Pediatric urinary tract infections (UTI) account for 0.7% of physician office visits and 5–14% of emergency department visits by children annually.1 Accurate diagnosis of UTI has important clinical implications; most febrile infants with UTI show evidence of renal parenchymal involvement (pyelonephritis).2 Nevertheless, the presenting signs and symptoms of UTI in childhood are often nonspecific and, among infants, definitive testing for UTI involves bladder catheterization. Accordingly, clinicians caring for young children are frequently faced with the decision of whether or not to obtain a urine sample for urinalysis and culture.
Knowledge of the prevalence of UTI among different subgroups of children can assist clinicians in selecting children who would benefit from further diagnostic testing. Using prevalence rates as an estimate of the prior probability of disease is the first step in evidence-based practice. In children with a very low pretest probability of disease, routine diagnostic testing is not necessary. In fact, in such children, an indiscriminate approach to diagnostic testing might lead to more harm than benefit. In contrast, in children with high pretest probability of disease, routine diagnostic testing would be appropriate. In a survey of 300 academic and community pediatricians regarding diagnostic testing in infants with unexplained fever, baseline risk was important in determining diagnostic decisions.3 Specifically, only 10% of clinicians believed that a urine culture was indicated if the probability of UTI was <1%, whereas 80–90% would obtain a culture if the probability of disease was 3–5%, and all would do so if the probability exceeded 5%. Whether a certain child has a 2% or a 10% baseline probability of UTI makes a difference to the practicing clinician.
Prevalence was defined as the proportion of children with the target disorder among patients undergoing diagnostic testing.4 This type of point prevalence, also known as pretest probability, provides clinicians with an estimate of the baseline risk of disease.
There are currently no pooled data available stratifying prevalence based on age, gender, race, or circumcision status, all of which can affect UTI risk. To address this we conducted a meta-analysis with the aim of providing clinicians with quantitative estimates of UTI prevalence for each subgroup.
As part of a larger meta-analysis on signs and symptoms of pediatric UTI, a large database was compiled. Two investigators searched MEDLINE and EMBASE for articles from January 1966 through October 2005. Search terms included the following: UTI, cystitis, pyelonephritis, prevalence, and incidence. This computerized search was supplemented with a manual review of references cited in the Technical Report on UTI from the American Academy of Pediatrics.5 From this database, we reviewed full-text versions of articles that might contain data regarding prevalence of UTI. A second MEDLINE search was conducted 6 months later to ensure that all relevant articles were obtained. For the second search, the following terms were used: incidence, prevalence, epidemiology, UTI, cystitis, and pyelonephritis. We reviewed 652 titles. No additional articles were identified in the second search. We did not specifically search for unpublished studies. Articles meeting inclusion criteria were then independently reviewed, rated, and had data abstracted by 2 of the authors (NS, NEM). Both authors had previous experience with data abstraction and quality rating in meta-analysis.
Explicit a priori inclusion and exclusion criteria were applied. We included articles in our review if they: (1) presented data on the prevalence of UTI in children 0–19 years of age presenting with symptoms of UTI (including fever alone), (2) used urine cultures as the gold standard, and (3) defined a positive urine culture as ≥104 for catheterized specimen, >103 for suprapubic specimen, and ≥105 for clean catch or bag specimens. Articles were excluded if they: (1) were in languages other than English, (2) evaluated only a high risk subgroup (eg, malnourished, premature, genitourinary or neurologic abnormalities, nosocomial infections, sexually abused), (3) were performed in developing countries, (4) evaluated only a low risk subgroup (eg, asymptomatic well appearing children), (5) evaluated only children with other symptomatic illnesses (febrile seizures, infectious diarrhea, bronchiolitis), (6) contained fewer than 10 subjects, and (7) used bags to collect urine specimen in more than 25% of subjects with UTI without confirming results using more accurate methods such as suprapubic catheterization or bladder catheterization.
Four of the authors whose prevalence papers are included in this analysis (Zorc, Newman, Hoberman, and Shaw) provided us with additional data from their studies, which we included in our analyses.
