- Design
- systematic review and random-effects meta-analysis with multivariate meta-regression
- Population
- 175 treatment arms across 163 studies of adults with PTSD
- Primary outcome
- PTSD symptom reduction (Hedges' g)
- Effect
- pooled g = 1.67 (95% CI 1.56-1.79); 1.55 after excluding 10 influential outliers
Emergency and critical care clinicians meet post-traumatic stress disorder from both sides: in patients after major trauma and intensive care, and in colleagues. A meta-analysis pooled 175 treatment arms across 163 studies comparing prolonged exposure and cognitive processing therapy. The overall pooled effect was large - Hedges' g 1.67 (95% CI 1.56 to 1.79) - with substantial symptom improvement across the board.
On the comparison between the two, the analysis is honest about its own instability. Across the full sample, multivariate meta-regression found no significant main effects or interactions. Excluding ten influential outlying studies lowered the pooled effect to 1.55 and produced a signal that prolonged exposure outperformed cognitive processing therapy in non-military samples, but the treatment-by-sample interactions were not significant in that trimmed model either. A finding that appears only after removing outliers, and then fails its own interaction test, is a hypothesis.
The defensible conclusion is the useful one. Both treatments produce large reductions in symptoms, and there is no dependable basis for choosing between them on patient characteristics. So refer to whichever is actually available and properly delivered, rather than waiting for the theoretically preferable one - which, given typical waiting times, is the decision that will determine the outcome.
- Refer to whichever trauma-focused therapy is genuinely accessible rather than holding out for a specific modality.
- Both are evidence-based first-line options; neither is a compromise.
- Do not use sex, military status or trauma type to select between them - the moderator findings did not survive testing.
- Consider structured follow-up after critical illness and major trauma, where PTSD is common and rarely asked about.
- The same applies to staff: access to treatment matters more than which of the two protocols is offered.
Why it matters
It removes the case for delaying a referral while waiting for a particular therapy, which is where most of the harm in PTSD care happens.
Don't overread it
The apparent superiority of prolonged exposure in non-military samples emerged only after excluding outliers and failed its own interaction test - treat it as a hypothesis, not a basis for selection.
The statistics, in plain English
A Hedges' g of 1.67 is a large effect - roughly, the average treated patient does better than about 95% of untreated controls. But the pooled figure across 163 studies conceals wide variation, and the fact that removing ten studies moved it to 1.55 and changed which treatment appeared superior shows how much a few influential studies drive these comparisons. Effects this large in psychotherapy meta-analyses also partly reflect within-group pre-post comparisons, which include natural recovery.
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 emergency & critical care, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free