Role of Salmonella Newport cell surface structures on bacterial attachment and transfer during cucumber peeling.

Fresh cucumbers have been recognized as a vehicle in foodborne disease outbreaks since several multistate outbreaks of salmonellosis linked to fresh cucumbers occurred in the United States. Little is known about how microbial cell surface characteristics that are known to affect adhesion can influence bacterial cross-contamination and transfer. This study investigated the role of S. Newport cell surface components on bacterial attachment and transfer in cucumbers. Wild type Salmonella Newport and its transposon mutants were used to inoculate cucumbers. Attachment strength of S. Newport wild type to cucumber was not significantly different than that of mutants. Log10 percent transfer of mutant strains to edible flesh was not different from the wild type. Significantly less wild type Salmonella remained on the peel and transferred to the peeler than one mutant did, but not the other. Our results suggest that while curli and cellulose enhance Salmonella attachment to surface of cucumbers, there appear to be other mechanisms and factors that govern Salmonella transfer in cucumbers.

Bootstrap parametric GB2 and bootstrap nonparametric distributions for studying shiga toxin-producing Escherichia coli strains growth rate variability.

Shiga toxin-producing Escherichia coli (VTEC) strains, including the serotype O157:H7, are considered foodborne pathogens. Transmission occurs through consumption of undercooked meat, unpasteurized dairy products, vegetables, or contaminated water. The variability of pathogenic and nonpathogenic E. coli strains growth parameters at different temperatures and in different media was studied. Bootstrap parametric (Generalized Beta of the Second Kind, GB2) or nonparametric models were used. GB2 estimations show increased growth rates and shortened lag times with increased temperature, as expected. Similar estimations were obtained using the nonparametric model. Parametric and nonparametric growth rate confidence intervals were wider with increased temperature; lag times confidence intervals were wider with decreased temperature. The nonparametric method gives similar confidence intervals to the parametric method, confirming its suitability for growth parameters estimation. The estimations obtained from nonpathogenic E. coli strains approximate distributions from pathogenic E. coli strains.

Bayesian modeling of two- and three-species bacterial competition in milk.

Listeria monocytogenes is a well-known food-borne pathogen and is among the bacteria best adapted to grow at low temperatures. Psychrotrophic spoilage microorganisms present in milk and milk products are primarily in the genus Pseudomonas, and their numbers increase during cold storage leading to deterioration and/or spoilage. The nature of the competition in two- or three-species bacterial systems with L. monocytogenes, L. innocua, and P. fluorescens in skimmed milk at 7 or 14°C was studied. The Baranyi growth model was used to estimate the growth rate and the maximum population density of the three microorganisms for each strain in single cultures or in two- or three-strains co-cultures. The highest Listeria populations were achieved by pure cultures, decreasing in co-culture with P. fluorescens at both temperatures. A modified deterministic logistic model was applied which includes inhibition functions for single cultures, and two- or three-species cultures. A subsequent Bayesian approach was applied for modelling the bacterial interactions. There was not a direct correlation between the growth rate of P. fluorescens and its inhibitory effect on Listeria species. The use of some species from the natural food microflora to inhibit pathogen growth may be an important tool to enhance the safety of refrigerated foods such as milk and dairy products.

Surfactant concentration and type affects the removal of Escherichia coli from pig skin during a simulated hand wash.

The effect of surfactant type and concentration on a bland soap formulation's ability to remove bacteria from hands remains largely unstudied. Several combinations of surfactants and water were combined to test bacterial removal efficacy using a hand-washing device (two pieces of pig skin and a mechanical motor) to simulate a hand wash. A nalidixic acid-resistant, nonpathogenic strain of Escherichia coli (ATCC 11229) was used. Two anionic surfactants, sodium lauryl sulphate and sodium stearoyl lactylate, and two nonionic surfactants, poloxamer 407 and sorbitan monostearate, each in concentrations of 2, 5 and 10% were studied. A slight positive (r2  = 0·17) but significant (P = 0·03) correlation was observed between hydrophile-lipophile balance value and mean log reduction. No correlation was observed between pH of the treatment solution and the mean log reduction (r2  = 0·05, P = 0·25). A 10% sodium lauryl sulphate mixture showed the highest log reduction (x¯ = 1·1 log CFU reduction, SD = 0·54), and was the only treatment significantly different from washing with water (P = 0·0005). There was a correlation between increasing surfactant concentrations above the critical micelle concentration, and mean microbial reduction (r2  = 0·62, P = 0·001). SIGNIFICANCE AND IMPACT OF THE STUDY: This study characterizes the role of surfactants in removing microbes during a hand wash. Numerous studies address how surfactants support antimicrobial effect in soap, or cause irritation of skin, but no published studies show which surfactants are best for removing microbes. We used pig skin as a model for human skin and a lathering device to simulate a hand wash. A 10% sodium lauryl sulphate mixture was the only treatment significantly different from a water wash. There was a strong correlation between increasing surfactant concentrations above the critical micelle concentration and mean microbial reduction.

