Fifty RCTs were included in the review (n=3,323). Seventeen RCTs had adequate allocation concealment, 26 had double or triple blinding and 14 had adequate allocation concealment and double blinding.
Corticosteroid prophylaxis was associated with a significant reduction in CRP concentration (WMD -44.2mg/L, 95% CI -15.4 to -72.9, p<0.01) and IL-6 concentrations (WMD -148pg/mL, 95% CI -114.8 to -181.1, p<0.01), although these were associated with significant heterogeneity (I2=90.7% for CRP and 98.1% for IL-6). There was a non-significant trend towards reduction of IL-8 after corticosteroid use, also with significant heterogeneity (I2=97.0%).
Corticosteroid prophylaxis was associated with a significant reduction in the risk of atrial fibrillation (RR 0.74, 95% CI 0.63 to 0.86, p<0.01, NNT 10; n=1,509) and hyperglycaemia that required insulin (RR 1.49, 95% CI 1.11 to 2.01, p<0.01; n=503). Corticosteroid prophylaxis did not have a significant effect on the risk of all-cause infection or mortality. No significant heterogeneity was found for these analyses.
Corticosteroid prophylaxis was associated with a significant reduction in duration of mechanical ventilation (WMD -0.68 hours, 95% CI -0.03 to -1.33, p=0.04; n=1,789). This outcome varied according to dose strata. High dose strata were associated with increased duration of mechanical ventilation (WMD 2.1 hours, 95% CI 1.76 to 2.52, p<0.01; n=276). The high dose corticosteroid effect was significantly different from low dose (p=0.03) or medium dose (p<0.01) effect. No significant difference was found between low and medium dose effects. No significant heterogeneity was found for these outcomes.
Corticosteroid prophylaxis was associated with a reduction in the length of intensive care unit stay (WMD -0.37 days, 95% CI -0.21 to -0.52, p<0.01) and hospital stay (WMD -0.66, 95% CI reported as -0.77 to -1.25, p=0.03). Both analyses were associated with significant heterogeneity (I2=89.1% for intensive care unit stay and 77.2% for hospital stay).
No significant difference between the three dose strata was found for the outcomes CRP, IL-6 and IL-8 concentrations, atrial fibrillation, hyperglycaemia that required insulin and length of intensive care unit or hospital stay.
Sensitivity analyses showed similar results. The funnel plot suggested publication bias. The adjusted effect of corticosteroid on atrial fibrillation (using the trim and fill method) remained significant (RR 0.75, 95% CI 0.63 to 0.89, p<0.01).