The predictive value of clinical parameters for frailty in emergency observation patients

Liu Xiaoying; Xu Jun; Liu Aihui; Zhou Wenhua; Chen Keyu; Tian Jian; Zhang Chen; Zhu Jing; Jia Zheng; Sun Xiaonan; Tian Liyuan

doi:10.3969/j.issn.1002-1949.2025.11.007

Abstract：Objective To investigate retrospectively the current status of frailty among emergency observation patients and to analyze the predictive value of clinical indicators for frailty in these patients.Methods A total of 194 emergency observation patients from a tertiary Grade-A general hospital in Beijing from April to September 2022 were selected as the research subjects.General information of the patients,FRAIL scale scores,Barthel index scores,emergency triage levels,and laboratory test results were collected.The patients were divided into the frailty group and the non-frailty group based on whether they developed frailty.Binary Logistic regression was used to analyze the risk factors for frailty in emergency observation patients.A predictive model was constructed by a nomogram.Use Hosmer-Lemeshow test and Boostrap 500 resampling to verify the fitting excellence and accuracy of the evaluation model.The predictive value and calibration performance of clinical indicators for the occurrence of frailty were evaluated by using the receiver operating characteristic(ROC)curve and calibration curve.Results A total of 139 cases(71.65％)of emergency observation patients experienced frailty.There were statistically significant differences in terms of triage level,Barthel index,memory loss,and albumin between the two groups(P＜0.05).With frailty as the dependent variable,Logistic regression analysis revealed that triage level,Barthel index score,and albumin were independent risk factors for frailty in emergency observation patients.The prediction equation for frailty in emergency observation patients was 6.087+0.585 × memory+1.295 × triage level-0.038 × Barthel index-0.092 × albumin,with the optimal cut-off values of Barthel index and albumin being 67.5 points and 35.5 g/L,respectively.The area under ROC curve of Logistic regression model was 0.820(95％CI 0.803-0.828),the accuracy was 0.805(95％CI 0.768-0.830),the sensitivity was 0.908(95％CI0.849-0.957),the specificity was 0.546(95％CI 0.455-0.636),optimal cut-off value was 0.818.Conclusion The incidence of frailty among emergency observation patients is relatively high.Clinical parameters such as self-care ability and nutritional status can predict the degree of frailty in emergency observation patients.