6.1 Introduction The modern food industry is called on to deliver seemingly contradictory market demands. On the one hand consumers want improved safety and sensory quality, together with increased nutritional properties, extended shelf-life and convenience in preparation and use. On the other they want food with a traditional, wholesome image, with less processing and fewer additives. In achieving safer and better quality food scientists and manufacturers apply intense optimisation and control of all the production and preservation parameters and additionally explore and benefit from innovative techniques to ensure safety and reduce food deterioration. Novel packaging such as active packaging is among such innovative tools. Producers and regulators rely on the development and application of structured quality and safety assurance systems based on prevention through monitoring, recording and controlling of critical parameters through the entire product life cycle. These systems should include the post-processing phase and ideally extend to the consumer’s table. The ISO 9001:2000 quality management standard (ISO 9001:2000; ISO 15161: 2001), widely adopted by the food industry, emphasises documented procedures for storage, handling and distribution. The globally recommended Hazard Analysis and Critical Control Point (HACCP) safety assurance system also focuses on this phase (93/43/EEC; Codex, 1997; US Federal Register, 1996). Certain stages of the chill chain are recognised as important critical control points (CCPs) for minimally processed chilled products such as modified-atmosphere packaged and other ready-to-eat chilled products. Monitoring and controlling these CCPs is seen as essential for safety. Research and industrial studies show that chilled or frozen distribution and 6 Time-temperature indicators (TTIs) P. S. Taoukis, National Technical University of Athens, Greece and T. P. Labuza, University of Minnesota, USA handling very often deviate from recommended temperature conditions. Since temperature largely constitutes the determining post-processing parameter for shelf-life under good manufacturing and hygiene practices, monitoring and controlling it is of central importance. The complexity of the problem is highlighted when the variation in temperature exposure of single products within batches or transportation sub-units is considered. Ideally, a cost-effective way to monitor the temperature conditions of food products individually, throughout distribution, is required to indicate their real safety and quality. This requirement could be fulfilled by Time Temperature Integrators or Indicators (TTIs). TTIs can be classified as active packaging. A TTI based system could lead to effective quality control of the chill chain, optimisation of stock rotation and reduction of waste, and provide information on the remaining shelf-life of product units. A prerequisite for the application of TTIs is the systematic study and kinetic modelling of the role of temperature in determining shelf-life. Based on reliable models of food product shelf-life and the kinetics of TTI response, the effect of temperature can be monitored, recorded and translated from production to the consumer’s table. 6.2 Defining and classifying TTIs A time temperature integrator or indicator (TTI) can be defined as a simple, inexpensive device that can show an easily measurable, time-temperature dependent change that reflects the full or partial temperature history of a food product to which it is attached (Taoukis and Labuza, 1989). The principle of TTI operation is a mechanical, chemical, electrochemical, enzymatic or microbiological irreversible change usually expressed as a visible response, in the form of a mechanical deformation, colour development or colour movement. The rate of change is temperature dependent, increasing at higher temperatures. The visible response thus gives a cumulative indication of the storage conditions that the TTI has been exposed to. The extent to which this response corresponds to a real time-temperature history depends on the type of the indicator and the physicochemical principles of its operation. Indicators can thus be classified according to their functionality and the information they convey. An early classification system introduced by Schoen and Byrne (1972) separated devices into six categories. Byrne (1976) revised this classification, realising that the main functional difference is whether the indicator responds above a preselected temperature, or responds continuously thus giving information on the cumulative time-temperature exposure. He proposed three types: 1. defrost indicators 2. time-temperature integrators 3. time-temperature integrators/indicators. A similar scheme recognised three categories (Singh and Wells, 1986): 104 Novel food packaging techniques 1. abuse indicators 2. partial temperature history indicators 3. full temperature history indicators (an alternative nomenclature for time- temperature integrators). A three-category classification will be used in this chapter (Taoukis et al., 1991). 6.2.1 Critical temperature indicators (CTI) CTI show exposure above (or below) a reference temperature. They involve a time element (usually short; a few minutes up to a few hours) but are not intended to show history of exposure above the critical temperature. They merely indicate the fact that the product was exposed to an undesirable temperature for a time sufficient to cause a change critical to the safety or quality of the product. They can serve as appropriate warning in cases where physicochemical or biological reactions show a discontinuous change in rate. Good examples of such cases are the irreversible textural deterioration that happens when phase changes occur (e.g., upon defrosting of frozen products or freezing of fresh or chilled products). Denaturation of an important protein above the critical temperature or growth of a pathogenic microorganism are other important cases where a CTI would be useful. The ‘critical temperature’ term is preferred rather than the used alternative ‘defrost’ that is too limiting. The term ‘abuse’ might be misleading as undesirable changes can happen at temperatures which are not as extreme or abusive as the term implies and which are within the acceptable range of normal storage for the product in question. 6.2.2 Critical temperature/time integrators (CTTI) CTTI show a response that reflects the cumulative time-temperature exposure above a reference critical temperature. Their response can be translated into an equivalent exposure time at the critical temperature. They are useful in indicating breakdowns in the distribution chain and for products in which reactions, important to quality or safety, are initiated or occur at measurable rates above a critical temperature. Examples of such reactions are microbial growth or enzymatic activity that are inhibited below the critical temperature. 