A systematic review of the international published literature relating to quality of institutional care for people with longer term mental health problems

We included papers that examined factors associated with quality of care, of adults of working age with longer term mental health problems living in institutional care in hospital or the community. Papers that examined the relationship between quality of care and operational systems, staffing, staff training, supervision and support were included as well as papers that investigated living conditions and those that investigated specific approaches to improve the quality of care. The review was limited to papers published since 1980 since much of the deinstitutionalisation across Europe has taken place in the last 30 years.

Papers were excluded if the focus was irrelevant to the aims of our systematic review due to one or more of the following:

A) the results were specific to a client group that did not meet our inclusion criteria (e.g. child or adolescent patients; patients in prison; patients with mental illnesses unlikely to require long-term institutional care; patients with dementia; patients with primary drug or alcohol problems) and could not be extrapolated to adults of working age with long term mental health problems living in institutional care in hospital or in the community;

the study was carried out in unrelated settings (e.g. short-term wards or specialist units not focusing on patients with long-term mental health problems or patients living at home or in non-institutional community settings);

the results reported were confined to an exceptional setting, culture, client group or intervention and could not be extrapolated internationally (e.g. national mental health legislation or a very specific service context);

studies that examined patients' quality of life or satisfaction in isolation from their context in institutional care, or whose focus was too broad for its results to be useful for the aims of this systematic review.

studies that reported on drug trials.

Where a systematic review was included, we did not examine each paper contained within it. Nor did we include editorials, letters, books or book chapters.

The following terms were used to identify relevant articles: mental patient*; mental* ill*; mental disease*; mental* deficien*; mental disorder*; schizophreni*; mental* disab*; mental* retard*; psycho*; severe mental illness; psychiatr*; mental health patient; delivery; standard*; quality; benchmark*; evaluat* near care; evaluat* near health care; guideline*; quality of life; treatment satisfaction; model; evaluation stud*; patient* satisfaction; clinical guideline*; evidence based medicine; psychiatric rehabilitation; rehabilitat*; activities of daily living; art therapy; bibliotherapy; dance therapy; exercise therapy; music therapy; occupational therapy; rehabilitation, vocation*; physical restrain*; hold* down; clinical hold*; human right*; patient right*; behaviour control; collaboration; recovery; empowerment; consumer movement; mental health care; mental health cent*; mental hospital*; psychiatric department*; community mental health; community mental health cent*; community psychiatric nurs*; mental health service*; hospital*; inpatient*; institut* care; institution*; deinstitution*; social work, psychiatric; managed care; community mental health care; architectural accessibility; elevator* and escalator*; floor* and floorcovering*; interior design and furnishing*; location directorie* and sign*; parking facilit*; health facility environment; patient* room*; rehabilitation center*; sheltered workshop*; residential facility*; assisted living facility*; group home*; halfway house*; homes for the aged; nursing home*; nursing care; nursing services; rehabilitation; activities of daily living; rehabilitation, vocational; self care.

All search terms were adapted for each database.

The following electronic databases were searched:

Medline: 1980 - May 2007

Embase: 1980 - May 2007

PsycINFO: 1980 - May 2007

CINAHL: 1982 - May 2007

The Cochrane Library as of Issue 2, 2007

Web of Knowledge: 1980 - June 2007

ASSIA: 1980 - July 2007

International Bibliography of the Social Sciences: 1980 - June 2007

Sociological Abstracts: 1980 - July 2007

Social Science Citation Index: 24 October 2007

Science Citation Index EXPANDED: 24 October 2007

Author or paper searches were clarified, where necessary, using Google scholar. First authors of included articles were contacted for additional published or unpublished material when appropriate. Principal investigators from each of the countries participating in the DEMoBinc study provided references or copies of relevant papers that had not been identified from the databases listed above. No relevant studies were found which had been missed by our search.

The American Psychological Association's (APA) guidelines for the treatment of schizophrenia suggest that, where patients require treatment in a residential facility, this should be in the least restrictive setting that will ensure patient safety and allow for effective treatment [22]. Overall, community residential facilities have been found to be less regimented than hospital wards and more facilitative of patient autonomy [23‐25]. Hawthorne et al [26] examined two community residential facilities in America which emphasized provision of treatment in the least restrictive environment and positive staff-patient relationships. In a repeated measures design, where patients acted as their own control, patient functioning significantly increased and rehospitalisation significantly decreased in less restrictive settings even when patient morbidity was taken into account.

