The epidemiology of carbapenem resistance in Acinetobacter baumannii complex in Germany (2014–2018): an analysis of data from the national Antimicrobial Resistance Surveillance system

The primary outcome is the proportion of carbapenem-resistant A. baumannii complex isolates among all A. baumannii complex isolates tested for carbapenem resistance. Additionally, we analysed factors that are associated with the likelihood of carbapenem resistance in A. baumannii complex isolates (see “Study variables” section below).

The secondary outcomes of interest are (1) the proportion of carbapenem-resistant non-baumannii complex Acinetobacter isolates among all non-baumannii complex Acinetobacter isolates tested for carbapenem resistance and (2) the proportional distribution of A. baumannii complex species among all Acinetobacter species in different clinical specimen materials.

We performed a retrospective observational study using data from the German Antibiotic Resistance Surveillance (ARS) system between 2014 and 2018. Laboratories that voluntarily participate in the surveillance system submit routine clinical microbiological data to the Robert Koch Institute (RKI) [27]. These data include results from pathogen identification and antimicrobial susceptibility testing, as well as pseudonymised information on health care facilities and patient characteristics, such as care setting type, hospital ward, age, gender, specimen materials and the geographical location of patient care [28]. Forty-eight laboratories contributed to the ARS system in 2018, which includes data from around 13% of all hospitals (389 out of ~ 3000) and around 16% of all outpatient clinics (16,016 out of ~ 100,000) in Germany [29].

Isolates obtained between 2014 and 2018 were selected for the primary analysis. We avoided including multiple isolates from individual patients’ single infection episodes by selecting only the patients’ first isolate per specimen and per quarter. Since this analysis focuses on clinical infections, isolates derived for multidrug-resistant pathogen screening were also excluded.

For the main analysis, we included all Acinetobacter species isolates that are part of the A. baumannii complex (i.e. A. baumannii, A. pittii, A. calcoaceticus and A. nosocomalis). In addition, we only selected isolates that were tested against at least one of the following carbapenems: meropenem, imipenem, and doripenem. Ertapenem was not selected since it has a considerably different pharmacology compared to the selected carbapenems [30]. We defined an isolate as carbapenem-resistant if it was tested as “resistant” (R) to at least one of the carbapenems of interest based on the standards used in the participating laboratories, such as the guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) or the Clinical and Laboratory Standards Institute (CLSI). In order to determine proportions of carbapenem resistance among non-baumannii complex Acinetobacter species, isolates from the respective species were included but the same selection criteria were otherwise used as described above.

For the analysis of the proportional distribution of A. baumannii complex and non-baumannii complex Acinetobacter species, only isolates that were identified at species level were included. Since complete species identification results were increasingly available after 2014, we only selected isolates from 2015 to 2018 for this analysis. In this analysis, we also included isolates without carbapenem testing.

Carbapenem resistance was analysed for the following variables: care setting type, year of sampling, region in Germany, age, gender and, clinical specimen materials. We grouped care setting type into the following categories: outpatient clinics (primary healthcare facilities), secondary care hospitals (offering basic and standard care), tertiary care hospitals (offering maximum care, such as university hospitals), specialist care hospitals (offering specialised care or private hospitals), prevention and rehabilitation care centres, and other hospitals (offering psychiatric, neurological and/or geriatric care only). Clinical specimen materials were categorised into the following groups: wound (swabs from wounds and abscesses), blood (blood cultures), urine (urine samples), upper respiratory materials (swabs from the upper respiratory tract), lower respiratory materials (bronchial lavage, bronchial secretion, bronchial rinse water, sputum and tracheal secretion), and other specimens that did not fit in the categories listed above (i.e. swabs, biopsy tissues, dialysate material, ejaculate, skin flakes, hair and nails, catheter, cerebrospinal fluid, puncture material, stool samples and non-specified materials). We grouped patient gender into female and male and categorised patient ages in six different groups (< 1, 1–19, 20–39, 40–59, 60–79, ≥ 80 years). Furthermore, to determine the isolate’s geographical origin (location of the healthcare facility), each of the federal states was categorised into one of five major regions in Germany: Northeast (Mecklenburg-West Pomerania, Brandenburg, Berlin, Saxony-Anhalt), Southeast (Bavaria, Saxony, Thuringia), Southwest (Hesse, Rhineland-Palatinate, Saarland, Baden-Wurttemberg), West (North Rhine-Westphalia) and Northwest (Bremen, Lower Saxony, Hamburg, Schleswig–Holstein). We considered all variables as categorical for statistical analyses, apart from year, which was treated as a continuous variable.

