Long-term outcomes of assertive community treatment in Japan: 7-year follow-up of a randomized controlled trial

Abstract

Background

Within the context of community care, assertive community treatment (ACT) targeting individuals with severe mental disorders is typically a prolonged endeavor. Despite its extended nature, evidence on the long-term effects of ACT remains limited. This study aimed to assess the long-term effects of ACT services in Japan over 7 years by comparing readmission outcomes between the ACT and treatment-as-usual (TAU) groups.

Methods

This study traced participants from a prior randomized controlled trial (RCT) examining the short-term effects of ACT program in Japan. We assessed their readmission experiences, including the frequency and length (days) of readmissions over 7 years through medical record review. Multivariable analyses were conducted to examine outcomes by group.

Results

Of the 63 participants tracked for 7 years of follow-up (ACT group = 34, TAU group = 29), there were no significant group differences in readmission rate (TAU = 79% vs. ACT = 53%; odds ratio = 0.41, 95% CI = 0.12 to 1.35, p = 0.144) or cumulative days of hospitalization (Β = -75.66, 95% CI = -184.41 to 33.09, p = 0.173). However, the ACT group had significantly fewer hospitalizations (Β = − 1.03, 95% CI = − 1.97 to − 0.10, p = 0.030) than the TAU group. In particular, readmissions in the ACT group substantially decreased at 2 years after initial randomization.

Conclusion

In the Japanese medical landscape, an ACT program for individuals with severe mental illness might help mitigate the frequency of readmission compared with TAU over the long term. In particular, the pronounced benefits of the ACT program appear to emerge approximately 2 years after the start of ACT services. Future research in different countries is needed to confirm the findings of this study, particularly the timing at which the long-term effects of ACT services emerge for various outcomes.

Trial registration

The trial was retrospectively registered on 5 August 2025 with the Japan Registry of Clinical Trials (Registration number: jRCT1030250281).

Supplementary Information

The online version contains supplementary material available at 10.1186/s12888-025-07311-3.

Background

Effective long-term community care for individuals with severe mental illness is an international issue. Medications and psychotherapies often yield only modest effect sizes, rendering them insufficient for improving the lives of such people [1]. Consequently, after deinstitutionalization, several community care models have emerged, such as assertive community treatment (ACT), supported employment, and family psychoeducation [2, 3]. In particular, the ACT program, which originated in the United States, is an effective case management model. It has been a leading evidence-based practice in community care for over a half century. While ACT services are expected to effectively support individuals living in the community with severe mental illness [4, 5], evidence of its long-term effects remains scarce.

ACT is a well-structured program. The core elements consist of 11 key principles: multidisciplinary staffing, integration of services, team approach, low patient–staff ratios, locus of contact in the community, medication management, focus on everyday problems in daily life, rapid access, assertive outreach, individualized services, and time-unlimited services [6]. These elements are reflected in the ACT program's Fidelity Scale [7, 8]. Indeed, several countries have implemented ACT services and investigated its effectiveness. Systematic reviews have consistently underscored its robust international evidence to diminish the duration of rehospitalizations [4, 9]. A higher fidelity score is associated with fewer days of rehospitalization [4]. In other words, ACT services are a replicable package of programs, leading to its widespread adoption and endorsement as an effective community care model in many countries.

ACT programs work effectively in Japan. Although the Japanese medical system has many psychiatric beds and prolonged inpatient stays [10], community care has been developing gradually in the twenty-first century [11]. For example, the current Japanese system includes some community services such as counseling by social workers, brokering case management services, sheltered workshops, and visiting nurses [12]. In this context, ACT services were introduced in Japan with the dual aim of fostering deinstitutionalization of long-term inpatients and curtailing extended hospitalizations for new inpatients [13]. Indeed, a randomized controlled trial (RCT) evaluating ACT in Japan has highlighted its efficacy in mitigating depressive symptoms and enhancing patient satisfaction compared to usual care, further indicating its potential to reduce the duration of readmission [14]. Furthermore, ACT services are potentially associated with improved functioning, fewer antipsychotic medication prescriptions, and increased attention to familial concerns [15–17]. Collectively, these findings underscore the versatility of ACT services in addressing the multiple needs of patients and ensuring safe lives in the community.

