How many steps along the path is too far?

Neely et al [1] have studied the disposition of voriconazole in children suspected or proven to have a fungal infection. Although this is important, and the pharmacokinetic data are most welcome in this difficult-to-study population, the ability to link the drug concentrations observed to outcome is the critical issue for this study. A number of other investigators have linked measures of voriconazole exposure to outcomes [2–5]. Most of these have linked exposure to some measure of response [2–4]. In this study, voriconazole exposure has been linked to time to death in a Cox proportional hazards model. Because the article also included a table detailing the outcomes for the patients examined and whether they had an actual trough concentration of voriconazole measured (thanks to the authors and the editors), it allowed secondary analyses to be performed. Of the patients examined, 24 had at least 1 trough concentration measured. In a logistic regression analysis, whether the measured voriconazole concentration exceeded 1.0 mg/L had a significant effect on the probability of death (odds ratio, 30.3; 95% confidence inter-