1. A predictive model for readmission within 1-year post-discharge in patients with schizophrenia
- Author
-
Mingru Hou, Yuqing Wu, Jianhua Xue, Qiongni Chen, Yan Zhang, Ruifen Zhang, Libo Yu, Jun Wang, Zhenhe Zhou, and Xianwen Li
- Subjects
Hospital discharge ,Psychiatric readmission ,Schizophrenia ,Predictive model ,Nomogram ,Psychiatry ,RC435-571 - Abstract
Abstract Background Schizophrenia is a pervasive and severe mental disorder characterized by significant disability and high rates of recurrence. The persistently high rates of readmission after discharge present a serious challenge and source of stress in treating this population. Early identification of this risk is critical for implementing targeted interventions. The present study aimed to develop an easy-to-use predictive instrument for identifying the risk of readmission within 1-year post-discharge among schizophrenia patients in China. Methods A prediction model, based on static factors, was developed using data from 247 schizophrenia inpatients admitted to the Mental Health Center in Wuxi, China, from July 1 to December 31, 2020. For internal validation, an additional 106 patients were included. Multivariate Cox regression was applied to identify independent predictors and to create a nomogram for predicting the likelihood of readmission within 1-year post-discharge. The model’s performance in terms of discrimination and calibration was evaluated using bootstrapping with 1000 resamples. Results Multivariate cox regression demonstrated that involuntary admission (adjusted hazard ratio [aHR] 4.35, 95% confidence interval [CI] 2.13–8.86), repeat admissions (aHR 3.49, 95% CI 2.08–5.85), the prescription of antipsychotic polypharmacy (aHR 2.16, 95% CI 1.34–3.48), and a course of disease ≥ 20 years (aHR 1.80, 95% CI 1.04–3.12) were independent predictors for the readmission of schizophrenia patients within 1-year post-discharge. The area under the curve (AUC) and concordance index (C-index) of the nomogram constructed from these four factors were 0.820 and 0.780 in the training set, and 0.846 and 0.796 for the validation set, respectively. Furthermore, the calibration curves of the nomogram for both the training and validation sets closely approximated the ideal diagonal line. Additionally, decision curve analyses (DCAs) demonstrated a significantly better net benefit with this model. Conclusions A nomogram, developed using pre-discharge static factors, was designed to predict the likelihood of readmission within 1-year post-discharge for patients with schizophrenia. This tool may offer clinicians an accurate and effective way for the timely prediction and early management of psychiatric readmissions.
- Published
- 2024
- Full Text
- View/download PDF