Assessing the effect of sodium dichloroisocyanurate concentration on transfer of Salmonella enterica serotype Typhimurium in wash water for production of minimally processed iceberg lettuce (Lactuca sativa L.).

UNLABELLED: This study evaluated the impact of sodium dichloroisocyanurate (5, 10, 20, 30, 40, 50 and 250 mg l(-1) ) in wash water on transfer of Salmonella Typhimurium from contaminated lettuce to wash water and then to other noncontaminated lettuces washed sequentially in the same water. Experiments were designed mimicking the conditions commonly seen in minimally processed vegetable (MPV) processing plants in Brazil. The scenarios were as follows: (1) Washing one inoculated lettuce portion in nonchlorinated water, followed by washing 10 noninoculated portions sequentially. (2) Washing one inoculated lettuce portion in chlorinated water followed by washing five noninoculated portions sequentially. (3) Washing five inoculated lettuce portions in chlorinated water sequentially, followed by washing five noninoculated portions sequentially. (4) Washing five noninoculated lettuce portions in chlorinated water sequentially, followed by washing five inoculated portions sequentially and then by washing five noninoculated portions sequentially in the same water. Salm. Typhimurium transfer from inoculated lettuce to wash water and further dissemination to noninoculated lettuces occurred when nonchlorinated water was used (scenario 1). When chlorinated water was used (scenarios 2, 3 and 4), no measurable Salm. Typhimurium transfer occurred if the sanitizer was ≥10 mg l(-1) . Use of sanitizers in correct concentrations is important to minimize the risk of microbial transfer during MPV washing. SIGNIFICANCE AND IMPACT OF THE STUDY: In this study, the impact of sodium dichloroisocyanurate in the wash water on transfer of Salmonella Typhimurium from inoculated lettuce to wash water and then to other noninoculated lettuces washed sequentially in the same water was evaluated. The use of chlorinated water, at concentration above 10 mg l(-1) , effectively prevented Salm. Typhimurium transfer under several different washing scenarios. Conversely, when nonchlorinated water was used, Salm. Typhimurium transfer occurred in up to at least 10 noninoculated batches of lettuce washed sequentially in the same water.

Assessing the Performance of Clostridium perfringens Cooling Models for Cooked, Uncured Meat and Poultry Products.

Heat-resistant spores of Clostridium perfringens may germinate and multiply in cooked meat and poultry products when the rate and extent of cooling does not occur in a timely manner. Therefore, six cooling models (PMP 7.0 broth model; PMIP uncured beef, chicken, and pork models; Smith-Schaffner version 3; and UK IFR ComBase Perfringens Predictor) were evaluated for relative performance in predicting growth of C. perfringens under dynamic temperature conditions encountered during cooling of cooked, uncured meat and poultry products. The predicted growth responses from the models were extensively compared with those observed in food. Data from 188 time-temperature cooling profiles (176 for single-rate exponential cooling and 12 for dual-rate exponential cooling) were collected from 17 independent sources (16 peer-reviewed publications and one report) for model evaluation. Data were obtained for a variety of cooked products, including meat and poultry slurries, ground meat and poultry products with and without added ingredients (e.g., potato starch, sodium triphosphate, and potassium tetrapyrophosphate), and processed products such as ham and roast beef. Performance of the models was evaluated using three sets of criteria, and accuracy was defined within a 1- to 2-log range. The percentages of accurate, fail-safe, or fail-dangerous predictions for each cooling model differed depending on which criterion was used to evaluate the data set. Nevertheless, the combined percentages of accurate and fail-safe predictions based on the three performance criteria were 34.66 to 42.61% for the PMP 7.0 beef broth model, 100% for the PMIP cooling models for uncured beef, uncured pork and uncured chicken, 80.11 to 93.18% for the Smith-Schaffner cooling model, and 74.43 to 85.23% for the UK IFR ComBase Perfringens Predictor model during single-rate exponential chilling. Except for the PMP 7.0 broth model, the other five cooling models (PMIP, Smith-Schaffner, and UK IFR ComBase) are useful and reliable tools that food processors and regulatory agencies can use to evaluate the safety of cooked or heat-treated uncured meat and poultry products exposed to cooling deviations or to develop customized cooling schedules.

Modeling the influence of temperature, water activity and water mobility on the persistence of Salmonella in low-moisture foods.