Despite robust evidence for the ACT program, most prior research studies have predominantly focused on short-term effectiveness. A few studies have addressed the long-term effects, but the results have been mixed. For instance, three longitudinal observational studies spanning over 5 years suggest that ACT services might be associated with lower readmission rates, fewer hospital stays, and enhanced function in patients with severe mental illness [18–20]. A Danish study with 4 years of follow-up showed a reduction in inpatient care costs after implementing ACT services, while overall psychiatric service use costs were not significantly different from those for traditional care [21]. However, when Killaspy et al. [22] undertook a decade-long evaluation of RCT participants in the United Kingdom, they found no clinical advantages of ACT services relative to standard care. Their research also underscored challenges in evaluating ACT's long-term effectiveness, noting that individuals persistently availing ACT services over the 10-year duration typically presented with complex service needs. Given the limited evidence and heterogeneous findings from previous studies, further studies, particularly trial-based research, are required.

Community care for individuals with severe mental illness is not typically a short-term endeavor; ACT services are often provided over extended periods. Therefore, elucidating the long-term effects of ACT is a significant issue. To add more evidence to the existing literature, this study aimed to investigate the long-term effects of ACT services by following RCT participants in Japan for 7 years and comparing readmission outcomes between the ACT and usual care groups.

Methods

Overall design

This study was a 7-year extended outcome evaluation of a prior RCT that employed a parallel-group design with a 1:1 allocation ratio and a 12-month follow-up period. The RCT was conducted in the Chiba prefecture of Japan from 2004 through 2007, wherein participants were randomly allocated to either the ACT group or the treatment-as-usual (TAU) group. Allocation concealment for participants and ACT staff members was not feasible in this study. The findings of the original RCT with 12 months of follow-up were published elsewhere [14]. This study reported that the ACT program significantly improved depressive symptoms and service satisfaction compared to the control group, but did not demonstrate a significant reduction in the number of readmission days. In this study, a 7-year follow-up was conducted. Medical records were traceable for participants recruited from Konodai Hospital, a public psychiatric hospital managing the ACT team. Due to a transition in the hospital managing the ACT team, it was not possible to conduct follow-up beyond an 8-year period. Our methodological approach and subsequent results adhered to the guidelines set forth by the CONSORT statement [23, 24]. All study procedures were approved by the ethics committee of the National Center of Neurology and Psychiatry (no. A2021-017).

Participants

Patients admitted to Konodai Hospital, situated in Ichikawa city, were recruited to be study participants from May 1, 2004, to April 30, 2007. The inclusion criteria were as follows: 1) residence in one of the three cities close to the hospital; 2) age of 18 to 59 years; 3) primary diagnosis of schizophrenia, delusional disorder, mood disorder, or other psychiatric condition, as defined by the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [25]; 4) hospitalized on at least two occasions over the last 2 years with accumulation of over 100 days in hospitalization, used a psychiatric emergency department a minimum of three times, or withdrawal from outpatient treatment for more than 3 months, and 5) score of 50 or below on the Global Assessment of Functioning [26] within the last year. The inclusion criteria varied slightly on the basis of the primary diagnosis. Individuals diagnosed with schizophrenia or mood disorders were qualified for participation upon meeting either the fourth or fifth criteria, along with the initial three. Conversely, individuals diagnosed with other psychiatric disorders were required to fulfill all listed criteria for inclusion. The exclusion criteria consisted of dementia, intellectual disability, and personality disorder. During the recruitment period, 1,932 candidates were screened, of whom 202 met the eligibility criteria. Subsequent to acquiring their written informed consent, 118 were included in the study. In this study, to prevent contamination, participants in the control group who used ACT services after the end of the original RCT were excluded from the analysis.

