Food Control 17 (2006) 935941 www.elsevier.

com/locate/foodcont

The establishment of critical control points at the washing and air chilling stages in poultry meat production using multivariate statistics
lez-Miret, M.L. Escudero-Gilete, F.J. Heredia M.L. Gonza
*
Laboratory of Food Colour and Quality, Department of Nutrition and Food Science, University of Seville, Faculty of Pharmacy, C/P. Garcia Gonzalez, 2, 41012 Seville, Spain Received 26 October 2004; received in revised form 31 May 2005; accepted 6 June 2005

Abstract The selection of control points is one of the most important steps in the design of a Hazard Analysis and Critical Control Points (HACCP) system. Total Count, Pseudomonas and Enterobacteriaceae are microorganisms frequently analyzed on carcasses in slaughterhouses. Its usefulness can be assessed by means of univariate and multivariate statistical methods. In this study, the use of these microbiological parameters for verication of the eect of the stages of washing with pressurized water and air chilling has been evaluated. It makes clear that multivariate statistics appears as a valuable tool to check which points and stages of the process must be controlled, demonstrating that the washing stage produces signicant decreases in contamination (so it must be considered a CCP). The air chilling stage maintains the decrease in the contamination as the carcasses come out of the washing process. This is due to the control of temperature under adequate limits. The chiller air temperature would be considered a CCP in a HACCP system. 2005 Elsevier Ltd. All rights reserved.
Keywords: HACCP; Poultry meat; Statistical process control

1. Introduction Hazard Analysis and Critical Control Points Systems (HACCP) have become mandatory for food industry in the European Union (European Community Directive, 1993). They are based on following the production line from the beginning (raw material included) to the nished product. The main objective is to ensure that the products are of high hygienic quality. These systems present some advantages over traditional methods. Results obtained in a study from eight slaughterhouses suggested that HACCP systems can maintain or even

Corresponding author. Tel.: +34 954556761; fax: +34 954557017. E-mail address: heredia@us.es (F.J. Heredia).

improve food safety (Cates, Anderson, Karns, & Brown, 2001). The design of a HACCP system includes dierent steps. One of the most important facts to be concerned about is the proper selection of eective control parameters and the establishment of the stages of the process that we can control (CCP) (ICMSF, 1991; McNamara, 1997; Mortimore & Wallace, 1996; National Advisory Committee on Microbiological Criteria for Food, 1998). An eective quality control system can be dened as one whose main objective is not only to avoid defects (such as HACCP systems) but also to increase the factors that improve the nal quality of the product. This could be achieved by means of the application of uni- and multivariant statistical studies considering dierent factors, points and processes, with regard to the production chain, obtaining results from which

0956-7135/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodcont.2005.06.012

936

lez-Miret et al. / Food Control 17 (2006) 935941 M.L. Gonza

lezone can take decisions about the process (Gonza lez-Miret, Miret, Alonso, & Heredia, 1998, 2000; Gonza lez-Miret, Coello, Alonso, & Heredia, 2001; Gonza Escudero, Alonso, & Heredia, 2001). There are several uses of the microbiological testing in HACCP systems: prerequisite programs, assessment and monitoring of critical control points, validation and verication (Kvenberg & Schwalm, 2000). Poultry carcasses usually have very high levels of contamination on the skin. They can present microorganisms that cause food-born illness as well as food spoilage. There are a series of microorganisms on the surface of the carcasses which can be analyzed in order to indicate the microbiological quality, the level of hygiene in production and handling, and the correct maintenance of the cold chain (Sandrou & Arvanitoyannis, 1999). Furthermore, they can help to predict products shelf-life. The aerobic mesophile bacteria (Total Count) grow at medium temperatures (optimum temperature from 30 to 45 C) and they can be used as indices of inadequate practices during the process. This analysis can reect the hygienic quality of the analyzed products, indicating the hygienic conditions of the raw material as well as the way in which they were handled during the manufacturing process (Grau, 1986; Pascual, 1992). Psychrotrophic microorganisms such as Pseudomonas have great importance in products that are stored at low temperatures. They are responsible for the supercial alteration of refrigerated poultry meat since in aerobic atmospheres they can grow and multiply at very low temperatures (4 C). Pseudomonas is one of the most important sources in the spoilage of refrigerated foods, causing putrefactive odors and slime (Sofos, 1994). This microorganism could be present in water (Lehallec & Colin, 1979). Total Enterobacteriaceae, coliforms and Escherichia coli are used as marker organisms in foods, and the detection of specic pathogens of the family Enterobacteriaceae is widely applied in many food control laboratories (De Boer, 1998). The testing of Enterobacteriaceae is a normal analysis for poultry meat contamination while E. coli is more specically analyzed for pathogen microorganisms. Normally the analysis of Enterobacteriaceae as an indicator of fecal contamination (whether pathogenic or not) is used, indicating a misshandling process (Forsythe & Hayes, 2002; Robach, 1996; Snijders & Van Knapen, 2002). Most of the Enterobacteriaceae found on the surfaces of the carcasses come from feathers and fecal contamination. Counts of these microorganisms have been used as indicators of contamination due to unhygienic manipulation or inadequate storage (Grau, 1986; Pascual, 1992). Chickens initially show a high level of microbial contamination in their digestive tract, as well as on their feathers, skin and paws, which comes feces and the atmosphere. These microorganisms will be redistributed

