Clinicians treating childhood anxiety have had little evidence on which treatment to start with, or what to do when three months of it has not worked. This pragmatic 24-week sequential multiple assignment randomised trial enrolled 316 youths aged 8-17 with DSM-5 anxiety disorders, notable for high sociodemographic disadvantage and frequent co-occurring diagnoses. They were randomised to fluoxetine or exposure-based CBT for 12 weeks, then non-remitters were randomised again to continue or to add the other modality.
Youth-reported anxiety fell 31.7% over 24 weeks. Initial treatment made no significant difference, with a non-significant advantage for CBT (difference in mean change 1.45, 95% CI -2.25 to 5.16). In week-12 non-remitters, combination therapy was no better than continuing monotherapy (-2.74, 95% CI -6.53 to 1.05). The authors' hypothesis — fluoxetine first, then combination — was not supported.
One subgroup finding deserves care rather than action: non-Hispanic White youths did better starting and continuing fluoxetine, while minority youths did better switching to combination therapy. That is a single subgroup analysis in a trial not powered for it, and should prompt further study rather than differential prescribing. The main message is liberating: since sequences performed similarly, treatment can genuinely be chosen on family preference, availability and cost.
- Fluoxetine and CBT performed equivalently as initial treatment
- Combination therapy no better than continuing monotherapy in non-remitters
- Overall anxiety fell 31.7% over 24 weeks whichever path was taken
- Choose by preference, access and cost — the evidence does not rank them
The statistics, in plain English
Both key intervals cross zero, so no difference was demonstrated, and both are narrow enough relative to the 31.7% overall improvement that a large advantage for any sequence is unlikely to be hiding in them. This is a genuinely informative null result rather than an underpowered one. The racial subgroup finding is the opposite: it comes from splitting an already modest sample, and such analyses fail to replicate far more often than not.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free