Background
People with serious mental illness, such as schizophrenia, face decisions to continue, reduce or discontinue antipsychotic medications daily. Many fail to formally consult their psychiatrists as part of this decision process. Few prescribers explicitly address this issue with patients during the clinical encounter [
1‐
3]. This situation illuminates two important points to consider in clinical practice. First, it raises concerns about the effectiveness of antipsychotic medications as a primary approach for preventing relapse in psychosis [
4]. Second, people with serious mental illness make treatment decisions about their use of medications independently, often without enough information gained in consultation with psychiatrists and other mental health providers about the likely risks and benefits [
5‐
7].
Many studies have addressed the benefits and risks of continuing antipsychotics [
8‐
13] as well as the effects of nonadherence to antipsychotic medications and its antecedents, factors, and outcomes [
3,
14,
15]. However, few have focused on the decision-making process in psychiatric consultation or on how to present and discuss the relevant options [
16‐
21]. Existing studies suggest that the use of a shared decision making (SDM) process in psychiatric medication consultations is preferred for making mental health care decisions characterized by uncertainty and when more than one reasonable option is available [
22]. Indeed, a recent review by the Australian National Mental Health Consumer Carer Forum on the topic of psychiatric medications concluded, among other things, that a SDM approach encourages the empowered use of psychiatric medications as part of the recovery process and positions patients as active agents in their own recovery process [
23].
SDM is a health communication model that helps patients and clinicians make treatment decisions jointly. It gives both parties a framework and legitimacy for the discussion [
24], focuses on the patient–clinician interaction, and encourages open dialogue in which both parties have expertise, which should lead to the choice of a consensus plan [
25‐
27]. From an ethical perspective, SDM embraces patients’ experiences, autonomy, and empowerment, which are aligned with personal recovery values and self-determination [
28,
29].
One promising approach to operationalizing SDM in the medical consultation is the use of an encounter decision aid (EDA) to help patients and clinicians discuss relevant treatment options, support patients in exploring what is clinically available, and incorporate patient preferences [
30,
31]. EDAs are evidence-based tools used before, during, or after a medical encounter to supplement the clinical consultation and to help patients make an informed, deliberate choice among treatment options [
32,
33]. EDAs differ from health education materials because they make explicit a specific decision [
34], and simplify the information about the patient’s condition and his or her options for treatment, screening, or diagnostic procedures. Previous research on the effects of EDA on various patient populations generally supports the potential positive effects of EDAs on patients’ medical decision-making: improving their knowledge regarding their treatment options and risk perceptions, reducing their level of decisional conflict, and decreasing the proportion of people who remain undecided [
31]. However, this review included only two studies of EDAs in mental health [
16,
35], both showing the effects of the EDA on uptake of psychoeducation [
16] or completion of psychotherapy [
35], not on antipsychotic medications adherence, which is strongly different from participation in psychotherapy or psychoeducation. Therefore, although Stacey et al. [
31] found strong evidence when the patient choice is about whether to undergo surgery or treatment for heart failure or diabetes, their conclusions cannot be easily extended to patients with mental disorders.
Indeed, previous reviews focusing specifically on SDM in mental health have recommended using decision aids when discussing treatment options [
17,
36], but similar biases and methodological limitations in the included articles, as with the findings by Stacey et al. [
31], still exist. A recent review of SDM interventions in mental health by Zisman-Ilani et al. [
37] included, in addition to randomized control trials (RCT), non-RCT studies and conceptual articles. The authors included 31 articles, of which 12 have described decision aids (or EDAs); only seven (of these 12 articles) were research articles and described an evaluation of a decision aid. Results were mixed; in three studies that used a decision aid, without supplemental elements before or after introducing it (such as goal setting), there was at least one significant outcome related to improved health behaviors (i.e. adherence, service utilization) [
19,
38], mental health symptoms [
39], and SDM-related outcomes (i.e. involvement, knowledge, self-efficacy). One study [
40] reported no significant effect of the decision aid on patient outcomes. Four additional studies, that used a decision aid with supplemental elements, were found to have at least one positive significant effect on SDM-related outcomes (i.e. involvement, knowledge, self-efficacy) [
16,
18,
41], and one study showed improved engagement in preferred treatment (psychotherapy, not medications) [
35].