during the stages of the production chain. At the same time, a cross contamination from one carcass to another can take place, as well as from surfaces, water and handlers. During the stages of sacrice, bleeding, scalding, plucking and gutting, important increases of the supercial contamination of the carcasses take place due to cross contamination and contamination from feces and atmosphere. After the gutting process, the carcasses are cleaned with pressurized water. Then the carcasses are chilled using cold air or cold water. A delay in the application of the cooling involves a possible microbial growth. The aim of this study is to determine if the cleaning and air chilling stages are CCPs or not, to nd out if there are signicant dierences between consecutive points in the production line, in addition to the importance of each microbiological variable in the discrimination, thanks to the application of multivariate statistical techniques to the microbiological data obtained.

2. Materials and methods 2.1. Process description The study has been carried out including two phases: washing with pressurized water and air chilling (Fig. 1). 2.1.1. Previous stages After the slaughter and bleeding the carcass was scalded to 52 C. This was followed by the plucking and automatic gutting before cleaning with pressurized water and air chilling stages. 2.1.2. Washing with pressurized water After these processes, carcasses are cleaned, both outside and inside, for two reasons: to clean them (remove any remains of blood, dirt and feathers) and to avoid , 1985; subsequent drying in the freezers (Buxade Vaquerizo, 1991). Just before the washing process, the carcasses are hung manually on stainless steel hooks. The washing zone is 225 cm long and inside has four blocks of sprinkles with a total of 32 nozzles in dierent directions, 28 sideways and 4 upwards. Therefore, the carcasses are washed for 8 s. In block 1 and 2 the water is at a pressure of 1215 kg/cm2 and in 3 and 4 it is at about 2 kg/cm2. The water temperature is approximately 17 C, the average temperature of the washing stage is approximately 1213 C and the carcasses enter the washing stage at a temperature of approximately 39 C and leave at approximately 37 C. This is a very important operation, since cleaning car lez-Miret casses eliminates many microorganisms (Gonza et al., 1998) before the air chilling stage. At the washing

 lez-Miret et al. / Food Control 17 (2006) 935941 M.L. Gonza

937

[A]

100 min. -6 - 2 C

CHAIN

[G] 8 seg. 12-13 C 12-15 Kg 2 Kg

[W]

Washing with pressurized water

Air chilling

Fig. 1. Washing with pressurised water and air chilling stages. Sampling points: [G], [W] and [A]. ([G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling.)

