The database search yielded 933 titles and abstracts in addition to another two articles from hand searching. Following the removal of duplicates, 890 titles and abstracts were screened, after which 874 were excluded. The full-text articles of 13 references were obtained and considered against inclusion and exclusion criteria. Eight studies were excluded for different reasons, as shown in PRISMA chart. These left five studies to be included in the final review (see Fig.1 for the PRISMA flowchart).
Insert figure 1 here
Five studies were retained as they met the inclusion criteria. Two studies were conducted in Jordan and three in Egypt. These were published between 2008 and 2018. In total, 344 dyads of patients and their caregivers were recruited into the studies. The designs were two RCTs (29, 30), two non-randomized trials (31, 32) and one qualitative study (33). See table 1 for the descriptive characteristics of studies.
Insert Table 1 here
First, as shown in table (2), the methodological quality was good for Hasan, et al. (2014) and poor for Rami, et al. (2018), which has a higher risk of bias. The study was not explicit about the method of randomization or allocation concealment. Furthermore, the study protocol was not available to assess the reporting bias. The study did not provide a hypothesis, power calculation or primary outcome. All statistical differences between arms were reported, but there was no report of effort to minimise bias.
Insert table 2 here
Second, for non-randomized trials, the two studies have a high risk of bias because two or more criteria are not met according to the JBI tools (see Additional file 2 for Methodological Quality of Non-randomized Trials).
Insert table 3 here
However, there is not enough data reported to judge the quality in many instances. The study, done by Soliman, et al. (2018), was reported as a cross-sectional interventional study, but it is more consistent with quasi-experimental design because of the lack of randomisation and inclusion of control and intervention group. The study risked bias in selecting participants because the sample was not randomized. Furthermore, the drop out was not reported, which could have affected the analysis. The study by El-Shafei, et al. (2008), was reported as a case-control design but the elements of control group and randomization make it more consistent with experimental studies. They did not report the difference in basic characteristics between participants in both groups, which may have introduced a selection bias. They did not report a sample calculation and they included 30 participants only. Furthermore, no details about attrition, loss to follow-up or outcome measurement were reported.
Overall, the quality of the included studies is poor, and none of these studies, except Hasan, et al. (2014) had the statistical power to detect the benefit of family interventions. This indicates that the included studies have a risk of overestimating the effect of interventions.
Third, the qualitative study done by Al-HadiHsan, et al. (2017) is consistent with good quality studies because there was congruency between the research methodology and objectives, and the method for data collection, analysis and interpretation. However, two of the questions in the tool were answered with ‘No’. The researcher did not follow any methodological theory for qualitative research because the authors were trying to answer the research question and explain the quantitative data without following specific methodological theory. (See table 4 for methodological quality of qualitative research using JBI tool).
Insert table 4 here
The interventions in the five studies were delivered in Egypt (3) and Jordan (2), and the content of the interventions differed across studies. All studies delivered individual family sessions, but these varied in terms of intervention characteristics like mode of delivery, duration and number of sessions. Four of the studies were delivered in a clinical setting in the outpatients’ department, and one was delivered using a booklet within patients’ homes. The duration of the interventions ranged from 12 weeks to 6 months. The duration of an individual session was reported in one study as 60 minutes. All the interventions were led by healthcare providers or researchers. None of the interventions were delivered in an inpatient setting. All the studies compared family interventions to standard care. See table 5 for the Intervention Characteristics Table.
Insert table 5 here
Contents and Components of the Interventions
Two studies reported the process of adaptation and modification of the original manuals (29, 30). Hasan, et al. (2014) used the framework of Atkinson and Coia, which covers Bloom’s Taxonomy of Learning domains, while Rami, et al. (2018) used the Behavioural Family Therapy (BFT) manual by Mueser and Glynn (1999). The components of the interventions are in Table 6.
