Prophylaxis Options in Children With a History of Recurrent Urinary Tract Infections: A Systematic Review

This study was conducted according to the Preferred Reporting Items for Systematic Reviews Incorporating NMA guidelines and the Cochrane Handbook for Systematic Reviews of Interventions.¹¹  A prespecified protocol has been registered in OSF (https://osf.io/m28t5). Our search strategy was based on the publications in the main medical e-databases (PubMed/Medline, Scopus and Cochrane Library) (Supplemental Table 3) from inception to November 26th, 2023 for studies examining the efficacy of any medications or supplements as prophylaxis for RUTI in children. There were no limitations for publication year. We screened all the references from the included studies and identified systematic reviews and meta-analyses for additional studies. Clinicaltrials.gov, PROSPERO, OSF, and “gray literature” were searched to identify relevant unpublished, ongoing, or published studies and trials to avoid duplication. Only studies published in the English language were included in our NMA.

In this systematic review, we included only randomized controlled trials (RCTs) examining the efficacy of different antibiotics and dietary supplements at any dose as prophylaxis in UTI and kidney scarring in pediatric patients with a history of RUTI (2 or more episodes of UTI). The diagnosis of UTI was defined as a positive result of a urine culture (clean catch, urine bag [≥10⁵ cfu/mL], catheter [≥10⁴ cfu/mL], or any number of bacteria after suprapubic bladder aspiration) in addition to any symptoms indicative of a UTI in children (fever, abdominal pain, changes in continence pattern, urine color or odor, etc). Studies that included patients with congenital urogenital anomalies were excluded. The research question (patients-intervention-comparison-outcome) was defined using the criteria we set in the prespecified protocol.¹² 

Three independent reviewers (A.G., S.C. and P.M.) performed the search of the literature, extracting and importing all records in a reference management tool (rayan.qcri.org), and duplicates were removed.¹³  The remaining studies were assessed initially at the title and abstract level and then by full-text reading. In case of disagreements, a fourth reviewer (N.G.) made the final decision. Finally, 2 reviewers (A.S. and K.C.) independently extracted the data (first author, publication year, identification number, journal, country, study design, number of patients included in each study arm, funding source, intervention and comparator details, population, treatment, and follow-up characteristics) of the eligible studies with the help of a prespecified data extraction form for all the reviewers. If any disagreements arose, they were solved with the help of a third reviewer (N.G.). In case of missing data relevant to study characteristics, we contacted the corresponding authors to obtain them.

Each treatment agent was handled as a different node in the network. Different doses of the same agent were merged into a single treatment node, whereas placebo and no intervention were also merged under the “control” node.

The risk of bias was assessed independently by 2 examiners (N.G. and P.M.) using the revised Cochrane risk-of-bias (RoB 2.0) tool for randomized trials for each outcome.¹⁴  Any discrepancies were resolved by a third reviewer (D.T.). Studies were graded as low risk when all RoB 2.0 domains were classified as “low risk,” “some concerns,” or “high risk” in studies which had 1 domain classified as “high risk,” or 3 domains were deemed as “some concerns.”

Evaluations were performed through the confidence in the results from NMA using the CINeMA (Confidence in NMA) framework (https://cinema.ispm.unibe.ch/), which is an adaptation of the Grading of Recommendations Assessment, Development and Evaluation for NMA.^15,16 

The primary outcome was the presence of symptomatic (fever, abdominal pain, changes in continence pattern, urine color or odor, etc) cultured-verified RUTI during prophylaxis. Prespecified secondary outcomes were episodes of febrile UTI, scarring incidence, symptomatic (as mentioned above) UTI after prophylaxis, and during follow-up and adverse events.

Α descriptive table is provided to summarize the baseline characteristics of the eligible RCTs (Table 1). We performed an NMA by employing R software (Version 4.3.0) to conduct the statistical analysis. The frequentist random effects model was applied, using the R package “netmeta,” to assess the effect estimate of each treatment choice.¹⁷  The relative treatment effects were estimated using odds ratio (OR) for dichotomous outcomes and mean difference and 95% confidence intervals for continuous outcomes. Global network homogeneity was assessed with the Cochran’s Q test. The design-by-treatment interaction model as a global method was used to calculate the total inconsistency in NMA.¹⁸  Possible effect modifiers and their distribution across treatments were examined to identify potential violations in the transitivity assumption. We used the P-score for intervention ranking, taking values between 0 and 1, with higher values representing higher-ranked interventions.¹⁹  Small study effects (including publication bias) were assessed visually with comparison-adjusted funnel plots and formally tested with the Egger’s test.²⁰  Heterogeneity between the studies in pairwise meta-analysis was assessed using the I² test. I² value < 40% was set as low, 30% to 60% as moderate, 50% to 90% as substantial, and 75% to 100% as considerable.²¹ 

A P value < .05 was considered statistically significant. Additionally, a subgroup analysis was conducted including only the studies with children with VUR.