stage the so called clean zone of the slaughterhouse begins. In this area the operations are carried out at controlled low temperatures to avoid contamination. 2.1.3. Air chilling It is the rst stage of the clean zone. Thanks to a dynamic cooling tunnel the carcasses are rapidly refrigerated. This consists on a refrigeration vault, at very low temperature (6 C to +2 C), a relative humidity of 69% and with a capacity for about 10,000 carcasses. A zigzag-shaped chain of 1300 m crosses it at three levels. The process occurs in approximately 100 min. At the end of this step the carcasses are between 4 and 9 C, depending on their size. It avoids the growth of the non psychrotrophic ora and it slows down the growth of the psychrotrophic ora. 2.2. Samples All samples were taken from breast skin in a normal working day in which no special cleaning took place in order to evaluate the real possible contamination. The same chicken, specially labeled for identication purposes, was sampled at three dierent moments of the process: after gutting [G], after washing with pressurized water [W] and after air chilling [A] (Fig. 1). At the same time, but over a period of dierent days, 30 carcasses were studied, which gave a total of 90 samples. All the samples were taken in aseptic conditions using dierent tweezers (they were sterilized in autoclave) and singleuse sterilized scalpels, and were placed in Petri plates. Therefore, the entire sampling process takes 2 min. The sample is immediately transferred to the laboratory where it is instantly analyzed.

2.3. Microbiological analysis A 10 g sample of each product was taken aseptically and placed in a sterile Stomacher bag containing a solution of 90 ml of 0.1% Peptone Water (PW, Oxoid, Basingstoke, Hampshire, England). The samples and the PW were stomached for 1 min. Decimal dilutions (102, 103, 104) were carried out using the same diluent. In this study ambient and carcass temperatures were controlled, and the following microbiological parameters were measured (Cuni, 1995; Forsythe & Hayes, 2002; Vanderzant & Splittstoesser, 1992) whose utility in the control system has been made clear in previous lez-Miret et al., 2001): studies (Gonza Total Count (Tc): Nutrient Agar (Oxoid, Basingstoke, Hampshire, England). Incubated at 30 C for 48 h. Pseudomonas (Ps): Pseudomonas Isolation Agar (Difco, Detroit, Michigan, USA). Incubated at 30 C for 48 h. Enterobacteriaceae (Eb): Violet Red Bile Glucose Agar (Oxoid, Basingstoke, Hampshire, England). Incubated at 37 C for 24 h.

2.4. Statistical analysis The data were processed by means of statistical software: Statistica v 5.5 (1995). A t-student test for the related groups and the stepwise discriminant analysis were carried out by means of SPSS v 10.0 (1999) statistical software.

938

lez-Miret et al. / Food Control 17 (2006) 935941 M.L. Gonza Table 1 KolmogorovSmirnovLilliefors normality test (jDmaxj) Variable Stage [G] Log Tc Log Ps Log Eb 0.096 0.144 0.129 [W] 0.131 0.152 0.146 [A] 0.119 0.150 0.139

A descriptive analysis of the data (means and standard deviations) has been done to give a general prelin, 2001). The objective of this minary view (Mart analysis is the description of data through the presentation, organization and summary of data to see the form of these. After this, a comparison among the three dierent sampling points was made by means of repeated measures analysis of the variance (ANOVA) (the same carcass was sampled at three dierent points and three dierent microorganisms were analyzed) in order to determine if signicant dierences (p < 0.05) exist among the means of the dierent groups (between points, between micoorganisms and between points for each microorganism) (Norman & Streiner, 1996). If differences (p < 0.05) are found we can state that the points are dierent and could be considered a CCP. The ANOVA statistic technique was carried out to check how different stages aect contamination carcasses. Furthermore, stepwise discriminant analysis (SDA) was carried out. This statistical technique requires a qualitative variable (dependent variable) and two or more quantitative or dichotomous variables (independent variables). It is a method of classication whose aim is to estimate through linear functions (discriminant functions) of the independent variables (microbiological variables) (Johnson, 2000; Tabachnick & Fidell, 1983) the probability that one of the cases belongs to each of the groups dened by the categories of the dependent variable (stage). There are as many groups as categories with the aforementioned dependent variable ([G], [W] and [A]). This classication is made according to the properties given by the independent variables common to each case (Tc, Ps and Eb). The aim of this analysis was to evaluate the capability for classication and prediction of the obtained functions in order to check which microbiological variables were better for discriminating among the dierent stages. What microbiological variable is valid to classify dierent stages. In stepwise discriminant function analysis, a model of discrimination is built step-by-step incorporating variables to the model in successive steps. Specically, at the rst step all variables are reviewed and evaluated to determine which one will contribute most to the discrimination between stages. That variable is included in the model, and the next step starts again (reviewing and evaluating the rest of the variables) (StatSoft, 2002).

