Non-Typhoidal Salmonella Infection in Children: Influence of Antibiotic Therapy on Postconvalescent Excretion and Clinical Course—A Systematic Review

(1) Background: Although published recommendations are available, the use of antibiotics in non-typhoidal Salmonella (NTS) infections in children is still controversially debated in clinical practice. Patients might even be put at risk, with necessary antibiotic therapy being withheld due to the widespread concern of prolonged post-convalescent shedding. The authors conducted a systematic review to assess whether antibiotic treatment influences fecal excretion or the clinical course in children with NTS infection. (2) Methods: The review was carried out following the PRISMA guidelines. In a Medline database search, studies assessing the influence of antibiotic therapy on excretion and/or the clinical course of NTS infections were selected. Studies reporting on adults only were not considered. Out of 532 publications which were identified during the systematic literature search, 14 publications were finally included (3273 patients in total). Quality and bias assessment was performed using the Newcastle-Ottawa scale (NOS) or the Cochrane risk-of bias tool (ROB-2). (3) Results: Four early studies from decades ago demonstrated a prolongation of intestinal NTS excretion in children after antibiotic treatment, whereas most studies published more recently observed no significant influence, which might be due to having used more “modern” antibiotic regimes (n = 7 studies). Most studies did not describe significant differences regarding the severity and duration of symptoms between untreated patients and those treated with antibiotics. Quality and bias were mainly moderate (NOS) or variable (ROB-2), respectively. (4) Conclusions: There is no substantial evidence of prolonged excretion of NTS in pediatric patients after treatment with newer antimicrobials. Consequently, clinicians should not withhold antibiotics in NTS infection for children at risk, such as for very young children, children with comorbidities, and those with suspected invasive disease due to concerns about prolonged post-convalescent bacterial excretion. In the majority of cases with uncomplicated NTS diarrhea, clinicians should refrain from applying antibiotics.


Introduction
Salmonellae are gram-negative facultative intracellular enterobacteria. As the taxonomic classification is somewhat intricate in daily practice, clinicians still subdivide the anthropopathogenic serovars of Salmonella enterica subspecies I enterica into typhoidal and non-typhoidal Salmonellae (NTS), with S. typhimurium and S. enteritidis being the most prominent non-typhoidal serotypes. NTS are one of the most common bacterial pathogens worldwide causing enterocolitis, with an estimated 95 million cases globally in 2017 being a burden both in developing and developed countries [1].
The search was not restricted by the date of publication or the type of studies. After the literature search, titles and abstracts were screened to identify potentially appropriate articles. Subsequently, a detailed evaluation of these articles including crosschecking of the references was performed. Only studies reporting on children and adults or children only were included. Studies about adults only were excluded from this review. Only original articles (no reviews or case reports) were included in this analysis. Data acquisition started in October 2020 and was finished in July 2021.
The assessed characteristics of the considered articles included the type of study (e.g., randomized controlled trial or retrospective case series), the pathogen, the number of patients, the ratio of children (<18 years), and the administered antibiotics. The findings of these studies were investigated with the focus on the influence of antibiotic treatment on the intestinal excretion of NTS or on the clinical course of patients, respectively.
Quality assessment was performed independently by two investigators (J.L.L. and R.A.). Any disagreement in scoring was solved by general consensus.
For non-randomized studies, the Newcastle-Ottawa scale [39] was used (Table S2); for randomized-controlled trials, the Cochrane risk-of bias tool for randomized trials (ROB-2) [40] was applied.

Results
A flow-chart illustrating the study selection is shown in Figure 1. The systematic literature search identified 532 records, of which 396 scientific articles were screened for inclusion in this review. After screening of titles and abstracts, 343 articles were excluded because of irrelevance or unsuitability according to the abovementioned selection criteria. Hereafter, 53 potentially appropriate articles were evaluated in detail. Furthermore, 39 articles were excluded for the following reasons: insufficient data, no original articles (reviews or case reports), only adult patients, and/or no systematic consideration of antibiotic therapy. A total of 14 scientific articles published between 1954 and 2021 were subsequently assessed for this review [13,23,31,32,37,[41][42][43][44][45][46][47][48][49].
The results of the quality assessment are described in Tables S3 and S4. Of the nonrandomized trials, eight studies were rated "moderate quality" and one study "low quality". Of the randomized-controlled trials, the overall risk of bias was "low" for two studies and "high" for three studies.
Altogether, the selected studies comprised 3273 patients, with 95.8% of them being children (12 studies included children only, while in two studies, the rates of children were 95.1% and 64.3%, respectively). An overview of these articles is shown in Table 1.
