Experimental condition.

In the experimental group, clinicians and patients were instructed to use DIALOG+ once per month over a six-month period and flexibly thereafter at their discretion. DIALOG+ is delivered using a tablet computer throughout to display each stage visually and record the outcome of the session. It includes an initial assessment of the patient’s satisfaction with eight life domains (e.g. mental health, physical health, job situation) and three treatment domains (medication, practical help, meetings with professionals) on a Likert scale. This is followed by a review of the ratings, positive feedback on high-scoring domains, and selection of the domains to discuss further in the meeting. Finally, a 4-step solution-focused approach is used to discuss the chosen domain and identify actions. The steps include (1) understanding the negative and positive aspects of the situation, (2) looking forward towards solution-focused goals, (3) exploring what the patient, clinician and others can do, and (4) agreeing actions. Further details of the intervention and training are reported elsewhere [6].

Control condition.

In the control group, patients independently completed the same ratings on a tablet computer at the end of their routine meetings once a month for six months. There was no collaboration or further discussion with the clinician.

Views of patients and clinicians.

Qualitative data on the views and experiences of patients and clinicians in the experimental group were collected shortly after their final DIALOG+ session. Patients took part in focus groups based on a semi-structured interview schedule that was developed by the research team and piloted with service users. Clinicians also participated in focus groups or, where this was not practically possible, individual interviews. For clinicians, a semi-structured interview schedule was also developed by the research team based on the DIALOG+ procedure and piloted with trainers. Focus groups were preferred due to the benefits of interaction among members [17].

Upon enrolment to the main trial, all participants were asked for additional written informed consent to participate in the focus groups or interviews. All clinicians and 80% of patients in the experimental group agreed to participate (the remaining 20% refused). A convenience sample was then used whereby consenting participants were allocated to focus groups based on availability. All sessions were conducted at Community Mental Health Teams by one facilitator (EG, male) and one co-facilitator (SW or LK, both female), who made notes throughout, with no non-participants present. Researchers were Graduate Research Assistants trained in qualitative methods. Most participants had met the researchers previously during the recruitment and outcome assessments for the trial. All participants were made aware that the purpose was to share their opinion on their experience of DIALOG+ and researchers were looking for no right or wrong answers. Sessions lasted approximately 60–90 minutes, were audio-recorded and transcribed verbatim. No transcripts were returned to participants for comment and no repeat interviews took place. Data collection took place between July 2013 and May 2014, ending once the research team agreed that data saturation had been reached.

Content of DIALOG+ sessions.

In the experimental group, the details of each DIALOG+ session were automatically recorded on the tablet computer. This included the agreed action items that were entered into the software by the clinician at the end of the 4-step approach. The data were synchronised to a server where they became available to the research team. This data reflects the content and outcome (i.e. the agreed solutions) of each DIALOG+ session over the 12 month study period. Equivalent data were not available in the control group.

SQOL assessment.

The primary outcome in the trial was SQOL as measured on the MANSA [16] during meetings with blinded assessors at baseline, three, six, and 12 months. Patients in the experimental and control group rated their satisfaction with different life domains on 12 Likert scales from 1 (“couldn’t be worse”) to 7 (“couldn’t be better”). This includes a general satisfaction item (“how satisfied are you with your life as a whole today?”) and 11 life domains (e.g. mental health, job situation, accommodation). This process evaluation focused only on the 12 month SQOL assessment. This is because a consistent effect was found across all three time points in the trial [6] but only the 12 month data reflect longer term outcomes after all patients had finished receiving DIALOG+.

View of patients and clinicians.

An inductive thematic analysis was used to analyse the focus groups and individual interviews, whereby the identified themes were derived from the data at a semantic level (i.e. what participants explicitly said) [18]. The facilitator and co-facilitator familiarised themselves with the transcripts and then coded them line by line, identifying a list of initial codes and condensing these into themes. All themes were entered into database software for ongoing comparison and reference. The patient and clinician data were initially analysed independently using this inductive approach. Following this, all themes relating to the research question (i.e. the mechanisms of DIALOG+) were combined across both sets of data. The process was iterative whereby the themes were regularly revisited and discussed among independent analysts and the wider research team to minimise bias and improve validity. Participants did not provide feedback on the findings. The final themes were summarised using verbatim quotes, with a focus on those related to the mechanisms of DIALOG+.

