Does Cognitive Behavioral Therapy Work or Not? The Plot Thickens
Over the past several months, we have reviewed several articles that have endorsed cognitive behavioral therapy for the treatment of depression. In some cases, CBT has appeared even more effective than antidepressants, at least for the long-term prevention of relapse. But now, along comes a paper that appears to show that CBT isn’t particularly effective after all. Well, at least this is how the media has been reporting it. The actual results are not cut and dried.
A team of researchers identified all controlled clinical trials of CBT for schizophrenia, depression, and bipolar disorder that had been published over the past 20 years, likely including all relevant studies ever done on these topics. For their meta-analysis, they included trials that compared CBT to some sort of nonspecific psychotherapy (e.g., supportive therapy, psychoeducation, recreational therapy), treatment as usual (TAU), or a placebo pill. They excluded trials that compared CBT to a wait list control because they were examining the effectiveness of CBT compared to other treatments as opposed to CBT compared to no treatment. What was “TAU” in these studies? TAU was the treatment that patients would typically get in the community if they were not in any study, including medication, miscellaneous forms of psychotherapy, and case management. Across nine trials examining psychosis (primarily schizophrenia) the researchers found that CBT was no more effective than nonspecific psychotherapies. While there seemed to be a small advantage for CBT in reducing positive symptoms, this advantage evaporated in studies where the clinical raters were unaware of whether patients were assigned to either CBT or nonspecific therapy. In 10 studies examining depression, CBT was more effective than nonspecific therapies and pill placebo by a statistically significant but clinically questionable margin (the effect size was 0.28). In the long term, CBT resulted in less depressive relapse than TAU. In four trials of bipolar disorder, CBT was not more effective than TAU for preventing a relapse of either mania or depression (Lynch D et al., Psychol Med; online ahead of print).
TCPR’s Take: Over the years, CBT has developed the reputation of being the most evidence-based psychotherapy, with the implication that it must be the “best” therapy out there, especially for major depression and the anxiety disorders. But this analysis shows that CBT beat non-specific depression therapies by only a small margin, although it has a more robust advantage for preventing depressive relapse in the long term. With regard to bipolar disorder and schizophrenia, it now seems crystal clear that CBT provides no benefit beyond treatment as usual. In a future issue of TCPR, we will look at whether CBT truly beats other forms of psychotherapy for panic disorder and obsessive compulsive disorder.
Study Links Ritalin to Sudden Death in Children; FDA Disagrees
Do stimulants cause serious cardiac problems in children? It is a controversial topic—and not just among psychiatrists. Cardiologists and pediatricians have done battle over this issue, and their respective medical societies have opposite recommendations on whether to get EKGs in children before treatment. The cardiologists say “yes,” while the pediatricians say, “no, not really necessary.”
In the past, the FDA has looked into this issue and decided not to issue a black box warning, but they recommended that manufacturers of stimulant medications add label warnings regarding the risk of sudden death in patients with known cardiac abnormalities. Now a new study complicates the debate. Funded by the FDA and the NIMH, epidemiologists searched state vital statistics for all cases of unexplained sudden death in children occurring in the U.S. from 1985-1996. They found a total of 564 cases of sudden death over that period. They then looked at how many of these children had been taking stimulants at their time of death. Using a variety of information sources, such as toxicology reports and parental reports, they determined that a total of 10 of these children—1.8% of the total—were taking stimulants when they died. In all 10 cases, the identified stimulant was Ritalin (methylphenidate). (They did not limit their analysis to Ritalin—this was simply the most commonly prescribed stimulant during the period they studied.)
The next step of the study was the crucial one, because the 1.8% figure means nothing by itself. It could mean that these 10 children happened to be taking Ritalin when they died, and that the cause of death had nothing at all to do with the medication. In order to shed light on this figure, the researchers found a comparison group of children who had also died suddenly, but for whom the cause of death was known. They chose as their control group 564 matched cases of children who died in motor vehicle accidents. They chose this comparison group because this was a close match for the children who died from unknown reasons. They figured that since both unexplained sudden deaths and deaths from car accidents involved parents or guardians experiencing a sudden and traumatic loss of a child, the accuracy of parental recall about whether the children were taking stimulants would be about the same. We’ll come back to this.
