What Good are Large Databases of Intraoperative Data?

In Response: We thank Drs. Luce, Auroy, and Benhamou for their comments (1) regarding our recent publication (2). Their investigation (3) did not yield significant associations between anesthetic choice and intraoperative hypotension. There are several important differences in the studies: They manually recorded hypotensive events in the database, with the likelihood that manual reporting significantly underestimates the true incidence of intraoperative hypotension (4,5). They sought the predictors of hypotension throughout the entire intraoperative period. Thus, the associations of induction drug choices with hemodynamic events later in the anesthetic may have been less strongly associated compared with our analysis, which was limited to the initial 10 min after anesthetic induction. They did not consider baseline mean arterial blood pressure as a risk factor. They used etomidate as the reference group, whereas we used etomidate or thiopental. Although their data do not confirm an increased risk for propofol or thiopental versus etomidate, the lower confidence bounds of 1.0 indicates a trend towards more hypotension with propofol and thiopental, compared with etomidate. The differences noted above may have caused their results to barely miss being statistically significant. The anesthesia information management system used to create our database is expensive, but it was not purchased solely as a research tool. This system provides economic benefits to our institution that offset the purchase and maintenance costs. It is secure in its protection of patients’ identities. We require IRB approval for database access for research purposes. Qualified individuals can readily extract data from the database for research. The letter also addresses the inability of anesthesiologists to affect important predictors of postinduction hypotension, including age and ASA physical status. Although this is true, our retrospective analysis was intended to inform anesthesiologists of the independent associations of the induction techniques with postinduction hypotension, controlling for the effects of the important risk factors. These data are preliminary by nature. However, they suggest that propofol should be avoided in certain situations, such as when hypotension is present before induction of anesthesia or in elderly patients with ASA physical status III or higher. These may not be novel concepts, but they could not have been identified without the data provided by our automated database. David L. Reich, MD Sabera Hossain, MA Carol A. Bodian, DrPH Department of Anesthesiology Mount Sinai Medical Center New York, NY [email protected]