Content analysis of DIALOG+ sessions.

A content analysis [19] of the action item data was conducted to simultaneously explore: (1) the nature of the agreed actions (at the level of individual action items), and (2) the domains discussed (at the level of each DIALOG+ session). This approach was appropriate given the large number of action items, their concise nature, and the need to condense the data to summarise the content.

For the content analysis of the agreed actions, each individual action item was coded for two pre-determined themes: (i) the person responsible for completing the action; and within that (ii) the type of action. The codes within each overarching theme, however, were generated through an inductive approach. Initially, each action item was coded according to who was responsible for the action and the type of action using preliminary codes. The preliminary codes were then condensed further into common categories. The frequency of action items within each category was counted.

For the content analysis of the domains discussed, a pre-determined coding framework was used based on the 11 specific domains in the MANSA scale [16]. Each DIALOG+ session was coded one-by-one according to the domains that were explicitly discussed (e.g. mental health, physical health, job situation). The frequency of sessions addressing each domain was counted. This analysis was later combined with the quantitative analysis of the SQOL domains.

Quantitative analysis of SQOL domains.

An exploratory analysis was conducted to identify which of the 12 MANSA items showed the largest effect of DIALOG+ compared to the control group, or if there was a consistent effect across all domains. This analysis included patients randomised to either the experimental or control group, with the outcome assessed at 12 months. Patients in the experimental group were excluded if they did not receive any DIALOG+ sessions. This allowed for better integration with the content analysis. Using the same methodology as the primary analysis [6], which was pre-specified in an analysis plan, this analysis was conducted at the level of the individual and based on available cases. Mixed-effects linear regressions were conducted for each MANSA item with a fixed effect for treatment group and the associated baseline value of the item, and a random effect for clinician to account for clustering. This analysis was combined with the content analysis to explore whether the effect was specific to the domains addressed or consistent across all domains. Given the exploratory nature, the focus was on the adjusted mean differences, although confidence intervals and significance levels are also reported.

A comprehensive structure.

This theme relates to the benefits of DIALOG+ as a comprehensive structure that forms the basis of routine meetings. Within this, many clinicians reported that the assessment of satisfaction provided a holistic structure that covers all aspects of the patient’s life, not just their mental health.

It is actually very beneficial because you have your questionnaire that has been structured in such a way that you can cover almost everything that you need to cover with a client, from the psychosocial intervention to other things like their physical health, their mental health, their social life, their relationships…(Clinician [C] 1, Focus Group [FG] 7)

In addition to the holistic assessment, some clinicians reported that the structure made the meetings more focused by setting an agenda based on important topics.

[DIALOG+ was] a mechanism for formulating an agenda and, you know, letting them know that “we’ve got an hour together, let’s make the most out of what you need to talk about” and this enables us to do that … It just was a catalyst, it was a useful way to focus on relevant topics rather than irrelevant topics. (C1, FG8)

Similarly, patients reported the benefits of DIALOG+ covering a wide range of topics.

It was more structured and more focussed. … And it, and it… And, we discussed different issues. Like, it was more of a holistic sort of approach? Like, discussing all areas of mental health. (Patient [P] 1, FG1)

It was thorough and it seemed like we went over everything. (P2, FG2)

Patients also reported that the structured approach facilitated the agreement of solutions by the end of the discussion, ensuring that change occurs.

[DIALOG+] was more focussed, and, um, there was like actions at the end where, after we had discussed, the few topics that we chose…she would make actions where she would say, “What could be done about it?” So, she would make notes in, like, say “Contact so and so for this”… And I think that was better, because things got done, in that way… Issues got addressed… Constructive things were being done about certain issues. So I think more and more was being done, with [DIALOG+] in place. (P1, FG1)

Some patients highlighted that the tablet computer itself was useful in facilitating this structure.

It stopped your key worker forgetting everything… I did feel more in control when she was using the iPad I felt like we were using a structure, I felt more secure. (P2, FG2)

Self-reflection.

