Urinary tract infection (UTI) is one of the most frequent bacterial infections in children together with bacterial respiratory infections. Underlying diseases such as vesicoureteric reflux (VUR) also arise in some patients with UTI.
VUR is an important risk factor for recurrent UTI, and recurrent UTI contributes to the development of renal scarring or deterioration in renal function. Therefore, it is important for pediatricians to diagnose VUR at an early stage and perform adequate follow-up renal function studies to prevent the development of complications. However, whether we should perform voiding cystography (VCG) in all patients during the first episode of UTI is controversial, as VCG is an invasive procedure that requires catheterization and exposure to X-rays. In addition, VUR is present in only 25–40% of children with febrile UTI and is often of grade I or II severity.
In this study, we aimed to evaluate whether sex, clinical variables, laboratory variables, or ultrasonography (US) could predict the presence of VUR in patients presenting with UTI for the first time and define the criteria for performing VCG in these patients.
We retrospectively studied the medical records of 286 patients (median age: 5 months; range: 0.5–169 months) who presented with UTI for the first time without other evidence of urologic disease and who were hospitalized in Kumamoto Regional Medical Center between January 2004 and December 2013. The data of these patients were retrospectively analyzed by reviewing their medical records.
A total of 286 patients presented with UTI for the first time, and of these, 200 patients (69.9%) underwent VCG. Sixty-nine patients (34.5%) (median age = 4 months; range = 1–122 months) had VUR, and 131 patients (65.5%) (median age = 5 months; range = 1–143 months) did not exhibit VUR. The presence or absence of VUR was determined by VCG and graded from I to V according to the International Reflux Study in Children guidelines. Patients who did not undergo VCG (median age = 7 months; range = 0.5–169 months), included those in whom VUR was considered absent on the basis of the first US (N = 65); patients who displayed transient abnormal US findings (N = 12); patients who did not undergo US during the first episode of UTI (N = 4); patients who were transferred from another institution for comprehensive examination and treatment because they developed AFBN, renal abscess, or an inherited disease of the urinary system (N = 4); and patients whose parents did not provide consent for participation in this study (N = 1). Abdominal US findings were classified as grade I if there were no abnormal findings; as grade II in cases of mild pyelectasis; as grade III in cases of pyelectasis and dilation of the urinary tract; as grade IV in cases of pyelectasis, dilation of the urinary tract, and abnormal movement of the ureter resembling urine reflux in the urinary tract ( Movie 1); and as grade V when grade ≥III hydronephrosis (13), hydroureter of ≥5 mm in diameter, or pathological features such as AFBN, renal abscess, or renal scarring were present. US findings were double-checked by 2 technicians.
This study included 200 children presenting with UTI for the first time and whose diagnosis of VUR was confirmed using VCG. Sixty-nine patients had VUR of various grades as follows: grade I (n = 3), grade II (n = 8), grade III (n = 17), grade IV (n = 26), and grade V (n= 15). Twenty of these patients had VUR on both sides, and we regarded such patients as having high-grade VUR. Sex (P = 0.001), peak blood CRP levels (P < 0.001), the duration of fever (days) after the administration of antibiotics (P = 0.007), and the presence of abdominal US findings (P < 0.001) (Table 1) were significantly different between patients with and without VUR. The US grade that most accurately discriminated subjects with and without VUR was grade IV. Ninety-four percent of patients with VUR (65/69) and 50% of non-VUR patients (65/131) presented with an abdominal US grade of III or greater. Peak blood CRP content was identified as the second-most discriminatory variable regarding the presence or absence of VUR. Seventy-one percent of patients with VUR (49/69) and 28% of patients without VUR (37/131) had peak blood CRP levels of ≥80 mg/L.
We performed multivariable logistic regression analysis using the variables age, sex, duration of fever, CRP level, and grade of US findings and calculated the probability of positivity for VUR. The grade of US findings most accurately predicted positivity for VUR. When we considered a x value of ≥0.5 as indicating VUR positivity (log (x/1 – x) = −0.0092 × [Age] + 0.9553 × [Sex] + 0.2138 × [Fever duration] + 0.00875 × [CRP] + 1.068 × [US grade] − 5.9493), the sensitivity, specificity, positive predictive value, and negative predictive value were 77.6, 87.0, 75.4, and 88.4%, respectively. Grade IV US findings and blood CRP levels of 80 mg/L produced a x value of ≥0.5.
In conclusion, sex, fever duration, blood CRP levels, and the grade of US findings were useful predictors of VUR according to multivariable logistic regression analysis. Among these variables, the grade of US findings was the most important factor for predicting VUR. VCG should be performed for pediatric patients with CRP levels ≥80 mg/L and with grade IV–V US findings, whereas VCG can be avoided in patients with CRP levels <80 mg/L and grade <III US findings. When either CRP levels are at least 80 mg/L or the US grade is IV–V, then the formula derived from our multivariable logistic regression analysis will be informative for predicting VUR.
Jun Kido and Masaaki Yanai
Department of Pediatrics, Kumamoto Regional Medical Center, Kumamoto City, Kumamoto, Japan.