Of these studies, five were prospective randomized controlled trials, five were retrospective case series, two were prospective controlled trials, and two were prospective case series. The randomized controlled trials were allocated between "antibiotic treatment" and "no antibiotic treatment" or placebo in patients with NTS infection [41][42][43]45]. In one of the prospective controlled trials, patients with severe enterocolitis were assigned to antibiotic treatment, while the rest received supportive treatment only [48]. A recent prospective case series compared children with appropriate, inappropriate, and no antibiotic treatment [47]. The other articles with an observational study design investigated the influence of antibiotics on the duration of NTS excretion without a randomization between antibiotic treatment and a control group.
The sub-group of Salmonella spp. was not specified in all of the investigated studies. In 12 of 14 studies, all NTS species were included, whereas in two studies S. typhimurium was the predominantly diagnosed pathogen. The antibiotic agents which were used changed over time. The most commonly administered substances before 1990 were chloramphenicol, ampicillin, and neomycin, while in more recent studies, fluoroquinolones and cephalosporins were predominantly applied. Altogether, the selected studies comprised 3273 patients, with 95.8% of them being children (12 studies included children only, while in two studies, the rates of children were 95.1% and 64.3%, respectively). An overview of these articles is shown in Table 1.
Of these studies, five were prospective randomized controlled trials, five were retrospective case series, two were prospective controlled trials, and two were prospective case series. The randomized controlled trials were allocated between "antibiotic treatment" and "no antibiotic treatment" or placebo in patients with NTS infection [41][42][43]45]. In one of the prospective controlled trials, patients with severe enterocolitis were assigned to antibiotic treatment, while the rest received supportive treatment only [48]. A recent prospective case series compared children with appropriate, inappropriate, and no antibiotic treatment [47]. The other articles with an observational study design investigated the influence of antibiotics on the duration of NTS excretion without a randomization between antibiotic treatment and a control group.
The sub-group of Salmonella spp. was not specified in all of the investigated studies. In 12 of 14 studies, all NTS species were included, whereas in two studies S. typhimurium was the predominantly diagnosed pathogen. The antibiotic agents which were used changed over time. The most commonly administered substances before 1990 were chloramphenicol, ampicillin, and neomycin, while in more recent studies, fluoroquinolones and cephalosporins were predominantly applied.
In 13 articles, the duration of excretion was investigated [23,31,32,37,[41][42][43][44][45][47][48][49]. Prolongation of NTS shedding after antibiotic therapy was found in four studies (including 127, 117, 70, and 239 patients, respectively) [32,37,41,49]. In 1965, Dixon et al. published an analysis of two similar outbreaks in schools after food poisoning caused by S. typhimurium [32]. Seven weeks after the outbreak, the percentage of positive fecal samples was 44.8% for children with antibiotic treatment versus only 4.7% for untreated  In 13 articles, the duration of excretion was investigated [23,31,32,37,[41][42][43][44][45][47][48][49]. Prolongation of NTS shedding after antibiotic therapy was found in four studies (including 127, 117, 70, and 239 patients, respectively) [32,37,41,49]. In 1965, Dixon et al. published an analysis of two similar outbreaks in schools after food poisoning caused by S. typhimurium [32]. Seven weeks after the outbreak, the percentage of positive fecal samples was 44.8% for children with antibiotic treatment versus only 4.7% for untreated children. Accordingly, two years later, Rosenstein et al. reported on an investigation of Salmonella excretion of 70 patients with symptomatic and asymptomatic NTS infections and found that patients who were treated with antibiotics had a substantially longer excretion time [37]. After twelve weeks, stool cultures were negative for all patients who did not receive antimicrobial treatment compared with 30.1% positive stool cultures in those patients who were treated.
These results were confirmed by a further study that demonstrated that patients who were given antibiotics were more likely to be symptomless excreters in a six-week period after treatment than those given a placebo (lactose) [41].
Ho et al. also reported a prolongation of fecal shedding, albeit only in children receiving inappropriate antibiotic treatment defined as either not active in vitro or given to patients without severe disease. They did not observe a prolongation of intestinal NTS excretion after adequate antibiotic therapy (active in vitro or given to patients with a high severity score) or in the group receiving no antibiotics [47].
