Pediatric anxiety treatment has long been guided by a simple assumption: combine therapy with medication and you get the best result. A large trial published in the American Journal of Psychiatry in September 2026 tested that assumption in the places children are actually treated — community clinics and primary care offices — and found something more complicated. No single approach came out clearly ahead.
The finding matters because it shifts the question parents and clinicians face. Instead of asking which treatment is best, the more useful question may be which treatment is realistic to start, and what to do twelve weeks later if it is only partly working.
What the pediatric anxiety treatment trial actually did
The study was a sequential multiple assignment randomized trial, or SMART — a design built specifically to test treatment sequences rather than single treatments. Led by researchers at Children’s Hospital Los Angeles, it enrolled 316 young people with severe anxiety and ran for 24 weeks across two stages.
In the first 12 weeks, participants were randomly assigned to either fluoxetine (an SSRI) or cognitive behavioural therapy. At the 12-week mark, they were randomised a second time: either continue what they had been doing, or add the other treatment to create a combination.
That second randomisation is the clever part. It mirrors the decision point clinicians genuinely face — not “what should we start with” but “this is helping somewhat, now what?”
Across the full 24 weeks, youth-reported scores on the SCARED anxiety measure fell by 31.7%. Anxiety improved substantially. But when researchers compared the pathways against each other, no route was the clear winner.
Why this sample is different from earlier research
Most of what we know about treating childhood anxiety comes from trials run at academic medical centres with relatively advantaged, relatively uncomplicated participants. This trial deliberately did the opposite, and the contrast with prior research is stark:
- Ethnicity: 64% of participants were Hispanic or Latino, compared with roughly 12% in earlier landmark trials.
- Economic circumstances: 56% were covered by Medicaid, against about 25% classified as low socioeconomic status previously.
- Co-occurring depression: 67% of this sample, versus 0% in the comparison trials, which typically excluded it.
- Co-occurring ADHD: 46%, compared with 12%.
- Setting: community clinics and primary care, not university research centres.
Earlier trials excluded children with depression alongside anxiety. Real clinics cannot. By enrolling the more complicated, more representative population, this trial answers a question the tidier studies could not: does any of this work outside ideal conditions? The answer appears to be yes — with the caveat that nothing works dramatically better than anything else.
What “no clear winner” does and does not mean
It is easy to misread a null result as evidence that treatment is pointless. That is not what happened here. Anxiety scores dropped by nearly a third overall. Every pathway led somewhere better than where families started.
What the trial undercuts is the idea of a single correct sequence that should be applied to every child. The researchers noted that individual characteristics and the order of treatments both influenced how young people responded — just not in a way that produced one universally superior route.
For families, that is arguably good news. If you cannot access CBT quickly because of waitlists, starting with medication is a defensible choice rather than a compromise. If you would prefer to avoid medication initially, beginning with therapy is equally defensible. The trial gives clinicians permission to work with what is actually available.
The limitations worth holding onto
This was a single-blind trial, meaning participants knew which treatment they were receiving. In studies of anxiety, where expectation shapes reported symptoms, that matters. Much of the outcome data was also self-reported by the young people themselves — appropriate for a subjective condition, but not the same as an objective measure.
Twenty-four weeks is a reasonable window but not a long one. Anxiety disorders in young people frequently recur, and this trial cannot tell us which sequence holds up after a year or five. The study also tested one specific SSRI, fluoxetine, and one structured form of therapy; other medications and other therapeutic approaches were not compared.
Finally, “no statistically significant difference between groups” is not proof that the groups are identical. It can also mean the trial lacked the statistical power to detect a modest difference. With 316 participants split across multiple pathways, some cells were relatively small.
What this means for parents navigating treatment
The most practical takeaway is that the first decision carries less weight than it feels like it does. Parents often agonise over the starting point as though a wrong choice forecloses recovery. This trial suggests the starting point is far less decisive than what happens at the review.
A few things follow from that. Build in a genuine checkpoint at around twelve weeks and treat it as a real decision rather than a formality. Track symptoms in some consistent way, because “does this seem better?” is hard to answer from memory. And if a child also has depression or ADHD alongside anxiety — as most children in this trial did — know that this is the norm, not a complication that puts them outside the evidence.
If your child is struggling with anxiety, this research is a reason for cautious optimism about several routes, not a substitute for assessment. Treatment decisions about children belong with a qualified clinician who knows the specific child. Our guides to high-functioning anxiety and dealing with anxiety attacks may help you describe what you are seeing more precisely when you get there.
Frequently asked questions
Does this mean therapy and medication are equally good for childhood anxiety?
Over 24 weeks in this trial, no pathway was clearly superior. That is not quite the same as proving equivalence — the study may not have been powered to detect smaller differences — but it does mean there is no strong evidence for insisting on one particular starting point.
Should we always add the second treatment at twelve weeks?
The trial did not find that combination therapy reliably beat continuing the original treatment. Adding the second option is one reasonable choice among several, and the decision should rest on how much the child has improved so far.
What is a SMART trial and why does it matter here?
A sequential multiple assignment randomized trial randomises participants more than once, at different decision points. It is designed to evaluate treatment strategies over time rather than one-off interventions, which is much closer to how chronic conditions are actually managed.
Does having ADHD or depression change what should be tried first?
Most participants in this trial had at least one co-occurring condition, so the results already reflect that reality. The researchers observed that youth characteristics influenced response, but the study did not produce a simple rule for matching profiles to treatments.
How long should we wait before concluding something is not working?
This trial used twelve weeks before reassessing, which is consistent with typical clinical practice for both SSRIs and structured therapy. A clinician may reasonably adjust sooner if a child is deteriorating.
The bigger shift
Research that produces no winner is unsatisfying to read and genuinely useful in clinic. By running the study in ordinary community settings with the complicated, co-occurring, under-resourced population that most trials screen out, this team produced something rarer than a novel treatment: evidence that the available options work reasonably well for the children who actually need them. For more on how treatment research is evolving, see our coverage of new depression research in 2026.
The full trial is published in the American Journal of Psychiatry, with a summary from Children’s Hospital Los Angeles.