Is my child hyperactive or is it simply that he is too small?

Pedro Rey

Parents are increasingly worried about their children being been diagnosed with attention deficit hyperactivity syndrome (ADHD). In fact, in the United States, the proportion of children between two and five years who were diagnosed with this syndrome increased by more than 50% from 2007 to 2012. It is estimated that in 2016 the worldwide prevalence of ADHD diagnosis in children under 18 years old was 5% and in the United States, for example, 5.2% of children of these ages take medication. Spain is one of the main consuming countries of methylphenidate, one of the medicines used to treat ADHD, whose consumption has multiplied by more than 30 in the last 15 years.

Although these figures could raise alarms about the existence of an epidemic, ADHD has also been one of the most debated examples when talking about overdiagnosis and over-medicalization of health, since the prevalence  of ADHD diagnosis variaes a lot geographically and over time, raising questions about whether there are consistent biological markers and whether its organic origin is grounded. Besides, the diagnostic criteria have fluctuated over time and no consensus has been reached on whether their diagnosis is related to relevant long-term behavioural variables such as social, academic or family dysfunction. Even a very successful documentary, Take Your Pills, which can be found on Netflix, deals with drug abuse to treat ADHD in undiagnosed populations.

Being an economist writing for a clinical management blog, and beyond my concern as a parent of school-age girls, my purpose in these lines isn’t to give an opinion on the prevalence and possible over-diagnosis of this syndrome, but to comment on a study recently published in the NEJM, Attention Deficit. Hyperactivity Disorder and Month of School Enrolment by Layton, T. et al., 2018, which uses an ingenious methodology to enrich the debate.

The study is based on the fact that we expect children of the same school year to behave similarly, so the anomalous behaviour related to children of the same group tends to stand out and lead teachers and psychologists of the school to do jump the alarms about the possible existence of a hyperactivity syndrome in certain students. However, these behavioural differences could also be because in the same school year there are children who can have an age difference of up to twelve months, so the differences in behaviour could be due to variations in the development of compatible children with similar age. Until recently, most of the studies that attempted to separate these two hypotheses used anecdotal evidence and data from questionnaires that could be affected by different biases. The article by Layton et al. stands out for using prescription and diagnostic data from medical insurers, which increases its reliability, and for exploiting what we call a "regression of discontinuity", which takes advantage of the differences in the legislation of the states regarding the age that children must be upon joining a school course. In this way, while in some states it is compulsory to have turned five on September 1 to join the nursery, in others it is not. Thus, we can compare the prevalence of diagnosis in children born in August and September in states where this legislation is in force and in states where it is not. The idea of the study is that there should be no discernible differences in the development of children born a few days apart (from August to September), but who, however, may be in a course with children up to one year older than them, what that could encourage overdiagnosis because the child behaved in a "different" way.

The authors find precisely this result. In a sample of more than 400,000 children from all states of the United States, the prevalence of diagnosis in states in which the mandatory age of five years to join the nursery is in force is 85.1 per 10,000 children among children born in August, while only 63.6 per 10,000 children among those born in September, a difference of 21.5 per 10,000 children. On the contrary, this difference is only 8.9 per 10,000 children in states where the legislation isn’t in vigour and, therefore, children born in August and September can be in the same course. Similar results are found on the prescription of ADHD treatment among children born in August (52.9 per 10,000) and September (40.4 per 10,000) in states where the legislation is applied, while no differences are found in states without it. The authors don’t appreciate any difference in these two variables when they compare children born in two contiguous months other than August and September. Moreover, and this is key, in the states whose legislation require that they have turned five on September 1 to join the nursery, they don’t find differences in the proportion of those diagnosed with other diseases, such as asthma, diabetes or obesity, which are not considered to have a direct connection with age.

These results remind me of other rich databases and exogenous sources of variability, such as differences in legislation caused by reasons unrelated with the phenomenon being studied, that serve to inform medical debates away from conspiracy theories and more focused on the possible mechanisms behind a medical phenomenon with important health (diagnostic), economic (prescription) and social consequences (stigma associated with psychiatric diseases)