All statistical analyses were performed using R version 3.6.1 [31] and the “survey” package (version 3.37) [32]. We used percentages with 95% confidence intervals (95% CI) to describe proportions of carbapenem-resistant isolates among all isolates tested against carbapenems. Univariable and multivariable logistic regression models were performed to identify patient and healthcare-related risk factors that are associated with carbapenem resistance in isolates from patients with A. baumannii complex infections. All previously listed variables were included in the univariable and the multivariable logistic regression model. We accounted for clustering at facility level in the carbapenem resistance proportions calculations and the analysis of associations between carbapenem resistance and the selected variables.

In total, 43,948 isolates from 12,169 and 26,840 patient visits in outpatient clinics and hospitals, respectively, were included in the primary analysis (Table 1). The number of isolates collected increased from 2014 to 2017, reflecting the increasing coverage of the ARS database. Almost half (45.4%) of the analysed A. baumannii complex isolates were from female patients, while 36.1% were from male patients. For 18.4% of all isolates the patient gender was unknown. Most isolates were collected from people in older age groups (median: 69 years, IQR: 52–79). A. baumannii complex isolates derived most frequently from wounds (27.9%), urine (20.2%) and respiratory materials (18.3%) (Table 1). Among the 30,867 isolates from hospitals, 5428 (17.6%) were derived from intensive care units.

In all clinical specimens, 60% of all Acinetobacter species were identified as species of the A. baumannii complex (Additional file 1: Fig. 1A). A. baumannii sensu stricto accounted for two-thirds of all A. baumannii complex isolates (19,522 out of 30,051 [65%]). Among blood isolates, A. baumannii and species of the A. baumannii complex accounted for 26% and 44% of all Acinetobacter species, respectively (Additional file 1: Fig. 1B). In contrast, in lower respiratory materials, A. baumannii complex species (76%) and A. baumannii sensu stricto (54%) were the most frequent Acinetobacter species (Additional file 1: Fig. 1C).

During the study period (2014–2018), the mean proportion of CRABC in Germany was 4.2% (95% CI 3.0–6.0%). In 2018, the CRABC proportion was 3.5% (95% CI 2.5–4.7%). In contrast, only 0.9% (95% CI 0.8–1.0%) of non-baumannii complex Acinetobacter isolates showed carbapenem resistance between 2014 and 2018. In the same time period, proportions of carbapenem resistance also varied among species of the A. baumannii complex. While carbapenem resistance was more pronounced in A. baumannii sensu stricto (5.7% [95% CI 4.2–8.0%]) and A. nosocomialis (7.0% [95% CI 1.7–25.0%]), carbapenem resistance proportions were lower in A. pittii (0.8% [95% CI 0.4–1.0%]) and A. calcoaceticus (0.6% [95% CI 0.2–2.0%]).

Differences in carbapenem resistance patterns between isolates from patients treated in outpatient clinics and different types of hospital are presented in Fig. 1. Carbapenem resistance in A. baumannii complex isolates was lower among isolates from outpatient clinics (1.3% [95% CI 1.1–1.6%]) compared to all types of hospitals, with CRABC proportions ranging from 4.9% (95% CI 3.2–7.5%) in secondary care hospitals to 8.7% (95% CI 4.1–17.5%) in prevention and rehabilitation care centres (Fig. 1). Univariable and multivariable regression analyses confirmed that A. baumannii complex isolates from hospitals were more likely to be carbapenem-resistant than isolates from outpatient clinics (Table 2).

Between 2014 and 2018, proportions of carbapenem-resistant A. baumannii complex in Germany decreased from 7.6% (95% CI 4.4–12.7%) to 3.5% (95% CI 2.5–4.7%) (Fig. 2). This decreasing trend was also supported by the multivariable analysis (adjusted OR: 0.85 [95% CI 0.79–0.93, p ≤ 0.001]) (Table 2). An additional sensitivity analysis of the temporal trend was conducted on isolates (n = 22,876) from healthcare facilities (1700 outpatient clinics, 194 hospitals) that provided data continuously for the entire study period (2014–2018). This sensitivity analysis showed similar temporal trends: CRABC proportions decreased from 7.8% (95% CI 4.2–14.1%) in 2014 to 4.4% (95% CI 2.6–7.2%) in 2018.

