To the Editor.

It is truly disappointing to be corresponding on this issue in the 21st century, but Cornfield et al continue to perpetuate the fallacy of developing a (false) sense of security by the lack of observed complications in a small study. In their review of 142 patients in their pediatric intensive care unit (PICU) receiving propofol by infusion, they found no complications attributable to the drug. Their conclusion that “propofol can be administered safely” ignores the concept of confidence intervals. In the eloquent article by Hanley and Lippman-Hand, the simple rule of thumb to calculate an upper 95% confidence interval for a zero complication rate is presented as 3/n. In Cornfield et al then, we can conclude with 95% certainty that the complication rate—which conceivably refers to the fatality rate with this drug—is no more than 2.1%? With thousands of PICU patients receiving this...

You do not currently have access to this content.