Over half of people with schizophrenia are cared for by family members (
1,
2). Despite the heavy burden, the involvement of family caregivers has resulted in improved patient recovery (
3,
4). Clinical guidelines and the efficacy of some family intervention models, such as psychoeducation, in reducing relapse rates and improving medication adherence of patients with schizophrenia indicate that family caregivers should be offered some form of psychosocial intervention (
4–
7). Although such family interventions enhance knowledge about the illness and reduce relapse rates and the number of rehospitalizations (
3,
5), evidence of their effects on other patient and family health-related outcomes, especially in the longer term (more than two years), is inconclusive (
3–
5,
8).
Peer support groups are a form of self-help for people affected by severe mental illness whose needs may have been inadequately addressed by routine mental health services (
9,
10). Family participation in peer support groups, such as the National Alliance on Mental Illness and affiliated peer support groups for families affected by severe mental illness (
10–
12), can reduce caregiving burden and improve family coping. Participation in peer support groups is less effective, however, in improving patients’ symptoms and psychosocial functioning (
13,
14). Family-led education and self-help groups require less intensive training for professionals as facilitators and provide a flexible, interactive client-directed approach for family caregivers to cope with their stress in caregiving (
15,
16).
Only a few studies of effectiveness of peer support groups have focused on Chinese populations (
1,
16) and included them in a psychoeducational group intervention. Chinese populations have culture-specific family structures, functions, and processes, such as extended and close interdependence and mutual support between family members, a collective identity and behaviors (
17,
18), and a strong sense of filial responsibility and obligations of caring for a disabled relative (
1,
18,
19). As suggested by Ran and colleagues (
18) and Xiong and colleagues (
19), Chinese people with mental illness who never marry are usually cared for by their parents and outlive them, and thus these parents are very keen to help their mentally ill children obtain stable employment and acquire better problem-solving and self-care skills.
Based on very positive findings of our initial 12-session family-led peer support group program (FPGP) (
1,
9,
13), this trial evaluated the effectiveness of a modified 14-session FPGP and compared it with a psychoeducation group program and with standard psychiatric care for Chinese patients with schizophrenia. The comparison focused on both patient and family health outcomes and functioning over a three-year follow-up. [Details and research evidence of the initial program are available online in appendix 1 of a
data supplement to this article.] The main hypotheses with this trial were that the FPGP intervention would significantly reduce patient symptoms, rehospitalizations, and missed outpatient appointments and improve patients’ and families’ functioning and their utilization of mental health services compared with psychoeducation or standard care.
Methods
A randomized controlled trial with a three-group repeated-measures design was conducted at three regional psychiatric outpatient clinics in Hong Kong between August 2007 and January 2011. Eligible patients and family caregivers (dyads) were randomly assigned to the FPGP intervention, to a psychoeducation group, or to standard care. Participants were assessed at recruitment and again one week (posttest 1), 18 months (posttest 2), and 36 months (posttest 3) after completion of the interventions by a trained research nurse who was independent from the participants’ recruitment procedure and blind to their intervention participation. During follow-up, participants received no further group intervention unless it was initiated and run by the group members themselves.
The trial was approved by the Clinical Research Ethics Committee of the Hong Kong Polytechnic University and registered with ClinicalTrials.gov.
Recruitment
Participants were selected randomly from the outpatient clinics’ lists, and the sample comprised more than 1,200 (13%) patients with a diagnosis of schizophrenia (
1). The clinics served all areas of the New Territories, Hong Kong (three million, or 40% of the total population). [Trial participant enrollment and flow, according to the revised version of the CONSORT statement (
20), are summarized in appendix 2 of the online
data supplement.]
Inclusion criteria were caregivers living with and caring for a relative with a primary diagnosis of schizophrenia that met DSM-IV criteria, patients with no other mental illness at baseline, age >17 years, and understanding of Mandarin or Cantonese. Exclusion criteria included caregivers who had mental illness themselves (N=58) or who had been primary caregivers for less than three months (N=40). Forty-five families refused to participate because of a lack of interest or time. For patients with more than one caregiver, we recruited the family member who had the major caring role as indicated by the patient.