A number of studies have found that the majority of patients with longer term mental health problems prefer living in community, rather than hospital, settings [18, 23, 24, 27, 28]. Community settings have also been reported to be associated with better client outcomes than hospital settings [29]. In a national study of community-based residential facilities for people with mental health problems in Italy, facilities with higher levels of restrictiveness and fewer links with community-based activities experienced higher rates of hospital readmission [30]. A Danish study found that community residential facilities were better able to promote residents' activities both within the facility and in the community than hospital-based psychiatric rehabilitation units [31]. Residents of a community hostel, which emphasised individualised care, were found to have a better quality of life and greater freedom compared to patients in hospital-based rehabilitation units with similar levels of psychopathology and impairment [23]. The hostel also had the highest rating of rehabilitation environment quality, with lower social distance between staff and residents, greater flexibility and greater promotion of community integration for its residents.

In a descriptive review of community residential programmes, patient characteristics were reported to have a weaker correlation with positive outcomes than environmental factors [29]. In "board and care" homes in the USA, a positive social climate characterised as cohesive, organised, comfortable and encouraging of residents' independence and involvement in decision making, was found to be associated with greater resident satisfaction with their living situation [32]. High levels of resident involvement, support, spontaneity, autonomy, organisation and programme clarity have been cited as important components of environmental quality in group homes by both staff and residents [33]. Similar elements have been found to be important for greater therapeutic alliance between staff and patients in inpatient settings [34]. Brunt and Hansson [33] also found that security, physical (built) environment and social interaction were considered important by both residents and staff but staff more often stressed the importance of supporting residents to gain practical skills. However, Cournos [29] found that concordance between staff and residents about the importance of specific environmental characteristics was only weakly correlated with resident outcomes.

In a study comparing community-based residential facilities for people with mental health problems in Andalusia (Spain) and London (UK), Spanish facilities were found to be more restrictive with more rules and less privacy [18]. However, Spanish residents had more favourable views than their English counterparts on their individual progress and enjoyment of the company of other residents, greater acceptance of house rules and routine and they reported greater benefits from their activities and medication. Spanish residences were closer to community amenities but twice as many UK participants reported involvement in community activities (such as attending day centres or sheltered employment) whereas Spanish residents made greater use of indoor recreational activities.

There is no clear evidence on the optimal number of residents in community-based residential mental health facilities. A study carried out in the USA found no association between the number of residents per facility and residents' integration in activities within the facility, after adjustment for other factors [25]. Another study found the number of residents in community-based "board and care" homes for veterans in the USA was positively correlated with social functioning in the community [32]. Although, an optimal number will depend upon "the prevalent philosophy of care, available resources and population need", density of occupation (the ratio of residents to available space) rather than a recommended specific number of residents is considered a better guide, since increased density increases residents' stress and decreases their privacy and control over their environment [35].

The effect of the physical environment on patient outcomes was examined in a systematic review of 30 controlled trials [36]. Participants included those with mental health as well as physical health problems. No trials were identified that exclusively investigated wall colour, pictures, plants, gardens, floor coverings or room size. Eleven trials investigated the effect of renovation or redecoration of a whole ward or treatment area on participants' social functioning and symptoms. Inconsistent findings were reported. Two trials found that the amount and timing of access to sunlight was associated with a reduced length of admission for depressed patients. One trial carried out in a psychiatric unit showed that seating arrangements in common areas that encouraged interaction (e.g. seating around small tables) increased patients' social interaction.

Baker and Douglas [37] carried out a large study in New York to investigate outcomes for people with mental health problems living in supported and unsupported community housing of varying condition (assessed by observer ratings of the property's location, exterior condition, interior condition and the condition of their personal property). Those in supported housing where physical conditions were rated below average displayed a significant increase in maladaptive behaviour over the nine-month study period compared to those in housing of average or above average quality.

In a systematic review of 28 studies of supported housing for people with mental health problems, both the quality of the physical environment and the degree of privacy were found to mediate patient outcomes [27]. Corrigan [38] investigated mental health inpatients' satisfaction with their accommodation and found lack of privacy to be a major concern, specifically having a place to be alone and secure storage for personal items.