We planned a priori to combine studies with both parallel-group and crossover designs. We analyzed all crossover studies as if they were parallel-group designs. According to the Cochrane Handbook for Systematic Review of Interventions, although confidence intervals (CIs) are likely to be too wide and the trials would receive too little weight, this analysis is conservative.^21,22 

In total, 12 014 articles were identified through our initial search. After duplicate detection, 10 822 records were removed (Fig 1). After title and abstract screening, we assessed 68 studies for eligibility (full-text), and finally, 23 primary studies, published between 1975 and 2023 and a total of 3335 participants included in the NMA (Supplemental Table 4).^7,8,23–47 

Of the 26 studies, 13 were conducted in Europe,^{24–28,31–33,35–39,41,42,48}  8 in Asia,^{23,29,30,40,44–47}  3 in the United States,^8,34,43  1 in Argentina,⁴⁹  and 1 in Australia.⁷  Studies’ treatment and follow-up duration ranged from 2 months to 3 years. Participants’ mean age ranged from 3.8 to 14.7 years. Table 1 presents the characteristics of the included studies.

Taking into consideration the quality of the studies and the information concerning the primary outcome, 14 out of 23 studies were deemed of “high risk of bias” based on the RoB 2.0 tool (Fig 2).^{23,28–30,33–36,39,41,42,44,45}  Among the remaining studies, 10 were appraised as “some concerns.”^{7,8,19,24–27,31,32,37,40,43,46,47}  The lack of a well-described study methodology or the lack of a prespecified protocol were the main reasons for the increased risk of bias.

Assessments on certainty of evidence were performed where a network meta-analysis was performed. The summary of Grading of Recommendations Assessment, Development and Evaluation assessments for each comparison are presented in the Supplemental Table 5. Comparison-adjusted funnel plot along with the Egger’s test P value for the primary outcome is presented in Supplemental Fig 8. No evidence of small study effect was identified for this outcome. No funnel plots are presented for the secondary outcomes because of the small number of included studies.

In the NMA of different treatments, 9 interventions (including placebo or no intervention) have been compared, as illustrated in the network graph in Fig 2A. The 9 interventions (including control group of placebo or no intervention) resulted in a total of 18 theoretical comparisons for this primary outcome of interest. The cranberry group included both cranberry juice and tablets as supplementation. The line width and the node size in Fig 2A are proportional to the number of studies, directly comparing the treatments and number of participants included in each intervention node, respectively.

When all interventions were considered, cranberry had 59% lower odds of symptomatic UTI episodes during prophylaxis compared with the control group (odds ratio [OR] = 0.41, 95% CI = 0.23–0.74, P = .003, P-score = 0.71). Moreover, nitrofurantoin group had significantly lower odds of symptomatic UTI episodes during prophylaxis when compared with control group (OR = 0.21, 95% CI = 0.07–0.65, P = .007, P-score = 0.91), trimethoprim or sulfamethoxazole (TMP/SMX) group (OR = 0.29, 95% CI = 0.10–0.82, P = .02, P-score = 0.40), or trimethoprim group (OR = 0.23, 95%CI = 0.08–0.66, P < .01, P-score = 0.29). Although other interventions had clinical significance, with respect to reducing UTI incidence, they did not reach statistical significance in this analysis. These results are summarized in the Table 2 and Fig 3. Moreover, nitrofurantoin was ranked as the best intervention for reducing UTI incidence as compared with all available interventions.¹⁹  The certainty of network estimates was very low, mainly because of major concerns in within-study bias and imprecision (Supplemental Table 4).

Total Q statistic was calculated equal to 21.08, P = .007. The heterogeneity (within-designs) is significant (P = .01), whereas inconsistency (between designs) is not (P = .07). The design-specific decomposition of QW shows that the within-design heterogeneity can largely be traced back to the comparison of TMP/SMX and cranberry versus placebo. The Q statistic decreases considerably when assuming a full design-by-treatment random-effects model (Q = 3.98), and the between-design inconsistency is still not significant (P = .26). Results for detaching single designs and the full design-by-treatment interaction model do not show evidence of inconsistency between designs.

