Informing the development of a decision aid: Expectations and wishes from service users and psychiatrists towards a decision aid for antipsychotics in the inpatient setting

Abstract Objectives Decision aids (DAs) are promising tools to foster evidence‐based shared decision‐making between practitioners and service users. Nevertheless, it is still obscure how an evidence‐based DA for people with severe mental illness, especially psychosis, should look in an inpatient treatment setting to be useful and feasible. Therefore, we conducted focus groups with psychiatrists and service users to collect and assess their expectations and wishes regarding an evidence‐based DA. From these findings, we derived immediate recommendations for the future development of DAs. Methods We held two group interviews with service users (n = 8) and three group interviews with psychiatrists (n = 10). We used an open, large‐scale topic guide. First, we presented data from a current meta‐analysis on antipsychotics to the interviewees and, in a second step, asked for their expectations and wishes towards a DA that integrates these data. Results Our thematic analysis revealed six key themes addressed by the respondents: (1) general considerations on the importance and usefulness of such a DA, (2) critical comments on psychiatry and psychopharmacotherapy, (3) communicative prerequisites for the use of a DA, (4) form and content of the DA, (5) data input, data processing and output as well as (6) application of the DA and possible obstacles. Conclusions Participants identified several important features for the development of DAs for selecting antipsychotics in inpatient psychiatric treatment. The digital format was met with the greatest approval. Especially the adaptability to different needs, users and psychopathologies and the possibility to outsource information dissemination via app seemed to be a decisive convincing argument. Further research is required to test specific features of DAs to be developed in clinical settings.


| INTRODUCTION
Shared decision-making (SDM) is a model for medical encounters in which patients and providers make decisions together as equals while sharing information, seeking consensus and ultimately agreeing on the treatment decision to be made. 1,2 In an initial review of SDM in psychiatry, Hamann et al. 3 found a high desire for participation on the part of patients but several open questions regarding the feasibility of SDM in this setting. As SDM is already successfully practised in other medical fields and its positive impact on health-related outcomes has already been demonstrated, they recommended the development of decision aids (DAs) as well as medical training in risk communication to overcome disease-immanent and systemic barriers. 3 Zisman-Ilani et al. 4 pointed out the tendency of research on SDM in mental health to focus on DAs to promote information sharing between patients and clinicians. At the same time, they also drew attention to approaches that focused more on the joint discussion of values and goals to strengthen the doctor-patient relationship. Strengthening the rapport was assessed as an important approach for implementing SDM in mental health, in line with findings from research on successful therapeutic processes. The authors suggested not to directly apply SDM outcomes from other medical fields. Rather, participation in SDM itself could be considered an important factor in mental health recovery as it promotes empowerment, whereas a reduction in decisional conflict or an increase in knowledge as a potential outcome is more difficult to achieve in this field given the many barriers. 5 Successful implementation of SDM in mental health can be hampered by several barriers. Hamann and Heres 6 described professionals' lack of time or staff's tendency to consider patients with acute psychotic episodes as unsuitable for SDM because of associated cognitive or emotional deficits. Barriers on the patients' side were described including passivity and loss of motivation. 6 However, there is no evidence to date, that cognitive impairments prevent participation in SDM. 7,8 Zisman-Ilani et al. 9 point out the role of stigmatisation on the physician side towards patients with severe mental illness (SMI) as a likely barrier for SDM. Further barriers include distorted information given by physicians, for example, regarding age-or personality-related aspects. 10,11 In addition, it is difficult for psychiatrists to keep track of the multitude of medications available, which also means that both sides are not always aware of the most current evidence on available treatment options. 12 Against the background that antipsychotics often show similar efficacy but different side effect profiles, the intake of antipsychotics is a preference-based decision that lends itself very well to the use of SDM. 13 Although nonadherence rates are high with estimations between 40% and 60%, 14 antipsychotics remain the treatment of choice for schizophrenia according to various treatment guidelines. 15 DAs are promising tools for addressing the above-mentioned problems. Nevertheless, only a few have been developed for the use of antipsychotics. In their scoping review, Müller et al. 16 compared existing tools with the established quality criteria for patient-based DAs with mixed results for the improvement of patient outcomes related to the use of a DA. In summary, the authors consider the current quality criteria for DAs for patients with schizophrenia as inadequate due to the complexities of factors influencing the decision for or against antipsychotics. In any case, they only found one DA that was clearly oriented on the International Patient Decision Aids Standards criteria, the 'APM-DA' or 'the Antipsychotic Medication Decision Aid'. 17,18 Testing of DAs for antipsychotics in inpatient settings is still rare, and existing tools are aimed at either service users or clinicians rather than involving both stakeholders in the process to address barriers on both sides. 19,20 To develop and test a new DA for antipsychotics in inpatient settings, a qualitative study was conducted to explore the expectations and desires for the DA to be developed amongst the two main stakeholders: People treated for psychosis and psychiatrists.

