Eight controlled studies (5 RCTs and three cohorts) have evaluated the effectiveness of prolonged GC treatment initiated before day 14 of early ALI/ARDS (N = 334)24,44,45,46 and late ARDS (N = 235).17,23,47,48,49 These trials consistently reported that treatment-induced reduction in systemic inflammation17,23,24,45,46,47,49 was associated with significant improvement in PaO2:FiO2,17,23,24,45,46,47,49 and significant reductions in multiple-organ dysfunction score,17,23,24,45,47,49 duration of mechanical ventilation,17,23,24,44,45,46 and ICU length of stay (all with P values < 0.05).17,23,24,44,45 Four of the 5 randomized trials provided Kaplan Meier curves for continuation of mechanical ventilation; each showed a 2-fold or greater rate of extubation in the first 5 to 7 days of treatment.17,23,24,45 In the ARDS Network trial, the treated group—before discontinuation of treatment—had a 9.5-days reduction in duration of mechanical ventilation (14.1 ± 1.7 vs 23.6 ± 2.9; P = 0.006) and more patients discharged home after initial weaning (62% vs 49%; P = 0.006).23 As shown in Figure 68–1, GC treatment initiated before day 14 of ARDS was associated with a marked reduction in the risk of death (Relative Risk (RR) = 0.68, 95%CI: 0.56-0.81; P < 0.001; I2 56%).9 There was, however, significant heterogeneity between studies. As a result of the marked differences in study design and patient characteristics, the limited size of the studies (fewer than 200 patients), the cumulative mortality summary of these studies should be interpreted with some caution. For this reason, a recent consensus statement recommended early initiation of prolonged GC treatment for patients with severe ARDS (PaO2:FiO2 < 200 on PEEP 10 cm H2O) and before day 14 for patients with unresolving ARDS.15 The ARDS Network trial reported that treated patients had increased mortality when randomized after day 14 of ARDS (8% vs 35%; P = 0.01).23 This subgroup (N = 48), however, had large differences in baseline characteristics, and the mortality difference lost significance (P = 0.57) when the analysis was adjusted for these imbalances.50