No evidence of carbapenemase-producing Enterobacteriaceae in stool samples of 1,544 asylum seekers arriving in Rhineland-Palatinate, Germany, April 2016 to March, 2017

Introduction: Since 2015, increased migration from Asia and Africa to Europe has raised public health concerns about potential importation of extended-spectrum β-lactamase-producing Enterobacteriaceae (ESBL-PE), specifically those producing carbapenemases (C-PE), into European hospitals. Aims: To inform infection control practices about ESBL-PE prevalence in asylum seekers and to investigate whether C-PE prevalence exceeds that in the German population. Methods: Cross-sectional study from April 2016–March 2017. Routinely collected stool samples from asylum seekers were tested for antibiotic resistant Enterobacteriaceae. Country/region of origin and demographic characteristics were explored as risk factors for faecal colonisation. Results: Of 1,544 individuals, 294 tested positive for ESBL-PE colonisation (19.0%; 95% confidence intervals (CI): 17.0–21.0). Asylum seekers originating from Afghanistan/Pakistan/Iran had a prevalence of 29.3% (95% CI: 25.6–33.2), from Syria 20.4% (95% CI: 16.1–25.2) and from Eritrea/Somalia 11.9% (95% CI: 8.7–15.7). CTX-M-15 (79%) and CTX-M-27 (10%) were the most common ESBL determinants. Highest ESBL-PE prevalences were observed in boys under 10 years and women aged 20–39 years (interaction: p = 0.03). No individuals tested positive for C-PE. Faecal C-PE colonisation prevalence in asylum seekers was not statistically significantly different from prevalence reported in German communities. Conclusion: In absence of other risk factors, being a newly arrived asylum seeker from a region with increased faecal ESBL-PE colonisation prevalence is not an indicator for C-PE colonisation and thus not a reason for pre-emptive screening and isolation upon hospital admission.


Introduction
Antibiotic resistance of pathogens and resulting limitations of therapeutic options increase morbidity, mortality and costs [1]. Since the beginning of this century, the number of infections caused by extended-spectrum β-lactamase-producing Enterobacteriaceae (ESBL-PE) has grown rapidly. Of these, carbapenemase-producing Enterobacteriaceae (C-PE) are of particular interest, as carbapenems are considered compounds of last resort against life-threatening infections. Major steps in the spread of antibiotic resistance in Enterobacteriaceae are horizontal exchange of mobile resistance genes into different clones and their dissemination over long distances, often facilitated by travel or migration of the colonised host [2]. International travel to south/ south-east Asia and Africa was found to be a risk factor for colonisation [3,4] and subsequent infection with ESBL-PE [5]. Introduction of these bacteria into unaffected hospitals is much dreaded since nosocomial outbreaks of C-PE have been reported world-wide [6].
In 2015, Europe was challenged by the arrival of a large number of asylum seekers, sparked primarily by the Syrian civil war, but also by other conflicts and humanitarian crises in southern Asia, western Asia, and Africa. Reports of high ESBL-PE and C-PE colonisation prevalences in hospitalised populations and among asylum seekers from these countries/regions [7][8][9][10][11][12], together with data on increased ESBL-PE and C-PE colonisation in returning travellers [3,4] led to discussions whether such migration may increase the risk of nosocomial transmission of multidrug-resistant bacteria in European countries with low C-PE prevalence [13].
In April 2016, the European Centre for Disease Prevention and Control (ECDC) recommended that individuals with recent exposure in high prevalence countries may also be considered for pre-emptive screening and isolation upon admission in European hospitals, even if they had no history of hospitalisation or antibiotic therapy before their arrival in Europe [14]. However, published studies regarding ESBL-PE/C-PE colonisation of migrants seeking asylum in the European Union are limited to research in hospitalised patients [9,10,15,16], do not stratify by country/region of origin [9,10,15,16], and lack statistical power [10,11,15]. Thus, while these studies provide rough estimates of the ESBL-PE/C-PE colonisation prevalence in populations that have a priori an increased likelihood of pre-morbidities, they do not represent newly arrived asylum seekers in general and thus cannot inform whether country/region of origin alone is a sufficient predictor for increased risk of ESBL-PE/C-PE colonisation. Therefore, research on colonisation status in sufficiently large populations recruited outside health care institutions is needed to determine whether newly arrived asylum seekers in general require pre-emptive screening and isolation upon hospital admission [14].
This study aimed to determine the colonisation prevalence of ESBL-PE/C-PE in asylum seekers newly arrived to the federal state of Rhineland-Palatinate, Germany, and to compare whether the prevalence of colonisation with C-PE exceeds that reported in the German community [17].

