Effectiveness of adjunctive traditional Chinese medicine in reducing psychiatric rehospitalization in patients with schizophrenia: A nationwide cohort study in Taiwan

Abstract

Background

Schizophrenia is a chronic psychiatric disorder marked by high relapse and rehospitalization rates. Although antipsychotic medications are effective, their adverse effects frequently compromise treatment adherence. Traditional Chinese medicine (TCM) is widely used as a complementary therapy in Taiwan, but real-world evidence regarding its long-term effectiveness remains limited.

Methods

Data were obtained from the Longitudinal Generation Tracking Database (LGTD), a two-million representative cohort within Taiwan’s National Health Insurance Research Database (NHIRD) spanning 2000 to 2021. Patients aged 18–65 years with newly diagnosed schizophrenia were identified. Following the application of exclusion criteria and 1:1 propensity score matching, 4578 patients were selected. TCM users were defined as those who received outpatient Chinese herbal medicine for >28 days. Outcomes included 5-year rehospitalization risk, length of stay (LOS), and admission frequency.

Results

The TCM group exhibited a significantly lower risk of schizophrenia-related rehospitalization (adjusted hazard ratio = 0.601; 95% CI, 0.482–0.863; p < 0.001). They also had shorter average LOS and fewer cumulative psychiatric admissions over 5 years.

Conclusions

Adjunctive TCM use was associated with better long-term outcomes in patients with schizophrenia. These findings support the potential value of incorporating TCM into standard psychiatric care. Further prospective studies are warranted to validate these results and elucidate the mechanisms contributing to these effects.

1. Introduction

Schizophrenia is a chronic psychiatric disorder with a global prevalence of about 1%, and is a major cause of disability worldwide.^{1, 2} Its positive and negative symptoms, cognitive impairment, and disorganization markedly impair daily functioning.^{3, 4} Hospitalization is warranted when severe psychosis or impaired self-care renders outpatient management unsafe or ineffective.^{5, 6}

Antipsychotics are the cornerstone of relapse prevention,⁷ but adverse effects such as weight gain, extrapyramidal symptoms, and sedation often compromise adherence and outcomes.^{1, 8} These limitations have led to increasing interest in complementary treatment strategies. In Taiwan, traditional Chinese medicine (TCM) is widely used as an adjunctive therapy in psychiatric care, delivered by licensed physicians and integrated into the National Health Insurance system.9, 10, 11

Preliminary studies have suggested that TCM may alleviate certain psychiatric symptoms12, 13, 14 and reduce the adverse effects associated with psychotropic medications.¹⁵ However, evidence specific to schizophrenia and its long-term clinical outcomes remains limited. The National Health Insurance Research Database (NHIRD), covering over 99% of the population and containing detailed longitudinal data, provides an ideal foundation for population-based research.16, 17, 18

This study evaluated whether adjunctive TCM reduces rehospitalization risk, a proxy for relapse severity and frequency, in patients with schizophrenia. The results are expected to clarify TCM’s potential benefits in schizophrenia care and support its integration into routine practice.

2. Methods

2.1. Data source

This study used data from the NHIRD in Taiwan.¹⁹ The National Health Insurance program was established in 1995 and provided coverage to over 23.88 million individuals by the end of 2023, representing more than 99% of Taiwan’s population.²⁰ We analyzed data from the Longitudinal Generation Tracking Database (LGTD), a nationally representative cohort of two-million individuals randomly sampled from the NHIRD. The dataset contains comprehensive longitudinal information on demographics, healthcare utilization, diagnoses, treatments, and prescriptions.

2.2. Study population and exposure definition

This study included patients newly diagnosed with schizophrenia between January 1, 2000, and December 31, 2021, based on LGTD data. For individuals who received TCM, exposure was defined as the date on which the cumulative duration of oral TCM use exceeded 28 days. This date was designated as the index date for exposed patients. To avoid immortal time bias, each exposed patient was matched 1:1 with a control who had not yet received TCM on the same calendar day. Controls were required to be alive, under follow-up, and free of schizophrenia-related rehospitalization. Each control was assigned the same landmark index date as the matched TCM user. Patients were matched on age, sex, Charlson Comorbidity Index (CCI), and calendar time. Matching incorporated psychiatric medication use, including first-generation antipsychotics (FGAs), second-generation antipsychotics (SGAs), and long-acting injectables (LAIs).

