Given method differences, sample sizes and the question of which trials are even published, it is difficult to draw conclusions about activation/mania by drug, age or diagnosis. There is some suggestion that young people become activated more frequently and while taking certain medications.
We examined drug-induced behavioral disinhibition (DIBD) in a well-characterized sample of 267 children hospitalized for psychiatric reasons (Carlson and Mick, 2003). We operationalized DIBD as a dramatic increase in irritability identified by increased children's "time-outs" for impulsive/aggressive behavior while on medication. The children's reactions were carefully measured and recorded weekly during unmedicated and medicated weeks by two shifts of nurses. Age, gender, diagnosis and medication were covariates. Using this definition, we found only 20 (7.5%) children with DIBD. Severe ADHD and pervasive developmental disorder (but not mania or depression) appeared to increase the risk for DIBD, while older age and stimulant use decreased the likelihood. A trend suggested an increase in DIBD with SSRIs. However, it was often difficult to distinguish true DIBD from the behavioral fluctuations of these disturbed children. Fifteen percent of children subsequently improved on the same regimen; 40% improved when the offending drug was stopped and another treatment started; and the remainder had adverse responses to many medications.
The most robust data on the vulnerability of young people to antidepressant-induced switching come from a large pharmacoepidemiologic study of 87,920 mental health users ages 5 to 29. Martin et al. (2004) defined a "manic" conversion by the presence of a mood stabilizer prescription three or more months after antidepressant prescription. (The three-month time period was chosen to eliminate activation that subsides on drug discontinuation.) A new diagnosis of BD was also present in the claims. An overall rate of 5.4% of this phenomenon was found. However, prepubertal children were found to be at highest risk for this occurrence with a "number needed to harm" rate of 10 (95% CI=9-12). In 15- to 19-year-olds, this rate dropped to 16; in 20- to 24-year-olds, to 31; and in 25- to 29-year-olds, to 29.
The question here is not so much whether the event took place, but whether children prescribed antidepressants at a younger age have a greater vulnerability to behavioral toxicity, whether this vulnerability will persist as they develop, and whether it means they even have BD. In nonresearch settings, the diagnostic accuracy of a BD diagnosis is poor (e.g., Pogge et al., 2001). As noted in the inpatient study described, a variety of disorders were found to be present when drug-induced disinhibition occurred.
Activation-like events should be defined, anchored and obtained systematically in future drug trials and subsequently in postmarketing surveillance efforts. In clinical settings, baseline measures of aggression, agitation, irritability and mood should be obtained before starting treatment. Ideally, these should be obtained prospectively for several weeks in order to determine whether an event is part of the natural history of the condition or due to the medication.
Activation events may occur more frequently in children and adolescents, especially those with developmental disabilities, than they do in adults. It is premature to conclude what the clinical or pharmacological significance of these events is because the data are so limited. Switching children from the frying pan of antidepressants to the fire of mood stabilizers and atypical antipsychotics on the basis of behavioral toxicity alone seems ill-advised.
How specifically to handle behavioral toxicity will depend on what the symptoms are, how severe they are and the effectiveness of the medication. Choices obviously range from stopping the drug, lowering the dose and/or changing the dosing schedule to adding another medication that mitigates the unwanted symptoms.