Despite analysing our data from multiple perspectives, we cannot rule out that sicker patients received antibiotics sooner and that the results are confounded by these patients being at higher risk of death;however, this confounding should bias toward the null.