© 2003 American Thoracic Society
Meta-analysis of tidal volumes in ARDSTo the Editor:
In their recent letter regarding trials of lower tidal volume ventilation in acute respiratory distress syndrome (ARDS) (1), Drs. Parshuram and Kavanagh asked, "... are the control groups representative of the standard of care?" None of the trials (2–6) included a standard or routine care study group because they were not designed to compare outcomes of patients who received lower tidal volumes with those who received routine care. Routine care encompassed broad ranges of tidal volumes and inspiratory airway pressures. These broad ranges reflected physicians' widely varying opinions regarding the value of different approaches to mechanical ventilation in acute lung injury (ALI) and ARDS (7, 8). Also, some physicians doubted the importance of adjusting tidal volumes and inspiratory pressures within these broad ranges. To determine if a lower tidal volume approach was superior to a traditional approach that used higher tidal volumes, the investigators for each of the trials designed study protocols to represent these two approaches. The tidal volumes and inspiratory pressure limits were consistent with contemporary opinion and within the scope of routine care (9, 10). How can the results of the trials be used to improve clinical practice? Our trial results strongly suggest that tidal volumes Drs. Parshuram and Kavanagh also asked, "Does entry into the trial confer a disadvantage to patients randomized to the control group (versus not being in the trial)?" This question has been raised only with hindsight, after the results of the trial were known. To avoid risks to study participants, trial methodology included many safeguards and good practice procedures, including standardized weaning, which improve the process of care relative to routine care (12, 13). Our trial procedures were reviewed for safety and approved by Institutional Review Boards (IRBs) for 24 hospitals and two independent committees of experts in critical care, clinical trial design, and biomedical ethics. It is notable also that in many patients randomized to the traditional study group, tidal volumes were actually decreased from their routine care values to 12 ml/kg PBW or lower.
In their recent letter (14), Eichacker and colleagues correctly estimated that in nearly 80% of patients in our traditional study group, tidal volumes were increased from their routine care levels to the levels stipulated by protocol rules. Mean routine care tidal volume (± SD) before patients were enrolled in the trial was 10.3 ± 2 ml/kg PBW (10). Mean tidal volume after randomization to our traditional study group was 11.8 ± 0.8 ml/kg PBW (6). Thus, there was substantial overlap in these ranges of tidal volumes. The value of 11.8 ml/kg PBW is equivalent to 9.9 ml/kg measured body weight, which is at the lower end of the range that had been recommended for the traditional approach that gave high priority to maintaining gas exchange, breathing comfort, and acid-base homeostasis (7). It is also lower than those used in ARDS patients in the early–mid 1990s (11). The modest increase in mean tidal volume after randomization to the traditional study group was associated with a modest increase in mean plateau pressure from 30 to 33 cm H2O. For comparison, 33% of plateau pressures during routine care were Eichacker and colleagues are correct that a small proportion of routine care tidal volumes was 6 ml/kg PBW or lower. We chose 6 ml/kg PBW as the tidal volume target in our lower tidal volume study group because it was consistent with the tidal volumes used in four encouraging reports of lower tidal volume ventilation in ALI/ARDS (8, 15–17) and with contemporary recommendations (18). Eichacker and colleagues compared the 31% mortality of patients who were eligible but declined to participate in our trial (19) to the 40% and 31% mortalities in our traditional and lower tidal volume study groups. These comparisons are misleading because they fail to account for important differences between participants and the nonparticipants on whom the data are available. IRBs did not grant permission to use nonparticipant data at some Network centers. At the centers where permission was granted, mortalities for both participant study groups were lower than the mortalities cited by Eichacker and colleagues, which are for all Network centers combined. At the centers where permission was granted, mortality in nonparticipants was not significantly different from the mortality in either of the participant study groups. Citing data from one of the ARDS Network centers, Eichacker and colleagues stated, "mortality associated with routine care reached less than 30% in 1996, the year that the ARDS Network started their trial of low tidal volume." This is misleading because it compares mortality for 1 year at a single center to the mortality in our lower tidal volume group at all Network centers without addressing differences in the populations. Subjects at the single center were younger than the patients at all centers who participated in the trial (48.6 ± 17.5 compared with 51.5 ± 17.2, p value = 0.0006). Regardless of primary risk for ARDS, 46% of the patients at this center were admitted after major trauma compared with 11% with trauma at all Network centers combined. Mortality from ARDS in younger patients and in those with trauma is lower than in other patient populations (20, 21). Finally, it is misleading to select the mortality in an outlier year at the single center for comparison with the overall mortality in our trial. From 1997 to 2001, mortality at the single center ranged from 36 to 41%. Eichacker and colleagues claim that the unimodal distribution of tidal volumes in routine care (Figure 1 of their letter) does not represent the traditional and lower tidal volume approaches described in our previous letter (11). A reasonable explanation for this distribution of tidal volumes is that some physicians adhered to a traditional approach, some adhered to the lung-protective approach, and many used approaches that represented hybrids between the two approaches because they were unsure which approach was better. Eichacker and colleagues suggested that one of the study groups in our trial should have used tidal volumes of 9 to 12 ml/kg PBW, a range that encompassed approximately 60% of the tidal volumes that were used before enrollment. This approach could potentially provide a method to test the same hypothesis addressed in our trial. Analyses of our trial database strongly suggest that, if properly designed and conducted in other respects, such a trial would have demonstrated superiority of the lower tidal volume approach to the 9–12 ml/kg approach (11). However, substantial practice variations would have occurred in the 9–12 ml/kg study group, reducing the "signal to noise ratio." Therefore, it would have been necessary to enroll more patients in the trial to demonstrate the same result. More patients would have received an inferior approach in the course of such a trial. At the time the trials were designed and conducted, nobody knew if the traditional or lower tidal volume approach was superior. Therefore, nobody knew if the trial design proposed by Eichacker and colleagues to avoid increases in tidal volumes from those used in routine care would be safer than a design that avoided decreases in tidal volumes from those used in routine care. We maintain that the approach we used was consistent with highest standards for clinical trial design because it addressed an important clinical problem using methodology that could answer the question while employing redundant mechanism of oversight and numerous safeguards and procedures to minimize risks to study participants. For the NIH NHLBI ARDS Network investigators:
a Johns Hopkins University Baltimore, Maryland REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
12 ml/kg predicted body weight (PBW) and inspiratory plateau pressure limits 