Salmonella can survive in low-moisture foods for long periods of time. Reduced microbial inactivation during heating is believed to be due to the interaction of cells and water, and is thought to be related to water activity (a(w)). Little is known about the role of water mobility in influencing the survival of Salmonella in low-moisture foods. The aim of this study was to determine how the physical state of water in low-moisture foods influences the survival of Salmonella and to use this information to develop mathematical models that predict the behavior of Salmonella in these foods. Whey protein powder of differing water mobilities was produced by pH adjustment and heat denaturation, and then equilibrated to aw levels between 0.19±0.03 and 0.54±0.02. Water mobility was determined by wide-line proton-NMR. Powders were inoculated with a four-strain cocktail of Salmonella, vacuum-sealed and stored at 21, 36, 50, 60, 70 and 80°C. Survival data was fitted to the log-linear, the Geeraerd-tail, the Weibull, the biphasic-linear and the Baranyi models. The model with the best ability to describe the data over all temperatures, water activities and water mobilities (f(test)

Development of a risk-ranking framework to evaluate potential high-threat microorganisms, toxins, and chemicals in food.

Through a cooperative agreement with the U.S. Food and Drug Administration, the Institute of Food Technologists developed a risk-ranking framework prototype to enable comparison of microbiological and chemical hazards in foods and to assist policy makers, risk managers, risk analysts, and others in determining the relative public health impact of specific hazard-food combinations. The prototype is a bottom-up system based on assumptions that incorporate expert opinion/insight with a number of exposure and hazard-related risk criteria variables, which are propagated forward with food intake data to produce risk-ranking determinations. The prototype produces a semi-quantitative comparative assessment of food safety hazards and the impacts of hazard control measures. For a specific hazard-food combination the prototype can produce a single metric: a final risk value expressed as annual pseudo-disability adjusted life years (pDALY). The pDALY is a harmonization of the very different dose-response relationships observed for chemicals and microbes. The prototype was developed on 2 platforms, a web-based user interface and an Analytica(R) model (Lumina Decision Systems, Los Gatos, Calif., U.S.A.). Comprising visual basic language, the web-based platform facilitates data input and allows use concurrently from multiple locations. The Analytica model facilitates visualization of the logic flow, interrelationship of input and output variables, and calculations/algorithms comprising the prototype. A variety of sortable risk-ranking reports and summary information can be generated for hazard-food pairs, showing hazard and dose-response assumptions and data, per capita consumption by population group, and annual p-DALY.

Extracting survival parameters from isothermal, isobaric, and "iso-concentration" inactivation experiments by the "3 end points method".

Theoretically, if an organism's resistance can be characterized by 3 survival parameters, they can be found by solving 3 simultaneous equations that relate the final survival ratio to the lethal agent's intensity. (For 2 resistance parameters, 2 equations will suffice.) In practice, the inevitable experimental scatter would distort the results of such a calculation or render the method unworkable. Averaging the results obtained with more than 3 final survival ratio triplet combinations, determined in four or more treatments, can remove this impediment. This can be confirmed by the ability of a kinetic inactivation model derived from the averaged parameters to predict survival patterns under conditions not employed in their determination, as demonstrated with published isothermal survival data of Clostridium botulinum spores, isobaric data of Escherichia coli under HPP, and Pseudomonas exposed to hydrogen peroxide. Both the method and the underlying assumption that the inactivation followed a Weibull-Log logistic (WeLL) kinetics were confirmed in this way, indicating that when an appropriate survival model is available, it is possible to predict the entire inactivation curves from several experimental final survival ratios alone. Where applicable, the method could simplify the experimental procedure and lower the cost of microbial resistance determinations. In principle, the methodology can be extended to deteriorative chemical reactions if they too can be characterized by 2 or 3 kinetic parameters.

Evidence for quorum sensing in Clostridium botulinum 56A.

AIMS: Experiments were designed to detect quorum-sensing signals produced by Clostridium botulinum. METHODS AND RESULTS: Clostridium botulinum 56A cell-free supernatants obtained at the end of lag phase, the mid-exponential phase and early stationary phase of growth were assayed for bioluminescence in the Vibrio harveyi quorum-sensing assay system. Twelve and 16-h culture supernatants induced bioluminescence in the auto-inducer 2 (AI-2) but not the auto-inducer 1 (AI-1) assay. Intra-species quorum sensing was also assayed as the ability of the supernatants to promote spore germination and outgrowth in a microtitre plate system. Spore populations exposed to C. botulinum supernatant from the end of lag phase became positive for growth sooner than controls. CONCLUSIONS: The influence of cell-free supernatant on ungerminated spores and detection of bioluminescence in the AI-2 assay are evidence for a signalling molecule(s) and provide a first step in characterizing C. botulinum quorum sensing. SIGNIFICANCE AND IMPACT OF THE STUDY: This study suggests that spores do not behave independently of each other and may explain the inocula size effects observed in challenge studies. Whether AI-2 production in C. botulinum serves as an inter-species signal or as a detoxification mechanism remains to be determined.