Of 550 patients (and family caregivers) eligible, 106 were randomly selected from the patient lists by means of computer-generated numbers, and both the selected patients and their family caregivers were contacted by the principal researcher. Written informed consent was obtained after the purpose and procedure of the study had been explained to each patient and caregiver. After completion of the pretest, participants were randomly assigned to receive the FPGP intervention (N=35), psychoeducation (N=35), or standard care (N=36).
Sample size and expected study power
Sample size calculations were performed a priori in terms of the study hypotheses and outcome variables on an intention-to-treat basis. Family caregiving burden and patient functioning were two main outcome variables of studies on family intervention for schizophrenia (
1–
5), and in this trial. A sample size of 35 families per intervention was required to provide 80% power (two-sided p<.05) to detect a difference on a change in the mean score of .6 point on patient functioning or family burden; the calculation assumes a standard deviation of their change scores of 1.2 (moderate effect size of .48) and 15% expected attrition. This effect size was consistently found to be sufficient to detect the mean score differences of the study outcomes in similar family intervention studies (
1,
9,
13).
Measures
At pre- and posttest, family caregivers completed the Family Assessment Device (FAD) and the Family Support Services Index (FSSI), and patients completed the Specific Level of Functioning Scale (SLOF). Their Chinese versions were used and had been tested among Chinese patients with schizophrenia and with satisfactory reliability and validity (
9,
13,
21). The average number and duration of patient rehospitalizations and number of default outpatient follow-up visits obtained from clinic records were noted by the research nurse over six months before recruitment (pretest) and three posttests.
Patients’ and families’ demographic data (age, gender, education, and duration of mental illness) were collected by the research nurse at baseline. Antipsychotic medications were noted from patients’ outpatient treatment sheets and dosages converted into haloperidol equivalents for comparison (
22). The research nurse also assessed severity of positive symptoms at pretest and the three posttests using five items of the Brief Psychiatric Rating Scale (
23).
The FAD (
24) is a 60-item scale measuring family functioning on a Likert scale (ranging from 1, strongly disagree, to 4, strongly agree). The Family Burden Interview Schedule, completed by family caregivers, is a 25-item semistructured interview schedule to assess burden in caring for a relative with schizophrenia on six domains, including family finance, routine, leisure, interaction, general medical health, and mental health. Its items are rated on a 3-point Likert scale (ranging from 0, no burden, to 2, severe burden). Both English and Chinese versions indicated satisfactory internal consistency (α=.88–.96) and significant correlations with patients’ psychopathology and social dysfunction (
13,
21).
The FSSI (
25) is a 16-item checklist (yes-no) measuring mental health service needs and utilization of families of people with mental illness. Its Chinese version showed satisfactory interrater reliability (r=.85) and internal consistency (α=.90) among Hong Kong Chinese patients with severe mental illnesses (
1,
9).
The SLOF (
26) is a 43-item assessment scale in which each item is rated on a Likert scale (ranging from 1, totally dependent, to 5, highly self-sufficient) in three functional areas (self-maintenance, social functioning, and community living skills) of psychiatric patients. Its Chinese version demonstrated satisfactory content validity and internal consistency (α=.88–.96) among Chinese patients with schizophrenia (
9,
13).
Intervention
Thirty-five family caregivers (and patients) received the nine-month FPGP program modified from our previous work (
1,
9,
13), which provided a hybrid model of care integrating peer support and education into the context of standard psychiatric care. The 14 group sessions (each lasting two hours) were mainly held every two to three weeks, and participants were encouraged to interact and have activities outside of these group sessions. A peer leader, who had more experiences than other participants in caregiving and was elected among the group members during the first group session, was trained by the researchers with a two-day group coordination and leading workshop to facilitate the group sessions. The principal researcher who acted as a resource person provided support for the group development at its early stage, as requested by the peer leader (
15,
27).
Table 1 summarizes the main content of the FPGP sessions in these five stages. Emphasis was given to problem-solving and caregiving practice, and to specific Chinese cultural issues, including high social stigma associated with mental illness and health services, interdependent family structure and collective behaviors, and high tendency to expect immediate practical help (
17–
19).
