Effectiveness of a Community-Based Intensive Case Management Model on Reducing Hospitalization for People With Severe Mental Illness in Seoul
Article information
Abstract
Objective
To overcome the limited effectiveness of standard case management services, the Seoul Intensive Case Management program (S-ICM) for patients with serious mental illnesses was introduced in 2017. This study aimed to evaluate its effectiveness in reducing the length of hospital stay.
Methods
Monitoring data from April 2019 to March 2020 were retrieved from the Seoul Mental Health Welfare Center. A total of 759 participants with serious mental illnesses were included. The average length of admission per month was compared between the pre-ICM (previous year) and during-ICM periods. For post-ICM observation subgroup, average length of admission per month was compared between pre-ICM, during-ICM, and post-ICM periods. To determine the relative contributions of risk factors for during-ICM and post-ICM admission, multivariate logistic regression analyses were performed.
Results
The average admission stay for pre-ICM period was significantly longer than that for during-ICM period (1.47 vs. 0.26 days). Among the predictors for during-ICM admission, pre-ICM psychiatric admission was the most important risk factor, followed by medical aid beneficiary and suicidal behavior. In the subgroup analysis of the post-ICM observation period, the pre-ICM, during-ICM, and post-ICM average admission stays were 1.45, 0.29, and 0.57 days/month, respectively. There was a significant difference in the average length of stay between the pre-ICM and during-ICM periods and between the pre-ICM and post-ICM periods. Post-ICM admission risks included pre-ICM admission, S-ICM duration <3 months, and chronic unstable symptoms.
Conclusion
The results suggest that the S-ICM effectively reduces psychiatric hospitalization duration, at least over a short-term period.
INTRODUCTION
Community mental health services were launched in Korea approximately 35 years ago. As of 2021, 244 community mental health & welfare centers serve more than 79,446 registered patients with mental disorders across Korea [1]. Incidentally, the number of psychiatric beds in Korea was also increased from 14,109 in 1990 to 75,474 in 2021, with most of this increase witnessed in the mental hospitals [1,2]. The simultaneous growth of community- and institution-based mental health services has a complex background. Low reimbursement rates for the psychiatric admissions, coupled with loose staffing and facility regulations for psychiatric admission facilities, pose as hindrances to improved quality of psychiatric treatment in the hospitals. Meanwhile, with only 134 psychiatric rehabilitation services for 4,727 patients and 229 residential services for 2,327 patients operating nationwide [1], there had been no alternative treatment options other than hospitalization for patients with severe mental illness.
Moreover, most community mental health & welfare centers in Korea had been overwhelmed with cases. In national data, the crude number of clients per case manager improved from 43.2 in 2017 to 26.6 in 2021 [1]. Despite the improvement, when adjusted to full-time-equivalent (FTE) numbers to represent real capacity, data from Seoul showed that the number of clients per FTE-case manager was still as high as 116.4 in 2021 [3]. According to the “clinical case management” model upon which the Korean case management system is based, the staff-to-patient ratio should ideally be 1:30. Even the basic broker model recommends a ratio of 1:50 [4]. Assuming sufficient staffing, non-intensive case management (non-ICM) models have shown mixed efficacy. Although a few studies reported reduced hospitalizations with non-ICM services [5], most others, including a meta-analysis, reported non-ICM services to be ineffective in shortening hospital stay [6-11].
Countries with well-established community mental health systems introduced intensive case management (ICM) programs, including Assertive Community Treatment (ACT), in the late 1970s [4,12,13]. In modern practice, case management packages with a staff-to-patient ratio <1:20 are categorized as ICM, regardless of the model, even though most ICM programs are based on ACT or a similar assertive outreach model [14,15]. In general, ACT and ICM have proven effective in reducing psychiatric hospitalizations, if not for reducing psychopathology [11,12,14-21].
In 2017, the Seoul Mental Health Welfare Center introduced the Seoul-ICM (S-ICM) to improve quality of care, and now community mental health & welfare centers of each district in Seoul are providing ICM services. Details of eligibility criteria for participation and the services provided by the S-ICM [22] are presented in Table 1. In this study, we evaluated the effectiveness of the S-ICM using monitoring data from April 2019 to March 2020.