The exclusion criteria included a diagnosis before 2000, schizoaffective disorder, pregnancy, age <18 or >65 years, unknown sex, and absence of medical records during the 5 years after the index date due to death or loss to follow-up. Patients with schizoaffective disorder were excluded because their symptomatology and clinical course differ substantially from those of schizophrenia, including more severe affective symptoms and less pronounced negative symptoms.²¹ To minimize exposure misclassification, patients with common non-psychiatric indications for TCM use were excluded (Supplement 1).¹⁹ Both cohorts were followed for 5 years from their assigned index date to assess psychiatric rehospitalizations. Outcomes were defined as schizophrenia-related rehospitalizations, with non-psychiatric admissions excluded. These admissions may postpone but are unlikely to preclude subsequent psychiatric rehospitalizations. The study design and exposure definition are illustrated in Supplement 2.

2.3. Outcome measures

The primary outcome was the cumulative risk of schizophrenia-related rehospitalization within 5 years following the index day.^{22, 23} Time to first rehospitalization was the interval from the index date to the next psychiatric admission. Kaplan–Meier survival analysis was used to estimate cumulative incidence, with differences between groups assessed using the log-rank test. Secondary outcomes included length of stay (LOS) for the first rehospitalization and the cumulative number of schizophrenia-related readmissions.

2.4. Statistical analysis

Baseline characteristics between groups were compared using the chi-square test or Fisher’s exact test for categorical variables and independent t-tests for continuous variables. Time-to-event outcomes were analyzed using Kaplan–Meier survival curves with log-rank tests. Adjusted hazard ratios (aHRs) and 95% confidence intervals (CIs) were estimated using multivariable Cox proportional hazards models. Sensitivity analyses were also performed according to the medication possession ratio (MPR) of FGAs and SGAs.²⁴ To evaluate the dose-response relationship, TCM exposure was further quantified using the cumulative defined daily dose (cDDD). The cDDD was defined according to the total prescribed days of TCM during the follow-up period and was assessed in stratified Cox models. For repeated measures of hospitalization frequency and LOS during the 5-year follow-up, generalized estimating equations (GEEs) were used to account for within-subject correlations. All models were adjusted for age, sex, CCI, index date, and psychiatric medication use. A two-sided p-value <0.05 was considered statistically significant. Statistical analyses were conducted using SPSS version 27.0 (IBM Corp., Released 2020., Armonk, NY, USA).

2.5. Ethics

This study was approved by the Institutional Review Board of Tri-Service General Hospital (TSGH IRB No. E202516030). The need for individual written informed consent was waived by the committee.

3. Results

3.1. Baseline characteristics

The cohort selection process is illustrated in Fig. 1. A total of 14,794 patients diagnosed with schizophrenia between 2000 and 2021 were identified from the national database. After applying the inclusion and exclusion criteria, 9,163 individuals met the eligibility requirements. Among these, 2,289 patients who used TCM for more than 28 days were matched 1:1 with non-users, yielding a final matched cohort of 4,578 individuals. Non-users served as the control group in all subsequent analyses. Baseline characteristics were comparable between groups (Table 1). The only significant difference was geographic location, with TCM users being more likely to reside in northern Taiwan. When stratified by antipsychotic medication adherence, the aHRs for psychiatric rehospitalization decreased with increasing MPR in both FGAs and SGAs (Table 2). Across all adherence strata, patients receiving adjunctive TCM consistently exhibited lower risks of rehospitalization. Analysis by cumulative exposure yielded similar findings (Table 2). Increasing levels of cDDD were associated with progressively lower risks of psychiatric rehospitalization, indicating a significant dose-response relationship relative to non-TCM users.

Fig. 1.

Flowchart of cohort selection from the NHIRD in Taiwan. Abbreviations: NHIRD, National Health Insurance Research Database; TCM, traditional Chinese medicine.

Table 1.

Baseline characteristics of the study population.

Variables	Total		With TCM		Without TCM		P
	n	%	N	%	N	%
Total	4578		2289		2289
Sex							0.999
Male	1878	41.02	939	41.02	939	41.02
Female	2700	58.98	1350	58.98	1350	58.98
Age (yr)	49.26 ± 18.86		49.25 ± 18.86		49.27 ± 18.87		0.971
Age groups (yr)							0.999
18 - 24	336	7.34	168	7.34	168	7.34
25 - 44	2146	46.88	1073	46.88	1073	46.88
45 - 65	2096	45.78	1048	45.78	1048	45.78
CCI	1.02 ± 1.71		1.03 ± 1.72		1.01 ± 1.70		0.648
FGAs							0.471
Without	3043	66.47	1510	65.97	1533	66.97
With	1535	33.53	779	34.03	756	33.03
SGAs							0.251
Without	1083	23.66	525	22.94	558	24.38
With	3495	76.34	1764	77.06	1731	75.62
LAIs							0.798
Without	3803	83.07	1898	82.92	1905	83.22
With	775	16.93	391	17.08	384	16.78
Season							0.999
Spring (Mar-May)	1118	24.42	559	24.42	559	24.42
Summer (Jun-Aug)	1174	25.64	587	25.64	587	25.64
Autumn (Sep-Nov)	1156	25.25	578	25.25	578	25.25
Winter (Dec-Feb)	1130	24.68	565	24.68	565	24.68
Location							0.031
Northern Taiwan	1508	32.94	764	33.38	744	32.50
Middle Taiwan	1257	27.46	635	27.74	622	27.17
Southern Taiwan	1296	28.31	658	28.75	638	27.87
Eastern Taiwan	481	10.51	222	9.70	259	11.31
Outlets islands	36	0.79	10	0.44	26	1.14
Urbanization level							0.825
1 (The highest)	1238	27.04	611	26.69	627	27.39
2	1517	33.14	752	32.85	765	33.42
3	792	17.30	406	17.74	386	16.86
4 (The lowest)	1031	22.52	520	22.72	511	22.32