To help address the concerns from the limited studies available from these reviews, our purpose in this study was to develop a tool to help patients with psychosis who have stabilized, their carers, and clinicians explicitly address the daily dilemma of continuing, adjusting, or discontinuing antipsychotics during the mental health consultation. Specifically, our aims were: (1) to develop an EDA for antipsychotic medications decision-making that formally addresses the three options (continue, adjust, or stop); and, (2) to revise the tool based on a qualitative user-testing study with potential users that was designed to evaluate the clarity, usability, and potential barriers to and facilitators of future implementation of the new EDA in psychiatric consultations.
Methods
The EDA was developed according to the recommendations of the International Patient Decision Aid Standards (IPDAS) collaboration [
42], the IPDAS instrument (IPDASi) [
43], and the previous experience of developing 50 Option Grids™ decision aids. We used a community-based participatory research approach both in developing the EDA and during the evaluation of the proposed EDA by clinicians, carers and consumers [
44]. The study had two phases: (1) EDA development, and (2) user testing.
Encounter decision aid development
We chose a specific format for the EDA, the Option Grid decision aid for clinical encounters, that is short and that can be implemented in routine care during medical encounters [
45]. This format provides a simple one-page table with rows containing frequently asked questions by patients about their treatment options and the benefits, risks, and implications of differing decisions. The columns display the treatment options available for the health care decision in question: here, continuing, adjusting, or discontinuing antipsychotic medications.
Creating an Option Grid decision aid of this nature requires a process that brings together the most current scientific evidence and filters that data through a team of health care experts. This team, which most often consists of service users, clinicians, and researchers, rigorously synthesizes evidence from systematic reviews and the most reputable clinical guidelines in an editorial process, hence, the name “editorial research team”.
Our multidisciplinary editorial research team included people with lived experience of serious mental illness, psychiatrists, psychologists, a pharmacist, and decision scientists. The team developed the EDA between March 2015 and March 2016 within a larger project to create technical assistance materials for use in first-episode-psychosis programs. As an editorial research team, we were not only involved in the development and user-testing of the tool but also highly involved in the process of summarizing the evidence in an editorial process, especially when we had to translate from primary or secondary research studies to words and sentences that are accessible to patients and other end-users of the tool. Our team met regularly to discuss development of the EDA and the user-testing strategy and employed an iterative process [
46] to create a final version for the study.
The team focused on the often-neglected decision node of what to do with antipsychotic medication management after a person has experienced symptom relief. After opting to focus on medication decisions after initial stabilization, the team identified a series of frequently asked questions that can typify an informed clinical interaction following stabilization from the first episode. The questions that were ultimately selected are portrayed in the rows of the EDA and address various aspects of the three decisions that service users can make following stabilization: Continuing on the medication regime that was used in their initial treatment; Adjusting their antipsychotic medications and/or adding medication for side effects; or Stopping the medications. Following the identification of these questions, the research editorial team reviewed research literature to determine the most accurate responses to the questions. Systematic reviews, treatment guidelines, meta-analyses, and individual studies were included. Based on this process, draft answers to each of the frequently asked questions were reviewed by the team for accuracy and clarity. Each cell in the EDA matrix is linked to an evidence document (Additional file
1) that references the research literature that supports the answers. When no research base is available for a question, the opinion of the editorial research team is cited.
User-testing study
Once an Option Grid decision aid is constructed, it is put through a series of user tests and is refined until a final version is agreed upon. This document is then used during clinical encounters as a scaffolding of information that prompts a deeper conversation about individuals’ specific desires for particular health care decisions. Ultimately, an Option Grid is meant not to simply be a stand-alone summary but a tool used to promote collaboration and deliberation.
