In Reply.—Dr Zuccalà and colleagues
report the results of their study, which suggested that perioperative transfusion
may be associated with an increased risk of mortality and postoperative fever.
These results are different than the results of our study of transfusion in
patients with hip fracture, which did not demonstrate a relationship between
transfusion and postoperative mortality.
We believe several methodologic differences may explain the conflicting
results. Our study included 8787 patients from 20 medical centers that had
significant variability in transfusion rates. The study by Zuccalà
et al included a much smaller number of patients (N=206) and restricted transfusion
to patients with hemoglobin levels less than 95 g/L. As a result, there will
be no variability in transfusion practice for patients with hemoglobin levels
greater than 95 g/L. Therefore, it is advisable to exclude from the study
those patients not eligible to receive a transfusion, as they contribute no
information to the comparison of patients receiving transfusions vs those
patients not receiving transfusions. Furthermore, to understand the clinical
relevance of the analysis, the association between transfusion and outcome
should be stratified by trigger hemoglobin level. The trigger hemoglobin level
is the principal measure of the severity of disease for which transfusion
might be administered and is likely to be related to both an increased risk
of transfusion and mortality. For example, in Table 4 of our article, we display
a stratified analysis of the effect of transfusion in patients with hemoglobin
levels of 70 to 79, 80 to 99, and greater than 100 g/L.