The Recovery Group Project: Development of an Intervention Led Jointly by Peer and Professional Counselors

Mental health care continues to move toward recovery-oriented, consumer-directed services (1–4). Central to this transformation are services that promote self-determination and mental health care–related activation, hope, and belief in recovery; are based on respect; and recognize that individuals have different needs, preferences, goals, and recovery pathways (1).

Few conventional mental health services can be characterized as meeting these objectives; thus new approaches are being developed. Peer-provided services represent one promising approach (5), with recovery supported by both providers and consumers, including through processes of experiential learning, natural supports, and peer role models (6). Recent systematic reviews suggest that peer-provided services can be as effective as conventional services for many outcomes (7,8); however, high-quality evidence remains limited, and randomized controlled trials (RCTs) are needed (8).

We report results of a project to develop and evaluate a group recovery intervention led jointly by peer and professional counselors. The intervention was based on the consumer-developed self-help recovery workbook Pathways to Recovery: A Strengths Recovery Self-Help Workbook (9). The goal was to promote recovery by providing a safe, supportive community environment and an atmosphere of acceptance; by fostering self-worth, dignity, and respect; and by facilitating learning from others with relevant knowledge and experience (10).

Methods

We first developed a facilitator’s manual (available at www.kpchr.org/RGP), although a different manual has since been published (11), and conducted a six-week feasibility study (cohort 1). We then conducted a small RCT of a ten-week intervention (cohorts 2 and 3), using a delayed-intervention control group that received the intervention immediately after the first follow-up assessment point (after completion of the intervention by participants in the intervention group). Both intervention and delayed-intervention control groups received the same assessments; the only difference between the two groups was the timing of the intervention. With the findings from these efforts, we obtained funding for a study to examine optimal intervention length (cohorts 4 and 5). All phases were approved and monitored by the Kaiser Permanent Northwest (KPNW) Institutional Review Board.

Intervention content

Groups were cofacilitated by a professional mental health counselor and a study-trained peer counselor. The leaders’ manual and group materials were informed by key features of successful adult learning (12,13) and are available from the lead author (CAG). Participants received the Pathways to Recovery workbook and participated in six weekly two-hour sessions; healthy snacks were provided. Weekly discussion topics followed the structure of the workbook and included opportunities to share recovery-related experiences, complete exercises to reinforce concepts, and engage in group problem solving. Workbook chapters and topics covered are shown in the box on this page. For cohorts that met for longer periods, workbook materials were spread out over the extended number of sessions to allow in-depth discussion of each chapter.

Group leaders and training

Two peer counselors were recruited through participating community mental health centers (CMHCs); the third was a project consultant and mental health care consumer activist. One professional counselor was a study staff member, and two were employed by participating CMHCs. Research staff provided eight hours of initial training for using the workbook and manual. Additional supervision and training were provided in weekly meetings in which participant evaluations and ideas for improving the intervention and manual were discussed. Peer and professional counselors were expected to function as a team; the one difference in roles was the expectation that the peer counselor brought relevant personal experience to the role and served as a role model for participants.

Study settings, recruitment, and retention

For cohort 1 (feasibility pilot conducted in six weekly sessions) we distributed flyers to participating CMHCs and clinicians and placed a brief story about the project in the local National Alliance on Mental Illness newsletter. We planned to enroll 20–25 participants, received inquiries from 34 individuals, and successfully enrolled 22. Of those enrolled, 21 (95%) completed postintervention follow-up interviews and 20 (91%) completed follow-up questionnaires. Nineteen attended the first group session, and an average of about 15 people attended remaining sessions (range 13–16 attendees).

Cohorts 2 and 3 were recruited for the RCT from KPNW, a not-for-profit integrated health care system serving about 480,000 members in Oregon and Washington State. Participants were assigned to either the intervention group (cohort 2) or the delayed-intervention control group (cohort 3).

We used health plan databases to identify individuals ages 18–65 with diagnoses of bipolar disorder, schizophrenia, or schizoaffective disorder who lived in the local area (14). [Recruitment outcomes are presented in a CONSORT diagram, available online as a data supplement to this article.] We mailed invitation letters with return postcards and followed up with telephone calls. Recruitment was stopped after 38 eligible persons agreed to attend group orientation meetings. Of these, 32 were enrolled and randomly assigned to a cohort, although two persons dropped out before the study began and were not included in analyses. For cohort 2, attendance over the ten sessions ranged from zero to ten sessions (mean±SD sessions attended=6.3±3.8). For cohort 3, attendance also ranged from zero to ten sessions, with 6.3±3.1 sessions attended. Twenty-eight participants completed questionnaires at follow-up 1 (93%); 30 participants (100%) completed questionnaires at follow-up 2.

