A pilot randomized controlled trial demonstrated that the CommonGround approach improved patient and prescriber communication when the personal medicine report, which describes individuals’ key wellness behaviors (such as painting and exercising) other than medication therapy, was referenced in the medical consultation (
3). Other studies have also reported potential benefits of the CommonGround approach based on patients’ subjective perceptions and self-reported outcomes (such as reduced concerns about use of mental health medication itself, concerns about adverse effects of medications, concerns about symptoms [hallucinations, disturbing thoughts, feeling nervous or anxious, good sleep, and so on], and improved recovery attitudes) (
4–
6). However, no existing trials outside the United States have examined the effects of a shared decision-making system that incorporates peer support specialists in a decision support center and a recovery-oriented decision aid. In addition, past studies of CommonGround have not assessed the relationship between users and doctors and have not assessed users’ perception of communication with doctors. Stein and colleagues (
7) also noted the lack of evidence for the impact of the CommonGround approach on clinical outcomes. Therefore, based on the CommonGround approach, we developed a Japanese-language shared decision-making tool called SHARE (Support for Hope And REcovery) and a comprehensive shared decision-making system comprising SHARE and a peer-led decision support center in a Japanese setting and conducted an exploratory pilot randomized controlled trial. The purpose of the study was to assess the effects of this system on implementation of shared decision making in medical consultations, the relationship and communication between users and doctors (coprimary outcomes), and other patient-level outcomes.
Methods
We conducted an individual-level randomized controlled trial with a parallel group design and 1:1 allocation ratio at two outpatient sites (one outpatient psychiatric clinic and one psychiatric hospital) in Tokyo over six months of follow-up. Both sites offered outpatient services, psychiatric day care, and visiting nurse services. [Further details about psychiatric day care and visiting nurse services in Japan are available in an online supplement to this report.] When patients who regularly used psychiatric day care or visiting nurse services arrived for their appointment with their psychiatrist, they first used SHARE with a peer support specialist in the decision support center, after which they were expected to discuss their treatment with a doctor during the medical consultation. We evaluated the intervention at process, proximal outcome, and distal outcome levels. The study was conducted between July 2014 and March 2016.
We included patients who regularly received medical care from one of the four participating doctors at the two sites, who received services from case managers in either a psychiatric day care or visiting nurse program, and who were age 20 years or older. We excluded patients with a primary ICD-10 diagnosis of substance abuse, dementia, or neurotic disorder. Written informed consent was obtained from all participants. This study was approved by the Research Ethics Committee at the National Center of Neurology and Psychiatry (A2014–001). A total of 111 patients were eligible, and 56 were enrolled. Participants were randomly allocated to either the shared decision-making system (intervention) group or a treatment as usual (control) group. The 53 participants who completed 6-month follow-up assessments were included in the final analyses. [Further details about the recruitment process, enrollment process, and random allocation to study groups are available in the online supplement.]
The intervention setting included four participating doctors who had a supervisor role in a psychiatric day care or visiting nurse program. Three of them worked in the psychiatric hospital. There were three peer support specialists in the psychiatric hospital and one in the psychiatric clinic, and all four participated in this project. Two peer support specialists had a primary diagnosis of schizophrenia, one had depression, and one had anxiety disorder. [Further details about the characteristics of the doctors and peer support specialists are available in the online supplement.] Peer support specialists began their service when the study was launched, and they received at least two training sessions prior to launch of the study.
The primary aim of SHARE is to facilitate the process in which patients and doctors make treatment decisions together, rather than to provide proper treatment information with the patient. The general framework of SHARE was developed focusing on recovery goals and self-management, like the CommonGround approach. Detailed contents and individual items where users rate their concerns about health status and medications in SHARE were developed by using Substance Abuse and Mental Health Services Administration guidelines (
8), after discussion among study team members. Another difference from CommonGround was that SHARE has self-rated items for problems with community life (e.g., home, job, or interpersonal relationships), which fit the Japanese context. We provided three two-day training sessions to peer support specialists, doctors, and case mangers during the study period. [Further details about the content of the training sessions are available in the
online supplement.]
