This is another prime example of statisticians throwing the baby out with the bath water. When one looks at the raw percentages, the suicide rate is clinically insignificant, and the number needed to treat is quite large. Again, this cost/benefit ratio is somewhat ignored at the expense of committees that feel compelled to "create" policies.
If your sample size is large enough, statistically significant results are bound to show. All this did, unfortunately, was again tie psychiatrists' hands and cause them a great deal of nervousness when prescribing anything other than fluoxetine to children...even when it clearly would not be the first drug of choice in a given case scenario.
There are neurochemical changes demonstrated in the brain even when given placebo. Brains in general are very plastic. The confounding variables in these SSRI-children studies are many. Thankfully, psychiatrists are continuing to do what they've previously done - take each case on an individual basis and weigh the pros and cons of various treatment options with ongoing evaluation to ensure the best treatment outcome.