Eligibility criteria and service principle of the Seoul Intensive Case Management program (S-ICM)
METHODS
Participants
The S-ICM monitoring data from the Seoul Mental Health Welfare Center from April 2019 to March 2020, were used for the analysis. These data were collected for the purpose of quality improvement and administrative management and contained information such as diagnosis, S-ICM start and end dates, frequency of patient interactions, psychiatric admission status, and clinical characteristics as eligibility criteria for program enrollment. These reasons of enrollment were collectively decided by the ICM-committee in each participating center, which was constituted by the multidisciplinary staff of the respective center, in consultation with the patients’ attending psychiatrists. The diagnosis of each patient was also obtained from the medical certificate issued by the attending psychiatrist. After the termination of S-ICM, patients received standard case management as usual, and were observed for post-ICM psychiatric admission. The post-ICM observation data, including psychiatric admission status were retrieved from the Seoul Mental Health Information System.
There were 1,059 S-ICM-completed cases during the study period. Of these, 73 patients aged <18 years and 118 patients aged >65 years were excluded. As adult case management services in the Korean community mental health are mainly concerned with the psychotic disorders (including International Classification of Diseases, Tenth Revision [ICD-10] codes F20–F29) and mood disorders (including ICD-10 code F30–F39), the 109 cases without any of these diagnoses were excluded. Cases with alcohol use disorders (n=75), anxiety disorders (n=13), attention-deficit hyperactive disorders (n=9), and unspecified mental disorders (n=12) were excluded. The final sample included 759 cases (age range: 18–65 years).
Data analysis
We examined the demographic factors, clinical characteristics, including diagnosis and reasons for enrollment, and service-related profiles of the study participants. Service-related profiles included S-ICM duration and the number and length of psychiatric admissions during the year preceding S-ICM enrollment (pre-ICM), during S-ICM (during-ICM), and after completion of S-ICM (post-ICM).
We used the Wilcoxon signed-rank test to compare average admission stay (days per month) and associated clinical characteristics of patients from the pre-ICM period to the during- and post-ICM periods. Significance of the multiple comparisons was determined by Bonferroni corrections. To determine the risks associated with during-ICM admission, we performed both univariate and multivariate logistic regression analyses with demographic and clinical characteristics as candidate predictors. Multivariate logistic regression model was constructed using the stepwise backward method.
To examine the post-ICM lingering effect, a subgroup of participants with data from a follow-up observation period >90 days after S-ICM completion was analyzed. Differences between observed and non-observed groups were analyzed using chi-square tests. The pre-, during-, and post-ICM hospital stays were compared using Wilcoxon signed-rank tests. To determine the risks of post-ICM psychiatric admission, we reperformed univariate and multivariate logistic regression analyses with demographic and clinical characteristics and intervention profiles including the S-ICM duration. Post-ICM multivariate model was also constructed using the stepwise backward method.
The study was approved by the Institutional Review Board of the Seoul Medical Center (IRB No. Seoul Medical Center 2020-12-005). Informed consent was waived by the Institutional Review Board of the Seoul Medical Center because of the retrospective nature of the study with the analysis using anonymous data. Statistical analyses were performed using the IBM Statistical Package for the Social Science, version 26.0 (IBM Corp., Armonk, NY, USA).
RESULTS
Demographic and clinical characteristics of the participants are presented in Table 2. Among the included participants, 40.5% were >51 years old, while 28.2% were <35 years old; 32.9% lived alone; and 58.1% were medical aid beneficiaries, indicating unemployment and lower socioeconomic status. Approximately half of the participants were diagnosed with psychotic disorders, and the other half with a mood disorder. Among reasons for S-ICM enrollment, 52.6% of the participants were enrolled due to chronic unstable symptoms, 17.8% for aggression, and 14.6% for suicidal behavior; first-episode status (9.1%) and frequent use of emergency services (5.1%) were the less common reasons. Chronic medical diseases were reported in 16.5% of participants. The median duration of S-ICM for the participants was 100.1 days. Approximately 50.3% of the participants had a 3–9-month-long service duration; however, 38.2% spent less than 3 months in the S-ICM. Another 11.5% had been in the S-ICM for more than 9 months. In the pre-ICM period, 21.6% of participants experienced more than one psychiatric hospitalization. From additional analysis, among participants with psychiatric hospitalization, 65.9% had a single admission, while 34.1% had multiple (2–5) admissions; furthermore, male participants were more likely than female participants to be psychiatrically admitted (χ2=4.207, p=0.04). With regards the during-ICM period, 3.4% of participants experienced psychiatric admissions, no difference was observed between males and females.
