Corresponding author: Ja-Wook Koo, MD, PhD. Department of Pediatrics, Inje University Sanggye Paik Hospital, Inje University College of Medicine, 1342 Dongil-ro, Nowon-gu, Seoul 139-707, Korea. Introduction
Urinary tract infections (UTIs) are common causes of generalized pyrexia in infants, occurring in approximately 5% of 2- to 24-month-old febrile children1,2). Fevers associated with UTIs are usually caused by pyelonephritis3,4), the early diagnosis and treatment of which are important for the prevention of complications, including renal scarring which are associated with later renal dysfunction and progressive septicemia5,6). The diagnosis of a UTI is most commonly accomplished using a urine culture, but the collection of urine samples from infants is difficult. Regardless, choosing an appropriate urine collection method for young children is important to avoid misdiagnosing a UTI. Over-diagnosis of UTIs can lead to unnecessary radiological evaluations and antibiotic treatment7), and the consequences of under-diagnosis were previously stated.
Suprapubic aspiration (SPA) is regarded as the gold standard method for collecting urine samples with minimal false-positive results2,8). However, this method is more invasive and painful than the others9). Transurethral catheterization is a less painful method that has comparable sensitivity and specificity as SPA, making it more routinely used2,6). The noninvasive and easily accessible collection and culturing of collecting bag urine (CBU) is widely used in outpatient clinics and emergency units, despite its high contamination rate2,10).
A few authors have performed paired examinations of the different urine culture methods used for diagnosing UTIs in children of various ages (9 days to 9 years)11) and in asymptomatic infants during voiding cystourethrograms7), and with prior selection (by urinalysis) for febrile infants12). In this study, we compared the results of CBU cultures and catheter urine (CATH-U) cultures, following the recovery of both sample types from the same patients. The culture results were used to estimate the ability of CBU cultures to diagnose UTIs, when CATH-U samples were considered the "gold standard" for diagnosing UTIs in infants.
Materials and methods
The study population consisted of infants examined in the pediatrics department of Inje University Sanggye Paik Hospital (Seoul, Korea), between September 2008 and August 2013. We retrospectively analyzed patient records to determine if the patients had a potential risk of UTI, based on the following inclusion criteria: (1) a tympanic fever >38℃ of unknown origin, (2) patient age of 2-24 months, (3) non-toilet-trained patients, and (4) all female infants and uncircumcised male infants.
Demographic and clinical profiles, laboratory findings, urinalyses, and urine culture results were reviewed for each patient. Only infants with collecting bag and CATH-U cultures, obtained on the same day, were included. Patients already receiving antibiotics or having incomplete data for evaluation were excluded from the study, as were patients with contaminated CATH-U or CBU cultures.
Urine specimens were collected using an aseptic technique, according to our departmental guidelines. After cleansing the perineum with antibacterial cotton (povidone iodine), a urine collection bag was attached to the patient's perineum by trained pediatric nurses. The urine bag was left in place for up to 1 hour, without changing the bag, unless there was evidence of leakage, stool contamination, or the bag separated from the skin. The perineum was also cleaned with povidone iodine before catheterization, which involved inserting an uncontaminated, lubricated, 5-Fr feeding tube into the urethra. Whether the urine was collected via a urine bag or catheterization, the first few drops were routinely discarded because of potential bacterial contamination from the distal urethra2,13). Within 30 minutes of collection, the urine specimens were sent to the laboratory for prompt urinalysis and culture. This study received ethical approval from the institution`s research board.
1. Definitions
The following definitions were based on published descriptions 2,5,14,15,16,17,18,19). Dipstick tests were defined as positive if either nitrite or leukocyte esterase (LE) was positive. Microscopy was positive if either the white blood cell (WBC) counts were ≥10/mm3 or bacteria were present. Positive urinalysis was defined when the dipstick and/or microscopy evaluations were positive. A CATH-U culture was considered indicative of a UTI if the culture grew only a single species of bacteria at a density of ≥104 colony-forming units (CFU)/mL. CBU cultures were stratified, based on the presence of a single uropathogen species at densities of ≥103, ≥104, and ≥105 CFU/mL, and were compared to the positive/negative CATH-U results. The definition of CBU contaminants was similar to that of CATH-U, except for the bacterial colony count. Contaminants of CATH-U results were defined as: (1) bacteria present at densities of <104 CFU/mL, (2) the presence of mixed pathogens, (3) recovery of organisms such as Lactobacillus spp., coagulase-negative staphylococci (CNS), and Corynebacterium spp., or (4) different pathogens were isolated from the two urine samples from the same patient. Urine cultures were considered to be negative if bacteria were not recovered.