| Study design and setting
To investigate the wishes and expectations of psychiatrists and service users towards an evidence-based DA, a qualitative design was chosen to explore the perspectives in the psychiatric inpatient setting of two hospitals in Munich, Germany. We chose to interview both parties involved separately, to allow the most open expression of opinions. The sample originally included 10 service users and 10 psychiatrists to ensure sufficient discussion and diversity of opinions.
Due to the COVID-19 pandemic restrictions, a maximum of five individuals were allowed to participate, resulting in a total of four interviews to achieve the sample size. One focus group was also split in two (centre 2), as several physicians were unable to attend at short notice during an agreed appointment. Therefore, five group interviews were held in the end. For more information on the dates and the exact number of participants in each focus group, see Table 1.
The first two focus groups were held in the same clinic where the research group is located (centre 1). For this reason, individual service users knew the interviewers, but no therapeutic relationship existed between the interviewers and study participants. For training purposes, the entire team of five researchers participated in the first group interview with psychiatrists in centre 1, consisting of the interviewer F. S., the assistant K. M., two chief physicians/project leaders and one other research assistant. In this case, the two project leaders present were also the supervisors of four of the five participating psychiatrists in their residency training. Afterwards, the research team reflected on the process and structure to inform the subsequent group interviews. On average, the group interviews lasted about 40 min. The study was approved by the local review board of the Technical University of Munich. For the reporting of the manuscript, we rely on the consolidated criteria for reporting qualitative health research checklists. 21

| Participants and recruitment
The inclusion criteria were 'experience in the treatment of psychoses' (psychiatrists) and 'current inpatient treatment' as well as a diagnosis of a schizophrenia spectrum disorder (International Classification of Diseases-10 F20-F29; service users). A convenience sampling strategy was implemented by inviting all psychiatrists eligible for participation until a sufficient number of people was reached). Service users were preselected by approaching the nursing staff on the wards for a brief assessment of which service users could attend a discussion for about an hour. Informed written consent was obtained by F. S., an experienced physician, who insured that the service users were able of consenting.
With the exception of two service users, all those approached were interested in participating. All interviews and recruitment were conducted by either K. M., a female psychologist who is pursuing her dissertation on the project or the second author (F. S.), a male psychiatrist, with the other one taking notes. Both K. M. and F. S. served as research assistants and have experience in inpatient schizophrenia treatment, whereby F. S. was also experienced in conducting and analysing qualitative interviews. Training and supervision were provided by J. H., a professor of psychiatry, with many years of experience in conducting and supervising qualitative studies.

| Data collection
Sociodemographic data were collected before the group interviews.
During the interviews, a broad topic guide was used. First, the evidence of a recently published meta-analysis on antipsychotics 12 was presented, before wishes and expectations for a DA using such data were asked. Questions were for example: How could you imagine this data being best prepared? What content would be particularly important to you? What other aspects can you think of? Audio recordings were collected and transcribed verbatim by using a simple transcription guide. 22

| Data analysis
Data analysis was carried out according to the principles of thematic analysis by Braun and Clarke, 23 using MAXQDA20. The first step was to become familiar with the content by listening to the audio T A B L E 1 Overview and assignment of all patient IDs to the respective focus groups. and their gender ratio was balanced. Amongst service users, the proportion of males predominated 5 out of 8, matching the current estimated gender distribution, which indicates a higher risk of schizophrenia in males. 24 The inclusion criterion for service users was a psychiatric diagnosis requiring treatment with antipsychotics. In addition to paranoid schizophrenia (six out of eight), also other patients (one with bipolar disorder and one with a schizoaffective disorder) were included in the study. The psychiatrists had been practising for about 6.5 years on average. Both clinics involved were teaching hospitals offering specialist training, therefore a correspondingly large number of the staff doctors were in training (see Table 2 for the summary characteristics of psychiatrists and service users).
By performing a thematic analysis with the data, six key categories were identified (see thematic map, Figure 1). In addition to wishes and expectations towards the design of the DA, the groups also shared their concerns and general opinions on the topic. Theme Psychiatrists and service users mentioned the desire for a reliable, easy-to-understand source of information on antipsychotics so that they would no longer have to resort to the summary of product characteristics or internet search engines.  Also, some psychiatrists expressed concern about late side effects of individual drugs, for example, in the form of severe weight gain.

