[Skip to Content]
[Skip to Content Landing]
Citations 0
January 26, 2011

Blood Transfusion and Cardiac Surgery—Reply

JAMA. 2011;305(4):357-359. doi:10.1001/jama.2011.15

In Reply: Drs McIlroy and Argenziano and Dr Shehata and colleagues raise questions related to the statistical methodology used in our study, in particular, the choice of a noninferiority margin of −8%. Based on previous reports,1,2 we estimated a 10% incidence of the primary outcome and calculated the sample size according to CONSORT guidelines.3,4 The noninferiority margin of −8% was chosen as the smallest value that would represent a clinically important effect; it was selected because it was the lowest of the upper confidence interval limits for the differences between groups in outcomes reported in the randomized controlled study by Hébert et al5 (upper limits of confidence intervals varying from 8 to 12). Because the lower limit of the confidence interval of the difference between groups for the primary outcome event rate in our study (−6% to 4%) was above the predefined −8% noninferiority threshold, we can state that the 1% difference in outcome was not clinically important, confirming the primary hypothesis of noninferiority between the liberal and restrictive groups.4

First Page Preview View Large
First page PDF preview
First page PDF preview