Letters Section Editor: Stephen J. Lurie,
MD, PhD, Senior Editor.
In Reply: In response to Drs Van den Berghe
and Bouillon, we indicated that APACHE II scores, SOFA scores, and reason
for admission were all included in the logistic regression analysis. Age was
removed as it did not improve the model fit, possibly because it is already
contained within the APACHE II instrument. Consequently, we believe our model
did examine the relationship between insulin administration, glycemia, and
mortality, and was not confounded by severity of illness. Furthermore, because
the predicted odds ratios of death attributable to insulin administration
were the same in each glycemic band examined, it is highly unlikely that our
results simply reflect insulin resistance.
Finney SJ, Evans TW. Optimal Control of Glycemia Among Critically Ill Patients—Reply. JAMA. 2004;291(10):1198-1199. doi:10.1001/jama.291.10.1199-a