We used a published quality rating system for prevalence articles.6 Two investigators (NS, NEM) assessed each study independently on a 5-point scale. We reviewed each article to determine whether: (1) study design was appropriate for obtaining prevalence estimates, (2) sample was representative of the general population of children presenting with a UTI, (3) UTI diagnostic criteria were acceptable (catheterized, suprapubic, or clean catch specimen with >105, >103, and >105 organisms, respectively), (4) urine culture was performed on a consecutive or random sample of subjects, and (5) the final diagnosis was known for >80% of eligible subjects. To avoid introducing bias, we included all articles meeting our inclusion criteria in the analysis, and used the quality rating to explore the effect of study quality on prevalence values.7 Disagreements were resolved by consensus of the authors.
Data were imported into STATA version 9.2 and a pooled estimate of UTI prevalence was calculated. To determine whether or not to use the fixed-or random-effects model, statistical heterogeneity between and within groups was measured using the Q statistic and assessed visually using the Galbraith plot of heterogeneity. If the Q test was not significant, the fixed effects methods were used. Otherwise pooled estimates and confidence intervals were calculated assuming a random-effects model with inverse-variance weighting using the DeSimonian and Laird method.8
Although we did not expect to see publication bias when assessing prevalence, this was assessed using the Begg rank correlation method and the Egger weighted regression method. We also looked at the cumulative effect of adding articles one at a time ordered by publication date on the pooled prevalence estimate. We also performed a similar analysis for quality ratings. To evaluate the weight of particular articles on the pooled estimate we performed influence analysis. This method recalculates the pooled prevalence estimate omitting 1 study at a time. Meta-regression was used to analyze the relationship between UTI prevalence and study quality, age, length of the study, setting (outpatient clinic versus ER), year of study, and whether the study was conducted in the United States or another country. All reported confidence intervals represent the 95% confidence intervals.
Description of Articles.
From a total of >4000 articles found through our search strategy, we retrieved 330 for full text review. Fifty-one articles contained prevalence data and of these 18 articles evaluating 22,919 children met all criteria for inclusion. The setting for all articles was either the outpatient clinic, emergency department or both. There was no evidence of publication bias. We categorized the studies into 2 groups based on the population of children enrolled.
There were 14 articles which enrolled infants (0–24 months, Table 1); in these articles, enrollment was based on the presence of fever (of at least 38.0°C). We further subdivided this group into 4 smaller subgroups: infants ≤3 months, infants 3–6 months, infants 6–12 months, and infants 12–24 months of age.
There were 4 articles which enrolled children older than 2 years of age (Table 2). In these studies, enrollment was based on the presence of signs and symptoms referable to the urinary tract. Although some infants (with fever) were included in these studies, the majority of subjects were children >2 years of age (with urinary symptoms).
Prevalence of UTI Among Febrile Infants <2 Years of Age.
Among the 14 studies of febrile infants <24 months of age, the pooled prevalence of UTI was 7.0% (CI: 5.5–8.4) (Fig. 1). The χ2 test of homogeneity was highly significant (P < 0.001). Accordingly random effects estimates were used. The Begg and Egger tests showed no evidence of publication bias. The Galbraith plot, identified the populations studied by Newman et al,9 Bachur and Harper,10 and Shaw et al11 as the most heterogeneous. However, influence analysis showed that no study, including these 3, significantly impacted the pooled prevalence estimate. Although high study quality was associated with lower prevalence of UTI (P = 0.001), only small changes in the pooled prevalence rate was observed after eliminating the 2 bag studies (6.6, CI: 5.1–8.1) or the 2 level 4 studies 6.6% (5.1–8.1%). Meta-regression showed that study year, study length, location within the United States, and study setting did not impact prevalence rates significantly.
The effect of age and gender on prevalence of febrile UTI in infants is shown in Table 1. Fourteen articles contained data regarding the prevalence of UTI among febrile infants, 9 of which presented data according to gender. Among males, prevalence rates were highest during the first 3 months of life and declined thereafter. Among females, prevalence rates were highest during the first 12 months.
Prevalence of Febrile UTI According to Circumcision Status.
