Prevalence of urinary tract infection (UTI) in sequential acutely unwell children presenting in primary care: Exploratory study

Abstract

Background. Due to the non-specific nature of symptoms of UTI in children and low levels of urine sampling, the prevalence of UTI amongst acutely ill children in primary care is unknown. Objectives. To undertake an exploratory study of acutely ill children consulting in primary care, determine the feasibility of obtaining urine samples, and describe presenting symptoms and signs, and the proportion with UTI. Design. Exploratory, observational study. Setting. Four general practices in South Wales. Subjects. A total of 99 sequential attendees with acute illness aged less than five years. Main outcome measure. UTI defined by >10⁵ organisms/ml on laboratory culture of urine. Results. Urine samples were obtained in 75 (76%) children. Three (4%) met microbiological criteria for UTI. GPs indicated they would not normally have obtained urine samples in any of these three children. However, all had received antibiotics for suspected alternative infections. Conclusion. Urine sample collection is feasible from the majority of acutely ill children in primary care, including infants. Some cases of UTI may be missed if children thought to have an alternative site of infection are excluded from urine sampling. A larger study is needed to more accurately determine the prevalence of UTI in children consulting with acute illness in primary care, and to explore which symptoms and signs might help clinicians effectively target urine sampling.

Key Words::

• Children
• diagnosis
• prevalence
• primary care
• urinary tract infection

Diagnosis of UTI in young children is difficult as symptoms are non-specific and urine samples often hard to obtain. The presentation of 99 unwell children in general practice is described.

• Urine samples were obtained in over 75% of cases; 4% of these met microbiological criteria for a UTI.

• Although more robust epidemiological studies are required, GPs should maintain a high index of suspicion of the possibility of UTI and have a low threshold for urine sampling.

Symptoms of UTI in young children tend to be non-specific [1]. If UTI is not suspected or there is difficulty obtaining a sample, the diagnosis will be missed. This is particularly the case in primary care where children frequently consult with non-specific symptoms and where appropriate paediatric equipment is not always available [2].

Urine is infrequently sampled from children in primary care [3]. In 2007, the UK National Institute for Clinical Excellence (NICE) issued a guideline concerning UTI in children [4]. This promotes increased urine sampling, particularly from primary care. It advises against urine sampling if an alternative site of infection is identified.

A recent review found that prevalences of UTI in children ranged from 2% to 20% [5]. Most of the included studies were from the US and included only febrile infants (temp >38°C; age <12 months). No studies have systematically sampled from consecutive acutely ill children in primary care, without selecting on the basis of presenting features.

We do not know how often UTI is the cause of illness in young children presenting in general practice-based primary care and which children should be targeted for urine sampling. We conducted an exploratory study of the prevalence of UTI describing presenting symptoms and the proportion found to have a UTI when urine samples were systematically requested from all eligible children.

Material and methods

Four general practices in South Wales (UK) recruited consecutive children aged less than five years presenting with an acute illness (<28 days duration) between February and April 2007. Presenting symptoms and signs and working diagnosis were recorded. A urine sample was requested from all children, using clean-catch or nappy-pad methods [4]. Clinicians were asked whether they would have requested a urine sample under conditions of usual care.

It was not always clear from the recorded diagnosis whether GPs had identified a specific site of infection or cause for the child's symptoms. In some cases, the diagnosis was vague (e.g. “viral illness”) or only seemed to relate to one of several presenting symptoms (e.g. “breast bud”). In view of the recent NICE guidelines we wanted to determine how many children appeared to have a specific site of infection to explain their symptoms. Two clinicians (KO and NS) independently assessed presenting symptoms, signs, and the GP's working diagnosis to determine whether a specific site of infection was a likely explanation for each child's illness. A Kappa statistic was calculated to assess inter-rater agreement. In the five cases where there was disagreement between the two raters, a consensus opinion was reached through discussion.

Data were entered into SPSS (version 12) for analysis. Results are presented descriptively and the confidence interval for prevalence calculated using Wilson's method [6]. Fisher's exact test was used to examine associations between symptoms and microbiological diagnosis of UTI.

Results

Participants

A total of 116 children aged under five were invited to participate. Seven parents did not provide consent, three were excluded, the treating clinician felt six were unsuitable and the documentation was lost for one, leaving 99 for analysis.

Forty-eight children (48%) were male and 51 were female. The median age was 20 months (IQR: 8–32 months).

