Objective: The use of corticosteroids is frequent in critically-ill patients. However, little information is available on their effects in patients with intensive care unit–acquired pneumonia. We assessed patients’ characteristics, microbial etiology, inflammatory response, and outcomes of previous corticosteroid use in patients with intensive care unit–acquired pneumonia.
Design: Prospective observational study.
Setting: Intensive care units of a university teaching hospital.
Patients: Three hundred sixteen patients with intensive care unit–acquired pneumonia. Patients were divided according to previous systemic steroid use at onset of pneumonia.
Measurements and Main Results: Survival at 28 days was analyzed using Cox regression, with adjustment for the propensity for receiving steroid therapy. One hundred twenty-five (40%) patients were receiving steroids at onset of pneumonia. Despite similar baseline clinical severity, steroid treatment was associated with decreased 28-day survival (adjusted hazard ratio for propensity score and mortality predictors 2.503; 95% confidence interval 1.176–5.330; p = .017) and decreased systemic inflammatory response. In post hoc analyses, steroid treatment had an impact on survival in patients with nonventilator intensive care unit–acquired pneumonia, those with lower baseline severity and organ dysfunction, and those without etiologic diagnosis or bacteremia. The cumulative dosage of corticosteroids had no significant effect on the risk of death, but bacterial burden upon diagnosis was higher in patients receiving steroid therapy.
Conclusions: In critically-ill patients, systemic corticosteroids should be used very cautiously because this treatment is strongly associated with increased risk of death in patients with intensive care unit–acquired pneumonia, particularly in the absence of established indications and in patients with lower baseline severity. Decreased inflammatory response may result in delayed clinical suspicion of intensive care unit–acquired pneumonia and higher bacterial count.