Identifying Modifiable Risk Factors for Relapse in Patients With Schizophrenia in China

Weifeng Mi, Xiaohong Chen, Tengteng Fan, Serik Tabarak, Jingbo Xiao, Yongzhi Cao, Xiaoyu Li, Yanping Bao, Ying Han, Lingzhi Li, Ying Shi, Lihua Guo, Xiaozhi Wang, Yongqiao Liu, Zhanmin Wang, Jing Chen, Fengchun Wu, Wenbin Ma, Huafang Li, Weidong Xiao, Feihu Liu, Wen Xie, Hongyan Zhang, Lin LüFrontiers in Psychiatry 2020 · DOI

Abstract

For schizophrenic patients discharged from hospital, who had good medication adherence, more higher household income, be employed and normal ability of daily living, would be less likely to relapse. Decision tree provides a new path for doctors to find the schizophrenic inpatient's relapse risk and give them reasonable treatment suggestions after discharge.

Introduction

Relapse in schizophrenia is a major clinical hurdle, leading to increased healthcare costs, lost productivity, and worsening long-term prognosis. While clinicians know that factors like medication and family support matter, most existing predictive models are linear, failing to account for how different risks interact.

This study specifically addresses the Chinese context, where healthcare infrastructure and socioeconomic factors—such as the shift between self-payment and medical insurance—create a unique risk landscape for the nearly 1,500 patients tracked in this study. By identifying modifiable risk factors, the research aims to provide a clinical roadmap for post-discharge care.

Methodology

The researchers employed a multi-stage statistical approach to move from broad associations to a predictive hierarchy. Data was collected from 1,487 schizophrenia patients (ages 18–65) across 10 psychiatric hospitals in China.

Figure 1: Flowchart of the study design, illustrating the progression from initial screening to decision-tree modeling.

The methodology involved:

  1. Initial Screening: χ² (Chi-square) tests to find factors significantly associated with relapse.
  2. Logistic Regression: Identification of "high-risk" factors while controlling for covariates like gender and medication patterns.
  3. Decision-Tree Model (CHAID): The core contribution of this study, creating a branching hierarchy that splits the patient population into subgroups based on the most significant predictors.

Results

The study identified a clear hierarchy of risk, with medication adherence emerging as the single most powerful predictor of relapse.

Figure 2: Decision-tree model showing the hierarchical interaction of medication adherence, employment, and daily living skills.

Key Insights:

  • The Primacy of Adherence: Medication adherence is the "root" of the decision tree. The relapse rate for non-adherent patients (55.7%) was more than double that of adherent patients (22.9%). Non-adherent patients were approximately four times more likely to relapse (OR = 4.07).
  • The "Social Safety Net" Effect: For patients who are medication-adherent, employment is the next biggest predictor of staying well. For those who are unemployed, the critical factor is their "Ability of Daily Living" (ADL). This suggests that social skills training and vocational rehabilitation are primary defenses against relapse.
  • Socioeconomic Factors: Higher household income and self-payment of medical costs were associated with lower relapse risk. The latter may reflect higher personal investment in treatment or higher baseline wealth.

Discussion

The study emphasizes that while some risk factors are static, many are modifiable. Clinicians can actively intervene in medication habits, social skills (ADL), and vocational support to improve outcomes.

Limitations and Future Work

The study is subject to recall bias due to telephone follow-ups and relies on subjective relapse criteria. As an observational study, it establishes associations rather than direct causality. Future research should focus on intervention studies to test whether targeted vocational or social support actually reduces the relapse rates predicted by this model.

Summary updated

Abstract

For schizophrenic patients discharged from hospital, who had good medication adherence, more higher household income, be employed and normal ability of daily living, would be less likely to relapse. Decision tree provides a new path for doctors to find the schizophrenic inpatient's relapse risk and give them reasonable treatment suggestions after discharge.
BibTeX
@inproceedings{mi2020identifying,
  author = {Weifeng Mi and Xiaohong Chen and Tengteng Fan and Serik Tabarak and Jingbo Xiao and Yongzhi Cao and Xiaoyu Li and Yanping Bao and Ying Han and Lingzhi Li and Ying Shi and Lihua Guo and Xiaozhi Wang and Yongqiao Liu and Zhanmin Wang and Jing Chen and Fengchun Wu and Wenbin Ma and Huafang Li and Weidong Xiao and Feihu Liu and Wen Xie and Hongyan Zhang and Lin Lü},
  title = {{Identifying Modifiable Risk Factors for Relapse in Patients With Schizophrenia in China}},
  booktitle = {Frontiers in Psychiatry},
  year = {2020},
  doi = {10.3389/fpsyt.2020.574763},
}
Figure