Clinical Edge

Summaries of Must-Read Clinical Literature, Guidelines, and FDA Actions

Fluoxetine Tied to Lower Obsessive-Compulsive Scores Among Children With ASDs

Key clinical point: Fluoxetine improved obsessive-compulsive or repetitive behaviors in children with autism spectrum disorders in a preliminary study.

Major finding: At 16 weeks, the mean Children’s Yale-Brown Obsessive Compulsive Scale–Modified for Pervasive Developmental Disorders (CYBOS-PDD) score fell to 9.02 points in the fluoxetine group and 10.89 in the placebo group. But the findings fell apart in secondary analyses.

Study Details: The findings come from a multicenter, randomized trial of 146 participants aged 7.5-18 years.

Disclosures: The study was supported by a federal grant from the Australian government. Dr. Reddihough had no financial disclosures.


Reddihough DS et al. JAMA. 2019;322(16):1561-9.


One could certainly take issue with the suggestion that this was presented as a positive trial on two levels: One, the prespecified primary outcome was not met, and two, the clinical significance (as distinct from statistical significance) of a 2-point change on that scale is problematic given the wide range of baseline scores allowed into the study.

The other thing that gets mixed up in this study is: Exactly what are obsessive-compulsive symptoms as distinct from repetitive behaviors? That question is a real challenge in this field when it comes to clinical trials for this target in autism, which tend to lump together heterogeneous repetitive behaviors.

The fact that there was absolutely no signal, or at least not a very strong one, is very challenging, considering how frequently this class of drugs is prescribed in autism.

This is not the first SSRI study for autism that’s come up empty. In this case, though, a negative study is still important because it confirms other negative studies. Another recently published study – the SOFIA fluoxetine study (J Autism Dev Disord. 2019 Jul 2. doi: 10:1007/s10803-019-04120-y) – also came up negative. SOFIA randomized 158 children to 14 weeks of fluoxetine or placebo. There were no significant differences on the primary endpoint, the Children’s Yale-Brown Obsessive-Compulsive Scale, and the placebo response rate was 41%.

However, it was clearly a heroic effort by Dr. Reddihough et al. to get this current study done: It took 7 years to get it over the finish line. This is probably because fluoxetine is so easily available. Why would a parent take a 50% chance of their child not getting a drug that might have some benefit – and that they could get without much trouble? And if it takes 7 years to complete a clinical trial, and we’re sitting around waiting for a definitive one, we are literally looking at potentially decades before we have some real answers that would inform your clinical practice in terms of this commonly prescribed drug.

As far as nailing the coffin shut on fluoxetine, I don’t think that will ever happen because some kids clearly improve. The placebo response in this population is very high. In our citalopram study (JAMA Pediatr. 2013 Nov;167[11]:1045-52), it was close to 33%. The improvement is dramatic and real, no less than any other response. If you see that response as a clinician and parent, it is very difficult to walk away from. Moreover, the population in clinical practice may be different from the population that shows up in a clinical trial specifically focused on restricted, repetitive behaviors.

One reason we may see a response in some is because SSRIs can help with anxiety, which is a common, arguably core symptom of autism. It appears to be part of the reason kids have catastrophic meltdowns when there are any changes in things they have come to expect, like a different route to school or a delay in their favorite TV show coming on. And if anxiety drives that, and an SSRI helps with anxiety, the child might be able to cope with something that would otherwise feel like the end of the world. Maybe that starts a positive feedback loop instead of a negative one, and maybe it propels more changes as the child and family experience success.

So, what are clinicians to do? The answer is still the same – they should use their best judgment about each child’s symptoms and about the risks and benefits that might occur with that individual. The fact that these trials are coming up negative for this indication in autism doesn’t mean that SSRIs might not be helpful for anxiety or depression, just as they are in the general population. I think we are back to basics. Clinicians need to use their best medical judgment according to each child’s unique needs.

These comments were adapted from an interview with Bryan H. King, MD, MBA. Dr. King is a professor of psychiatry at the University of California, San Francisco. He reported receiving personal fees from Genentech. Dr. King also commented on the study in an accompanying editorial (JAMA. 2019;322[16]:1557-8).