Statistical Power and Polyvalent Immunoglobulin

Abstract

<h3>To the Editor.—</h3> In a report on their randomized trial of the use of polyvalent immunoglobulin (IgG) for prevention of cytomegalovirus infection after renal transplantation,1 Kasiske et al report that incidence rates of cytomegalovirus infection and moderately severe cytomegalovirus infection were not significantly different between the control group (84% and 77%, respectively) and the group treated with intravenous IgG (67% and 53%, respectively). I was not surprised that these decreases of 21% and 31% from baseline were not significant, since the sample size was quite small (15 treated with intravenously administered IgG, 13 controls), but I was surprised by their statement that this study had a power of .86 to detect a (clinically) significant difference in infection incidence and a power of .91 to detect such a difference in moderately severe infection rates. I therefore worked back from their data to calculate what they had considered to be a significant difference in infection