Similarly, 35 psychoeducation group participants received 14 bi- or triweekly, two-hour sessions of psychological support, education, problem-solving activities, and routine outpatient care. Content and format were based on previous programs (
1,
28,
29) with reference to those conducted by Li and Arthur (
17) and Ran and colleagues (
18) in mainland China. The program emphasized the importance of family environment and relationships and involvement in caregiving and provided information about mental illness and its treatment, problem-solving and coping skills training, and discussion of and practice with caregiving skills, including effective communication, medication adherence, crisis intervention, and Chinese culture, such as family collective behaviors and needs for practical assistance. [Details on the psychoeducation program are provided online in appendix 3 of the
data supplement.]
For both the FPGP and psychoeducation groups, patients were instructed to attend at least five sessions in which knowledge of the illness and treatment, drug compliance, and availability of mental health services were addressed. They were also invited to participate in the other nine group sessions as preferred and agreed to by the family participants. Seven experts (psychiatrists, psychologists, and nurse specialists) in psychiatric rehabilitation rated independently the appropriateness of the contents of two programs and treatment progress by regularly reviewing the audiotaped sessions and clarifying problems encountered. Amendments were made on one topic in each program.
Participants in all three groups received routine psychiatric care, consisting of medical consultation and advice, individual nursing advice on medication and community services, and social welfare and counseling services by social workers.
Statistical analyses
Analysis of variance with the Kruskal-Wallis test by ranks was used to test any differences in demographic characteristics of the three groups. We used a repeated-measures mixed-model multivariate analysis of variance (MANOVA) test for the dependent variables to determine whether the interventions produced the within- and between-groups and interactive group-by-time effects that we postulated. A Helmert contrast test was used to examine any significant differences in the means of each outcome measure within and between groups over time (
30). The level of statistical significance was set at p<.05.
Mean scores of the outcome variables were normally distributed, indicating satisfactory homogeneity of variance with only slight or moderate correlations (Pearson’s correlations ranged from .22 to .38) at pretest. Data analyses were on an intent-to-treat basis, and there were only a few missing data, which were replaced by group mean imputation (
9,
30) with little difference to the results.
Results
For all groups (
Table 2) the mean±SD age of family caregivers was 42.7±6.0 (range 22–60), two-thirds (65%, N=69) were female, and their relationship with the patient was mainly parent (32%, N=34), spouse (30%, N=32), or child (24%, N=25). The mean age of patients was 27.3±5.6 (range 20–49), nearly two-thirds (62%, N=66) were male, and over half of them were taking medium dosages of antipsychotics (haloperidol equivalent mean values range from 8.3±4.8 to 10.1±5.0) (
14). Over two-thirds (N=24–26) were taking oral medication, and about one-fifth (N=6 or 7) were taking oral and depot intramuscular medications. On average, two family members lived with each patient (68%, N=72), and the mean duration of illness was 2.5±1.2 years (range .5−6.0 years).
Of the participants, 31 (89%) family caregivers in the peer support group and 32 (91%) in the psychoeducation group completed the intervention. They, together with those who dropped out or were absent for more than five group sessions—FPGP (N=4), psychoeducation (N=5), and standard care (N=3)—were evaluated at pretest and three posttests (details are available online in appendix 2 of the
data supplement). Reasons for dropout or discontinuation were similar: insufficient time to attend (N=4), patient’s mental state worsened (N=4), not interested (N=3), and being the only person taking care of patient (N=3). In addition, 33 (94%) patients in both the peer support and psychoeducation groups attended five sessions as instructed, and their average attendance rates were 9.8±2.8 and 9.3±3.9 sessions, respectively (range six to 14 sessions). Therefore, patient attendance at both group programs was very similar.
There were no significant differences in any demographic variables among the groups, nor significant correlations (Spearman’s r<.20) between the demographic variables and patient and family measures at baseline, thus indicating no covariate effect.