6.2.3 Time temperature integrators or indicators (TTI) TTI give a continuous, temperature dependent response throughout the product’s history. They integrate, in a single measurement, the full time-temperature history and can be used to indicate an ‘average’ temperature during distribution and possibly be correlated to continuous, temperature dependent quality loss reactions in foods. In the remainder of this chapter, the term TTI will refer to this type of indicator, unless otherwise noted. A different method of classification sometimes used is based on the principle of the indicators’ operation. Thus, they Time-temperature indicators (TTIs) 105 can be categorised as mechanical, chemical, enzymatic, microbiological, polymer, electrochemical, diffusion based, etc. 6.3 Requirements for TTIs The requirements for an effective TTI are that it shows a continuous change, the rate of which increases with temperature and which does not reverse when temperature is lowered. There are a number of other desirable attributes for a successful indicator. An ideal TTI would have all the following properties: ? It exhibits a continuous time-temperature dependent change. ? The change causes a response that is easily measurable and irreversible. ? The change mimics or can be correlated to the food’s extent of quality deterioration and residual shelf-life. ? It is reliable, giving consistent responses when exposed to the same temperature conditions. ? It has low cost. ? It is flexible, so that different configurations can be adopted for various temperature ranges (e.g., frozen, refrigerated, room temperature) with useful response periods of a few days as well as up to more than a year. ? It is small, easily integrated as part of the food package and compatible with a high-speed packaging process. ? It has a long shelf-life before activation and can be easily activated. ? It is unaffected by ambient conditions other than temperature, such as light, humidity and air pollutants. ? It is resistant to normal mechanical abuse and its response cannot be altered. ? It is non-toxic, posing no safety threat in the unlikely situation of product contact. ? It is able to convey in a simple and clear way the intended message to its target, be that distribution handlers or inspectors, retail store personnel or consumers. ? Its response is both visually understandable and adaptable to measurement by electronic equipment for easier and faster information, storage and subsequent use. 6.4 The development of TTIs The drive for development of an effective and inexpensive indicator dates from the time when the importance of temperature variations to final food quality during distribution became apparent. Initially, the interest was focused on frozen foods. The first application of a ‘device’ to indicate handling abuse dates from World War II when the US Army Quartermaster Corps used an ice cube placed inside each case of frozen food. Disappearance of the cube indicated 106 Novel food packaging techniques mishandling (Schoen and Byrne, 1972). The first patented indicator goes back to 1933 (Midgley, 1933). Over a hundred US and international patents relevant to time-temperature indicators have been issued since. During the last 30 years numerous TTI systems have been proposed of which only few reached the prototype and even fewer the market stage. Patents dating up to 1990 are tabulated in the literature (Byrne, 1976; Taoukis, Fu and Labuza, 1991). Byrne (1976) gives an overview of the early indicators and Taoukis (1989) presents a detailed history of TTI. Table 6.1 lists significant recent TTI patents classified according to type and principle of operation. The first commercially available TTI was developed by Honeywell Corp (Minneapolis, MN) (Renier and Morin, 1962). The device never found commercial application, possibly because it was costly and relatively bulky. In the early 1970s, the US government considered mandating the use of indicators on certain products (OTA, 1979). This generated a flurry of research and development. Researchers at the US Army Natick Laboratories developed a TTI that was based on the colour change of an oxidisable chemical system controlled by the temperature dependent permeation of oxygen through a film (Hu, 1972). Field testing over a two-year period with the TTI attached to rations showed their potential for use (Killoran, 1976). The system was contracted to Artech Corp (Falls Church, VA) for commercial development. By 1976 six companies were making temperature Table 6.1 List of recent TTI patents and classification according to type and mode of response. Date Inventor Principle of operation Patent No 1991 Jalinski, T.J. Chemical (TTI) US5,182,212 1991 Jalinski, T.J. Chemical (TTI) US5,085,802 1991 Thierry, A. Chemical (CTI) US5,085,801 1991 Swartzel, K.R. Physicochemical (TTI) US5,159,564 1992 Jalinski, T. Chemical (CTI) EP497459A1 1993 Veitch, R.J. Physicochemical (CTI) EP563769A1 1993 Loustaunau, A. Physical (CTI) EP615614A1 1994 Loustaunau, A. Physical (CTI) US5,460,117 1994 Veitch, R.J. Physicochemical (CTI) US5,490,476 1995 Prusik, T. Physicochemical (TTI) US5,709,472 1996 Cannelongo, J.F. Physical (CTI) US5,779,364 1996 Veitch, R.J. Physical (CTI) EP835429A1 1997 Arens R. et al. Physicochemical (TTI) US5,667,303 1997 Schneider, N. Physical (CTI) US6,030,118 1999 Simons, M.J. Physicochemical (CTI) EP930488A2 2000 Schaten, B.B. Physical (CTI) EP1053726A2 2000 Prusik, T. Physical (CTTI) US6,042,264 2000 Ram, A.T. Chemical (TTI) US6,103,351 2000 Bray, A.V. Physical (TTI) US6,158,381 2001 Simons, M.J. Physicochemical (TTI) US6,214,623 2001 Qiu, J. Physicochemical (TTI) US6,244,208 2002 Qiu, J. Physicochemical (TTI) US6,435,128 Time-temperature indicators (TTIs) 107 indicators at least at the prototype stage (Kramer and Farquhar, 1976). The Artech, the Check Spot Co (Vancouver, WA) (US patent 2,971,852) and the Tempil (S Plainfield, NJ) indicators could be classified as CTI. The I-Point (Malmo¨, Sweden), the Bio-Medical Sciences (Fairfield, NJ) (US patents 3,946,611 and 4,042,336) and the 3M Co (St Paul, MN) indicators were TTI. The Tempil indicator could function as a CTTI. It involved a change to a red colour and subsequent movement when exposed above the critical temperature. The I-Point was an enzymatic TTI, and the 3M, a diffusion based TTI. By the end of the 1970s, however, very little commercial application of the TTI had been achieved. Research and development activity subsided temporarily, noted by a decrease in the relevant publications and in the new TTI models introduced. However, the better systems remained available and development continued, aiming at fine tuning and making performance more consistent. In the early 1980s, there were four systems commercially available including the I-Point and the 3M TTI. Andover Labs (Weymouth, MA) marketed the Ambitemp and Tempchron devices up to 1985. Both were for use in frozen food distribution and could be classified as CTTI. Their operation was based on the displacement of a fluid along a capillary. 6.5 Current TTI systems In the last fifteen years three types of TTI have been the focus of both scientific and industrial trials. They claim to satisfy the requirements of a successful TTI and have evolved as the major commercial types on the market. They are described in detail in the following sections. 