Shortening of the duration of excretion was found in one study (including 30 patients) published by Chiu et al., who observed a significantly higher rate of clearance of NTS from stools after antibiotic therapy using ceftriaxone in children with bacteremia [48]. The authors compared two groups of 15 patients with NTS enterocolitis and bacteremia receiving treatment with ceftriaxone or no treatment, respectively. One month after therapy, the rate of NTS excretion was 20% for patients treated with ceftriaxone versus 63% for those without antibiotic treatment.
No influence of antibiotics on NTS excretion was found in six studies (including, in total, 1865 patients) [23,31,[42][43][44][45]. In the largest study focusing on this topic, comprising 1543 patients, of which 21.2% had received antibiotics during an acute NTS infection for a mean duration of six days, Barbara et al. reported that antibiotic treatment did not affect the fecal shedding at any time point up to three months post-infection as compared with age-matched untreated controls [23].
All four studies that observed a prolongation of NTS excretion after antibiotic treatment in children were published in 1974 or earlier. No such phenomenon was reported since then in the literature assessed for this review.
The impact of antibiotic treatment on the clinical course of the infection was reported in 11 studies [13,23,31,[41][42][43][45][46][47][48][49]. Six out of these studies did not observe a significant influ-ence of antibiotic treatment on the severity and/or duration of symptoms [31,[41][42][43]45,48]. The first of these studies was published by Macdonald et al. in 1954 [31]. In this early study, comprising 51 children, the mean duration from the beginning to clinical cure was similar: 19.5 days for treated children and 19.0 for the control group. The largest of these studies (168 patients, 64% children <12 years) that assessed the duration of symptoms also did not observe an overall significant difference between patients treated with neomycin and those treated with placebo [41]. Most of the works which did not observe a clear influence of antibiotic treatment on the clinical course were performed in earlier years.
In one study, a higher frequency of persistent symptoms after antibiotic therapy was observed [23]. In this investigation, performed three months after NTS infection, 6.0% of the subjects still complained of digestive symptoms, such as vomiting, abdominal pain, and diarrhea. The rate of patients reporting such symptoms was 9.5% in the group that received antibiotic therapy, compared with 2.9% in those who did not receive it.
A recent study observed a worse clinical outcome only in children with inappropriate antibiotic treatment, but not after treatment with appropriate antibiotic agents [47]. The authors stratified 126 children into "no antibiotic therapy" (n = 69 patients), "appropriate antibiotic therapy" (bacteremia or severe cases receiving antibiotics active in vitro; n = 24 patients), and "inappropriate antibiotic therapy" (cases with mild or moderate severity receiving antibiotics resistant in in vitro testing; n = 33 patients) groups. The "appropriate antibiotic therapy" group had a similar fecal excretion time compared to the "no antibiotic therapy" group (9.64 ± 6.22 days versus 11.13 ± 7.01 days). In a multivariate analysis, the authors showed that an inappropriate antibiotic therapy significantly prolonged the fecal excretion time (18.24 ± 24.67 days).
Further subsequent works on this subject, such as the ones published by Lin et al. in 2003 and by Tsai et al. in 2011, reported potential positive effects of a short-term antibiotic therapy for three to five days [13,46]. Lin et al. analyzed patients with a severe clinical cause which was defined as high C-reactive protein (CRP) levels, high fever, and signs of dehydration, and showed that a therapy using short-term ceftriaxone was clinically beneficial for this group of patients (e.g., rapid defervescence after 1.6 ± 1.4 days (mean ± standard deviation)). Based on a retrospective analysis of 683 children (386 with antibiotic treatment and 297 without), Tsai et al. similarly concluded that critically ill patients with high CRP levels and longer febrile duration prior to admission might benefit from the administration of third generation cephalosporins or fluoroquinolones. Treatment with these antibiotic agents resulted in a shortened duration of fever and in fewer days of hospitalization. Table 2 summarizes the findings of this systematic review.

Discussion
The adequate therapy for children suffering from infections caused by NTS species is principally well described in published recommendations [50][51][52][53][54]. International guidelines are scarce, though, and rarely address invasive NTS infections specifically [55]. The risk of prolonged excretion of Salmonella in pediatric patients treated with antibiotics is still a relevant matter of debate in routine clinical practice. Substantial insecurity about this issue among colleagues is not rare (as perceived in cases admitted to our university children's clinic as a tertiary referral hospital specialized in infectious diseases). Warnings to varying extend about this assumed risk can be found in virtually all publications on the subject of NTS infections in children. In addition to the aspect of prolonged shedding, a second issue frequently discussed is the effect of antibiotic therapy on the severity and duration of symptoms in pediatric patients with invasive NTS infection.