Our user-testing included individual semistructured interviews with people from six stakeholder groups: persons with first and long-term psychosis, family members of patients, psychiatrists, mental health counselors, and administrative staff at mental health clinics. Our team (YZI, DS, MM, GE) developed an interview guide with three broad sections: (1) general impression of the EDA and its purpose; (2) content, wording, accuracy, and visual appearance; and (3) perspectives on usability and implementation in psychiatric settings (Table
1). We adapted the interview guide to fit the six groups of interviewees.
Table 1
The semistructured interview guide for patients
General impression and purpose: |
• Can you describe in your own words the purpose of the grid? |
• What is your overall impression of the grid? |
• In your opinion, who is the grid intended for? |
• How helpful do you think such a tool would be for you for decisions about the use of antipsychotic medication? |
Content, wording, and appearance: |
• Is there anything you’d add, delete, or change about the opening instructions of the DA? |
• Is there anything you’d add, delete, or change about the information in the row ‘What does this involve?’ † |
• Are there other questions that you feel would be important to include? |
Usability and potential contribution to the psychiatric consultation: |
• In your opinion, how would this tool best be incorporated into aclinic’s workflow? (prompts: Before the visit at home? Before visit by clinic staff? During visit? After visit?) |
• Do you think there is room for such a tool in psychiatric care? |
• Would you prefer a static version (paper and pencil) or an interactive paper (for example on a tablet or mobile device)? |
• Would you like the clinician to use this tool as part of your meeting/consultation? (Why?) |
• How do you think patients will respond to such a tool? |
• How do you think clinicians will respond to such a tool? |
†Four more questions like this were asked for each of the additional rows in the decision aid. |
Recruitment and procedures
Our team advertised the study and recruited potential interviewees with the help of colleagues at two community mental health clinics in rural New Hampshire (patients with long-term illness, psychiatrists, mental health counselors, and administrators), a mental health clinic for young adults with first-episode psychosis in the New York City area affiliated with Mental Health America (patients with first-episode psychosis), and a support group for family members run by the National Alliance on Mental Illness (family members of people with a history of psychosis). We chose these sites based on the relationships between some members of the editorial research team and these settings; some members of the team helped open doors and access the specific clinics and sites. However, these members had no direct interaction with the study participants. In addition, the lead author (YZI) conducted all interviews and had no prior relationships with the clinics or the participants.
Using non-probability purposive sampling, we recruited a convenience sample of interviewees based on population characteristics and on the study’s objective to achieve a final sample of diverse participant groups [
47‐
49]. Specifically, we used expert sampling because our research requires first-hand experts in relation to antipsychotics (prescribing, consulting, using, and caring for users) to capture knowledge rooted in a particular form of expertise. It is common to use this purposive sampling technique in the early stages of a research process, when the researchers are seeking to become better informed about the topic at hand before embarking on a study. Eligibility criteria varied by group category. Patients (first-episode psychosis and long-term illness) had experienced at least one episode of psychosis for which they took antipsychotic medications, were age 18 years or older, and had a chart diagnosis of schizophrenia, schizoaffective disorder, schizophreniform disorder, delusional disorder, or psychosis not otherwise specified. Interviewees were excluded if they could not speak, write, or read English or could not participate in an interview because their symptoms of psychosis were not stabilized by the time of the interview. Patient interviews were conducted face-to-face. Interviews with family members, clinicians, and administrative staff were conducted face-to-face or by telephone.
We obtained verbal consent from all respondents at the time of the interviews and written consent for face-to-face interviews. All interviewees were assured that participation was voluntary and that all information collected would be confidential and used for research purposes only. Interviewees were offered a $20 honorarium for their participation after the face-to-face interviews or $10 for telephone interviews (via mail). Each interview lasted about 30 min. The Committee for the Protection of Human Subjects at Dartmouth College approved the study.
