Validation of the Giannella Risk Score for the Prediction of Infection by Carbapenemase-producing Enterobacteriaceae in the Pediatric Population.

BACKGROUND Despite efforts made to prevent the spread of multi-drug-resistant bacteria, carbapenemase-producing Enterobacteriaceae (CPE) has become one of the most dangerous threat worldwide. However, data on the epidemiology of CPE and on the correlation between CPE colonization and infection are scanty. The objectives of this study were first to describe the epidemiological characteristics of colonizations and invasive CPE infections in the pediatric population, and second, to apply the Giannella Risk Score (GRS) to the pediatric population for the assessment of the risk of invasive CPE infection in patients with already known colonization. METHODS Pediatric patients with evidence of colonization by CPE were retrospectively enrolled. For each colonized patient, the subsequent development of an infection by CPE was then assessed for a 90-day period after the first CPE isolation; GRSs were compared between patients who had developed any type of CPE infection and those without infection. RESULTS A total of 215 patients (113 males and 102 females) with at least 1 isolation of CPE during hospitalization were analyzed. Median age was 5.6 years [interquartile range (IQR), 1.89-12.2 years]. Overall, 28 CPE infections (13%) were documented: 23 blood stream infections and 5 complicated urinary tract infections. The 30-day mortality of invasive CPE infections was 34.8%. The GRS values in patients with any CPE infection were statistically higher than in noninfected patients: median GRS 9 (IQR, 4-12.5) versus 4 (IQR, 2-4), respectively; P < 0.0001. The analysis of the receiver operating characteristic curves identified a GRS cut-off value ≥8 as the best predictor of CPE infection. The likelihood ratio of the results was <2 and the informedness of the test had a value <0.50. CONCLUSIONS Our study confirms that the spread of CPE is an impelling problem also in the pediatric population, with a high mortality rate of invasive infections. However, the application of the GRS appears to be poorly informative in the pediatric setting; it might sometimes help to identify patients at very low risk of CPE infection, in whom it is reasonable to spare targeted antimicrobial treatments.