Short‐ versus standard‐course antimicrobial therapy for children with urinary tract infection: A meta‐analysis

Abstract

Aim

To determine if short‐ (2–5 days) course antimicrobials are as effective as standard‐ (6–14 days) course antimicrobials in the treatment of symptomatic UTI in children.

Methods

MEDLINE and EMBASE were searched from their origin to January 2024. We only considered randomised controlled trials in children <18 years of age. The main outcomes of interest were UTI or bacteriuria at the end of therapy.

Results

Nine studies were included. Compared to children treated with a standard course of antimicrobials, those treated with shortened courses of antimicrobials did not have significantly different risks of UTI at the end of therapy (risk difference 2.2%, CI: 0.0–4.3). Risk of bacteriuria at end of therapy (RD = 8.7%, CI: 5.3–12.2) was slightly higher in children treated with shorter courses. In children with fever at baseline (two studies), there was no significant difference in risk between short and standard duration treatment (RD = 0.4%, CI: −2.8 to 3.6).

Conclusion

In children without fever at the time of presentation, treatment with shorter courses of antimicrobials appears reasonable. More studies of febrile children are needed before shorter courses could be recommended for febrile children.

Abbreviation

UTI

Urinary Tract Infection

Key Notes.

• Two new studies investigating the efficacy of short‐course antimicrobials to treat urinary tract infections in children have been published since the last meta‐analysis.

• In children without fever at the time of presentation, treatment with shorter courses of antimicrobials appears reasonable.

• More studies of febrile children are needed before shorter courses could be recommended for febrile children.

1. INTRODUCTION

Urinary tract infections (UTI) are among the most common serious bacterial infections in children. Approximately, 7% of infants presenting with fever and 7.8% of older children presenting with urinary symptoms are found to have UTIs. ¹ The last meta‐analysis, conducted in 2012, found standard‐course antimicrobial therapy was associated with a lower incidence of bacteriuria at the end of therapy compared to short‐course therapy. ² However, two new studies investigating the efficacy of short‐course antimicrobials to treat urinary tract infections in children have been published since. Our objective was to systematically review the data to date and to compare the benefits and harms of using shortened courses of antimicrobials for the treatment of UTI in children.

2. METHODS

2.1. Types of studies included

We considered randomised controlled trials that evaluated the efficacy of short‐ (2–5 days) versus standard‐ (6–14 days) course oral antimicrobials for the treatment of acute urinary tract infection in children. We included studies of children from birth to 18 years of age with UTI defined by bacteriuria in a child with at least one symptom of a UTI. Symptoms included fever, dysuria, urinary urgency, frequency, or incontinence, and abdominal pain or flank pain. We excluded studies that gave a single dose of antimicrobials as the intervention. Bacteriuria was defined as a urine culture with over 10 000 CFU/mL of growth. We excluded studies in which the short course consisted of 1 day and/or 1 dose of antimicrobial treatment because previous meta‐analyses had convincingly demonstrated that this duration was not effective for treatment of UTI in children. The protocol for this review was submitted to PROSPERO (https://www.crd.york.ac.uk/PROSPERO/) on December 05, 2022 (CRD42022380787).

2.2. Search strategy

We searched MEDLINE and EMBASE on 2 January 2024. Search strategies are presented in the electronic supplement (Table S1). We also reviewed the reference lists of all included articles and relevant systematic reviews to identify additional studies in grey literature. ² , ³ We contacted authors when relevant data was not included in the published source. Two authors (GM and AG) independently applied the selection criteria to all citations (titles and abstracts). The full text of all articles identified by either author was retrieved and reviewed. We limited the review to articles written in English, Spanish, French or Italian.

2.3. Data extraction and management

For each study meeting the inclusion criteria, we extracted the following covariates: age range of children enrolled, gender of children enrolled, diagnostic criteria for UTI, urine collection method, antimicrobial(s) used, dosage of the antimicrobial(s) used, number of children in each arm, and whether children with fever were eligible to enrol in the study. We examined the following outcomes: UTI at the end of therapy (primary outcome), bacteriuria at the end of therapy, UTI 25–60 days after end of therapy, bacteriuria 25–60 days after end of therapy and adverse events (i.e. diarrhoea, increased antimicrobial resistance). Differences were resolved by discussion.

2.4. Statistical analysis and data synthesis

Because of the large number of small studies, we pooled data using fixed‐effects models. We focused on risk differences (as opposed to risk ratios) because it allowed for easy translation of results to clinical practice. For example, a risk difference of 10% in rate of UTI between two treatments translates into a number needed to treat of 10 (i.e. 1 divided by 0.1), which means that 10 extra children will need to be treated with the superior treatment to prevent one child from experiencing a UTI. Smaller differences in risk would result in a higher number needed to treat. We used the I ² statistic to assess the degree of heterogeneity between studies. We did not use meta‐regression to investigate reasons for heterogeneity because the number of studies were limited. Instead, we conducted a sensitivity analysis in which we excluded studies with a high risk of bias.

