Development and validation of predictive model for 28-day mortality in very older patients with sepsis
10.3760/cma.j.issn.1671-0282.2024.04.015
- VernacularTitle:高龄脓毒症患者28 d死亡预测模型的构建及验证
- Author:
Qiujing LI
1
;
Na SHANG
;
Zhen WANG
;
Tiecheng YANG
;
Shubin GUO
Author Information
1. 首都医科大学附属北京世纪坛医院急诊科，北京 100038
- Keywords:
Sepsis;
Elderly;
Cognitive impairment;
Frailty;
Risk factor;
Predictive model;
Nomogram;
Prognosis
- From:
Chinese Journal of Emergency Medicine
2024;33(4):542-548
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To develop and validate a predictive nomogram for 28-day mortality among very older patients with sepsis, to identify high-risk patients early and improve prognosis.Methods:This study was conducted from January 1, 2022, to November 30, 2022. Very older patients aged≥80 years with sepsis admitted to the emergency department of Beijing Chao-Yang Hospital, Capital Medical University were consecutively recruited. Their clinical data within 24 h of admission and 28-day mortality was recorded. The participants were divided into training (70%) and validation cohort (30%) (random number). In the training cohort, the risk factors of 28-day mortality were selected via least absolute shrinkage and selection operator (LASSO) regression analysis and multivariable Cox proportional hazard model, and a nomogram was developed. The prediction model was verified in validation cohort.Results:In total, 507 very older patients with sepsis were included, among which the mortality rate was 31.2%. In training cohort, the independent risk factors for 28-day mortality were identified: increased age [hazard ratio ( HR)=1.059, 95% confidence interval (95% CI)=1.017-1.103, P=0.005], cognitive impairment ( HR=2.100, 95% CI=1.322-3.336, P=0.002), frailty ( HR=2.561, 95% CI=1.183-5.545, P=0.017), decreased mean arterial pressure ( HR=0.987, 95% CI=0.976-0.998, P=0.017), decreased prealbumin ( HR=0.997, 95% CI=0.994-1.000, P=0.040), increased blood urea nitrogen ( HR=1.028, 95% CI=1.010-1.045, P=0.001), increased procalcitonin ( HR=1.008, 95% CI=1.001-1.016, P=0.019) via LASSO regression analysis and multivariable Cox regression analysis. The nomogram was developed using these seven predictors. In the training and validation cohorts, the calibration curves, time-dependent AUC curves, and decision curve analysis showed that the nomogram had good calibration degree, discrimination and clinical net benefits. Conclusions:Increased age, cognitive impairment, frailty, decreased mean arterial pressure, decreased prealbumin, increased blood urea nitrogen, and increased procalcitonin are independent risk factors for 28-day mortality in very older patients with sepsis. The nomogram, which included the seven predictors, have good predictive performance, and might be helpful for prognosis assessment.