Participants in cohorts 4 and 5 (pre-post design) were recruited by distributing flyers at two CMHCs, which resulted in 36 calls from potential participants. Of these, three refused to participate, three called after recruitment closed, and 30 were eligible and enrolled in the study (23 women and seven men). Participants were informed that the group would last from ten to 18 weeks and that they would determine the length and pace. Cohort 4 continued for 17 weeks, and attendance ranged from zero to 16 sessions (with 10.43±4.9 sessions attended). Cohort 5 continued for 18 weeks, and attendance ranged from zero to 15 sessions, with 10.11±6.25 sessions attended.

Data collection

Data were collected from 2005 through 2007. Assessments included semistructured qualitative interviews and paper-and-pencil questionnaires at baseline and follow-ups. For cohort 1 (feasibility study), data were collected at baseline and after completion of the intervention (95% follow-up rate). For cohorts 2 and 3 (the RCT), assessments were completed at baseline, about three months later (after the intervention group completed the intervention; 93% follow-up), and six months later (after the delayed-intervention control group completed the intervention; final follow-up=100%). For cohorts 4 and 5 (pre-post design), assessments were completed at baseline and ten months after enrollment (final follow-up=83%, N=25). Participants completed evaluation forms for each session and overall evaluations after final sessions.

Assessments and measures

Qualitative assessments and analyses.

Baseline interviews were conducted in person and averaged 30 minutes. Interviewers entered responses to questions on laptop computers, recording salient quotes verbatim. Interview topics included physical and mental health and feelings about health care, psychiatric diagnoses and feelings about diagnoses, problems with alcohol and other substances, mental health–related symptoms, quality of life and happiness, desired changes and life improvements, goals for the year, barriers to attaining goals and help needed to achieve them, feelings about the cofacilitator arrangement, and expectations for the group.

Follow-up interviews were conducted by telephone. Participants were asked about changes in their lives since the previous interview, particularly quality of life and happiness, and progress with goals. After group participation, we asked participants about their perceptions of the program, whether participating helped or did not help them make changes and pursue long-term goals, what was most and least helpful, whether they would recommend the group to others, suggestions for improvement, feelings about the cofacilitator arrangement, and reactions to the workbook. Those who did not attend the group or who dropped out were asked to describe their reasons for doing so.

The team read interview responses and generated a list of preliminary descriptive codes. Codes were tested, refined, and defined by having team members code the same interview responses and discuss agreements and disagreements. Responses were coded by using Atlas.ti (15), and queries were generated and reviewed for common themes, for contradictory cases, and for ideas and feedback that appeared important for intervention development.

Quantitative assessments.

Questionnaires collected sociodemographic information and measures of recovery, mental health, quality of life, patient activation, and functioning. Measures were selected that measured factors associated with recovery, including the Colorado Symptom Index (CSI) (16,17), the Wisconsin Quality of Life Index (W-QLI) general life satisfaction subscale (18–20), the Recovery Assessment Scale (RAS) (21,22), the Patient Activation Measure–Mental Health (PAM-MH) (23), and the 24-item Behavior and Symptom Identification Scale (BASIS-24) (24).

Results

Participants

Across all cohorts, 82 people participated in the study (27 men and 55 women). Table 1 provides demographic information. Cohorts 2 and 3 (from the RCT) did not differ significantly at baseline on sociodemographic, diagnostic, or mental health functioning characteristics, although cohort 2 appeared to have slightly better scores than cohort 3 on measures of recovery, functioning, and life satisfaction.

Cohort 1 (feasibility pilot and six-week intervention)

Qualitative analyses of postintervention interviews provided information about participants’ feelings about the workbook, the group, and the session format. Participants liked the workbook and group but wanted the group to last longer than six weeks to allow more time for reflection, for working through exercises, and for in-depth discussions about key topics. Participants reported finding fellowship in the group and validation of experiences, and they liked learning from others with experiences similar to their own. They appreciated the strengths-based focus, contrasting this to other groups they had attended that focused more on symptoms or problem solving. Several reported feeling more hopeful about their future, and some reported feeling sad when the group ended. About a third of participants indicated that they would prefer a smaller group.