The patients were categorized into three groups according to the number of bacteria recovered from their urine cultures. The UTI group was subdivided into patients with high CBU colony counts (CBU ≥105 CFU/mL and CATH-U ≥104 CFU/mL, group A) and those with low CBU colony counts (CBU <105 CFU/mL and CATH-U ≥104 CFU/mL, group B). If bacteria were not cultured from the CATH-U samples, the patients were classified into the non-UTI group (group C).
2. Statistical analysis
The data were analyzed using IBM SPSS Statistics ver. 21.0 (IBM, Armonk, NY, USA). The chi-square test or Fisher exact test for small numbers was used to compare the percentages of infants with positive CBU and CATH-U cultures. One-way analysis of variance was used to compare the differences between the groups (A vs. B vs. C); a P value <0.05 was considered significant. We calculated the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for the prediction of UTIs using the CBU culture method; the results were based on a UTI defined by the CATH-U culture results; likelihood ratios (LRs) were also calculated with a 95% confidence interval (CI).
Results
A total of 246 infants were potential participants in this study. We excluded 36 patients; 15 infants had contaminated CATH-U cultures (CFU <104/mL, 1 patient; growth of CNS, 2 patients; growth of different microorganisms in the 2 cultures, 12 patients), 13 received antibacterial agents before urine collection, 6 were circumcised male infants, and 2 patients had CBU cultures positive for CNS. Of the remaining 210 patients, there were 86 female infants and 124 uncircumcised male infants. The median age of the included children was 6.0 months (range, 1-23 months).
The CBU and CATH-U culture results are shown in Table 1. The UTI group (those with CATH-U cultures ≥104 CFU/mL) comprised 53% (111/210) of the infants. Of these, the high CBU colony count UTI group accounted for 84 children (76%) and the low CBU colony count UTI group accounted for 25 (21.6%). The non-UTI group, with negative CATH-U culture results, comprised the remaining 99 infants (47%). The CBU results for patients diagnosed as having a UTI, based on the CATH-U culture results, showed varying densities of bacteria. If a positive CBU culture was indicated by a bacterial density ≥105 CFU/mL, the false-positive and false-negative rates were 18% and 24%, respectively. However, if the cutoff value for the CBU cultures was changed to ≥103 CFU/mL, the false-positive and false-negatives rates were 38% and 3%, respectively. False-positive UTI results were associated with CBU CFU counts of 103-104 CFU/mL in 64% (7/11) of the cases and with CFU counts of 104-105 CFU/mL in 44% (16/36) of the cases.
Table 2 shows the sensitivity and specificity of the different CBU cut-off values among female infants and uncircumcised male infants. For male infants, a cutoff value of ≥105 CFU/mL resulted in a missed diagnosis in 24% of those with UTIs and falsely diagnosed a UTI infection in 18% of those without a UTI. The PPVs for the uncircumcised male infants and female infants were 91% and 69%, respectively. Using a CBU cutoff value of ≥103 CFU/mL, the sensitivity approached 97%, but misdiagnosed a UTI in 41% of patients without a UTI.
Table 3 shows that the probability of UTI increased when the CBU bacterial count was ≥105/mL for all infants, uncircumcised male infants, and female infants, with LRs of 4.16, 4.11, and 4.11, respectively. Overall, a CBU cutoff value of 104-105 CFU/mL had a borderline probability (LR 1.11) of correctly diagnosing a UTI; a cutoff value of 103-104 CFU/mL had a low probability (LR, 0.51) of correctly diagnosing a UTI. Although UTIs could not be excluded in male infants when the CBU bacterial density was 103-104 CFU/mL (LR, 1.06), the probability of a UTI was lower when the CBU bacterial density was 104-105 CFU/mL (LR, 0.93). For female infants, a bacterial count of 104-105 CFU/mL showed a borderline UTI probability (LR, 1.40), but a UTI was unlikely for female infants, if the bacterial density was 103-104 CFU/mL.
