In the late 1990s and continuing on into the past decade, bipolar disorder started being diagnosed more and more in children. This became a problem only because the criteria for bipolar disorder in children have never been firmly established. Researchers developed their own set of criteria which contradicted the official diagnostic criteria for the disorder. The research criteria basically did away with the need for a manic or hypomanic episode, and instead replaced it with irritability and anger.
Coincidentally, a few pharmaceutical companies also released a set of medications — called atypical antipsychotics — which can be used to treat certain symptoms of bipolar disorder.
Doctors started diagnosing bipolar disorder in children with a looser set of criteria, and felt more at ease prescribing a treatment for it because these new medications had become available.
This set of circumstances led to a reported 40-fold increase in the past decade of bipolar disorder diagnoses in children. This suggests a pretty obvious problem in the diagnostic criteria, since nothing has changed so much in the past decade to offer a reasonable explanation for this sort of increase.