Four articles contained information regarding UTI prevalence in circumcised and uncircumcised males in infants <3 months of age (Fig. 2).9,11–14 The UTI prevalence rates for circumcised and uncircumcised males were 2.4% (CI: 1.4–3.5) and 20.1% (CI: 16.8–23.4), respectively. The prevalence of UTI among circumcised males was relatively similar across the articles. Elimination of the bag studies from this analysis did not significantly alter the results: prevalence of UTI among uncircumcised (2 studies) and circumcised infants (1 study) was 20.7% (CI: 16.7–20.8) and 2.3% (CI: 1.1–5.4), respectively.
One study provided data regarding prevalence of UTI among older infants by circumcision status. In that study the prevalence rates of UTI among circumcised and uncircumcised males 6–12 months of age were 0.3% and 7.3%, respectively. None of the studies included data regarding the prevalence of UTI in males >12 months of age.
Prevalence of Febrile UTI According to Race.
Four articles provided data regarding race.9,11,14,15 The prevalence of UTI among whites (8.0%; CI: 5.1–11.0) was significantly higher than the prevalence of UTI among African Americans (4.7%; CI: 2.1–7.3). However, there was heterogeneity among the articles. Three articles found significantly higher rates of UTI among whites than African Americans.9,11,15 In the remaining study,14 which used Hispanics as a race category, whites were less likely to have a UTI than nonwhites. Because Hispanic males are less likely to be circumcised,16,17 the higher rate of UTI among nonwhites may be partially the result of the Hispanic male subgroup. Among female children, all 4 studies found higher rates of UTI among whites than African Americans.
Prevalence of UTI Among Older Children.
Four studies presented data on the prevalence of UTI among children <19 years of age with signs or symptoms referable to the urinary tract (Table 2). The χ2 test indicated that the studies were homogeneous (P = 0.36). The pooled prevalence of UTI (febrile and afebrile) was 7.8% (CI: 6.6–8.9).
This analysis found that the reported prevalence of UTI varies widely by age, gender and circumcision status. This confirms the importance of demographic and clinical characteristics when considering further diagnostic testing. The quantitative prevalence estimates presented in this article provide the clinician with a better sense of the relative importance of each of these factors. By incorporating information about the patient's age, race, gender and circumcision status, clinicians can make more informed decisions on a case-by-case basis.
We found slightly higher prevalence rates than a 1999 study of UTI prevalence conducted by the American Academy of Pediatrics (AAP) in which the pooled prevalence of UTI in 10 studies was 5%.5 We used explicit a priori inclusion and exclusion criteria specifically related to prevalence studies, which led us to exclude several articles included in the AAP report. Furthermore, we were able to provide pooled estimates of prevalence by age, race and gender.
Female infants with fever had a relatively high prevalence rate of UTI, especially during the first year of life. Our results are consistent with data from large epidemiologic UTI studies,18,19 which have shown a decreasing incidence of febrile UTI among females during the first 2 years of life. Accordingly, it would be reasonable to consider obtaining a urine specimen from febrile females younger than 1 year of age. Knowledge of the baseline probability of UTI, along with information on the unique clinical presentation, can help the clinician decide whether obtaining a urine specimen is indicated.
Among febrile males, circumcision status was important in determining risk for UTI. Uncircumcised male infants ≤3 months of age had the highest prevalence of UTI of any group, male or female, whereas circumcised males had one of the lowest rates. Among febrile male infants ≤3 months of age, 20.1% of uncircumcised males had a UTI. This finding suggests that clinicians need to carefully ascertain circumcision status in all male infants with unexplained fever. Among studies with circumcised male infants ≤3 months of age, the prevalence of UTI was 2.4%. Accordingly, approximately 42 such infants will need to undergo catheterization to detect a single UTI. Although the prevalence of UTI is relatively low in this subgroup, the risks of a misdiagnosed neonatal UTI (hematogeneous dissemination, sepsis, missed high grade vesicoureteral reflux) are high. Furthermore, in this age group more than any other, appropriate therapy depends on determination of the exact site and bacterial etiology of the infecting agent. Accordingly, catheterization of all febrile male infants <3 months of age should be considered.
