精神分裂症患者再入院风险预测模型的系统评价

Risk Prediction Models for Readmission in Patients with Schizophrenia: A Systematic ReviewYE Junjie1,HUANG Sirui1,HE Jiaojiao1,WANG Ying2

中文关键词:  精神分裂症  再入院  预测模型  系统评价

英文关键词:Schizophrenia  Readmission  Prediction model  Systematic review

中文摘要:

      背景 精神分裂症患者再入院率高，是精神卫生领域面临的重要临床挑战与社会负担。近年来，多种再入院风险预测模型被开发以辅助临床决策，但其性能与适用性仍需全面评估。目的 系统评价精神分裂症患者再入院风险预测模型，为未来开发出性能优良且适用性高的预测模型提供参考。方法 于2025年7月5日，计算机检索PubMed、Embase、Cochrane、Web of Science、CINAHL、中国知网、中国生物医学文献数据库、万方数据库和维普数据库中有关精神分裂症患者再入院风险预测模型的文献,检索时限为建库至2025年7月1日。由2名研究者按照纳入与纳排标准独立筛选文献和提取数据,并评价纳入文献的偏倚风险和适用性。结果 共纳入9篇文献,涉及18个风险预测模型。4个模型报告了建模的AUC(0.734～0.820),16个模型报告了内部验证的AUC(0.642～0.879),1个模型报告了外部验证的AUC(0.841)。核心预测因子包括病程和抗精神病联合用药。所有研究偏倚风险均为高风险。结论 精神分裂症患者再入院风险预测模型的研究尚处于探索阶段,模型预测性能良好,但偏倚风险较高,模型的预测性能评估显著不足。

英文摘要:

      Background Schizophrenia patients face high readmission rates, posing significant challenges to mental health services and societal resources. Accurately predicting readmission risk is crucial for early intervention and optimized resource allocation. Objective To systematically review risk prediction models for readmission in patients with schizophrenia,providing reference for the development of high-performance and highly applicable prediction models in the future. Methods On July 5, 2025, a comprehensive search was conducted in PubMed, Embase, Cochrane Library, Web of Science, CINAHL, CNKI, CBM, WanFang Data, and VIP databases for studies on risk prediction models for readmission in schizophrenia patients, from database inception to July 1, 2025. Two researchers independently screened the literature according to inclusion and exclusion criteria, extracted data, and assessed the risk of bias and applicability of the included studies. Results A total of 9 studies were included, involving 18 risk prediction models for schizophrenia patient readmission. Four models reported the area under the receiver operating characteristic curve (AUC) during model development, ranging from 0.734 to 0.820; 16 models reported AUC values for internal validation (range: 0.642–0.879), and one model reported an AUC for external validation (0.841). Key predictors included disease duration and combination therapy with antipsychotic medications. All included studies were assessed as having a high risk of bias. Conclusion Research on risk prediction models for readmission in schizophrenia patients is still in its exploratory stage. While the models demonstrate good predictive performance, they exhibit a high risk of bias and significant limitations in predictive performance evaluation.