Analysis
We used an iterative coding and analysis process with constant comparison [
50,
51]. Two authors (YZI, KG) coded the interviews to ensure intercoder reliability and conferred weekly [
52]. Coders resolved discrepancies through discussion. For example, codes were combined or split after conferring with the other coder. After we reached agreement on the themes, YZI wrote an evolving qualitative memo, which was reviewed by KG biweekly. Other authors (GE and DS) commented on ongoing drafts of the results in both content and presentation. These versions of the tables and figures were presented to the entire editorial research team to gather their feedback.
Discussion
The newly developed EDA was positively evaluated by the different stakeholders who participated in user-testing. The EDA was perceived as usable, context-appropriate, and potentially feasible in psychiatric consultations. Some respondents, particularly patients and family members, were concerned that the early version of the EDA strongly advocated against antipsychotic medications, which led to revisions to the final version. Interviewees encouraged us to present the information neutrally. This recommendation is aligned with previous work on the role of persuasive approaches in medical decision-making and the need to determine whether medical decisions are the result of a doctor–patient partnership or of persuasive tactics based. on power and hierarchical relationships [
54]. Persuasive theories are one subset of health communication theories; they can be applied at many levels including intrapersonal, interpersonal, organizational, and mass communication [
55]. Accordingly, persuasion can play a critical role in doctor–patient communication, as it is often the intent of a practitioner to shape, reinforce, or change a patient’s behavior [
56], especially when discussing medication and treatment adherence [
55], and even more in psychiatric settings, where patients are often questioned about their ability and capability to make treatment decisions [
17,
57].
Our respondents raised the issue of patients’ ability to make decisions. This is particularly relevant to patients with schizophrenia spectrum disorders and serious mental illness, who may lack awareness of their illness [
58‐
60]. However, findings from the last decade about SDM in mental health provide evidence that people with mental conditions, including schizophrenia [
16], can effectively participate in SDM with their health provider [
37] and will make medication decisions independently without clinical guidance [
23]. Therefore, because people with long-term illness provided most of the revisions to the EDA’s wording and content, this may demonstrate their ability and capacity to engage in these ideas as well as anyone else, challenging perhaps the assumption that they lack capacity.
Family members were guarded in their initial reactions to the EDA, perhaps fearing that discussion of stopping medication might suggest an alternative that would harm their family member. Another explanation might be related to a sense of family burden in caring for a loved one with a mental illness [
61‐
63], and a fear of a sudden increase in this burden may explain family members’ concerns about the EDA. Because family members play a major role in a triadic decision-making in mental health care [
64,
65], their concerns are valid and should be addressed to facilitate and support the discussion on medication management. The present EDA and future SDM tools should include family members as targeted end-users, along with patients and clinicians, to support implementation and increase buy-in.
Our study has several strengths: First, we presented a heterogeneous overall sample representing unique access to the different stakeholder groups that created an important diversity of participant groups in the final sample. Second, we have described the development of a tool using a participatory approach that considered participants “active consultants” rather than “research subjects.” This research approach allows people with severe mental illness / psychotic illness, who may be largely excluded from decisions about their treatment, to become actively involved in the research and provide important feedback. Therefore, not only may SDM and tools such as decision aids help shift the power balance in medical encounters to help address coercive treatment, but also the participatory approach may help patients feel valued for their input, which, in turn, may further empower them. These strengths allowed us to explore and compare the different perceptions and perspectives of people with lived experiences of antipsychotic medication management. Yet, several limitations should be noted: First, the relatively small sample in each group may affect the generalizability of the results. Second, the intention of this article is not to venture into making suggestions about clinical impact, as this was a small, qualitative user-testing. Future field-testing research in various settings, such as clinics, hospitals, and community programs, is needed. In addition, future potential research would require a further study with much larger samples to test the effectiveness of the EDA and should include clinical and more psychosocial impact measurements.