6.5.1 Diffusion-based TTIs The 3M Monitor Mark (3M Co., St Paul, Minnesota) (US Patent, 3,954,011, 1976) is a diffusion based indicator. One of the first significant applications of TTI was the use of this indicator by the World Health Organization (WHO) to monitor refrigerated vaccine shipments. The response of the indicator is the advance of a blue dyed ester diffusing along a wick. The useful range of temperatures and the response life of the TTI are determined by the type of ester and the concentration at the origin. Thus the indicators can be used either as CTTI with the critical temperature equal to the melting temperature of the ester or as TTI if the melting temperature is lower than the range of temperatures the food is stored at, e.g., below 0oC for chilled storage. The same company has marketed the successor to this TTI: the Monitor Mark Temperature Monitor (Fig. 6.1) and Freshness Check, based on diffusion of proprietary polymer materials (US patent 5,667,303). A viscoelastic material migrates into a diffusely light-reflective porous matrix at a temperature dependent rate. This causes a progressive change of the light transmissivity of the porous matrix and provides a visual response. The response rate and temperature dependence is controlled by the tag configuration, 108 Novel food packaging techniques the diffusing polymer’s concentration and its glass transition temperature and can be set at the desirable range (Shimoni, Anderson and Labuza, 2001). The TTI is activated by adhesion of the two materials. Before use these materials can be stored separately for a long period at ambient temperature. 6.5.2 Enzymatic TTIs The VITSAB Time Temperature Indicator is an enzymatic indicator. It is the successor of the I-Point Time Temperature Monitor (VITSAB A.B., Malmo¨, Sweden). The indicator is based on a colour change caused by a pH decrease which is the result of a controlled enzymatic hydrolysis of a lipid substrate (US Patents 4,043,871 and 4,284,719). Before activation the indicator consists of two separate compartments, in the form of plastic mini-pouches. One compartment contains an aqueous solution of a lipolytic enzyme, such as pancreatic lipase. The other contains the lipid substrate absorbed in a pulverised PVC carrier and suspended in an aqueous phase and a pH indicator mix. Glycerine tricapronate (tricaproin), tripelargonin, tributyrin and mixed esters of polyvalent alcohols and organic acids are included in substrates. Different combinations of enzyme-substrate types and concentrations can be used to give a variety of response lives and temperature dependencies. At activation, enzyme and substrate are mixed by mechanically breaking the barrier that separates the two compartments. Hydrolysis of the substrate (e.g., tricaproin) causes acid release (e.g., caproic acid) and the pH drop is translated in a colour change of the pH indicator from deep green to bright yellow. Reference starting and end point colours are printed around the reaction window to allow easier visual recognition and evaluation of the colour change (Fig. 6.2). The continuous colour change can also be measured instrumentally (Taoukis and Labuza, 1989). The TTI is claimed to have a long shelf-life if kept chilled before activation. 6.5.3 Polymer-based TTIs The Lifelines Freshness Monitor and Fresh-Check indicators (Lifelines Inc, Morris Plains, NJ) are based on a solid state polymerisation reaction (US Patent, 3,999,946 and 4,228,126) (Fields and Prusik, 1983). The TTI function Fig. 6.1 Diffusion based TTI. Time-temperature indicators (TTIs) 109 is based on the property of disubstituted diacetylene crystals (R C = C C = C R) to polymerise through a lattice-controlled solid-state reaction proceeding via 1,4-addition polymerisation and resulting in a highly coloured polymer. During polymerisation, the crystal structure of the monomer is retained and the polymer crystals remain chain aligned and are effectively one dimensional in their optical properties (Patel and Yang, 1983). The response of the TTI is the colour change measured as a decrease in reflectance. The Freshness Monitor consists of an orthogonal piece of laminated paper the front face of which includes a strip with a thin coat of the colourless diacetylenic monomer and two barcodes, one about the product and the other identifying the model of the indicator. The Fresh-Check version, for consumers, is round, and the colour of the ‘active’ centre of the TTI is compared to the reference colour of a surrounding ring (Fig. 6.3). The laminate has a red or yellow colour so that the change is perceived as a change from transparent to black. The reflectance of the Freshness Monitor can be measured by scanning with a laser optic wand and stored in a hand-held device supplied by the TTI producer. The response of Fresh Scan can be visually evaluated in comparison to the reference ring or continuously measured by a portable colorimeter or an optical densitometer. Before use, the indicators, active from the time of production, have to be stored deep frozen where change is very slow. Fig. 6.2 Enzymatic TTI. Fig. 6.3 Polymer based TTI. 110 Novel food packaging techniques 6.6 Maximising the effectiveness of TTIs Despite the potential of TTIs to contribute substantially to improved food distribution, reduce food waste and benefit the consumer with more meaningful shelf-life labelling, their application up to now has not lived up to the initial expectations. The main reasons for the reluctance of food producers to adopt the TTI have been: ? cost ? reliability ? applicability. Cost is volume dependent, ranging from 2 to 20 US cents per unit. If other questions are resolved, cost-benefit analysis should favour use of TTIs. The reliability question has its roots in the history of indicators, due partly to lack of sufficient data, both from studies and from the suppliers. Initial attempts at using TTI as quality monitors were not well designed and hence unsuccessful. Re-emerging discussions by regulatory agencies to make TTI use mandatory, before the underlying concepts were understood and their reliability demonstrated, resulted in resistance by the industry and may have hurt TTI application up to the present time. Current TTI systems have achieved high standards of production quality assurance and provide reliable and reproducible responses according to the specifications stated. Testing standards have been issued by the BSI and can be used by TTI manufacturers as well as TTI users (BS 7908:1999). The question of applicability, however, has been the most substantial hurdle to TTI use. Earlier studies have been ineffective in establishing a clear methodology on how the TTI response can be used as a measure of food quality. The initial approach was to assume an overall temperature dependence curve (or zone) for the shelf-life of a general class of foods, e.g., frozen foods, and aim for an indicator that has a similar temperature dependence curve for the time to reach a specific point on its scale. Such a generalisation has proved insufficient, as even foods of the same type differ significantly in the temperature dependence of the deterioration in their quality. What is needed is a thorough knowledge of the shelf-life loss behaviour of the food system through accurate kinetic models. It has been widely assumed that the behaviour of a TTI should strictly match that of the particular food to be monitored at all temperatures. This approach, even if feasible, is impractical, and requires an unlimited number of TTI models. Instead of a TTI exactly mimicking quality deterioration behaviour of the food product, a meaningful, general scheme of translating TTI response to food status is needed. This should be based on systematic modelling of both the TTI and the food. Advances in modelling are now making this possible. Current developments in this area are reviewed by Taoukis (2001). The following sections discuss how modelling contributes to the practical use of TTIs. Time-temperature indicators (TTIs) 111 6.7 Using TTIs to monitor shelf-life during distribution TTIs can be used