Science1 publisher2 min readPublished
A 54% low-birth-weight gap in Danish ADHD and autism diagnoses narrowed to 17% over ten years
ISGlobal and Aarhus researchers tracked 71,000 Danish ADHD and autism diagnoses from 2012 to 2022 and found the low-birth-weight excess fell from 54% to 17%. The team says it cannot weigh how much of that narrowing comes from wider case-finding.
The Scientist · Science desk
What happened
- Researchers at ISGlobal and Aarhus University compared more than 71,000 Danish young people diagnosed with ADHD or autism between 2012 and 2022 against more than 713,000 undiagnosed peers.
- Those diagnosed were still more likely than their peers to carry the characteristics long associated with the conditions, but the size of those differences declined steadily across the decade.
- Children born at low birth weight were 54% more likely to receive one of the diagnoses at the start of the period and 17% more likely by the end.
- The authors name rising awareness, more effective case-finding in health and education systems, wider access to diagnostic evaluation, and a moving impairment threshold as possible contributors.
- They say rising diagnosis rates should not automatically be read as evidence that the underlying occurrence of ADHD or autism rose by the same amount.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- constraint The registry records diagnoses, not symptoms, so whether more Danish children actually have these traits stays an open question.
- decision Services and screening pathways that treat low birth weight or family income as flags for neurodevelopmental follow-up are working with markers that now separate the groups much less sharply; a 17% excess is thin ground for a referral rule.
- precedent Aetiological studies that pool diagnoses made across 2012 to 2022 will see attenuated associations for reasons unrelated to the exposure being tested. Diagnosis year is a variable such studies have to stratify on.
- contradiction The same result is the strongest evidence yet that the diagnosed population widened and, in the senior author's own words, not evidence of overdiagnosis, so both sides of the public argument can cite it.
Matched registry studies depend on the comparison group. Here it is about ten undiagnosed children for every diagnosed one [16], sampled from a source population of more than 2.1 million Danish children and adolescents [3]. That leaves roughly 784,000 people in the analysis, about 37% of the source cohort [19]. What gets estimated, year by year, is the association between things recorded before diagnosis and the diagnosis itself: gestational age, birth weight, parental education, household income, parental psychiatric history, and healthcare use in the run-up [5]. Preterm birth and several measures of socioeconomic disadvantage moved the same way as birth weight [8].
The direction of this finding is on firmer ground than its size. These are relative comparisons, and a risk ratio is one plus the absolute excess divided by the baseline rate [18]. Diagnosis rates climbed steeply over the same decade [15]. If the baseline rises and the absolute gap between low and normal birth weight children stays put, the ratio falls toward 1 without any change in who is being diagnosed [18]. The ISGlobal summary reports only the ratios.
Tarp, a PhD student at Aarhus University and the paper's first author, set the boundary himself. "The key message is not that these risk factors are no longer important," he said [9]. "What we found is that people diagnosed in recent years resemble the general population more closely than those who received the same diagnoses a decade ago," he said [10].
Reading the surge as a widening of thresholds takes one item off the authors' list of candidates and promotes it. The study cannot say how much any single factor contributed [14]. Plana-Ripoll, the senior author, said "Our findings help us better understand why ADHD and autism diagnoses have increased so markedly in recent years" [11], and he said the results do not show that these conditions are being overdiagnosed [12].
Separating a moving threshold from wider access would need symptom severity at the point of diagnosis, measured the same way in 2012 and in 2022, and the summary has none of it. The composition of the diagnosed group changed, and I would treat the magnitude as provisional until absolute rates are published: the excess associated with low birth weight fell by about 69% over the decade, from 54 points to 17 [17].
What to watch
- Whether the full JAMA Psychiatry paper reports absolute diagnosis rates and risk differences by year. Those figures would show how much of the narrowing is a rising baseline rather than a changed population.
- Whether the same attenuation appears in other national registries, where access to diagnostic assessment expanded on different timetables.
- Whether follow-up work can attach severity at diagnosis to registry records, the measure that would separate a moving threshold from wider access.