Risk factors for community-onset pneumonia caused by drug-resistant pathogens: A prospective cohort study.

Introduction: There is no consensual definition of risk factors for drug resistant pathogens (DRP) in community-onset pneumonia (COP). Healthcare-associated pneumonia criteria have been abandoned because they were found to have weak discriminative power. Our aim was to identify risk factors for DRP in COP.

Methods: Prospective cohort study, conducted over a two years' period, in a community-based hospital, including all adult patients with COP criteria. Univariate and multivariate logistic regression modeling were performed to understand the association of risk factors (demographic, clinical and epidemiological) with COP by a DRP (PES: Pseudomonas aeruginosa, extended-spectrum ß-lactamase producing Enterobacteriaceae, Methicillin-resistant Staphylococcus aureus; and other non-fermenting gram-negative bacteria, namely Acinetobacter baumannii).

Results: A total of 660 cases of COP were included, with a mean (±SD) age of 74±15 years and 58.9% of males. Microbiological documentation was possible in 32.6% of the cases. There were 197 cases selected for further analysis, of which 37 were cases of PES. The multivariate logistic regression model retained antibiotic use in the previous 90 days (adjusted OR=4.411, 95%CI [1.745-11.148]) and being bed-ridden (adjusted OR=5.492, 95%CI [2.121-14.222]), adjusted for Charlson's Index, CURB 65 and provenience from a long-term care facility. The area under the ROC curve for this model was 0.832, 95%CI [0.756-0.908], higher than the application of the HCAP criteria (AUROC = 0.676, 95%CI [0.582-0.770]).

Conclusion: In this study, antibiotic use in the previous 90 days and being bed-ridden were independently associated with COP caused by DRP, after adjustment for Charlson's Index, CURB 65 and provenience from a long-term care facility.