Meta-analyses of cognitive behavioural therapy (CBT) for patients with schizophrenia and other related psychoses have found consistent evidence for its efficacy [39, 40]. Descriptions of these studies as well as other evaluations of CBT can be found in Table 2. Pilling et al [39] found CBT had a small effect on improving positive symptoms during treatment (n = 273, effect size 0.27, CI 0.15, 0.49; NNT 5, CI 4, 9) and nine to 18 months after treatment (n = 119, effect size 0.25, CI 0.10, 0.64; NNT 6, CI 3, 27) but was not associated with reduced relapse rates. Pfammatter et al [40] examined three meta-analyses of CBT's effect on positive symptoms. The effect varied from small (effect size 0.33, CI 0.14, 0.51) to large (effect size 0.93, CI not reported). In their own meta-analysis of 17 randomised controlled trials (RCTs) they reported that consistent (but small) effects for CBT could only be established when provided to individuals with persistent positive symptoms (during treatment: n = 486, effect size 0.47, CI 0.29, 0.65; post-treatment: n = 335, effect size 0.39, CI 0.17, 0.61) [40].

Turkington et al [41] carried out a multicentre RCT to investigate the efficacy of CBT for patients with schizophrenia who had ongoing positive and/or negative symptoms or were at risk of relapse and found CBT to be associated with improved insight (ANCOVA 0.711, CI 0.11, 1.31, p = 0.021) and fewer negative symptoms (eta² -0.773, CI -1.27, -0.28, p = 0.002) than participants assigned to usual care at 12-month follow-up. It was also found to be protective against depression and relapse.

An RCT of group CBT for people with schizophrenia found no effect in terms of improvement in symptoms, functioning or relapse rates, but a significant increase in self esteem (n = 94, effect size -1.51, CI -2.84, -0.18) and decrease in hopelessness (n = 94, effect size -1.62, CI -3.06, -0.18) at 12-month follow-up [42].

Both NICE [43] and the APA [22] recommend offering CBT to individuals with schizophrenia, especially those with persistent positive symptoms, with NICE recommending treatment over at least six months comprising at least ten sessions.

Many family interventions involve psychoeducation and many trials of psychoeducation involve family members. Therefore, we have included studies of both family interventions and psychoeducation in this section (see Table 3).

Meta-analyses show that, compared to usual care, family interventions (including psychoeducation, crisis management work and problem solving) for people with a diagnosis of schizophrenia reduce the risk of relapse (n = 3838, effect size 0.42, CI 0.35, 0.49) [40] and readmission (6-12 month follow-up: n = 3789, effect size 0.22, CI 0.14, 0.29; 18-24 month follow-up: n = 445, effect size 0.51, CI 0.32, 0.70) [40] and improve medication adherence (n = 393, effect size 0.63, CI 0.40, 1.01; NNT 10, CI 6, 90) [39] (n = 369, RR 0.74 CI 0.6, 0.9; NNT 7 CI 4, 19) [44].

Family interventions that include patients and their relatives are more effective than those for relatives alone [43]. Both single and multiple family interventions are efficacious but drop-out from multiple family interventions is high. Lehman et al [22] suggest the best time to engage families is during the acute phase of the illness or at times of crisis.

In a meta-analysis of 31 RCTs of family psychoeducation carried out by Pfammatter and colleagues [40], improvements were shown in family members' understanding of the disorder (n = 3662, effect size 0.39, CI 0.31, 0.46) and expressed emotion (n = 284, effect size 0.59, CI 0.36, 0.83) and in patient's social functioning (n = 3362, effect size 0.38, CI 0.30, 0.46) and general psychopathology (n = 178, effect size 0.40, CI 0.10, 0.70).

In a Cochrane review of 10 RCTs of psychoeducation for service users, which included interventions where family members also participated, psychoeducation was found to significantly decrease relapse rates at nine to 18 months follow-up (n = 720, RR 10.8, CI 0.70, 0.92; NNT 9, CI 6, 22) and increase global psychosocial functioning (Weighted Mean Difference (WMD) 5.2, CI -8.8, -1.7) at one year follow-up [45].

An RCT examining the effectiveness of psychoeducation provided to patients and families in separate groups compared to standard care found patients in the experimental group had significantly lower rehospitalisation rates than the standard care group at 12 (N = 163, RR 0.56, CI .033, 0.92) and 24 (N = 153, RR 0.70, CI 0.50, 0.97) month follow-up [46].