to monitor the temperature exposure of food products during distribution, from production up to the time they are displayed at the supermarket. Attached to individual cases or pallets they give a measure of the preceding temperature conditions at selected control points. Information from TTIs can be used for continuous, overall monitoring of the distribution system, leading to recognition and correction of weak links in the chain. Furthermore, it serves as a proof of compliance with contractual requirements by the producer and distributor. It can guarantee that a properly handled product was delivered to the retailer, thus eliminating the possibility of unsubstantiated rejection claims by the latter. The presence of the TTI itself would probably improve handling, serving as an incentive and reminder to distribution employees throughout the distribution chain of the importance of proper temperature storage. The same TTIs can be used as shelf-life end point indicators readable by the consumer and attached to individual products. Tests using continuous instrumental readings to define the end point under constant and variable temperatures showed that such end points could be reliably and accurately recognised by panellists (Sherlock et al., 1991). However, for a successful application of this kind there is a much stricter requirement that the TTI response matches the behaviour of the food. To achieve this the TTI end point should coincide with the end of shelf-life at one reference temperature and the activation energy should differ by less than 10kJ/mol from that of the food. In this way the TTI attached to individually packaged products can serve as active shelf-life labelling instead of, or in conjunction with, open date labelling. The TTI assures the consumer that the product was properly handled and indicates the remaining shelf-life. Consumer surveys have shown that consumers can be very receptive to the idea of using these TTI on dairy products along with the date code (Sherlock and Labuza, 1992). Use of TTI can thus also be an effective marketing tool. Diffusion-based TTIs have been used in this way by the Cub Foods Supermarket chain in the USA and polymer-based TTIs by the Monoprix chain in France and the Continent stores in Spain. A number of experimental studies have sought to establish correlations between the response of specific TTIs and quality characteristics of specific products. Foods have been tested at different temperatures, plotting the response of the TTI v. time and the values of selected quality parameters of the foods before testing the statistical significance of the TTI response correlation to the quality parameters. Foods correlated to TTI include: ? pasteurised whole milk (Mistry and Kosikowski, 1983; Grisius et al., 1987; Chen and Zall, 1987) ? ice cream (Dolan et al., 1985) ? frozen hamburger (Singh and Wells, 1985) ? chilled cod fillets (Tinker et al., 1985) ? refrigerated ready to eat salads (Cambell, 1986) 112 Novel food packaging techniques ? frozen bologna (Singh and Wells, 1986) ? UHT milk (Zall et al., 1986) ? refrigerated orange juice (Chen and Zall, 1987b) ? pasteurised cream (Chen and Zall, 1987a) ? cottage cheese (Chen and Zall, 1987; Shellhammer and Singh, 1991) ? frozen strawberries (Singh and Wells, 1987) ? chilled lettuce and tomatoes (Wells and Singh, 1988) ? chilled fresh salmon (Ronnow et al., 1999). Such studies offer useful information but do not involve any modelling of the TTI response as a function of time and temperature. They are thus applicable only for the specific foods and the conditions that were used. Extrapolation to other similar foods or quality loss reactions, or even use of the correlation equations for the same foods at other temperatures or for fluctuating conditions is not accurate. A kinetic modelling approach allows the potential user to develop an application scheme specific to a product and to select the most appropriate TTI without the need for extensive testing of the product and the indicator. This approach emphasises the importance of reliable shelf-life modelling of the food to be monitored. Shelf-life models must be obtained with an appropriate selection and measurement of effective quality indices and be based on efficient experimental design at isothermal conditions covering the range of interest. The applicability of these models should be further validated at fluctuating, non- isothermal conditions representative of the real conditions in the distribution chain. Similar kinetic models must be developed and validated for the response of the suitable TTI. Such a TTI should have a response rate with a temperature dependence, i.e., activation energy E A1 , in the range of the E A of the quality deterioration rate of the food. The total response time of the TTI should be at least as long as the shelf-life of the food at a chosen reference temperature. TTI response kinetics should be provided and guaranteed by the TTI manufacturer as specifications of each TTI model they supply. The above concepts have been applied in studying the suitability of TTIs in monitoring the seafood chill chain within the FAIR-CT96-1090 research project funded by the European Commission entitled ‘Development, Modelling and Application of Time-Temperature Integrators to monitor Chilled Fish Quality’. The fish chill chain is noted for substantial losses by spoilage. As part of this programme the shelf-life of different fresh and minimally processed fish products was systematically studied and modelled. Shelf-life analysis requires establishing a time correlation between measured chemical/biochemical changes, microbiological activity and sensory quality. Although each type of fish product, depending on the particular intrinsic and extrinsic factors, has its own specific spoilage pattern, investigation of the influence of each of these factors provides the fundamentals for understanding the spoilage phenomenon and for reliable shelf-life predictive modelling (Dalgaard, 1995; Dalgaard and Huss, 1995). Models of sensory quality and growth of spoilage microflora were Time-temperature indicators (TTIs) 113 developed and validated in dynamic temperature conditions for a variety of different fish. In this context the natural microflora of different Mediterranean fish of commercial interest such as boque, seabass, seabream and red mullet was studied and growth of the specific spoilage bacteria Pseudomonas spp. and Shewanella putrefaciens was modelled and correlated to organoleptic shelf-life (Taoukis et al., 1999; Koutsoumanis and Nychas, 2000; Koutsoumanis et al., 2000). Arrhenius and square root functions were used to model temperature dependence of maximum growth rates. For example experimental data for growth of the different measured constituents of the boque natural microflora showed that, at all temperatures, growth of Pseudomonads and Shewanella putrefaciens followed closely the decrease of average sensory score of the cooked fish. End of shelf-life coincided with an average level of 10 7 for these two bacteria from 0o to 15oC. At 0oC it was determined at 174 hr. The Arrhenius temperature dependence of the rate of sensory degradation and Pseudomonads and Shewanella putrefaciens exponential growth rate was determined in terms of activation energy (E A ) as 86.6, 81.6 and 82.7kJ/mol respectively. Based on this kinetic data the effect of the difference in the activation energies of TTI response and spoilage rate of