Participants who were assigned to the intervention group visited decision support centers located in one section of the psychiatric day care center at each site. They first met with peer support specialists, who helped them use SHARE by sharing their own recovery experiences. SHARE guided patients in identifying personal values and treatment preferences. Before medical consultations, patients also used SHARE to rate their condition and concerns about community life. During medical consultations, doctors were strongly encouraged to confirm the patient’s personal recovery goals and the number of times the patient performed key behaviors identified in the program. Doctors then proceeded with their medical consultation on the basis of the participant’s condition and concerns as entered in SHARE. In addition, as part of shared decision making, doctors were expected to discuss treatment or self-management behaviors based on the participant’s individual personal recovery goals. At the end of the medical consultation, the patient and the doctor determined the treatment (such as medication type and timing/use of medication) or self-management behavior for follow-up at the next consultation, after which the doctor confirmed shared decision-making content with the patient and entered it into SHARE. [Further details on the process of the comprehensive shared decision-making system are available in the online supplement.]
We used the decision support center fidelity scale to rate the shared decision-making system (
9). The scale consisted of 13 items, with scores ranging from 13 to 65. Higher scores indicated closer adherence to the CommonGround approach. The mean±SD overall score of the two sites was 47.50±.71. Although scores for the domains of structural components, tool process, and peer support were relatively high, the score for direct service staff (case management involvement) was low. (Further details about the fidelity assessment are available in an
online supplement to this article.) Research team members performed the ratings, although they were not independent assessors trained for fidelity assessment.
Participants in the control group received the usual medical consultation with the same doctors as the intervention group. To avoid contamination between the arms as much as possible, control group participants were not allowed to use SHARE or the decision support centers during the study period.
We assessed whether the intervention facilitated implementation of shared decision making at the process level, whether it affected the patient-doctor relationship and communication at the proximal level, and whether it affected other patient-level outcomes at the distal level. We assessed a range of patient-level outcomes, including clinical outcomes (weight, symptoms, overall functioning, medication side effects and adherence, service satisfaction) and recovery-related outcomes (quality of life, recovery stage).
At the process level, we employed the SDM-18. SDM-18 and its coding system for a psychiatric setting were developed by Salyers and her colleagues (
10), based on the Elements of Informed Decision Making Scale, which has nine items identifying whether a clinical decision is present and assessing quality of the clinical decision in a medical consultation. This was used to evaluate the core components and process of shared decision making by scoring transcripts of conversations between patients and doctors. Transcripts were obtained by recording medical consultations during the study period. To assess the coprimary outcomes of patient-doctor relationship and communication, we used the Scale To Assess Therapeutic Relationships in Community Mental Health Care (STAR) (
11), the Interpersonal Processes of Care Survey Short Form (IPC) (
12), and the Japanese version of the Patient Activation Measure (PAM) (
13). STAR is used by both patients (STAR-Patient) and doctors (STAR-Clinician) to rate their relationship. Higher scores on these scales indicate better relationships and communication. To evaluate medication side effects, the four study doctors rated the severity of side effects among study participants by using the Drug-Induced Extrapyramidal Symptom Scale (DIEPSS). Higher DIEPSS scores indicate more severe medication side effects (
14). [Further details about the scales are available in the
online supplement.]
We conducted a linear mixed-effect model for longitudinal data with restricted maximum likelihood to assess the effects of the intervention. The model included group, time, and baseline outcome score. When the data were skewed, we used generalized linear models for longitudinal data.
Results
There were no significant differences in demographic characteristics between the intervention and control groups at baseline [see online supplement]. The mean±SD number of medical consultations differed significantly between the groups during the study period (intervention group, 9.85±3.41; control group, 7.89±2.33; t=2.45, df=51, p=.018), although there was no significant difference before the intervention period (intervention group, 4.54±1.73; control group, 4.07±1.71). The mean number of SHARE uses in the intervention group was 9.23±3.33 during the study period.
