Design: Multisite, cluster-randomised superiority trial
Population: 494 adults presenting in crisis to community mental health services in England
Primary outcome: Time to first relapse after recovery over 2 years
Effect: Marginal HR 0.95 (0.67 to 1.32)

The ODDESSI trial, published in August 2026, randomised 30 primary care clusters across five English mental health trusts to Open Dialogue or the usual functional team model. Open Dialogue organises care around network meetings with the person, their family and social network, and usually the same two practitioners throughout. 494 adults in crisis took part and were followed for two years.

The primary outcome, time to first relapse after recovery, did not differ (marginal HR 0.95, 95% CI 0.67 to 1.32). Open Dialogue was associated with lower probability of psychiatric admission and of re-referral to crisis or secondary services, and with better self-rated recovery, quality of life and satisfaction. Social network size and quality did not change. Most serious adverse events, 376 of 386, were judged unrelated to the intervention.

The headline is a null result, and that should be said plainly. The secondary signals on bed use and experience are encouraging but are exactly the kind of finding that needs confirmation. For services considering network-based crisis care, the trial shows it can be delivered at scale in a public system without obvious harm.

Do not expect network-based crisis care to reduce relapse on this evidence; the primary outcome was null.
Consider involving the family and social network in crisis meetings where the person agrees.
Aim for continuity of the same practitioners through a crisis episode where staffing allows.
Treat the lower admission rate as a hypothesis worth testing locally, not a settled benefit.

Why it matters

A service model can improve experience and bed use without changing the outcome it was designed to move.

The statistics, in plain English

A hazard ratio of 0.95 with an interval from 0.67 to 1.32 crosses 1.0, so no difference in relapse was shown. The admission and recovery benefits are secondary outcomes; because many were tested, some may be chance findings.