluded cases “would arguably represent low persistence rather than low adherence” (p. 8). As a methodological distinction, that holds.

The trouble is the summary. Inside the paper, “high” carries its condition; in the abstract and the conclusion, the condition is gone. Adherence was high among people who stayed on treatment for at least 180 days after titration, and staying on treatment means repeatedly collecting prescriptions, which is what the measure counts. That’s very nearly a definition. The selection rule had largely decided the verdict before any days covered were counted.

What the 397,445 shorter episodes were is a question the paper doesn’t answer, and it would be easy to supply an answer it doesn’t support. The paper does list possibilities — a change in circumstances, adverse effects, contraindications, an active decision, a forgotten renewal — but as reasons for treatment gaps its data cannot tell apart, not as an account of who stopped. It also cites a global study (Brikell et al., 2024) which, as the paper reports it, found fewer than a quarter of patients still on treatment after five years. The Directory has reported a separate review which found that stimulant benefits for real-world functioning were smaller and more variable than the benefits for attention. That review is a reason to ask the question. It doesn’t answer it.

Inside the selected group: adolescents, the very young and amphetamine

Within its own terms, the paper does find variation, and the adolescent dip is real. Across the study years, the share of patients at or above 80% days covered was:

  • 77.8% to 88.3% at ages 0 to 5
  • 60.9% to 63.5% at 6 to 11
  • 49.7% to 52.4% at 12 to 17
  • 58.0% to 58.9% at 18 to 24
  • 69.3% to 71.8% at 25 and over

Even among adolescents still collecting after the minimum period, only about half reached the threshold. The authors say this “warrants further study”.

The highest figures, in the youngest group, rest on at most 489 patients — 0.4% of the cohort by age at study entry. That’s a group for whom, the paper notes, pharmacotherapy “is not indicated” under the age of six, though it adds that treatment may start sooner for children with greater functional impairment (p. 2).

Amphetamine stands apart, with average adherence of 7.2% across 1,227 episodes, and 95.1% of those episodes fell in the lowest band. The paper explains that amphetamine is not commercially available in Sweden and needs an import licence before a psychiatrist can prescribe it. It offers that as the likely reason prescriptions were few. It gives no separate explanation for the adherence figure. Adherence did not differ by sex and was stable over time, including in sensitivity analyses allowing for summer breaks and longer supplies.

What a pharmacy register can see, and who funded the study

About this source

The paper’s conflict-of-interest statement reads: “This study was funded by Takeda Pharma AB, Sweden, who were involved in the study design, collection, analysis, interpretation of data, and drafting of the manuscript.” Two authors are Takeda Pharma employees, and one was employed by Shire Sweden (a Takeda company) at the time of the research. Two were at Parexel, which received funding from Takeda to conduct the study. The original research was funded by Shire Development LLC, now part of Takeda; the current analysis, and the medical writing support, were funded by Takeda Pharmaceuticals International AG.

That is reported as the paper reports it, and it doesn’t make the figures wrong: the Figure 1 flow chart is there because the authors put it there. The question the study answers is how consistently people already on treatment keep it going.

That question has a narrow reach. In the UK, the route to a prescription starts with an assessment; the Directory’s guide to getting an ADHD diagnosis sets it out, and its assessment listings show who provides one. Beyond the prescription, the Directory has covered a Norwegian registry study in which medication made no measurable difference to school grades for children at the margin of treatment, and a review flagging the cardiovascular risk of cumulative exposure. Support that isn’t a prescription is in the counselling and therapy listings.

The Neurodiversity Book (chapter XXI) describes the capacity a dose restores as “imported, metered, and renewed at a pharmacy counter”. That renewal is exactly what a prescribing register records. It can show that renewals continued for the people who kept collecting. It can’t say anything about the treatment starts that ended early. In this study those were 62.3% of all initiations, and the abstract’s headline doesn’t mention them.

Citations

Giacobini, M., Zhao, J., Freilich, J. et al. (2026) — Medication Adherence in Children and Adults Receiving Treatment for Attention Deficit Hyperactivity Disorder (ADHD) in Sweden: A Nationwide Study — Journal of Attention Disorders

News and updates.

Straight to your inbox.