Leader observations suggested that the group stimulated important thinking about recovery processes and about taking personal responsibility and control of recovery. Many participants reported they had not encountered support for empowerment or personal activation before their participation. Group leaders agreed that the group should be longer and noted differences in participant functioning and recovery stage that affected needs for additional time with the material. In particular, those with poorer functioning and in earlier stages of recovery appeared to need more time than others. A few participants reported that a more homogeneous group (with respect to functional status or recovery) would be helpful; a few wanted the consumer cofacilitator to share fewer personal experiences.

Pre-post intervention quantitative measures showed statistically significant improvements in hope for improved quality of life, perceived control over important areas of life, symptom scores, and increased willingness to rely on others (an item on the RAS subscale). No change was observed in general life satisfaction or happiness. On the basis of these findings, we concluded that the intervention was feasible and acceptable and that six weeks was inadequate to cover the material in the workbook. On the basis of this assessment, we increased the length of the intervention to ten weeks to provide additional time to cover materials while minimizing intervention costs.

Cohorts 2 and 3 (RCT with ten-week intervention)

Baseline assessment.

When participants (N=30) were asked about hopes and expectations for the recovery group, the most common reasons given were wanting to help others (N=11, 37%), learning new ways to cope with mental health problems (N=7, 23%), improving symptom management (N=5, 17%), establishing an identity beyond that of the mental health diagnosis (N=4, 13%), and acquiring social connections and support (N=4, 13%). When asked about goals and desired changes, participants mentioned increasing activities (N=11, 37%), obtaining paid employment (N=9, 30%), and achieving education goals (N=6, 20%).

Follow-up 1 qualitative assessment.

After group participation, intervention group members reported primarily positive reactions. Among the 28 participants who responded, the most common theme (N=7, 25%) centered on the feeling that group interactions stimulated new ideas for coping with mental illness and new perspectives on recovery. One participant stated, “What has been lacking from my experience as a patient were concrete examples of success. It [the group] made mental illness seem workable in a way that it had not seemed to me previously.” Ten of 12 participants voiced strongly positive opinions about the value of having both professional and peer cofacilitators. “I actually liked that part of the group. The co-leader seemed very knowledgeable, and it was nice to know that she had been in our shoes but she also wasn’t looking down on us. I felt like they did a good job of going back and forth.”

When asked what they did not like about the program, many participants noted difficulties with group dynamics; more talkative members dominated some discussions. Another commonly expressed concern was that ten weeks was insufficient time to complete the workbook.

Follow-up 1 quantitative assessment.

Twenty-eight (93%) participants completed questionnaires at follow-up 1. Significant improvements were seen over time in the intervention group in scores on the CSI, the RAS general scale, the RAS personal confidence and hope subscale, and the W-QLI relationships scale (Table 2). Marginally significant improvements were seen in the RAS subscale for willingness to ask for help, the W-QLI general life satisfaction scale, and the PAM-MH. In contrast, for the delayed-intervention control group (which had not yet received the intervention), participants showed improvement on only the RAS willingness to ask for help subscale, and that improvement was marginal.

Results from analysis of covariance models that were adjusted for baseline values showed net improvements in the intervention group compared with the delayed-intervention control group. The initial intervention group improved by a net 6.2 points on the RAS overall scale, by a net 2.2 points on the RAS not dominated by symptoms subscale, and by a net .44 points on the W-QLI general life satisfaction scale. No significant between-group differences were found for other outcomes.

Follow-up 2 qualitative assessment.

After participation in the intervention, cohort 3 (delayed-intervention control group) reported favorable reactions, noting that the group’s “positive approach” was helpful (comments indicated that the group focused on being “happy instead of just coping. Not just trying to get through and not hurt[ing] yourself, but asking what does it mean to lead a normal life and be happy?”) Members liked the cofacilitator model but felt there was too little time to cover the material. A few cohort 3 members suggested allowing greater flexibility during group discussions rather than following a topical outline; two participants were uncomfortable with the topic of sexuality in a mixed-gender group.

Follow-up 2 quantitative assessment.

Thirty participants (100%) completed questionnaires after the delayed intervention. Analyses focused on trajectories from follow-up 1 rather than on group differences at the final time point. In cohort 2 (intervention group) there was a marginal worsening of social relations on the W-QLI social relations subscale; no other differences were found compared with the first postintervention follow-up assessment. In cohort 3 (delayed-intervention control), significant improvements were seen for the RAS personal confidence and hope subscale and for the BASIS-24 depression and functioning subscale and the substance abuse subscale. Marginally significant reductions were seen in the CSI and the BASIS-24 overall scales (Table 2).

Cohorts 4 and 5 (pre-post study of intervention length)

Feedback from participants in the RCT that the ten-week group was still not long enough to adequately cover the workbook material, together with additional evidence that the intervention was feasible and effective, led us to conduct two additional groups to determine optimal intervention length. For both, we allowed a range of ten to 18 weeks, which we based on participants’ comfort with the pace of working through the workbook.

