First-episode psychosis programs, such as RAISE (Recovery After an Initial Schizophrenia Episode), EASA (Early Assessment and Support Alliance), and others, are finding that early effective engagement improves clinical and functional outcomes (
1). Open Dialogue (OD) is a clinical model, developed in Finland, to improve functional outcomes for young people experiencing acute psychosis or another psychiatric crisis (
2). OD provides services in “network meetings” that involve the person in crisis, family members, and others in the support network. When contacted concerning a psychotic crisis, a mobile, multidisciplinary psychiatric team rapidly engages the young person and the network in a meeting—often in his or her home.
A five-year outcomes study of a Finnish cohort of 42 young persons with nonaffective psychosis found that 86% were working or in school, only 14% were disabled, and only 17% were still being prescribed antipsychotics (
2). There was no randomized control group; however, a historical control group had worse occupational functioning and symptoms at two-year follow-up (
4). Other groups in the United Kingdom, Italy, Norway, Denmark, Germany, and Poland are implementing OD, but they have not yet reported outcomes.
Implementing and Adapting OD in the United States
We conducted a 12-month feasibility study to adapt and implement the OD approach at a mental health agency in the United States and examined preliminary clinical outcomes. All procedures were approved by the Boston University Institutional Review Board and were overseen by a Data Safety Monitoring Board.
Our team included seven master’s-level clinicians and a psychiatrist, who all completed two years of training in OD at the Institute for Dialogic Practice (
www.dialogicpractice.net). Eligible patients engaged in a complete psychiatric examination and risk assessment prior to enrollment. This initial evaluation was an adaptation of the Finnish model, in which the first meeting is focused on developing a connection with the individual and his or her network.
The program, named the Collaborative Pathway (CP), consisted of an established mobile crisis team, which operated around the clock and 365 days a year, and outpatient services. Network meetings occurred in the clinic and in persons’ homes. Unlike Finnish practice, the treatment team did not provide inpatient care but stayed engaged with participants and their providers during hospitalizations.
We assessed the feasibility of implementing CP by using qualitative interviews with participants and staff, monitoring service use, and tracking service costs. Six participants and their family members engaged in qualitative interviews; all seven program clinicians participated.
Feasibility and effectiveness were assessed through surveys at baseline and at three, six, and 12 months with the Brief Psychiatric Rating Scale (BPRS) (
6), Revised Behavior and Symptom Identification Scale (BASIS-R) (
7), Strauss-Carpenter Level of Function Scale (SCLFS) (
8), Decision Self Efficacy Scale (DSES) (
9), Shared Decision Making Questionnaire (SDMQ) (
10), Autonomy Preference Index (
11), and Client Satisfaction Questionnaire (CSQ) (
12). Using clinical records, we assessed psychiatric medications prescribed, school and work participation, and psychiatric hospital days in the six months before the start of CP services and during the 12 months of the study. Safety was assessed by tracking adverse events, including suicidal acts or violence and unplanned hospitalizations.
Study participants were ages 14 to 35, experiencing psychotic symptoms (or had experienced them within one month of intake), presenting for emergency services voluntarily or involuntarily, able to provide informed consent, and willing to have family participate. Exclusion criteria were severe substance abuse, active suicidal or homicidal ideation or similar risks requiring inpatient care, and neurological disability or significant developmental disability. We enrolled 16 participants. Of note, another ten were excluded because they did not meet study inclusion or exclusion criteria or declined participation. Fourteen of 16 patients completed the full year of the study, and two withdrew but remained in standard care. Data analysis was based on the 14 completers, although only six completed all four assessments (two completed three assessments, three completed two, and three completed only one). Clinical record data were available at all four assessment points for 11 of the 14. We lacked data for three individuals at 12 months because they were not in contact during that period.
