Service Users’ Priorities and Preferences for Treatment of Psychosis: A User-Led Delphi Study

Recent U.K. government initiatives affirm that the perspectives and preferences of service users should play an important part in mental health decision making (1,2). However, people seeking help for psychosis- or schizophrenia-associated difficulties may find that their personal priorities and preferences differ from those of clinicians or that these factors are not adequately considered (3).

Clinicians and researchers in this field have primarily focused on the positive symptoms of psychosis (such as hearing voices or experiencing paranoia), and antipsychotic medication is often the only active treatment offered (4). Although antipsychotic medication is helpful for some, it is ineffective for others and is associated with significant adverse effects that may outweigh perceived benefits and lead to discontinuation (5,6). It has been suggested that there is an excessive reliance on antipsychotic medication in the treatment of schizophrenia and related disorders and that there is “little or no choice offered to service users who meet criteria for such diagnoses, despite policies that actively promote patient-led care, collaborative decision-making and provision of choice” (7).

Although there is a need, therefore, to consider service users’ priorities and choice of treatment, research assessing these factors is limited. A number of studies have compared the priorities of service users, caregivers, and mental health professionals (8,9) and have reported differences between these groups. Differences between individual service users have also been found in ratings of treatment priorities and preferences (10), although the most recent research in this area suggests it may be possible to identify areas of consensus among a large group of service users (11).

Several studies have used the Delphi method to seek expert consensus regarding treatment for psychosis or schizophrenia (12,13). However, such studies tend to include only professional participants, such as psychiatrists or psychologists, who may not prioritize elements of treatment that are important from a service user perspective. This study aimed to address this potential source of research bias by conducting a user-led Delphi method evaluation of user-defined treatment outcome priorities and treatment preferences.

Methods

Delphi method studies are consensus-seeking surveys of expert opinion, often conducted in three stages (12). We sought to recruit a sample of individuals with experience of receiving treatment for psychosis or schizophrenia (“experts by experience”) to complete a Delphi method survey examining treatment priorities and preferences. Appropriate ethical approvals were obtained from local university and National Health Service research ethics committees. During the first stage of the study, a literature search was conducted to identify treatment factors of potential importance to service users. Relevant research surveyed included quantitative studies of service users’ treatment priorities and preferences and qualitative studies of service users’ experiences of treatment and recovery. A diverse range of treatment factors was identified and listed until thematic sufficiency was achieved (that is, until no new items were identified).

These items (N=77) were listed in a numbered series of statements (for example, “It is most important for me to get help with depression”) alongside a Likert rating scale (1, essential, to 5, not required). This statement list was then sent to members of a local Service User Reference Group for consultation (N=6), each with experience of treatment for psychosis or schizophrenia. After group rating and discussion, 13 items were removed from the statement list, and a final statement list comprising 64 items was organized into four sections: current treatment priorities, long-term priorities, treatment preferences, and additional treatment preferences.

The second and third stages of the study were conducted online between January and August, 2012, and recruitment was achieved with the help of the Rethink Mental Illness charity, which advertised the study through e-mail and online social networks (for example, Facebook and Twitter). Online study material consisted of a participant information sheet and a consent form, along with the four-part statement list and accompanying Likert rating scale. Before they could progress to the survey section, participants were required to complete the consent form (online “tickbox” format) and were asked to provide an e-mail address to which we could send the next stage of the survey. All participants completed the online consent form. Among the 32 stage 2 participants, 50% were female and 59% were between the ages of 20 and 39. Among the 21 stage 3 participants, 45% were female and 55% were between the ages of 20 and 39. Race-ethnicity was not recorded for this study. [A table with more data on participant characteristics is included in an online data supplement to this report.]

