Introduction
Infants (< 18 months old) have more and closer interpersonal contacts than older children and adults. Additionally, they are not yet aware of personal hygiene, resulting in a higher chance of faecal-oral transmission of microorganisms within this age group [
1]. For this reason, bacteria can pass easily among children in day care centres (DCCs) [
2] and, subsequently, children attending DCCs are at higher risk for infections [
1]. Moreover, the declining maternal immunity, in combination with an immature immune system, makes infants even more susceptible to infections and, therefore, more likely to receive antimicrobial treatment. This treatment exerts selection pressure on the intestinal microbiota and promotes carriage of resistant bacteria [
1]. This makes DCCs ideal settings for drug-resistant strains of bacteria to emerge [
2]. Still, few studies have focussed on the prevalence and risk factors for resistant bacteria in young children. An overall prevalence of extended-spectrum beta-lactamase (ESBL)-producing bacteria in DCC-attending children of 4.0–16.8% was established in the Netherlands, in France and in Sweden between 2012 and 2016 [
1,
3‐
5]. In Sweden, a more than six-fold increase in the prevalence of ESBL-producing
Enterobacterales (ESBL-E) was determined in 2016 compared with a similar study conducted 6 years earlier [
5].
Finally, the infectious disease burden not only concerns the attending child. Infectious pathogens, including their antimicrobial resistance properties, may transmit via children, caretakers, parents and families [
3,
6] into the society at large. Therefore, it is important to intensify our efforts in infection control and antimicrobial stewardship [
7] and to support DCCs in their efforts to control and prevent infectious disease transmission within their facility.
In this point prevalence study we measured different infection risk factors in a standardized way based on the Infection RIsk Scan (IRIS), a tool to measure the quality of infection control and antimicrobial use, both at individual and DCC group level [
8]. This study included DCCs in three southern provinces of the Netherlands (Limburg, Noord-Brabant and Zeeland) and the five Flemish provinces of Belgium (Antwerpen, Limburg, Oost-Vlaanderen, Vlaams-Brabant and West-Vlaanderen). This study was part of a larger Interreg project, which aimed at broadening the knowledge regarding antimicrobial resistance and use in different healthcare and veterinary settings among cross-border countries, specifically Belgium and the Netherlands [
9].
The study aimed to determine the prevalence of faecal antimicrobial resistant bacteria carriage in children attending DCCs and to assess and identify infection risk factors within DCCs in the Netherlands and Belgium.
Data analysis
Descriptive analyses were used to show the presence of infection risk factors in DCCs and AMR risk factors in children. A prevalence of children in DCCs carrying ESBL-E, CPE, VRE, and CipR-E was calculated, by dividing the number of children testing positive (based on phenotype) by the total number of included children. Risk ratios (RR) (binomial variables) or univariate logistical regression (odds ratio: OR) (continuous variables) were used to assess the difference between the measured variables between both countries.
Multilevel logistical regression analyses were used to define which child/DCC characteristics predicted the presence/absence of ESBL-E, CPE, VRE, and CipR-E. Multilevel analyses take the hierarchical structure of the data into account [
14]. A two-level model was used: (1) child level (Level 1) and (2) DCC group level (Level 2). This means that children are nested within groups. Variables at child level were age, days per week attending the DCC, gender, antimicrobial use, hospital admission, parental occupation, animal contact and travel abroad. Variables at group level were the country and variables related to toilet hygiene, food hygiene, hand hygiene and environmental contamination [
15]. For the individual data (questionnaires) the missingness was considered as completely at random (MCAR), as incomplete questionnaires were totally at random. For the data on group level, the missingness was considered at random (MAR), as the missing data was mostly linked to the group itself. No imputation techniques were performed, but a complete case analysis was used.
Two multilevel models were built, one with ESBL-E as the dependent variable and one with CipR-E as the dependent variable. For both models, predictors were first assessed in univariate models (see Additional File
1) with the relevant dependent variable and checked for multicollinearity. Independent variables significant at the 0.25 level were included in the final model. The variable “Staff wash their hands after going to the toilet” was excluded from the model because of the high correlation with “Liquid soap available at the sink at the staff toilet”. The general variables “animal contact” and “travel abroad” were disaggregated into “place where the animal lives” and “travelling to continent”. Some variables were only observed on infant groups (< 18 months old) (e.g. only formulae in powdered form are accepted), while other variables were only observed on toddler groups (≥18 months and < 4 years old) (e.g. no potties are used). Because these variables introduced a lot of missing values and reduced the number of observations in the model, only the variables who were available for all kind of group types were included in the models.
