Prehospital versus emergency department noninvasive ventilation: oxygen, outcomes, and mediator
Year: 2024
Type: article
Abstract: We have read with great interest the article ‘Effect of early initiation of noninvasive ventilation in patients transported by Emergency Medical Service for acute heart failure’ by Gorlicki et al. [1]. We commend them for this insightful analysis comparing early (prehospital) versus delayed (intrahospital) use of noninvasive ventilation (NIV) and reinforcing the importance of prospective studies on the topic. We, however, have three points we would like to discuss. The first point concerns the treatment administered by Emergency Medical Services, particularly the use of oxygen. We were surprised to find that patients receiving NIV were less likely to receive supplemental oxygen. Generally, patients with acute heart failure receive supplemental oxygen when NIV is initiated. We, therefore, wonder if, by ‘oxygen’, the authors meant inhalation via a mask or nasal cannula, thereby excluding NIV by definition. The second issue relates to the choice of some outcomes, surprisingly described as ‘short-term’ outcomes. While mortality is indeed a crucial outcome, we believe the time frame between early and delayed NIV initiation might be too short to significantly impact 30-day mortality rates. More immediate outcomes, such as ICU admission rates or the need for invasive mechanical ventilation, might be more appropriate as they are emergency department-oriented and associated with prehospital NIV [2,3]. It would be helpful to understand why these outcomes were not considered. Our final point concerns the analytical approach, specifically the selection of covariables in the models. While model selection based on the statistical significance of univariate analysis is debatable, our main concern is the adjustment for ICU admission. We believe that ICU admission should be considered a mediator rather than a confounder. ICU admission likely occurs as a result of the initial treatment decision (whether to initiate NIV early or later) and subsequently influences the primary outcome. Therefore, ICU admission acts as an intermediary step in the causal pathway from the type of NIV initiation to the adverse outcomes. For ICU admission to be a confounder, it would need to independently affect both the decision to initiate NIV early or later and the primary outcome. Adjusting for ICU admission as a confounder in the statistical models could introduce bias, as demonstrated in simulation studies [4]. ICU admission was included in the first two models, but it is unclear whether it was also included in the third model due to discrepancies between the method description and Table 3. We would be very interested to see whether the results differ when ICU admission is excluded from the model. Despite these considerations, we are grateful for this paper that reinforces both the importance of NIV in emergency medicine and the need to continue research in this field. Acknowledgements Conflicts of interest There are no conflicts of interest.
Institutions: University of Geneva, University Hospital of Geneva
Language: English
Subfield: Pulmonary and Respiratory Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W4405832436