Four RCTs (n=1,261) were included in the review. Studies were judged to be of high quality. All studies reported adequate allocation concealment, randomisation and outcome assessment and intention-to-treat analysis. Losses to follow-up ranged from 0 to 2.3%.
Short-term mortality rates ranged from 10% to 54%. There was no significant difference between corticosteroids and placebo in short-term mortality (25.9% versus 27.9%, RR 0.81, 95% CI 0.54 to 1.20, I2=54%). Corticosteroids were associated with a significant reduction in risk of hearing loss (RR 0.73, 95% CI 0.55 to 0.97, I2=0%) and a non-significant reduction in risk of neurologic sequelae other than hearing loss (23.5% versus 30.8%, RR 0.67, 95% CI 0.45 to 1.01, I2=56%).
A significant interaction was found between the effects of corticosteroids and income status of the country and between effects and prevalence of HIV infection. In high-income countries, corticosteroids were associated with a significant reduction in short-term mortality (RR 0.5, 95% CI 0.27 to 0.92, I2=0%, NNT to prevent one death=12.5) and a significantly lower risk of neurologic sequelae other than hearing loss (RR 0.58, 95% CI 0.36 to 0.94, I2=0%, NNT=11.0). In studies assumed to have a low-incidence of HIV (European studies), corticosteroids were associated with significantly lower short-term mortality compared to placebo (RR 0.66, 95% CI 0.44 to 0.99, I2=0%).