Many patients stated that DIALOG+, especially the assessment of satisfaction, encouraged self-reflection. They reported an increased awareness of their current situation and where changes were needed.

Sometimes you get so caught up dealing with things on a daily basis that you don’t really check yourself, and when you’re asked these questions on a scale of 1–10 (sic) or whatever it kind of gives you more of an insight into how you are actually feeling. (P3, FG3)

It made me really stop and look at my life, and basically, like, the progress I need to make, or um, things that I need to stop doing … You make good use of the knowledge, that you need to improve yourself … Maybe prior to using [DIALOG+] it never actually crossed your mind. But the questions are so in your face that now you realise that, yeah, this is something that I need to work on. (P4, FG4)

Additionally, patients found it useful to monitor changes over time and this helped them to identify improving areas, which also instilled hope.

It helped me track my progress…. On a monthly basis…My lifestyle, my accommodation, my safety, medication, all those things. (P3, FG4)

Sometimes you get so caught up in life you’re just praying for the good days, so it’s nice to know that … you can reflect and say, “well, last month I was feeling shitty but this month I’m all right”, and it gives you a bit more hope for the future, so in that way it’s good. (P3, FG3)

Clinicians also reported that the initial assessment helped patients to reflect on their situation and identify positive changes, which promoted hope and improved self-esteem.

It gives them the opportunity to think, what makes it 3, how can it be 5, or 7, or even when it was 3 and now it’s 7, what has happened for it to go from 3 to 7 … It gives them thought to what has happened in the weeks or so since you’ve not met. (C1, FG7)

They would rate it, say, 2, and then when you try to go, when you try to explore number 2, they say “oh, I should’ve been on 4” because they look at the negative aspect rather than the positive aspect of their lives … So it actually help them to think about the positive aspect rather than the negative aspects … You can actually have a look over the length of time where they are and they can actually see the positive aspect of what they’ve done … It shows them they’ve improved … It increases their self-esteem. (C2, FG7)

Therapeutic self-expression.

Some patients reported that DIALOG+ encouraged them to express themselves, which had a therapeutic impact. The process of exploring and expressing their thoughts and feelings on a wide range of topics helped improve their affect.

Sometimes you’re… not feeling good inside and you’re holding things inside and… once you come in front of the computer, she would ask you questions and… that would… help you to express your feelings and things … I think it was the best way to… get yourself to express the feelings, held inside. (P3, FG1)

It’s like an offload, isn’t it, dust yourself off, you’re up to speed … It’s like you’ve been rebooted with [DIALOG+], it’s like a different kind of therapeutic feeling… This cheers me up… I feel happy to do it. (P2, FG5)

Clinicians also reported the benefits of DIALOG+ in encouraging patients to express themselves and this helped them to learn more about their patients.

I thought it was really good … you got a lot of information. One client actually, I got so much more information, background information out of him because he really responded to this structure actually very well … A client who usually when you ask him “How are you feeling” usually answers “I’m fine” … he was giving away so much more about his past history. (C2, FG6)

Empowerment.

Many clinicians reported that DIALOG+ empowered their patients to become more involved in their treatment and decision-making.

It was empowering because you are … shifting the care co-ordination responsibility and letting them make decisions … it was positive. (C2, FG8)

I found it the most empowering tool in the 10 years I have been qualified as a psychiatric nurse. By far. It helped me to empower my client … By the time we had got to the end, he had taken the reigns into his own hands … It definitely changed our therapeutic relationship… By the end really he was very very much in control of his own care. (C4, FG6)

Clinicians reported that the solution-focused approach, especially, helped to shift the focus of the meetings towards the patient generating ideas and taking responsibility.

Normally some of them will want you to do everything for them but with this solution focused therapy it did help them to see how to actually help themselves rather than being spoon-fed by the care coordinator … It did help them to understand how to actually come up with a solution rather than me saying “this is how to do it” … They come to you thinking “how can I do this” but … it’s like an eye-opener to some of them … They felt empowered to say that actually “I can do this by doing this or doing that, I can actually find a solution to my problem.” (C3, FG7)

Some patients also reported a sense of empowerment, although this was less pronounced than the clinician interviews. This included an increased involvement in deciding which domains to talk about during the meetings.

