We would like to thank Dr. Singh and colleagues for their letter regarding our manuscript.1We believe their concerns are justified and have the following responses.
First, with regard to the number of patients excluded, we attempted to exclude all patients who had data that was not appropriate. Because our anesthesia information system is designed to provide care for patients of all ages, heights, and weights, the values allowed for multiple variables are quite robust, which does allow for potential errors in data entry. Cases were excluded based on extremes of recorded height and weight. The 320 cases that were excluded because of excessively low tidal volume were either completed using an intensive care unit ventilator (which did not automatically port data to the anesthesia information system) or had erroneous electronic data collection, which results in tidal volumes being recorded in very low increments of (less than 100 ml).
Patients were included in the study if they had clear preoperative arterial blood gas data that the treating anesthesiologist would have used in planning clinical care. As Singh et al. point out, it is possible that not all hypoxic patients were identified preoperatively; however, the focus of the study was to evaluate anesthesiologist behavior in cases in which the presence of acute lung injury (ALI) could be considered. Although surrogate measures such as SpO2/FiO2could have increased the size of the study group, inclusion of these patients would have introduced bias, as to be included in the comparator ALI database, patients must have met formal American-European Consensus Conference criteria, including an arterial blood gas with a PaO2/FiO2ratio less than 300.2
Regarding mode of ventilation, for a significant portion of the study, exclusively volume-controlled anesthesia machines (Narkomed 2B; Dräger Medical, Inc., Telford, PA) were used. In 2007, anesthesia machines with a pressure-controlled mode became available (Aisys; General Electric Healthcare, Madison, WI). Unfortunately, the anesthesia information system does not record the mode of ventilation, and we agree that this would better illustrate the care provided to patients. As we described in the manuscript,1the use of plateau pressures would be preferable to peak inspiratory pressures, but again, these data are not collected by the anesthesia information system. We believe peak inspiratory pressure is a reasonable surrogate because it has been used by several authors in the anesthesia literature to investigate ventilator-induced lung injury and postoperative ALI.3,–,5Regardless, we believe that our study documents that intraoperative tidal volumes used in patients with known preoperative hypoxia and ALI are not consistent with lung-protective ventilation strategy settings and that this is an important consideration for anesthesiologists.
Following the landmark ARDSNet publication in 2000,6the gradual adoption of lung-protective ventilation strategies in the intensive care unit has been reported in several studies.7,–,10Of particular concern in our study is the finding that patients who were receiving low Vt ventilation in the intensive care unit typically had their Vt increased in the operating room to 7.5 ml/kg predicted body weight, again reinforcing the need for anesthesiologists to consider ALI management as part of their operative care.
Finally, we sincerely regret the typographical errors in the manuscript and appreciate the readers allowing us the opportunity to clarify these discrepancies. The abstract indeed and figure 1 legend should read “PaO2” rather than “PaCO2,” and in figure 1 and table 1 “PaOC” should read “PaO2,” and FIO2is expressed as a percentage.