the monitored fish on the shelf-life predictive ability of TTI can be assessed. The actual effective temperature (based on growth kinetics) of variable temperature profiles is compared to the one calculated from the response of the TTI. The total shelf-life at 0oC is 174 hr based on the Pseudomonads growth with N 0 1000 and N max 10 7 . This coincides with the sensory shelf-life. Setting these limits allows the estimation of remaining shelf-life at 0oC after the ‘abusive’ storage conditions of the first 24 hr. Table 6.2 shows T eff for the fish, after exposure for 24 hours at the variable temperature profiles (shown in Fig. 6.4), is given. It can be seen that, for the first temperature profile TD 1 , the enzymatic TTI Model C has an activation energy more than 40kJ/mol different from the fish spoilage and gives a T eff error of more than 1oC. This results in a prediction of remaining shelf-life of 74 hrs compared to the 90 hrs of actual remaining shelf- life. T eff and prediction of remaining shelf-life from the enzymatic TTI Model M and the polymer-based TTI Model A6 (92 and 91 hr respectively) are very close to the actual. It should be noted, however, that even the erroneous estimations Table 6.2 Effective temperature and remaining shelf-life (t r ) of boque at chilled conditions of 0oC, for variable storage temperatures (TD1, TD2) during the initial 24 hr estimated by different TTI TD1 TD2 T eff (oC) t r (hr) T eff (oC) t r (hr) ‘ACTUAL’ 8.93 90 9.8 81 Enzymatic TTI – Model C 10.50 74 10.0 79 Enzymatic TTI – Model M 8.74 92 9.8 82 Polymer-based TTI – Model A6 8.95 91 9.8 91 114 Novel food packaging techniques from the TTIs with different activation energies are in practice much better than the 150 hr that would be presumed for shelf-life if no indication of improper storage was available. For the second temperature profile, TD 2 , predictions from all TTIs are good. This illustrates the fact that the error depends on the actual temperature distribution. TD 1 and TD 2 are qualitatively different in that the first represents a profile with more abrupt changes than the second. The problem is that the temperature profile in a real situation is unknown. It is therefore advisable to select the TTI that, in addition to other requirements, has an activation energy close to the one of the quality loss rate of the food. Alternatively, response of two or three TTIs (i.e. a multiple TTI) with different E A1 s could provide a corrected estimate of T eff giving a reliable estimate of the food quality even when these E A1 s differ substantially from the E A of the food (Stoforos and Taoukis, 1998). The example shows the potential and demonstrates the methodology of monitoring the chill chain based on a continuous scale TTI response, translatable to an effective temperature history. This methodology can be applied to chilled products other than fish if appropriate quality loss models are available. Long shelf-life chilled foods can benefit from the ability to monitor their temperature history by the introduction of a TTI based distribution control and stock rotation system. Such a system will be described and evaluated in the next section. Frozen foods can also be monitored based on the same approach. Diffusion- based and enzymatic TTIs have been tested and modelled at temperatures in the range of 1 to 30oC (Giannakourou et al., 2000; Giannakourou and Taoukis, 2002). Diffusion-based TTIs can respond above a temperature at which diffusion commences. This temperature can also be set, based on the type of polymer Fig. 6.4 Variable temperature scenarios of storage of boque during 24 hours. Time-temperature indicators (TTIs) 115 materials used and their glass transitions. Certain caveats should, however, be taken into account when TTIs are applied to frozen foods. These are related to the applicability of Arrhenius models developed under isothermal conditions to the prediction of the effect of dynamic, fluctuating temperature storage. The response of tested TTIs shows Arrhenius behaviour also at the subfreezing range. Furthermore, the response models can reliably be used in non-isothermal conditions. However, foods can seriously deviate from such behaviour. Deviations from Arrhenius type temperature dependence can be due to freeze concentration effects, recrystallisation dependent quality deterioration and glass transition phenomena (Taoukis et al., 1997). For products like frozen vegetables, assuming that thawing is avoided (which can be verified with the aid of a CTI or a CTTI), Arrhenius behaviour in the range of 1 to 30oC has been modelled and validated for fluctuating conditions (Giannakourou et al., 2000; Giannakourou and Taoukis, 2002a). TTI monitoring can effectively be applied in these cases. In cases such as ice cream or other frozen desserts where recrystallisation phenomena seriously affect, if not determine, the product quality, a cumulative effective temperature (obtained by a single response TTI) might not be sufficient to predict accurately the quality loss. 6.8 Using TTIs to optimise distribution and stock rotation The information provided by a TTI, translated to remaining shelf-life at any point of the chill chain, can be used to optimise distribution control and apply a stock rotation system. Such an inventory management and stock rotation tool at the retail level was initially proposed by Labuza and Taoukis (1990). The approach currently used is the First In First Out (FIFO) system according to which, products received first and/or with the closest expiration date on the label are shipped, displayed and sold first. This approach aims in establishing a ‘steady state’ with all products being sold at the same quality level. The assumption is that all products have gone through uniform handling. Quality is basically seen as a function of time. The use of the indicators can help establish a system that does not depend on this unrealistic assumption. This approach is called LSFO (Least Shelf-life First Out). The LSFO system would reduce the number of rejected products and largely eliminate consumer dissatisfaction since the fraction of product with unacceptable quality at consumption time can be minimised. The development of LSFO system is based on validated shelf-life modelling of the controlled food product, specification of the initial value of the quality index, A 0 , and the value A s at the limit of acceptability (end of shelf-life), and temperature monitoring in the chill chain with TTI. The above elements for the core of integrated software that allows the calculation of the actual remaining shelf-life of individual product units (e.g. small pallets, 5–10 kg boxes or even single product units) at strategic control points of the chill chain. Based on the distribution of the remaining shelf-life, decisions can be made for optimal 116 Novel food packaging techniques handling, shipping destination and stock rotation, aimed at obtaining a narrow distribution of quality at the point of consumption. The diagram of the decision- making process is illustrated in Fig. 6.5. For example, at a certain point, e.g. at the supermarket, one half of a shipment could be forwarded to retail display immediately, the other half the next day. The split could be random according to conventional FIFO practice or it can be based on the actual individual product quality and LSFO. For all units the response of the TTI, cumulatively expressing the temperature exposure of the product, is