Carrà et al [47] found no statistically significant differences in patients' relapse or readmission rates in an RCT in which families were assigned to attend either a psychoeducation group with the patient, a psychoeducation group with the patient plus a support group without the patient, or treatment as usual. Patient adherence with standard care was better for families who received the psychoeducation plus support group intervention, although carer burden increased. However, several studies have found that both service users and their family members receiving psychoeducation show an improved level of knowledge about the relevant psychiatric condition [45, 48] but no consistent improvement in insight or adherence to medication [45].

As well as reduced relapse rates and improved symptoms and social functioning, other reported benefits of multiple family psychoeducation groups are improved well-being for family members and increased service user participation in vocational rehabilitation and competitive employment [49].

Although psychoeducation has been shown to have beneficial effects on patient outcome, it is not regularly provided in inpatient care [48]. Rummel-Kluge et al [21] used a large postal survey to investigate difficulties in implementing a psychoeducational intervention in psychiatric hospitals in Germany, Austria and Switzerland. Although 86% of the institutions offered psychoeducation, only 21% of patients with schizophrenia and 2% of their family members had received the intervention in the previous year. Staff stated they lacked resources and training.

Clinical guidance from the UK [43] and US [50] recommends family interventions last over six months and with a minimum of ten sessions [43, 50].

Supported employment is an approach to improve vocational functioning among people with various mental health problems including schizophrenia [22]. Evidence is strongest for programs that encourage direct entry into competitive employment and provide individualised workplace support rather than models which offer step-wise progression towards employment [51, 52]. In two meta-analyses of different approaches to vocational rehabilitation, supported employment was found to be three to four times more successful in achieving competitive employment than other forms of vocational training such as sheltered workshops, psychosocial rehabilitation work programmes and transitional employment schemes (RR 0.76, CI 0.64, 0.89; NNT 4.5, CI 4.48, 4.63) [53] (OR 4.14, CI 1.73, 9.93) [54]. Descriptions of studies relevant to vocational therapy can be found in Table 4.

Individual Placement and Support (IPS), a specific, manualised version of supported employment, has been shown to be more effective than prevocational training in terms of participants achieving competitive employment (n = 295, RR 0.79, CI 0.70, 0.89; NNT 5.5, CI 3.6, 10.2) [53] and their number of days in employment but there is insufficient evidence as to whether IPS is more effective than other less carefully specified forms of supported employment [53, 54]. The components of IPS that are most beneficial are: rapid job search; elimination of prevocational preparation; sensitivity to the client's job preferences, strengths and work experience; integration with mental health services and time-unlimited support [50, 52].

Integration of supported employment programmes within other mental health services is more successful in engaging and retaining clients in vocational rehabilitation than when these services are separately provided. Integrated programmes also reduce problems with communication between services and raise mental health staff's awareness of the achievability of clients' vocational goals [55].

Supported employment is recommended by the APA [22]. In the UK, NICE [43] recommends the provision of supported employment for individuals who wish to work. However, they also recommend that other vocational rehabilitation resources are available for those who are unable to work.

Social skills training (SST) aims to improve social functioning for people with a diagnosis of schizophrenia by teaching them skills to improve their social performance in activities of daily living, employment, relationships and leisure [56]. Descriptions of studies examining SST can be found in Table 5.

The effectiveness of SST has been examined in two meta-analyses [40, 57] which reached different conclusions regarding evidence for its efficacy. Pilling et al. [57] included nine RCTs of SST and found no clear evidence of benefit for relapse rates, global adjustment, social functioning, quality of life or treatment adherence. In contrast, Pfammatter and colleagues' [40] meta-analysis of 19 SST studies (quasi-experimental studies as well as RCTs) found beneficial effects for the acquisition of social skills (during treatment: n = 688, effect size 0.77, CI 0.62, 0.93; post-treatment: n = 295, effect size 0.52, CI 0.28, 0.77), improvement in social functioning (during treatment: n = 342, effect size 0.39, CI 0.19, 0.59; post-treatment: n = 210, effect size 0.32, CI 0.08, 0.56) and reduced hospitalisation (post-treatment: n = 110, effect size 0.48, CI 0.11, 0.86).

Bustillo et al [56] included five SST studies (two were RCTs) in a systematic review of psychosocial treatment for schizophrenia. They noted that although social skills were usually enhanced when assessed, this did not generalise to social competence in the community.