The results? Out of 564 children who died of MVA, only two children were identified as having used stimulants (0.4%), fewer than the 10 children stimulant-taking children who had died of unknown causes. They concluded that stimulants led to a six- to seven-fold increased risk of sudden death.
Sounds bad. So why did the FDA immediately come out with a statement urging “caution” in interpreting these results, and why did Dr. Robert Temple, the director of medical policy for the FDA’s Center for Drug Evaluation and Research, discount the study by saying, “I don’t think it makes the case that there is real risk here.”
Various reasons. One is the very small number of events. After all, the researchers were basing their conclusions on a total of 12 children who had died over an 11 year period in the United States. Many millions of children were prescribed stimulants over that time, so to base conclusions of risk on only 12 cases is a stretch. In addition, Temple and others believe that the problem of recall bias is significant. After all, if your child died suddenly of unexplained causes, wouldn’t you spend quite a bit of time and energy trying to figure out the cause of death? That might lead you to blame a medication, even if your child was no longer taking it. (Gould MS et al. Am J Psychiatry 2009; online ahead of print).
TCPR’s Take: We agree with the FDA that the small number of events and possible recall bias in this study are serious and limit the confidence we can have in the results. Having said that, we do know that stimulant medications are linked to increased blood pressure and heart rate in children, so it is biologically plausible that such medications could lead to increased risk of sudden death. We recommend asking patients about a history of cardiac problems before starting stimulants, and following up with an EKG if indicated. But screening every potential stimulant-taker with an EKG could end up leading to multiple false positive findings, which would in turn lead to unnecessary and expensive cardiac work-ups. While one might argue that even a single prevented death would make mass screening worthwhile, given the methodological problems, these data do not prove that screening would provide any benefit.
Citalopram Not Effective for Autism in Children
In the November 2008 TCPR, we wrote about various topics in autism diagnosis and treatment.
Autism is one of five conditions classified by DSM-IV-TR as a “pervasive developmental disorder.” The condition can occur in children or adults, and it is diagnosed by a characteristic triad of symptoms: 1. Social interaction problems (e.g., poor eye contact, aloofness); 2. Impaired communication (e.g., delayed speech development, odd speech patterns, echolalia, inability to sustain a normal conversation); 3. Rigid or repetitive patterns of behaviors and interests (rocking, hand flapping, lining toys up, preoccupation with one particular interest). It is a frustrating condition to treat. Currently, the only FDA-approved medication is risperidone, and that is effective only for the agitation that accompanies autism. The SSRI fluvoxamine (Luvox) has been shown effective for reducing repetitive behaviors in adults, but thus far studies of SSRIs in autistic children have been mixed (McDougle CJ et al., Arch Gen Psychiatry 1996;53:1001-1008). This new study is the largest trial of an SSRI in autistic children to date, and the news is bad—citalopram (Celexa) doesn’t work.
In this NIH-funded trial, 149 children and adolescents with variants of autism were randomly assigned to either citalopram (average dose, 16.5 mg) or placebo. After 12 weeks of treatment, citalopram was no better than placebo on the Clinical Global Impressions scale or on the Yale Brown Obsessive Compulsive Scale modified for autism. Citalopram caused more adverse events than placebo, especially on measures of activation, including increased motor activity, impulsiveness, and insomnia (King BH et al., Arch Gen Psychiatry 2009;66:583-590).
TCPR’s Take: An accompanying editorial by Dr. Fred Volkmar of the Yale Child Study Center concluded that the results of this study are definitive: SSRIs do not work in children with autism. We have to agree. Intensive behavioral treatment is probably a more effective option, but it is time consuming and expensive.