Application of WISCA (Weighted Incidence Syndromic Combination Antibiogram) to guide empiric therapy in oncological paediatric patients with febrile neutropenia

Tags: | April 26th, 2020

Authors: E. Barbieri, D. Bottigliengo, P. Costenaro, A. Marzollo, M. Petris, M. Pierobon, G. Biddeci, C. Giaquinto, A. Biffi, D. Donà

Published in: 30th European Congress on Clinical Microbiology and Infectious Diseases (ECCMID), April 2020

Background: Febrile neutropenia (FN) is an acute potentially life-threatening oncological complication which should be treated promptly with antibiotics. With the spread of antibiotic resistance, the choice of an empiric therapy is driven by local epidemiology usually described by cumulative pathogens susceptibilities antibiograms. The WISCA attempts to address the unmet need for syndRome-specific local susceptibility data to guide empirical prescribing, providing estimates for different treatment reg- imens as a weighted average of pathogens susceptibilities. Our aim was to create a WISCA model to inform empirical antibiotic regimens selection for FN in children.

Materials/methods: We included all non-duplicate blood cultures from patients aged 0-17 years with FN admitted to the paediatric oncology/hematology wards in Padua from January 2016 to August 2019. WISCA was developed by estimating the sensibility of 29 antibiotic regimens with a Bayesian probabilities distribution. Moreover, we created a second model with 57 blood cultures excluding potentially contaminant bacteria.

Results: We collected 69 blood cultures, 41 Gram- and 28 Gram+ bacteria. Considering most used combinations such as piperacillin-tazobactam + amikacin the median sensibility was 58% (BUI 33-84%) that increased to 70% (BUI 42-85%) in the second model. When adding a glycopeptide to this combination the median sensibility increased dramatically (Figure 1). The highest median sensibility for a beta-lactam + aminoglycoside combination was 66% (BUI 37-86%; meropenem + amikacin) in the first model and 75% (BUI 46-85%) in the second model; the lowest was 42% (BUI 26-75%; ceftriaxone + amikacin) and 50% (BUI 32- 76%) respectively. Overall mono-treatments had median sensibility lower than 50%, exept meropenem (65%; BUI 35-85) and gentamycin (60%; BUI 33-84%), but in the second model most median sensibilities increased above 50%. WISCA model with median sensibilities and uncertainty intervals is shown in Figure 1.

Conclusions: WISCAs represent a valid tool to maximize the clinical utility of microbiological surveillance data supporting appropriate empirical antibiotic treatment selection, while contributing to conservation of broad-spectrum antibiotics.