A scoping review of hypoxemia risk prediction models for postoperative patients
10.3760/cma.j.cn115682-20240312-01292
- VernacularTitle:术后患者低氧血症风险预测模型的范围综述
- Author:
Xiangyuan WANG
1
;
Hongxia GE
1
;
Liying SHI
1
;
Ke SHAO
1
;
Wenzi WANG
1
;
Shutao LI
1
;
Wei WANG
1
Author Information
1. 山东中医药大学护理学院，济南 250014
- Publication Type:Journal Article
- Keywords:
Review;
Postoperative hypoxemia;
Risk assessment;
Prediction model
- From:
Chinese Journal of Modern Nursing
2025;31(3):398-404
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To summarize the risk prediction models for postoperative hypoxemia and provide a reference for clinical nursing practice and future research on hypoxemia risk prediction models for postoperative patients.Methods:A systematic literature search was conducted in CNKI, CBM, Wanfang, PubMed, Web of Science Core Collection, Cochrane Library, Embase, and CINAHL databases, covering publications up to January 31, 2024. Two researchers independently screened the literature, extracted data, performed integrative analysis, and evaluated the risk of bias in the included studies.Results:Seventeen studies were included, involving 17 different prediction models. The study populations were primarily adult patients, with hypoxemia incidence rates ranging from 2.40% to 49.30%. Modeling methods included Logistic regression and decision tree algorithms. The presentation formats of the models included risk scoring formulas, nomograms, decision tree diagrams, and web calculators. The five most frequently identified predictive factors were body mass index, age, comorbidities, duration of intraoperative cardiopulmonary bypass, and preoperative white blood cell count. Sixteen models reported the area under the receiver operating characteristic curve ranging from 0.667 to 0.916. All 17 studies exhibited varying degrees of bias risk.Conclusions:Existing risk prediction models for postoperative hypoxemia demonstrate good performance; however, the bias risk level of all studies was high. Future research should standardize the model development process according to bias risk assessment checklists to establish models with low bias risk and strong clinical applicability.
