High rates of meticillin-resistant Staphylococcus aureus among asylum seekers and refugees admitted to Helsinki University Hospital, 2010 to 2017

Introduction Antimicrobial resistance is increasing rapidly in countries with low hygiene levels and poorly controlled antimicrobial use. The spread of resistant bacteria poses a threat to healthcare worldwide. Refugees and migrants from high-prevalence countries may add to a rise in multidrug-resistant (MDR) bacteria in low-prevalence countries. However, respective data are scarce. Methods We retrospectively collected microbiological and clinical data from asylum seekers and refugees treated at Helsinki University Hospital between January 2010 and August 2017. Results Of 447 asylum seekers and refugees (Iraq: 46.5%; Afghanistan: 10.3%; Syria: 9.6%, Somalia: 6.9%); 45.0% were colonised by MDR bacteria: 32.9% had extended-spectrum beta-lactamase-producing Enterobacteriaceae (ESBL-PE), 21.3% meticillin-resistant Staphylococcus aureus (MRSA), 0.7% carbapenemase-producing Enterobacteriaceae (CPE), 0.4% multiresistant Pseudomonas aeruginosa (MRPA), 0.4% multiresistant Acinetobacter baumannii (MRAB); no vancomycin-resistant Enterococcus (VRE) were found. Two or more MDR bacteria strains were recorded for 12.5% of patients. Multivariable analysis revealed geographical region and prior surgery outside Nordic countries as risk factors of MRSA colonisation. Young age (< 6 years old), short time from arrival to first sample, and prior hospitalisation outside Nordic countries were risk factors of ESBL-PE colonisation. Conclusion We found MDR bacterial colonisation to be common among asylum seekers and refugees arriving from current conflict zones. In particular we found a high prevalence of MRSA. Refugees and migrants should, therefore, be included among risk populations requiring MDR screening and infection control measures at hospitals.


Introduction
Antimicrobial resistance (AMR), a major health problem worldwide, surges most rapidly in regions with low level of hygiene and poor control of antimicrobial use [1]. AMR spreads across the globe and its extent has been recognised by international bodies at the highest level: in 2016 AMR was addressed at a General Assembly session of the United Nations as the greatest and most urgent global health risk [2].
A major concern about AMR is its spread to healthcare settings in low-prevalence countries with severe consequences: treatment failures, increase in the number of serious infections, and dramatic cost implications [3]. To prevent the spread of AMR to hospitals, patients with particular risk of colonisation and infection with multidrug-resistant (MDR) bacteria should be identified and subjected to infection control measures at the admission stage. Numerous studies have identified international travel as a major risk factor for colonisation: approximately one third of visitors to high-prevalence regions acquire MDR bacteria during ordinary tourist travel [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18]. Refugees and migrants who have lived for years in high-risk regions are presumed to have even higher levels of colonisation.
The European migrant crisis began in 2015 when over 1.2 million first-time asylum seekers (this group includes both refugees and migrants) applied for international protection in European Union countries [19,20]. Finland received 32,476 applications for asylum in 2015, an almost 10-fold increase on previous years [21].
Like international travel, migration may contribute substantially to the spread of AMR [22][23][24][25][26]. Refugees and migrants mostly come from countries with considerably higher rates of MDR bacteria than Finland and, moreover, they may have journeyed through other high-prevalence regions [1]. Accurate data on colonisation rates in this population are required to estimate transmission risk and to prepare infection control guidelines for hospitals. This retrospective study investigates the prevalence of various MDR bacteria among asylum seekers and refugees hospitalised in Finland, and seeks risk factors that can be used to identify those at highest risk of colonisation.