[G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

of Dmax which indicate the dierence between the samn, 2001; Statpled and theoretical distribution (Mart Soft, 2002; Vives-Rego, Resina, Comas, Loren, & , 2003) and it can be seen that these values are Julia not signicant (p < 0.05) in all cases. In other words, the dierence is not signicant (p < 0.05) and therefore the data show a normal distribution since null hypothesis is accepted (H0 = normal distribution of samples). Table 2 shows the statistical values of contamination of the samples at dierent sampling stages ([G], [W] and [A]). A decrease in contamination in stage [W] compared to stage [G], and in stage [A] compared to [W] is observed in all the cases. This decrease in contamination is due to the washing with pressurized water and to the low temperatures (between 6 and +2 C) as was lez-Miret et al., made clear in previous studies (Gonza lez-Miret, Alonso, & Heredia, 2000). It is 1998, Gonza observed that Pseudomonas is the least aected variable in both stages (decreases of 0.2 log cfu/g due to washing stage, and 0.1 log cfu/g due to air chilling stage). Enterobacteriaceae is the microbiological variable most affected by washing stage (decreases of 0.6 log ufc/g). The application of washing with pressurized water and low temperatures during a short period of time mainly have an eect on Enterobacteriaceae, and Pseudomonas is the least aected of the three microbiological variables studied. This can be due to the psychrotrophic nature of Pseudomonas. Assuming that the distribution is normal, univariate Analysis of Variance (ANOVA) of each transformed variable was carried out, with stages as a xed factor of repeated measure, and carcasses (chickens) as a random eect. Results obtained by means of the General Linear Model regression (GLM) (StatSoft, 2002) are

3. Results and discussion Transformation of raw data was conducted to approach normal distribution. Log 10 of the raw data was used. The KolmogorovSmirnovLilliefors test was used to evaluate the normality of every microbiological variable transformed at each stage, obtaining the results shown in Table 1. This table shows the values

Table 2 Mean and standard deviations for microbiological variables Variable Log Tc Log Ps Log Eb [G] 5.27 (0.33) 3.51 (0.42) 4.18 (0.31) [W] 5.00 (0.31) 3.34 (0.33) 3.54 (0.31) [A] 4.80 (0.30) 3.20 (0.26) 3.40 (0.21)

[G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

 lez-Miret et al. / Food Control 17 (2006) 935941 M.L. Gonza Table 3 Results of univariate ANOVA for microbiological variables Variables Log Tc Log Ps Log Eb Eect Stage Stage Stage F 13.07 3.97 47.65 p-Level 0.0000 0.0261 0.0000 g2 0.362 0.153 0.684

939

Table 5 Results of ANOVA with two repeated measure factors: stage and microbiological variable (p-level < 0.05 is signicant) Eect All stages included F Stage Microbiological variable (MV) Stage MV 24.74 840.29 11.35 p-Level 0.0000 0.0000 0.0000 g2 0.529 0.974 0.340 [G] excluded F 7.04 817.82 0.38 p-Level 0.014 0.000 0.685 g2 0.234 0.973 0.160

Tc = Total count; Ps = Pseudomonas; (p-level < 0.05 is signicant).

Eb = Enterobacteriaceae

presented in Table 3. Signicant dierences (p < 0.05) for all the microbiological variables among the means of the three stages ([G], [W] and [A]) were detected. The size of the eect (g2) was quite high for Eb and very low for Ps, meaning that Enterobacteriaceae has an important inuence on the response. The Bonferroni test was used to determine signicant dierences among the means of the three groups ([G], [W] and [A]). Pairs comparison using this test showed that Tc and Eb presented a signicantly higher mean at the stage [G] than at the stage [W], but [W] and [A] do not dier signicantly for Tc, Ps and Eb (Table 4). This could indicate that the decontamination eect of washing with pressurized water is very important, since the clean zone begins from this stage; later, air chilling maintains the decrease in the contamination as carcasses come out of the shower, due to the control of temperature under adequate limits. Cold temperatures ability to control microbial growth is well known. If the air chiller does not use cold air, microbial growth would have surely occurred. It could be expected an increase of the number of Pseudomonas on the carcasses after air chilling because of their psychrotrophic behavior. However, this increase does not occurs and this could be due to the air chilling stage is a rapid cooling process (a great decrease in temperature in a short period of time), and not prolonged storage in a refrigeration vault, a situation in which this lez-Miret microorganism (Ps) usually increases (Gonza et al., 2000). A 3(stages) 3(microbiological variables) repeated measures ANOVA was carried out with cases (chickens) as a random factor, for the purpose of establishing dif-