Methods
We conducted a cross-sectional study from April 2016-March 2017, on the prevalence of ESBL-PE and C-PE in stool samples from asylum seekers arriving in the federal state Rhineland-Palatinate, Germany.

Study population
According to federal state law, a medical examination including the analysis of one stool sample is mandatory for each individual within 1 week of seeking asylum [18]. Upon arrival in one of the 29 refugee reception centres in Rhineland-Palatinate, each asylum seeker is provided with a sample tube, pictogram and pre-paid packaging to be sent to the federal state public health laboratories for further processing. There, native stool samples are routinely tested for Salmonella spp., Shig ella spp. and helminth eggs [19]. Stool samples from asylum seekers were split into a routine and a study aliquot, given there was sufficient material. We recorded country of origin (self-reported), year of birth and sex from routinely collected data in pseudo-anonymised form using an eight digit number code. Samples from 42 asylum seekers providing incomplete personal information (n=22) or insufficient material (n=20) were excluded. Using data from the first 6 months, we decided to restrict sampling to subjects from the most frequently occurring regions of origin: Syria (Western Asia), Afghanistan, Pakistan and Iran (all South Asia), as well as Eritrea and Somalia (both East Africa).

Sample size
We powered our study to provide evidence against the null hypothesis: 'there is no difference in the prevalence of C-PE in asylum seekers arriving in Rhineland-Palatinate compared to that of the German community'. We used population prevalence estimates based on a study from Valenza et al. [17], who reported one C-PE in 3,344 individuals (0.03%; 95% CI: 0.00-0.17) in a community-based sample in Germany. Based on this, we considered a prevalence of 0.20% or higher in asylum seekers to be of public health relevance. Using 80% power, an α-error of 5%, and a one-sided comparison against a fixed population value, we calculated to enrol 1,512 individuals using the 'sampsi' command in Stata 14 (StataCorp LP, College Station, Texas, United States of America). A maximum period of 12 months was predetermined in case the desired sample size would not be met.

Microbiological analyses
Stool samples were sent to a diagnostic laboratory (Labor 28 GmbH, Berlin, Germany) for isolation of Enterobacteriaceae and ESBL phenotyping. All samples were spread on chromogenic screening agar (Brilliance ESBL agar, Oxoid, Wesel, Germany). Single colonies of different morphotypes were further cultured and bacterial species were identified using the Vitek 2 system (bioMérieux, Nürtingen, Germany). Susceptibilities to amoxicillin, amoxicillin/clavulanic acid, mezlocillin, piperacillin, piperacillin/tazobactam, tigecycline, cefuroxime, cefotaxime, cefepime, ceftazidime, cefpodoxime, ceftriaxone, imipenem, meropenem, ertapenem, gentamicin, tobramycin, ciprofloxacin, moxifloxacin, levofloxacin, trimethoprim and trimethoprim/sulfamethoxazole were determined and the results interpreted according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints [20]. All isolates with ESBL phenotype (resistance to cefotaxime and/or ceftazidime and/or ESBL phenotype according to the automated expert system implemented in Vitek 2 system) were sent to the Robert Koch Institute, Wernigerode, Germany, for molecular analyses.

Data analysis and statistics
All records were stored in an EpiData database and imported into Stata 14. Countries were grouped by region according to United Nations (UN) standard M49 [31]. Age was categorised in steps of 10 years, with individuals aged 40 years and over being combined in the '40+' group. Based on the binomial distribution, exact confidence intervals (CI) were calculated for prevalence estimates, using 95% CI as default and one-sided 97.5% CI for the upper limit of zero frequencies.
To determine potential risk factors for ESBL-PE/C-PE colonisation, we analysed the influence of age, sex and geographic origin on colonisation using cross tabulation as well as uni-and multivariable logistic regression modelling. We allowed for interaction between the explanatory variables and kept the interaction term, if a likelihood ratio test indicated improved model fit.

Ethical considerations
The study protocol was reviewed and approved by the ethics commission, board of physicians, Federal State of Rhineland-Palatinate, Germany (ref. number 837.487.15).