Selection of participants
Helsinki University Hospital (HUCH) provides secondary and tertiary care for the 1.6 million inhabitants of southern Finland. During the study period, from January 2010 to August 2017, our hospital's infection control guidelines stated that all asylum seekers and refugees admitted to hospitals should be screened at entry for meticillin-resistant Staphylococcus aureus (MRSA), vancomycin-resistant Enterococcus (VRE), extended-spectrum beta-lactamase-producing Enterobacteriaceae (ESBL-PE), carbapenemase-producing Enterobacteriaceae (CPE), multiresistant Acinetobacter baumannii (MRAB) and multiresistant Pseudomonas aeruginosa (MRPA). The same screening guidelines were applied to patients who had been hospitalised outside the Nordic countries during the previous 12 months before admission to our hospital.
Using the HUCH infectious diseases database, SAI, we compiled a list of patients who had been sampled for both MRSA and multiresistant Gram-negative (MRGN) bacteria at hospital admission. Among these, we selected those with a non-Finnish/non-Swedish name and, after screening their patient charts, included only those who were asylum seekers or refugees ( Figure 1). According to the Finnish Medical Research Act, review by an ethics committee is only required for research involving an intervention. The study protocol was approved by the research board of the Inflammation Center, Helsinki University Hospital, Finland.

Collection of patient data
For background information, we collected data from the patient records on sex, age, country of origin, date of arrival in Finland, prior hospitalisation and surgery (as recorded by the clinician), and determined Charlson Comorbidity Index (CCI) [27] for each subject. In addition, we collected data covering the results of bacterial cultures (blood, urine, stool), reason for admission, clinical diagnosis (ICD-10) [28] at discharge, MDR bacterial infections identified, and deaths. For further analysis, patients were grouped by geographical region according to their country of origin. Here we applied a classification ( Figure 2) modified from United Nations geoscheme (Europe, North Africa and Middle East, sub-Saharan Africa, Asia, other) [29].

Microbiological methods
According to the hospital guidelines, swabs for screening patients for MDR bacteria are collected as follows: MRSA samples are taken each with a separate swab from the nostrils (one swab for both), pharynx and rectum or perineum. MRGN bacteria samples are collected as rectal swabs. In addition, swabs are taken from wound infections when applicable. The screening comprises two sets of samples, and where possible the samples are to be collected on consecutive days.
While some minor modifications to the routine laboratory practices took place during the study period, at the time of the last sampling, MDR bacterial analyses were carried out as follows: MRSA was screened by overnight enrichment on eMRSA broth (Copan Italia, Brescia, Italy) or selective in-house MRSA enrichment broth [30] followed by culture on CHROMagar MRSA (CHROMagar, Paris, France), and confirmed with S. aureus-specific nuclease and mecA gene qPCR [30]. VRE were screened by enrichment Enterococcosel Broth (BBL, Cockeysville, MD, United States of America (USA)) and followed by culture on CHROMagar VRE media. Positive findings were confirmed by in-house PCR as described by Suppola et al. [31].

Figure 1
Flowchart showing multidrug-resistant bacteria found in samples from asylum seekers and refugees admitted to Helsinki University Hospital, Finland, January 2010 to August 2017 (n = 447) For nine patients, country of origin was not known. One patient was from Mexico (not shown in the map). ESBL-PE and CPE were analysed by plating directly on CHROMagar ESBL and CHROMagar KPC or CHROMagar mSuperCARBA, respectively. ESBL-PE species identification was confirmed by MALDI-TOF (Vitek-MS, bioMérieux, Marcy l'Étoile, France) and resistance by standard EUCAST method [32]. CPE species were confirmed with in-house carbapenemase gene PCR. MDR-P. aeruginosa and MDR-A. baumannii were screened from ESBL and KPC SuperCARBA plates. Cultures were tested by MALDI-TOF for species identification. Isolates resistant to meropenem for Acinetobacter, and both meropenem and ceftazidime for Pseudomonas, were analysed by PCR for carbapenemase genes as previously described [33].