After completing the preplanned analysis, we realised that for the UTI outcome at the end of therapy, between study heterogeneity was high in the two studies that reported these data. ⁴ , ⁵ We postulated that this could be due to the difference in patient characteristics included in each study. Namely, the study by Zaoutis included both febrile and afebrile children whereas the study by Montini only included febrile children. Thus, we contacted the authors of the Zaoutis study to provide summary data according to whether children had fever at the time of enrolment. Once these data were provided, we conducted a stratified analysis; this analysis was not planned a priori.

3. RESULTS

The results of the search strategy are shown in Figure 1. Of the 2388 records identified, 60 were assessed for eligibility, and of these 51 were excluded. Reasons for exclusion are detailed in Figure 1. Exclusion of three studies merit further explanation because some of these studies were included in past meta‐analyses. We excluded the study by Zaki because children were randomised to two antibiotics, not to short versus standard courses of antimicrobials. ⁶ We excluded the study by Petersen because it included children with asymptomatic bacteriuria. ⁷ Finally, the study by Kornberg was excluded because of inconsistencies in the number of children reported to have received standard‐course treatment. ⁸ A total of nine studies met our inclusion criteria and their characteristics are summarised in Table 1. All but one study required children to have at least 10 000 CFU/mL of a pathogen for inclusion; the study by Tambic did not report criteria used to define a positive culture. ¹⁴ One study required pyuria for inclusion. ⁴ Of the nine included studies, four were determined to be at high risk for bias (Figure S1). ⁹ , ¹¹ , ¹⁴ With the exception of two studies, ⁴ , ⁵ randomisation occurred on the day of presentation. In these two studies, randomisation occurred after ensuring the child's fever had resolved (on Day 4 or 5).

FIGURE 1.

Study selection process.

TABLE 1.

Characteristics of placebo‐controlled randomised trials that examined duration of antimicrobial therapy for symptomatic urinary tract infection.

Author, Year	Included febrile children	Age	Day randomisation occurred	No. short therapy/standard therapy	Antimicrobial(s) used	Outcome(s) reported at end of therapy a	Outcome(s) reported 25–60 days after end of therapy b
Wientzen 1979 9 , c , d	No	3 months–16 years	Day 1	24/28 e	Amoxicillin	Bacteriuria f	None
Helin 1981 10 , c , d	No	3 months–16 years	Day 1	23/20	Trimethoprim/sulphadiazine	Bacteriuria	Bacteriuria
Khan 1981 11	Yes	6 months–15 years	Day 1	13/14 g	Ampicillin, sulfisoxazole, cephalexin	Bacteriuria	Bacteriuria
Lohr 1981 12	No	2 years−18 years	Day 1	26/23	Nitrofurantoin	Bacteriuria f	None
Gaudreault, 1992 13	No	2 years−18 years	Day 1	20/20	Trimethoprim/sulphadiazine	Bacteriuria	Bacteriuria h
Tambic 1992 14	Yes	4 months–14 yr	Day 1	58/59	Co‐trimoxazol	Bacteriuria f	Bacteriuria
Johnson 1993 15	No	1 year−13 years	Day 1	20/17	Amoxicillin‐clavulanate	Bacteriuria	Bacteriuria
Zaoutis 2023 4 , d , i	Yes	2 months −10 years	Day 5 with clinical improvement	336/328	Amoxicillin‐clavulanate, cefixime, cephalexin, trimethoprim/sulfamethoxazole	UTI, bacteriuria	UTI
Montini 2024 5	Yes	3 months–5 years	Day 4 with clinical improvement	72/70	Amoxicillin‐clavulanate	UTI	UTI

^a End‐of‐therapy outcomes (UTI or bacteriuria) were assessed 6 days after end of therapy (± 4 days) in all studies except for the studies by Tambic and Gaudreault in which this was assessed immediately after end of therapy.

^b Late outcomes were assessed 25–60 days after end of therapy except for Johnson which reported outcomes at 30–47 days after end of therapy and Montini which assessed outcomes approximately 30 days after end of therapy. Stool antimicrobial resistance in the Zaoutis study was assessed between days 24 and 30.

^c Diagnostic and follow‐up criteria included UTI/bacteriuria measured in urine collected using strap on bag method.

^d Diagnostic criteria included UTI/bacteriuria measured in urine collected by methods other than urinary catheter.

^e Only includes CRP negative children because CRP positive children were not randomised.

^f Only includes recurrence of infection by initial infecting organism.

^g Asymptomatic group not included.

^h Assumed bacteriuria occurred at study visit (at least 28 days after end of therapy).

ⁱ Follow‐up criteria included bacteriuria measured in urine collected using strap on bag method.