| Theme 5: Data input, processing and output
Thoughts were expressed on what information should be included as input, how this information could be processed and how a possible output could be presented, including possible difficulties associated with it. Figure 2 illustrates the ideas on input and output sorted by frequency of utterance. Most often, the individual patient history was mentioned as a desired input of the app: Various fears were expressed by psychiatrists in the context of a clear drug recommendation, such as the concern that service users might become fixated on the outcome and that they as physicians might be deprived of control over the decision-making process; although the objection was also raised several times that this need not be a concern in the context of shared use. Occasionally, psychiatrists also suggested not making a recommendation but presenting the different antipsychotics on an equal footing.

| Theme 6: Type of use of the DA and influencing factors
Theme 6 includes perceptions from both sides about who should primarily use the DA. In this context, psychiatrists also expressed various concerns or possible barriers to use.

| Type of use
The idea most frequently mentioned by service users and physicians was to hand out the DA to service users for their sole use before or after a joint discussion with the psychiatrist.
T A B L E 3 Wishes regarding the format, content or related to "extra features". 5 most frequent mentioned wishes for the format, content and "extra features" presented with the most frequent mentions first Format Digital format ("App") Explicitly referring to an "App" as a format for the DA or advocating a digital format SU: Yes, I meant that you can also simply include the doctor's report in the app in order to adjust the app to me as a person.

SU & Psychr
Various display options More than one version or interface of the DA (e.g. the idea of extension, reduction options, different logins for service users and psychiatrists, etc.) Psychr: I think there can simply be two user-modes. One for service users to log-in and one for doctors to log-in, so you can start with the simple things for the drugs and then click on them and then get more detailed information and maybe also a quick overview and then again in detail when you can deal with it more closely and have more time. (G2Psychr02, female, paragraph 83)

SU & Psychr
Simple presentation, clarity The desire for clarity in presentation Psychr: Not all physicians are familiar with these study images but it has to be more instinctive for the service users, so with colours, with bars, because not to forget that the service users, even if they are still being cared for by us, are not yet in full remission. Or are just in remission and you really just have to work with more colours, with more shapes, perhaps, so that they can also perceive it better visually.

SU & Psychr
Analogue, on paper The explicit wish for an analogue version of the DA SU: Well, yes, more on paper, you don't always have access to a website or whatever.

SU & Psychr
Avoid deficit orientation The concern that mentioning too many side-effects or focusing on the negative aspects of a drug could lead to a loss of motivation and/or the recommendation to emphasise positive aspects of drug treatment. Psychr: One thing that would still be important to me is when I show this to the patient or when I then have the possibility to print out their profile, then it shouldn't be too deficit-oriented because then it's probably like this: I have the effect size on positive symptoms, on negative symptoms and then there are all the side effects-and that probably overwhelms in proportion and then the patient will just say like with the package insert where everything is red: "yes sorry people, I'm not taking that drug". So there are two things that are green and the rest is simply a disaster and no matter which drug they would say "it's all a disaster, I don't want to take any of them now". Reversibility of side effects The desire for guidance on the management of potential side effects in the DA. Psychr: Reversibility. That when I would say, "okay that side effect, I'll risk it, that I know that I will gain weight but if I then know that 95% of the weight will never come off, then I might deal with it differently". Or in the case of sexual dysfunction, that I know that it is potentially reversible. How many of them manage to get rid of it after stopping the medication? That would also be extremely important for me. I always assess the risk consciously. On the one hand I have to decide on a medication, but then I also have to say that the side effects show up, even if it has perhaps not felt that important to the patient-I take the risk and then I also have to know, if I should change the medication, how sure I can be that these side effects are then also limited in time. I would find that important. Side-effect profile The desire for a risk profile ("side-effect profile") of each antipsychotic SU: Or you can, for example, if we just say weight gain, you can of course also say: "okay, there is weight gain and these and those drugs do that" and then compare the options.