In this outcome, 6 interventions have been compared, as illustrated in the network graph in Fig 2B. The line width and the node size are proportional to the number of studies directly comparing the treatments and number of participants included in each intervention node, respectively.

The 6 interventions (including control group) resulted in a total of 8 theoretical comparisons for this secondary outcome of interest. The total Q statistic was insignificant (Qw = 0.67, P = .412), indicating no inconsistency in this network. When all interventions were considered, the various interventions did not reach statistical significance in this analysis. These results are summarized in Table 2 and Fig 4. Moreover, cefadroxil was ranked as the best intervention for reducing UTI incidence as compared with all available interventions, followed by nitrofurantoin (Fig 5).

Three interventions (including control group) have been compared to estimate their treatment effect on scars. The odds of new scars in lactobacillus group were 45% lower compared with the control group (OR = 0.55, 95% CI = 0.19–1.60, P value = 0.27, P-score = 0.86), whereas the TMP/SMX group had 9% lower odds of new scars compared with the control group (OR = 0.91, 95% CI = 0.54–1.51, P value = 0.71, P-score = 0.4). These interventions did not reach statistical significance in this analysis. The results are summarized in Fig 2C and 6.

The RCTs assessed this outcome of interest presented variability in the intervention groups because the different pharmacological agents have been administrated as a combination. Because of this fact, the 4 RCTs that assessed this outcome were analyzed as a comparison between the prophylaxis (antibiotic) group and the no prophylaxis(control) group. A pairwise meta-analysis was conducted using the “meta” package. The prophylaxis group has 39% lower odds of febrile UTI incidence; however, this result has not reached statistical significance (OR = 0.61, 95% CI = 0.24,1.57, P = .3). Substantial heterogeneity between the included RCTs has been recorded (I² = 60.8%). The summary of these findings is presented in Fig 6.

When only studies with children suffering from any grade of VUR were analyzed, nitrofurantoin, TMP/SMX, and lactobacillus had significantly lower odds of symptomatic UTI episodes during prophylaxis when compared with trimethoprim (Fig 7). However, there was significant heterogeneity in the grade of VUR, and the number of included studies is too small to assess the certainty of this result.

The latest data suggests there may be a clinical benefit of antibiotics and cranberry supplements for RUTI prophylaxis in the general population.^10,50,51  We collected all the available RCTs regarding antibiotic and nonantibiotic prophylaxis options against UTI episodes and kidney scarring in children with a history of RUTI and proceeded to the only available NMA making the first direct comparison of available prophylactic agents.

As a result of the careless use of antibiotics, the effectiveness of antibiotics is limited.^3,10  Resistance of pathogens, especially multidrug-resistant Gram-negative bacteria, to common antibiotics used for UTI and UTI prophylaxis, such as β-lactams, aminoglycosides, nitrofurantoin, and trimethoprim, has been identified worldwide.⁵²  In the future, combination therapy for UTI management may become a routine process, as resistance because of antibiotic overuse has increased.⁵³  Alternative prophylaxis options in RUTI and treatment of acute UTI are needed.

In the past, the biggest trials on antibiotic prophylaxis, RIVUR, and PRIVENT trials showed that prophylaxis with TMP/SMX may lead to a significant reduction of the risk of RUTI but no reduction in the risk of kidney scarring.^7,8  According to Williams et al, long-term antibiotic prophylaxis, especially with nitrofurantoin, may reduce the risk of further UTI in children with previous episodes of UTI or RUTI when compared with other antibiotics, such as TMP/SMX or no treatment or placebo.¹⁰  The authors underline that the benefit may be small and the increased risk of microbial resistance must be considered. Wang et al included 8 studies comparing antibiotic prophylaxis with no treatment in their meta-analysis, showing an improvement in the risk of febrile and symptomatic UTI in children with VUR.⁵⁴  No significant effect was demonstrated in kidney scarring or other adverse events. On the other hand, the most recent and well-designed meta-analysis (Autore et al) in children for the prevention of RUTI and long-term complications revealed a limited effect of antibiotic prophylaxis, stressing the need for new prophylactic agents.⁵⁵  Autore et al collected the response of a panel of experts in a list of clinical questions related to the management of UTI and congenital anomalies of the kidneys and the urinary tract, suggesting continuous antibiotics prophylaxis only in children with significant obstructive uropathies until surgical correction. The current guidelines suggest that a selective and individualized approach to long-term antibiotic prophylaxis should be considered, based on age, toilet training status, the risk of microbial resistance, and parental choice.^2,9  Other approaches, such as adequate fluid consumption, prevention of constipation, avoidance of urine withholding, and good hygiene, may decrease the risk of RUTI.³ 