First, a null (intercept only) model with no predictors at any level was built. The null model serves as comparison for other models and provides information on how much variation in the outcome exists between level-2 units [
14]. Second, child characteristics (Level 1 variables) were added to the model. Finally, a third model was built, including risk factors at DCC group level (Level 2 variables). The likelihood ratio (χ
2) test was used to test the improvement of fit for each model [
14].
All analyses were performed in SAS Enterprise Guide 7.1 (SAS Institute Inc., Cary, North Carolina, USA). A p-value of 0.05 was considered statistically significant.
Discussion
Main results
In children attending DCCs we showed an overall ESBL-E prevalence of 16% in Belgium versus 6% in the Netherlands, and an overall CipR-E prevalence of 17% in Belgium versus 8% in the Netherlands. The use of antimicrobial agents and hospital admissions among children attending DCCS was significantly higher in Belgium, compared to the Netherlands. For both child and DCC group characteristics, some significant differences could be observed between both countries. The final multilevel models indicated that, children who travelled to Asia within the previous 6 months had a higher odds to be a carrier of ESBL-E, whereas children who received an antimicrobial treatment within the previous 6 months, had a higher odds to be a CipR-E carrier. Attending DCC groups where the changing mat was cleaned after each use was found as a protective factor for carriership of CipR-E. No rectal carriage of VRE and of CPE was found in children attending a DCC on a phenotypical level.
a.
Antimicrobial use and hospital admission
Antimicrobial use differed significantly between the two countries. In 2014, 41% of the Belgian children (age < 16 years) received a systemic or ophthalmological antimicrobial, delivered by the home pharmacy (no hospital consumption) [
16]. This included both antimicrobials prescribed in primary care and secondary care (paediatricians). A Dutch study showed that 15% of Dutch children (age ≤ 12 years), received at least one oral antimicrobial prescription per year during 2000–2010 in primary care [
17]. The prescription rates for oral antimicrobials among Dutch children decreased significantly during the period 2006–2010 [
17]. In both countries, the percentage of oral prescriptions was the highest in the youngest age groups (0-4 years) [
16,
17]. Even though these studies do not describe the exact same population and more recent studies on antimicrobial use in young children in Belgium and the Netherlands are lacking, they seem to confirm that antimicrobial use among children differs between both countries. However, in this study parents were only questioned whether the child had received an antimicrobial agent in the previous 6 months, without asking further details about the drug used or when it was administered. Therefore, a recall bias might have been introduced. In addition, a period of 6 months may have been too long or the question too broad to show a significant effect.
There is also a significant difference in the number of hospital admissions in both countries. Additional information to explore further the difference in hospital admissions between both countries is missing. Only a general question was asked to the parents and no reason and/or date of admission were asked. Additionally, there was no distinction between ambulatory care and hospital admissions with at least one overnight stay. However, these forms of bias might have been introduced in both countries, allowing us to assume that there is indeed a substantial difference between the two countries.
Additionally, there are differences in the organization of DCCs for infants and toddlers, where Belgian DCCs have lower levels of preventive hygiene and have younger children who, on average, spend more days per week in day care. This suggests that cultural differences between both countries might play an important role in the use of antimicrobials and the emergence of antimicrobial resistance. We, therefore, recommend exploring further the difference in antimicrobial use and hospital admissions in children between the two countries more in-depth, in order to improve the policy on antimicrobial use and hospital admissions in children.
b.