Statistics
Data were entered on Microsoft Excel 2013 spreadsheets, and statistical analyses were conducted using SPSS 24.0.0.0 software (IBM Corp., Armonk, NY, USA). In univariate analyses for categorical variables, chi-squared test, Fisher's exact test or binary logistic regression analysis was applied. For continuous variables, we used binary logistic regression. Chi-squared test and Fisher's exact test were two-sided. Variables for the multivariable model were selected using the p value limit of 0.2 in the univariate model. Time between arrival and first sample was not known for all cases; missing values were taken into account by multiple imputations, assuming that data were missing at random. In the multivariable model, we used backward selection with Akaike information criteria (AIC) so as to choose the most informative explanatory variables for the final model. From several highly correlated variables, only one was included.

Subject characteristics
According to the HUCH infectious diseases database SAI, 6,423 patients were screened for both MRSA and MRGN bacteria at admission between 1 January 2010 and 23 August 2017. A total of 2,319 patients with a non-Finnish/non-Swedish name were selected and their patient records screened ( Figure 1). The final study population included 447 patients with refugee or asylum seeker status stated in their patient records. The increase in the number of refugees and asylum seekers admitted during the study period was evident: the vast majority (86.8%) were hospitalised after the beginning of 2015. As shown by the demographic data presented in Tables 1 and 2 Figure 2).

Risk factor analysis of meticillinresistant Staphylococcus aureus colonisation
In multivariable analysis, geographical region and prior invasive procedure outside the Nordic countries were found to be independent risk factors of MRSA colonisation. No other risk factors were identified. (Table 6)

Risk factor analysis of extended-spectrum beta-lactamase-producing Enterobacteriaceae colonisation
We identified by univariate analysis the following factors as predisposing to ESBL-PE: young age, Table 4 The number of carriers of multidrug-resistant bacteria among asylum seekers and refugees admitted to Helsinki University Hospital, as presented by five most common countries of origin a , Finland, January 2010-August 2017 (n = 447 patients)    The spa-types of some MRSA samples were not available.

Table 5
The results of univariate and multivariable risk factor analyses of extended-spectrum beta-lactamase-producing Enterobacteriaceae carriage among asylum seekers and refugees admitted to Helsinki University Hospital, Finland, January 2010-August 2017 (n = 447) geographical region, short time from arrival to first sample, prior hospitalisation, prior invasive procedure abroad, and prior ICU care ( Table 5). The final multivariable analysis revealed geographical region of origin, age under 6 years, short time from arrival to first sample and prior hospitalisation abroad as independent risk factors for colonisation. As for the geographical region of origin, the ESBL-PE colonisation rates were significantly higher among North African, Middle Eastern and Asian patients than Europeans (Table 5).
Of factors with p value under 0.2 in univariate analysis, male sex, prior invasive procedure abroad, prior ICU care, and general health (CCI score) were eliminated from the final model on the basis of the AIC.

Discussion
Data remain scarce on AMR carriage rates in refugees and migrants [24]. The present investigation reveals MDR bacterial colonisation and a high prevalence of MRSA of 21.3% among asylum seekers and refugees seeking healthcare at a tertiary hospital in Finland.
Although focusing on AMR, we also collected general data on refugees and migrants and causes of their hospitalisation. Although 79% of the asylum seekers who have arrived in Finland since 2015 have been male [34], in our data no clear sex difference in hospital admission was seen; the discrepancy can probably be explained by our high number of obstetric hospitalisations. Overall, our patients were in good general health; the greatest number of visits (29.8%) were reported for the specialty of obstetrics, while infectious diseases accounted for fewer than one in five cases. Respiratory, skin, and urinary tract infections were the most common causes among infectious diseases. Obviously, these are also common causes of hospitalisation among the host population. The proportion of minors corresponded to that of refugee and migrant minors arriving in Finland (24.9%) [35].