put in either electronically as a signal to a suitable optical reader or keyed in manually based on visual readings. This information directly fed into a portable unit with the LSFO programme, is translated to quality status, A t , based on the kinetics of the used TTI, which integrates the time-temperature history of each product into an effective temperature value, T eff , and the shelf-life model of the product. Having estimated A t for all the n product units, the actual quality distribution for the products at the decision point is constructed. Based on the quality of each product unit relative to this distribution, decisions about its further handling are made. For the scenario illustrated in Fig. 6.5, products B with less remaining shelf- life, i.e., higher A t , will be displayed first in the retail display cabinets of the supermarket and will therefore be consumed sooner whereas products with longer remaining shelf-life (lower A t ) will be displayed later. The decision process can involve more options with regard to, e.g., handling methods, Fig. 6.5 Logical diagram of the decision-making routine of LSFO system at important control points of the distribution chain. Quality at time t (A t ) is computed for all n product units. The computation is based on the response of the TTI, translated to the effective storage temperature (T eff ) of the product. The distribution function of quality is constructed and decision for the further handling of each unit is taken based on its value within this function. Time-temperature indicators (TTIs) 117 shipping means or destinations, stock rotation timing and planning. Points of the chill chain where actions are taken with regard to handling, transportation, distribution and stocking of products can be designated and used as decision points of the LSFO system. In order to evaluate the results of the application of the LSFO system and quantitatively prove its effectiveness a Monte Carlo simulation can be applied, with data and information provided by surveys on the conditions of the distribution chain. It is based on the generation of hypothetical ‘scenarios’. Values of the controlling parameter, temperature, are treated as probability distributions, which represent uncertainty or the commonly encountered variation in the parameter. The procedure, repeated many times, requires the random selection of a value from each of the probability distributions assigned for the input parameters, in order to calculate a mathematical solution, defined by the shelf-life model used. At each iteration, a value is drawn from the defined distribution, calculations are performed and the results are stored. Eventually, the analysis provides a frequency distribution for the output of interest (quality status and remaining shelf-life), that has taken into account the probability distribution of temperature conditions, instead of using a single-point estimate. The results for the simulated application of LSFO in the cases of two long shelf-life chilled products are shown in Fig. 6.6. These products have both a shelf-life of three months at 4oC and their quality loss rate shows a low and high temperature dependence (low and high E A respectively). Russian salad, a chilled product widely consumed in Greece with a shelf-life of three months, was used as a case study (Taoukis et al., 1998). For this microbiologically stable, complex food, modelling of shelf-life was based on overall organoleptic deterioration and development of rancidity. Use of Weibull Hazard Analysis facilitated shelf-life determination and modelling of sensory evaluation data. The activation energy of shelf-life loss was estimated at 31.5 kJ/mol. For a realistic estimate of the storage temperature conditions at the different stages, data of chilled product temperatures previously collected at the commercial level and from a survey of home refrigerators was used. The temperature condition distributions are illustrated in Fig. 6.7. To demonstrate the effectiveness of the LSFO approach as compared to FIFO, a 60-day cycle, from production to consumption, was used. This consisted of three stages: ? Stage 1: 30 days at local distribution centres ? Stage 2: 15 days at the supermarket storage ? Stage 3: 15 days at the domestic refrigerator Based on a Monte Carlo simulation approach, 2000 temperature scenarios were run, using a program code written in FORTRAN 77. The temperatures used were obtained at random from the distributions of Fig. 6.7 (distribution 10a for stages 1 and 2 and distribution 10b for stage 3). The results of this simulation are illustrated in Fig. 6.6 which shows the probability for the product to be 118 Novel food packaging techniques Fig. 6.6 Distribution of quality of Russian salad products after 60 days distribution, retail and domestic storage. For each point the percentage of the products that have a remaining shelf-life in the range of 2.5 days of the abscissa value can be read on the vertical axis. The line with solid circles corresponds to the FIFO and with open circles to the LSFO system based on actual temperature monitoring or a TTI with E A I E A . Open diamonds line is the LSFO line based on the TTI, Type B (practically coincides with the actual LSFO line). Fig. 6.7 Left: Temperature distribution in commercial chilled storage. (Measurements in 150 supermarkets in the metropolitan area of Athens). Right: Temperature distribution in domestic refrigerators. (Based on measurements in 40 households). (Adopted from Taoukis et al., 1998.) Time-temperature indicators (TTIs) 119 consumed at a certain quality level, expressed as Shelf-Life Remaining (SLR). The FIFO approach shows a significant portion of products (8%) consumed with quality lower than the one used to set end of shelf-life (expressed as a negative remaining shelf-life). Using the LSFO approach, products in the second state are advanced to retail cabinets for sale every five days based not on FIFO but on the response of the attached TTI (an enzymatic TTI (Type B) Models C and M) showing which products should be advanced first. This system leads to a narrower range of quality at consumption time (less than 1% unacceptable products) and can practically eliminate the ‘tails’, i.e., the portions of products consumed at extreme qualities. Thus a situation where products are consumed at a uniform quality, with no ‘below standard’ products can be obtained. As tools for a comparative selection of least shelf-life products at control points both TTIs can be used effectively. The same system was applied to distribution and stock rotation of shrink- packed marinated seafood products (marinated fish fillets, shrimp, squid, octopus) with a target shelf-life of three months at 4oC. Shelf-life temperature dependence of such products, based on sensory evaluation, varied but was in the high range of activation energies. An E A value of 110kJ/mol, a distribution cycle of 35 days, consisting of the same three stages and temperature distributions as above (10 days at stage 1, 15 days at stage 2 and 10 days at stage 3) and enzymatic TTIs (Models C, S and L with activation energies 48.3, 102 and 160 kJ/mol respectively) were used in the Monte Carlo simulation to assess application of LSFO. Results are shown in Fig. 6.8. It can be seen that in products with high activation energies the distribution of quality at consumption time is much wider as temperature variation affects more intensely the rates of quality loss. Application