LAUSR

e1222 www.ccmjournal.org December 2018 • Volume 46 • Number 12 The authors reply: We thank Karakonstantis et al (1) for raising important points in their letter to the editor regarding our report (2) recently published in Critical Care Medicine. Karakonstantis et al (1) wrote: “Of interest is the finding that the higher mortality was not driven by the delayed initiation of antimicrobial therapy. This finding was unexpected as it contrasts findings from prior larger studies.” We would emphasize that our report asserted that “the substantial difference in mortality between the two cohorts (vague vs explicit) appears to be due to factors beyond just sepsis recognition and antibiotic delay.” In multivariate analysis, vague symptomatology was associated with mortality, whereas time-to-antibiotics was not significant. This was not to imply that there is no causal relationship between delayed antibiotics and mortality, and we stated that “this study was not powered to show effect sizes associated with hourly delays that have been previously published.” We assessed for collinearity in our model. Collinearity means that predictor variables are highly correlated with each other such that values of one can be predicted by that of the other Timing of Antibiotic Administration and Mortality in Septic Patients Presenting With Vague Symptoms