Construction and validation of predictive model for acute respiratory failure in adult patients with community-acquired pneu-monia
10.13602/j.cnki.jcls.2025.08.07
- VernacularTitle:社区获得性肺炎成人患者并发急性呼吸衰竭的预测模型构建及验证
- Author:
Ziming WANG
1
;
Yue QU
;
Dandan LI
;
Huicong ZHOU
;
Binbin WU
;
Wei YU
Author Information
1. 南京大学医学院附属泰康仙林鼓楼医院检验科,南京 210046
- Publication Type:Journal Article
- Keywords:
community acquired pneumonia;
acute respiratory failure;
prediction model
- From:
Chinese Journal of Clinical Laboratory Science
2025;43(8):586-590
- CountryChina
- Language:Chinese
-
Abstract:
Objective To explore the risk factors of acute respiratory failure(ARF)in adult patients with community-acquired pneu-monia(CAP),and thereby construct and validate the efficacy of nomogram model.Methods The clinical and laboratory data of 172 adult CAP patients admitted to Taikang Xianlin Drum Tower Hospital affiliated to Nanjing University School of Medicine from January 2018 to December 2021 were retrospectively collected.The patients were divided into two groups based on whether they had concurrent ARF.After the comparison for the differences of single factor between the two groups,collinearity analysis was assessed.The risk fac-tors were then screened by binary logistic regression analysis with forward stepwise regression method.A nomogram model was subse-quently constructed and the discrimination and accuracy of the model were evaluated by ROC and colibration curves.Results Among the 172 CAP patients,53 cases(30.8％)developed ARF.The results of univariate analysis showed that the CAP patients with concur-rent ARF group had higher age,CURB-65 score and inflammatory markers than the non-concurrent ARF group,and the incidence of complex infection(culturing two or more pathogenic bacteria)was high.The values of CRP(C-reactive protein)and BUN/Alb(blood urea nitrogen/albumin)were significantly different between the two groups(53.910[25.900,101.200]vs.23.300[6.800,48.930],0.231[0.160,0.302]vs.0.123[0.089,0.171],P＜0.05).Multivariate analysis indicated:glucose(Glu)≥6.06 mmol/L(odds ra-tio(OR):2.737,95％confidence interval(CI):1.116-7.037),AST(aspartate aminotransferase)≥22.5 U/L(OR:4.291,95％CI:1.779-11.120),fibrinogen(Fib)≤3.83 g/L(OR:3.955,95％CI:1.631-10.237),uric acid(UA)188.07 μmol/L(OR:4.617,95％CI:1.859-12.489),BUN/Alb≥0.15 mmol/g(OR:6.381,95％CI:2.423-18.513),total number of multicomor-bidity≥3(OR:6.191,95％CI:2.088-21.905)were the risk factors(P＜0.05).All the screened indicators were incorporate into the nomogram model and its efficacy was verified.The results showed that the area under the curve of the model was 0.888[95％CI:0.840-0.935](P＜0.05),the sensitivity was 0.868,and the specificity was 0.790.The calibration curve showed that the predicted probability of adult CAP patients-associated with ARF was in good consistency with the observed probability(Briser Score:0.125;H-L test:x2=7.563,P=0.477).Conclusion The established model has a good ability to predict adult CAP associated with ARF,and can provide a reference basis for early clinical prediction and intervention treatment.