Roder et al [19, 20] carried out an evaluation of a four stage skills training program focused on improving either recreational skills, vocational skills or residential skills for patients at eight institutions in Germany, Switzerland and Austria. Participants were assigned to the group that most addressed their goal of interest and matched for age, psychopathology, duration of illness and motivation. Group and individual sessions, in-vivo exercises and homework assignments were used to focus on clients' most frequent problems. Small to medium effect sizes for cognitive and social functioning for all three programs were found at three (recreational skills: effect size 0.35, CI not reported, vocational skills: effect size 0.40, CI not reported, residential skills: 0.51, CI not reported), six (recreational skills: effect size 0.48, CI not reported, vocational skills: effect size 0.47, CI not reported, residential skills: effect size 0.60, CI not reported) and 12-month (recreational skills: effect size 0.58, CI not reported, vocational skills: effect size 0.66, CI not reported, residential skills: effect size 0.73, CI not reported) follow-up.

The American Psychiatric Association states "SST may be helpful in addressing functional impairments with social skills of activities of daily living" [22] but it is not recommended by NICE [43].

A meta-analysis of five RCTs of cognitive remediation plus standard care found no benefit in terms of attention, verbal memory, visual memory, mental state or executive functioning over standard care alone [57]. However, in Pfammatter et al's [40] review of six meta-analyses small to medium effects of cognitive remediation on general cognitive functioning were found, as well as an indication of a possible transfer of these effects to social functioning. Through a further meta-analysis of 19 studies, cognitive remediation was found to have a small effect on attention (n = 539, effect size 0.32, CI 0.15, 0.49), executive functioning (n = 606, effect size 0.28, CI 0.12, 0.44), memory (n = 704, effect size 0.36, CI 0.20, 0.51) and social cognition (n = 228, effect size 0.40, CI 0.13, 0.68) [40]. A moderate transfer effect on social functioning (n = 306, effect size 0.49, CI 0.27, 0.70) and small reductions in overall psychopathology (n = 452, effect size 0.20, CI 0.01, 0.38) and negative symptoms (n = 394, effect size 0.24, CI 0.04, 0.44) were found. Descriptions of studies relevant to cognitive remediation can be found in Table 6.

More recently, Wykes et al [58] conducted a single blind RCT comparing outcomes for participants assigned to receive 40 sessions of cognitive remediation therapy with participants receiving standard care. A small effect on working memory was found (effect size 0.34, CI 0.1, 0.55) but there were no differences between groups in social functioning.

Gold et al [59] conducted a meta-analysis of four RCTs comparing music therapy for inpatients with a diagnosis of schizophrenia plus standard care with standard care alone. A minimum of 20 sessions was associated with significant improvement in positive and negative symptoms while findings for interventions with less than 20 sessions were inconclusive. Recipients of music therapy had significantly improved global functioning (n = 72, RR 0.10, CI 0.03, 0.31; NNT 2, CI 1, 2) and individuals receiving "high dose" music therapy (average 78 sessions) showed significant improvement in social functioning (Standardised Mean Difference -0.78, CI -1.27, -0.28). Descriptions of studies relevant to arts therapies can be found in Table 7.

A Cochrane review and meta-analysis of two RCTs of art therapy for people with schizophrenia found marginally beneficial effects on mental state but no effect on social functioning or quality of life [60]. The need for further RCTs was recommended. A Cochrane review of drama therapy identified five RCTs but, with minimal extractable data, no conclusions regarding efficacy could be made [61].

Integrated therapy, which incorporates psychosocial and pharmacological interventions, has been evaluated in a number of studies. Descriptions of these studies can be found in Table 8.

The most widely implemented model is integrated psychological therapy (IPT), a group-based CBT programme for people with schizophrenia, which integrates neurocognitive remediation with social cognition, problem solving and social skills training. Roder et al [62] conducted a meta-analysis of 30 studies of IPT, then a second meta-analysis using only the highest quality studies (n = 7) to determine whether or not the results would confirm the findings of the first meta-analysis. In comparison to standard care or placebo-attention control interventions, medium effect sizes were reported for participants who received IPT for global effect (N = 253, effect size 0.65, CI 0.39, 0.74) and psychopathology (N = 638, effect size 0.58, CI 0.39, 0.61), small to medium effect sizes were reported for functioning (neurocognition: N = 633, effect size 0.61, CI 0.43, 0.65; psychosocial functioning: N = 530, effect size 0.43, CI 0.29, 0.54) and small effect sizes for symptoms (positive symptoms: N = 424, effect size 0.42, CI 0.32, 0.60; negative symptoms: N = 277 effect size 0.46, CI 0.24, 0.57). Inpatients showed greater improvement at follow-up than outpatients (inpatient weighted effect size [at 10-month follow-up] 0.79, CI 0.43, 1.16 vs. outpatient weighted effect size [at 7.5-month follow-up] 0.44, CI 0.07, 0.80). Studies including only social skills training and problem solving sub-programmes showed no effect on neurocognition. Effects at follow-up were stronger when all five sub-programmes (cognitive differentiation, social perception, verbal communication, social skills and interpersonal problem solving) were provided. Longer term therapy had a beneficial effect on functional outcome. However, individuals with longer illness durations were less likely to benefit from IPT.

