© 2007 American Thoracic Society
Plateau Pressures in the ARDSnet Protocol: Cause of Injury or Indication of Disease?From the Authors:Drs. Loring and Weiss argue that in our recent article (1) we neglect to consider that worse outcome (as measured by cytokine concentrations and ventilator-free days) in patients with plateau pressure (Pplat) values ranging between 28 and 30 cm H2O may simply be due to the fact that they were more severely ill rather than due to the lack of protection produced by limiting tidal volume (VT) to 6 ml/kg and Pplat to 30 cm H2O. As indicated in the METHODS and RESULTS sections of the article, multivariate stepwise regression analysis with backward elimination was used to explain whether differences in cytokine concentrations and number of ventilator-free days were related to disease severity or to ventilator-induced lung injury (VILI). The primary goal of linear multiple regression analysis is the identification of relevant independent variables that contribute to criterion score variance from a larger initial set. The regression equation for the full set of predictors is initially determined. Then, for each variable in turn both the reduction in R2 that would occur if the variable were deleted and the F ratio associated with that decrease are computed. The variable yielding the smallest nonsignificant F ratio is deleted. The entire process is then repeated with the remaining predictors and continues until all F ratios are significant or until the deletion of another variable would produce an R2 significantly lower than the R2 for the full set. Output of this analysis may clarify the interpretation of criterion performance as well as isolate possible independent variables for future study.
Dependent variables entered in the regression analysis were indices of disease severity (weight of the entire lung, PaO2/FIO2 ratio, Pplat) and indices of VILI (amount of "protected tidal inflation" and "tidal hyperinflation"). We found that tidal hyperinflation was the only variable independently associated with concentration of IL-6 (p = 0.001), IL-1 Rather than being conclusive, these findings may (1) support Hager and coworkers' data suggesting that mortality decreases as Pplat declines from high to low levels at all levels of Pplat (2) and (2) provide the rationale for a future clinical trial to evaluate whether maintaining a safe limit of 25 cm H2O can provide better protection against VILI in patients with acute respiratory distress syndrome (ARDS) than the limit of 30 cm H2O (3). According to this reasoning, in the DISCUSSION section of the article we stated, "prospective clinical trials are required to prove that ventilation at the lower Pplat would have improved outcomes in those patients ventilated with Pplat ranging between 28 and 30 cm H2O."
Università di Torino, Ospedale S. Giovanni Battista-Molinette, Torino, Italy
Università di Milano, Ospedale Maggiore Policlinico, Mangiagalli, Regina Elena, Milano, Milan, Italy
University of Toronto, St.Michael's Hospital, Toronto, Ontario, Canada FOOTNOTES Conflict of Interest Statement: V.M.R. served as a consultant to Maquet and received grant support from Tyco. L.G. received consulting and lecture fees from Kinetic Concepts (KCI). A.S.S. received consulting fees from BOC Medical, Hamilton Medical, Maquet, and KCI. REFERENCES
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(p = 0.0025), IL-8 (p = 0.001), tumor necrosis factor (TNF) soluble p55 receptor (p = 0.007), TNF soluble p75 receptor (p = 0.001), and number of ventilator-free days (p = 0.005).