Predictive Performance of Risk Factors for Multidrug-Resistant Pathogens in Nosocomial Pneumonia

Rationale: In 2017, the International European Respiratory Society/European Society of Intensive Care Medicine/European Society of Clinical Microbiology and Infectious Diseases/Latin American Thoracic Society (European) guidelines defined new risk factors for multidrug-resistant (MDR) pathogens in patients with nosocomial pneumonia.Objectives: To assess the predictive performance of these newly defined risk factors for MDR pathogens.Methods: We enrolled 507 adult patients with nosocomial pneumonia who were treated in six intensive care units at the Hospital Clinic of Barcelona in Spain. Of the 503 patients at high MDR pathogen and mortality risk, 275 (54%) had no septic shock and 228 (46%) had septic shock.Results: Admission to hospital settings with high rates of MDR pathogens (n = 421; 83%) and prior antibiotic use (n = 399; 79%) showed the highest prevalence in the overall population, with sensitivities of 92% and 85% and negative predictive values of 85% and 82%, respectively. However, low specificities and low positive predictive values were found. Previous respiratory MDR pathogen isolation was less common (n = 17; 3%) but presented a specificity and positive predictive value of 100%. The area under the receiver operating characteristic curve was less than 0.6 for all risk factors and combinations.Conclusions: The risk factors proposed by the European Respiratory Society/European Society of Intensive Care Medicine/European Society of Clinical Microbiology and Infectious Diseases/Latin American Thoracic Society showed low accuracy for predicting MDR pathogens in intensive care unit acquired pneumonia (ICU-AP). Admission to hospital settings with high rates of MDR pathogens and prior antibiotic use were the most prevalent risk factors, with a high sensitivity for predicting these microorganisms; prior positive cultures for MDR pathogens showed high specificity but very low sensitivity. Combinations of risk factors did not show any great accuracy for predicting these microorganisms. Further studies assessing combined strategies of risk stratification and complementary methods are now warranted.