2216. Pneumonia in the ICU: Epidemiology and Outcomes

Abstract

BackgroundThe understanding of pneumonia epidemiology has been evolving in recent years with an increased awareness of viral respiratory pathogens since the introduction of respiratory pathogen panels. However, epidemiological and clinical data on pneumonia due to co-infection are lacking.MethodsA single-center retrospective analysis of mechanically ventilated adult patients treated in critical care units from January to October 2018 was performed on patients with one or more pathogen identified via microbiological testing of a bronchoalveolar lavage (BAL) sample. SOFA and APACHE II scores at the time of admission to the critical care unit and SOFA the day the of BAL were obtained, along with ICU length of stay, duration of ventilation, and pathogens. Associations between categorical variables and co-infection status were assessed using Chi-Square tests, Fisher exact tests, and t-tests. Differences in counts of days between coinfection status groups were analyzed by generalized linear models.Results140 bronchoalveolar lavage samples met inclusion criteria, of which 31 were determined to have co-infection with two or more pathogens identified. Of the two methods used to obtain BAL samples, co-infection was found in a higher proportion of patients undergoing bronchoscopic BAL as compared with blinded (35.6% vs. 7.5%; P < 0.0001). Patients with co-infection were determined to be statistically more likely to require a longer duration of mechanical ventilation (P = 0.03); however, there was no difference overall seen in 30 day mortality rate (23.4% vs. 19.57%; P = 0.61).ConclusionCo-infection is a significant risk factor for a prolonged duration of mechanical ventilation compared with patients with single pathogen pneumonia. In addition, the use of bronchoscopic BAL appears to be a significant factor in identifying higher rates of co-infection compared with blinded. Although our study failed to demonstrate a significant difference in mortality rates, this could have occurred due to the small sample size. Disclosures All authors: No reported disclosures.