Open Dialogue approaches to psychosis – reflections on a two day workshop

I spent two days this week attending a workshop entitled – Developing Open Dialogue; so I thought I would use this blog post as a means of reflection on the course and the feelings I was left with in relation to open dialogue as an approach to mental distress.

The workshop

The workshop was hosted in Sheffield and was coordinated by an experienced family therapist, a clinical psychologist and a narrative therapist.

The audience was mixed – consisting of psychiatric nurses, clinical psychologists, psychotherapists and two medics, including myself. 

Goals discussed for the workshop included gaining a greater understanding of the role of open dialogue and also means of working with services to incorporate the approach into clinical practice.

The theory

The underlying theoretical assumptions presented in relation to open dialogue were as follows:

  • Psychotic behaviour represents a strategy to manage behavioural experience (response to trauma for example)
  • Hallucination represents a metaphor for actual experience
  • Overall psychotic experience represents a mental phenomena that has not yet been linguistically processed

The practice

Building on this theory the open dialogue approach was described not as an intervention per se, but instead as providing a space in which ideas could be communicated between group members. 

The support offered centred around the conduct of meetings between clinical staff and identified members of the distressed individual’s social network. The following principles were proposed as being fundamental to the approach:

  • Immediate help
    • To be received within 24 hours
    • That all participants be involved from the outset (client and family members for example)
  • Social network perspective
    • All those identified as having a stake in defining the problem to be included in the process
    • Psychotic stories to be discussed in open dialogue with all those present
  • Flexibility and mobility
    • Response is adapted to the needs of every social group
    • Other therapies are integrated and not excluded
  • Responsibility
    • The first person contacted should ideally, where logistically possible, be responsible for arranging meetings
    • All issues are to be openly discussed in meetings
  • Psychological continuity
    • Integrated team approach – both inpatient and outpatient
    • Meeting as often as needed
  • Tolerance of uncertainty
    • Generation of safe space for all to be heard
  • Dialogism
    • Allow a space for new narrative development

I have highlighted what I understand to the the key change mechanisms in the process.

The evidence

Two papers were cited in the workshop as providing evidence for the efficacy of the approach – Aaltonen and Seikkula (2011 – both open access).

I supplemented this with a PubMed search today, which sadly allowed for little more insight.

There is a third outcome paper that I could not access through the University of Manchester library, but an author proof version is linked to here (Seikkula 2003).

I’m going to focus primarily on the Seikkula 2003 paper here as this attempts comparison between open dialogue and other treatment approaches, the other two papers rely mainly on historical cohort analysis and I feel are quite difficult to interpret. 

Seikkula 2003

This paper aimed to:

The study reported in this article aimed at clarifying the effectiveness of OD in treatment of first episode psychosis.

The authors used data from three different cohorts of participants. The first – labelled the API (1992-1993) group were recruited as part of a larger trial looking at different approaches to the treatment of psychosis in different geographical regions of Finland. Similarly the comparison group was taken from this trial and represented a “standard treatment method” – primarily it seems with neuroleptic medication. The final group, ODAP (1994-97), represented a local continuation of the original trial that sought to follow the development of the open dialogue principles. The authors comment that many of the principles of open dialogue were present in the original API cohort, but that the training and methods developed by the time of the ODAP cohort. 

Participants were assessed at baseline recruitment and then at 2 year follow up. All referred patients with psychotic experiences were included initially in the assessment. Clinical assessment by a psychiatrist led to an initial “hypothesis” which was then re-assessed after 6 months to reach a final diagnosis (the assessing psychiatrist was the second author on the paper). It is stated that “an experienced psychiatrist from outside the two regions served as an outside rater to test the reliability of diagnosis. The agreement of diagnoses in the case of Schizophrenia was 78% for the API and 80% for the comparison group. Agreement on other diagnoses (schizoaffective disorder for example) are not given.

The API cohort consisted of 39 participants, of whom 5 were lost to follow up, while the ODAP cohort consisted of 54 with 10 lost to follow up. While the authors had access to data on 21 participants in the comparison group on comparing the groups they found that;

 only the schizophrenic group (N=14) was comparable to both the Western Lapland groups

As such the final comparison was made between API (N=22), ODAP (N=23) and comparison (N=14). Neuroleptic medication was more likely to be prescribed in the comparison group, there was no difference in rates of individual therapy between groups. 