of the LSFO system reduces the percentage of unacceptable products to less than 5% compared to 22% with the FIFO approach. It can also be seen that even TTIs that differ from the food in terms of E A approximately 50 kJ/mol can serve as tools for the relative comparison of the shelf-life of the products at the control points of the LSFO system. The LSFO system can also be applicable for frozen foods ( Giannakourou and Taoukis, 2003). A further development of LSFO is an intelligent system known as the Shelf Life Decision System (SLDS) (Giannakourou et al., 2001a,b; Koutsoumanis et al., 2002). SLDS integrates predictive kinetic models of food spoilage, data on initial quality from rapid techniques and the capacity to monitor continuously temperature history of the food product with time temperature integrators (TTIs). It provides an effective chill chain management tool that leads to an improved distribution of quality at consumption, significantly reducing the probability of products past their shelf-life reaching consumers. For most processed food products, ‘zero time’ post-processing parameters, including a target range of initial microbial load, can be fixed and achieved by proper design and control of the processing conditions. This is the working assumption of LSFO. However, initial microflora in fresh foods such as fish or meat can fluctuate significantly, depending on a number of extrinsic factors at slaughter or 120 Novel food packaging techniques catch, and subsequent handling and processing (Eisel et al., 1997; Huss, 1995). SLDS takes not only the history of the product in the distribution chain into account but also this variability of initial contamination. Rapid methods of microbial enumeration can be employed to provide such information as input. The Shelf Life Decision System can incorporate other parameters in the calculation of the quality distribution at each control point. Such parameters can include variation of initial pH, water activity, packaging gases composition, provided the shelf-life predictive models can account for the effect of these parameters on the microbiological and chemical reactions responsible for the loss of quality. 6.9 Future trends TTIs will inevitably find wider application as tools to monitor and control distribution as their potential is thoroughly understood by the food industry. Progress on both the variety, reliability and flexibility of TTIs, and on better quantitative shelf-life characterisation of food products, will allow successful application of chill chain optimisation tools such as the LSFO and the intelligent Fig. 6.8 Distribution of quality of shrink-packed products after 30 days distribution, retail and domestic storage. For each point the percentage of the products that have a remaining shelf-life in the range of 5 days of the abscissa value can be read on the vertical axis. The line with solid circles corresponds to the FIFO and with open circles to the LSFO system based on actual temperature monitoring or the TTI B- Model S with E A I E A . Open diamonds and open triangles lines are the LSFO lines based on the TTI, Type B–Models L and C. Time-temperature indicators (TTIs) 121 Shelf Life Decision System. Research progress in the area of quality kinetic modelling and predictive microbiology will show how the TTI concept can be meaningfully and safely expanded to contribute to the quality assurance of more foods. User friendly softwares will integrate support systems designed to predict effects of processing parameters and product design to food product quality (Wijtzes et al., 1998). Such systems could provide the data input on initial product quality distribution, based on processing and raw material parameters, that is needed for the SLDS calculations at the control points of the chill chain on which the TTI based management of the products occurs. The state of TTI technology and of the scientific approach with regard to the quantitative safety risk assessment in foods will also allow the undertaking of the next important step, i.e., the study and development of a TTI based management system that will assure both safety and quality in the food chill chain. The development and application of such a system coded with the acronym SMAS is the target of a new, multipartner research project funded by the European Commission titled ‘Development and Modelling of a TTI based Safety Monitoring and Assurance System (SMAS) for chilled Meat Products’ (project QLK1-CT2002-02545, 2003-2005). The main objectives of this project are: ? modelling the effect of food structure, microbial interactions and dynamic storage conditions on meat pathogens and spoilage bacteria. ? combination of validated pathogen growth models with data on prevalence/ concentration, dose response and chill chain conditions for risk assessment with and without SMAS application. ? development, modelling and optimisation of TTI with accuracy to monitor microbiological safety of meat products. ? development of SMAS into a user-friendly computer software. ? evaluation of the applicability and effectiveness of SMAS in real conditions of meat distribution. ? assessment of the industry acceptance of the TTI and the concept of chill chain management and ? evaluation of the consumer attitude on use of TTI and correlation to quality. Information on the outputs of this project will be available on the Web (www.cordis.lu/life/src/pub_qol.htm). 6.9 References 93/43/EEC, 1993, ‘Hygiene of foodstuffs’, General Directive, 1993. BS 7908: 1999, ‘Packaging–temperature and time–temperature indictors – performance specification and reference testing’, British Standard, 1999. BYRNE C H, 1976, ‘Temperature indicators – the state of the art’, Food Technology, 30 (6) 66–8. CAMBELL L A, 1986, Use of a time–temperature indicator in monitoring quality of refrigerated salads, M. S. Thesis, Michigan State University. 122 Novel food packaging techniques CHEN J H and ZALL R R, 1987a, ‘Packaged milk, cream and cottage cheese can be monitored for freshness using polymer indicator labels’, Dairy Food Sanitation, 7 (8) 402–5. CHEN J H and ZALL R R, 1987b, ‘Refrigerated Orange Juice can be monitored for freshness using a polymer indicator label’, Dairy Food Sanitation, 7 280– 3. CODEX COMMITTEE ON FOOD HYGIENE, 1997, ‘Revised International Code of Practice – General Principles of Food Hygiene’, WHO/FNU/97, 1997. DALGAARD P and HUSS H H, 1995, ‘Mathematical modelling used for evaluation and prediction of microbial fish spoilage’, in: Seafood safety, processing and biotechnology, Kramer, D, Shahidi, F and Jones, Y (eds) Technomic Pub Co Inc, Lancaster, Canada, 73–89. DALGAARD P, 1995, ‘Modelling of microbial activity and prediction of shelf–life for packed fresh fish’, Int J of Food Microbiol, 26 305–17. DOLAN K D, SINGH R P and WELLS J H, 1985, ‘Evaluation of time–temperature related quality changes in ice cream during storage’, J Food Proc Preserv, 9 253–71. EISEL W G, LINTON R H and MURIANA P M, 1997, ‘A survey of microbial levels for incoming raw beef and ground beef in a red meat processing plant’, Food Micro, 14 273–82. FIELDS S C and PRUSIK T, 1983, ‘Time–Temperature monitoring using solid–state chemical indicators’, Intl Inst Refrig Commission C2 Preprints, 16th Intl Cong Refrig, 636–40. GIANNAKOUROU M C, TAOUKIS P S, 2002. Systematic application of Time Temperature Integrators as tools for control of frozen vegetable quality. J.Food Sci. 67(6): 2221–8. GIANNAKOUROU