Treatment programmes that specifically combine pharmacological and psychosocial interventions (including supportive psychotherapy, family interventions, SST and CBT) have been reported to have additive positive outcomes for patient with schizophrenia [63].

The Illness Management and Recovery programme, a medication adherence intervention comprised of SST, coping skills training, recovery strategies, a relapse prevention plan and a cognitive behavioural approach, was developed by Gingerich and Mueser [64]. The programme was found to be beneficial in individuals receiving psychiatric care in non-residential community settings who reported high levels of satisfaction and found the programme to be useful, respectful, helpful for managing their symptoms and beneficial in making progress towards their goals [65].

In their review of treatments for people with severe mental illnesses and co-occurring substance misuse, Drake et al [66] concluded that the available evidence supported the integration of mental health and substance misuse interventions into a single care package. Patient-centred programmes were especially effective as they addressed the individual's stage of motivation for change. Programmes that incorporate some form of motivational counselling and outreach to engage the individual were recommended [66].

An integrated approach is supported by the APA [22]. Consensus expert guidance also endorses integrated treatment plus screening all patients with psychiatric symptoms for substance misuse and assessing the type, amount and mode of substance use; the client's motivation for change; physical sequelae of substance use, medication and psychosocial treatments including motivational therapy, modified CBT and relapse prevention [67]. Descriptions of studies relevant to the treatment of comorbid substance misuse can be found in Table 9.

Evidence-based clinical guidance and treatment algorithms assist clinicians in the choice and administration of antipsychotic medications [22, 43]. These are described in Table 10. These guidelines also stress the importance of discussion and negotiation with the patient about the choice of medication, changes in medication and regular review of its effects and side effects [22, 43].

A Cochrane review of compliance therapy for patients with schizophrenia [68] included only one RCT comparing it with non-specific counselling [69]. The manualised intervention was based on cognitive behavioural interventions and included aspects of motivational therapy, cognitive therapy and psychoeducation. No evidence of efficacy was found. Descriptions of studies relevant to compliance therapy can be found in Table 11.

Compliance therapy comprising four to six individual sessions exploring the pros and cons of patients' medications is more effective than supportive counselling [70], with advantages in terms of increased compliance (mean difference 19%; CI 0.9, 1.6), insight (mean difference 18.8%; CI 12.3, 25.2), attitudes towards medication (mean difference 15.6%; CI 2.5, 7.2) and global functioning (mean difference 2.4%) from baseline to 18 months after the end of the intervention.

A clinician administered, behaviourally-oriented programme on medication management and patient outcomes lead to improvements in patients' knowledge about medication, medication management and compliance [71]. Eckman et al [72] subsequently compared outcomes for male veterans receiving social and independent living skills training in groups or supportive group psychotherapy in an RCT. In group by trial analyses, individuals assigned to skills training showed significant improvement in medication management (F(1, 34) = 75.1, p < 0.0001) and symptom management (F(1, 32) = 36.23, p < 0.0001). Significant differences in medication management in favour of skills training were found between groups at six-month follow-up (F(1, 30) = 56.45, p < 0.0001) and 12 (F(1, 25) = 40.28, p < 0.0001). Significant differences between groups in favour of skills training were also found for symptom management at six-months follow-up (F(1, 28) = 6.34, p < 0.02) and 12 (F(1, 22) = 9.41, p < 0.005).

Whether psychoeducation on medication is structured or not seems to make no difference to patient knowledge about medication or positive attitudes [73].

An RCT comparing medication adherence for patients receiving either psychoeducation (medication information talks given by medical or nursing staff) or treatment as usual found significantly more control group participants were non-adherent at five-month follow-up based on pill count (N = 32, effect size 0.13, CI 0.03, 0.53, p < 0.001) and urine testing (N = 38, effect size, 0.22, CI 0.03, 1.73, p < 0.001) [74].

