Open Dialogue and ODDESSI: What Have We Learned?

In late August, the results of a large randomized controlled study of Open Dialogue, known as the ODDESSI trial, were published. As I awaited them, I reflected on my own experiences with Open Dialogue (OD).

I first learned of OD when I read Anatomy of an Epidemic. That book challenged current standard psychiatric practice with respect to the long-term use of psychiatric drugs. At the time I first read Anatomy, I considered antipsychotic drugs (APDs) essential to the treatment of people who experienced psychosis or mania. Anatomy argued that long-term use of APDs worsened outcomes. I found the book sufficiently compelling to prompt my own review of the research data and to formulate my own conclusions drawn from those data. I have written about this in many earlier blog posts.

One challenge for me is that I had limited tools beyond APDs for helping people who were in the midst of psychotic states. At the end of Anatomy, several alternative approaches were discussed, including Open Dialogue, which had been developed in northern Finland in the 1980s.

Several studies on Finnish OD have been published. The subjects in those studies were individuals experiencing their first episodes of psychosis. Since OD was the only treatment offered in the region, the researchers compared their outcomes to those of people who had been treated in another area. The outcomes were stunning. After five years, only about 20% of those treated with OD were on disability, versus about 60% in the comparison group. About 20% of people in the OD group were ever treated with antipsychotic drugs and, after 5 years, about 17% were still taking them. In the comparison group, those proportions were 93% and 75%, respectively, numbers which mirror current standard treatment.

I was fortunate to attend conferences in Finland and Norway, where I met clinicians who were experienced in OD. I later received training and was able to bring that mode of practice (or at least aspects of it) to my own clinic. What I learned was that the use of medications was only one part of a broader shift in both the conceptualization of mental illness and the implementation of services.

OD involves having a team respond to a person in crisis within 24 hours of learning of its onset. The team arranges an initial meeting in a location that is most convenient for the person and invites family members and/or other important people in the person’s life to participate. This same team stays with this person and the network through the crisis.

The conceptual shifts are related to the attitude of the clinicians. They engage with a sense of curiosity, uncertainty, and humility. OD does not embrace a conventional clinical approach where the experts suggest treatment based on the diagnoses they formulate. Rather, it recognizes that there might be many ways of understanding and making sense of a crisis situation. It was reported that, often, the crisis would resolve and medications were not needed or could be stopped. In the conventional approach, the diagnosis dictates the recommended treatment. If the diagnosis is schizophrenia, for example, the expert consensus is that antipsychotic drugs should be started as soon as possible and, often, are continued indefinitely.

I found OD to be a compelling and deeply humane way of working but I also learned how hard it is to implement and sustain. The reluctance to assign a psychiatric diagnosis quickly is problematic in part because (at least in the US) diagnosis is integrally tied to how payment is made. Training is long and, with the turnover in our systems, it sometimes felt like we take one step forward as we train people and two steps back as they leave. And then there are the logistics of having a team that is able to respond quickly and remain with a network through the crisis. Bringing many people together to a meeting is a challenge.

Mental health systems are large and expensive. There are often many ideas about effective treatment. To implement OD, one needs to convince people who are unfamiliar with it of its value. When the current ODDESSI research team was organizing their study, they published a review of existing published studies and concluded, “It is vital that an extensive evaluation of its efficacy take place.” For these reasons, many of us have eagerly anticipated the results of the ODDESSI trial.

ODDESSI was an audacious and remarkable study whose aim was “to determine whether Open Dialogue was more clinically effective than treatment as usual.” We owe those who conducted and participated in the study—through COVID no less!—much respect and gratitude. Clinicians in five mental health trusts in and around London were trained. The researchers identified clusters of general medical practices (GPs) and randomization took place at that level. For instance, if an individual was identified as a candidate for the study, the person was not randomly assigned to OD or treatment as usual (TAU). Rather, treatment assignment was dependent on that of the GP cluster. This is an important distinction that makes sense for this type of team intervention. But it does mean that individuals and clinicians knew what treatment they were receiving.

This study was not designed to be a replication of the Finnish OD studies. Rather than focus on individuals experiencing a first episode of psychosis, a broad swath of people was eligible. Participants were not recruited based upon psychiatric diagnosis. The criterion for inclusion was that a person was presenting in a “mental health crisis.”

Ultimately, 494 individuals consented to participate and 354 attended an initial interview. I am not going to delve further into the characteristics of this cohort with the exception of pointing out that a minority were diagnosed with psychotic disorders—31 (11%) in the OD group and 46 (21%) in TAU. In addition, most people in this study (71%) were employed at the time they enrolled.

The primary outcome measure was time to relapse following initial recovery; raters were blinded to the treatment. There was no difference between the groups. There were also no differences in secondary outcomes (for which raters were also blinded), which included time to first recovery and total days in recovery. The risk of having an additional inpatient admission was lower in the OD group.

There were a number of questionnaires completed by participants (who were not blinded to their treatment assignments) and the results of those self-rated measures, which included satisfaction, overall health, and recovery, favored OD.

The initial research on OD had been on individuals experiencing their first episodes of psychosis. The use of OD, however, is not confined to that patient cohort; it was the treatment model for everyone who sought care in and around Tornio, Finland. ODDESSI is a test of that system of care. This is valuable but I am not sure the results reported thus far will be compelling enough to influence those who fund mental health care systems. The researchers promised future reports addressing the economic aspects. The OD cohort had fewer hospitalizations; if they can demonstrate significantly lower costs, that will create a powerful economic incentive for implementation.

I hope there will be future papers that address the logistics of implementation. Having teams who respond to a crisis and stay with the network through the crisis is a logistical challenge. I would like to know more about how this was accomplished. In their initial press conference, the researchers said that only a small number of network teams included family members. They alluded to lower use of medication, improved clinician satisfaction, and reduced staff turnover in the context of OD. These are all important topics and I look forward to learning more.

At the same time, many questions remain about the value of OD in the treatment of psychosis. Kim Mueser wrote in a commentary about the available pre-ODDESSI data that OD “promises to fundamentally alter the long-term trajectory of schizophrenia.” Can a majority of individuals experiencing psychosis recover with limited use of APDs?

In my own experience, OD has helped me to meet individuals and their families in a thoughtful and respectful manner that acknowledges that when a person experiences psychosis, it is often a crisis not only for that person but also for the family. It provides a framework in which I can hold off on using antipsychotic drugs and remain open to finding other ways of addressing the crisis. This seems important because we already know that there is a sizable number of individuals who go on (regardless of treatment) to have good outcomes. Employment of the principles that characterize OD enhances my ability to identify them and avoid assigning them prematurely a dire diagnosis. But that is different from altering the trajectory of those who would have fared poorly with traditional treatment.

Although some of the people I have treated using OD principles have done well, the outcomes I observe do not come close to matching the OD study outcomes. That might be a reflection of my limited clinical skills but, despite being someone who appreciates OD, I await replication of the Finnish data.

OD is a manifestation of a collection of values. Many of us who support the broader implementation of OD do so on the basis of these values: reducing fragmentation of care; elevating the voices of the people at the center of concern; inclusion of families; practicing with transparency, respect, and humility; and embracing uncertainty and flexibility. We might not need research to inform us of the importance of those values, but we do need more evidence if we want to make claims about the impacts those values—when embedded into a treatment approach—have on outcomes.