A study from Michigan State University strongly suggests that up to one million school children have been diagnosed with ADHD simply because they are the youngest and most immature students in their class.

All states have a cutoff birth date for when a child must enroll in kindergarten. In a state with a cutoff date of September 1, all children born between January and September 1, 2000 entered kindergarten in 2005. Children born September 2-December 31 entered kindergarten in 2006, a full year later and older. The children born on September 1 or in August are the youngest students in their kindergarten class, up to eight months younger than their fellow kindergartners.

If a child is behaving poorly, if he's inattentive, if he can't sit still, it may simply be because he's 5 and the other kids are 6.

According to the study, these youngest students are the ones most likely to be diagnosed with ADHD. And this may be simply because the children are acting their age.

Todd Elder, an assistant professor of economics at Michigan State University, looked at a sample of 12,000 school children from the Early Childhood Longitudinal Study Kindergarten Cohort. Elder found that the youngest kindergartners were 60% more likely to be diagnosed with ADHD than the oldest students in their class. And when these students reached the fifth and eighth grade, they were twice as likely to be prescribed stimulants, such as Ritalin, as their older classmates were.

"If a child is behaving poorly, if he's inattentive, if he can't sit still, it may simply be because he's 5 and the other kids are 6," said Elder. "There's a big difference between a 5-year-old and a 6-year-old, and teachers and medical practitioners need to take that into account when evaluating whether children have ADHD."

Teachers do not diagnose ADHD, but their opinions and evaluations may lead to a child being sent to a mental health professional who may make such a diagnosis.

If these youngest children truly are being misdiagnosed, the study suggests they account for about 20% of all children diagnosed with ADHD. According to current estimates, that's 900,000 misdiagnoses.

Several different comparisons, within a state and between states, led Elder to conclude that these diagnoses are misdiagnoses.

In Michigan, where the cutoff date is December 1, kindergartners born on December 1 had much higher rates of ADHD diagnosed than those born on December 2, who entered kindergarten a full year later.

Illinois kindergartners born in August were much more likely to be diagnosed with ADHD than their Michigan counterparts. This is presumably because Illinois has an earlier cutoff date than Michigan, September 1. The August-born students were the youngest in their class in Illinois, but not in Michigan.

Both these comparisons suggest that the problem lies not with the children, but with their age.

There is no medical test for ADHD. Diagnosis is somewhat subjective, relying on evidence of multiple symptoms of inattention or hyperactivity, with these symptoms persisting for at least six months and in at least two settings, by the time a child is seven.

This has led to a great deal of debate, among both the public and health professionals, over whether ADHD is underdiagnosed, overdiagnosed or even a valid medical condition.

One point being made by many is that the long-term effects of stimulants such as Ritalin on children's health are unknown, because the practice of prescribing them to children has begun so recently. In effect, a long-term study on this is now being conducted on all the children who have been prescribed stimulants as treatment for ADHD.

Elder's study only focuses on one small part of this debate. And it can't prove that these 900,000 children have been misdiagnosed, it can only show that younger children are more likely to be diagnosed with ADHD. What it should do is make the people involved in these diagnoses think about whether they're penalizing children for behavior that's well within the spectrum of normal human behavior: acting their age.

An article detailing the study will be published in a future issue of the Journal of Health Economics. A corrected proof of the article was published online by the journal on June 17, 2010.