Psychr
Link to other databases Other tools that are already established in clinical practice and currently play a role in medication decision-making (e.g. "AID Clinic", PSIAC) and/or the desire to link the DA with their contents. Psychr: Well, there is, for example, "psychiatry-to-go" or these classic ones, where all the side effects are shown in a clock and then the points go up in different heights and then you have such a graphic construct, and so that's something, for example, I always have it printed out and lying there and then, before I prescribe something, I look at it again, and I haven't really forgotten anything. So that is, I think, already very simply represented graphically.  Table 4). What united both sides was the desire for an easy-to-understand source of reliable information, which seems to indicate a general openness to the use of an evidence-based DA. In particular, the ability to customize the complexity by fading in individual content or setting up different logins for different users seemed to be a decisive advantage of a digital tool. However, the desire for two formats (digital and analogue) was also expressed, to In this context, however, the danger was also pointed out that physicians may then only be able to influence the treatment decision to a limited extent if the users develop 'tunnel vision' as a consequence of the app's results. Psychiatrists, therefore, often also called for a gradual phasing out of medication that would lead to a comprehensible end result.
Both sides expressed the idea that the app could be used by service users alone before or after the consultation, for example, to approach. Indeed, it should not be forgotten that treatment decisions in acute psychiatric settings often involve risk-taking on the part of both the patient and the clinician, for example, when symptom exacerbation may be accompanied by suicidality or homicidality, which could entail liability consequences and thus be a potential impeding factor for the implementation. They, therefore, advocate for distinguishing between decisions of higher risk and those of lower risk as well as communicating this clearly and adapting SDM for people with SMI accordingly.

| Implications for clinical practice and the development of DAs
Based on our interviews, we draw the following conclusions regarding the development of a DA for antipsychotics in the inpatient setting: A digital tool seems to be the most suitable to combine the different expectations of both sides.
We propose an app that can be used and viewed separately by both physicians and service users, but at the same time has a common, scalable interface.
T A B L E 4 Differences and common grounds in wishes and expectations towards the decision aid (DA) of psychiatrists and service users. Our proposal is that patients receive well-prepared information that can be obtained without the presence of psychiatrists and that provides basal information about treatment options and the disease. Experience knowledge with different drugs should also be queried. In parallel, physicians should be able to provide information on possible interactions or comorbid diseases. Both the medical information and the empirical knowledge are to be recorded in the app and can be taken into the joint discussion. There, the evidence should be easily processed and colourfully presented to stimulate joint discussions. Different information should be viewed depending on psychopathology, cognitive abilities, and information needs, by selecting different tabs.
In particular, the scalability of information seems to us to be most easily implemented in the form of a digital tool, as well as best suited to meet the specific requirements in the acute setting.

| Limitations
Several limitations of this study need to be addressed. Contrary to what has recently been emphasized as 'triadic shared decisionmaking', 28 we limited the sample to service users and psychiatrists as the main stakeholders of the decision against or for antipsychotics, although caregivers and family members also play an important role in treatment planning in the inpatient setting. 29 A certain selection bias must be noted in the recruitment of patients. By asking the nurses about their assessment of the patient's ability to concentrate and selecting suitable candidates based on this, those with stronger positive symptoms such as inner restlessness or concentration problems were excluded. As the research group was located at one of the recruitment centres where half of the interviews took place, biases in the response behaviour of patients and medical colleagues can also not be ruled out. In addition, we showed subjects study results that should find a place in future DA but did not present an actual, already existing tool. Another restriction is that the coding of the data material was carried out by one investigator alone. However, regular interdisciplinary team meetings were held to ensure intersubjectivity in the key steps of the data analysis, such as setting up the coding scheme and eliminating ambiguities.

| CONCLUSION
Although patient DAs are becoming increasingly popular in the field of schizophrenia treatment, there is still the need to develop and test new tools that address the complexity of the decision on antipsychotic medication and eliminate hindering factors responsible for the poor implementation of SDM in psychiatry, especially for people with SMI. For this reason, time constraints as well as cognitive deficits, amongst others, should be considered. Nevertheless, the aim must be to provide those involved in the decision-making process with the highest possible data quality to enable a personalised, wellinformed decision. Consulting the main stakeholders involved in this decisional process, we gained important insights to develop an evidence-based DA with the potential to be highly useful to both parties, and also feasible in future implementation.

AUTHOR CONTRIBUTIONS
Katharina Müller planned the study, managed data acquisition, interviewed participants, planned the analysis, analysed the results and drafted the manuscript. Florian Schuster planned the study, managed data acquisition, interviewed participants, analysed the results and critically revised the manuscript. Prof. Silvia Krumm planned and supervised the project together with Prof. Johannes Hamann with her expertise as a qualitative researcher critically revised the manuscript. Prof. Stefan Leucht planned the study, managed data acquisition and critically revised the manuscript. Spyridon Siafis planned the study and critically revised the manuscript. Stephan Heres planned the study, managed data acquisition and critically revised the manuscript. Peter Brieger planned the study, managed data acquisition and critically revised the manuscript. Johannes Hamann planned the study, managed data acquisition, supervised the whole process of data acquisition and analysis as a qualitative expert, analysed the results and critically revised the manuscript.