Assessing the health burden of infections with antibiotic-resistant bacteria in the EU/EEA, 2016-2020

Our study methods were based on the methodology to estimate the burden of infections with antibiotic-resistant bacteria from the Burden of Communicable Diseases in Europe (BCoDE)i, using data on bloodstream infections (BSIs) caused by selected antibiotic-resistant bacteria, as reported to the European Antimicrobial Resistance Surveillance Network (EARS-Net). We estimated the total incidence of infections by taking the annual number of BSIs with antibiotic-resistant bacteria, adjusted the data with the estimated population coverage reported to EARS-Net, converted the population coverage-corrected number of BSIs to other types of infections, and deducted the estimated number of secondary BSI. The conversion multipliers and reduction factors were derived from the ECDC point prevalence survey of 2016-2017 and from previous literature. We then used the estimated annual number of infections with disease models based on literature reviews and ran 10 000 iterations of Monte Carlo simulations to acquire the health burden estimates and their respective 95% uncertainty intervals (UIs).

Our main results include the annual number of infections with antibiotic-resistant bacteria, the number of attributable deaths, the number and rate of disability-adjusted life years (DALYs), and the age-group-specific DALY rates. We used the median estimates derived from the simulations, and reported the 95% uncertainty interval (UI) derived using 2.5% and 97.5% percentiles for the uncertainties. The trends were assessed using Poisson regression. The proportion of healthcare-associated infections was estimated by information derived from the literature and, when using existing data, defined as infections for which the onset of symptoms was on day three or later after the start of the current admission, or where the patient has been re-admitted less than 48 hours after a previous discharge or transfer from a healthcare facility.

We estimated that between 2016 and 2020, the annual number of cases of infections with the included bacterium– antibiotic resistance combinations in the EU/EEA ranged from 685 433 (95% UI 589 451 – 792 873) in 2016 to 865 767 (95% UI 742 802 – 1 003 591) in 2019, with an annual number of attributable deaths ranging from 30 730 (95% UI 26 935 – 34 836) in 2016 to 38 710 (95% UI 34 053 – 43 748) in 2019. When analysed as DALYs, the infections led to an annual health burden ranging from 909 488 (95% UI 813 858 – 1 013 060) in 2016 to 1 101 288 (95% UI 988 703 – 1 222 498). We estimated that 70.9% of cases of infections with antibiotic-resistant bacteria (95% confidence interval (CI) 68.2 – 74.0%) were healthcare-associated infections.

Between 2016 and 2020 in the EU/EEA, there were significantly increasing trends in the estimated number of infections (p < 0.001), attributable deaths (p < 0.001) and DALYs (p < 0.001) per 100 000 population due to antibiotic-resistant bacteria, although numbers decreased slightly from 2019 to 2020. There was a significant increasing trend in estimated number of infections for almost all bacterium–antibiotic resistance combinations. To note, at both EU/EEA and national level, the estimates had wide uncertainty intervals (UIs). The largest burden of disease was caused by third-generation cephalosporin-resistant Escherichia coli, followed by meticillin-resistant Staphylococcus aureus and third-generation cephalosporin-resistant Klebsiella pneumoniae. The total age-group-specific burden was the highest in infants and the elderly. Adjusted for population size, the overall burden of infections with antibiotic-resistant bacteria was estimated to be the highest in Greece, Italy and Romania.

This study confirms the considerable health burden of infections with antibiotic-resistant bacteria in the EU/EEA. It also highlights the increasing burden of these infections, with an exception in 2020 when the overall burden was estimated to decrease. Changes in annual burden estimates can be affected by changes in surveillance, or changes in healthcare practices, such as in 2020 when the COVID-19 pandemic put pressure on all health services in EU/EEA countries. Part of the decrease in 2020 can also be explained by measures taken to control the spread of COVID-19, including changes in infection prevention and control, and changes in the patient mix in hospitals due to the different hospitalisation practice during the pandemic.

Our estimates for the burden of antibiotic-resistant bacteria for 2016 remains similar to 2015 data, whilst the estimate has increased in the following years. Also, for individual bacterium–antibiotic resistance combinations, our results were similar with the previous burden analysis and other published sources for data. For example, the observed increase in the burden of vancomycin-resistant E. faecalis/E. faecium is in line with what is reported elsewhere. Also, as reported in EARS-Net data, most of the individual bacterium–carbapenem resistance combinations showed an increasing trend in 2016-2020.

The limitations of this study include the limitations in the data and the evidence used for estimating the burden of disease. For example, data on the age-specific effects of different infections with antibiotic-resistant bacteria remain relatively scarce. Also, whilst some countries have surveillance systems with good population coverage, the incidence of BSIs often remains a crude estimate if the population coverage is low and the representativeness is poor. Furthermore, the changes in the surveillance and the participation in surveillance in general, affect year-to-year comparisons, and our uncertainty estimates at EU/EEA level remain comparatively wide.

Our main strength is the well-established, internationally approved methodology and a strong collaboration with participating national networks. Updated studies on the burden of antibiotic resistance in the EU/EEA, including relevant updates on the evidence and methodology, can provide important evidence for policy formulation and evaluation.