Carriage of resistant bacteria
Rectal carriage of ESBL-E and CipR-E was significantly higher in Belgium than in the Netherlands. In the larger Interreg project, similar results were found in different healthcare and veterinary settings [
9,
18,
19]. In addition, the study done in the hospital setting showed that the number of ESBL-E carriers tested within 24 hours of hospital admission, representing a community carriage rate, was also higher in Belgium than in the Netherlands [
9]. A possible hypothesis is that the results found in children attending DCCs and found in the other settings are a reflection of carrier status in the community. Unfortunately, a surveillance network is lacking in Belgium, which makes it difficult to monitor the epidemiology of carriage of antimicrobial resistance in the community [
20]. Studies conducted between 2003 and 2018 reported higher ESBL carriage rates in the Belgian (11.6%) than in the Dutch community (5–10%) [
21‐
25]. However, it should be noted that the prevalence in the Belgian community is based on one study with a sample of patients who were admitted to a geriatric unit in one teaching hospital in Belgium [
24]. As the prevalence of faecal carriage among healthy individuals has increased (eight-fold) during the last two decades [
26], the cited studies probably underestimate the current situation.
An alternative hypothesis that may explain the high prevalence in this study is the dissemination via direct contact as suggestions for household transmission have been described [
3,
18]. Moreover, the results of molecular typing and whole genome sequencing observed transmission of ESBL-E in Belgian DCCs with a high ESBL-E prevalence [
27].
Surprisingly, the carrier rate for CipR-E in children attending a DCC was so high in both countries. The use of ciprofloxacin and by extension fluoroquinolones in children under 16 years of age induces an increased risk of cartilage damage [
28]. In practice, fluoroquinolones are only used in strict indication in children [
16,
17,
29,
30]. This strengthens the hypothesis that the results established in children are a possible reflection of carrier status in the community.
Numerous enzymes associated with ESBL activity (mainly CTX-M) can diffuse easily due to their mobile genetic elements that mediate rapid dissemination. These are also linked with transfer of other genes that confer resistance to beta lactams as well as other antibacterial agents such as quinolones [
26].
c.
Risk factors for carriage
Risk factors for carriage have mainly been studied in adult hospitalized patients [
3,
21]. Reported risk factors for carriage of ESBL-producing bacteria in healthy adults are travelling [
22,
23], being owner of an animal [
31], poor kitchen hygiene [
23] and having a child attending a DCC [
3]. Attending a DCC was determined to be a risk factor for carriage of ESBL-producing bacteria in children [
3]. While looking at the children attending a DCC, being less than 1 year old and having paper towels available in the DCC increased the odds to be a carrier of ESBL. Prohibiting ill children from entering the DCC, extra supervision on handwashing of sick children and consistently reporting to local health authorities have been found as protective factors which lower the odds of being a carrier of ESBL [
1]. Our study confirmed travelling to Asia as a risk factor for carriage of ESBL-producing bacteria in children attending a DCC. Despite recent travel to Asia being a clear significant risk factor for carriage of ESBL-E and CipR-E, this risk factor will not have a major impact on carrier status of resistant bacteria, as only a small number of children travel to Asia. To our knowledge, no studies evaluated risk factors for carriage of CipR-E in children.
It was unexpected to see that antimicrobial use did not emerge as a risk factor for ESBL-E and as a (borderline) significant risk factor for CipR-E. This might be explained by the lack of detailed information about the use of antimicrobials. A surveillance study in the community might give additional information about carriership of antimicrobial resistance in the community and children and provide insights in possible risk factors [
25].
Strengths and limitations
To the best of our knowledge, this is the first study that assessed ESBL-E, VRE, CPE and CipR-E carriage rates in children attending a DCC in Belgium and compared the results with the Netherlands. Additionally, this study has explored risk factors for AMR carriage, including ATP measurements. Besides the child related risk factors for AMR carriage, we included several pillars of infection prevention and control on the level of the DCC groups simultaneously, which can be seen as an added value. Two central laboratories analysed the samples, using a standardized protocol, which can be seen as another strength.
One of the limitations of the study is that, due to the lack of existing evidence, the included risk factors are mainly based on the experiences of the implementation of the IRIS in hospitals, national guidelines, and expert opinions. A lot of the risk factors on DCC group level, were measured through observation by an IPC expert. To improve the standardisation of data collection, all local IPC experts received thorough training and revision moments were provided to discuss certain issues in group.