Intervention

The intervention was provided by ACT-J, the pioneer ACT team in Japan. Established at Konodai Hospital in 2003, ACT-J transitioned to community-based services in 2008. It continued to provide vital ACT services to individuals with severe mental illness. Initially, the team comprised two nurses, four social workers, one clinical psychologist, one occupational therapist, one rehabilitation counselor, and one psychiatrist. Additional part-time members included two employment specialists, another psychiatrist, and a program manager. Following the end of a research grant in 2008, the team’s size was scaled down. Nevertheless, ACT-J consistently maintained a multidisciplinary team encompassing at a minimum nurse(s), social worker(s), occupational therapist(s), and a psychiatrist. Moreover, the team continued to offer extensive outreach services tailored to patient needs, delivering holistic support that covered areas such as strength-based assessments and planning, medication assistance, symptom management, daily life support, familial interventions, and employment-related services. During the initial 12-month follow-up period, participants in the ACT group primarily received community services from the ACT team. Subsequently, depending on their individual needs and reduction in symptoms, some participants transitioned to receiving services from other community service agencies.

ACT-J underwent a fidelity review based on the Dartmouth Assertive Community Treatment Scale (DACTS) [7]. During the research period (2004 to 2013), annual fidelity score, which represent the average score of all items from 1 to 5, ranged from 3.55 to 4.32. ACT-J fidelity scores tended to improve gradually during the study period. These scores surpassed the average score of 3.38 (SD = 0.69) obtained from 50 ACT teams in the United States at the time DACTS was developed. This suggests that ACT-J consistently upheld a certain quality of ACT service throughout the research period (Table 1).

Table 1.

Fidelity scores by year from 2004 to 2013

Fiscal year	2004	2005	2006	2007	2008	2009	2010	2011	2012a	2013
Average score	3.68	3.85	3.89	3.55	3.65	3.71	3.86	4.02	―	4.32

^aFidelity reviews were not conducted for all assertive community treatment teams across Japan

Participants in the control group accessed counseling services from the hospital’s social workers and availed community services distinct from ACT, as part of TAU services such as brokering case management services and sheltered workshops. The majority of these services required participants to visit either the hospital or community service agencies, with minimal emphasis on outreach services.

Measures

The outcomes of this study were participants'readmission experiences, including the frequency and length (days) of readmissions over a 7-year follow-up period. The first author (NS) retrospectively assessed outcome data using clinical records from Konodai Hospital and the ACT-J team. Socio-demographic information of the participants was sourced from the previous RCT [14].

Statistical analyses

We included the participants who could be tracked over 7 years in the analyses. We illustrated monthly readmission rates using a line graph. To examine readmission experiences, we used a logistic regression model. A significant age difference was observed between the two groups in the baseline assessment (p = 0.012). Alongside age, past hospitalization experience has been documented as predictive variable for readmission [27, 28]. These variables were incorporated as covariates in the logistic regression model. The frequency and length (days) of readmissions exhibited a Poisson distribution with noticeable outliers (Online supplementary figures). To address this, we used Poisson regression with robust standard errors to evaluate the number of readmissions, thereby mitigating the effect of outliers. Regarding the duration of readmissions, we used a generalized linear model (GLM) with robust standard errors. Both models included the following variables: group, age, and hospitalization days in the year prior to the baseline survey. Furthermore, we conducted sensitivity analyses for each model, adjusting for additional variables such as gender, diagnostic categorization (schizophrenia or other diagnoses), living situation, and social security. In addition, we summarized the status of the 12 participants who maintained engagement with ACT services over a 7-year follow-up period. Using medical records, we classified these participants into five categories on the basis of 1) reduction in symptoms, 2) living alone, 3) having issues engaging with other services, 4) low level of social activities, and 5) limited human interaction and loneliness. Next, we assessed the presence of these characteristics in each participant. All analyses were conducted using Stata version 16. Statistical significance was set at 5% (p < 0.05). Statistical analyses were carried out by the third author (SS), who was blinded to group allocations.

Results

Of the 118 participants enrolled in the original RCT, 2 in the ACT group and 3 in the TAU group withdrew from the study (Fig. 1). Moreover, nine participants in the TAU group who had utilized ACT services after initial randomization were excluded from the analysis. During the 7-year follow-up period, three participants in the ACT group and seven participants in the control group died. Additionally, follow-up data could not be obtained for 20 participants in the ACT group and 11 participants in the control group. Consequently, 63 participants (34 from the ACT group and 29 from the TAU group) were tracked for 7 years and included in the analysis. Within the ACT group, 12 participants (35%) continued to use ACT services, 17 (50%) transitioned to conventional medical or community services, and 5 (15%) ceased using ACT services during the 7-year follow-up. Suicide or unexplained death occurred in 2 participants (4%) in the ACT group and 7 participants (15%) in the TAU group.

