Methicillin‐resistant Staphylococcus aureus and Pseudomonas aeruginosa community acquired pneumonia: Prevalence and locally derived risk factors in a single hospital system

OBJECTIVES: Current American Thoracic Society/Infectious Disease Society of America (ATS/IDSA) community‐acquired pneumonia (CAP) guidelines expand the CAP definition to include infections occurring in patients with recent health care exposure. The guidelines now recommend that hospital systems determine their own local prevalence and predictors of Pseudomonas aeruginosa and methicillin‐resistant Staphylococcus aureus (MRSA) among patients satisfying this new broader CAP definition. We sought to carry out these recommendations in our system, focusing on the emergency department, where CAP diagnosis and initial empiric antibiotic selection usually ooccur. METHODS: We performed a retrospective cohort study of patients admitted with CAP through any of 3 EDs in our hospital system in Northern California between November 2019 and October 2021. Inclusion criteria included an ED admission diagnosis of pneumonia or sepsis, fever or hypothermia, leukocytosis or leukopenia, and consistent chest imaging result. SARS‐CoV‐2‐positive cases were excluded. We abstracted variables historically associated with P. aeruginosa and MRSA. Outcome measures were prevalence of P. aeruginosa and MRSA in the overall clinically defined cohort and among microbiologically confirmed cases and predictors of P. aeruginosa or MRSA isolation, as determined by univariate logistic regression, bootstrapped least absolute shrinkage and selection operator, and random forest analyses. Additionally, we describe the iterative process used and challenges encountered in carrying out the new ATS/IDSA guideline recommendations. RESULTS: There were 1133 unique patients who satisfied our definition of clinically defined CAP, of whom 109 (9.6%) had a bacterial pathogen isolated. There were 24 P. aeruginosa isolates and 11 MRSA isolates in 33 patients. Thus, the prevalence P. aeruginosa and MRSA was 2.9% in the overall CAP cohort, but 30.3% in the microbiologically confirmed cohort. The most important predictors of either P. aeruginosa or MRSA isolation were tracheostomy (odds ratio [OR] 22.08; 95% confidence interval [CI] 10.39–46.96) and gastrostomy tube (OR 14.7; 95% CI 7.14–30.26). Challenges included determining the suspected infection type in patients admitted simply for “sepsis”; interpreting dictated radiology reports; determining functional status, presence of indwelling lines and tubes, and long‐term care facility residence from the electronic health record; and correctly attributing culture results to pneumonia. CONCLUSION: Prevalence of MRSA and P. aeruginosa was low among patients admitted in our medical system with CAP – now broadly defined – but high among those with a microbiologically confirmed bacterial etiology. Our locally derived predictors of MRSA and P. aeruginosa can be used to aid our emergency physicians in empiric antibiotic selection for CAP. Findings from this project might inform efforts at other institutions.