A comparison of clinical and laboratory findings between the UTI and non-UTI groups is shown in Table 4. Group A UTIs were more common among younger (P<0.01) patients and uncircumcised male infants (P<0.001) than were non-UTIs. However, the fever duration among the 3 groups was not different. The initial WBC counts and C-reactive protein (CRP) levels were higher in the UTI groups than in the non-UTI group. Similarly the UTI groups had higher rates of positive LE, nitrite levels, pyuria, and bacteriuria than did the non-UTI group (all, P<0.001). The clinical and laboratory findings of group B were similar to those in group A.
The imaging study results of UTI groups are described in Table 5. There were no statistically significant differences between groups A and B relative to the number of abnormal ultrasonography, technetium99m dimercaptosuccinic acid scintigraphy (DMSA), or voiding cystourethrography findings.
In Table 6, the LRs for predicting a UTI with either a positive dipstick test or a positive urinalysis were 4.24 and 3.11, respectively. The probability of the determination of an elevated nitrite level was very high (LR, 35.68), but their sensitivity and NPV were low (36% and 58%, respectively.)
Discussion
In a febrile infant with suspected UTI, prompt antibiotic administration is important for the eradication of the infection, after obtaining a confirmatory urine culture. According to the American Academy of Pediatrics guideline, urine cultures by CATH-U or SPA are necessary to accurately diagnose UTIs in infants2). Although SPA is regarded as the "gold standard" for collecting urine specimens, its success rate in correctly diagnosing a UTI has been reported to range from 23% to 90%. Because of the variable success rates, ultrasonographic guidance and greater technical experience are required for the conduct of this sample collection technique2,8,20,21,22). Moreover, obtaining parental consent is more difficult because SPA is a more invasive and painful urine collection method than CATH-U9). CATH-U has a sensitivity of 95% and a specificity of 99% for detecting UTIs, compared with urine collection via SPA2,20,23); the extent of contamination is less than for CBU7). Hence, CATH-U is considered an alternative method to SPA. In this study, we used CATH-U as the "gold standard" for diagnosing UTI in the infants.
This study was performed retrospectively and in a non-selective fashion for collecting and evaluating the urine culture samples. McGillivray et al.16) suggested that "selective catheterization" is a more reasonable strategy than catheterization of all infants for avoiding unnecessary, invasive examinations of febrile infants. However, a negative urinalysis result does not guarantee the absence of a UTI in a febrile infant2). The collection of urine samples for urinalysis is still a time-consuming procedure for non-toilet-trained children and their parents. Moreover, obtaining the urinalysis results and the subsequent decision to require CATH-U collection indicates that physicians, patients, and parents have to wait for a longer period without treatment. A prior urinalysis could also lead to a selection bias in the evaluation. Therefore, a policy recommending catheterization of all febrile infants, suspected of having a UTI, may be more convenient for the patients and their parents. In this study, the infants had a very high prevalence of UTIs (52.5%) compared with previous studies1,16). The main reason for this high prevalence may be related to a selection bias in the sample population (all uncircumcised male infants) or the small sample sizes. Other factors, possibly contributing to the high prevalence of UTIs, include the physicians being aware of which infants were at high risk of having a UTI, the catheterizations being performed by the treating physician, and a CATH-U-based definition of UTI.
Our study indicated that CBU colony counts do not always match those of CATH-U samples. A CBU cutoff value ≥105 CFU/mL would miss approximately one-fourth of the UTIs. This was suggested by the fact that 24 patients had low CBU bacterial counts (<105 CFU/mL), despite clinical and laboratory findings comparable to those for group A. Low-colony-count UTIs predominate in infants and young children24), and there is an increasing rate of gram-negative bacteria, other than Escherichia coli, involved in these infections25). However, our study showed that E. coli was the principal microorganism in this group (88%), as it was in the patients with high-colony-count UTI. Factors causing low bacterial colony counts include urine concentration defects, urine samples obtained following hydration, and reduced bladder incubation times5,24,25). The collection of CBU after CATH-U, or vice versa, may also affect the results of bacterial colony counts because of the reduced urine dwell time16). However, the sequence of each urine sample collection was not determined, nor was the interval between each procedure estimated in this study. Additionally, the specific gravities of the urine samples from the individuals in the three groups A, B, and C were (1.012, 1.010, and 1.013) respectively, were similar (P>0.05). Thus, the involvement of the previously enumerated factors in causing low bacterial colony counts in our patients is uncertain. Febrile infants with low bacterial colony count UTIs should not be overlooked, as the prevalence of pyelonephritis and vesicoureteral reflux (VUR) is similar to that in patients with high-colony-count UTIs24).
