Blood stream infections (BSI) remain a major cause of morbidity and mortality in immunocompromised children and increase the length of hospitalisation as well as the cost of treatment. The difficulty of documenting bacteremia at the onset of fever in neutropenic patients led to an early administration of broad-spectrum antibiotics, without waiting for further clinical or microbiological evidence of infection. Furthermore, gut decontamination is performed in children with a high-risk of BSI. Does this practice trigger the emergence of multidrug resistant bacteria?

We performed a monocentric, retrospective and descriptive study between January 2014 and December 2017, using a large cohort of pediatric patients with hematological and oncological diseases in a tertiary care pediatric center in France, the Institute of Hematology and Oncology Pediatric (IHOPe) in Lyon. This includes patients with acute leukemia, lymphoma, children who underwent hematopoietic stem cell transplantation (HSCT) (allogenic and autogenic), children with severe combined immunodeficiency syndromes as well as children suffering from solid tumors. A bacteremia was defined by a positive blood culture sample, associated with fever. For coagulase-negative staphylococci (CoNS), two distinct positive blood cultures within 48h were mandatory. The children were divided into 4 infectious risk groups regarding their immunosuppression level. Groups 3 and 4 received non-absorbable antibiotics for gut and oral decontamination. The empirical antibacterial therapy was adjusted to the risk group, as well as a prior identified bacteria colonisation: all patients with a central venous access received vancomycin (stopped after 48hours in the absence of bacterial identification), amikacin and a broad-spectrum penicillin.

During these 4 years, our institution recorded more than 8000 hospitalisations distributed within 3 units: 2 conventional hospitalisation wards and the HSCT unit. Our protected unit is equipped with spacious positive air pressure rooms, for child and parent, with high-efficiency particulate air (HEPA) filters where only a clean non-sterile outfit (gown and mask) are required. We identified 350 cases of BSI, thus an overall incidence of 4,5%. Fifteen to twenty per cent of these children suffered from a second or more re-infection. The incidence of BSI was stable throughout the four years of our study. Gram-positive bacteria represented more than 50% of BSI and CoNS was the most frequent pathogen agent found. Among the others gram-positive isolates, we counted around 5 to 6 cases/year of Streptococcus and Staphylococcus aureus. Among Gram-negative bacteria, enterobacteries accounted for 14 to 20 BSI per year and Pseudomonas aeruginosa for 2 to 4 cases per year. Among the identified bacteria, only 6.5% were multidrug resistant for all our study time. Three highly resistant bacteria were identified in 2 children priorly treated in Romania and Tunisia: 2 Klebsiella pneumoniae producing Carbapenemase (KPC), 1 Enterococcus faecium resistant to glycopeptides. Interestingly, 35% of the BSI occurred in the less immunocompromised group. The other cases were evenly distributed between the 3 more severely immunocompromised groups.

Our empirical initial broad-spectrum antimicrobial coverage is well adapted to the epidemiology of our bacterial agents and shortens hospitalisations. Furthermore, it is not associated with the emergence of antibiotic resistance patterns. The microbiological spectrum of incriminated bacteria is relatively stable over time and we do not observe the emergence of multidrug resistant gram-negative bacteria as well as the absence of vancomycin-resistant enterococci (VRE) in our tertiary center.