Fig. 1.

Participant recrement flow diagram

Table 2 presents the baseline demographic and clinical characteristics of the participants. Within both the ACT and TAU groups, females constituted approximately half of the cohort. The majority of participants in both groups were diagnosed with schizophrenia. The ACT group had a significantly higher mean age (42.59 years, SD = 11.2) compared to the TAU group (35.53 years, SD = 10.46) (t = 2.583, p = 0.012). No significant differences were observed between the groups for other variables.

Table 2.

Characteristics of participants at baseline assessment

	TAU		ACT
	(n = 29)		(n = 34)		Test statistic	d	p
Sex, n (%)
Female	13	(44.83)	17	(50.00)	X2 = 0.168	1	0.682
Male	16	(55.17)	17	(50.00)
Age, Mean (SD)	42.59	(11.2)	35.53	(10.46)	t = 2.583	61	0.012
Diagnosis, n (%)
Schizophrenia	25	(86.21)	31	(91.18)	X2 = 1.925	2	0.382
Mood disorder	4	(13.79)	2	(5.88)
Organic, Including Symptomatic, Mental Disorders	0	(0)	1	(2.94)
Living situation, n (%)
Living alone	8	(27.59)	8	(23.53)	X2 = 0.136	2	0.712
Living with family	21	(72.41)	26	(76.47)
Marital status, n (%)
Not married	23	(79.31)	30	(88.24)	Fisher's exact =		0.492
Married	6	(20.69)	4	(11.76)
Social security, n (%)
Not received	24	(82.76)	31	(91.18)	Fisher's exact =		0.453
Received	5	(17.24)	3	(8.82)
Hospitalization in past 12 months, n (%)
Not have been hospitalized	18	(62.07)	21	(61.76)	X2 = 0.001	1	0.980
Have been hospitalized	11	(37.93)	13	(38.24)
Days of Hospitalization in past 12 months, Mean (SD)	24.59	(44.04)	37.79	(62.04)	t = 0.958	61	0.342
GAF, Mean (SD)a	46.44	(10.57)	46.64	(10.33)	t = 0.063	45	0.950

ACT assertive community treatment, GAF Global Assessment of Functioning, TAU treatment as usual

^an = 18 in TAU, n = 28 in ACT

Figure 2 shows the proportion of patients with readmission in both the ACT and TAU groups over a 7-year follow-up period by month. In the ACT group, approximately 15% of the participants were readmitted up to 20 months after discharge for the index admission. However, this proportion decreased to below approximately 5% afterward and slightly increased (5–10%) after 5.5 years (65 months) post-intervention. In contrast, in the TAU group, there was an increase in the hospitalization rate 2 years after discharge from the index admission, even reaching up to 20% temporarily (39 months and 61 months).

Fig. 2.

Monthly hospitalization rate after discharge from index admission

Concerning psychiatric hospitalizations over 7 years (Table 3), no significant differences in readmission rate were observed between the two groups (TAU = 79% vs ACT = 53%, odds ratio = 0.41, 95% CI = 0.12 to 1.35, p = 0.144). On the other hand, compared to the TAU group, the number of hospitalizations in the ACT group was significantly lower (Β = − 1.03, 95% CI = − 1.97 to − 0.10, p = 0.030). While there were fewer days of hospitalization in the ACT group compared to the TAU group, no significant differences were observed (Β = −75.66, 95% CI = −184.41 to 33.09, p = 0.173). Trends were similar in the sensitivity analyses.

Table 3.

