Long-term Consequences of ADHD Medication Use for Children’s Outcomes

One of the most publicly debated new medical technologies is pharmacological treatment of children diagnosed with Attention-Deficit/Hyperactivity-Disorder (henceforth ADHD). There are concerns, even among some professionals in the field, that children are over-diagnosed with ADHD; that we know too little about effects of ADHD medication and that children are, in fact, over-treated with ADHD medication. This is the first paper to investigate longer-run effects of pharmacological treatment of ADHD on a series of key human capital variables. More generally, our paper speaks to the literature on the impacts of early health interventions.

Our main identification strategy exploits variation in access to pharmacological treatment generated by hospital variation in specialist physicians’ propensities to prescribe to estimate effects of treatment; see for example Duggan (2005) who uses the same type of variation to investigate effects of second-generation antipsychotics on spending on other types of medical care, Doyle (2007, 2008) who uses variation in investigator assignment to estimate causal effects of foster care, and Maestas, Mullen, and Strand (forthcoming) who rely on examiner variation to look at the effects of disability insurance on labor market attachment. Such variation may stem from hospital level differences in treatment culture and knowledge spillovers; see Coleman, Katz and Menzel (1957) and Soumerai et al. (1998). Our analysis of health outcomes also allows for an individual level panel data strategy.

We use a combination of Danish registers that apart from rich socio-economic background variables include the following key information: 1) psychiatric history and diagnoses for children and their parents including information about the hospital where a diagnosis was first established, 2) history of prescription drug usage for children and their parents, and 3) measures of health, crime, and for a limited set of cohorts also school performance.

We find that the behavior of specialist physicians varies considerably across hospitals and that the prescribing behavior does affect the probability that a given child is treated. Results show that children diagnosed with ADHD on the margin of receiving pharmacological treatment have fewer hospital contacts if treated and are less likely to be charged with crime.

In the analysis of health outcomes, we have sufficient data to meaningfully distinguish between birth cohorts. We document that effects are smaller in later cohorts where more children are diagnosed and treated pharmacologically before the age of ten. There are still significant gains from treatment in the later cohorts, but the results support a hypothesis of diminishing returns to broadening the group of treated.