451Case Finding for Invasive Aspergillosis in Children

Background. Invasive aspergillosis (IA) continues to pose a serious threat to immunocompromised children. We sought to determine the recent epidemiology of IA and compare the utility of multiple diagnostic strategies for IA case finding. Methods. A retrospective case series of patients ≤18 years of age with proven, probable or possible IAdiagnosed from 2004-2013 was performed at NewYork-Presbyterian Morgan Stanley Children’s Hospital, Columbia University Medical Center, New York, NY. Case finding was performed by review of positive cultures and serum galactomannan and by natural language processing to detect relevant pathology findings from autopsy or surgery (e.g., septated hyaline branching, angioinvasion) and from radiographic findings (e.g., halo, air-crescent, nodules). Results. Twenty-one children were identified by culture (n = 8), galactomannan (n = 10), pathology (n = 14), and/or radiology (19) of whom 13, 4, and 4 cases met criteria for proven, probable, and possible IA, respectively. Natural language processing detected 6 additional cases not detected by either culture or galactomannan. Of the 21 children, 9, 2, and 2 had undergone bone marrow (BM), heart, and liver transplant, respectively, of whom 2 had both BM and solid organ transplant. The remainder were on chemotherapy (n = 5) or had other comorbid conditions including lupus (n = 2). IA rates were 2.8%, 0.9%, and 1.2% in BM, heart, and liver transplant recipients (p = 0.2). The most common sites of infection were lung (n = 12), sinuses (n = 3), and heart (n = 3); 5 patients had multi-site infections. Most (81%) had received antifungal agents within 90 days of IA diagnosis. Prior use of antifungal agents was similar in those with (n = 7/8) and without (10/13) positive cultures. Crude mortality was 67% at 3 months and 86% at 6 months. Conclusion. In this 10 year study, IA rates appeared to be lower than previously reported, although crude mortality was similar. The use of natural language processing for case finding could improve IA surveillance efforts. De Pauw B, et al. Clin Infect Dis 2008; 46:1813-21. Disclosures. All authors: No reported disclosures.