Geographical differences of carbapenem non-susceptible Enterobacterales and Acinetobacter spp. in Germany from 2017 to 2019

Gram-negative bacteria that are non-susceptible to carbapenems pose a significant threat to patients and healthcare systems and are a major public health issue [1]. In the priority pathogens list of the World Health Organization (WHO) carbapenem-resistant Acinetobacter spp. and Enterobacterales are classified as the most critical group of bacteria that pose the greatest threat to human health [2].

Whereas Acinetobacter spp. is mainly disseminated via environmental contamination, Enterobacterales are mainly transmitted person-to-person, i.e. via the hands of health care workers [3]. These bacteria occur especially in hospital settings due to antibiotic use and co-morbidities of the patients, but new studies suggest that also non-healthcare associated transmission outside health care settings occur [4‐6]. Identifying risk groups for carriage of carbapenem non-susceptible Acinetobacter spp. (CRA) and Enterobacterales (CRE) might guide local public health authorities to implement preventive strategies to prevent further spread. Identifying differences in geographical distribution of CRA and CRE helps to adapt control measures to local situations.

Carbapenem non-susceptibility varies between countries within the European Union and between bacterial species [1, 7]. In many European countries, these resistant bacteria have increased since 2012 and especially high levels are reported for Klebsiella pneumoniae and Acinetobacter spp. [7‐9].

Until 2016, surveillance data on antimicrobial resistance in Germany were only available from the German Antimicrobial Resistance Surveillance (ARS) System which collects routine data of antimicrobial susceptibility testing from voluntarily participating German laboratories [10]. Based on data from ARS, carbapenem non-susceptibility in K. pneumoniae was present in 0.5% of all clinical isolates in 2016 [11]. The percentage of invasive carbapenem resistant Acinetobacter spp. was 4.5% in 2018 [7]. Although Germany has a low proportion of carbapenem resistant bacteria, it has been slowly increasing over the past years [11].

In 2016, a new comprehensive surveillance for CRA and CRE was implemented in Germany to assess the epidemiological situation based on nationwide data [12, 13]. The mandatory notification requirements include infection as well as colonisation. These surveillance data focus primarily on the epidemiology and do not aim to identify what was driving the carbapenem non-susceptibility (i.e. causing genes, clonal or plasmid mediated dissemination). The early identification and notification of cases to local public health authorities (LPHA) aims to facilitate rapid implementation of control measures to avoid further spread.

The aim of our study was to assess the epidemiology of CRA and CRE and to identify risk groups for carriage of CRA and CRE. A special focus was laid on geographical differences from 2017 to 2019 in Germany to support public health authorities to target infection control measures locally and nationally and to determine the influence of hospitals on CRA and CRE notifications.

In total, 2278 carbapenem non-susceptible Acinetobacter spp. (CRA) and 12,282 carbapenem non-susceptible Enterobacterales (CRE) were notified in Germany from 2017 to 2019 (Fig. 1). The number of CRA notifications was stable over this time period, whereas the number of CRE notifications increased from 2017 to 2019. Of 2102 CRA and 11,100 CRE with information, carbapenem non-susceptibility was determined primarily with culture-based antimicrobial susceptibility testing in 97.9% (n = 2057 and n = 10,869, respectively).

We described the epidemiology of carbapenem non-susceptible Acinetobacter spp. (CRA) and Enterobacterales (CRE) that were notified in Germany from 2017 to 2019.

The notification incidence was 0.95, 0.94 and 0.86 for CRA and 4.23, 4.80 and 5.72 for CRE in 2017, 2018 and 2019, respectively. Whereas the notification incidence of CRA decreased over the study period, the notification incidence of CRE increased by 35%. This is in accordance with the finding that the proportion of carbapenem non-susceptible K. pneumoniae increased in Germany in the last years [11]. Comparison with data from other states is difficult as surveillance systems, test strategies, access to healthcare and clinical routine differ substantially. A study in the United States found an annual notification incidence rate of CRE infections of 2.93 per 100,000 population in 2012, but included certain infections only and did not include colonisations [24].

Median age of notified CRA cases was 66 years and of CRE cases 68 years. The age and gender structure of notified cases reflect that of hospital patients in Germany [25]. High CRA and CRE incidences were found in patients over 70 years and for CRE in infants under one year of age. Among the older age groups, men were more affected. This is in accordance with the literature and emphasizes that hospitalisation might be the main driver of CRA and CRE [26]. The notification incidence of CRE in infants below one year of age might also be elevated because of the recommended screening of newborns for certain bacteria in neonatal intensive care [27]. This is supported by the finding that 86% of cases below one year of age were reported to be colonised. In all age groups, most cases were colonised with CRA or CRE, but about a third of cases had an infection. Death due to the notified illness should be interpreted with caution as it is unknown if and how LPHA follow-up on cases and underreporting with huge regional variation is assumed.

The most common reported bacteria were Klebsiella pneumoniae and Acinetobacter baumannii-complex. The most common reported carbapenemase in CRE was OXA-48 and in CRA OXA-23. Both results are in accordance with the findings of the National Reference Centre for Multidrug-resistant Gram-negative bacteria (NRC) in Germany [28]. The NRC receives isolates sent voluntarily from diagnostic laboratories to confirm carbapenem non-susceptibility.

