Risk factors associated with life-threatening infections in children with febrile neutropenia: a data mining approach

To determine risk factors (RFs) and their relationship with life-threatening infection (LTI) in children with febrile neutropenia (FN).

In this cross-sectional study, from December 2008 to November 2009, all children with FN admitted to Dr Sheikh Pediatric Hospital were enrolled. For each patient, demographic, clinical, and laboratory data were recorded and they were followed up for occurrence of LTI.

One hundred and twenty episodes of FN in 68 patients were analyzed. The most common underlying disease was acute lymphoblastic leukemia (53.3%), 9 (7.5%) died from an infection and 35 patients (29.1%) had a LTI. Five variables were identified as RFs for LTI, that is, body temperature ≥39°C (P=0.000), presence of mucositis (P=0.000), abnormal chest x-ray (P=0.001), platelet count <20,000/mm (P=0.000), and absolute neutrophil count <100/mm (P=0.001). Risk of LTI was increasing according to number of RFs presented at the beginning of admission (from 2.8% in patients without RF to 100% in patients with 5 RF). Data mining analysis showed relationship between RFs with platelet count as the most important variable in the high-risk group for LTI.

Evaluation of important RFs and judging the severity of patients' condition by studying the importance and relationship between RF at the time of admission can be a useful method for screening LTI in children with FN.