Xi 2010.

Methods	Multicentre retrospective cohort study
Participants	Country: China (Beijing) Setting: in‐hospital Number of sites: 23 Number of individuals: 155 Inclusion criteria: adults aged >= 18 years admitted to hospital Definition of influenza: laboratory confirmed Influenza type: 2009 influenza A H1N1 virus (H1N1pdm09) Mean (SD) age (years): cohort 43 (±18.6) Male sex: cohort 90 (58.1) Comorbid illness: COPD: cohort 10 (6.5%) Disease severity: acute respiratory failure (24/103 (23.3%) in the no CS group versus 38/52 (73.1%) in the CS group); septic shock (5/103 (4.9%) in the no CS group versus 13/103 (25.0%) in the CS group); invasive ventilation (16/103 (15.5%) in the no CS group versus 27/52 (51.9%) in the CS group)
Interventions	Groups: primary comparison was survivors versus non‐survivors. Secondary comparison made of CS treatment (n = 52) versus no CS treatment (n = 103). Dose: daily median dose equivalent to 80 mg (IQR 80 mg to 160 mg) of methylprednisolone Co‐interventions: antivirals given to n = 132 (85.2) of cohort; antibiotics given to n = 139 (89.7) of cohort
Outcomes	Hospital mortality: raw numbers for mortality were derived from data provided in article. Adjusted odds ratio for mortality given for CS use in multivariate analysis. Adjustment was made for ethnicity, comorbid illness (hypertension, diabetes), symptoms at disease onset (dyspnoea, sore throat), clinical presentation (dyspnoea), laboratory testing (lactate dehydrogenase), and CS treatment in the multivariate analysis. No difference in mortality for low‐dose CS (< 80 mg of methylprednisolone) versus high‐dose CS in a subgroup analysis (9/30 versus 8/22, P = 0.854)
Risk of bias (Newcastle‐Ottawa Scale)	Selection domain score (max 4): 3 Comparability domain score (max 2): 1 Outcome domain score (max 3): 3
Notes	—