Level of colonisation by any multidrugresistant bacteria
Nearly half of the patients were colonised by MDR bacteria, a finding consistent with recent research focusing on refugees admitted to hospitals [36][37][38]. in two studies with both adult and paediatric patients [37,38]. The various countries of birth may account for the differences: in the previous studies, most of the patients were Syrians and Afghans, whereas in ours, nearly half were Iraqis. The high overall rate of MDR bacterial colonisation among refugees and asylum seekers justifies infection control measures, according with rates of 55.2% demonstrated in another group routinely subjected to isolation at hospitals in lowprevalence countries, namely travellers with a recent history of hospitalisation in the (sub)tropics [39].
As we expected, the clinical ESBL-PE infections generally proved to be UTIs; MDR bacterial wound infections were mostly attributed to MRSA. Data on the frequency of MDR bacterial carriers developing an infection with the MDR bacterial strain are scarce. Among returning travellers treated on an infectious and tropical diseases ward in France from 2012 to 2013, Epelboin et al. [40] found that one in five MRGN bacterial carriers identified (21.7%, 5/23) also had a clinical MRGN bacterial infection, albeit in a very small sample. The difference between our rates and those of Epelboin et al. (5.0% vs 21.7%, respectively) can probably be explained by two points related to the selection of population. Firstly, our data included a high proportion of healthy obstetric patients with no MDR bacterial infections. Secondly, in the investigation by Epelboin all patients had a clinical infection, while ours looked retrospectively at all patients, admitted for any reason, who at admission to a tertiary hospital had been sampled for both MRSA and MRGN bacteria, and whom we could identify as asylum seekers and refugees from their patient records. Indeed, of our patients who had infectious disease as primary diagnosis, 10.8% had an MDR bacterial infection. In our previous study looking at 1,122 Finnish patients hospitalised abroad for diverse reasons, we found that 11.4% (38/333) of those colonised with MDR bacteria had an infection with the same MDR bacteria [39].

Colonisation by MRSA
A major finding emerging from our investigation was that MRSA carriage in asylum seekers and refugees is considerably more common than expected. This proportion significantly exceeds those of German studies that have reported rates from 5.6% to 9.8% (8/143 and 32/325) [37,41], and rates of 15.7% recorded at Swiss asylum seeker reception centres [42]. MRSA carriage also proved more prevalent in the study population than among regular travellers hospitalised in the (sub)tropics: Khawaja  An analysis of spa-types revealed that the MRSA strains differ from those most commonly seen in clinical samples in Finland (t008, t172, t067) [46]. Clear differences were found between those with different countries of origin, suggesting that the strains had not originated in Finland but that refugees had been MRSA carriers on arrival. The median time from arrival in Finland to first sample was almost twice as long among MRSA carriers than among non-carriers. A tendency was seen for MRSA rates to grow as their stay in Finland lengthened, but the difference was not statistically significant (Table 6).

Colonisation by multiresistant bacteria
The rising number of CPEs has aroused great concern in Europe [47]. CPE findings have been reported for 0-2.1% of refugee patients [37,41,48]. Likewise, our study found the CPE rates to be low. Of course, even at these levels, they exceed the background rate in Finland [49]; special attention is warranted, since they may rise further, running parallel to increasing AMR prevalence in the various countries of origin.
As for MRAB and MRPA colonisation, two cases of each were identified; none with VRE were found, a valuable piece of information for professionals planning infection control measures in hospitals.
ESBL-PE were the most commonly recorded MDR bacteria. Detected among one third of the refugees and migrants, the rates remained lower than in two studies by Reinheimer et al. [37,38] [50]. The colonisation rates among refugees resembled those reported for regular travellers (20-70%) [51,52], the figures probably reflecting country-related background colonisation rates. Traveller studies show significant differences depending on destination [51,52]; the highest numbers are seen among visitors to major risk regions such as the Indian subcontinent.
It appears, however, that the initial ESBL-PE rates of asylum seekers and refugees on arrival in Finland may actually have been higher than recorded here: those with longer time since arrival had a lower ESBL-PE carriage frequency than those sampled soon after immigration. Indeed, recent follow-up studies show that travellers' ESBL-PE carriage tends to be transient and detectable only for a few months after return [15,53]. While the acquisition rates reported in traveller studies are based on samples collected soon after travellers' return, in the present study the median time from arrival to sampling was approximately 8 weeks. Further research is needed into carriage duration among refugees and migrants or other people with a recent history of long-term exposure to MDR bacteria.
A substantial proportion of our ESBL-PE isolates proved co-resistant to levofloxacin, tobramycin or cotrimoxazole, which accords with the results of studies exploring colonisation among travellers [4,9,11,12]. Co-resistance to levofloxacin correlated with resistance to tobramycin but not co-trimoxazole. This finding may be related to a genetic linkage between the resistance mechanisms of the first two [54,55].

