While there have been substantial increases in available research in pediatric pharmacology over the past 10–15 years, the majority of research in this age group occurs after approval of a medication for an adult indication. Initially, information on treatment indications and dosing from adults is extrapolated to children, and then it is hoped that at a later date, specific efficacy and safety studies are conducted in children and adolescents. Decisions about the need for studies in pediatrics frequently depend on whether the disorder for which the treatment is approved actually occurs in the pediatric age group or is unique to the adult population.

Emphasis on the longitudinal nature of psychiatric disorders became more evident with the advent of DSM-IV, in which continuity across the lifespan is more apparent (e.g., overanxious disorder of childhood was replaced by either generalized anxiety disorder or social anxiety disorder). Similarly, in depression, the criteria for a major depressive episode were modified to include irritable mood in children. The concept of utilizing adult criteria also affected medication development, with studies of treatment for obsessive-compulsive disorder (OCD), major depressive disorder, generalized anxiety disorder, and bipolar disorder in children. Essentially, in most cases, adult criteria were simply extrapolated to children.

In psychiatry, commonly seen disorders such as depression and anxiety frequently begin in childhood or adolescence in a substantial proportion of individuals, so clinical trials are generally first conducted in adults and then some time later (often 5–10 years) are completed in the pediatric age group. This has also led to some interesting anomalies. For example, Food and Drug Administration approval of antidepressants for treatment of major depressive disorder in children and adolescents generally requires two randomized controlled trials because at some level childhood depression is considered noncontinuous. However, use of antidepressants for OCD requires only one bridging study in children, as the phenomenology and pathophysiology of OCD are considered identical across the lifespan.

In recent years, there has also been increased interest in adult outcomes of disorders that begin only in childhood, e.g., attention deficit hyperactivity disorder (ADHD) and autism spectrum disorders. The increased awareness of adult outcomes of pediatric disorders raises similar issues with regard to the complexity of extrapolating from children to adults. For example, a relatively small number of children diagnosed with the combined type of ADHD meet the full criteria for ADHD in adulthood, but approximately two-thirds have continued residual symptoms (1). Also, it is likely that these symptoms have different effects in a child's school setting and in a work or home setting in adulthood.

In this issue, Hollander et al. address similar issues in a report on a randomized controlled trial comparing fluoxetine and placebo for treating repetitive behaviors and global severity in adults with autism spectrum disorders (2). In this study, fluoxetine demonstrated a significant decrease in repetitive behaviors as measured by the Yale-Brown Obsessive Compulsive Scale, with a predicted endpoint difference of 3.7 points. It is interesting that the dose of fluoxetine was relatively high (mean endpoint dose of 64.8 mg/day) but was well tolerated even at these higher doses. Previously, Hollander and colleagues reported on a small randomized crossover study of fluoxetine in children with autism spectrum disorders (3). In the pediatric study, fluoxetine, which was titrated slowly at low doses (mean end dose of 9.9 mg/day), had a modest effect on reducing repetitive behaviors (based on the compulsion scale of the Children's Yale-Brown Obsessive Compulsive Scale), with few side effects reported at those doses.

The findings from these studies of fluoxetine in these different age groups are similar to the results of earlier trials. In studies by McDougle et al. using fluvoxamine, 53% of the adults (eight of 15) responded to fluvoxamine (4), compared to 6% of children (one of 18) (5). Again, adults had a proportionally higher dose of fluvoxamine, and in this case, there were substantially more adverse events in the children. In a large National Institute of Mental Health multisite trial of citalopram (up to 20 mg/day) in youth with autism spectrum disorders, there was no demonstrated superiority of efficacy over placebo for repetitive behaviors, although there was a higher rate of adverse events, including activation, than was seen in the adult studies (6). Overall, it appears that higher doses of selective serotonin reuptake inhibitors (SSRIs) (similar to doses used in OCD) are effective in adults and that these higher doses are well tolerated. Alternatively, in children with autism spectrum disorders, efficacy of SSRIs is limited and even moderate doses of SSRIs lead to higher rates of adverse events than are seen in other studies of SSRIs in children.

Interpreting differences between studies of children and adults goes beyond pharmacokinetic and pharmacodynamic differences. There are major, but often subtle, differences in diagnosis and assessment of outcomes that need to be understood. This is particularly exemplified in studies of autism spectrum disorders, where study methods appear similar. While the studies by Hollander et al. in both age groups used the Autism Diagnostic Interview–Revised and the Autism Diagnostic Observation Schedule, the former requires a parent interview, which was not available for all adult subjects. Also, some adults historically meet the full criteria but have amelioration of symptoms across time and therefore may not meet the full criteria at the time of the evaluation. Additionally, in these studies, essentially the same outcome measure was used across different ages (the compulsion scale of the Yale-Brown Obsessive Compulsive Scale). This has the advantage of being a clinical interview and not a parent or self-report, though the availability and type of other informants are likely to be different across the age range (i.e., more reliance on patient interview in adults and more reliance on parent interview in children). Another issue that is likely common to both groups is difficulty with recruitment. The small study groups make it difficult to identify subgroups of individuals in such heterogeneous samples. This is further compounded by the differences in age ranges in pediatric versus adult studies. The adult study had an age range of 18–60 years (mean age=34.3 years, SD=14.3), whereas the pediatric fluoxetine study had an age range of 5–17 years (mean age=8.2 years, SD=3.0).

Finally, the large disparity in dosing across the age span is difficult to interpret. Differential tolerance of medication, as already mentioned, means there is little ability to compare efficacy across age groups. In the two fluoxetine studies of autism spectrum disorders, the dose for adults was six times as high as the dose for children, with a body mass only two to three times as high.

In summary, it is evident that in disorders starting in childhood, it is not possible to assume that what works or does not work in children will have the same efficacy and adverse effect profile in adults. Similarly, extrapolating from adults to children without doing the pediatric studies does a substantial disservice to the children in need of treatment. However, having studies across the lifespan informs our understanding of the developmental nature of psychiatric disorders and provides a richer picture for all ages. Information on children informs our treatment of adults, and treatment studies in adults inform our treatment of children. Such studies also raise important questions about definitions of continuity of illness across the lifespan, which has already been addressed in other medical conditions, such as juvenile rheumatoid arthritis and juvenile diabetes. Are juvenile-onset illnesses just more severe, or are they different disorders? While there are few clear answers, there are a lot of compelling questions that need to be considered to improve our future research strategies.