Residential program for long-term hospitalized persons with mental illness in Japan: Randomized controlled trial


Shinji Shimodera, MD, PhD, Department of Neuropsychiatry Kochi, Medical School, Kohasu Oko-cho, Nankoku, Kochi 783-8505, Japan. Email:


Abstract  Research on the merits of long-term group residences is inconclusive. The purpose of the present paper was to investigate the effects of supported group residence on the symptoms, social function, quality of life, general health quality, and the medical/psychiatric cost in Japan of a large number of psychiatric beds and long average length of stay. Patients were assessed every 6 months for 2 years using Positive and Negative Syndrome Scale, Katz Adjustment Scale, World Health Organization Quality of Life (WHO-QOL) and General Health Questionnaire 12-item version. Patients discharged to the supported group residence (SGR) significantly improved with regard to positive symptoms, the level of socially expected activities and free-time activities. The QOL physical domain of the inpatients was significantly more deteriorated compared to the SGR group. The total psychiatric/medical cost of the SGR group was approximately one-third that of the inpatient group, while the cost of the SGR to treat physical comorbidity was much higher. The present findings indicate that SGR has advantages for mental and social function but not for physical health. A major limitation of the present study was the high mean age (>60 years) of the subjects who had been hospitalized for a long period (mean, 24 years).


In most Western countries for several decades there has been a policy of deinstitutionalization involving discharge of patients from psychiatric hospitals into the community. This social movement has generated much related research. Earlier community care studies mainly focused on the rate of psychiatric bed use, the rate of service use in psychiatric hospitals, or changes of symptoms,1 and found that community care has lowered the rate of re-hospitalization2 and improved clinical symptoms.3 In addition, some studies showed that community care could be provided at a lower cost.4,5 The patients' or relatives' satisfaction with treatment was another area of interest, and it was shown that they much preferred community care to traditional hospital care.6,7 More recently, studies have been refined, further showing the predominance of community care in cost-effective studies8,9 and the quality of life (QOL) of patients and relatives.10 Studies of educational intervention have also enforced the flow from hospital to community life.11,12 Although research findings on community care have not been numerous in Asia compared to the West, there were some studies modeled on the Western approaches in community care. Shu et al. found that total QOL in patients receiving home care programs was higher than those receiving half-way house services,13 while Chan et al. showed that community-based treatment settings had a positive impact on objective QOL.14 In other studies using psychosocial interventions, long-term hospitalized patients were successfully discharged through the Community Re-Entry Program,15,16 or family interventions could prevent patients from having to undergo re-hospitalization.17,18 Japan has just begun to reduce the number of psychiatric beds and to try to transfer long-term hospitalized patients into community settings. The Japanese psychiatric care delivery system has been characterized as having many beds, many closed wards, a long average length of stay, and poor community resources.19,20

After 1994, however, the number of beds has been gradually decreasing with more programs of psychiatric rehabilitation introduced for the inpatients,21 which aroused interest regarding the effects of housing services on chronic patients.

The purpose of the present study was to investigate the effects of a supported group residence (SGR) on symptoms, social function, quality of life, and the general health quality of long-term hospitalized patients, and to compare the psychiatric/medical cost between hospital and residential care. Such studies would be significant not only in examining deinstitutionalization in an Asian country, but also in the understanding of relevant community care in general.



The patients who had been hospitalized in Tosa hospital, a private psychiatric hospital with 229 beds, for ≥1 year were selected and included in the present study. They had to be between 30 and 80 years old, without serious symptoms, such as being dangerous to oneself or others, persistently troubling others or showing bizarre behavior, and not requiring regular nursing. The subject selection criterion was patients with schizophrenic disorders including epileptic psychoses. Patients with personality or depressive disorders were excluded. There were 30 patients who met this standard. After the purpose of the present study was explained and the consent form was presented, 28 patients agreed and signed the form (16 male, 12 female). They were randomly allocated into a group moving to the SGR (n = 14) or a group continuing in hospital (n = 14).

Because the number of rooms for men and women differed in the residence, random distribution was performed according to sex. Consequently, in the SGR group, there were 10 men and four women, while six men and eight women continued in the hospital. The diagnosis of patients was DSM-IV schizophrenia in 27 and epileptic psychoses in one patient who belonged to the hospital group. The DSM-IV diagnosis of the patient with epileptic psychoses was psychotic disorder due to epilepsy.