Comparison of readmission outcomes between ACT and TAU during 7-year follow-up

	TAU		ACT		Statistics
	(n = 29)		(n = 34)		OR/Bab	95%CIs	P
Psychiatric readmission in past 7 years, n (%)
Had not been hospitalized	6	20.69	16	47.06	0.41	0.12, 1.35	0.144
Had been hospitalized	23	79.31	18	52.94
Number of psychiatric readmissions in past 7 years, Mean (SD)	3.66	7.29	1.44	2.20	−1.03	−1.97, −0.10	0.030
Length of psychiatric readmissions in past 7 years (days), Mean (SD)	177.83	423.83	71.15	133.03	−75.66	−184.41, 33.09	0.173

^aOR (odds ratio) from Logistic regression for psychiatric readmission

^bB (regression coefficient) from Poisson regression for the number of psychiatric readmissions and generalized linear model for length of psychiatric readmissions

Table 4 presents the status of 12 participants who continued to use ACT services over a 7-year period. Reduction in symptoms was observed in five of these participants, while approximately half showed reluctance to engage with other community services and maintained a low level of social activities such as employment. Notably, all 12 participants experienced minimal or no human interactions, leading to prevalent feelings of loneliness.

Table 4.

Status of 12 participants who continued using ACT services for 7 years

Participant	Reduction in symptoms	Living alone	Having issues engaging with other services	Low level of social activities	Limited human interaction and loneliness
A	X		X		X
B			X		X
C			X		X
D			X		X
E	X			X	X
F	X		X	X	X
G	X			X	X
H		X	X		X
I		X		X	X
J				X	X
K	X			X	X
L			X		X
Overall	41.7%	16.7%	58.3%	50.0%	100.0%

Discussion

This study examined the long-term effects of ACT on outcomes related to readmission over a 7-year follow-up period. The analysis included approximately 60% of the participants initially enrolled in the RCT. The findings indicate that an ACT program might help mitigate the frequency of readmissions, particularly 2 years after initial randomization. Although the mean number of readmission days for the ACT group was roughly 3 months less than that of the TAU group, no statistical significances were observed between the groups.

The findings of this study partly align with findings from prior research. In this trial, ACT services were not associated with a significant reduction in the number of readmission days. Previous research has yielded mixed results regarding the long-term impact of ACT on readmission [18, 20, 22]. Notably, a rigorously designed RCT conducted in the United Kingdom found no substantial differences in the length of readmission between the ACT and TAU groups [22]. Thus, ACT services might not necessarily lead to a reduction in the duration of hospital admissions over the long term.

Unlike a simplistic perspective on readmission duration, this analysis highlighted less frequent readmissions in the ACT group compared to the TAU group over the 7-year follow-up. We found a distinct longitudinal trend in readmission patterns between the groups. Within the first 2 years of the 7-year follow-up, the ACT group had a high readmission rate (approximately 15%), which decreased until year 5 of follow-up. By contrast, the TAU group persistently demonstrated a greater rate of hospitalizations than the ACT group, starting 2 years after discharge. This suggests that ACT services might effectively mitigate against frequent readmissions, with the impact possibly emerging later in the Japanese context.

The postponed effect of ACT services might be linked to the specific demographics of the participants in this study and the unique service structure in Japan. In general, individuals with severe mental illnesses have multifaceted and intricate needs [29–31], often face isolation in their communities [32], and struggle with forming therapeutic relationships in a short time frame [33]. Furthermore, service users in Japan might be unfamiliar with the comprehensive outreach services offered by a multidisciplinary team, given the absence of such community mental health services in the Japanese system [11]. As a result, ACT staff members might require an extended period to meet the diverse needs of participants and build a trusting relationship. These could be intertwined with the observed lag in the effects of ACT services in this study.

We also found that the readmission rate increased again in the ACT group after 5.5 years. There are three potential explanations for this finding. First, participants might face challenging socio-environmental transitions, such as securing employment or living independently during ACT service use. Indeed, detailed descriptions of participants who continued using ACT services revealed that despite some showing a reduction in symptoms, many struggled with engaging in services, social activities, and maintaining human relationships. Such social deprivation has been reported among Japanese patients with psychiatric admissions [34] and is internationally recognized as a potential factor associated with psychiatric readmission [35]. Therefore, these social aspects could have influenced the results of this study. Second, residing in the community can occasionally present challenges for these individuals due to prevalent stigmas and insufficient social support systems [36]. Third, organizational factors might have impacted the outcomes. The observed gradual improvement in fidelity scores suggests that structural changes within the ACT program had limited effects on the readmission rate during both the initial and final 2-year periods of the study. Instead, other organizational factors related to service quality, which were not measurable with the fidelity scale, might have influenced the outcomes. In other words, even with comprehensive ACT services, there might be intervals when the program fell short in fully facilitating continuous safe community living. However, the current study was unable to determine the precise causes and organizational factors associated with readmission, marking this as a topic for future exploration.