The method in which statistical comparisons are made between groups is somewhat confusing. Most comparisons are made between API and comparison group. Occasionally they are made between ODAP and comparison and finally they are also made between ODAP and API.

In terms of results the author’s suggest a statistically significant reduction in days in hospital between API and comparison group. It should be noted that the standard deviation on these figures is enormous however and it would have been interesting to see these figures presented with 95% confidence intervals, as many publications would now require. More family meetings were held for the API group than for the comparison group, as would likely be expected given the nature of the approach. Finally the mean BPRS scale for the API group at 2 year follow up appears to be greater than for the comparison group, however the comparison statistical test for this is not given in the paper. 

The number of relapsed patients in the comparison group was greater than in the API group. The number of patients in employment or education was quoted as being greater in the ODAP group than in the comparison group – I would question the significance of this finding given that the ODAP group was drawn from a different time period? Finally the number of residual psychotic symptoms appears to be fewer in the ODAP group than comparison group.

Finally an analysis was made for variance in Global assessment of functioning (GAF), psychotic symptoms and employment status in terms of time, group and interaction effects. A significant time main effect was observed for GAF and psychotic symptoms. A group effect was not observed for GAF but was seen for psychotic symptoms.

Summary – Seikkula 2003

I found this paper incredibly difficult to interpret. At the outset of the the methods section all individuals presenting with non-affective psychotic symptoms were included, however by the time of the analysis only those with a schizophrenia diagnosis were included. The statement;

only the schizophrenic group (N=14) was comparable to both the Western Lapland groups

I feel this is somewhat confusing as well – why were the remaining patients not comparable? How do we know that all those who responded well to treatment with neuroleptics were not simply excluded from analysis at this point? I have tried to access other reports of the API project from Finland, but hit a paywall that I can’t get through in each case… (this is using a University of Manchester subscription which has its limits but generally accesses most material). 

In terms of statistical comparison between groups I found the methods difficult to follow again. Most comparisons were made between the API and comparison group, who were recruited within the same time frame and these comparisons seem reasonable. However comparisons between the ODAP group and comparison group, for example in terms of employment rates, seems more questionable? 

Some of the results were also confusing to me – for example it seems strange that the numbers of patients with 0-1 residual psychotic symptoms was greater in the API and ODAP group than in the comparison group, yet the mean BPRS score appeared higher in the API than comparison group?


Throughout the two day workshop I struggled to work out whether open dialogue represented a therapy in its own right, or whether it was simply proposed as a service provision model with other therapies to be provided in parallel?

As a service provision model it seems to me that open dialogue has a great deal to say in commentary on our current practice. For example to me it seems self apparent that care decisions must be made in a transparent manner with service users? Also the involvement of family or other social network members in the care process also seems a self-apparent good? However I am not certain that what open dialogue is laying out is novel in this regard? Do we really need this model to tell us that involving the family and client directly in the care process is necessary? Is this not self-apparent? There are limitations sadly present within the mental health system – however it would seem to me that these represent complexities of the social system and require a much greater scrutiny in general, not just by mental health professionals but by society as a whole?

Perhaps open dialogue represents a novel therapeutic approach? Again I’m not clear on this. The approach makes theoretical claims regarding the nature of psychosis – inability to clearly express suffering through dialogue for example and the role of hallucinatory experience as metaphor? There is much in this approach that I can see as valuable however I am not clear that there is the theoretical depth here that is present in other formulations of psychosis – for example psychodynamic approaches?

Unfortunately if open dialogue is to position itself as a novel therapeutic approach it will need to play the evidence base game. I have strong reservations about the role and interpretation of randomised control trials in mental health – however these do represent the current gold standard methodological approach. Compared to these standards I found the Seikkula 2003 paper unfortunately lacking and I can not make a clear, personal, interpretation of the evidence presented. 