Streicker et al [75] investigated effects of a ten session medication education programme on participants assessed as non-adherent at baseline in a non-randomised controlled study and found that those who received the intervention gained knowledge about their medication but were more likely to be non-adherent post-intervention and at follow-up and as likely to be admitted to hospital as controls who received no intervention.

There is very little evidence on the effectiveness of occupational therapy in this patient group. Descriptions of included studies can be found in Table 12.

In a small (N = 26), non-blinded RCT carried out in Brazil, participants were assigned to six months of occupational therapy plus clozapine or clozapine alone [76]. Outcomes included activity, symptoms, social interaction and personal care. In an intention-to-treat analysis at six months, a combined rating of symptoms and social functioning showed significant advantage for those who received occupational therapy (effect size -1.44, p = 0.001).

Oka et al [77] retrospectively examined "before and after" outcomes over 17 years, for 52 patients with a diagnosis of schizophrenia discharged from a Japanese hospital between 1976 and 1990 who had taken part in inpatient occupational therapy, six days per week, in an integrated supported employment programme prior to discharge. Measures of social adjustment and rehospitalisation showed significant improvement pre and post-treatment.

In a Cochrane review, Buckley and colleagues [78] examined the effectiveness of supportive therapy for schizophrenia, defined as any "one-to-one" intervention which aimed to improve or maintain the patient's functioning (see Table 13 for study descriptions). There was no evidence that supportive therapy was more beneficial than standard care or other psychological or psychosocial therapies and, in fact, supportive therapy was more likely to cause social impairment (n = 39, RR 1.46, CI 1.0, 2.0; Number Needed to Harm (NNH) 4, CI 3, 39), treatment-related early termination (n = 151, RR 2.15, CI 1.1, 4.3; NNH 8, CI 3, 128), an episode of affective symptoms (n = 151, RR 1.84, CI 1.6, 2.9; NNT 5, CI 3, 27) and poor medication compliance between 13 to 26 weeks follow-up (n = 39, RR 2.63, CI 1.3, 5.4; NNT 3, CI 2, 12).

Significant differences favouring other psychological or psychosocial therapies were found: reduced chance of relapse (n = 39, RR 1.87, CI 1.1, 3.2; NNT 3, CI 3, 21) and hospitalisation (n = 241, RR 2.12, CI 1.2, 3.6; NNT 8, CI 4, 35) at six-month follow-up; greater general functioning at three months (n = 70, WMD -9.5, CI -16.1, -2.9) and six months (n = 67, WMD -12.6, CI -19.4, -5.8) follow-up; better general mental state (n = 194, RR 1.27, CI 1.0, 1.5; NNT 7, CI 4, 43), reduced symptoms (n = 12, WMD 17.10, CI 13.8, 20.4), better attitudes to medication (n = 74, WMD -4.50, CI -6.8, -2.2) and greater treatment satisfaction (n = 45, RR 3.19, CI 1.0, 10.1; NNT 4, CI 2, 736) were found at six-month follow-up [78].

The effectiveness of a coping skills training programme for service users comprising 24 twice weekly sessions (n = 55) was tested against treatment as usual (n = 44) in an RCT [79]. The approach consisted of training in seven components for coping with daily life: identification of symptoms; cognitive appraisal of stress, change and resources; selection of a coping strategy; use of strategy and evaluation of the results. Although no difference between groups was found, participants in the experimental group showed greater change scores over time as compared to the control group in hygiene (F(2,120) = 4.25, p < 0.05), self-esteem (F(2,140) = 3.08, p < 0.05) and delusions (F(2,132) = 3.16, p < 0.05). However, these findings are limited as the analyses were performed secondary to the main trial. Descriptions of relevant studies can be found in Table 14.

The effectiveness of a training programme focused on encouraging service users' self-esteem and empowerment was assessed in an RCT [80]. Although the intervention had no significant advantage compared to the control intervention, group by time analyses found a significant improvement in coping (F(1,70) = 4.01, p < 0.05). Positive symptoms were also significantly decreased in group by time analyses in the experimental arm (F(1,68) = 7.72, p < 0.01) but at six-month follow-up had reverted to baseline scores. The intervention was most beneficial in improving coping skills among participants with severe symptoms and lower functioning.