The very first study that systemically assessed these factors was the work by Mac-Donald et al. published in 1954 [31]. The authors observed neither a prolongation of NTS excretion nor an influence of antibiotic therapy on the clinical course. The first study which reported a prolongation of fecal shedding after antibiotic therapy was published by Dixon et al. in 1965 [32]. However, the cohorts described were most likely infected by different NTS strains, as the two outbreaks examined occurred ten years apart and in different geographical regions and the antibiotic regimens used were highly diverse, with one to three courses of various antibiotic agents applied. Furthermore, as almost all of the affected school children received antimicrobial medication, it is likely that students not severely affected were treated as well. In consequence, the conclusion of the study of a possible prolongation of fecal NTS shedding after antimicrobial therapy has to be perceived with caution.
Another study describing protracted clearance of NTS after antibiotic treatment was published by Rosenstein et al. in 1967 [37]. Limitations of this study include different age spectra of the untreated and the treated group (e.g., 68% vs. 10% in the age group 15 years or older), detection of NTS infection by a single culture, often from a rectal swab, and antibiotic treatment of relevant numbers of asymptomatic patients.
Further work propagating a lower clearance of NTS in antibiotically treated patients was published by the "Joint Project by Members of the Association for the Study of Infectious Disease" in 1970 in a randomized controlled trial, in which 168 patients (108 < 12 years) were assigned to receive either oral neomycin or placebo (lactose). Clearance rates and differences in fecal NTS excretion for the whole cohort including adults were calculated after just three weeks. For a fraction of 76 patients, there was a tendency for prolonged shedding after six weeks following neomycin treatment (36% vs. 25%). No clear differentiation between age groups was performed and severity of symptoms was not assessed. Therefore, the conclusions drawn by the authors regarding the influence of antibiotics on NTS shedding is debatable.
One more study described a prolongation of NTS excretion after antibiotic treatment in 117 children. Here, the tendency of protracted Salmonella clearance was discrete and the authors themselves stated that this effect may not have been related to antibiotic therapy alone, as it was applied most frequently in infants less than 3 months of age and mainly in very sick patients [49].
All studies that reported a prolongation of fecal NTS excretion in pediatric patients were published no later than 1974. Since then, only one observational study, which predominantly included adult patients, also observed a prolongation of excretion. In this observation, 51 out of 105 patients were followed with stool cultures subsequently to a foodborne S. typhimurium outbreak. Fifteen children seven years or older were included, eight hospitalized and antibiotically treated, of whom three children excreted NTS longer than four weeks. It remains unclear whether disease severity itself led to hospitalization or the antibiotic therapy per se was responsible for prolonged shedding in the small number of children [36]. Although frequently cited in the literature in support of the notion that administration of antimicrobial drugs extends NTS excretion, such a conclusion should be appreciated with caution, as only a fraction of the original cohort was followed and the number of children was small.
Inappropriate use of antibiotics, however, does seem to significantly prolong the time of excretion, as shown in patients with mild disease receiving antibiotics or patients with severe disease receiving antibiotics later being found to be resistant in vitro [47,60]. So, in accordance to published guidelines and generally accepted assumption, these results confirm that, especially for mild clinical courses of NTS diarrhea, empiric antibiotic treatment should generally be avoided [33,38,52].
Another parameter identified as a risk factor for a prolonged excretion was an antibiotic therapy course longer than seven days, suggesting that a short-term antibiotic treatment might be beneficial [61]. The adequate duration of the treatment has not yet been systematically investigated in larger clinical studies, though.
The majority of the studies assessed which were published in the last decades, including three randomized controlled trials, could not demonstrate any significant influence of antibiotic therapy on the duration of excretion [23,42,43,45]. One prospective controlled trial even observed a significantly shorter duration of NTS excretion after antibiotic treatment using ceftriaxone [48].
When weighing the possible risk of bacterial shedding and the potential influences of an antibiotic therapy, it also has to be considered that even without antibiotic treatment, the excretion of NTS is often prolonged, lasting up to 20 weeks especially in young children (under the age of five) [62].
A possible explanation for the higher rate of prolonged fecal excretion in patients treated with antibiotics can be derived from the design of most of the non-randomized studies. In many studies, the indication for an antibiotic treatment was mainly based on the severity of the disease and the patients' age. Consequently, especially very sick and young children had received antibiotics. Therefore, one possible explanation for the observed prolongation of bacterial shedding could be a presumably higher bacterial load in these more severely affected and young children leading to higher concentrations of NTS in the stool independent from an antibiotic intervention.