You would choose them from the list, yea, your care co-ordinator doesn’t choose anything for you, she gives you options, how do you feel. (P2, FG5)

It also included increased reflection on what actions they can take to improve their situation.

It can make you think about what you’re going to do for your life, obviously when you’re asked questions it can make you think about what you can do for yourself as well. (P4, FG3)

Domain-specific change.

The quantitative and content analyses suggest that the effect of DIALOG+ is, to an extent, domain-specific to two areas that are commonly addressed during sessions (accommodation and mental health). This is in contrast to a ‘general appraisal tendency’ whereby a consistent effect might be found across all subjective patient-reported outcomes, often more dependent on the patient’s overall mood than objective change [20–22]. DIALOG+ may instead bring about improvements to patients’ quality of life through addressing a specific concern and initiating positive change in that area. The thematic analysis of participants’ views supports this. Among the theme regarding a comprehensive structure, participants reported that DIALOG+ focused the discussion on the main issues and ensured constructive actions were agreed. Although physical health was also commonly addressed, and healthy lifestyle changes were the most frequent action items, there was no significant effect on this domain—perhaps due to the numerous barriers to improving such outcomes in this patient group [23].

A comprehensive, solution-focused structure.

Previous attempts to improve routine meetings in community mental health care have had a limited effect, possibly because they fail to influence clinician and patient behaviour [3–5]. In contrast, DIALOG+ provides a model to directly guide the discussion that takes place. Simply incorporating such a model is important to maximise the therapeutic effect of any patient-clinician communication [24]. Both the thematic analysis and content analysis indicate that the DIALOG+ model is comprehensive, addressing a wide range of potential concerns. This includes practical issues as well as psychological ones, which will likely have a real impact on the patient’s life. Indeed, actions involving practical assistance (e.g. help paying bills or completing forms) were among the most commonly agreed during sessions. This holistic approach is important to address the complex range of health and social needs in this patient group [25].

The intervention also goes beyond an initial holistic assessment [26]. Participants reported that, in addition to setting an agenda for discussion, the structure encourages the identification of constructive solutions, ensuring that change occurs. This is reflected both in the moderate number of action items agreed (about five per session) and the apparent domain-specific effect. Although not central to the intervention, the computer technology may help to facilitate this comprehensive, solution-focused structure. This is in line with findings that a computer-assisted approach can maximise the effectiveness of therapeutic encounters [27, 28].

Self-reflection and therapeutic expression.

Two unanticipated mechanisms that arose through the qualitative findings were self-reflection and therapeutic self-expression. Although perhaps not central to the domain-specific effect, these mechanisms might play a key role in improving general psychopathological symptoms (including depression and anxiety) as found in the main trial outcomes [6]. The initial assessment encouraged patients to reflect on their current situation and facilitated better insight into both the problematic and improving areas. Awareness of positive progress instilled hope which might have contributed to improved mood [29, 30]. The initial assessment also encouraged patients to express themselves which itself, as reported by patients, can have a therapeutic impact on affect [31].

Empowering patients.

The content analysis found that patients receiving DIALOG+ took responsibility for the majority of agreed actions (65%). In comparison, previous findings showed patients are responsible for only a minority of actions (8%) in community mental health meetings [10]. This striking difference suggests DIALOG+ facilitates a shift toward the patient taking control. This is supported by the qualitative findings in which patients and clinicians commonly reported a change from clinician-led meetings towards empowering the patient.

The process of empowerment may be a key mechanism through which DIALOG+ initiates positive change. Empowerment itself is often identified by patients as important to recovery [32] and is associated with better quality of life and symptoms [33]. DIALOG+ appears to facilitate two key aspects of this process: participation in decisions and self-reliance [34]. Patients prefer shared-decision making [35] and a clinical orientation towards the patient taking more control can increase satisfaction [36]. Yet, patients often report a lack of involvement in treatment decisions [37]. The nature of DIALOG+ encourages patients to identify their own needs, goals and solutions and encourages them take more responsibility, both during treatment meetings and in the actions required to initiate change. This process of empowerment supports the benefits of DIALOG+ adopting a resource-oriented approach, whereby utilisation of the patient’s own strengths and resources is maximised [9].