Pooled results for all analyses performed are summarised in Table 2; Figures provide details for each analysis. Early outcomes reported were UTI and bacteriuria within 2–10 days of completing therapy.

TABLE 2.

Summary of findings.

Question: Short‐ versus Standard‐course antimicrobials for children with acute urinary tract infection. Population: Children with acute symptomatic urinary tract infection. Setting: Outpatient. Intervention: Short‐ (2–5 days) course antimicrobial. Comparison: Standard‐ (6–14 days) course antimicrobial
Outcomes evaluated	Risk in short course	Risk in standard course	Risk difference (95% confidence interval)	Number of children (studies)	Certainty of the evidence (GRADE)
UTI a at end of therapy	3.7% (15/408)	1.5% (6/398)	2.2% (0.0, 4.3)	806 (2)	★★★☆ b
In children with fever at enrolment	3.0% (6/200)	2.6% (5/192)	0.4% (−2.8, 3.6)	392 (2)	★★☆☆ b , c
In children without fever at enrolment	4.3% (9/208)	0.5% (1/206)	3.8% (0.9, 6.8)	414 (1)	★★★☆ b
Bacteriuria d at end of therapy	13.3% (69/520)	4.5% (23/509)	8.7% (5.3, 12.2)	1029 (8)	★★★☆ e
UTI 25–60 days after therapy	3.6% (14/394)	4.5% (18/396)	−1.0% (−3.7, 1.7)	790 (2)	★★★☆ b
Bacteriuria 25–60 days after therapy	20.9% (28/134)	16.2% (21/130)	5.2% (−3.9, 14.4)	264 (5)	★★☆☆ b , e
Diarrhoea during therapy	10.1% (34/336)	13.1% (43/328)	−3.0% (−7.9, 1.9)	664 (1)	★★★☆ b
Antimicrobial resistance in organisms in stool 1 month after therapy	9.0% (28/310)	7.7% (23/298)	1.3% (−3.1, 5.7)	608 (1)	★★★☆ b

Note: Two stars defines Low certainty and every instance of three stars defines Moderate certainty.

^a UTI defined by presence of significant bacteriuria in a child with symptoms of a urinary tract infection.

^b Downgraded due to imprecision.

^c Downgraded due to heterogeneity.

^d Bacteriuria was defined as ≥10 000 CFU/mL of a pathogen on urine culture.

^e Downgraded due to high number of studies at high risk of bias.

Incidence of UTI at the end of therapy was examined in the studies by Zaoutis and Montini. ⁴ , ⁵ Children randomised to short‐course therapy had a 2.2% (CI: 0.0–4.3, I ² = 82.8%) higher risk of UTI compared with children randomised to standard‐course therapy (Figure 2). Because of the high unexplained heterogeneity between the results of these two studies, we then examined the risk of UTI at end of therapy according to the presence of fever at the time of enrolment. In febrile children, there was no significant difference in risk between short and standard duration treatment (RD = 0.4%, CI: −2.8 to 3.6) (Figure 2). Of note heterogeneity between the results from the two studies decreased only slightly (from 82.8% to 76.9%) in children with fever, indicating substantial unexplained heterogeneity between the results of the two studies. In afebrile children, there was 3.8% (CI: 0.9–6.8) higher risk of UTI at end of therapy in children randomised to short‐course treatment (Figure 2).

FIGURE 2.

UTI and Bacteriuria at end of therapy in children randomised to short‐ versus standard‐course antimicrobial therapy (panel A). UTI at end of therapy in children randomised to short‐ versus standard‐course antimicrobial therapy stratified according to the presence of fever at enrolment (panel B).

Bacteriuria at the end of therapy was reported in eight studies. Children treated with a short course of an antimicrobial had a 8.7% (CI: 5.3–12.2, I ² = 49.9) higher risk of bacteriuria at end of therapy than children treated with a standard course of antimicrobial (Figure 2); in a sensitivity analysis, in which studies with a high risk of bias were excluded, the difference in risk was slightly lower (RD = 7.3%, CI: 3.4–11.1, I ² = 26.6%) (Figure S2).

Late outcomes reported were UTI and bacteriuria 25–60 days after completing treatment (Figure 3). The risk of UTI and bacteriuria 25–60 days after end of therapy did not differ significantly (RD = −1.0%, CI: −3.7 to 1.7, I ² = 73.1%, and RD = 5.2, CI: −3.9, 14.4, I ² = 0%, respectively) between the two treatments. Results were similar when studies at high risk of bias studies were excluded (Figure S3).

FIGURE 3.

UTI and Bacteriuria 25–60 days after end of therapy in children randomised to short‐ versus standard‐course antimicrobial therapy.