Apart from the impact of antibiotic treatment on fecal excretion, the influence on patients' symptoms and the clinical course of the disease might be even more important, especially for patients at risk, such as very young children as well as immunocompromized patients and those with co-morbidities. The majority of the analyzed studies reported no significant difference regarding the severity or duration of the children's symptoms after antibiotic therapy. The most recent studies even reported potential benefits of an antibiotic treatment using ceftriaxone for a subgroup of patients with a severe clinical course [13,46,63]. However, clinical symptoms of the patients that were included in this review were relatively heterogeneous, ranging from asymptomatic courses to severe disease, which was mainly defined by fever, CRP levels and/or bacteremia, as well as the need for hospitalization and the severity of diarrhea.
As mentioned above, the armamentarium of antibiotic agents has broadened in the last decades, with regimes being refined regarding the duration of treatment and the substances used. Chloramphenicol, for example, is nowadays considered as reserve antibiotic and is even not officially approved for young children in some parts of the world. "Newer" antibiotic agents, such as third-generation cephalosporins, did not cause a prolongation of excretion and were associated with potentially beneficial effects on the patients' symptoms [46,48]. Adequately powered clinical studies are warranted to evaluate the efficacy of these drugs for patients with NTS infections.
Another promising group of antibiotics for the treatment of NTS infection are fluoroquinolones [63,64]. However, the studies that investigated these agents included adult patients only. Fluoroquinolones should be prescribed with caution in children because of potential side effects, albeit their application should be considered in life-threatening instances involving the central nervous system, such as NTS-meningitis, because of their favorable blood-brain-barrier penetration [63][64][65].
A systematic Cochrane review published by Onwuezobe et al. in 2012 which analyzed antibiotic treatment for symptomatic NTS infection concluded that there was no evidence for the benefit of antibiotics in NTS diarrhea in otherwise healthy people. However, the authors were unable to comment on the antibiotic therapy of very young and very old patients or people with severe and extraintestinal disease as the trials included did not specifically address these subgroups of patients [33].
According to our literature review, adequate antibiotic treatment in vulnerable groups or children already presenting considerably ill with substantially elevated CRP-levels or fever seems to outweigh the questionable risk of a prolongation of fecal NTS excretion. Targeted antibiotic therapy should be performed whenever possible according to antimicrobial activity in vitro susceptibility testing [47,60].
The current recommendation to administer third generation cephalosporins (and ciprofloxacin) for complicated NTS infections should be followed in the light of antimicrobial resistance to these drugs being reported in still low but increasing numbers [66][67][68]. Considered critically important antimicrobials for human medicine, these drugs have been listed in the "Watch"-group of the 2019 WHO AWaRe classification and should be closely monitored in antibiotic surveillance and stewardship programs [69]. As antibiotics do not provide a beneficial effect in the vast majority of uncomplicated Salmonella enteritis cases, patients should be selected with scrutiny and restricted use of antibiotics should be a major goal to prevent further development of antibiotic resistance, even more as vaccines against NTS infections are still not available yet [70].
We acknowledge that this review has some limitations. The designs of the included studies were heterogeneous, comprising randomized trials and retrospective cases series, which limits the comparability and the conclusions drawn. Besides that, the quality of the included non-randomized studies was only moderate (n = 8) or low (n = 1) and the risk of bias for randomized trials was high for three of the five included studies. Another limitation is the heterogeneity of the antibiotic agents administered. Especially antibiotics that had been applied in earlier studies are nowadays no longer common in clinical practice. Furthermore, in the non-randomized studies, the indication for an antibiotic treatment was made depending on the clinical status of the patients (e.g., fever or CRP levels), which is a potential bias for the effect of an antibiotic therapy. In addition, no study explicitly examined other factors, such as coinfections, possibly influencing prolonged excretion of NTS or persistence of symptoms.

Conclusions
The evidence of protracted NTS excretion after antibiotic treatment in children is mainly derived from older studies using antimicrobials not routinely applied in today's clinical practice. Recent studies did not observe a prolongation of excretion of NTS and could even show benefits of antibiotic treatment (e.g., ceftriaxone) on the patient's symptoms in selected cases. This discrepancy between earlier and later observations can be attributed to considerable changes in the antibiotic agents used over time. The restriction of antibiotic treatment for NTS infection, especially in very young children, children with comorbidities, and in those with suspected invasive infection should be critically questioned and the ostensible risk of prolongation of excretion should not lead to a withholding of a treatment with modern antimicrobial agents in these vulnerable subgroups of patients.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/