Ventilator-Free Day Outcomes Can Be Misleading

Introduction: Acute respiratory distress syndrome often requires invasive mechanical ventilation, with both mortality and mechanical ventilation duration as outcomes of interest. The concept of ventilator-free days has been proposed as an outcome combining these two outcomes. Here we analyzed the construction of the ventilator-free day outcome and provided a hypothetical scenario to alert physicians that such an outcome can lead to misleading interpretations. Methods: We proposed the isoventilator-free day curve concept and, using an analytical development, illustrated how a median ventilator-free day value can actually result from very different combinations of death rates and mechanical ventilation durations. We also used a hypothetical example to compare the Student t test, Wilcoxon rank-sum test, and Gray test (which accounts for death as a competing event with extubation) in comparing exposition to mechanical ventilation. Results: A median ventilator-free day value of 10 days may mean that 10% of the patients died while survivors were ventilated during a median of 14 days or that 40% died while survivors were ventilated during a median of 5 days. Changing the time horizon affected the Student t test but not the Wilcoxon rank-sum result. The Gray test was more relevant than both the Student t test and Wilcoxon rank-sum test in identifying differences in groups showing highly different mechanical ventilation duration, despite equal median ventilator-free days. This approach was also illustrated using real data. Conclusions: Use of ventilator-free days as an outcome appears to have many drawbacks. Suitable methods of analyzing time to extubation should be preferred. Drs. Laetitia and Bruno designed the study. Drs. Laetitia, Denis, and Elsa performed the study. Dr. Laetitia drafted the article. Drs. Fanny, Yohann, and Bruno provided critical review of the article. All authors read and approved the article. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website (http://ift.tt/29S62lw). The authors have disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: laetitia.contentin@univ-tours.fr Copyright © by 2017 by the Society of Critical Care Medicine and Wolters Kluwer Health, Inc. All Rights Reserved.

http://ift.tt/2BLwAxD