Saved in this browser. Sign in to sync across devices.
Key finding: At eight weeks, the RELAX group scored an adjusted average of 1.5 points lower on the GAD-7 anxiety scale than usual-care participants (95% confidence interval −2.4 to −0.6). Meaningful, but not an individual guarantee.
Who took part?
The UK multisite study randomly assigned 268 adults (134 per group; mean age 33.7 years) between 16 and 28 weeks of pregnancy. They reported high repetitive negative thinking—persistent worry and rumination—but had no more than moderate anxiety at enrollment. People with higher anxiety scores, certain current psychiatric diagnoses or specified recent self-harm risks were excluded.
One group received usual maternity care. The other received usual care plus 12 self-directed online sessions, approximately 17 minutes each, over four weeks. RELAX used cognitive bias modification for interpretation: structured practice considering less threatening interpretations of ambiguous situations.
The numbers—and what they mean
MAIN OUTCOME · EIGHT WEEKS
−1.5 GAD-7 points
Adjusted average difference in anxiety scores (95% CI −2.4 to −0.6; p = .001) compared with usual maternity care.
EXPLORATORY DETERIORATION ANALYSIS
1.5% vs 8.2%
Two of 134 RELAX participants versus 11 of 134 usual-care participants showed a four-point or greater worsening in GAD-7 scores by eight weeks.
Some anxiety and repetitive-thinking differences persisted at later follow-up, but they were smaller. Everyday functioning did not differ significantly between groups on the trial's work-and-social-adjustment measure. Several depression, worry and perinatal-anxiety differences were not sustained after birth. The researchers' proposed interpretation-bias mechanism was not confirmed as mediating the anxiety difference.
How strong is the evidence?
Evidence-supported result for one studied program and population. Randomization and a prespecified anxiety endpoint strengthen confidence in the measured group difference. But this was a single phase 2b trial in UK maternity services, participants could not be blinded to receiving a program, secondary comparisons were numerous, and eligibility excluded people with higher-risk symptoms. These findings do not establish benefits for teens, severe anxiety, OCD, nonpregnant adults or general-purpose AI chatbots.
RELAX is a specific research intervention; this report does not establish that it is available to the public or a substitute for a clinician-led therapy. Do not change an established treatment plan because of this study.
Three practical steps for a care conversation
- Name the pattern. Does worry recur around health, uncertainty, relationships or being a good parent? What helps you feel supported without dismissing your concerns?
- Ask about evaluated options. A maternity or mental-health clinician can discuss whether evidence-based CBT or perinatal mental-health services fit the situation and how to access them.
- Track what matters. Beyond symptom scores, note sleep, daily functioning, quality of life and the ability to ask for help.
Questions to bring to an appointment
“What level of anxiety needs additional assessment?” “Which perinatal treatments or digital programs are appropriately evaluated?” “What should I do if symptoms worsen or I feel unsafe?” Save these in an appointment sheet →
Original sources
- Hirsch and colleagues, JAMA Network Open, published November 4, 2026 ↗
- RELAX trial registry ISRCTN12754931 ↗
- Practical caregiver and parent support resources →
Editorial note: Original trial publication November 4, 2026; NeuroCoach summary added November 5. Independent clinical review pending. Educational information, not a personal treatment plan or clinical service.