*P indicates Chi-square test or Fisher’s exact test for categorical variables, and t-test for continuous variables. Abbreviations: CCI, Charlson Comorbidity Index; FGAs, first-generation antipsychotics; LAIs, long-acting injectables; SGAs, second-generation antipsychotics; TCM, traditional Chinese medicine.

Table 2.

Schizophrenia rehospitalization associated with TCM use by antipsychotic adherence and cumulative TCM exposure.

Variables	aHR*	95% CI	p
Overall	0.617	0.489–0.895	< 0.001
FGAs (MPR)
< 20%	0.650	0.513–0.941	0.021
20-80%	0.627	0.497–0.910	0.005
> 80%	0.579	0.458–0.842	< 0.001
SGAs (MPR)
< 20%	0.663	0.526–0.962	0.031
20%-80%	0.619	0.490–0.897	< 0.001
> 80%	0.581	0.462–0.843	< 0.001
Cumulative TCM exposure (cDDD)
< 33%	0.646	0.512–0.938	0.019
33-65%	0.618	0.490–0.895	< 0.001
> 66%	0.592	0.463–0.861	< 0.001

^⁎aHR compares TCM use with non-TCM use (reference).

aHR, adjusted hazard ratio; cDDD, cumulative defined daily dose; CI, confidence interval; FGAs, first-generation antipsychotics; MPR, medication possession ratio; SGAs, second-generation antipsychotics; TCM, traditional Chinese medicine.

3.2. Risk of schizophrenia rehospitalization, length of stay, and rehospitalization frequency

Kaplan–Meier analysis demonstrated a significantly lower cumulative risk of rehospitalization in the TCM group (log-rank p < 0.001) (Fig. 2A). In the multivariable Cox proportional hazards model, the aHR for rehospitalization in the TCM group was 0.601 (95% CI: 0.482–0.863; p < 0.001), representing a 39.9% relative reduction in risk. The number of events and incidence rates are presented in Supplement 3A.

Fig. 2.

(A) Kaplan–Meier curves showing 5-year cumulative risk of schizophrenia-related rehospitalization. (B) LOS for first schizophrenia-related rehospitalization by follow-up year. (C) Cumulative number of rehospitalizations stratified by year of first relapse. Abbreviations: LOS, length of stay; TCM, traditional Chinese medicine.

The mean LOS for the first schizophrenia-related rehospitalization was consistently shorter in the TCM group across all follow-up years. Both groups exhibited significant within-group reductions over time (P for trend < 0.001). However, the difference in trend between groups was not statistically significant (P = 0.371) (Fig. 2B, Supplement 3B).

Over the observation period, the cumulative number of schizophrenia-related rehospitalizations remained consistently lower in the TCM group than in the non-TCM group (Fig. 2C, Supplement 3C). Patients were stratified based on the timing of their first rehospitalization relative to the index date. Both groups exhibited significant within-group increases in cumulative rehospitalizations over time (P for trend < 0.001), and the between-group difference was also statistically significant (P < 0.001).

4. Discussion

4.1. Summary of key findings

This study demonstrated that adjunctive TCM use was associated with a significantly lower risk of schizophrenia-related rehospitalization over 5 years. Patients receiving TCM also had shorter average LOS during rehospitalizations and fewer cumulative psychiatric admissions. Together, the results indicated that TCM can improve long-term stability and reduce healthcare burden in patients with schizophrenia. In addition, higher MPR of psychiatric drugs was linked to lower risks of rehospitalization, with greater benefits observed in the TCM group. Increasing cumulative exposure to TCM was correlated with progressively better outcomes. These findings underscore the significant benefits of integrated Chinese and Western medicine.

