Fig. 2.

Time course of mean arterial pressure (A), log-transformed vasopressor infusion rate in patients who received vasopressors and no./total no. (%) of patient-days of vasopressor use (B), and log-transformed interleukin (IL)-6 (C) in the Steroids and Control group. Data are presented as mean (95% confidence interval). ITT analysis: intention-to-treat mixed-model analysis corresponding to the first 72 hours postresuscitation; EXPL: exploratory mixed-model analyses corresponding to days 1–10 (A, B) or to days 1–7 (C) postresuscitation. In B, the nos./total nos. (%) of patient days of vasopressor use were compared by Fisher’s exact test. ROSC, return of spontaneous circulation. AIC: Akaike information criterion; %VE: percent variance explained, which reflects the squared Pearson correlation coefficient determined by linear regression with observed variable values as dependent variable and mixed model-estimated variable values as independent variable. A, B, and C: The sequences of numbers just above the horizontal axes represent nos. of patients participating in the analysis at the respective follow-up time points. B: The numbers on the top of the bars reflect numbers of patients not receiving any vasopressor support at the corresponding time-points of follow-up. ^†, Additional model information: A, ITT and EXPL analysis: effect of center and insulin infusion rate, P = 0.17 to 0.52. ^‡, Additional model information: B, ITT and EXPL analysis: effect of center, insulin infusion rate, and blood glucose, P = 0.22 to 0.71. ^§, Additional model information: C, ITT and EXPL analysis: effect of center P = 0.10 to 0.11. ^†,^‡,^§, Log-transformed values were actually used in all analyses, because data exhibited skewed distributions; in A, we present the actually observed/recorded values of mean arterial pressure in mmHg, solely for the purpose of a simplified and clear presentation.