I believe that continuity of care as well as active involvement and encouragement of dialogue between service providers, service users and wider social networks are essential features of good care in the mental health services. I think there is much therapeutic benefit to be gained through these approaches. Unfortunately I do not feel that open dialogue offers me anything novel in these areas. Yes – I agree that many of the principles outlined are beneficial, and certainly need wider adoption, but does it need to be called open dialogue instead of simply “good practice”? Put simply I feel that we have known this for a long time but that the complexities of service provision and social pressures have limited its general application (not seeking to defend this, merely a comment). Beyond emphasising these features of good practice I am left uncertain with regard to my reading of the evidence base. Unfortunately I feel that a well designed RCT with integrated qualitative analysis and service user led outcome measure definition is essential here. 

I await this trial eagerly and will read it with great interest…

5 thoughts on “Open Dialogue approaches to psychosis – reflections on a two day workshop

  1. Hi,
    I began writing a summary that will highlight the main empirical findings (positive or negative) of the Open Dialogue. I have downloaded up about a dozen of articles so far. I am far from being done, but let me know if you would like me to send it to you when done:

    BTW, I know that you wrote on this blog about the Wunderink study on the discontinuation treatment arm for patients with psychosis, and I have this lingering question that I would love you to discuss given your interest in evidence-based psychiatry.

    The Wunderink study has clearly randomised patients in two amrs, but does it qualify as a controlled study? What is controlled, other than making sure that the two arms have approximately the same group of patients? The health-care providers were not the same between the two arms, as far as I can recall. The relationships between the mental health team and the patient, necessarily different as a consequence of a different treatment paradigm, were not controlled for a myriad of variables (warmth, trust, confidence, time spent together, and so on and so forth). In my opinion, the Wunderink study should be qualified as a randomised uncontrolled trial, shouldn’t?

    • Dear Marc,

      Many thanks for your comment on my blog – I would be most interested to hear your thoughts on the evidence for open dialogue approaches.

      Regarding the Wunderink study I have also written about this study for the Mental Elf – the post can be found here (

      Briefly though I think there are some methodological limitations to the Wunderink long term follow up that prevent firm conclusions being drawn – there is obviously immense treatment variation between individuals within both arms of the trial, although there is some evidence that the initial variations in prescribing may have been maintained over the follow up period.

      I do think that the study warrants closer attention though we certainly need more, in-depth, research as this is clearly a pertinent clinical problem to consider. Long term studies are clearly difficult to arrange, monitor and, by definition, take a long time to report findings, but given the number of people on long term psychotropic meds I really do think that we need a clear long term evidence base on which to base discussion and prescribing considerations.

      I look forward to hearing your thoughts,

  2. Hi Andrew, interesting post – (also interested in your Mental Elf review re the ethnographic study on prison officers and prisoners with mental health issues, which is how I found your blog-site). I take your point about ‘but does it need to be called open dialogue instead of simply “good practice”‘ , but I think that it isn’t just lack of resource that prevents us from taking this kind of approach.

    As a systemic psychotherapist, I would think this, but I do think that a) clinicans generally tend to avoid ‘whole-system approaches’ – we tend to feel much safer with one client with one diagnosis, b) we aren’t much good with letting those systems define their own solutions – it can make us feel inept or redundant and c) we convince ourselves that it is we who have to ‘manage’ risk, oblivious to the fact that it’s only really clients and others who can define or ‘manage’ risk (though we can – of course – help, if asked to do so).

    All those beliefs tend to militate against any use of a more Open Dialogue, or open, or dialogue, approach – and perhaps that’s why using a ‘model’ rather than simply trying to change belief, may have some benefit.

    I know that for those of us not living in Lapland, that the full implementation of the approach isn’t likely this side of hell freezing over, but I think that the obstacles to implementing some of the ‘good practice’ take a lot of belief, attitudinal and service alteration.

    Thanks – good site.

    • Hi Maeve,

      Many thanks for your comment – I suspect that I share some of your beliefs regarding our fear as clinicians about embracing the idea of whole system involvement. In my own research that I’m conducting at the moment there is certainly a good deal to support the idea that we can not consider the individual as being divorced from their social networks and surroundings. I do wonder if we are still too wed to the idea that there is a disorder or illness in the individual that can be corrected through technological intervention.

      Best wishes

  3. Hi Andrew, yes, agree entirely. Hope your research goes well – it’s hard not to become wholly obsessed, I find – I think that’s the only way one ever survives the process…good luck,


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s