Databases suggest that 60-70% of patients with severe respiratory failure due to H1N1 received steroids. Brun-Buisson et allooked back at a French registry including 208 patients with H1N1 and ARDS, 83 of whom received steroids. They found a hazard ratio of 2.4 for death associated with steroid administration, rising to 2.8 after applying their propensity scoring model. Receipt of steroids <3 days after intubation increased the association. AJRCCM 2011;183:1200-1206.
Meanwhile, in South Korea, Kim et al retrospectively observed 245 consecutive patients admitted to ICUs with H1N1 in 2009-2010 (162 of whom were intubated). They constructed a case-control analysis (steroids-no steroids, respectively). The steroid group was far sicker, with more ARDS, mechanical ventilation, secondary pneumonias, and also had a high prevalence of receipt of prior corticosteroids. After applying their propensity-matching analysis, the authors conclude that steroids were independently associated with mortality and superinfections (odds ratio 2.2 for death in 90 days). AJRCCM2011;183:1207-1214.
Neither of these studies were randomized or seemed convincingly able to demonstrate control for severity of illness as a confounder (viz., higher steroid use in patients less likely to survive). A spirited PRO (D. Annane) / CON (M. Matthay) Blue Journal editorial exchange ensued.