Major discrepancy between factual antibiotic resistance and consumption in South of France: analysis of 539,037 bacterial strains

Ousmane Diallo ,Raymond Ruimy,Hervé Chaudet,Didier Raoult,Philippe Halfon,Jean-Pierre Arzouni,Grégory Dubourg ,E Delaunay ,Béatrice Comte,Philippe Seyral,Arnaud François,Stéphanie Joubert,Joseph Carvajal,Sophie Burignat,François Parisot,Pierre‐Yves Levy ,Philippe Stolidi,Sabine Camiade,Claude Bosi,Jean‐Paul Casalta ,Jean-Christophe Achiardy,Philippe Colson,P Roussellier ,Patrick Brunet,Pierre Hance,Sandra Meyer,Christophe Maruéjouls ,Sophie Alexandra Baron,Jean‐Marc Rolain

doi:10.1038/s41598-020-75158-7

Abstract

The burden of antibiotic resistance is currently estimated by mathematical modeling, without real count of resistance to key antibiotics. Here we report the real rate of resistance to key antibiotics in bacteria isolated from humans during a 5 years period in a large area in southeast in France. We conducted a retrospective study on antibiotic susceptibility of 539,107 clinical strains isolated from hospital and private laboratories in south of France area from January 2014 to January 2019. The resistance rate to key antibiotics as well as the proportion of bacteria classified as Difficult-to-Treat (DTR) were determined and compared with the Mann–Whitney U test, the χ2 test or the Fisher’s exact test. Among 539,037 isolates, we did not observe any significant increase or decrease in resistance to key antibiotics for 5 years, (oxacillin resistance in Staphylococcus aureus, carbapenem resistance in enterobacteria and Pseudomonas aeruginosa and 3rd generation cephalosporin resistance in Escherichia coli and Klebsiella pneumoniae). However, we observed a significant decrease in imipenem resistance for Acinetobacter baumannii from 2014 to 2018 (24.19–12.27%; p = 0.005) and a significant increase of ceftriaxone resistance in Klebsiella pneumoniae (9.9–24.03%; p = 0.001) and Enterobacter cloacae (24.05–42.05%; p = 0.004). Of these 539,037 isolates, 1604 (0.3%) had a DTR phenotype. Over a 5-year period, we did not observe a burden of AR in our region despite a high rate of antibiotic consumption in our country. These results highlight the need for implementation of real-time AR surveillance systems which use factual data.

Highlights

The burden of antibiotic resistance is currently estimated by mathematical modeling, without real count of resistance to key antibiotics
A recent survey conducted in 251 intensive care units (ICU) in France estimated about 45 deaths attributable to antibiotic resistance without alternative treatment over a 10-years period contradicting the prediction based on mathematical models[6]
A total of 711,031 strains were isolated in all laboratories of the PACASurvE network, including 539,037 that belong to the 15 most common bacteria plus ESKAPE and had at least one key antibiotic tested (Fig. 1)

Summary

Introduction

The burden of antibiotic resistance is currently estimated by mathematical modeling, without real count of resistance to key antibiotics. Kadri et al suggested a new definition as “difficult-to-treat” (DTR) bacterial infections i.e. infections due to bacteria that are in vitro resistant to all antibiotics tested in 3 classes of first-line antibiotics (β-lactams, carbapenems and fluoroquinolones)[4]. According to this definition, they have reported only 1% of Gram-negative bacteria (GNB) classified as DTR in a large series of bacterial isolates from 173 hospitals in the USA over a 3 years period. A recent survey conducted in 251 intensive care units (ICU) in France estimated about 45 deaths attributable to antibiotic resistance without alternative treatment over a 10-years period contradicting the prediction based on mathematical models[6]

Methods

Results

Conclusion