There are several housing programs for persons with severe mental illness in Japan. Starting in 2000, Fukushi-home B (Japanese name of SGR)is a facility for people with psychiatric symptoms such as avolition or bizarre behavior, and/or those who need some daily support due to aging. The Fukushi-home B system was established for patients with volitional disorder or mild deviant behavior who often have difficulty living in group homes, or who require assistance due to aging. There are no special rules for the provision of meals. In the facility in the present study, meals were provided at patients' request. The institution standard is to maintain >23.3 m2 per occupant, to equip the consultation room, cooking room, manager's office and other key rooms, to appoint one manager, one or more physicians, three guides, including one psychiatric social worker (Seishin-hoken-fukushishi), to have a capacity of approximately 20, to provide programs, such as helping the patients acquire daily life skills such as cleaning or washing, and give advice on interpersonal relationships.

Before the SGR patients moved to the facility, all patients, including the comparative group, received training to acquire some basic skills, such as taking medication, money management, or personal self-care. The programs for the SGR group consisted of volunteer work, such as cleaning the neighboring park, which might help to promote good relationships with the local residents, tea meetings/birthday parties at the SGR to strengthen friendships, and attending day care programs at Tosa Hospital three times a week. Because the staff working hours were only during the day, the patients were instructed to use a direct phone line to the Tosa Hospital in the case of an emergency at night.

When their psychiatric symptoms resurged after moving to the SGR they were usually readmitted to Tosa Hospital, but for physical illness they visited another hospital. The patients of the hospital group attended the occupational therapy department three times a week and the day care department once a week. We followed them for 2 years. The SGR of this case is in the town outside the site of the hospital.


The Positive and Negative Syndrome Scale (PANSS)22 was used to assess patient symptomatology.

The interrater agreement was excellent (κ = 0.84). To evaluate social function we used the Katz Adjustment Scale (KAS).23 There are five evaluation domains in the KAS, and in the present study we used 16 social activity items and 22 leisure activity items and evaluated the actual activity and the degree of expectation separately, for 38 items in two domains, to give a total of 76 items. We asked the staff in the SGR and the hospital to substitute for the relatives in the family version. For assessment of the quality of life and general health condition of the patient, we administered World Health Organization Quality of Life (WHO-QOL)24 and the General Health Questionnaire 12-item version (GHQ-12),25 respectively. For GHQ score calculation we used (0-0-1-1) for the grading method.

We used PANSS, KAS, WHO-QOL and GHQ-12 every 6 months for 2 years. Concerning the blindness of the assessment, we ensured that the PANSS rating was made by a psychiatrist unaware of the group allocation.

The dosage of the prescribed antipsychotic(s) was converted into chlorpromazine equivalent based on the power value conversion table by the treatment resistance schizophrenia research group and the equivalent conversion table of an oral antipsychotic by the 2001 version Keio University Psychopharmacology Research Group.26 We compared the dosage between baseline and at 2 years.

Regarding the cost, we initially planned to examine both psychiatric/medical treatment cost and living cost. We excluded the cost of living, however, because the expense of psychiatric/medical care was precisely and easily available, while calculation of the living cost was complicated and inaccurate. Further, compared to the psychiatric/medical cost, the living costs were limited to the purchasing of cigarettes or sweets in the present study. Patients' psychiatric/medical services were recorded continuously, and data were collected monthly to calculate psychiatric/medical treatment costs for 2 years.

Statistical analysis

The χ2 test was used to compare the differences by gender, and Student's t-test was used to compare age, duration of hospitalization, and antipsychotic dose at baseline between the two groups. Repeated measure anova (times × groups) was used to compare PANSS, KAS, WHO-QOL, and GHQ-12 scores and the dose of antipsychotic(s) between baseline and at 2 years. All statistical analysis was carried out using SPSS for Windows, version 12.0 (SPSS, MapInfo, Troy, NY, USA).


There were no significant differences in sex, age, and duration of hospitalization between the SGR and hospital groups (Table 1). During the follow-up period of 2 years, three out of 14 patients in the SGR group were readmitted to Tosa Hospital. The duration of re-hospitalization of the three patients was 4.9 months, 2.9 months and 0.5 months, respectively. For the three re-admitted patients in the SGR group, the hospitalization cost was added to the ‘medical cost in Tosa Hospital’ in the SGR group for analysis. Therefore, the actual hospitalization cost in each patient was added, and the living cost in Fukushi-home B was also added because they did not go through the cancellation procedure. Meanwhile, all patients in the hospital group continued hospitalization for 2 years.

Table 1.  Comparison of patients characteristics in SGR and hospital groups
 SGR group (n = 14)Hospital group (n = 14)Significance
  • † 

    Student's t-test;

  • ‡ 

    χ2 test

  • SGR, supported group residence.

