Methods: Patients hospitalized at our institution from 2004-2014 with a diagnosis of probable or proven IA by EORTC criteria were identified. Patients were included if culture data from the site of IA were available. Coinfection was defined as microbiologic evidence of a likely pathogen identified within 7 days of IA diagnosis. Proportions were compared with Fisher exact test. Variables with P<.1 were used in multivariable logistic regression analysis.
Results: Coinfection occurred in 140 (37.0%) of 378 patients meeting inclusion criteria. Bacteria were found in 95 patients (25.1%, gram negative rods accounted for 71.4% of all bacteria), mycobacteria in 21 (5.6%), viruses in 35 (9.3%), fungi in 25 (6.6%), and a parasite in 1 (0.3%). >1 type of copathogen was found in 34 patients. The prevalence of coinfection varied significantly by site of IA (lung 35.6%, pleura 29.4%, sinus 20.0%, skin 44.8%, central nervous system (CNS)/eye 7.4%, bone 5.3%; P<.01) as well as the major IA risk factor (hematologic malignancy 26.9%, solid organ transplantation [SOT] 68.6%, other factors 41.2%; P<.01). The following independently predicted coinfection: SOT (adjusted odds ratio [aOR] 4.03, 95% confidence interval [CI] 1.95-8.64), A. fumigatus species (aOR 2.25, 95% CI 1.23-4.12), and cytomegalovirus (CMV) seropositivity (aOR 2.40, 95% CI 1.24-4.85). Neither G-CSF administration nor the use of broad-spectrum antimicrobials prior to diagnosis protected against coinfection.
Conclusion: Coinfection was most prevalent in IA with skin or lung involvement. SOT recipients, CMV seropositivity, and A. fumigatus as the cause of IA are independent risk factors. Coinfection was less frequent in patients with hematologic malignancy and when IA involved the CNS, eye, or bone. Bacteria, gram-negative rods in particular, are the most common copathogen. Our findings highlight the clinical settings in which clinicians should have a heightened suspicion for coinfections.
See more of: Poster Abstract Session