Cao 2016.

Methods	Multicentre retrospective cohort plus propensity score‐matched case‐control study
Participants	Country: China (mainland) Setting: in‐hospital Number of individuals: 288 Inclusion criteria: individuals aged ≥ 14 years admitted to hospitals throughout China with pneumonia and laboratory‐confirmed H7N9 influenza infection. All had radiological evidence of pneumonia. Exclusion criteria: no data due to being treated outside mainland China, aged under 14 years, incomplete data, physicians declined to participate Influenza type: A/H7N9 Median age (years): 58 (IQR 45 to 68); not reported for individual treatment groups Male sex: 201 (69.8%) Comorbid illness (cohort): asthma/COPD 14 (4.9); immunosuppressed 3 (1.0) Disease severity: moderate to severe ARDS (Berlin definition of severe hypoxaemia defined as ratio of PaO2/FiO2 of ≤ 200 mmHg and Positive End‐Expiratory Pressure (PEEP) of ≥ 5 cm H2O, associated with bilateral opacities on chest x‐ray that could not be fully explained by cardiac failure or fluid overload) 207 (71.9)
Interventions	Groups: CS (n = 204) versus no CS (n = 84) Dose: low‐moderate dose 168 (82.4); high dose 36 (17.6). Median dose (mg/day) 80 (IQR 40 to 120) Definition of dose: low‐moderate dose 25 to 150 mg/day of methylprednisolone or its equivalent; high dose > 150 mg/day of methylprednisolone or its equivalent Duration given for: median (days) 7 (IQR 4.0 to 11.3) Time to initiation from onset of illness: median (days) 7.0 (IQR 5.0 to 9.4) Antibiotics (cohort): total given antibiotics 261 (90.6); "appropriate" antibiotics 133 (46.2) Antivirals (cohort): 285 (99.0); median initiation time (days from onset) 6.3 (IQR 4.7 to 8.2)
Outcomes	Mortality: 30‐day and 60‐day mortality adjusted hazard ratios reported (multivariate Cox regression analysis). Also compared low‐moderate‐dose CS versus control and high‐dose CS versus control, and included a propensity score‐matched case‐control analysis (65 pairs) Adverse events: hospital‐acquired infections in CS versus no CS groups assessed by propensity matched pairs analysis Viral shedding: in CS group versus no CS group assessed by propensity matched pairs analysis with further stratification according to high/low‐moderate CS dose
Risk of bias (Newcastle‐Ottawa Scale)	Mortality Selection domain score (max 4): 4 Comparability domain score (max 2): 2 Outcome domain score (max 3): 3 Hospital‐acquired infection Selection domain score (max 4): 4 Comparability domain score (max 2): 2 Outcome domain score (max 3): 2 Viral shedding Selection domain score (max 4): 4 Comparability domain score (max 2): 2 Outcome domain score (max 3): 2
Notes	Adjusted estimates for 30‐day and 60‐day mortality presented following logistical regression in a cohort study. Separate estimates given in a propensity matched case‐control study. In this study outcomes were stratified according to different corticosteroid regimens (high dose and low‐to‐moderate dose).