During the last 20 years, nonantibiotic prophylaxis options for RUTI have been proposed for adults but also children.^50,56–58  A Cochrane review by Jepson et al included adults and children showing no risk reduction of RUTI with cranberry prophylaxis when compared with antibiotics or no treatment.⁵⁶  However, there was high heterogeneity in UTI definitions, patient population, and cranberry-containing products between studies, and no confident conclusion was drawn. Xia et al, in a recent meta-analysis with trial sequential analysis, showed a relative risk reduction of 45% in children with RUTI.⁵⁹  The results from the trial sequential analysis showed that the effects were conclusive, revealing the possible benefit of cranberry products as adjuvant therapy for UTI prevention. On the other hand, only 3 pediatric trials, with high heterogeneity, were included, revealing the limitations after careful inspection of the study.^37,42,60  Finally, the updated Cochrane review by Williams et al showed that compared with antibiotics, cranberry products have little to no difference in the risk of UTI episodes.⁵⁰  When cranberry products were compared with probiotics, they showed a significant reduction in the risk of UTI. Finally, the authors concluded that cranberry products may reduce the risk of UTI episodes in children and other population groups included in this review. This is the most recent review including 6 studies with children who had experienced more than 1 episode of UTI, making this study the most comprehensive regarding cranberries’ effectiveness in children with a history of RUTI.^{35,37,40,42,43} 

Another nonantibiotic prophylactic option that has been studied for UTI prevention in children is probiotic administration.^57,58  In a meta-analysis of 10 studies, Hosseini et al revealed no significant benefit in the incidence of UTI.⁵⁷  When probiotics were used as monotherapy, no significant effect was reached. On the contrary, when probiotics were coadministered with antibiotics, they showed a statistically significant effect on UTI prevention. The limitations of this study were the differences in probiotic strain, population, duration of treatment, control groups, and quality of studies. The same conclusion was drawn in the Cochrane review by Schwenger et al in 2015. Only 3 trials with high heterogeneity were included in the review, leading to poorer conclusions compared with the Hosseini et al study.^42,44,57,61 

Finally, many other nonantibiotic options of prophylaxis have been studied through the years for the prevention of UTI in children.^46,47,62,63  Yilmaz et al showed that a single dose of 200 000 IU vitamin A may have an adjuvant effect on the RUTI prophylaxis when compared with placebo.⁴⁷  In contrast, Merrikhi et al did not identify a positive effect of a daily 1000 IU dose of vitamin D compared with placebo in the prevention of UTI in children with a history of RUTI.⁴⁶  Elo et al compared methenamine hippurate after previous treatment with antibiotics with sulfafurazol and nitrofurantoin, suggesting a possible effect if it is used as an adjuvant treatment to antibiotics, showing a low incidence of side effects.⁶²  Finally, Arslanoglu et al revealed a decreased but nonsignificant number of UTI episodes in a group of young infants that were under a mixture of neutral short chain galactooligosaccharides and long-chain fructooligosaccharides compared with placebo (2 episodes to 7 episodes respectively).⁶³  The possible immune-modulating effect of probiotics may play a key role in the prevention of different infections during infancy.

Previous systematic reviews and meta-analyses in children susceptible to UTI tried to figure out if different nonantibiotic options of prophylaxis are effective in preventing RUTI.^51,53  Shira et al included a few pediatric data, and no safe conclusions on nonantibiotic prophylaxis in children can be drawn by this review.⁵³  Meena et al conducted a meta-analysis of randomized controlled trials in children with a history of RUTI or after a first episode of UTI who were treated with long-term nonantibiotic prophylaxis.⁵¹  Nonantibiotic prophylactic options, such as cranberry products, probiotics, and vitamin A or D were compared with no-treatment or placebo or antibiotics. According to this meta-analysis, cranberry products were found to be more effective in reducing the RUTI in children when compared with no treatment or placebo. Additionally, cranberry products were as effective as antibiotics in prophylaxis of RUTI. Finally, probiotics were better than no treatment or placebo in decreasing the risk of RUTI, but the evidence of data were of low quality.

