Burden of infections related to antibiotic resistant bacteria in France in 2015: Results from the French Hospital discharge database

Introduction The massive use of antibiotics has led to a significant increase of bacterial resistance to these drugs, making infections more difficult to treat; antimicrobial resistance has become a major public health issue. Few studies have assessed the total burden of antimicrobial resistance, and there is a paucity of comprehensive population-based data. In 2014, because of the impact of antimicrobial resistance on public health, the World Health Organization revised its codification of infections and antimicrobial resistance in the 10th revision of the international classification of diseases (ICD-10), which is used in the French Hospital Discharge Database (PMSI). In France, special attention was given to resistance causing therapeutic difficulties leading to additional specific codes in ICD-10. These improvements are an opportunity to use national healthcare database in order to reduce assumptions and to base burden estimations on actual nationwide data. This study aims to estimate the overall situation of antimicrobial resistance in France based on the PMSI. Methods Study population was incident hospitalizations with an acute infection caused by Streptococcus , Staphylococcus , Enterobacteriaceae or another Gram negative bacteria in 2015. Hospitalizations with infections were identified in the PMSI database from established lists of ICD-10 corresponding to infections, microorganisms and resistances codes. Stays (> 1 day) with an admission between January 1 and December 31, were selected when at least one code of infection was present in principal, related, or significant associated diagnosis. Patients hospitalized with infections in the 12 months preceding the admission were excluded. Using medical-administrative databases for a different objective from their original purpose leads to difficulties, in particular in the context of infectious disease: diagnoses are not related, and some information could be unknown. First, stays were usable only if the infection, the microorganism and the potential resistance could be related. For these stays, bacteria were considered susceptible when no resistance was recorded. Second, if no bacterial species was noted (i.e. because the physician considered irrelevant/unnecessary to obtain microbiological sample or because the culture results may be negative, etc.), the causal agent and its resistance status were extrapolated from the stays with usable information, according to sex, age (7 categories) and site of infection (11 categories). Resulting stays were described, characteristics of infections with resistant bacteria analyzed and the burden of resistance in France estimated. Results In 2015, 979,902 incident hospitalizations with an acute infection were selected. Bacteria species was recorded in 30% of the cases. After extrapolation, it was estimated that bacterial resistance was associated with 111,177 incident infections in France, corresponding to 11.3% of all incident infections. These infections were mostly of the urinary and genital tracts (25.2%), lower respiratory tract (21.1%), gastrointestinal and abdominal (20.7%) and skin and soft tissues infections (14.2%). Among resistant bacteria, the most frequently detected were ESBL-producing Escherichia coli (28.6%), methicillin resistant Staphylococcus aureus (MRSA, 19.9%), and Klebsiella pneumoniae ESBL (5.2%). MRSA and ESBL-producing Enterobacteriaceae (ESBL-PE) were estimated to cause 22,081 and 39,600 infections annually, respectively. Conclusions This study is the first study to evaluate the burden of resistance from a national comprehensive database. Resistance rates among blood infection estimated from the sample (ESBL-PE, MRSA and VRE) were in accordance with those provided by EARS-net, validating the extrapolation. This study also confirms the importance of ESBL-PE and MRSA in the overall burden of resistance.