ferences between the stages and the three microbiological variables, as well as to determine the interaction between this factor and the stages (StatSoft, 2002). Results are shown in Table 5. Signicant dierences (p < 0.05) among the stages were found, as well as among the microbiological variables and between both factors. Pairs comparison using Bonferroni test (Table 6) indicate that the stage [G] presents a signicantly higher mean than stage [W], and the mean of [W] is higher than the one in stage [A], although it is not signicant (p < 0.05). Comparison between pairs of microbiological variables (Table 7) revealed that the means were signicantly dierent (p < 0.05) from each other, Tc showed the biggest mean and Ps the smallest one. The comparison between pairs of stages regarding each one of the microbiological variables (Table 8) showed that all the means of the microbiological variables decrease signicantly between [G] and [W], except Ps. This decrease also exists between [W] and [A] for all microbiological variables studied, but is not signicant (p < 0.05). The eect of the stage was a still signicant issue when stage [G] was not included to apply ANOVA, but the size of this eect diminishes a lot (g2 = 0.234)
Table 6 p-Level (<0.05 is signicant) of comparison [G]/[W], [W]/[A], and [G]/[A] using Bonferroni test Stage [G]/[W] [W]/[A] [G]/[A] p-Level 0.036 0.533 0.001

[G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling. Table 4 p-Level (<0.05 is signicant) of comparison [G]/[W], [W]/[A], [G]/[A] and [G]/[W]/[A] for microbiological variables using Bonferroni test Stage Variables Log Tc [G]/[W] [W]/[A] [G]/[A] [G]/[W]/[A] 0.026 0.067 0.000 0.000 Log Ps 0.987 0.241 0.024 0.026 Log Eb 0.000 0.287 0.000 0.000 Table 7 p-Level (<0.05 is signicant) of comparison between microbiological variables using Bonferroni test Stage Log Tc/ Log Ps Log Ps / Log Eb Log Tc/ Log Eb p-Level 0.036 0.533 0.001

[G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

940

lez-Miret et al. / Food Control 17 (2006) 935941 M.L. Gonza Table 11 Results of t-test of related group [G]/[W] for microbiological variables Dependent variable Log Tc Log Ps Log Eb p-Level 0.001 0.120 0.000

Table 8 p-Level (<0.05 is signicant) of comparison [G]/[W], [W]/[A] and [G]/ [A] for microbiological variables using Bonferroni test Stage Variables Log Tc [G]/[W] [W]/[A] [G]/[A] 0.020 0.149 0.000 Log Ps 1.000 0.640 0.011 Log Eb 0.000 1.000 0.000

[G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

[G] = after gutting; [W] = after washing with pressurized water; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae (plevel < 0.05 is signicant).

and the eect of the interaction of the stage with the microbiological variables disappears (Table 5). Therefore, the stage [G]/[W] is decisive on the production chain, [G] has an important inuence on the response. In this case the Bonferroni test (Table 9) indicates that the stage [W] presents a higher mean than stage [A] but this dierence was not signicant (p < 0.05). The comparison between pairs of microbiological variables revealed that all the means were signicantly dierent (p < 0.05). The comparison between pairs of stages ([W]/[A]) regarding each one of the microbiological variables (Table 10) showed that all the means of the microbiological variables decrease signicantly between [W] and [A], except Eb. Observing the estimates of the size of the eect (g2) presented in Table 5, one can see that the size of the effect of the stage and of the interaction of the stage with the microbiological variable decreases notably when the point [G] is excluded from the study. However, the decrease in size of the eect for microbiological variables is less important. A t-student test for the related groups [G] and [W] was carried out to determine the eect of the washing
Table 9 p-Level (<0.05 is signicant) of comparison [W]/[A] between microbiological variables using Bonferroni test with [G] excluded Stage [W]/[A] Log Tc/Log Ps Log Ps/Log Eb Log Tc/Log Eb p-Level 0.139 0.000 0.007 0.000

