Research review 035
Randomised study
Short answer
The Circle of Security-Parenting support programme did not produce a clear additional improvement in mental health symptoms when added to existing specialist treatment. No clear additional benefit was found for the other outcomes examined either. The result concerns this programme in this setting, not all parenting support or every form of postnatal treatment.
What did the researchers investigate?
COSI was a large treatment trial in routine healthcare services. Researchers examined a group programme intended to help parents understand their child's needs and the relationship between parent and child. The programme was added to treatment participants were already receiving.
Participants were randomly assigned. The people assessing outcomes were not supposed to know which group participants belonged to. Mental health symptoms and other outcomes were assessed at three, seven and twelve months. This design is better suited to assessing additional benefits than a survey asking only whether participants liked a course.
What did they find?
No clear additional symptom benefit was demonstrated
The programme did not produce a clear additional reduction in mental health symptoms during follow-up. The difference was neither statistically clear nor large enough to be considered important for patients. Researchers also found no clear differences in the other outcomes examined.
The group programme consisted of ten sessions. Participants in the intervention group attended approximately six on average. The findings therefore describe the effect of offering the programme within routine care, with the attendance that actually occurred. They do not describe only a small group that completed everything.
Both groups received treatment. The absence of a difference therefore does not mean that participants did not need help or that receiving no treatment would have produced the same result. The question was whether this particular addition improved outcomes.
How certain can we be?
Random allocation, multiple treatment sites and outcome assessors unaware of group assignments strengthen the findings. Nevertheless, many participants did not provide follow-up data. Approximately 75% participated in the assessment after three months and 63% after twelve months. Missing data may influence the findings if those who did not respond had different experiences.
The study population had substantial and complex needs. Findings may differ for parents not receiving this level of specialist care. The study also cannot establish that the effect is precisely zero in every subgroup. It found no evidence of a clear average additional benefit in this service.
Who do the findings apply to?
The trial included 371 birthing parents at ten specialist perinatal mental health services in England. Of these, 248 were offered the programme alongside usual treatment, while 123 received usual treatment alone.
Participants were at least 18 years old, had a child younger than one year and met the study thresholds for mental health symptoms and difficulties in their relationship with the child. Active psychosis and certain serious conditions affecting the child were among the exclusion criteria. Most identified as women, and some as non-binary. Approximately 89% reported White ethnicity.
What might this mean in everyday life?
Services can use the findings when considering whether to devote time and resources to this additional programme. A well-known, meaningful or popular service is not, on its own, evidence of improved health. At the same time, the existing treatment and other support services need to be described before the findings are applied to a new setting.
What do we still need to know?
We need better answers about what additional support benefits this patient group alongside treatment they already receive. It is important to investigate whether the content, timing or delivery of an intervention suits their needs.
Researchers should record who attends, who stops attending and why. They can also examine whether other approaches work better for particular groups. These possibilities need testing in new trials. They do not become reliable answers simply because a small subgroup appears to improve in a post-hoc analysis.
Sources
The publication covered by this review
Rosan C, Alyousefi-van Dijk K, Cornelius V, Waddingham E, Darwin Z, Babalis D et al. Clinical effectiveness of the Circle of Security-Parenting group intervention for birthing parents in perinatal mental health services in England (COSI): a pragmatic, multicentre, assessor-masked, randomised controlled trial. The lancet. Psychiatry, 2025.
Other studies and professional sources mentioned in the text
Background source: Public Health Nurse-delivered Group Cognitive Behavioural Therapy for Postpartum Depression: A Randomized Controlled Trial. 2022. Read the publication.
Background source: Enduring effects of brief interpersonal therapy on depression from pregnancy through 1-year postpartum: A randomized clinical trial. 2026. Read the publication.
Last reviewed: 30 October 2026.
Source basis: The publication's full text.
Editorial update, 1 November 2026: Language, explanations and references were reviewed. The source-review date is stated above.