The mean±SD age of the 14 completers was 22.67±4.99. Eleven were male, and 13 were Caucasian. Eight were on antipsychotic medications at enrollment, and 11 had prior psychiatric hospitalizations (mean of 1.07±.88 in the preceding year). Mean duration of illness was 41.1±40.7 weeks. Intake diagnoses were schizophrenia spectrum disorders (N=12) and bipolar spectrum disorders (N=2). Diagnoses by the treating psychiatrist at one year were schizophrenia spectrum disorders (N=7), bipolar spectrum disorders (N=5), autism spectrum disorders (N=1), and obsessive-compulsive disorder (N=1).
OD network meetings were successfully implemented and integrated into the clinical program, with good client retention and engagement. Over one year, OD network meetings were held a mean of 12.53 times (range five to 28), and 66 of the meetings (36%) involved a psychiatrist. Scores for client satisfaction (CSQ) and perceptions of shared decision making (SDMQ) were high throughout (CSQ=3.23±.36 on a 4-point scale and SDMQ=5.29±.501 on a 6-point scale; higher scores on both indicate better outcomes).
In qualitative interviews, participants and family members appreciated the openness and transparency of the approach and felt part of decision making. They felt cared for rather than being “on the clock” and appreciated that treatment was not just medication focused. Families cited meeting in their homes and observing the clinicians’ “reflections” as promoting a collaborative atmosphere. There were few criticisms; one family member expressed frustration about a lack of clear direction about medications, and two families cited a need for supplementary social services.
Staff satisfaction was high. The team was well trained by completing two years of training. Staff expressed satisfaction with their ability to better engage patients and families. They enjoyed working with families and in teams. Clinicians reported that this structure promoted nonhospital options by affording additional support and safety. Staff had some concerns about scheduling urgent network meetings while managing their other cases.
Clinical outcomes were generally positive. Results of linear mixed-model analyses showed a significant positive change in symptoms, functioning, and need for care, as measured by the BPRS (p<.001), BASIS-R (p=.002), and SCLFS (p<.001), respectively; average work or school hours per month (p<.001); and hospital days (p=.023). The change in DSES score approached significance (p=.07). Nine of 14 participants were working or in school at one year. Of note, four individuals had six short-term psychiatric hospitalizations (two involuntary).
Three of the six individuals who were not on antipsychotics at program entry started antipsychotics. Of the eight already on antipsychotics, four had no change in their medication, and four elected to stop during the year. Both groups of four had similar outcomes and continued to be followed in treatment. Shared decision making and toleration of uncertainty contributed to these choices.
Per-person costs varied with service intensity, from a low of $5,126 to $10,236 for the year. Third-party reimbursement covered only 23% of service costs, and foundation grants supported uncovered costs.
Lessons Learned and Paths Forward
Results of this feasibility study suggest that the OD model can be successfully integrated into an established U.S. outpatient and crisis program, with satisfaction for participants, families, and staff, and that the model appears to be reasonably safe and clinically effective when implemented with appropriate risk assessment and crisis team availability. However, serious barriers to implementation remain. Training costs and time were substantial. We were fortunate to obtain training from an expert who worked relatively nearby, and we had grant support. Shorter training models are being created, implemented, and tested nationally and internationally.
The model involves costs traditionally uncovered by insurance, such as having at least two clinicians in network meetings, which were often longer and more frequent than covered by insurance. Travel time for home-based services, scheduling off-hour appointments, and supervision added substantial costs, and these were managed with foundation support. Capitated systems, such as accountable care organizations and the Department of Veterans Affairs, may be able to support the OD model.
Engaging the person in crisis and the family by means of support and deep listening, shared decision making, and investment of substantial time, especially in their homes, contributed to collaboration. Network meetings appeared to provide a holding environment to understand the psychotic crisis, explore treatment options, deal with conflict, and process setbacks. Toleration of uncertainly by the family and clinician appeared to allow time for finding solutions that faster decision making might have foreclosed.
This study had important limitations, including a small sample, diagnostic heterogeneity, lack of a control group, missing data, and unblinded clinical ratings.