A prospective, descriptive observational study was carried out in a provincial reference PICU with seven beds. The Unit treats medical and surgical disorders, with the exception of cardiovascular surgery. We included all patients between 15 days and 14 years of age admitted for over 48h in the period between 1 January and 31 December 2009. The exclusion criteria were as follows: age under 15 days or over 14 years, the presence of urinary infection or pyelonephritis upon admission or in under 48h after admission, and patients remaining in the PICU for less than 48h.

The following data were collected from all subjects: age, gender, reason for admission, medical or surgical disease, severity in the first 24h (PRISM III and PIM II scores), and personal history of relevance (infant brain paralysis, premature delivery, spina bifida, heart disease, neuromuscular disease, liver or kidney disease, immunodeficiency or hematological disease). We documented the possible factors related to nUTI: bladder catheter, previous hospital stay, corticosteroid treatment, immunosuppressors, surgery in the 2 months prior to admission, neutropenia, and antibiotic treatment received. Clinical information indicative of UTI was collected on a daily basis in all cases. The patients with a bladder catheter were subjected to daily urine culture until discharge from the PICU or removal of the catheter. In patients without bladder catheterization, or after removal of the catheter, urine cultures were obtained upon admission and every 48h until discharge. In addition, a urine culture was obtained 48h after transfer to the hospital ward.

Nosocomial infections were defined as all infections not present at the time of admission but which appeared within 48h after admission to the PICU. The diagnosis was based on the criteria of the CDC.6,7

Contamination was defined by urine cultures with low counts (≤104cfu/ml) of microorganisms normally present in the skin or external or internal genitalia. Mixed growth cultures are usually indicative of fecal microbiota contamination, though clinical evaluation is required in such cases.

Colonization in turn was defined as the presence of microorganisms in urine culture, without clinical repercussions, and which disappear after removal of the bladder catheter.

The severity of the disease upon admission was assessed using the PRISM III and PIM II scales, both validated for pediatric patients.8

In catheterized children urine cultures and systematic urine tests were made on a daily basis. In order to avoid false-positive readings due to colonization, in those children with positive urine culture findings we replaced the bladder catheter and then repeated urine culture for confirmation purposes. If bladder catheterization was discontinued during admission, sample collection was carried out as in the group described below.

In non-catheterized children with preserved sphincter control, we collected the middle portion of spontaneous micturition (preferably in the morning), after washing the genitals with water and soap, rinsing with sterile water and drying with sterile gauze. In no case were antiseptics capable of falsifying bacterial growth employed. In patients without sphincter control, the sample was collected with a perineal adhesive bag after cleaning of the zone, with replacement every 30min. In this latter case urine culture collection was systematically spaced to once every 48h in order to avoid irritation caused by the adhesive bag in the perineal region. If the urine sample proved doubtful or showed contamination, point catheterization was carried out.

The urine samples were collected under sterile conditions and transported to the laboratory in under 1h. A closed system with a 500-ml capacity was used to control hourly diuresis.

New variables were calculated for data analysis, based on differences in dates referred to patient age, days of hospital admission, and days of bladder catheterization.

The descriptive analysis was based on calculation of the mean, median and standard deviation (SD) and interval for quantitative variables, while qualitative variables were reported as frequencies and percentages.

For each categorical variable of interest we explored associations with urinary infection based on the comparison of proportions with the Mantel–Haenszel chi-squared test or the Fisher test in the case of variables including less than five individuals.

The Kolmogorov–Smirnov test was used to assess normal distribution of the study variables.

Comparisons between means of quantitative variables in the study of children with or without infection were based on the Student's t-test or the Mann–Whitney U-test in the case of variables with a non-normal distribution.

Nonparametric tests were used for the study of factors associated to bacterial or fungal nUTIs.

The level of significance was established as 95% (p<0.05). The SPSS version 15.0 statistical package was used throughout.

During the study period a total of 178 patients were admitted to the PICU, with a mean stay of over 48h. Seventy-four patients were excluded from the study due to some of the following reasons: age under 15 days (17 patients) or over 14 years (24 patients), presence of UTI upon admission (patients 22), or other reasons (patients 11). The final sample consisted of 104 patients. The general data are reported in Table 1. The children with nUTI were younger than the children without infection (15.8 vs 35.9 months), though the difference was not statistically significant. El group of children with nUTI showed a significantly higher percentage of personal antecedents of relevance than the group without nUTI. Upon admission, the children with and without nUTI showed similar median scores on the PRISM III and PIM II severity scales (3% and 1.4%, respectively), though the interval (range) was markedly greater in the children without infection. The children with nUTI presented significantly longer stays. The mortality rate among the children with and without nUTI was 0% and 8.7%, respectively.

A total of 68% of the patients carried a bladder catheter, with a mean of 7 days and a total duration of 492 days. The bladder catheter utilization rate was 0.41. The mean time from admission to the appearance of nUTI was 3.83 days. The urine cultures showed negative conversion after 7 days on average.

All the patients with nUTI had received a bladder catheter, and the catheterization period was longer than in the children without infection, though the difference was not statistically significant (6.5 vs 5 days). Likewise, the days of corticosteroid treatment and of previous hospital stay showed no significant differences. These data, compared with the patients without nUTI, are shown in Table 2. Three of the six cases of nUTI (50% vs 88.5% of the patients without nUTI) were receiving antibiotic treatment at the time of diagnosis, though only in the two documented cases of fungal infection did the treatment have to be changed. In turn, 4.9% of the patients without nUTI had undergone surgery in the previous 60 days, 1% presented neutropenia, and 1.9% suffered immune deficiency.

A total of 27 patients showed positive urine cultures (26%). Two of these cases were classified as colonization (1.9%), 19 as contamination (18.3%), and six as nosocomial urinary tract infection (5.8%).

The six cases of nUTI presented fever, with no identification of urinary symptoms (urgency, abdominal pain, pollakiuria, dysuria, or others). The colonized patients suffered no symptoms, while one of the 19 cases of contamination coincided with a fever process that did not meet the CDC criteria of nUTI.

We documented six cases of nUTI, all in catheterized children—this representing 5.8% of all the admissions for over 48h, and 8.6% of all children with bladder catheters. A total of 1188 days of hospital stay were registered (5.05/1000 patients/day), together with 492 days of bladder catheterization (incidence density 12.19/1000 days of catheterization).

Two of the urinary infections were recorded in one same girl with spina bifida, admitted on two occasions for different reasons.

Regarding the etiology of nUTI, we identified four cases attributable to Escherichia coli (multiresistant (BLEE) in once case) and two to Candida albicans.