A novel predictive model for safe discharge after upper gastrointestinal bleeding
10.3760/cma.j.issn.1671-0282.2022.12.019
- VernacularTitle:新的评估上消化道出血安全出院的预测模型
- Author:
Yajie LI
1
;
Yawen ZHAO
;
Mingyang SONG
;
Kexuan WU
Author Information
1. 东南大学附属中大医院老年医学科，南京 210009
- Keywords:
Emergency;
Upper gastrointestinal bleeding (UGIB);
Predictive model;
Safe discharge
- From:
Chinese Journal of Emergency Medicine
2022;31(12):1680-1684
- CountryChina
- Language:Chinese
-
Abstract:
Objective:Upper gastrointestinal bleeding (UGIB) is a common gastrointestinal disease in the emergency department. Identifying low-risk patients suitable for outpatient treatment is the focus of clinical and research. A simple predictive model was developed to identify patients with UGIB who could safely avoid hospitalization, thus providing a feasible basis for triage by emergency physicians.Methods:A retrospective cohort study was conducted on patients with UGIB treated at Zhongda Hospital Southeast University from January 2015 to December 2020. Baseline demographic data and clinical parameters at the initial presentation were recorded. Multivariate logistic regression model was performed to identify predictors of safe discharge.Results:Six hundred and twelve patients (45.9%) were not safely discharged. There were significant differences in age, Charlson comorbidity index, systolic blood pressure, pulse rate, hemoglobin, albumin, blood urea nitrogen, creatinine and international normalized ratio between the safe discharge group and the non-safe discharge group ( P<0.05). Using multivariate logistic regression analysis, a total of 7 variables were included in the clinical prediction model of UGIB risk stratification: Charlson comorbidity index > 2, systolic blood pressure < 90 mmHg, hemoglobin < 10 g/dL, blood urea nitrogen ≥6.5 mmol/L, albumin <30 g/L, pulse ≥100 beats/min and international normalized ratio ≥1.5. The sensitivity, specificity, positive predictive value, and negative predictive value for predicting unsafe discharge were 98.37%, 24.10%, 52.3%, and 94.6%, respectively, with the best cutoff value ≥1. The area under the receiver operating characteristic (AUROC) curve was 0.822, which was significantly higher than Glasgow Blatchford score (GBS) 0.786 (95% CI: 0.752-0.820, P< 0.01) and AIMS65 0.676 (95% CI: 0.638-0.714, P< 0.01). Conclusions:The predictive model has a reliable predictive value, which can provide references for emergency medical staff to triage patients with UGIB, thereby reducing medical expenses and having certain social and economic benefits.