Risk factor analysis
To identify potential risk factors of MRSA and ESBL-PE colonisation, we conducted univariate and multivariable analysis of the items derived from the patient records. To our knowledge, until now, risk factor analyses have not been included in refugee/migrant studies [36][37][38]41,48,50].
As independent risk factors of MRSA colonisation, we recognised geographical region and prior invasive procedure outside the Nordic countries. The latter accords with previous studies showing prior healthcare contact to be a risk factor of MRSA colonisation at hospital admission [56,57]. In our research, the highest risk of MRSA colonisation was seen among patients from North Africa and Middle East, which is in line with an investigation by Stenhem et al. analysing imported MRSA cases in Sweden from 2000 to 2003 [58]. Some of the other risk factors of MRSA colonisation such as prior infections or antibiotic treatments, occupation (e.g. healthcare worker) or contact with livestock could not be covered, due to the retrospective nature of our study [57].
For ESBL-PE colonisation, the final multivariable analyses revealed as independent risk factors geographical region, young age (< 6 years old), short time from arrival to first sample, and prior hospitalisation outside the Nordic countries. Geographical region has been identified in virtually all traveller studies analysing ESBL-PE risk factors [5,[12][13][14]16,52]. The highest risk has been linked with Asia and the Middle East [5][6][7][9][10][11][12][13][14][15][16][17][18], a finding that agrees with our results. Likewise, prior hospitalisation has been established as a risk factor for MDR bacterial carriage among travellers [59]. One previous result contradictory to ours showed a correlation between young age and lower rates of ESBL-PE colonisation [13]. Indeed, refugee children and travelling minors cannot be regarded as comparable: tourist children's exposure to food/drink contaminated with intestinal microbes is of shorter duration, and, moreover, trying to avoid diarrhoea, their parents probably select less risky food for them.
Traveller studies have also identified a number of other risk factors, such as occurrence of diarrhoea [5,6,9,10,[13][14][15] and use of antibiotics [6,13,[15][16][17]. Unfortunately, these predisposing factors could not be included in our analyses since such data could not be consistently drawn from our patient records.

Limitations
Due to the retrospective design of our study, the data were limited to those available in patient records. Some relevant factors such as antibiotic use could not be analysed. Information concerning the itineraries of the refugees and asylum seekers was lacking, and dates of arrival in Finland were recorded for only 63.3% of the patients. The research was conducted in a tertiary hospital which is reflected in the selection of the patients. Females were over-represented in our study population (53.0%), probably because of the large proportion of pregnancy-related hospital visits. After the beginning of 2015 only 21.2% of all asylum seekers arriving in Finland were female [60].

Conclusions
Our study shows considerable carriage rates for MDR bacteria among refugees and asylum seekers admitted to a tertiary hospital in Finland. The data suggest that these patients should be considered a risk group that requires both screening of MDR bacteria and infection control measures at entry to hospitals in low-prevalence countries. In particular the refugee and migrant population's considerable MRSA colonisation rate warrants attention in healthcare settings.
Note *This designation is without prejudice to positions on status, and is in line with United Nations Security Council Resolution 1244/99 and the International Court of Justice Opinion on the Kosovo Declaration of Independence.