4.2. Comparison with previous studies

Reported rehospitalization rates for schizophrenia vary widely, ranging from 35.9% to 95.6% in the United States and up to 81.9% in Japan.^{25, 26} Taiwanese studies have reported 4-year rates of 62.1%²⁷ and 1-year rates of 61–65.9% in public psychiatric hospitals.²⁸ In contrast, our study found a substantially lower 5-year rehospitalization rate in the TCM group (13.2%) than in the non-TCM group (21.1%). This heterogeneity may reflect differences in study design, demographic characteristics, and healthcare systems, including the integrative nature of Taiwan’s insurance-covered TCM services and sample selection factors.^{25, 29} Beyond relapse prevention, the finding of reduced LOS corroborates prior reports that functional impairment predicts longer inpatient stays in psychotic disorders. It suggests that TCM helps accelerate stabilization and reduce healthcare burden.³⁰

4.3. Clinical and scientific implications

Despite persistent stigma toward psychiatric illness in many Chinese societies, 85% of outpatients report strong family support.³¹ This caregiving context aligns with TCM’s holistic principles and may support treatment continuity. LOS is driven by acute episode severity and system-level discharge constraints, which may mask slope differences.³² Cumulative rehospitalization mirrors relapse propensity, and the widening gap may reflect putative resilience-enhancing effects of TCM.^{33, 34} This mechanism may buffer stress-related relapse over time, with benefits more evident in long-term stability. Adjunctive TCM conferred substantial protection even in patients with poor antipsychotic adherence, suggesting a potential safety net in vulnerable populations.³⁵ The stronger association in highly adherent patients is compatible with incremental benefit beyond optimized pharmacotherapy.³⁶ Concordance across classes reduces concern that findings reflect a single subgroup. Joint evaluation of antipsychotic dose and TCM exposure intensity enabled interpretation beyond a simple use–nonuse comparison. However, these findings warrant cautious interpretation because sociocultural caregiving may contribute to a healthy-user effect. TCM users may have greater health awareness, more proactive care-seeking, and higher adherence, which could bias effect estimates.^{37, 38} Prospective studies and randomized controlled trials with finer longitudinal exposure measurement are needed to delineate trajectories and improve causal inference.

4.4. Strengths and limitations

To our knowledge, this is the first nationwide study to jointly evaluate the dose-dependent effects of both antipsychotic adherence and cumulative TCM exposure. This dual-dose modeling distinguishes the specific benefits of TCM from general medication compliance. However, several limitations should be acknowledged. The database lacked information on adherence, symptom severity, lifestyle factors, and family history. It also did not capture out-of-pocket or non-reimbursed TCM use. To ensure relevance to schizophrenia, visits with unrelated diagnoses were filtered out, although some of these conditions could represent antipsychotic side effects or comorbid symptoms. The analysis was limited to oral herbal medicines. Other modalities, such as acupuncture or moxibustion, were not included. Patients who died during the 5-year follow-up were omitted from the cohort, which may limit generalizability and partly explain the lower rehospitalization rates compared with other countries. In addition, other indicators of illness severity and treatment engagement, such as psychotherapy, community follow-up, and discharge planning, were not available in the NHIRD, which may have resulted in residual confounding.

4.5. Conclusion

This nationwide cohort study provides important real-world evidence supporting the integration of TCM into conventional psychiatric care. The observed dose-dependent patterns are consistent with incorporating TCM into routine maintenance regimens, prioritizing sustained stability over episodic symptom control. Current evidence supports TCM as a potentially synergistic adjunct, not merely an alternative therapy. Consequently, this study offers a pragmatic rationale for clinicians and public health researchers to consider holistic, integrative strategies in mental health management.

Author contributions

Conceptualization: WTS, NST, YCH. Formal analysis: CHC. Data curation: CHC. Writing – Original Draft: WTS. Writing – Review & Editing: WCC, YCH. Visualization: CHC. Supervision: NST, WCC, YCH. Project administration: YCH. Funding acquisition: NST, WCC, YCH.

Ethical statement

This study was approved by the Institutional Review Board of Tri-Service General Hospital (TSGH IRB No. E202516030). The requirement for individual informed consent was waived due to the use of de-identified secondary data from the National Health Insurance Research Database (NHIRD) of Taiwan.

Funding

This study was supported by grants from the Medical Affairs Bureau, Ministry of National Defense (MND-MAB-D-113059); the Tri-Service General Hospital Research Foundation (TSGH-A-114010, TSGH-B-114022, TSGH-E-113301, TSGH-E-114286, TSGH-E-115219); and a joint research program from the Cardinal Tien Hospital–National Defense Medical Center (CTH-113A-NDMC-2224). The sponsors had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflict of interest

The authors declare that they have no conflicts of interest.

Acknowledgments

We appreciate the Health and Welfare Data Science Center, Ministry of Health and Welfare (HWDC, MOHW), Taiwan, for providing access to the National Health Insurance Research Database (NHIRD).

Data availability

The data used in this study are not publicly available due to legal and ethical restrictions, as they contain sensitive and confidential information.