[W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

n, stage on the contamination of the carcasses (Mart 2001; StatSoft, 2002). Looking at the results presented in Table 11 one can nd that means of variables Eb and Tc before washing with pressurized water are significantly (p < 0.05) dierent from those obtained after the cleaning. These results were in accordance with those obtained with the Bonferroni test of the repeated measures ANOVA. The stage of washing with pressurized water is very important on the production of poultry meat to decrease the supercial contamination of Eb and Tc signicantly (p < 0.05). The stepwise discriminant analysis (SDA) was used in order to explore the extent to which the studied microbiological variables are able to discriminate between stages (the higher the standardized coecient the higher contribution to the discrimination between groups) (StatSoft, 2002; Tabachnick & Fidell, 1983). Forward stepwise discriminant analysis was carried out and the results are shown in Table 12. It can be observed that Eb is able to discriminate between stages [G]/[W], [G]/ [A] and [G]/[W]/[A] with high signicance levels (p < 0.01, r > 0.7). Eb is the microbiological variable that presents the highest contribution to the discrimination between all groups except the [W]/[A] relationship. In this case Tc is the variable that contributes in a significant way, although with a very small correlation (0.317). However, Ps was not included in the discrimination function. These results indicate that microbiological similarity exists in the stages [W]/[A] and that Eb is the microbiological variable that presents the highest contribution to the discrimination in the [G]/[W] relationship with quite high correlations. In conclusion it can be established that the washing stage is a CCP to be checked in an adequate HACCP
Table 12 Results of comparison between groups using stepwise discriminant analysis Stage [G]/[W] [W]/[A] [G]/[A] [G]/[W]/[A] Variable Log Log Log Log Eb Tc Eb Eb p-Level 0.000 0.037 0.000 0.000 Canonical correlation 0.723 0.284 0.831 0.777

Table 10 p-Level (<0.05 is signicant) of comparison [W]/[A] for microbiological variables using Bonferroni test with [G] excluded Stage Variables Log Tc [W]/[A] 0.003 Log Ps 0.025 Log Eb 0.094

[W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae.

[G] = after gutting; [W] = after washing with pressurized water; [A] = after air chilling; Tc = Total count; Ps = Pseudomonas; Eb = Enterobacteriaceae) (p-level < 0.05 is signicant.

 lez-Miret et al. / Food Control 17 (2006) 935941 M.L. Gonza

941

system, since contamination decreases signicantly due to the eect of this stage. The air chilling stage can be eective for preventing the multiplication of the microbial ora when properly controlled. It is reasonable to assume that failure of temperature control in the air chiller would have result in microbial growth and a potential hazard. In fact, the air temperature in the chiller is critical to controlling microbial growth and as such, the chiller air temperature would be considered a CCP in a HACCP system.

Acknowledgements The authors are indebted to SADA p.a. SUR, S.A., a company of NUTRECO (Sevilla, Spain) for supplying the samples.

References
, C. (1985). El pollo de carne, Sistemas de explotacio n y te cnicas Buxade n. Madrid: Ediciones Mundi-Prensa. de produccio Cates, S. C., Anderson, D. W., Karns, S. A., & Brown, P. A. (2001). Traditional versus hazard analysis and critical control point-based inspection: results from a poultry slaughter project. Journal of Food Protection(64), 826832. Cuni, P. (Ed.). (1995). Ocial methods of analysis of the association of ocial analytical chemistry international. Arlington: Virginia. De Boer, E. (1998). Update on media for isolation of Enterobacteriaceae from foods. International Journal of Food Microbiology(45), 4353. European Community Directive. (1993). 93/43/EEG. Forsythe, S. J., & Hayes, P. R. (2002). Higiene de los alimentos, a y HACCP. Zaragoza: Editorial Acribia. microbiolog lez-Miret, M. L., Alonso, S., & Heredia, F. J. (1998). Sistemas Gonza n de alimentos. Efecto del Continuos de Control en la Produccio nivel de cloro en el control de calidad de carne de ave. Alimentaria(Noviembre), 9198. lez-Miret, M. L., Alonso, S., & Heredia, F. J. (2000). Sistemas Gonza n de alimentos. (II). Efecto Continuos de Control en la Produccio del refrigerado en el control de calidad de carne de ave. Alimentaria(Marzo), 3135. lez-Miret, M. L., Coello, M. T., Alonso, S., & Heredia, F. J. Gonza (2001). Validation of parameters in HACCP verication using univariate and multivariate statistics. Application to the nal phases of poultry meat production. Food Control(12), 261268. lez-Miret, M. L., Escudero, M. L., Alonso, S., & Heredia, F. J. Gonza n de (2001). Sistemas Continuos de Control en la Produccio alimentos. (III). Efecto del manipulado en el control de calidad de carne de ave. Alimentaria(Marzo), 4347. Grau, F. H. (1986). Microbial ecology of meat and poultry. In A. M. Pearson & T. R. Dutson (Eds.). Advances in Meat Research (vol. 2, pp. 136). Connecticut: AVI Publishing Company.