Table 1 presents outcome scores of the two groups at baseline and 6 months and results of the multivariate analyses. No statistical differences between the two groups in any outcome scores were observed at baseline. Analysis of the audio transcripts revealed a significantly higher SDM-18 score in the intervention group versus the control group (B=1.17, p<.001). We also found significant differences in scores on STAR-Patient, in which patients assess their relationship with their doctor. Specifically, the intervention group had higher scores on the positive collaboration subscale (
B=1.31, p=.048) and positive clinician input subscale (
B=.72, p=.027), compared with the control group. In terms of communication, intervention group participants had higher positive scores on the IPC communication subscale than individuals in the control group (
B=1.54, p=.005).
The only other outcome measure to yield a significant difference between the two groups over time was DIEPSS. At baseline, the DIEPSS score of the intervention group was higher than that of the control group, but this difference was not statistically significant. At follow-up, however, the intervention group had a significantly lower DIEPSS score than the control group (B=–.79, p=.023).
Discussion
This study examined the impact of a comprehensive shared decision-making system, based on the CommonGround approach, that incorporated peer support with SHARE, a recovery-oriented shared decision-making tool developed in Japan. This pilot trial found the system had significant effects on the implementation of shared decision making at the process level and on some relationship and communication outcomes at the proximal level. However, we did not observe any significant effects on any patient-level outcomes at the distal level except in the area of medication side effects. Although the results of this pilot study are not definitive, the findings suggest directions for future studies, particularly in the area of selecting outcomes of comprehensive shared decision-making systems.
In general, based on the fidelity scale and the SDM-18 results from medical consultation transcripts, the intervention appeared to promote shared decision making. Changes in the medical consultation process associated with the intervention appear to have had a positive impact on some coprimary outcomes. In the intervention group, participants’ perceptions of their relationship and communication with their doctors improved, compared with the control group’s perceptions, particularly as reflected in responses to STAR-Patient and IPC subscale questions about the positive aspects of relationship and person-centered communication. Although perceptions about relationship and communication may be intermediate variables, changing these variables is important in an actual care setting because they are predictors for other clinical outcomes (
15). In this context, our comprehensive shared decision-making system may foster behavioral changes among doctors, leading patients to view the relationship and communication with their doctors as reliable.
We found no significant impact of the intervention on doctors’ perceptions of their relationship with participants. Increased attention to individual personal values may not be associated with doctors’ perceptions of their relationship with their patients, so doctors may have positively changed their communication style without recognizing an improved relationship with their patients. This effect may be attributed to the comprehensive shared decision-making system strongly facilitating the confirmation of recovery goals and discussion of treatment by doctors with their patients regardless of the quality of the patient-doctor relationship.
In terms of clinical outcomes, we found a significant reduction in adverse effects associated with medication. However, the effect of the intervention on this outcome is not conclusive, because the DIEPSS score was rated by doctors who knew the group allocation, and the DIEPSS score in the intervention group was relatively high at baseline. There are two potential reasons for the lack of observed improvement in most of the patient-level outcomes. First, improvements in some outcomes (such as recovery-related scales) often require case manager involvement, and, as the fidelity score indicates, a weakness in our system was the lack of systematic cooperation between case managers and doctors/peer support specialists. Second, a six-month follow-up period may be too short to identify significant changes in clinical outcomes. As past studies have consistently emphasized, enhancement of the relationship between service providers and patients may lead to improved outcomes in the long term (
15).
This study had several limitations. First, it may have lacked statistical power because of the relatively small sample size. Second, symptoms, functioning, and side effects were assessed by doctors who were not blind to participants’ study groups. Third, the evaluation of SDM-18 used a mean score obtained from several medical consultations, regardless of the level of decision complexity. Future studies should incorporate more in-depth analysis of the effects of shared decision-making content. Fourth, although control group participants could not use the decision support centers and SHARE, they received some services from the doctors, peer support specialists, and case managers who attended the training sessions that were a part of this study. This study assessed only the effects of a decision support center, not the combination of training for performing shared decision making and decision support. In addition, scale scores (such as STAR, IPC, and SDM-18) in the control group were slightly lower at the six-month follow-up point than at baseline. Because of the study design, it is not possible to rule out provider bias caused by medical consultations being performed by doctors who were aware of the group allocation. A cluster randomization design at the organization level would be suitable for examining the effects of a comprehensive shared decision-making system. Future studies should address these issues to produce high-quality evidence on the clinical and economic benefits of a peer-led, recovery-oriented shared decision-making system.