Abstract Rationale The role of glucocorticoids in severe community acquired pneumonia (CAP) remains unclear despite recent randomized clinical trials. The CAPE-COD trial demonstrated reduced 28-day mortality with hydrocortisone in severe CAP patients, but similar benefits were not noted in ESCAPe or REMAP-CAP. Real-world evidence with larger study cohorts may provide additional insights into potential benefits of glucocorticoids. In this study, we evaluated the effect of glucocorticoids in severe CAP using methodologies designed to reduce confounding bias. Methods We conducted a retrospective cohort study of severe CAP patients admitted to 12 UPMC Health System hospitals from 2008 to 2016. We included critically ill adults with a primary ICD-9/10 diagnosis of pneumonia present on admission. We excluded “comfort measures only” patients. Glucocorticoid exposure was defined by receipt of at least one oral or intravenous glucocorticoid medication within 72 hours of admission. We used doubly robust estimators to determine the effectiveness of glucocorticoids on a primary outcome of 28-day mortality. All analyses accounted for potential baseline confounders, including demographics, vital signs, comorbidities, lab values, and organ support. Results 11,048 patients met inclusion criteria for severe CAP with 4,868 (43.7%) receiving early glucocorticoids. Patients with glucocorticoids had a higher prevalence of chronic obstructive pulmonary disease (72% versus 46%) and increased initial mechanical ventilation use (78% versus 65%), but were otherwise similar in baseline demographics and initial vasopressor use (∼9% in each group). Receipt of glucocorticoids was associated with higher crude 28-day mortality (23.1% vs. 21.0%). In doubly robust analysis adjusted for baseline confounders, there was no significant difference in mortality, with estimated absolute risk difference of -1.3% (-2.8% , 0.2%). Conclusion In critically ill patients with severe CAP, glucocorticoid use was not shown to impact 28-day mortality outcomes. These real-world observations do not show benefit to glucocorticoid use in ICU patients with severe CAP. Limitations of this study include use of ICD-9/10 codes for identification of the patient cohort and possible presence of additional unmeasured confounders. Future studies should better characterize real-world prescribing patterns and investigate how specific glucocorticoid regimens impact outcomes. This abstract is funded by: 5R01GM141081
Poluch et al. (Fri,) studied this question.