At the prespecified end date for completion of stage 2 of the study, responses were analyzed by calculating percentages for the group (N=32 participants). Statements rated as essential or important by at least 80% of participants were included in a final statement list. Statements rated as essential or important by between 70% and 79% of participants were included in the stage 3 survey for rerating. Statements that did not meet either of these conditions were excluded. Participants were then e-mailed a direct link to the stage 3 survey Web page. The same rating criteria as above were used to analyze stage 3 responses (N=21 participants).

Results

Sixty-four treatment-related statements were generated during stage 1. During stage 2 of the study, 15 of these items were rated essential or important by at least 80% of participants and included in the final statement set. Twelve statements were rated essential or important by 70%−79% of participants and rerated at stage 3, after which two additional statements were included in the final statement set (Table 1). [Tables presenting participants’ ratings for these statements and for statements not included in the final set are included in the online data supplement.]

Strong consensus was noted for particular treatment factors in a number of areas. The most frequently endorsed current treatment priorities included paranoia, stress and anxiety, confusion, and concentration-memory. Long-term priorities included improved understanding (of mental health), coping ability, emotional well-being, and staying out of the hospital. Frequently endorsed treatment preferences included individualized care and collaborative decision making, greater information and treatment choice, privacy (when meeting with mental health staff), and age-appropriate care.

Discussion and conclusions

There are few existing studies with which to compare our findings, and it is difficult to compare these findings directly because methodologies in this topic area vary. Nonetheless it is worth highlighting a number of general similarities. Rosenheck and colleagues (10) reported that among a large sample (N=1,200) of individuals with schizophrenia diagnoses, the most common treatment priority was reducing confusion, which is echoed in this study and in a number of studies in this area (11,14). Fischer and colleagues (8) found that service users and their family members more often prioritized social support, housing, and medical services, compared with service providers. Shumway and colleagues (9) found that across groups of service providers, users, policy makers, and family members, participants valued social and functional improvements more than reductions in psychotic symptoms. Although our participants prioritized social and functional improvements less clearly, they prioritized treatment of psychotic symptoms less frequently than more general psychological and emotional improvements (such as reduced confusion and stress and improved understanding).

The clearest limitation of this study was our small sample, because analysis of data drawn from a larger sample may have enabled identification of differences between groups of participants. In addition, online social network–based recruitment may have excluded some potential participants. People with psychosis or schizophrenia may be economically disadvantaged (15) and may not have private access to the Internet, whereas older people may be less likely to access social networking media in general. It is also possible that some people in our target population may not have been comfortable participating in this kind of study over the Internet. The option to request paper copies of study materials was offered in our advertising information to reduce the risk of excluding potential participants, but none were requested. The validity of our results is also limited by the inclusion of several participants who identified as having primary diagnoses other than psychosis or schizophrenia (for example, bipolar disorder). Control of inclusion was limited by the open-access nature of our recruitment strategy. This could be addressed in similar studies conducted online with more effectively targeted recruitment strategies. It may also have been desirable to collect more detailed information about participants’ current difficulties and self-defined recovery status, because treatment priorities may change over time.

Our results identified a number of commonly valued treatment priorities, suggesting these as important areas for clinical attention (for example, confusion, anxiety, and stress). The most highly valued treatment preferences we identified reflect a desire for more information, choice, and collaboration in treatment decision making, suggesting that participants are not routinely satisfied in these areas. It is also worth noting that among the survey items most frequently endorsed as unnecessary, the use of medical terminology and attending appointments at mental health service premises were identified, suggesting that some aspects of routine psychiatric care may not be considered helpful by service users.

The relatively low levels of group consensus for many elements of treatment reinforce the need to evaluate the individual’s personal priorities in clinical practice. It is reasonable to suggest that greater concordance between an individual’s priorities and preferences and the treatment that he or she is offered will enhance continued engagement. As Bailey and colleagues (6) suggested in a recent article addressing the risks of general medical illness among people with serious mental health problems, especially those prescribed antipsychotic medication, “it is perhaps now worth reflecting on what might most trouble the young person facing you in the consulting room—the weight gain or the reduction in mental health symptoms.”