Mean score differences on pretest data between groups were compared with a between-groups MANOVA and univariate contrasts tests (
30) and were not significantly different (p>.20). There was a significant difference between groups on the combined dependent variables (F=5.32, df=5 and 101, p=.003; Wilks’ λ=.93; partial η
2=.35, a large effect). Results of MANOVA tests for each of the dependent variables at the four measurements (
Table 3) indicated that there were significant differences between groups over time (group × time interactions), including improved family functioning (FAD score; F=5.32, df=2 and 103, p<.005) and patients’ functioning (SLOF score; F=6.33, df=2 and 103, p<.001) and reduced duration (F=4.82, df=2 and 103, p<.01) and number (F=4.78, df=2 and 103, p<.01) of patients’ rehospitalizations. The MANOVA on SLOF subscale scores also indicated significant differences between groups for all subscales (p<.001 or p<.01).
For the outcome variables indicating significant results in the MANOVA test, Helmert contrast tests were conducted to detect any significant differences on changes in mean scores within and between groups at pretest and posttests. The mean differences were statistically significant between the FPGP and the other two groups (
Table 4). The FPGP’s family (FAD score) and patient (SLOF score) functioning improved significantly at both posttests 2 and 3 (FAD, posttest 2, p=.008, and posttest 3, p=.003; SLOF, posttest 2, p<.001, and posttest 3, p=.008) and at posttests 2 and 3 (FAD at both p=.01; SLOF at p=.003 and .001), compared with standard care and psychoeducation, respectively; and second, rehospitalizations of FPGP patients decreased significantly at the three respective posttests in number (p=.04, .009, and .008) and duration (p=.04, .009, and .005) compared with standard care and decreased significantly at posttests 2 and 3 (p=.04 and .01; p=.04 and .03, respectively) compared with psychoeducation.
Neuroleptic medication scores changed slightly over time and did not differ between groups (
Table 3). Types of neuroleptics and default follow-up rates did not differ between groups. The FSSI scores of the three groups (3.6–4.2 out of 16 services listed) were stable between pretest and posttest 3, indicating no significant change in patient and family demands for mental health services over time. The most frequently requested services included day treatment programs and social programs for patients (including, for example, occupational training and recreational activities), family counseling, home visits by community psychiatric nurses, and respite care. There was also no difference between groups over time in severity of patients’ positive symptoms.
Discussion
This trial was one of very few to test the effectiveness of a family-led peer support group intervention in providing family-centered care for patients with schizophrenia. The findings are encouraging for the use of FPGP for Chinese with schizophrenia in community care to improve long-term psychosocial health outcomes for patients and their families. Although the original (six-month) FPGP we developed provided preliminary evidence of enhanced family functioning and reduced patient symptoms in the short to medium term (<18 months) (
1,
9,
13), the modified nine-month FPGP indicated superior effects on more diverse patient and family outcomes plus a 36-month follow-up. Compared with families that received family psychoeducation and standard care, those who received the FPGP intervention showed significant and consistent improvements in overall family functioning, patients’ rehospitalization, and all aspects of patient functioning, including self-maintenance, social functioning, and community living skills. Whereas there were significant improvements in patient functioning, there was no difference in the severity of patients’ positive symptoms between the three groups.
Although current patient treatment plans emphasize medication adherence and symptom control, this trial indicated that social support, problem solving, family needs recognition, caregiving role adoption, and interpersonal and caregiving skills modeling and sharing in the FPGP may intensify the group effect on patient care. Further research is needed to explore the differential outcomes we found across groups while positive symptoms remained constant. Investigation of this phenomenon, along with research to better understand the therapeutic components and process of change in family peer support groups, is needed, as is study of other potential mediating variables such as culture and acculturation status, negative symptoms, medication adherence, and characteristics within the patient-caregiver dyad.
