Abstract |
Background: Urinary tract infections (UTIs) are one of the most common serious
bacterial infection in childhood. UTIs are usually caused by E. coli organisms and run
an uncomplicated course. Non-E. coli UTIs have been associated with urinary tract
abnormalities, infection recurrences and complications. Moreover, the increasingly
complex resistance mechanisms of Gram (-) uropathogens challenge the usual
therapeutic choices in UTIs.
The aim of this study was the development of a predictive model for unusual/resistant
uropathogens in hospitalised children with UTI with the ultimate goal to select the
proper empiric treatment in children with UTIs.
Materials and methods: A retrospective cohort study of all children aged 30 days-
15,9 years hospitalised for UTI at Heraklion Univercity Hospital of Crete, from
January 2007-December 2019 (13years) was carried out. Three distinctive types of
uncommon uropathogens were studied, non-E. coli, ESBL phenotype and unusual to
the community pathogens, such as P. aeruginosa and Enterococcus spp. Risk factors
included, gender, age, delivery type, hospitalisation in neonatal intensive care unit
(NICU), exposure to antibiotics either as chemoprophylaxis or as a short therapeutic
course, abnormal urinary tract imaging and UTI recurrence, which was also studied as
a dependent variable. The identified risk factors, after they were controlled by logistic
regression analysis, were used for the construction of comparative receiver operator
curves (ROC). The predictive power of a model was regarded as adequate when the
area under the curve (AUC) was ≥ 0,8.
Results: The study included 866 UTI episodes (44,7% males) with a mean age of 2,08
years (95% CI 1,88-2,29). Non-E. coli pathogens were isolated in 36,4%, ESBL
phenotype in 11,1% and unusual community pathogens in 13,6% of cases. Caesarian
delivery was reported for 49,5% and NICU hospitalisation in 25,7%. Re-currences
represented 21,9% of the episodes, abnormal imaging was recorded in 32% and
antibiotic exposure in 33% of the episodes. Male gender, short course of antibiotic abnormal imaging and recurrences were identified as risk factors for non-E. coli and
unusual community pathogens while gender and antibiotics exposure were identified
as risk factors for ESBL phenotype in univariate analyses. Age, antibiotics and
presence of vesicouriteral reflux (VUR) were identified as risk factors for recurrences.
The risk factors for each type of resistant pathogen and recurrences were used for the
construction of predictive models with the use of comparative ROC curves, which
were all of adequate predictive power except for ESBL phenotype pathogens. For
non-E. coli AUC was 0.80 (p< 0,0001, 95% CI 0.75-0.84) with 69,94% sensitivity
and 76,41% specificity, for ESBL AUC was 0,73 (p=0,0001, 95% CI 0,67-0,78) with
56% sensitivity and 83,47% specificity, for unusual community pathogens AUC was
0,80 (p< 0,0001, 95% CI 0,76-0,84) with 79,71% sensitivity and 70,24% specificity
and for recurrences AUC was 0,82 (p< 0,0001, 95% CI 0,77-0,86) with 74,81%
sensitivity and 77,09% specificity.
Conclusion: A thorough previous history (gender, age, type of delivery, NICU
hospitalization, antibiotic use, abnormal imaging) can predict the probability of
resistant or unusual pathogen in a UTI episode. The inclusion of additional parameters
from perinatal or family history could increase the models’ predictive power and
contribute to wiser use of antibiotics in UTI management.
|