C-PE and ESBL-PE colonisation
As shown in Figure 3, prevalence of ESBL colonisation varied by age group and sex, with a range over categories of age and sex from 13% to 27%. Highest prevalences were observed in boys under 10 years ( (Figure 3).
No C-PE were detected in any of the samples, resulting in a prevalence estimate of 0.0% (upper 97.5% confidence limit 0.2%) for the overall population of asylum seekers. There was no evidence to reject the null hypothesis of C-PE in asylum seekers being equal to that previously reported in the German population (fixed population prevalence 0.20%, p value = 0.75). Based on sample size and according precision in estimating the proportion of C-PE in respective strata, the 0% C-PE prevalence estimates had upper onesided 97.5% confidence limits of 1.1% for Syria, 0.6% for Afghanistan/Pakistan/Iran, and 1.0% for Eritrea/ Somalia, respectively.

Genetic determinants of antibiotic resistance
Likelihood ratio test for interaction between age and sex gives a p value of 0.03.

Discussion
We conducted a large cross-sectional study on the prevalence of C-PE in newly arrived asylum seekers in Rhineland-Palatinate. Despite ample sample size and statistical power, our study did not find evidence supporting a higher C-PE prevalence in asylum seekers compared with that reported in the German community. Hence, it is unlikely that being a newly arrived asylum seeker from a country/region with increased prevalence of ESBL-PE colonisation alone, i.e. in the absence of morbidity that increases the likelihood of carbapenem use and/or hospitalisation before arrival in Europe, is a risk factor for C-PE colonisation and thus, should not be a reason for screening on admission to regular hospital wards.
A recent study reported C-PE colonisation in six of 290 asylum seekers (2.1%) who were screened upon hospital admission in Germany [16], which is contrary to what we found. The difference, however, could be explained by the different populations studied, with results from the hospital-based study likely influenced by patients with pre-morbidities and associated risk of C-PE colonisation, and therefore not representative for the overall population. Such differences re-inforce the need for both, community-based prevalence studies and an individual risk assessment upon hospital admission.
We found that the prevalence of ESBL-PE (i.e. non-C-PE) colonisation in newly arrived asylum seekers exceeds that of the population residing in Germany, which is assumed to be 2-6% in the community [17,32] and slightly higher (7-10%) in hospitalised patients [33,34]. With more than 29% of asylum seekers from Afghanistan, Pakistan and Iran, 20% from Syria and 12% from Eritrea and Somalia testing positive for ESBL-PE colonisation, the prevalences detected in our Table 1 Extended-spectrum β-lactamases in Enterobacteriaceae from 1,544 newly arrived asylum seekers, by country/region of origin, Germany, study are in line with estimates for south-east Asia (22%, 95% CI 7-44), eastern Mediterranean (15%, 95% CI 4-31), and Africa (22%, 95% CI 5-47) published by Karanika et al. [35] who combined the evidence of research on ESBL prevalence in 28,909 community dwellers in their large meta-analysis.
On the other hand, our estimates are considerably lower than those reported in a systematic review of ESBL colonisation prevalence in long-distance travellers returning to Europe; prevalence's were reported to be over 70% in travellers returning from south-and south-east Asia [4]. Travellers' diarrhoea and antibiotic use, both very common in travellers, were identified as risk factors for ESBL-PE colonisation [36]. The majority of travellers returning from long-distance travel clear ESBL-PE colonisation within the first month, with less than 10% still being colonised after 3 months [37]. Hence, apart from geographic variation in the colonisation risk, a predominance of ground-vs air-travel could explain the lower ESBL-PE colonisation prevalence observed in our sample of asylum seekers compared with long-distance travellers.
From a public health point of view, this puts our findings into perspective. First, considering the large number of air travellers returning to Europe from high prevalence countries each year, ESBL-PE faecal colonisation in asylum seekers is probably a minor contributor to the overall burden of ESBL-PE import. Second, as observed in travellers [37], ESBL-PE colonisation in asylum seekers is likely temporary. Thus the prevalence of colonisation should decrease to that of the population living in the host country after a few months. To confirm the latter hypothesis, research on the duration and risk factors of faecal ESBL-PE carriage in newly arrived asylums seekers is needed.
