Thread by @timothyjlayton

And now for your evening commmute 🚨NEW PAPER ALERT 🚨

New work with @AnupamBJena, @ml_barnett, and Tanner Hicks in @NEJM on age at school entry and diagnosis and treatment of ADHD out this evening.

Link to article: nejm.org/doi/full/10.10…

Initiate thread.

Headline finding: Children born in August are 32% more likely to be diagnosed with ADHD than children born in September in states that have a 9/1 Kindergarten age cutoff.

Children born in August are the youngest in their class, while children born in September are the oldest, suggesting that ADHD diagnosis is strongly influenced by a child's age relative to his peers.

We find no difference in diagnosis or treatment rates between children born in any other consecutive pairs of months (Jan vs. Feb, Feb vs. Mar, etc.). We also document that the clinical conditions of the parents of August vs. September-born children are identical.

Additionally, there is no August vs. September difference in ADHD diagnosis or treatment in states that do not have a 9/1 cutoff for Kindergarten entry. There is also no difference in diagnoses for non-behavioral conditions like asthma and obesity.

Finally, the August vs. September difference is not present at age 4 prior to children entering school, but is strongly present at age 7, after children have been in school for a few years.

We also find that among treated children, August-born children are treated *more* intensively (measured by days supply of ADHD drgus) than September-born children.

This rules out possibility that the additional August ADHD diagnoses were only slightly treated and suggests marginal kids' drug treatment was long term.

We interpret the results as suggestive evidence that teachers and parents experience cognitive biases when assessing the likelihood of ADHD. Instead of considering how a child's behavior compares to normal behavior for his precise age (in months)...

the teacher compares the child's behavior to the behavior of his peer group, *even if most of the children in the peer group are (as much as a year) older than the child*.

The big important caveat here is that we do not know if the August-born kids are over-diagnosed or the September-born kids are underdiagnosed (or both).

We also can't say whether the additional August-born children diagnosed with ADHD benefited from the diagnosis (via additional attention) or treatment

What we can say is that our results provide clear evidence ADHD diagnosis AND treatment seems highly subjective and context-specific. This seems troubling given that the long-term consequences of ADHD drug treatment seem somewhat uncertain.

At a minimum, it seems like doctors should pause when considering assigning an ADHD diagnosis to a child with a summer birthday and take the child's age relative to his peers into account when making the diagnosis.

More broadly, our results add to a growing body of evidence that children who are young for their grade face an uphill battle in school (see additional recent evidence here from Florida here: nber.org/papers/w23660).

Most interesting thing about this paper (to academics): About 2 weeks before submitting, we found out that there are two (!) JHE papers from 2010 showing exactly what we show.

But our results were just so clean and perfect, we couldn't bear throwing in the towel. So, we submitted to NEJM anyway, and it was accepted! So all you researchers out there who do really great work that you're proud of and then find out it's already been done, there's hope!