Age63.01 ± 7.9561 ± 9.41n.s.
Duration of hospitalization (years)24.23 ± 15.7024.18 ± 16.73n.s.
Gender (%)
 Female4 (29)8 (57)n.s.
 Male10 (71)6 (43)n.s.

Changes in psychiatric manifestations

Table 2 shows the mean value of the PANSS subscale scores at baseline and at 2 years. There was a significant improvement in positive syndrome in the SGR group (F = 6.43, P < 0.05) and a trend toward aggravation of negative syndrome in the hospital group (F = 3.54, P = 0.07).

Table 2.  PANSS, KAS, WHO-QOL, and GHQ-12 scores (mean ± SD)
 SGR groupHospital groupAnalysis
Baseline2 yearsBaseline2 yearsFP
  1. GHQ-12, General Health Questionnaire 12-item version; KAS, Katz Adjustment Scale; PANSS, Positive and Negative Syndrome Scale; R-LEPSA, relative's rating of level of expectations for performance of social activities; R-LPFA, relative's rating of level of performance of free-time activities; R-LPSA, relative's rating of level of performance of socially expected activities; R-LSFA, relative's rating of level of satisfaction of free-time activities; SGR, supported group residence; S-LPSA, subject's rating of level of performance of socially expected activities; S-LSPSA, subject's rating of level of self-expectation for performance of social activities; S-LPFA, subject's rating of level of performance of free-time activities; S-LSFA, subject's rating of level of self-satisfaction of free-time activities; WHO-QOL, World Health Organization Quality of Life.

 Positive syndrome17.7 ± 4.113.6 ± 4.718.5 ± 3.017.6 ± 3.26.430.02
 Negative syndrome24.4 ± 6.423.5 ± 6.127.6 ± 3.930.4 ± 4.23.540.07
 General psychopathology41.7 ± 7.142.1 ± 5.946.6 ± 4.449.9 ±
 S-LPSA13.7 ± 2.124.9 ± 6.414.4 ± 1.917.7 ±
 S-LSPSA16.1 ± 3.526.1 ± 6.617.2 ± 3.021.6 ± 9.92.420.13
 S-LPFA25.4 ± 3.629.6 ± 6.424.0 ± 4.424.7 ±
 S-LSFA20.2 ± 5.722.9 ± 7.718.9 ± 5.718.8 ± 7.80.610.44
 R-LPSA13.4 ± 2.319.4 ± 4.112.1 ± 2.615.1 ±
 R-LEPSA14.9 ± 3.021.6 ± 3.614.6 ± 1.415.7 ±
 R-LPFA24.0 ± 4.028.1 ± 5.321.6 ± 3.019.9 ±
 R-LSFA27.7 ± 4.328.4 ± 4.024.9 ± 5.923.0 ±
 Physical21.0 ± 4.921.4 ± 4.623.1 ± 4.019.8 ± 3.44.390.046
 Psychological16.6 ± 4.317.9 ± 4.817.9 ± 9.015.9 ±
 Social7.9 ± 2.48.4 ± 2.39.1 ± 2.210.0 ±
 Environment24.8 ± 3.924.1 ± 6.224.6 ± 5.023.0 ±
 Life satisfaction5.6 ± 1.55.9 ± 1.86.3 ± 1.35.8 ±
GHQ-122.6 ± 2.51.6 ± 2.42.5 ± 2.53.1 ± 2.60.640.43

Changes of social function

As presented in Table 2, the level of performance of the socially expected activities assessed by the subjects was more enhanced in the SGR group after 2 years (F = 9.22, P < 0.01). In addition, the level of both expectations for performance of social activities and performance of free-time activities assessed by the staff was also more enhanced in the SGR group (F = 6.28, P < 0.05, F = 6.14, P < 0.05, respectively) and level of performance of social activities assessed by the staff tended to increase in the SGR group (F = 3.8, P < 0.06).

Change of QOL

Table 2 also presents the change of QOL measured with WHO-QOL and shows that the score of the hospital group significantly declined in the physical domain (F = 4.39, P < 0.05). No differences were found in other domains.

Changes to general health condition

There was no significant difference in the general health condition measured with GHQ-12 between the two groups (Table 2).

Comparison of dosage in antipsychotic

When the dose of antipsychotic(s) was compared at baseline, there was no difference between the two groups. The mean ± SD dose of the SGR group was 568.8 ± 292.42 mg, while that of hospital group was 508.2 ± 312.4 mg. After 2 years the doses in the SGR and the hospital group were 562.0 ± 294.7 mg and 439.8 ± 396.6 mg, respectively (no significant difference; F = 4.24, P = 0.61).