lisis de Riesgos y Puntos Cr ticos. Su ICMSF (1991). El sistema de Ana n a las industrias de alimentos. Zaragoza: Editorial Acribia, aplicacio S.A. todos multivariados aplicados al ana lisis de Johnson, D. E. (2000). Me datos. Madrid: International Thomson Editores, S.A. Kvenberg, J. E., & Schwalm, D. J. (2000). Use of microbial data for hazard analysis and critical control point verication. Food and drug administration perspective. Journal of Food Protection(63), 810814. Lehallec, C., & Colin, P. (1979). Bacterial ora of poultry: changes due to variation in ecological conditions during processing and storage. Archiv fu r Lebensmittelhygiene(30), 9598. n, Q. (2001). Curso de tratamiento de datos estad sticos con SPSS. Mart rides. Salamanca: Editorial Hespe McNamara, A. M. (1997). Generic HACCP application in broiler slaughter and processing. Journal of Food Protection(60), 579604. ctico. Mortimore, S., & Wallace, C. (1996). HACCP: Enfoque pra Zaragoza: Editorial Acribia, S.A. National Advisory Committee on Microbiological Criteria for Food (1998). Hazard analysis and critical control points principles and application guidelines. Journal of Food Protection, 61, 12461259. stica. Madrid: Norman, G. R., & Streiner, D. L. (1996). Bioestad Mosby/Doyma Libros, S.A. Times Mirror International Publishers. a Alimentaria. Metodolog a anal Pascual, M. R. (1992). Microbiolog az de Santos. tica para alimentos y bebidas. Madrid: Ed. D Robach, M.C. (1996). Microbiological indicators for meat and poultry. In IFT annual meeting: Book of abstracts. ISSN 10821236, p. 105. Sandrou, D. K., & Arvanitoyannis, I. S. (1999). Implementation of hazard analysis critical control point in the meat and poultry industry. Food Review International(15), 265307. Snijders, J. M. A., & Van Knapen, F. (2002). Prevention of human diseases by an integrated quality control system. Livestock Production Science(76), 203206. Sofos, J. N. (1994). Microbial growth and its control in meat, poultry and sh. In A. M. Pearson & T. R. Dutson (Eds.). Advances in meat research (vol. 9, pp. 359403). London: Blackie Academic & Professional. SPSS Inc. (1999). Statistical Package for the Social Sciences SPSS v. 10.0 for Windows. Chicago, IL. StatSoft Inc. (1995). Statistica 5.5 for Windows Computer Program Manual, Tulsa, OK. StatSoft, Inc., 2300 East 14th Street, Tulsa, OK 74104. StatSoft, Inc. (2002). Electronic Statistics Textbook. Tulsa, OK: StatSoft. Available at http://www.statsoft.com/textbook/ stathome.html. Tabachnick, B. G., & Fidell, L. S. (1983). Using multivariate statistics. New York: Harper & Row Publishers. Vanderzant, C., & Splittstoesser, F. (Eds.). (1992). Compendium of methods for the microbiological examination of foods. Washington, D.C: American Public Health Association, Inc. n de carne de pollo. Madrid: Real Vaquerizo, J. M. (1991). Produccio Escuela de Avicultura. , O. (2003). Vives-Rego, J., Resina, O., Comas, J., Loren, G., & Julia Statistical analysis and biological interpretation of the ow cytrometric heterogeneity observed in bacterial axenic cultures. Journal of Microbiological Methods(53), 4350.