The problem isn't the statistics, and it isn't the solution either. Its the diagnosis data.

How do you quantify if someone is depressed? How depressed are they? How do you compare two people depression? It is an incredibly nebulous disease.