Demographic and clinical characteristics of the S-ICM study participants
The average number of admission days per month was 1.47 in the pre-ICM period and 0.26 in the during-ICM period, rates that were determined by Wilcoxon signed-rank test were comparable (Table 3). As shown in Table 3, the differences in hospitalization stay between the pre- and post-ICM periods persisted across patient diagnosis and clinical characteristics.
Comparison of mean psychiatric admission days per month between pre-ICM and during-ICM periods
In the univariate logistic regression analysis, pre-ICM psychiatric admission and medical aid beneficiary status were significant risk factors for during-ICM admissions, and no clinical characteristics predicted during-ICM admissions. In the subsequent multivariate logistic regression, pre-ICM psychiatric admission (prior to 1 year) remained as the most important predictor of during-ICM admissions (adjusted odds ratio [aOR], 36.24; 95% confidence interval [CI], 10.43–125.91), with medical aid beneficiary status (aOR, 8.29; 95% CI, 1.86–36.96). However, in the multivariate model, suicidal behavior before S-ICM enrollment emerged as a significant predictor of during-ICM admissions (aOR, 6.00; 95% CI, 1.76–20.43). Moreover, chronic unstable symptoms made it into the final model, but failed to demonstrate significance (aOR, 2.37; 95% CI, 0.84–6.68) (Table 4).
Univariate and multivariate logistic regression analyses for risk of during-ICM admission
For the subgroup analysis, participants with post-ICM observation >90 days were examined. The mean post-ICM observation period for all the participants was 198.5 (standard deviation [SD] 97.0) days, 230.2 (SD 75.2) days in the >90-day group, and 51.8 (SD 24.5) days in the <90-day group. Outside of S-ICM duration, there were no demographic or clinical differences between these two subgroups; in the >90-day group, a higher proportion of participants had spent either 6–9 months or >9 months in the S-ICM (Table 2). In this post-ICM observation group, pre-ICM, during-ICM, and post-ICM average admission stays were 1.45 (SD 4.19), 0.29 (SD 2.35), and 0.57 (SD 3.99) days/month, respectively. Wilcoxon signed-rank tests revealed a significant difference in average admission stay between the pre-ICM and during-ICM periods (z=-8.249, p<0.001) and between the pre-ICM and post-ICM periods (z=-7.089, p<0.001). No significant differences were observed between the during-ICM and post-ICM periods (Table 5).
Comparison of mean admission days per month between pre-ICM, during-ICM, and post-ICM periods (only including participants with more than 90-days post-ICM observation)
In the univariate logistic regression analysis, presence of pre-ICM psychiatric admission, less than 3 months of ICM service, medical aid beneficiary status, chronic unstable symptoms, and diagnosis of psychotic disorders were found to be significant risk factors for the post-ICM admissions. Subsequent multivariate logistic regression analyses revealed that the presence of pre-ICM psychiatric admission and medical aid beneficiary status were significant risk factors, consistent with the during-ICM findings. However, duration of S-ICM <3 months, compared to S-ICM duration >9 months, was found to be the most important risk factor for post-ICM admission (aOR, 11.24; 95% CI, 1.42–89.25). Intermediate service durations of 3–6 months and 6–9 months were not significant predictors. Among clinical characteristics, only chronic unstable symptoms predicted post-ICM hospitalization. Diagnosis of psychotic disorders was a significant factor in the univariate model, but not in the final multivariate regression model. Pre-ICM suicidal behavior, which was a significant risk factor for the during-ICM admission, was not a significant risk factor for the post-ICM admissions (Table 6).
Univariate and multivariate logistic regression analyses for risk of post-ICM admission
DISCUSSION
One of the primary objectives of community mental health services is to shorten psychiatric admissions. In this study, patients in the during-ICM period experienced shorter hospital stays than those in the pre-ICM period. Furthermore, the subgroup analyses showed that psychiatric admissions in the post-ICM observation period were shorter than those in the pre-ICM period. Despite the absence of control groups, these results suggest that S-ICM is effective, at least over short-term, as the participants in this study can be considered vulnerable to repeat psychiatric hospitalizations in the Korean healthcare setting. These results are congruent with those from a recent meta-analysis in the Cochrane database which indicated that the average hospital stay over a 2-year period was 0.86 days/month shorter in the ICM compared to that in the standard care. However, their review found no significant difference in the average number of hospitalization days per month between the ICM and non-ICM groups [14].
