Construction of a risk prediction model for peripheral intravenous infusion phlebitis in cervical carcinoma

Abstract

Objective To explore the risk factors of peripheral intravenous infusion phlebitis (IP) in cervical carcinoma (CC) patients, and construct a risk prediction model of peripheral venous IP.Methods Information of CC patients in the obstetrics and gynecology department of the First People's Hospital of Suqian city from January 2019 to May 2024 was retrospectively collected. CC patients from January 2019 to December 2023 were included in the training set, and CC patients from January 2024 to May 2024 were included in the internal validation set. Patients were divided into IP and non IP groups based on whether IP had occurred. Single factor analysis and stepwise Logistic regression were used to screen the influencing factors of peripheral venous IP in CC patients, and prediction model was constructed. The receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC), calibration curve, and decision curve were used to evaluate the predictive performance, accuracy, and clinical adaptability of the risk prediction model.Results A total of 349 CC patients were included, and the incidence of peripheral venous IP was about 14.33%. In the training set, there were 289 CC patients, and peripheral venous IP occurred in 41 cases. In the validation set, there were 60 CC patients, and peripheral venous IP occurred in 9 cases. Stepwise multivariate Logistic regression analysis showed that the combination of diabetes, right puncture, lower limb puncture, and total amount of infusion ≥ 2 000 mL were risk factors for the peripheral vein IP in CC patients, while infusion heating was a protective factor for the peripheral vein IP in CC patients. Probability of peripheral venous IP =1/[1+e^(-(1.231×Diabetes+1.075×Puncture site+2.935×Puncture limb +0.856×Infusion volume-0.983×Infusion heating))]. ROC analysis showed that the AUC (95%CI) of the risk prediction model for peripheral venous IP in the training set and internal validation set was 0.807 (0.738, 0876) and 0.838 (0.719, 0.956), respectively. The calibration curve indicated that the predicted probability and actual occurrence probability were roughly consistent, and the decision curve indicates that within a certain threshold range, it could benefit clinical practice. The results of calibration curve and clinical decision curve in the internal validation set were not good.Conclusion The risk prediction model of peripheral venous IP constructed in this study has certain predictive ability and clinical applicability, which can help nursing staff identify peripheral venous IP in CC patients in the early stage. Further validation is still needed through large-scale research in the future.

Keywords

• cervical carcinoma
• peripheral intravenous infusion phlebitis
• risk factors
• predictive model