Both voluntarily and involuntarily admitted service users report high levels of perceived coercion during the admission process and over the course of their admission (
1,
2). Our understanding of how individuals view their involuntary admission to a hospital and the use of coercive measures is growing. Studies have shown that between 39% and 72% of service users report that their involuntary admission was necessary (
3,
4). Service users express conflicting views of their involuntary admission, with some describing it as a positive experience or a “necessary emergency break” and others as a negative “unnecessary overreaction”; others are ambivalent or describe it as a “practice in need of improvement” (
5–
7). Service users find the use of physical coercion dehumanizing. It has proved difficult to predict who is more likely to experience physical coercion.
The study of service user perspectives is complex, and there is evidence that the background of the individual who formulates and asks the questions can influence the information elicited. A systematic review of service users’ perspectives on electroconvulsive therapy found that when service users led the investigation, lower rates of perceived benefit were reported than in studies that were led by clinician-researchers (
8). It has been noted that “coercion is in the eye of the beholder” and that the background of the researcher may bias information provided to him or her about perceived coercion (
9). In contrast, other studies have found no difference in the level of perceived coercion, discrimination, or satisfaction with services reported by service users when study interviews are conducted by service user–researchers or by clinician-researchers or staff members (
10–
12). In one study, service users provided more negative responses to service user–researchers; however, the study found no difference in overall satisfaction with services (
9).
We aimed to determine whether service users would provide different accounts of their perception of coercion experienced on admission when interviews were conducted by service user–researchers or by clinicians. This study had three unique aspects: it was a randomized study, the comparison was between service user–researchers and clinicians, and we included both voluntarily and involuntarily admitted service users. We also sought to determine whether there was a difference in the levels of procedural justice, perceived pressures, satisfaction with services, and the proportion of individuals consenting to participate in the study when interviews were conducted by service user–researchers or by clinicians.
Methods
Setting
This study was conducted in three hospitals: St. Vincent’s University Hospital, which is a general hospital with a psychiatric ward; Newcastle Hospital, and St. John of God Hospital. These three hospitals receive admissions from three community mental health services in Ireland: Dublin South East mental health service and the Newcastle Mental Health Services in County Wicklow and the Cluain Mhuire Mental Health Service in County Dublin. These services cover a combined catchment area population of approximately 390,000. In addition to treating service users from the local catchment area, St. John of God Hospital is an independent psychiatric hospital and receives referrals on a national basis. We included service users who were admitted privately to St. John of God Hospital.
Participants
All individuals age 18 years and older who were involuntarily admitted to St. John of God Hospital between May 1, 2010, and June 30, 2011, St. Vincent’s University Hospital between August 1, 2010, and June 30, 2011, and Newcastle Hospital between November 1, 2010, and June 30, 2011, were eligible for the study. In order to have a comparable number of voluntarily admitted participants, the next voluntarily admitted service user after each involuntary admission was selected. We also included individuals who were admitted voluntarily but were subsequently detained involuntarily during the admission. According to the Irish Mental Health Act of 2001, individuals with a sole diagnosis of a personality disorder or substance misuse cannot be admitted involuntarily. Therefore, to have comparable samples, we excluded voluntarily admitted service users with either a sole diagnosis of a personality disorder or substance misuse. We also excluded individuals with dementia and intellectual disabilities; those with a first episode of psychosis were excluded because they were involved in another study protocol.
Ethics approval
Written informed consent was obtained from all study participants, and ethics approval was granted at each of the three study centers. We did not offer compensation for participation in the interviews before discharge. We were not granted permission to obtain information about individuals who declined to participate in the study, and therefore we could not compare those who participated with those who declined.
Study design
Participants were randomly assigned to an interview with a service user–researcher or a clinician. Participants were interviewed once a discharge date was set, and interviews occurred in the days before discharge from the hospital. Each participant had a second interview with a clinician before discharge to confirm the diagnosis and measure the severity of symptoms. Participants were not blind to their randomization, because it was an important aspect of the study for participants to know that they were being interviewed by a service user–researcher or a clinician. All service user–researchers informed potential participants that they were a service user–researcher, and all clinicians informed them that they were a doctor or nurse. We did not accommodate requests by potential participants to be interviewed by the other type of interviewer. The interviewing clinician was not a member of the participant’s treatment team. Service user–researchers were involved in all stages of the study design, implementation, and dissemination of results.
To randomly assign participants, a list of the numbers 1 and 2 was generated in random sequence by a Web site (
www.random.org); 1 indicated a service user–researcher and 2 indicated a clinician. This list was generated and kept by an individual who was independent of the study and based in a separate building. The list was concealed from both the researchers and principal investigator.