In our analysis, the most important risk factor for the during-ICM admissions was pre-ICM psychiatric admissions, followed by patient medical beneficiary status. Among clinical characteristics, only suicidal behavior before S-ICM enrollment predicted during-ICM admissions. Aggressive behavior was not identified as a significant risk factor in the univariate analysis. Furthermore, presence of chronic unstable symptoms made into the final multivariate model, but failed to achieve statistical significance. Therefore, the clinical characteristics for readmission, except suicidal behavior, appear to have been mitigated by the S-ICM, at least during the tenure of the program. As it remained an independent risk factor for psychiatric readmission in the multivariate model, pre-ICM admission status may represent important elements of a patient’s social context (e.g., level of interpersonal support at home) rather than the patients’ clinical context. Medical aid beneficiary status, which signals economic disadvantage, is also an independent risk factor for during-ICM admissions. Overall, these findings may point to the significance of socioeconomic inequity as an important determinant in psychiatric admissions, even with dedicated support such as the S-ICM.
During the post-ICM observation period, pre-ICM admission and medical aid beneficiary status remained important risks for re-hospitalization; however, the OR was much lower than that for during-ICM period. Hence, it can be postulated that if S-ICM runs for a sufficient duration (>3 months), it would be able to mitigate at least a part of the short-term readmission risk among the socioeconomically disadvantaged population. The results of the post-ICM multivariate logistic regression analyses suggest that the benefits of S-ICM may be lost in the post-ICM observation period in the patients with chronic unstable mental symptoms more easily than in the other patients. Hence, for this group of patients, longer S-ICM duration than that for the others may be recommended, though a long-term effectiveness study would be needed to confirm this speculation.
However, the most important risk for psychiatric admission in the post-ICM observation period was having participated in the S-ICM for less than 3 months. This result can also can be regarded as the evidence that S-ICM is more effective in reducing hospitalizations than the standard case management approach in Korea, since an earlier report determined that standard case management in Seoul may need at least 2 years to be effective in reducing psychiatric admissions [23].
As there is some evidence that heavy service use settings, such as high baseline psychiatric bed use, are related to the effectiveness or cost-effectiveness of ACT programs [14-16], the heavy psychiatric bed use in Korean psychiatric services might play a role in reducing hospitalization, even with a short-term S-ICM duration of <3 months. Although the average psychiatric stay was much reduced from 110 days in 2005 [24] to 28 days in 2021 1 , due to recent de-institutionalization policies and the 2019 coronavirus pandemic, the psychiatric services in Korea are still dominated by mental hospitals rather than community mental health services. One study speculated that, in Korea, not just ICM, even low-intensity standard case management has been effective in preventing hospitalizations due to this heavy service environment [23]. Contrarily, a meta-analysis including studies from other countries reported that standard case management did not help shorten hospital stay or improve mental, global, or social state [20].
This study has some limitations; the most important being that it was an observational study without control groups. Due to its low fidelity to ACT, the S-ICM needs to be studied more carefully, and randomized controlled trials are needed to confirm its effectiveness. In addition, this study only observed participants for approximately 1 year, when factoring together the S-ICM and post-S-ICM periods. Therefore, longer-term observations are required. Finally, we did not examine the S-ICM effect on other important outcome variables such as psychiatric symptom severity, quality of life, social and vocational functioning, legal or police contact, and suicide. However, this study does provide very encouraging, albeit preliminary, evidence of the effectiveness of S-ICM in Korea.
Notes
Availability of Data and Material
The datasets analyzed during the study are not publicly available due to privacy and ethical restrictions, but are available from the corresponding author on reasonable request.
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Jee Hoon Sohn, Sung Joon Cho, Hae Woo Lee, Seung Yeon Lee, Yoomi Park. Data curation: Hyun Kim, Seung Yeon Lee. Formal analysis: Jee Hoon Sohn. Investigation: Hyun Kim, Seung Yeon Lee. Methodology: Jee Hoon Sohn. Project administration: Hae Woo Lee, Yoomi Park. Supervision: Sung Joon Cho, Hae Woo Lee. Validation: Jee Eun Park, Bong Jin Hahm. Writing—original draft: Jee Hoon Sohn. Writing—review & editing: Sung Joon Cho, Hae Woo Lee, Hwo Yeon Seo, Eun Soo Kim.
Funding Statement
None
Acknowledgements
This study did not receive grant funding, although the data were collected during the development and refining stages of the S-ICM by the Seoul Mental Health Welfare Center. The authors thank the staff at the Seoul Mental Health Welfare Center and the Bureau of Citizens’ Health in the Seoul Metropolitan Government for their tireless efforts to improve mental health in the community.