In addition, we found similar temporal trends in both outpatient clinics and hospitals. In isolates from outpatient clinics, proportions of carbapenem resistance declined from 2.3% (95% CI 1.3–4.2%) in 2014 to 1.2% (95% CI 0.8–1.7%) in 2018 (adjusted OR: 0.80 [95% CI 0.66–0.97, p ≤ 0.001]) (Fig. 2). In hospitals, carbapenem-resistant proportions decreased from 9.6% (95% CI 5.4–16.5%) to 4.4% (95% CI 3.1–6.1%) during the same period (adjusted OR: 0.85 [95% CI 0.78–0.93, p ≤ 0.001]).

Mean CRABC proportions (2014–2018) varied between German regions, with higher CRABC proportions in the West and Northwest of Germany than the other regions (Fig. 3). The multivariable analysis confirmed that A. baumannii complex isolates from the Northeast, Southeast and Southwest were less likely to be resistant to carbapenems in comparison to isolates from the West (Table 2). Regional variations were most pronounced amongst hospitals, with CRABC proportions ranging between 2.4% (95% CI 1.6–3.5%) in the Southeast and 8.8% (95% CI 4.2–17.3%) in the Northwest (Additional file 1: Fig. 2). In contrast, no geographic variation was observed amongst outpatient clinics.

CRABC proportions were significantly lower in young patients between < 1 year (0.6% [95% CI 0.2–1.3%]) and 1–19 years (1.3% [95% CI 0.7–2.5%]) compared to adult patients between 20 and 79 years (20–39 years: 7.7% [95% CI 4.4–13.0%]; 40–59 years: 6.2% [95% CI 4.2–8.9%]; 60–79 years: 5.8% [95% CI 4.0–8.3%]). Patients older than 79 years showed lower proportions (1.9% [95% CI 1.3–2.7%]) than patients between 20 and 79 years. The lower likelihood of carbapenem resistance among patients aged 80 years and older compared to those between 40 and 59 years was also confirmed by the multivariable analysis (Table 2). Moreover, we identified male gender as an independent risk factor for carbapenem resistance in A. baumannii complex isolates (adjusted OR: 1.91 [95% CI 1.55–2.36, p ≤ 0.001]). Differences in proportions between genders were especially pronounced in patients between 20 and 39 years (Men: 14.6% [95% CI 8.6–23.6%] vs. Women: 2.5% [95% CI 1.3–4.5%]). In contrast, we found no variations in proportions between male and female patients in people younger than 20 years or for those 80 years and older (Fig. 4).

Our data show that A. baumannii complex isolates from lower respiratory samples (11.4% [95% CI 7.9–16.2%]) exhibited higher carbapenem resistance proportions than isolates from upper respiratory material (4.0% [95% CI 2.7–6.0%]), urine (2.3% [95% CI 1.6–3.3%]) and wound (2.8% [95% CI 2.0–3.9%]) (Fig. 5). This was also confirmed by the multivariable analysis (Table 2). Although our data indicate that blood isolates had higher CRABC proportions (8.1% [95% CI 4.4–14.6%]) than upper respiratory samples, the multivariable analysis did not indicate statistical significance (adjusted OR: 1.25 [95% CI 0.91–1.73, p = 0.162]).

To our knowledge, this study is the most comprehensive analysis of the current epidemiology and risk factors of carbapenem resistance in A. baumannii complex isolates in Germany. Based on data from the German ARS system, we analysed almost 44,000 clinical A. baumannii complex isolates from more than 39,000 patient visits in hospitals and outpatient clinics.

Our study is also subject to several limitations. Firstly, since no information on diagnoses is available in the ARS database, it can only be assumed that the clinical specimens represent infectious diseases. Although we excluded all isolates labelled as screening samples, it is also possible that some of the included isolates (e.g. swabs of upper respiratory materials) actually represent screening samples that were not assigned as such by the hospital or laboratory [51]. Secondly, because participation in the ARS system is voluntary, the coverage of healthcare facilities providing ARS data may differs across regions and may not be representative. However, numbers of isolates from the analysed German regions roughly reflect population sizes in these regions. Finally, no clinical information is available in the ARS system, such as use of medical devices and co-morbidities, which are known to be associated with carbapenem resistance in A. baumannii complex infections.