Use of global sensitivity analysis in quantitative microbial risk assessment: Application to the evaluation of a biological time temperature integrator as a quality and safety indicator for cold smoked salmon M. Ellouze a, b, c, * , J.-P. Gauchi b , J.-C. Augustin c a CRYOLOG S.A. Département R&D. 58, Boulevard Gustave Roch, 44261 Nantes, France b Unité MIA (UR341), INRA, Domaine de Vilvert, 78352 Jouy en Josas, France c Unité MASQ Microbiologie des Aliments e Sécurité et Qualité, Ecole Nationale Vétérinaire d’ Alfort, Pôle HQSA, 7, Avenue du Général de Gaulle 94704 Maisons Alfort, France article info Article history: Received 23 February 2010 Received in revised form 19 May 2010 Accepted 22 May 2010 Available online 1 June 2010 Keywords: Global sensitivity analysis Biological TTI Contamination assessment model Cold smoked salmon Listeria monocytogenes abstract The aim of this study was to apply a global sensitivity analysis (SA) method in model simplification and to evaluate (eO) Ò , a biological Time Temperature Integrator (TTI) as a quality and safety indicator for cold smoked salmon (CSS). Models were thus developed to predict the evolutions of Listeria monocytogenes and the indigenous food flora in CSS and to predict TTIs endpoint. A global SA was then applied on the three models to identify the less important factors and simplify the models accordingly. Results showed that the subset of the most important factors of the three models was mainly composed of the durations and temperatures of two chill chain links, out of the control of the manufacturers: the domestic refrigerator and the retail/cabinet links. Then, the simplified versions of the three models were run with 10 4 time temperature profiles representing the variability associated to the microbial behavior, to the TTIs evolution and to the French chill chain characteristics. The results were used to assess the distri- butions of the microbial contaminations obtained at the TTI endpoint and at the end of the simulated profiles and proved that, in the case of poor storage conditions, the TTI use could reduce the number of unacceptable foods by 50%. Ó 2010 Elsevier Ltd. All rights reserved. 1. Introduction Time Temperature Indicators (TTIs) are simple and user friendly devices put on the food packaging at the end of the manufacturing process. They reflect measurable and irreversible time temperature dependent changes, usually expressed by a mechanical deforma- tion, a color development or a color change to indicate when a food is spoiled (Taoukis and Labuza, 1989). Several types of TTIs, mainly chemical or physical systems, were commercialized since the early sixties (Taoukis and Labuza, 2003). Recently, a novel class of bio- logical TTIs has been developed (Ellouze et al., 2008; Vaikousi et al., 2008; 2009) and some of them were commercialized. (eO) Ò (CRY- OLOG, France) is a microbiological TTI made of selected strains of lactic acid bacteria. In the event of temperature abuse, or when the product reaches its use by date, the temperature dependent growth of the micro-organisms inside the TTI causes a pH drop into the tags leading to an irreversible color change of the medium chromatic indicator from green to red (Ellouze and Augustin, 2010). In this study, a focus was made on a particular setting of the TTI (eO) Ò to trace cold smoked salmon (CSS). In fact, this food is a ready-to-eat product with a rather long manufacturing process (up to 24 h) including handling operations at relatively high temperatures (16  Ce25  C) which increases the risk of microbial contamination and development (Richard et al., 2004). Moreover, it has a relatively long shelf life (up to 28 days in the French market) during which microbial growth is likely to occur. Many papers were published over the last years on CSS alteration (Dondero et al., 2004; Gonzalez-Rodrıguez et al., 2002; Gram and Huss, 1996; Joffraud et al., 2006; Leroi et al., 1998; Stohr et al., 2001; Truelstrup-Hansen and Huss, 1998). Besides, several authors studied the evolution of Listeria monocytogenes in this food (Augustin et al., 2005; Cornu et al., 2006; Dalgaard and Jorgensen, 1998; Delignette-Muller et al., 2006; FAO/WHO, 2004; Lindqvist and Westöö, 2000; Pouillot et al., 2007) as this kind of product can be contaminated with this pathogen. In order to evaluate (eO) Ò as a quality and safety indicator for CSS, two contamination assessment models describing the behavior of the indigenous microflora and that of L. monocytogenes were * Corresponding author at: Ecole Nationale Vétérinaire d’Alfort, Unité MASQ, Avenue du Général de Gaulle 94704 Maisons Alfort, France. Tel.: þ 33 1 43 96 71 78; fax: þ33 1 43 96 70 90. E-mail address: mellouze@vet-alfort.fr (M. Ellouze). Contents lists available at ScienceDirect Food Microbiology journal homepage: www.elsevier.com/locate/fm 0740-0020/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.fm.2010.05.022 Food Microbiology 28 (2011) 755e769