Insert table 6 here
First, psychoeducation components included signs, symptoms, aetiology, diagnosis, treatment, and relapse signs and management strategies for schizophrenia. Furthermore, it included truths and myths about schizophrenia, and how these affect the persons’ thoughts, emotions, and behaviour. The treatment component includes information about medication, its side effects, likely benefits of the medicine, adherence to treatment, the importance of follow-up, and information regarding prognosis. Furthermore, leaflets, which contain information about schizophrenia, high expressed emotions families, notes and homework assignments for the problem-solving and communication skills training, were distributed to participants during the sessions (Rami, et al., 2018). Second, communication enhancement training included learning skills for active listening, delivering positive and negative feedback, and requesting changes in each other’s behaviours. Third, problem-solving skills training included identification of speciﬁc family problems and practical advice for solving them. Fourth, the stress vulnerability model discusses the role of the family, burden of care, and stress management skills and strategies.
Strategies Used to Adapt the Intervention
The strategies for adaptation included different themes. First, language adaptation was reported in all studies. The manuals were modified and translated into simple Arabic including folk stories relevant to the cultural beliefs of the participants. Second, the explanatory models of illness were incorporated into the adaptation process. Rami, et al. (2018) increased the number of sessions regarding the biological basis of the illness from one in the original BFT manual to two sessions because of the attribution of mental illnesses aetiology to magic and Jinn in Egyptian culture. Due to the expected low literacy levels in the Arab world, the complexity of psychoeducation was simplified, and the tools and educational material were examined for acceptability, practicality and linguistic accessibility (29, 30). Third, all the studies delivered the interventions in individual therapy sessions instead of groups to facilitate the cultural context of Arabs. Further to these adaptations, the program in Rami, et al. (2018) was shortened to 6 months instead of 9 because of practical and financial reasons that may influence adherence and attendance. See table 7 for the adaptation strategies table.
Insert table 7 here
Feasibility and Acceptability of the Interventions
Feasibility included the assessment of recruitment, attendance, retention (the proportion of participants who complete therapy sessions) and the compatibility of the interventions with the available resources. All the studies reported a feasible recruitment process without major barriers or difficulties. The attendance was also feasible because two of the studies (29, 31) delivered the interventions during the follow-up appointment, which ensured a high level of attendance. The third study by Hasan, et al. (2014) was delivered via a booklet to patients’ homes. The study by El-Shafei, et al. (2008) did not report attendance. The assessment of retention was reported in two studies only (29, 30). Rami, et al. (2018) reported that four subjects from the case group and six subjects from the control group missed their regular sessions. The dropout in the Hasan, et al. (2014) study was six from the intervention group and ten from the control group. All the studies reported compatibility of the intervention with the available resources. The study by Rami, et al. (2018) reported that the intervention was applicable and accessible because of the brevity of the program. Furthermore, meeting the needs of caregivers enhanced the feasibility of the program.
Acceptability is defined as "a multi-faceted construct that reflects the extent to which people delivering or receiving a healthcare intervention consider it to be appropriate, based on anticipated or experienced cognitive and emotional responses to the intervention" (34). Hasan, et al. (2014) followed his trial with a qualitative study to assess the acceptability of interventions. The qualitative interviews with service users and caregivers confirmed the acceptability of the interventions. They found that interventions using booklets was appropriate and valuable. No other studies examined acceptability.
Effect of Interventions
The outcomes reported across the studies vary a great deal and most of them did not distinguish primary from secondary outcomes. The most frequently reported primary and/or secondary outcome is severity of symptoms using the Positive and Negative Syndrome Scale (PANSS). The four studies found a statistically significant difference between the two groups concerning positive and negative symptoms experienced by service users, favouring the intervention group. Furthermore, Hasan, et al. (2014) found a reduction in the severity of symptoms at three months follow-up. Other frequently reported outcomes were social functioning, adherence to medication, quality of life and knowledge of schizophrenia. One study only assessed family outcomes including family burden of care and carers’ quality of life. (See Table 8 for the results of each outcome).
Insert table 8 here