Three infants with CATH-U-positive and CBU-negative culture results were included in the group having true UTI. Our review of their medical records showed that they experienced leukocytosis, elevated CRP levels, and pyuria. The cleansing materials, containing antimicrobials, used prior to sample collection may have impacted the urine culture results, with false-negative cultures ensuing12,26). A previous report mentioned that high number of false-negative CATH-U cultures resulted from povidone-iodine cleansing27). Therefore, false-negative CBU cultures could be caused by repeated application of disinfectants such as povidone iodine, before collecting urine samples. However, the influence of the disinfectant was uncertain, since this study was not designed to compare the effects of cleansing materials. None of the 3 patients demonstrated abnormal DMSA findings, but 1 patient showed VUR. In febrile children with an initial UTI, only 57% showed abnormal DMSA results28), and the absence of DMSA abnormalities could not guarantee the absence of high grade VUR29). Normal DMSA findings cannot exclude the possibility of UTI or the presence of VUR; therefore, clinicobiological features suggestive of UTI require administration of appropriate antibiotics.
In this study, conflicting results were found when comparing the colony counts of CBU specimens with those collected by CATH-U. Since the CBU results from patients with CATH-U bacterial counts ≥104 CFU/mL demonstrated a wide range of colony counts, from no growth to ≥105 CFU/mL, the interpretation of the CBU results was difficult. A CBU colony count ≥105/mL had an intermediate probability of correctly diagnosing UTl, whereas counts of 104-105 CFU/mL and 103-104 CFU/mL showed borderline and low UTI probabilities, respectively. The gender difference in the UTI probability associated with counts of 104-105 CFU/mL and 103-104 CFU/mL may be related to the small number of patients with urine samples having each bacterial density; more patients in these groups would allow better determination of the reasons for these results. Therefore, the establishment of a valid cutoff level at which UTI could be excluded, based on the CBU culture results, was somewhat vague and confusing.
Urinalysis cannot replace urine culture for diagnosing UTI, but it may predict urine culture and may discriminate true UTI from bacteriuria resulting from contamination or colonization. This would facilitate the initiation of prompt treatment2,18). Our results suggest that urinalyses (LE or nitrite, pyuria or bacteriuria) have high probabilities of predicting true UTIs in febrile infants. The sensitivity and PPV of a positive dipstick test (83% and 82%, respectively) and of a positive urinalysis (90% and 78%, respectively) were comparable to those associated with CBU bacterial densities ≥105/mL (76% and 82%, respectively). However, 10 patients (9%, 10/111) in the UTI group showed negative dipstick and urinalysis results. The prevalence of VUR in infants with positive urinalyses and those with negative urinalyses were 14% (13/90) and 10% (1/10), respectively (P=0.76). Therefore, a negative dipstick and/or microscopic urine examination cannot exclude the risk of UTI and the presence of VUR.
This study had a low CBU culture contamination rate compared with that in other studies (28% vs. 88%)2). Although the reasons were uncertain, numerous UTIs with low bacterial counts were observed, raising questions about the validity of CBU cultures.
There is a risk of contamination when collecting CATH-U samples as the catheter passes through an area of urethral contamination; this has been shown to be more likely in uncircumcised male infants than in female infants7). In this study, the contamination rate of the CATH-U cultures was estimated to be approximately 12% (15 patients were excluded, including 10 male infants), despite the absence of mixed bacterial populations in the CATH-U cultures. Moreover, the CATH-U sample collection is more difficult in male infants with moderate or severe phimosis and in female infants with tight labial adhesion2). A meticulous approach is required to avoid contamination and the resulting possibility of unnecessary treatment19).
Our study has several limitations. Data analyses were performed retrospectively using patient records, and involved a small sample size. Selection bias also existed, including the exclusion criteria for contaminated samples and the exclusion of circumcised males.
In conclusion, positive urinalyses help health care personnel to determine the necessity of further investigations because they have greater probabilities of predicting a UTI than does CBU culture, alone. The study also confirmed that physicians should not ignore UTIs associated with low CBU colony counts, especially if there are also indicative laboratory findings, such as leukocytosis or elevated CRP levels. Nevertheless, the validity of obtaining urine samples from sterile bags remains questionable. Inconclusive culture results from CBU samples should be confirmed using either retests or more reliable diagnostic methods, such as CATH-U or SPA5,12,16,25).
PDF Links
PubReader
XML Download
PubMed
Download Citation
