CHEST Guidelines-Factors-Associated-With-VTE-in-Patients-Who-Are-Cr

Factors-Associated-With-VTE-in-Patients-Who-Are-Cr

Pdf Summary

In this correspondence published by the American College of Chest Physicians, Guanyu Yang critiques a study conducted by Al-Dorzi et al. that analyzed risk assessment models for venous thromboembolism (VTE) in critically ill patients receiving pharmacologic thromboprophylaxis. Yang acknowledges the study's contributions but points out two major concerns. Firstly, the absence of critical factors such as recent surgery and history of VTE from the multivariate logistic regression analysis. Although these factors had P values greater than 0.1 in the univariate analysis, their clinical relevance in developing VTE was well-established in a referenced meta-analysis. Yang suggests that their inclusion could have provided novel insights if the sample size had allowed for it.<br /><br />Secondly, Yang emphasizes the importance of conducting multicollinearity tests among the variables used in the logistic regression analysis. Multicollinearity could lead to unreliable estimates, affecting the reliability of the study's findings. Conducting such tests could enhance the robustness of the research outcomes.<br /><br />The response from the original authors acknowledges these considerations. They explain that while the mentioned factors were predictive of VTE according to a recent meta-analysis, they were not included in their multivariate model. Nonetheless, they express appreciation for the feedback, which aims to improve the diagnostic performance of existing risk models in the clinical setting of critically ill patients. <br /><br />This discussion underscores the complexity of risk assessment in VTE and the necessity of continual evaluation and refinement of existing models to ensure optimal patient outcomes.

Keywords

venous thromboembolism

risk assessment

critically ill patients

pharmacologic thromboprophylaxis

multivariate logistic regression

multicollinearity

meta-analysis

clinical relevance

risk models

patient outcomes