The notification incidence of CRA and CRE varied geographically with the highest notification incidence in the federal states of Berlin, Hamburg and Hesse. The geographical variation could be due to different screening practices in health care facilities, a better reporting compliance in these federal states or because of a real higher occurrence of these bacteria. German hospitals are encouraged to standardise their screening practices according to the recommendations of the Commission for Hospital Hygiene and Infection Prevention (KRINKO), but screening practices might differ between hospitals [29]. Enhanced surveillance due to active surveillance in outbreak situations could also increase the incidence. Hamburg and Berlin are two city states with big university hospitals. Berlin published outbreaks with extended-spectrum β-lactamase producing Enterobacterales and CRE in 2016 [30, 31]. Hesse had implemented a mandatory notification for carbapenem non-susceptible Gram-negative bacteria already in 2011, therefore it seems possible that reporting compliance might be better established than in other federal states [32]. Within the notification data, the district (unit of one responsible LPHA) is mainly allocated according to the patient’s permanent address and does not necessarily represent the place of exposure. A local study showed that almost 60% of cases notified from the laboratory to the LPHA in the district of Frankfurt (Hesse) were not residents in Frankfurt [33]. These cases might have been forwarded to national level by the LPHA responsible for the district of residence of the patient and would therefore be allocated to another district than the one where the exposure occurred. We did not include place of exposure in our analysis, as only less than 1% of cases (n = 95/14,560) reported place of exposure.

Mean yearly notification incidence per 1000 patient-days was 0.01 for CRA and 0.03 for CRE. In international studies the notification incidence of carbapenem non-susceptible Gram-negative bacteria varied from 0.007 to 35.2 per 1000 patient-days, but these studies included Pseudomonas aeruginosa and high risk areas only, but often did not include colonisation [34‐37].

As transmission of CRA and CRE occurs mostly within healthcare settings, one would assume that the number of notifications is highest in areas with a high hospital density and therefore more hospitalised patients [4]. We could support this finding; hospitals and hospital beds increased the notification incidence of CRA and CRE. This is in line with our finding that more often hospitalised age groups show a higher notification incidence. Whereas the notification incidence of CRA seems to be linked to the number of hospitals and hospital beds more generally, the incidence of CRE seems to be linked to university hospitals. This is in accordance with an analysis of carbapenem non-susceptible Klebsiella pneumoniae in Germany that identified highly specialised hospitals and intensive care units as risk factors [11]. The association between hospital type and the incidence of CRA notifications was not significant for year-wise but for cumulative analyses. Therefore, the small sample size in university hospitals and CRA cases might have led to a false non-significant result. Additionally, intensified infection prevention measures and antimicrobial stewardship programs in university hospitals impact the transmission within the hospital and might decrease the notification incidence. Nonetheless, the impact of hospitals might be underestimated because place of exposure could not be analysed. This could impact especially university hospitals as those specialised units have large catchment areas and cases might be allocated to their district of residence. Complicated and difficult to treat cases are often referred to highly specialized and university hospitals. These cases often have a prolonged healthcare journey and were more exposed to healthcare settings. Furthermore, specialized and university hospitals might have a more intensive screening strategy in place. The directory of hospitals includes every hospital in Germany irrespective of number of beds. This finding might also lead to an underestimation of the association of the notification incidence and general hospitals. Geographical differences in health care seeking behaviour might impact the number of hospitalised patients and the notification incidence of CRA and CRE, although a significant association between the social deprivation index and the proportion of carbapenem non-susceptible Klebsiella pneumoniae in Germany could not be shown [11]. Hospitalisations in Germany differ regionally due to demographic and social structure, urbanisation, population density, and health status and supply of hospital beds impacts hospital utilisation (supply-sensitive care) [38]. This should be investigated further. It is known that CRE are also spread in the community, but data on community-acquired CRE are scarce [39, 40]. As health care structure also reflects social structure and population density, this may especially influence the notification incidence of CRE. As university hospitals might have intensified surveillance programs, the strong association of the number of university hospitals and the notification incidence of CRE might reflect a much higher incidence of CRE within the community. The national recommendation which risk groups should be screened in hospital does not differ for CRA and CRE. Bias due to systematic differences in targeted risk groups of screening for these two bacterial groups is therefore not assumed. Nonetheless, screening practices and the association between highly specialised clinics and the notification incidence of CRA and CRE should be investigated further on hospital level. The analysis should be repeated with notification data from the following years to decide whether the geographical differences of CRE and CRA cases remain stable. Whether other highly specialised hospitals show a similar association as university hospitals, could not be analysed with the available data.

There are several limitations to this study. Completeness of information in the surveillance system is crucial to verify, analyse and interpret notification data. The initial notification from laboratories to LPHA contains usually demographic as well as laboratory information. Additional data, such as clinical information, have to be investigated by the LPHA and the results of specific laboratory tests, such as the test for carbapenemases, is often added to the initial notification at a later time. Although most variables were 75% or more complete, there were missing values in particular with regards to infection status and test for carbapenemases. This additional information was often incomplete. This limits the generalisability of these study results. LPHA as well as hospital and laboratory staff members are encouraged to share information in a timely manner and work closely together. The information investigated by the LPHA should be entered into the notification software and forwarded to national level. Several notifications might have been forwarded for one patient so one case does not necessarily represent one patient. The number of CRA and CRE cases per inpatients and patient-days as well as local risk factors should be investigated and monitored on hospital level. This will be important to assess and support local preventive control measures.

This is the first report on CRA and CRE notifications in Germany since mandatory notification was established. The notification incidence of CRA and CRE varied geographically and was correlated with the type and number of hospitals and hospital beds per region.The results support the hospital as main driver of CRE and CRA. Whereas CRA seems to be correlated more generally with hospital density, CRE seems especially to be impacted by university hospitals. Both vulnerable groups, the elderly and the infants below one year of age, should be especially targeted when implementing infection control measures. In areas with a high incidence targeted preventive strategies and early control measures should be reviewed and adapted to prevent further spread. Geographical differences should be investigated further to exclude variation in testing and/or notification procedures. Furthermore, the impact of underestimation of cases should be investigated. We also encourage molecular surveillance of CRA and CRE to better understand and prevent transmission.