The study by Zaoutis was the only one to report the incidence of diarrhoea and the rates of stool antimicrobial resistance. ⁴ There was no apparent difference in the risk of developing diarrhoea (RD = −3.0%, CI: −7.9 to 1.9) in children treated with antimicrobials or placebo. Similarly, rates of antimicrobial resistance in organisms recovered from the stool after therapy did not differ significantly between treatment groups (RD = 1.3%, CI: −3.1 to 5.7).

4. DISCUSSION

In this meta‐analysis of nine randomised controlled trials, which included 1171 children, most of which were afebrile at the time of presentation, we found that children treated with a short course of antimicrobials when compared to those treated with standard‐course antimicrobials had similar risk of UTI at the end of therapy (risk difference of 2.2%; n = 806). Of note, in the subgroup of children with fever at the time of presentation (n = 392), the risk of having a UTI at the end of therapy was similar (risk difference of 0.4); however, the size of this subgroup was relatively small and the results from the two trials reporting these data were heterogeneous. Accordingly, further studies of febrile children are needed to increase confidence in the pooled results prior to making any new recommendations.

Risk of bacteriuria at the end of therapy in children receiving short‐ versus standard‐course therapies differed by 8.7%. This was calculated from eight randomised controlled trials including 1029 children. While this risk difference is higher than what was observed for UTIs, it is still below the 10% risk difference threshold (i.e. which translates into a number needed to treat of 10) that many would consider clinically significant. Furthermore, the clinical significance of bacteriuria on urine samples obtained only for research purposes (i.e. in the absence of any symptoms and in many cases using samples obtained using a urine bag) is unclear. In studies that have followed these children (e.g. the SCOUT study), ⁴ no increase in rates of symptomatic UTIs were noted in the weeks to months following this episode of bacteriuria. Moreover, some of the difference was driven by studies with a high risk of bias. Accordingly, the observed difference in risk on bacteriuria notwithstanding, the evidence presented here supports that use of shorter courses of therapy.

At 25–60 days after therapy, outcomes (i.e. both UTI and bacteriuria) were comparable in children treated with short and standard courses of antimicrobials further supporting the use of shorter courses for the treatment of UTI in children.

Regarding difference in risk of adverse outcomes, we found no differences in rates of diarrhoea during therapy, or in the rates of antimicrobial resistance in organisms in stool the stool 1 month after the initiation of therapy. However, these data were reported in only one study. ⁴

A recent meta‐analysis ¹⁶ also analysed short‐course versus Standard‐course antimicrobials for UTI. That meta‐analysis differs from ours in multiple ways. First, it included multiple studies that used single dose treatment courses which were not included in our analysis. Second, it lacked a stratified analysis based on presence of fever at the time of diagnosis. Third, it did not include the study by Montini et al.

This meta‐analysis had several limitations. First, we did not have access to individual‐level patient data. Second, the outcomes were not uniformly defined across all included studies. However, variations in these outcomes were relatively minor (Table 1) which justifies pooling of the data.

Review of the literature to date suggests some improvements for future studies. First, because of the unexplained heterogeneity between the rates of UTI at the end of therapy in children with fever at the time of presentation, and because of the smaller number of children in these analyses, future studies should focus on enrolling children with fever. Second, focusing on UTI at the end of therapy as the main outcome of interest would allow not only the use of more definitive methods of urine collection but also for capturing more clinically meaningful data. Third, because of the lack of differences in late outcomes in the studies to date, future studies need not follow children for an extended period.

In conclusion, the bulk of the evidence to date suggest that clinically relevant outcomes of children without fever at the time of presentation who are treated shorter courses (i.e. 2–5 days) of antimicrobial therapy are similar to that of children treated with standard duration of antimicrobial therapy. Although the data suggest that the same may be the case in children who have fever at the time of presentation, our confidence in these results is lower because of the smaller number of children and because of the unexplained heterogeneity among studies. More trials of children with fever seem to be warranted.

AUTHOR CONTRIBUTIONS

Grace D. Mueller: Writing – original draft; data curation; formal analysis; writing – review and editing; investigation. Shannon J. Conway: Investigation; writing – original draft; writing – review and editing; formal analysis; data curation. Asumi Gibeau: Writing – review and editing; writing – original draft; formal analysis; data curation; investigation. Nader Shaikh: Conceptualization; investigation; funding acquisition; methodology; visualization; formal analysis; project administration; data curation; supervision; writing – review and editing; writing – original draft.

FUNDING INFORMATION

This study was supported by R01DK118033.

CONFLICT OF INTEREST STATEMENT

The authors have no conflicts of interest to report.

Supporting information

Data S1. Supplementary information.

Mueller GD, Conway SJ, Gibeau A, Shaikh N. Short‐ versus standard‐course antimicrobial therapy for children with urinary tract infection: A meta‐analysis. Acta Paediatr. 2025;114:479–486. 10.1111/apa.17546