Comparison of psychiatric/medical cost

As is shown in Table 3, there was a significant difference in the psychiatric/medical cost in Tosa Hospital between the two groups. The expense of the hospital group was more than threefold greater than that of the SGR group. In contrast, the SGR group cost excluding Tosa Hospital was more than 10-fold higher than the hospital group.

Table 3.  Psychiatric/medical costs in the first and second years (mean ± SD)
 SGR groupHospital groupSignificance
  1. Yen per person per month. SGR, supported group residence.

Psychiatric/medical cost in Tosa Hospital
 First year128 522 ± 10 435414 866 ± 28 983<0.000
 Second year124 883 ± 24 918445 512 ± 100 771<0.000
Medical cost excluding Tosa Hospital
 First year14 074 ± 24 2711 318 ± 49310.065
 Second year34 646 ± 37 2152 480 ± 80590.004
Total302 125 ± 66 149864 177 ± 95 086<0.000


The present study compared psychopathology, social function, QOL, general health condition, dose of prescribed antipsychotic(s) and psychiatric/medical cost for long-term hospitalized patients moving to SGR with those of the patients continuing hospitalization. The patients receiving care in the SGR improved in their positive symptoms, performance of socially expected activities, expectations for performance of social activities and the performance of free-time activities and QOL physical domain. The present results suggest that SGR is a suitable facility for aged persons with chronic and stable mental illness without serious physical complications.

The results of the improvement in positive symptoms do not agree with previous findings4,6,27 and were contrary to expectation. Still, there may be some interpretations; first, there were a few reports that showed the improvement of psychotic symptoms through some psychosocial treatment28 or social environmental treatment.29 Patients in the present study had been hospitalized for an average of >24 years, and their psychopathology might have been greatly affected by the change of the therapeutic environment. Second, conducting a national survey on Japanese psychiatric hospitals based on the method of Wing and Brown,30 Oshima et al. demonstrated that there was a weaker correlation between positive symptoms and the social environment of psychiatric hospitals in Japan.31 Although their study was concerned with negative symptoms, positive symptoms might also be related to the hospital environment given the severe and significant degree of ward restrictions and the understimulating social environment. Third, there were many staff members involved in the evaluation process, such that perfect blindness of evaluation could not be maintained.

The results of the improvement in social function among the SGR group agree with the previous findings.32,33 The present findings suggest that the SGR patients had greater contact with the neighboring society than the patients in hospital, resulting in on-site training for development of interpersonal and occupational skills for the SGR patients. Further, the situation whereby the SGR patients were left without duty staff at night might promote their independence, leading to an increase of social function.6

In addition, it is easy to consider the possibility of assessment bias of the SGR staff, in that they wanted to see the patients living more actively at the residence.

The finding that the patients in hospital group had a poorer QOL physical domain is consistent with the previous report. For example, Anderson and Lewis showed that intermediate care facility patients reported higher QOL scores than state hospital patients.34 That the SGR patients might have been spending their social and leisure time more actively than hospital patients may have led to the difference of the QOL physical domain. The inpatients, however, seemed to be more physically healthy by reasoning from the difference of medical cost. There may be some discrepancy between the subjective QOL and objective condition.14

The fact that the total psychiatric/medical cost was smaller in the SGR group is not contradictory to the previous studies.4,5,8 However, it was noted that the medical cost, excluding Tosa Hospital, was much greater in the SGR group, suggesting that this group was much more vulnerable to physical complications. Those continuing hospitalization were regularly monitored by nursing staff, and their physical complications were recognized early and treated. In the long run, the medical cost becomes much greater for the SGR group and their life expectancy might be shorter. The management of physical health in community institutions such as SGR is a challenge for community care.

The present study had some limitations. The first is the high mean age (>60 years) of the subjects who had been hospitalized for a long period (mean, 24 years). The second is that because many investigators were involved, independent assessment might not have been assured during the assessment stage. Third, the small number of subjects weakened the statistical power, leading to failure to find significant differences of negative symptoms or other related variables.


We investigated the effects of SGR on the symptoms, social function, QOL, general health quality, and the medical/psychiatric cost and showed that the patients receiving care in the supported group residence improved in their positive symptoms assessed with PANSS, performance of socially expected activities, expectations for performance of social activities and the performance of free-time activities assessed with KAS, and QOL physical domain assessed with WHO-QOL. The total psychiatric/medical cost of the SGR was approximately one-third of the inpatient group, while the cost of treating physical complications for the SGR group was much higher. The present findings suggest that SGR has advantages for mental and social function but not for physical health.


The authors express their thanks to Dr Koichiro Suto, chief director of Tosa Hospital and Ms Hiroko Itoh, Manager of Community Liaison Room of the hospital for allowing access to patients in their care.