Big news this past week on medication and psychotherapy. In its on-line, advance publication mode, the New England Journal of Medicine published a multi-center study of cognitive behavior therapy (CBT) and Zoloft, alone and in combination, for anxiety disorders in children and adolescents. Administered singly, each intervention worked. Given together, the antidepressant and brief (14 one-hour sessions) therapy brought very substantial improvement to 80 per cent of patients over an interval of 12 weeks.

The last major study to show improvement at these rates was a trial of CBT and Serzone for chronic depression, published in 2000, also in the New England Journal. That research has been considered suspect for a number of reasons, including the drug company ties of some of its authors and the remarkably high response rate (73 per cent) in the combined group. But the criteria for improvement in the 2000 study were weaker than those in the current one - and the response rate in the new study is higher. The new study does include authors who have done work for drug companies. But it is NIMH-sponsored, registered in advance, and large enough, with 488 subjects (ages seven to 17), that its results would never have been relegated to a file drawer. The current study suggests that psychiatry's treatments may be quite powerful - certainly much more helpful than recent press has led the public to believe.

As for the details: Readers of this blog know that the skeptical approach to outcome studies demands "intention to treat" analyses that include reporting of effect sizes and the "number needed to treat," or NNT. "Intention to treat" means that you report the response rate relative to all subjects who are accepted into a given arm of the study, including those who then drop out. Effect size indicates the power of the intervention relative to the intractability of the disorder. NNT shows how many patients must be exposed to the therapy in order to produce recovery in one patient who would not have gotten better on his or her own or in response to a placebo.

In the current study of anxious children, the NNT was 1.7 for the combination of drug and talk therapy, with an effect size of .86, a large effect. These are favorable numbers for any treatment, the numbers you would expect when 80 per cent of treated children do very well, as opposed to 23.7% on placebo. If I'm reading the effect size right, most children on combined treatment did better than any children on placebo, and the average child in combined treatment did better than 80 per cent of children on placebo.

The drug alone and therapy alone were superior to placebo. Slightly more children responded to psychotherapy (59.7%) than to Zoloft (54.9%), but the responses to medication were stronger; neither of these differences was statistically significant. The NNT for psychotherapy was 2.8 with an effect size of only .31 (low-to-medium); the comparable numbers for Zoloft were 3.2 and .45 (medium). In their summary, the researchers put the NNT for either intervention alone at "three."

As for negative effects, there were no suicide attempts. Five children in CBT alone, five on combination therapy, and one on placebo (from a smaller group of subjects) developed suicidal ideation - but none on Zoloft alone. These differences were not statistically significant, but obviously the trend was for medication to protect against suicidal thought, not to arouse it. Four children spread over the groups engaged in non-suicidal self-injurious behavior, and two children on Zoloft developed homicidal ideation, a disturbing development. Regardless of the intervention, it seems that anxious children are at some risk for violent or injurious thoughts and behavior at a rate on the order of three per cent. But (in contrast to what studies of depressed children have sometimes shown) in this trial medication did not play a special role in suicidality. This data also suggest that, as was the investigators' intent, the study had recruited children with reasonably severe mental disorders.

What were those disorders? Interestingly, when they tried to to aggregate kids with moderate to severe anxiety disorders, the researchers came up with a group in which over 80 per cent of subjects had social phobia as part of the picture. Almost all of these children also had other forms of anxiety, such as separation anxiety or generalized anxiety; many had other diagnoses, such as dysthymia and attention deficits. A requirement of entry into the study was that these collateral diagnoses be milder than the anxiety.

Many articles in the popular press argue that social phobia is a weak concept, on the level of shyness, rather than a psychiatric condition. But the children in this study suffered impairment at a level that indicates current suffering and predicts mood and anxiety disorders later in life. So: social phobia can be a substantial disorder, and over a period of weeks, a combination of medication and psychotherapy is the preferred treatment for it. The same conclusion seems to hold across the anxiety disorders. Past studies mainly involved milder levels of anxiety; this trial shows that the treatments work in children with substantial disability.

Diagnosis matters. In studies of depression in children, CBT has looked somewhat less effective than medication alone, although combination treatment (CBT and antidepressant) was a safer and yet more effective than medication alone.

Across the common mental illnesses in children and adolescents then, the research suggests treatment with a combination of medication and psychotherapy. In the era of parity, strong evidence now justifies reimbursement for both modalities as elements in the treatment of mental illness in childhood. Actually, the conclusions might be broader. It is impossible to test every treatment for every indication in every age group. The targeted, careful studies should have broad impact; the current paper should stand as general support for a broad utility for antidepressants and talk therapy in combination.

Those are the bold-face take-home messages. But questions remain: How solid is the evidence? Do we really want children on antidepressants? If so, at what doses? Is CBT more effective than other psychotherapies? In subsequent postings, I will break out the caveats.

Most Recent Posts from In Practice

A Tribute to Tom Magliozzi

An appreciation of Click and Clack

Bell of Freedom

Pete Seeger

Kenneth C. Edelin

Remembering a Women's Rights Pioneer