Although we could not directly calculate the pooled prevalence of UTI in circumcised and uncircumcised males older than 1 year of age from the included articles, several important pieces of information are available. Among the studies compiled for this report, the overall prevalence of UTI among males decreased rapidly with age (Table 1). This is consistent with previous epidemiologic studies of UTI in which the highest rates of UTI were in the first month of life.18,20 The decreasing prevalence of UTI with age has been documented in uncircumcised18–21 and circumcised males.10,11 In these epidemiologic studies, the rates of UTI decrease considerably after 6 or 12 months of age. Only one of the studies that included older male infants reported UTI prevalence by circumcision status; among circumcised male infants 6–12 months of age, the prevalence of UTI was 0.3% among uncircumcised males and 7.3% among circumcised males.19 Accordingly, it would be reasonable to assume that the prevalence of UTI among circumcised males >12 months of age is <1%.
The available data suggest that race is associated with UTI prevalence. Although more data are needed to clarify the mechanism by which race affects baseline risk of UTI, based on the available data, white children can be considered at higher risk of developing UTI than African American children.
Among older children with signs and symptoms referable to the urinary tract, the prevalence of UTI was 7.8% (CI: 6.6–8.9). In contrast, among adult females presenting with genitourinary symptoms, approximately 50% are ultimately diagnosed with UTI.22 The discrepancy between children and adults could be secondary to biologic differences such as sexual activity or bacterial flora. Alternatively, it could be related to the better ability of adults to recognize and communicate symptoms of UTI. Whatever the reasons, the difference in pretest probability dictates a different approach to diagnosis. Specifically, the relatively low prevalence of UTI in children calls for the use of more accurate tests to minimize false positive and false negative results. Furthermore, because of the relatively low prevalence of UTI in children, diagnosis of UTI based on signs or symptoms alone is unlikely to be accurate.
Our analysis had several limitations. First, the heterogeneity (among studies of infants ≤2 years) could be considered a limitation. Although pooled estimates and the confidence intervals include adjustments for the between study variances, clinical judgment is warranted to decide whether, in fact, there are studies that were “too different” to be included. We found little difference in study design or quality between studies having the greatest and least impact on heterogeneity. Second, because most of the articles about older children did not differentiate cystitis from pyelonephritis, the estimates provided include both. Finally, we could not directly calculate a prevalence rate for verbal children according to gender as no study that included older children reported this data.
This analysis, however, has several strengths. By using rigorous methodology (comprehensive search strategy and use of a priori inclusion/exclusion criteria), our results provide a more updated picture of UTI prevalence. The pooled estimates have relatively narrow confidence intervals and are consistent with previous epidemiologic studies. Pooled prevalence values provided in this study can be used as an estimate of baseline probability in an evidence-based approach. Neither cumulative incidence nor population prevalence, often reported in epidemiologic articles, can be used in this fashion. In pediatrics, signs and symptoms of disease are often nonspecific. Consequently, more prevalence meta-analyses are needed to provide clinicians with baseline estimates of risk for common diseases.
Dr. Shaikh had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors thank Drs. Gorelick, Newman, Zorc, Lin and Hoberman for providing us with additional data from their studies.
1. Freedman AL. Urologic diseases in North America Project: trends in resource utilization for urinary tract infections in children. J Urol
2. Hoberman A, Charron M, Hickey RW, Baskin M, Kearney DH, Wald ER. Imaging studies after a first febrile urinary tract infection in young children. N Engl J Med
3. Roberts KB, Charney E, Sweren RJ, et al. Urinary tract infection in infants with unexplained fever: a collaborative study. J Pediatr
4. Haynes RB, Sackett DL, Guyatt GH, Tugwell P. Clinical Epidemiology: How to Do Clinical Practice Research.
3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2006.
