Survival with optimal medical management in a cohort of severe necrotizing bacterial lung infections

Abstract

Background: Necrotizing pneumonia and lung gangrene represent a continuum of severe lung infection. Traditionally, severe cases have been referred for surgical debridement. However, this has been linked to high mortality. Some groups have published encouraging results using a conservative medical approach. Unfortunately, lack of a standardized definition of necrotizing pneumonia has precluded meaningful comparison between medical and surgical approach in severe cases. Our objective was to describe the outcome of a cohort of severe necrotizing pneumonia treated with optimal medical management. Methods: We conducted an observational retrospective study by reviewing charts and radiology records of patients hospitalized between 2006-2019 in a tertiary center. We included all patients with severe necrotizing infection, defined as a necrotizing cavity involving at least 50% of a lobe, or smaller multilobar cavities. We made no distinction between necrotizing pneumonia and gangrene as there are no standardized criteria. Results: A total of 50 consecutive patients were included. On imaging, 42% had multilobar cavities and mean diameter of the largest cavity in each case was 5.9 cm. 50% required mechanical ventilation (median duration 12 days) and 44% needed vasopressors. Four patients (8%) had decortication surgery, while none underwent lung resection. Four patients (8%) died. The extent of infiltrates and number of cavities were not associated with mortality but the extent of infiltrates was associated with risk of intubation (P=0.004). Conclusions: We presented one of the largest series of medically-treated severe necrotizing lung infections in the pre-coronavirus disease-2019 (COVID-19) era. The overwhelming majority of patients recovered with optimal medical management alone. Our results strongly support avoiding pulmonary resection in patients with severe necrotizing bacterial lung infections.

Topics & Concepts