Author(s): Marco R Oggioni [*] 5 , Leonardo Furi 1 , Joana R Coelho 2 , Jean-Yves Maillard 3 , José L Martínez 4
Biocides have been widely used for decades, but despite this widespread and increasing use, most bacterial and fungal species remain susceptible to biocides. However, decreased susceptibility of bacteria and fungi to biocides has been reported and occasionally linked to antibiotic resistance. These reports have raised concern about a risk of increasing antibiotic resistance connected to the use of disinfectants in the environment, hospitals and industry. In view of this growing concern and the new licensing requirements recently posed in Europe, protocols are urgently needed to allow risk assessments for the use of biocidal products. The workshop entitled 'Biocides: do they select for antimicrobial resistance?' opened the second International Conference on Antimicrobial Research (ICAR)  held on 21-23 November 2012 in Lisbon. The workshop and the meeting in general provided a strong conceptual and experimental framework within which to answer these questions.
We have to recognize that over the last four decades, biocides have been increasingly used in a number of consumer products, and yet our understanding of bacterial behavior when exposed to these agents is limited. Factors affecting a decrease in bacterial susceptibility to biocides have been described and can be split into those depending upon the biocide (knowledge of the concentration exponent and mechanism of action, i.e., cellular targets, are key), those depending upon the microorganisms (a number of cellular mechanisms conferring an increase in insusceptibility have been reported), and those depending on application, such as exposure parameters and formulation of the biocide [1-3] . Although the potential for bacteria to survive biocide exposure is real, there is a lack of harmonization in the test methodologies used to measure the extent of bacterial resistance or decrease in biocide susceptibility, with protocols used often lacking a realistic approach [1-3] . There is a need for the development of predictive methodologies, based on ex situ test principle. This would require generating appropriate baseline data from using harmonized protocols and a better understanding of the conditions of use of biocides to set up appropriate test parameters.
Methods for modeling of resistance data and risk assessment approaches were among the first presentations to be discussed at the meeting. In order to find new methodologies useful for the study of impact of biocide use, Joana R Coelho (Technical University of Lisbon, Lisbon, Portugal) proposed the application of different machine learning methods, namely decision trees validated with permutation tests and clustering techniques, to investigate a large dataset of 1600 clinical isolates of Staphylococcus aureus  . Results showed new associations between antibiotic resistance and reduced susceptibility to biocides, where the standard statistical approaches had failed to deal with the complexity of the data. More precisely, a high MIC for benzalkonium chloride and chlorhexidine, but not of other biocides, was identified as rule for the prediction of multidrug resistance (MDR) in staphylococci.
The molecular analysis of biocide-induced antibiotic resistance was introduced by José L Martínez (Centro Nacional de Biotecnologia-CSIC, Madrid,...