5. Downs SM. Technical report: urinary tract infections in febrile infants and young children. Pediatrics
6. Richardson WS, Polashenski WA, Robbins BW. Could our pretest probabilities become evidence based? A prospective survey of hospital practice. J Gen Intern Med
7. Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring the quality of clinical trials for meta-analysis
8. DerSirmonian R, Laird N. Meta-analysis
in clinical trials. Control Clin Trials
9. Newman TB, Bernzweig JA, Takayama JI, Finch SA, Wasserman RC, Pantell RH. Urine testing and urinary tract infections in febrile infants seen in office settings: the Pediatric Research in Office Settings’ Febrile Infant Study. Arch Pediatr Adolesc Med
10. Bachur R, Harper MB. Reliability of the urinalysis for predicting urinary tract infections in young febrile children. Arch Pediatr Adolesc Med
11. Shaw KN, Gorelick M, McGowan KL, Yakscoe NM, Schwartz JS. Prevalence of urinary tract infection in febrile young children in the emergency department. Pediatrics
12. Crain EF, Gershel JC. Urinary tract infections in febrile infants younger than 8 weeks of age. Pediatrics
13. Lin DS, Huang SH, Lin CC, et al. Urinary tract infection in febrile infants younger than eight weeks of Age. Pediatrics
14. Zorc JJ, Levine DA, Platt SL, et al. Clinical and demographic factors associated with urinary tract infection in young febrile infants. Pediatrics
15. Hoberman A, Chao HP, Keller DM, Hickey R, Davis HW, Ellis D. Prevalence of urinary tract infection in febrile infants. J Pediatr
16. Risser JM, Risser WL, Eissa MA, Cromwell PF, Barratt MS, Bortot A. Self-assessment of circumcision status by adolescents. Am J Epidemiol
17. Laumann EO, Masi CM, Zuckerman EW. Circumcision in the United States. Prevalence, prophylactic effects, and sexual practice. JAMA
18. Jakobsson B, Esbjorner E, Hansson S. Minimum incidence and diagnostic rate of first urinary tract infection. Pediatrics
19. Marild S, Jodal U. Incidence rate of first-time symptomatic urinary tract infection in children under 6 years of age. Acta Paediatr
20. Winberg J, Andersen HJ, Bergstrom T, Jacobsson B, Larson H, Lincoln K. Epidemiology of symptomatic urinary tract infection in childhood. Acta Paediatr Scand Suppl
21. Uhari M, Nuutinen M. Epidemiology of symptomatic infections of the urinary tract in children. BMJ
22. Bent S, Nallamothu BK, Simel DL, Fihn SD, Saint S. Does this woman have an acute uncomplicated urinary tract infection? JAMA
23. Krober MS, Bass JW, Powell JM, Smith FR, Seto DS. Bacterial and viral pathogens causing fever in infants less than 3 months old. Am J Dis Child
24. Bonadio WA, Hennes H, Smith D, et al. Reliability of observation variables in distinguishing infectious outcome of febrile young infants. Pediatr Infect Dis J
25. Bonadio WA, Webster H, Wolfe A, Gorecki D. Correlating infectious outcome with clinical parameters of 1130 consecutive febrile infants aged zero to eight weeks. Pediatr Emerg Care
26. Bonadio WA, Smith DS, Sabnis S. The clinical characteristics and infectious outcomes of febrile infants aged 8 to 12 weeks. Clin Pediatr
27. Kadish HA, Loveridge B, Tobey J, Bolte RG, Corneli HM. Applying outpatient protocols in febrile infants 1–28 days of age: can the threshold be lowered? Clin Pediatr
28. Herr SM, Wald ER, Pitetti RD, Choi SS. Enhanced urinalysis improves identification of febrile infants ages 60 days and younger at low risk for serious bacterial illness. Pediatrics
29. Dayan PS, Bennett J, Best R, et al. Test characteristics of the urine Gram stain in infants less-than or equal to 60 days of age with fever. Pediatr Emerg Care
30. Heale WF, Weldon AP, Hewstone AS. Reflux nephropathy. Presentation of urinary infection in childhood. Med J Aust
31. Dickinson JA. Incidence and outcome of symptomatic urinary tract infection in children. Br Med J
32. Shaw KN, McGowan KL. Evaluation of a rapid screening filter test for urinary tract infection in children. Pediatr Infect Dis J
33. Struthers S, Scanlon J, Parker K, Goddard J, Hallett R. Parental reporting of smelly urine and urinary tract infection. Arch Dis Child