M C, TAOUKIS P S. 2003. ‘Application of a TTI–based distribution management system for quality optimisation of frozen vegetables at the consumer end’. J.Food Sci. 68(1): 201–9. GIANNAKOUROU M C, SKIADOPOULOS A, POLYDERA A and TAOUKIS P S, 2000, ‘Shelf–life modelling of frozen vegetables for quality optimisation with Time–Temperature indicators’, Eighth International Congress on Engineering and Food, Puebla, Mexico. GIANNAKOUROU M C, KOUTSOUMANIS K, DERMESONLOUOGLOU E, and TAOUKIS P S, 2001a, ‘Applicability of the intelligent Shelf Life Decision system for control of nutritional quality of frozen vegetables’. Acta Horticulturae, 566: 275–80. GIANNAKOUROU M, KOUTSOUMANIS K, NYCHAS G–J E and TAOUKIS P S, 2001b, ‘Development and assessment of an intelligent Shelf Life Decision System for quality optimisation of the food chill chain’, J Food Protection, 64(7): 1051–7. GRISIUS R, WELLS J H, BARRET E L and SINGH R P, 1987, ‘Correlation of time– temperature indicator response with microbial growth in pasteurised milk’, J Food Proc Preserv, 11 309–24. HU K H, 1972, ‘Time–Temperature indicating system ‘‘writes’’ status of product Time-temperature indicators (TTIs) 123 shelf–life’, Food Technology, 26 (8) 56–62. HUSS H H, 1995, ‘Quality and quality changes in fresh fish’, FAO Fisheries Technical Paper, No. 348, FAO, Rome. ISO 9001:2000, ‘Quality management systems – Requirements’, International Standard, 2000. ISO 15161:2001, ‘Guidelines on the application of ISO 9001: 2000 for the food and drink industry’, International Standard, 2001. KILLORAN J, 1976, ‘A time–temperature indicating system for foods stored in the nonfrozen state’, Activities Report, US Army Natick Labs, 137–42. KOUTSOUMANIS K and NYCHAS G–J E, 2000, ‘Application of a systematic experimental procedure to develop a microbial model for rapid fish shelf– life procedure’, Int J Food Microbiol, 60 171–84. KOUTSOUMANIS K, GIANNAKOUROU M C, TAOUKIS P S and NYCHAS G–J E, 2002. ‘Application of Shelf Life Decision System (SLDS) to marine cultures fish quality’, Int J Food Micro. 73: 375–82. KOUTSOUMANIS K, TAOUKIS P S, DROSINOS E and NYCHAS G–J E, 2000, ‘Applicability of an Arrhenius model for the combined effect of temperature and CO 2 packaging on the spoilage microflora of fish’, Appl Env Microbiol, 66 3528–34. KRAMER A and FARQUHAR J W, 1976, ‘Testing of time–temperature indicating and defrost devices’, Food Technology, 32 (2) 50–6. LABUZA T P and TAOUKIS P S, 1990, ‘The relationship between processing and shelf–life’, in: Foods for the 90’s, Birch, G G, Campbell–Platt, G and Lindley, M G (eds), Developments Series, Elsevier Applied Science Publishers, London and New York, Ch. 6, 73–105. MIDGLEY T, 1933, ‘Telltale means’, US Patent 1917048. MISTRY V V and KOSIKOWSKI F V, 1983, ‘Use of time–temperature indicators as quality control devices for market milk’, J Food Protection, 46 (1) 52–7. OTA, 1979, Open Shelf Life Dating of Food, Office of Technology Assessment, Congress of the US, Library of Congress Cat. No. 79–600128. PATEL G N and YANG N, 1983, ‘Polydiacetylenes: An ideal system for teaching polymer science’, J Chem Ed, 60 (3) 181–5. RENIER J J and MORIN W T, 1962, ‘Time–temperature indicators’, Intl Inst Refrig Bull Annex, 1 425–35. RONNOW P, SIMPSON R and OTWELL S, 1999, ‘The use of an enzymatic TTI to monitor time–temperature exposure in distribution of chilled seafood’, in: Predictive Microbiology Applied to Chilled Food Preservation. C M Bourgeois, and T A Roberts (eds) Refrigeration Science and Technology Proceedings Series, International Institute of Refrigeration (IIR), Paris, France, 308–15. SCHOEN H M and BYRNE C H, 1972, ‘Defrost indicators: many designs have been patented yet there is no ideal indicator’, Food Technol, 26 (10) 46–50. SHELLHAMMER T H and SINGH R P, 1991, ‘Monitoring chemical and microbial changes of cottage cheese using a full history time–temperature indicator’, J Food Sci, 56 402–5. 124 Novel food packaging techniques SHERLOCK M and LABUZA T P, 1992, ‘Consumer perceptions of consumer time– temperature indicators for use on refrigerated dairy foods’, J Dairy Science, 75, 3167–76. SHERLOCK M, FU B, TAOUKIS P S and LABUZA T P, 1991, ‘Systematic evaluation of time temperature indicators for use as consumer tags’, J Food Protection, 54 (11) 885–9. SHIMONI E, ANDERSON E M and LABUZA T P, 2001, ‘Reliability of Time Temperature Indicators Under Temperature Abuse’ J. Food Sci. 66(9): 1337–1340. SINGH R P and WELLS J H, 1985, ‘Use of time–temperature indicators to monitor quality of frozen hamburger’, Food Technology, 39 (12) 42–5. SINGH R P and WELLS J H, 1986, ‘Keeping track of time and temperature: new improved indicators may become industry standards’, Meat Processing, 25 (5) 41–2, 46–7. SINGH R P and WELLS J H, 1987, ‘Monitoring quality changes in stored frozen strawberries with time–temperature indicators’, Int J Refrig, 10 296–9. STOFOROS N G and TAOUKIS P S, 1998, ‘A theoretical procedure for using multiple response time–temperature integrators for the design and evaluation of thermal processes’, Food Control, 9 (5) 279–87. TAOUKIS P S and LABUZA T P, 1989, ‘Applicability of Time Temperature Indicators as shelf–life monitors of food products’, J Food Sci, 54 783–8. TAOUKIS P S, 1989, Time–temperature indicators as shelf life monitors of food products, PhD Thesis, University of Minnesota, St Paul. TAOUKIS P S, 2001, ‘Modelling the use of time temperature indicators in distribution and stock rotation’, in Tijskens L, Hertog M and Nicolai B (eds), Food process modelling, Woodhead Publishing Ltd, Cambridge. TAOUKIS P S, BILI M and GIANNAKOUROU M, 1998, ‘Application of shelf–life modelling of chilled salad products to a TTI based distribution and stock rotation system’, in: Proceedings of the International Symposium on Application of Modelling as an Innovative Technology in the Agri–Food– Chain, Tijiskens, L M M and Hertog, M L A T M (eds), Acta Horticulturae, 476 131–40. TAOUKIS P S, LABUZA T P and SAGUY I, 1997, ‘Kinetics of food deterioration and shelf-life prediction’, in: Handbook of Food Engineering Practice, Valentas K J, Rotstein E and Singh R P (eds), CRC Press, Boca Raton, Ch. 10, 361–403. TAOUKIS P S, FU B and LABUZA T B, 1991, ‘Time–Temperature Indicators’, Food Technology, 45 (10) 70–82. TAOUKIS P S, KOUTSOUMANIS K and NYCHAS G–J E, 1999, ‘Use of Time Temperature Integrators and Predictive Modelling for Shelf Life Control of Chilled Fish under Dynamic Storage Conditions’, Int J Food Microbiol, 53 21–31. TINKER J H, SALVIN J W, LEARSON R J and EMPOLA V G, 1985, ‘Evaluation of automated time–temperature monitoring system in measuring the freshness of chilled fish’, IIF-IIR Commissions C2, D3 4, 286–90. Time-temperature indicators (TTIs) 125 US FEDERAL REGISTER, 1996, ‘Pathogen Reduction; Hazard analysis and Critical Control Point (HACCP) Systems; Final Rule’, Vol. 61, No. 144, July 25, 1996. WELLS J H and SINGH R P, 1988, ‘Application of time-temperature indicators in monitoring changes in quality attributes of perishable and semi-perishable foods’, J Food Sci, 53 148–52. WIJTZES T, VAN’T RIET K, HUIS IN’T VELDAND J H J and ZWIETERING M H, 1998, ‘A decision support system for the prediction of microbial food safety and food quality’, Int J Food Microbiol, 42 79–90. ZALL R R, CHEN J and FIELDS S C, 1986, ‘Evaluation of automated time– temperature monitoring systems in measuring freshness of UHT milk’, Dairy and Food Sanitation, 6 (7) 285–90 126 Novel food packaging techniques