Although the psychoeducation program used in this study may differ from those recommended by the Schizophrenia Patient Outcomes Research Team (
31) as part of patients’ treatment regimens, the psychoeducation program used in this trial also demonstrated significant improvements in family and patient functioning compared with routine care over the first 18-month follow-up. Family psychoeducation is well accepted and widely used in the United States (
22,
32), the United Kingdom (
33), and mainland China (
19,
26,
34), although reviews and meta-analyses have concluded that it produces only modest effects on patient and family psychosocial health, particularly in the longer term (more than two years) (
2–
5). Our findings suggest that family psychoeducation without full integration into patients’ treatment may be unable to demonstrate substantial benefits beyond 18 months. In contrast, with the significant beneficial effects found in this trial at the three-year posttest, FPGP may be considered a superior alternative to standard schizophrenia care. It is worth noting that the FPGP intervention was culturally adapted to meet Chinese family values and needs, whereas the psychoeducation program was not (
4,
17,
18).
There was very little change in patients’ symptom severity and demand for family support services in both the FPGP and psychoeducation groups. Intervention completion was high (FPGP, 88%; psychoeducation, 90%), and attrition was very low (6% and 9%, respectively). These high rates may reflect the high motivation and enthusiasm of the families who participated, as found elsewhere (
10,
35), and may indicate that family caregivers could better cope with their caring role and manage patients’ illness-related behaviors successfully, with increased caregiving knowledge and skills and thus more appropriate use of available services as needed (
35,
36). Peer support groups can be a stress mediator for family members in a life crisis because of the provision of information and instrumental support between group members to meet family needs, on top of the community support resources offered by mental health professionals (
9,
36).
Despite endorsement of the use of peer support groups for families with schizophrenia, few trials have examined its effectiveness in community mental health services (
1,
33). This trial is an exception. In contrast to psychoeducation and behavioral models of family intervention, mutual support groups do not require a family therapist with extensive training or a highly structured working manual. Clinicians’ hesitation in using family intervention may be related to concerns about the experience and training required for a therapist (
5,
9), or families may be unwilling to attend a program with an expected modest benefit. The type of peer support in this study provides a flexible and interactive self-help group environment and only a nine-month period of intervention, which may be more acceptable and feasible in community care with limited resources. It may be better oriented to family needs and improve family functioning through two mechanisms: increased social support, resulting in an enhanced sense of control over interpersonal skills and family care, and a shared-experience, “all in the same boat” belief, providing effective social learning of patient care for other group members (
9,
12,
36).
Even when in need of professional help in caring for a relative with a mental illness, many Chinese families are reluctant to seek help because of strong perceived stigmatization and an unwillingness to expose family weaknesses to nonfamily members (“saving face”) (
9,
16). It is usual for Chinese families to feel guilty and blame themselves for the mental illness of a member, which may prevent them from living harmoniously with their mentally ill relative. If family intervention is to be effective, it should be culturally sensitive. Family peer support groups for Chinese with mental illness are effective in offering practical assistance and advice on caregiving, enhancing help-seeking, and allaying feelings of guilt (
9,
13,
17). These improvements may be a result of an increased willingness to share caregiving experiences openly with those facing similar difficult life situations.
A few limitations of this trial are noteworthy. First, the patients were recruited from three of 18 outpatient clinics in Hong Kong, and most had a short duration of illness (six years or less), when families might be more optimistic and motivated about the potential for change compared with later stages of illness (
9). Family caregivers were volunteers and were young (over half age <40 years), were typically female (>63%), and had a satisfactory education level and monthly income; combined, these factors may have made this group of caregivers more motivated than others to participate in the FPGP (very low dropout rates). With the highly selective sample, many patients were more mentally stable, male (>60%), and younger (mean age of 26–28 years) than those in previous studies. Their gender pattern and illness condition might not be considered representative of this Chinese patient population, and the potential effects of these variables on patients’ treatment responses and course of the illness are worth further exploration. In contrast to most psychosocial interventions, such as behavioral management family programs (
8,
25,
32), the peer support group had a fairly structured treatment protocol, which could be flexible to allow optimal autonomy and self-empowerment (
35,
36), and group dynamics and interactions within and outside support group meetings (
8,
37). Therefore, the contributions of these features merit further investigation by interview and observational studies. Another limitation of this study is that the FPGP intervention but not the psychoeducation intervention underwent a careful and deliberate process of cultural adaptation. It is unclear to what extent the FPGP benefits were the results of the mutual support components or of the cultural adaptation, which would be a topic for future research.