Baseline qualitative analyses.

Consistent with previous cohorts, the most common baseline goal and desired change among the 30 respondents was to increase activities (N=13, 43%). This was followed by improve mental health (N=9, 30%); increase education (N=8, 27%); and improve finances (N=5, 17%), housing (N=5, 17%), or employment (N=5, 17%). Common expectations were to learn self-help methods by learning from others (N=10, 33%); to improve sense of self, self-confidence, and self-direction (N=9, 30%); and to learn coping skills, particularly for controlling emotions and redirecting oneself (N=5, 17%).

Follow-up 1 qualitative analyses.

After group participation ended, when participants (N=26) were asked what they liked about it, some (N=10, 39%) indicated that they liked having a place to meet others with similar problems, because they felt less alone; some (N=9, 35%) reported liking learning new skills, gaining perspectives on managing mental health, and advocating for themselves. Participants appreciated support and nurturing from others (N=7, 37%); five of 26 (58%) participants responded that they liked the strengths and problem-solving focus of the group. Finally, 17 of 26 (65%) reported that they liked having a peer cofacilitator. Some participants (N=5, 19%) mentioned that sensitive topics (such as sexuality and spirituality) made them uncomfortable, four (15%) indicated the pace was too fast, and three (12%) felt that sessions were too short for in-depth exploration.

Quantitative analyses.

Analyses of changes from baseline to the ten-month follow-up showed significant reductions in mental health symptoms on the CSI and significant improvements in functioning, as indicated on the BASIS-24 overall functioning measure and the depression and functioning, interpersonal relationships, self-harm, and emotional lability subscales. We found significant improvements in social relationships, as indicated on the W-QLI subscale, and in overall recovery as indicated on the RAS and the RAS not dominated by symptoms subscale. We found marginally significant improvements in personal confidence and hope and in being goals and success oriented, as measured with RAS subscales (Table 2).

Discussion

Results suggest that recovery-focused group interventions that include peer-developed materials and a combination of peer and professional coleaders have potential to facilitate recovery across multiple dimensions among people with a range of psychiatric diagnoses. Qualitative analyses show that the intervention was feasible and acceptable, valuable, and helpful to participants and that the consumer-developed workbook was useful. Participants also reported that they liked having a consumer coleader and found these coleaders helpful as role models.

The duration of the intervention was also important. Our results suggest that interventions based on lengthy materials, such as the Pathways to Recovery workbook (which runs nearly 450 pages), should be long enough to allow for extended reflection and to work through exercises and delve more deeply into key topics. For this intervention, participants indicated that 17 or 18 weeks was adequate. This may be particularly important because group attendance on a week-to-week basis was affected by instability in participants’ lives; the longer intervention period allowed people to attend more sessions overall.

Difficulties with group dynamics and personal disclosures by peer facilitators suggest that leader training should specifically address these topics. Positive responses to intervention content, combined with evidence that the intervention facilitated recovery across multiple domains, suggest that interventions of this type have value in a patient-centered service system. In addition, the findings across the different cohorts, drawn from different settings and with different diagnoses, were similar. We approached the project with the assumption that people with different diagnoses have more in common than their diagnoses might suggest. The consistency between our qualitative and quantitative findings suggests this assumption was correct and that a strengths-based approach, designed for a wide range of individuals with different psychiatric symptoms and diagnoses, can be effective. Moreover, making such interventions available to a broader range of individuals could reduce costs compared with providing a variety of more narrowly targeted programs. Finally, our findings are especially promising given the limited duration of the intervention and the stability of outcomes over moderately long follow-up periods.

Conclusions

In mental health systems with shrinking resources, the availability of less expensive yet effective recovery interventions is increasingly important. Programs with professional counselor and peer counselor coleaders could be valuable adjuncts to mental health care.

The sample sizes for the cohorts were small and the number of sessions in cohorts 1–3 limited. Thus our results may underestimate effects. Small samples also prevented us from examining differences based on participant characteristics, such as diagnosis. Future research is needed to evaluate effectiveness of the 17- or 18-week intervention, to examine differences by key participant characteristics (including age, gender, and diagnosis), and to assess long-term outcomes. Women were overrepresented; thus information is needed regarding engaging men or adapting the intervention to meet men’s needs. Finally, research on the relative effectiveness of leadership configurations (led by a consumer, a professional, or both) could help determine the most cost-effective allocation of resources.