542 - Predictive factors of invasive infection in febrile oncology patients based on the oncology process

Publication Number: 2527.542

Borja Gómez, Hospital Universitario Cruces, Barakaldo, Pais Vasco, Spain; Jose Antonio Alonso Cadenas, Hospital Infantil Universitario Niño jesús, Madrid, Madrid, Spain; Monica Sancosmed, HOSPITAL VALL D'HEBRON, BARCELONA, Catalonia, Spain; Ana Jové-Blanco, Hospital General Universitario Gregorio Marañón, Madrid, Madrid, Spain; Mirian Moreno, Hospital Cruces, Bilbao, Pais Vasco, Spain; Oriol Quintana, Hospital Universitario Cruces, Bilbao, Pais Vasco, Spain; Fernando Almarza, Hospital Universitario Cruces, BILBAO, Pais Vasco, Spain; Ana María Martínez Álvarez, Sección de Urgencias Pediátricas, Servicio de Pediatría, Hospital Clínico Universitario Virgen de la Arrixaca (Murcia), Murcia, Murcia, Spain; Paula García-Sánchez, La Paz University Hospital, Madrid, Madrid, Spain; Izaskun Olaciregui, Donostia University Hospital, San Sebastian, Pais Vasco, Spain; Luisa Barón, Hospital Universitario 12 de Octubre, Madrid, Madrid, Spain; Andrea Riego, Hospital Universitario 12 de Octubre, Madrid, Madrid, Spain; Manuel Gijón, Hospital Universitario 12 de Octubre, MADRID, Madrid, Spain; Nerea González, Hospital Universitario de navarra, Bilbao, Pais Vasco, Spain; Nuria Gilabert Iriondo, Hospital Universitario Son Espases, Calvia, Islas Baleares, Spain; Irene Cinta Barceló Carceller, Hospital de la Santa Creu i Sant Pau, Barcelona, Catalonia, Spain; María Fernandez, H, Oviedo, Asturias, Spain; Elena Aquino Oliva, Hospital Universitario de Toledo (Spain), Toledo, Castilla-La Mancha, Spain; Maria Esteban, Hospital Universitario de Toledo, Las Rozas, Madrid, Spain; Emilia Urrutia, Hospital universitario Virgen de las Nieves, Granada, Andalucia, Spain; Santiago Mintegi Raso, Hospital Universitario Cruces, Bilbao, Pais Vasco, Spain

Background: oncology patients frequently experience febrile episodes during their chemotherapy treatment. The prevalence of invasive bacterial infection and sepsis during these febrile episodes varies by cancer type. We hypothesize that predictive factors for invasive infection also differ between these two groups.

Objective: to identify clinical and laboratory independent risk factors for invasive infection in febrile oncology patients, according to cancer type.

Design/Methods: multicenter prospective study including 15 hospitals members of the Spanish Pediatric Emergency Research Group (RISEUP). We included oncology patients receiving chemotherapy and presenting with fever to the Emergency Department between November 2022 and October 2024. Invasive infection was defined microbiologically (isolation of a bacterial pathogen in a sterile fluid or a commensal bacterial species in two different blood cultures) or clinically (meeting criteria for sepsis, receiving inotropic support or developing sequelae).
We performed separate multivariate analyses for hematologic cancers and solid tumors. Hematologic cancers were further classified in two subgroups based on the myelotoxicity of the chemotherapy being received. The potential predictive factors included were age, sex, fever degree, Pediatric Assessment Triangle (PAT), presence of tachycardia adjusted for age and temperature, leukocyte, neutrophil, monocyte and platelets counts, and C-reactive protein and procalcitonin values. Since risk of infection increases with both high and low neutrophil counts, this variable was analyzed both as continuous and categorical (profound, severe, moderate and mild neutropenia, normal count, neutrophilia). Optimal cut-off points for quantitative variables were determined.

Results: we included 1197 episodes: 747 hematologic cancers and 450 solid tumors. The median duration of fever was 3 hours (IQR 1-8) and in 95.2% of them the patient was stable by initial PAT. Rate of severe neutropenia ( < 500 neutrophils/mcL) was 46.6%.
We diagnosed 113 invasive infections (9.4%), 106 (93.8%) with microbiological confirmation. Eighteen episodes (1.5%) met clinical severity criteria. Table 1 shows the independent risk factors identified by cancer type. Table 2 presents the performance of each model.

Conclusion(s): predictive factors for invasive infection differ between patients with hematological malignancies and solid tumors. The neutrophil count alone appears less predictive than expected. Two models incorporating fever degree and blood biomarkers allow identifying a low-risk subgroup of patients.

Table 1: independent risk factors identified in the multivariate analyses for hematologic cancers and solid tumors
*the presence of pallor as the only abnormal finding was not considered indicative of an altered Pediatric Assessment Triangle if it was reported by parents as baseline in the patient and it was consistent with a known anemia related to his/her oncological disease or treatment.
For continuous variables, the OR refers to each increase indicated in brackets.

Table 2: performance of the model including the following low-risk factors
* the presence of pallor as the only abnormal finding is not considered indicative of an altered Pediatric Assessment Triangle if it is reported by parents as baseline in the patient and it is consistent with a known anemia related to his/her oncological disease or treatment.