We used postrandomization consent (Zelen’s design [
13]) because the delivery of consent is part of the difference between interviews conducted by service users and by clinicians. Had a service user–researcher or clinician elicited consent before randomization, this may have introduced bias because individuals may have refused consent if asked by a clinician but not by a service user–researcher, or vice versa. This procedure also allowed us to test a secondary outcome—that is, whether participation differed depending on whether consent was sought by a service user–researcher or a clinician.
Informed consent
Once a discharge date had been set, potential participants were informed by nursing staff on the ward that there was a researcher who would like to meet with them about a study. At this point, the potential participant was not informed whether the researcher was a service user–researcher or a clinician. If the potential participant agreed, a time was arranged for the meeting. Potential participants were provided with oral and written information about the study and were given an opportunity to ask questions and decide whether they wished to participate. There was no time constraint within which potential participants had to decide. Participants could withdraw consent at any stage, which was clearly stated in the letter of information and the consent form.
Instruments
We used the MacArthur Admission Experience Survey (MacArthur AES) to determine the level of perceived coercion, perceived pressures, and procedural justice experienced by an individual on admission to the hospital (
14). The primary outcome was level of perceived coercion, which is measured on a scale from 0 to 5, with higher scores indicating a higher level. The level of perceived coercion is obtained from five statements related to the individual’s experience of being admitted to the hospital. The five statements are: “I had more influence than anyone else on whether I came into hospital,” “I had a lot of control whether I went into hospital,” “I chose to come into hospital,” “I felt free to do what I wanted about coming into hospital,” and “It was my idea to come into hospital.” Answers of “true” are scored 0 and answers of “false” are scored 1. The level of perceived pressures is assessed with four yes-no questions and scored from 0 to 4, with higher scores indicating a higher level. Procedural justice reflects a person’s belief that others are acting out of genuine concern for him or her and that the person is being listened to and being treated respectfully and fairly. This is measured on a scale from 1 to 4, with higher scores indicating a lower level of experienced procedural justice (the person perceives that he or she was less fairly treated) (
15).
To evaluate the perceived necessity of the admission, participants were asked, “Do you think it was necessary for you to have been admitted [involuntarily or voluntarily–depending on legal status] to hospital?” We accepted “yes,” “no,” or “don’t know” as answers.
We used the Client Satisfaction Questionnaire (CSQ-8), a self-report, eight-item instrument, to measure satisfaction with services received (
16). The CSQ-8 uses a 4-point Likert scale (range 1–4), and a total score (range 8–32) is calculated from summing the scores on the eight items. There is no reliable or valid tool to assess whether an individual perceives the hospital admission to have been necessary. Therefore, we asked a closed question about whether the individual perceived the admission to have been necessary or unnecessary.
All individuals were assessed using the Structured Clinical Interview for DSM-IV by a trained clinician. This included the Global Assessment of Functioning, which ranges from 0–100, with 100 being the highest level of functioning (
17). Positive and negative symptom severity was measured with the Scale for Assessment of Positive Symptoms and the Scale for Assessment of Negative Symptoms (
18,
19).
We measured depressive symptoms with the Beck Depression Inventory, and categories of depressive symptoms were classified as minimal depression (score of 0–9), mild depression (score of 10–18), moderate depression (score of 19–29), and severe depression (score of 30–63) (
20). Insight was measured with the Birchwood Insight Scale, an eight-item scale on which scores range from 0 to 12, with higher scores indicating good insight (
21).
Interviewers
Six clinicians (five physicians and one nurse) from the local mental health services involved in the study and four service user–researchers from SOURCE (Source of User Research, Collaboration and Expertise) conducted the interviews. SOURCE is a research group staffed by a clinical researcher and a group of individuals with past or current experiences of mental health difficulties. The main focus of the group is to use research methods to capture the lived experiences and perspectives of individuals who have experienced mental health difficulties, with the objective of improving mental health services.
Before the start of the project, SOURCE had affiliations with only one of the hospitals involved in the study—St. John of God Hospital. To our knowledge, research involving service user–researchers had not been conducted at the other two hospitals before this project. None of the hospitals employed service user–researchers in professional roles before the start of this project.
Interrater reliability
All interviewers underwent training in the MacArthur AES before the study. Interrater reliability was obtained for all service user–researchers and clinician researchers via interviews with eight inpatients. We developed videos with actors for any interviewers who joined the study after the start date and for retraining. The interrater reliability (kappa) at baseline ranged from .77 to 1.00 for the perceived coercion component, from .77 to 1.00 for perceived pressures, and from .66 to 1.00 for procedural justice.
