In February 2005, the FDA added its strongest warning, a so-called black box, on the use of all antidepressants in children and teens to draw attention to the possible risks of these medications. In May 2007, it extended the warnings to young adults aged 18 to 24.
While they saw steady increases in the rate of depression diagnoses in each group between 1999 through 2004, the rate of diagnoses in children and teens fell sharply starting in 2004.
Because the team looked at health insurance claims data, they could not address why doctors diagnosed fewer cases of depression. "We don't know why. What we can say is that the warnings definitely changed behavior," Libby said in a telephone interview.
So the warning labels in 2005 caused fewer diagnoses in 2004?
Yeah right.
Correlation does not imply causation.
It's more likely that the warning labels and decreased diagnoses were both independently caused by a third factor.
It may have been new scientific studies which resulted in both.
Maybe it was a change in insurance company policies.
edit to add: Maybe it had something to do with political events which started in 2004 and culminated in election 2008 which gave people hope and reduced the amount of depression.