A recent review discussing enhanced contact precautions for all in-patients colonised with ESBLproducing E. coliconcluded that the evidence base for their implementation is 'rather weak' [38] and currently these are not recommended by European guidelines [39]. In Germany, screening for ESBL-PE other than CPE is currently not recommended upon hospital admission and enhanced infection control measures for patients colonised with these bacteria are only recommended on high-risk wards (i.e. neonatology, haematology and intensive care units) [40]. Hence, our findings on ESBL-PE (i.e. non-C-PE) colonisation in newly arrived asylum seekers being predominantly caused by E. coli, do not warrant additional infection control measures. However, they may help to guide the choice of diagnostic tests, calculated antibiotic therapy, as well as the planning of invasive procedures in newly arrived asylum seekers.  [31]. c Includes isolates from countries with less than five positive stool samples. d P value for chi-squared test testing the null hypothesis: 'Presence of any β-lactamase/any plasmid-encoded quinolone resistance genes among all ESBL isolates is equally distributed over countries of origin'. e bla TEM and bla OXA-1-group genes were not completely sequenced. Data are number (column %) of detected β-lactamase/plasmid-mediated quinolone resistance (PMQR) genes among ESBL-producing Enterobacteriaceae (total N), by country of origin, unless indicated otherwise. All bla TEM , bla OXA-1-group , bla SHV , bla CMY and PMQR genes occurred in combination with ESBL genes (bla CTX-M ).
We observed that sex modified the effect of age on ESBL-PE colonisation prevalence. The odds of ESBL-PE colonisation were highest in young males under 10 years and between the age of 30 and 39 years and in females between 20 and 39 years. The latter finding has also been described in other studies [41] and could potentially be explained by a higher incidence of urinary tract infections and associated antibiotic therapy in women of reproductive age.
Molecular analyses of ESBL-PE isolated in this study showed a dominant proportion of CTX-M-15 followed by CTX-M-27. CTX-M-15 is the most prevalent ESBL determinant worldwide and was also reported in up to 50% of ESBL-PE positive stool samples from Germany [17,42,43]. We found 24% of epidemic lineage E. coli-ST131 among ESBL-PE of asylum seekers, which is comparable to proportions in isolates from community dwellers, ambulatory and hospitalised patients in Germany [43]. From a public health point of view, this is relevant, since E. coli-ST131 and particularly the sublineage ST131-O25b is more virulent [44] and known to cause large outbreaks in both, the community and healthcare setting [45,46].We further observed that E. coli-ST131 mainly produced CTX-M-15 and CTX-M-27 (Supplementary Table S1). Of note, CTX-M-27 accounted for a third of all ESBL genes harboured by ST131 and half of those in ST131-O25b. The finding of CTX-M-27 in ESBL-PE from asylum seekers from all geographic regions supports the notion of its successful worldwide expansion, as proposed by Matsumura et al. [47].
Our findings support other studies that propose human travel as a main driver for the temporal and geographical shift in CTX-M-producing Enterobacteriaceae [48] and the E. coli-ST131 epidemic lineages [49]; a public health response that goes beyond targeting asylum seekers is now awaited.
This study has some limitations. First, precise information on travel routes and time would have been desirable, but could not be collected as stool samples from routine screening were used. However, this does not impact the main finding of a low prevalence of C-PE colonisation among asylum seekers in Germany. Second, we have no information regarding recent contact to the healthcare setting (e.g. hospital) or use of antibiotics.
With regards to C-PE prevalence, our main outcome of interest, inclusion of 'pre-morbid asylum seekers' with an increased risk of antibiotic intake and/or hospitalisation compared to 'community dwellers' would lead to an overestimation. Therefore, we assume that this information bias has not affected our 0% prevalence finding. Finally, like other studies on the import of ESBL-PE and C-PE to Europe through international travel [3], we were not set up to detect OXA-48-like producers without ESBL phenotype through the use of ESBL screening agar.
In summary, our study detected no faecal colonisation with C-PE in 1,544 newly arrived asylum seekers. These data support the notion that being an asylum seeker from a country/region with increased ESBL-PE colonisation prevalence alone is not an indicator for C-PE colonisation upon arrival in Germany and thus not a reason for pre-emptive screening or isolation upon hospital admission.
Note *This designation is without prejudice to positions on status, and is in line with UNSCR 1244/1999 and the ICJ Opinion on the Kosovo declaration of independence. ESBL: extended-spectrum β-lactamase; H0: null hypothesis. a The sequence type of one out of n = 310 E. coli isolated from asylums seekers could not be determined by PCR (ambiguous results). b Grouped according to United Nations Regions M49 standard [31]. c Includes isolates from countries with less than five positive stool samples. d P value from chi-squared test testing H0: 'The proportion of E. coli ST131 among all ESBL-producing E. coli (100%) is equally distributed over countries of origin'. e P value from chi-squared test testing H0: 'Among ESBL-producing E. coli ST131 (100%), the proportions of E. coli ST131 serotypes O16 and O25b are equally distributed over countries of origin'. Data are number (column %) of clonal lineages ST131-O16 and ST131-O25b among ESBL-producing E. coli (total N), by country of origin.