Statistical analysis
Data were entered into a MS Access database and exported to PASW, version 18, for analysis. For the primary outcome, our a priori power calculation indicated that 102 participants (51 in each arm) were required to detect a difference of 1 in the perceived coercion scale of the MacArthur AES (measured 0–5) with 90% power at the 5% significance level. The null hypothesis states that there is no difference in the information elicited from interviews conducted by service user–researchers and by clinicians. We used chi square tests to determine associations for dichotomous variables and t tests to compare means on continuous variables between the two groups. The Mann-Whitney test was used to determine differences in nonparametric data.
Ideally, the randomization should have controlled for known and unknown confounders between the groups, and the characteristics in both groups should have been similar. However, differences between the groups were noted; individuals interviewed by service user–researchers were younger and more likely to be male and to have more depressive symptoms. These differences may have occurred by chance or as a result of the method of postrandomization consent. To control for these factors, we used analysis of covariance (ANCOVA) for the MacArthur AES and CSQ-8 because the dependent variable was quantitative and there was a combination of qualitative and quantitative explanatory variables. When depressive symptoms were added to the ANCOVA model in which perceived coercion was the dependent variable, a test indicated that the data violated the assumption of equality of error variation. Therefore, for perceived coercion, depressive symptoms were examined separately as a potential confounder and as a dichotomous variable (absent or mild versus moderate or severe depressive symptoms). We used binary logistic regression analysis to control for age, gender, and depressive symptoms for the secondary outcome of the perceived necessity of the admission. [Tables presenting the results of the ANCOVA analysis and the binary logistic regression are available online as a data supplement to this article.]
Results
Description of the sample
A total of 231 service users were eligible for study participation, of whom 70% (N=161) were interviewed. [A flowchart showing study recruitment and reasons for nonparticipation is included in the online data supplement.] Most of those who were not interviewed were discharged before randomization or before they were able to provide consent (N=23), or they declined to participate (N=31). Of the 161 participants, 48% (N=77) were interviewed by service user–researchers, and 52% (N=84) by clinicians. Those interviewed by service user–researchers were more likely to be male, to be younger, and to have more severe depressive symptoms. However, in terms of legal status, diagnosis, clinical characteristics, and the use of coercion, the groups were similar (
Table 1). For the 161 patients in the full sample, the median duration of admission was 37 days (interquartile range 19–58 days)
Primary outcome: perceived coercion
Table 2 summarizes information elicited during the interviews by the service user–researchers and the clinicians. No differences were found by interviewer status in the reported level of perceived coercion on admission, either for individuals who were involuntarily admitted or for those who were voluntarily admitted. The findings were the same when the analysis controlled for age, gender, and depressive symptoms.
Secondary outcomes
No differences were found by interviewer status in the levels of procedural justice, perceived pressures experienced on admission, and satisfaction with services, either for individuals who were involuntarily admitted or for those who were voluntarily admitted. The findings were the same when the analyses controlled for age, gender, and depressive symptoms.
Among the 161 service users, 118 (73%) viewed their admission to the hospital as necessary. No differences were found by interviewer status. Five (3%) individuals reported that they did not know whether their admission had been necessary, and three (2%) did not provide an answer. Of the 78 service users admitted involuntarily, 54% (N=42) viewed their admission as necessary, and no difference was found by interviewer status. Of the 83 service users admitted voluntarily, 92% (N=76) viewed their admission as necessary, and no difference was found by interviewer status. Results were the same when the analyses controlled for age, gender, and depressive symptoms.
Service users were more likely to decline to participate in the study when consent was sought by a service user–researcher compared with a clinician (24% versus 8%, χ2=9.1, df=1, p=.003). Similarly, the decline rates were higher for service user–researchers when consent was sought from involuntarily admitted service users (31% versus 14%, χ2=3.8, df=1, p=.05) and those who were voluntarily admitted (14% versus 2%, χ2=4.65, df=1, p=.05).
Acknowledgments and disclosures
The project was partially funded by a grant from the Mental Health Commission in Ireland. Support was also provided by the Cluain Mhuire Mental Health Service, which employed a service user–researcher part time. The authors are grateful to the clinical directors of the mental health services for facilitating this study: Siobhan Barry, M.R.C.Psych., Bernadette Mangan, M.R.C.Psych., Anthony McCarthy, M.R.C.Psych., Freda O’Connell, M.R.C.Psych., and Justin Brophy, M.R.C.Psych. They also thank Daria Brennan, M.L.I.S., Ph.D., for managing the randomization and Darach Murphy for conducting interviews on behalf of SOURCE. They are grateful to the late Eadbhard O’Callaghan, D.Sc., F.R.C.Psych., for facilitating this study. The project would not have been possible without the administrative skills of Karen Cobbe of SOURCE.
Mr. Shannon’s work toward a doctorate in the area of seclusion is funded by the Mental Health Commission. The other authors report no competing interests.