The most common presenting symptom was nasal congestion (present in 77%), followed by cough (75%), and fever (69%). Non-specific constitutional symptoms, e.g. irritability, clinginess, crying more than usual, tiredness, poor sleep, and poor appetite, were present in over 50% of children. Specific urinary symptoms were uncommon: smelly urine (15%), increased frequency (9%), cloudy/dark urine (8%), pain/crying on passing urine (4%), and day or bed wetting when previously dry (1%).

Microbiological analyses

Urine samples were obtained in 75 (76%). Urine samples leaked in transit in three cases (4%) and were lost in transit in one (1%). There was a full laboratory result for 71 (72%).

A microbiological diagnosis of UTI was made in three (4%; 95% CI for prevalence: 1–12%). There was no significant association between any individual symptom and microbiological diagnosis of UTI. All of those with UTI had nasal congestion, poor feeding, and poor sleeping. Other constitutional symptoms – fever, irritability, clinginess, tiredness, and crying more than usual – were each present in two out of the three with UTI.

Working diagnoses

The pre-consensus discussion Kappa statistic regarding whether a specific infection site was considered evident to explain a child's illness was 0.79, indicating substantial agreement. Following discussion and consensus on those where there was some disagreement, a specific site of infection was felt to explain the presenting symptoms, examination findings, and working diagnosis in 87 (88%) children. The most frequent sites of infection were upper respiratory tract infection (49/99, 49%), lower respiratory tract infection (12/99, 12%), and otitis media (9/99, 9%). In the 12 cases where there was felt to be no clear site of infection, the GP working diagnoses included UTI (4), viral illness (2), PUO (2), thrush (1), cold sore (1), breast buds (1), and hernia (1).

All three children subsequently diagnosed with UTI had specific sites of infection identified and had been empirically prescribed amoxicillin, two for a working diagnosis of otitis media and one for tonsillitis.

Reported usual practice

GPs indicated that they would have normally requested urine samples in 12 (12%) of the presenting children. GPs indicated that they would not normally have requested urine samples on any of the children subsequently found to have a UTI.

Discussion

Principal findings

In this exploratory study, we have described the symptoms of 99 children aged less than five years consulting in primary care with an acute illness. All were asked to provide a urine sample. We determined the proportion of those who provided urine samples (76%) and found that three (3/71, 4%) met microbiological criteria for a diagnosis of UTI.

Strengths and weaknesses of the study

To our knowledge, this is the first study in primary care that has systematically sampled urine from sequentially presenting, acutely unwell children. However, the sample was small and not adequately powered to estimate the prevalence of UTI (95% CI 1–11%) or the predictive value of symptoms and signs. Normal urine sampling practice as indicated by GPs was much higher than suggested by a previous study [3]. GPs may have requested more samples in this previous study than were provided. In our study, GPs may have over-estimated their normal urine sampling practice due to the raised awareness of UTI created by the study. Urine sampling behaviour may have changed since this study as a consequence of the publication of the NICE guideline.

Context of other literature

According to the NICE guideline, urine sampling would not have been recommended at first presentation for any of those children in this study who were found to have a UTI [4]. Hospital-based studies, where urine has been systematically sampled, also found that UTI cannot be ruled out by the apparent presence of an alternative site of infection [7,8]. Hoberman et al. (1993) found that 3.5% of infants identified as having an apparent alternative site of infection had a UTI [7]. Most clinicians would consider that a urine sample is indicated for a pre-test probability of UTI this high [9]. However, false positives do occur, and some recommend a second urine sample or a higher microbiological threshold [10].

Further research

A large-scale epidemiological study of UTI prevalence, presentation, and outcome in acutely unwell children presenting in general practice is needed to better inform guidelines for diagnosis and management. The effect of the NICE guidelines on urine sampling and UTI diagnosis in primary care should also be evaluated.

Conclusion

We have shown that it is feasible to obtain urine samples from young children in primary care. However, any strategy to promote increased urine sampling should be based on robust evidence and give clear, applicable guidance for clinicians.

Acknowledgements

The authors would like to thank the parents of the children, the general practitioners, nurses, and practice staff for participating in the study.

Statement of financial support for the reported project

Cardiff University funded the project. The WORD (Welsh Office of Research and Development) funded South East Wales Trials Unit (SEWTU) and the Wales School of Primary care Research supported the study.

Ethics approval

Ethical approval for the study was obtained from the South East Wales Local Research Ethics committee (ref no.06/WSE03/117). Informed, written consent was obtained from all participants.

Competing interests

None.