Sommers-Flanagan and Sommers-Flanagan published a review of the research on antidepressants and child in 1996. They examined all the studies evaluating antidepressant medication for child and teen clients they could find published between 1985 and 1994.

They make several important notes about understanding the research:

First, some people respond to a placebo effect. They take "sugar pills" which have no clinical effect and still get better because they expect that they will. It is common to use a "wash out period." the researchers give the client placebos for the first week or two, and if they get better they are removed from the study. The rationale is that they were not seriously depressed, and so they should be removed from the study. Second, some people drop out of the study. They find the side effects of the medication are too upsetting or difficult to deal with. Typically, researchers omit these people from their final analysis because they can not evaluate how they would have responded to the medication had they stayed on it. Third, sometimes research is conducted by comparing children and teens on antidepressants to children and teens on placebos or "sugar pills." This is done only after informing the clients at the start of the study that they may not be receiving "real" medicine, but they should stay in the study. After several weeks of analysis, the researchers provide real medication to the children and teens who received sugar pills for free. This way, they are not deprived of treatment that could be effective and they receive it for free to compensate for any additional suffering that participating in the research caused. Conducting the research in this way allows for analysis of the effects of the drug with a comparison group of similarly depressed teens and children that aren't taking the drug but are seeing doctors. Often, the people who watch the children and teens know if they are on medications or on sugar pills.

the authors correctly note that these practices, while "purifying the research," could mislead people as to the effectiveness of the drug.

First, some studies removed 15% to 30% of their subjects due to placebo effects. Translated to real life, this could mean that many of the children and teens referred for medications might not really need them even if they look depressed. They may feel depressed in response to life events like changes in schools or homes, or such factors as divorce. Their depression might have improved in a few weeks without giving medication.

Second, side effects of antidepressant drugs include rash, nausea, vomiting, decreased energy, sleep disturbances, irritability, and sometimes decreases in social inhibitions and increases in acting out behaviors. They note one study where 50% of children and teens on antidepressants began showing at least two of these symptoms, and over all the studies they reviewed, 20%-30% of children and teens showed sufficiently severe responses that the doctors discontinued medication altogether, and they dropped out of the study. Translated to real life, counting these teens and children in the success and failure rates of the drug may give more realistic estimates of what is likely to happen if you take your child in for a medication evaluation. If they can not tolerate the drug in their systems, then the drug offers them no benefits.

Third, knowing that one child is on medication and another is not can lead people to assume that the child on medication is getting better while the one not on medication is getting worse. This judgment can effect one's observations in subtle but significant ways. A better research design includes conducting a "double blind" study in which neither the client nor the reviewer know if the child is receiving real medication or sugar pills. Thus, they argue that studies where observers of children and teens were not "objective" or "blind" to the nature of the pills given to the children (real antidepressant or sugar pill) are poorer quality, and should be given less weight.

The authors note that all of the studies supporting the use of antidepressants in children and teens show this last weakness. In other words, reviewers who knew which children were on real medication rated them as improved.

None of the studies that were conducted "blind" support the effectiveness of antidepressants with children and teens. In other words, reviewers who did not know which children were receiving real medications and which were receiving sugar pills could not tell the difference between them.

Further, even in studies that concluded that medication was helpful, if you recalculate their success percentages to include clients that got better on sugar pills and that dropped out due to side effects of the drugs, the percentages of improvement were sometimes cut by a third or more.

they conclude that before giving children or teens antidepressants, several conditions should be met:

1) the child or teen is seriously depressed in the absence of any life stressors, like family conflict or divorce 2) the depressive symptoms are strongly related to physical problems such as disturbed sleep, and significant loss of weight and appetite 3) 10 to 15 sessions of individual or family therapy have been tried and produced no success 4) the client prefers medications over other interventions

If you are interested, read Sommers-Flanagan and Sommers-Flanagan original article in the journal Professional Psychology: Research and Practice, Volume 27, Issue 2, pages 145-153.