An unexpected result of this study was the higher rate of loss to follow-up in the ACT group compared with the control group, which might be associated with the specific characteristics of the ACT team and study design. In this trial, the ACT program was administered by a public hospital instead of a local government agency, and participant data was sourced from the hospital’s medical records. In Japan, the sharing of medical information between medical institutions is extremely limited by law. Consequently, we were unable to track participants who either transferred to other hospitals or moved out of the ACT team’s catchment area. These structural aspects of the study, rather than the inherent quality of ACT services, likely explain the higher attrition rate observed in the ACT group.

Strengths and limitations

The strengths of this research are twofold. First, we were able to track participants from the original RCT, enabling a comparative analysis between ACT and TAU services. Second, the study provided insights into the timing of readmissions among participants, going beyond a mere comparison of outcomes. Given that only a handful of studies have explored the long-term impacts of ACT services in comparison to TAU, this study enhances the understanding of when ACT services are most efficacious over extended periods.

We also recognize three major limitations of this study. First, the follow-up rate was relatively low. As our data were sourced from medical records, participants who either moved to a different hospital or ceased receiving medical services altogether were not incorporated into our analysis. Consequently, the sample size was limited, resulting in inadequate statistical power to clearly determine the long-term effects of ACT services. This limitation highlights the need for future replication with a larger trial. Second, outcome types were limited in this study. While hospitalization constitutes just one aspect of life for individuals with mental illness and might not always be necessarily a negative outcome [37], assessing a broader range of outcomes, such as patient-reported outcome measures, is imperative for comprehensively understanding the long-term impacts of ACT services [12, 34, 38]. The third limitation pertains to participant selection and contamination. Even though we excluded participants in the TAU group who had accessed ACT services after the original RCT ended, our analysis did incorporate participants in the ACT group who had sought other medical or community services outside of the ACT program (50%). This transition to alternative services complicates the evaluation of the long-term effectiveness of ACT services, posing challenges in controlling for potential confounders. On the other hand, prolonged use of ACT services is a common concern [39]. In addition, there were ethical concerns about retaining participants in the ACT program who no longer required its services. Therefore, following the initial RCT with a 1-year follow-up, no specific efforts were made to keep participants within the ACT program. Rather, this study showed that the Japanese ACT program can be considered a service from which patients can graduate. Consequently, those who continued with ACT were more likely to experience social deprivation.

Conclusion

This study evaluated readmission outcomes associated with ACT services in Japan over a 7-year period. ACT services are potentially more effective in preventing frequent readmissions compared to TAU services. In particular, the pronounced benefits of the ACT program appeared to emerge approximately 2 years after the start of the ACT service use. Future research in different countries is needed to confirm the timing at which the long-term effects of ACT services emerge for various outcomes.

Abbreviations

ACT

Assertive community treatment

CI

Confidence interval

CONSORT

Consolidated Standards of Reporting Trials

DACTS

Dartmouth Assertive Community Treatment Scale

GLM

Generalized linear model

RCT

Randomized controlled trial

TAU

Treatment-as-usual

Authors’ contributions

N.S. designed the study and collected the data. S.S. conducted data analysis. N.S., S.Y., and S.S. wrote the first draft of the manuscript. All authors contributed to data interpretation and approved the final draft of the manuscript.

Funding

This work was not supported by any academic grants.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The authors assert that all procedures contributing to this work complied with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The ethical considerations of the current study, including the informed consent process and patient privacy measures, were based on the ethics guidelines for medical research in Japan. The study protocol was approved by the ethics committee of the National Center of Neurology and Psychiatry (no. A2021-017).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.