The sampling method differed between the two countries, both having their limitations. For Belgium, only a subsample of the DCCs were recruited and for the Netherlands a non-probability sampling was used with similar coverage of the six regions involved. The representativeness of the DCCs operating in these two regions can be questioned, although we do not expect a strong association between the selection of DCCs and prevalence of AMR. Moreover, only in a part of both countries, DCCs were recruited to participate in the study for practical reasons (Flanders represents 58% of inhabitants of Belgium and the southern provinces in the Netherlands 23%), which may compromise the generalizability of the results.
Detailed information about the use of antimicrobials and hospital admissions is lacking, which might have been of added value to further explore the difference in antimicrobial use and hospital admissions between both countries.
Acknowledgements
i-4-1-Health Study Group:
Lieke van Alphen14, Nicole van den Braak15, Caroline Broucke2, Anton Buiting16, Liselotte Coorevits17, Sara Dequeker1,2, Jeroen Dewulf18, Wouter Dhaeze2, Bram Diederen19, Helen Ewalts-Hakkoer6, Herman Goossens11, Inge Gyssens20, Casper den Heijer3,4, Christian Hoebe3,4,13, Casper Jamin14, Patricia Jansingh21, Jan Kluytmans9,12, Marjolein Kluytmans-van den Bergh9,12, Stefanie van Koeveringe11, Sien De Koster11, Christine Lammens11, Isabel Leroux-Roels17, Hanna Masson2, Ellen Nieuwkoop16, Anita van Oosten22, Natascha Perales Selva23, Merel Postma18, Stijn Raven5, Veroniek Saegeman24, Paul Savelkoul14, Annette Schuermans24, Nathalie Sleeckx25, Tijs Tobias26, Paulien Tolsma7, Jacobien Veenemans22, Dewi van der Vegt27, Martine Verelst24, Carlo Verhulst9, Pascal De Waegemaeker17, Veronica Weterings9, Clementine Wijkmans6, Patricia Willemse–Smits28, Ina Willemsen8,9.
1Department of Epidemiology and public health, Sciensano, Brussels, Belgium. 2Agency for Care and Health, Infection Prevention and Control, Government of Flanders, Brussels, Belgium. 3Department of Sexual Health, Infectious Diseases and Environmental Health, Living Lab Public Health, South Limburg Public Health Service, Heerlen, the Netherlands. 4Department of Social Medicine, Care and Public Health Research Institute (CAPHRI), Maastricht University, Maastricht, the Netherlands. 5Department of Infectious Diseases, Public Health Service region Utrecht, Zeist, the Netherlands. 6Public Health Service Hart voor Brabant, Tilburg, the Netherlands. 7Public Health Service Brabant-Zuidoost, Eindhoven, the Netherlands. 8Contrain Infectiepreventiecoach, Breda, the Netherlands. 9Amphia Hospital, Breda, the Netherlands. 10Public Health Service Zeeland, Goes, the Netherlands. 11University of Antwerp, Antwerp, Belgium. 12UMC Utrecht, Utrecht, the Netherlands. 13Department of Medical Microbiology, Infectious Diseases and Infection Prevention, Care and Public Health Research Institute (CAPHRI), Maastricht University Medical Centre (MUMC+), Maastricht, the Netherlands. 14Maastricht University Medical Center +, Maastricht, the Netherlands. 15Avans University of Applied Sciences, Breda, the Netherlands. 16Elisabeth-TweeSteden Ziekenhuis, Tilburg, the Netherlands. 17Ghent University Hospital, Ghent, Belgium. 18Ghent University, Ghent, Belgium. 19ZorgSaam Hospital, Terneuzen, the Netherlands. 20Hasselt University, Hasselt, Belgium. 21Regional Public Health Service Limburg Noord, Venlo, the Netherlands. 22Admiraal de Ruyter Hospital, Goes, the Netherlands. 23Antwerp University Hospital, Antwerp, Belgium. 24University Hospitals Leuven, Leuven, Belgium. 25Experimental Poultry Centre, Geel, Belgium. 26Utrecht University, Utrecht, the Netherlands. 27PAMM Laboratory for Pathology and Medical Microbiology, Veldhoven, the Netherlands. 28Elkerliek Ziekenhuis, Helmond, the Netherlands.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.