Compared with previous studies, we have conducted a systematic review collecting all the available RCTs of children with a history of RUTI proceeding in an NMA, making the first direct comparison of all available prophylactic options (antibiotic and nonantibiotic) of all the appropriate studies examining all the main outcomes (symptomatic UTI episodes and kidney scarring incidence). We have tried to address the long-term concerns regarding UTI prophylaxis in children with a history of RUTI; which is the best prophylactic option in preventing UTI and kidney scarring in children with a history of RUTI? Is prophylaxis superior to no treatment or placebo in children with a history of RUTI and with or without any other predisposing factors for RUTI? We have tried to limit the total heterogeneity by including only children with a history of RUTI and not children with any episode of UTI with or without predisposing factors or vulnerable children to UTI in general. An advantage of our study is that we collected the largest pool of studies after a broad search through the main online databases. We have also screened all the available systematic reviews of our systematic search and meta-analyses (44 studies in total) regarding UTI prophylaxis, trying to eliminate the risk of any study missing. Finally, many interventions (9 interventions), theoretical comparisons (19 comparisons), and a substantial number of studies (23 studies) and patients (3335 patients) were included, highlighting the results as the most conclusive and valid regarding UTI and scarring prophylaxis in children with history of RUTI. No evidence of inconsistency was identified in this NMA.

As demonstrated by the results of this study, cranberry products and nitrofurantoin lead to lower odds of symptomatic UTI episodes during prophylaxis compared with the control group and control, TMP-SMX, or trimethoprim groups accordingly. Nitrofurantoin may be the best option for UTI incidence reduction compared with all available interventions in the literature. These results agree with the results of individual studies that were described previously in this discussion.^10,50,51  In respect of kidney scarring incidence, lactobacillus and TMP/SMX groups showed lower odds of new scars compared with the control group, but no statistical significance could be reached. In general, no benefit was detected after any kind of prophylaxis in children with history of RUTI and subsequent kidney scar formation.^7,8,30,44,45  Kidney scar formation after APN is a complex procedure. According to Murugapoopathy et al, females may be at a higher risk for RUTI because of the shorter length of the urethra and the proximity to the anus.⁶⁴  The disturbance of bladder epithelium function and immune cell response (chemotaxis of neutrophils, macrophage kidney response, toll-like receptors signaling resident immune cell response). Nevertheless, our NMA has limitations that should be acknowledged. A few RCTs did not include only children with a history of RUTI.^8,31–33,65 Additionally, the risk of bias (RoB) in more than half of the included studies raised many concerns (high RoB).^{23,28–30,33–36,39,41,42,44,45}  Another issue is that definitions of RUTI in different studies included were not consistent (difference in threshold [cfu/mL]of positive urine culture, episodes of RUTI definition and urine sampling method).^10,50,58  The small number of left RCTs with “some” concers RoB was not enough for a sensitivity analysis. In addition, only a few studies reported the incidence of scarring during or after prophylaxis.^7,8,30,44,45  Because of a small number of studies assessing direct comparisons, especially for secondary outcomes, the reliability and ability to generalize our results is decreased, despite the fact that no significant between-design inconsistency was detected. Furthermore, variations in different therapies and dosages, especially in cranberry products, were observed. In addition, Mohseni et al could not be included in NMA as only the group of children who received nitrofurantoin were eligible for analysis.⁴⁴  None of the studies showed superiority in decreasing the risk of scarring compared with the control group. The only effective method, according to the current literature, in kidney scarring after UTI in children remains corticosteroid administration during the acute phase of an episode.⁶⁶ 

In conclusion, nitrofurantoin and cranberry products may decrease the incidence of symptomatic UTI episodes in pediatric patients with a history of RUTI. No prophylaxis option can lead to a reduction of the risk of kidney scarring after RUTI. Decisions should be individualized based on the patients’ profile (VUR incidence). Future studies with optimal methodology, studying nonantibiotic prophylaxis options, focusing on children with RUTI, and the risk for kidney scarring are needed to draw further conclusions.

APN

acute pyelonephritis

NMA

network meta-analysis

OR

odds ratio

RCT

randomized controlled trial

RoB

risk of bias

RUTI

recurrent urinary tract infection

TMP/SMX

trimethoprim or sulfamethoxazole

VUR

vesicoureteral reflux