Risk prediction model development for requiring unplanned psychiatric readmission in bipolar disorder

The high readmission rate of bipolar disorder (BD) has imposed a heavy burden on the country and patients. Current studies mainly examine readmission influence factors but neglect risk prediction models. This study aims to develop a prediction model requiring unplanned psychiatric readmissions (RUPR) within 1 year of BD. We screened the third people's hospital of tianshui inpatients between January 1, 2021 and November 10, 2023 via hospital records and phone follow-ups, collecting required participant data. Based on their demographic and scale score, multiple variable logistic regression analysis was used to develop a nomogram-based prediction model. The study included 448 cases with 153 events. The analysis results show that group comparison revealed a significant difference in gender distribution, age stratification, comorbid chronic somatic diseases (CSD), partnered status (PS), social support rating scale (SSRS) scores, insight and treatment attitude questionnaire (ITAQ) scores, modified overt aggression scale (MOAS) scores. The result of multiple variable logistic regression analysis indicates that male gender and comorbid CSD increased readmission risk; Age > 60 years decreased risk; Higher MOAS scores, lower SSRS scores and ITAQ scores were significantly associated with elevated readmission risk. Model evaluation demonstrated that area under the receiver operating characteristic curve was 0.85(95% CI 0.81-0.89) ; Hosmer-Lemeshow test (χ2 = 6.15, P = 0.63) indicates a good fit. This prediction model helps identify high-risk cases and simplifies BD management.