Solute-specific effects of osmotic stress on Staphylococcus aureus.

AIM: To determine if cell death from osmotic stress is because of lack of sufficient energy to maintain cell metabolism. Additionally, the solute-specific effect of five humectants on bacterial osmoregulation and cell survival was examined. METHODS AND RESULTS: Staphylococcus aureus was placed into 84% relative humidity (RH) broth (five humectants used individually). ATP, ADP and cell viability measurements were determined over time. The results indicate that ATP is not the limiting factor for cell survival under excessive osmotic stress. Although the same RH was achieved with various humectants, the rates of cell death varied greatly as did the sensitivities of the cell populations to osmotic stress. CONCLUSIONS: The results from this study provide strong evidence that mechanisms of osmotic inactivation depend on the solute. The molecular mobility of the system may be an important means to explain these differences. SIGNIFICANCE AND IMPACT OF THE STUDY: By bringing together an understanding of solute-specific effects, microbial physiology and genetics, the mechanisms of inactivation of micro-organisms by solute-specific osmotic stress may be elucidated, and this knowledge may then be exploited to ensure the production of high quality, safe foods.

Inoculum size influences bacterial cross contamination between surfaces.

Many factors have been shown to influence bacterial transfer between surfaces, including surface type, bacterial species, moisture level, pressure, and friction, but the effect of inoculum size on bacterial transfer has not yet been established. Bacterial cross contamination rates during performance of common food service tasks were previously determined in our laboratory using nalidixic acid-resistant Enterobacter aerogenes. Eight different transfer rates were determined, each involving a minimum of 30 volunteers. The influence of source inoculum level on the percentage of bacteria transferred (percent transfer rates) and log10 CFU per recipient surface was determined using statistical analysis. The effect of inoculum size on transfer rate was highly statistically significant (P < 0.0001) for all transfer rate data combined (352 observations) and for each individual cross contamination rate, except for data on contamination via transfer from chicken to hand through a glove barrier (P = 0.1643). Where inoculum size on the source was greater, transfer rates were lower, and where inoculum size on the source was less, transfer rates were higher. The negative linear trend was more obvious for activities that had a larger range of inoculum sizes on the source surface. This phenomenon has serious implications for research seeking to determine bacterial cross contamination rates, since the different transfer efficiencies that were previously shown to be associated with certain activities may actually be the result of differing initial inoculum levels. The initial inoculum size on the source and the amount of bacteria transferred must both be considered to accurately determine bacterial transfer rates.

Computer simulation of Clostridium botulinum strain 56A behavior at low spore concentrations.

It is generally assumed that spore behavior is independent of spore concentration, but recently published mathematical models indicate that this is not the case. A Monte Carlo simulation was employed in this study to further examine the independence assumption by evaluating the inherent variance in spore germination data. All simulations were carried out with @Risk software. A total of 500 to 4,000 iterations were needed for each simulation to reach convergence. Lag time and doubling time from a higher inoculum concentration were used to simulate the time to detection (TTD) at a lower inoculum concentration under otherwise identical environmental conditions. The point summaries of the simulated and observed TTDs were recorded for the 26 simulations, with kinetic data at the target inoculum concentration. The ratios of the median (R(m) = median(obs)/median(sim)) and 90% range (R(r) = 90% range(obs)/90% range(sim)) were calculated. Most R(m) and R(r) values were greater than one, indicating that the simulated TTDs were smaller and more homogeneous than the observed ones. R(r) values departed farther from one than R(m) values. Ratios obtained when simulating 1 spore with 10,000 spores deviated the farthest from one. Neither ratio was significantly different from the other when simulating 1 spore with 100 spores or simulating 100 spores with 10,000 spores. When kinetic data were not available, the percent positive observed at the 95th percentile of the simulated TTDs was obtained. These simulation results confirmed that the assumption of independence between spores is not valid.

Time-to-detection, percent-growth-positive and maximum growth rate models for Clostridium botulinum 56A at multiple temperatures.

We previously developed models for the influence of inoculum size on the growth kinetics (time-to-detection and maximum growth rate) and percent-growth-positive samples of Clostridium botulinum 56A with factors of inoculum size (1, 100, and 10,000 spores/sample). pH (5.5. 6.0 and 6.5) and sodium chloride concentration (0.5%, 2% and 4%) at 30 degrees C. In this present study, data were collected at two more temperatures (15 and 22 degrees C), making the final design a complete 3 X 3 X 3 X 3 factorial with a total of 81 conditions. Growth was followed hourly as change in A620. The Gompertz equation was fit to the growth data, and the parameters derived were used to calculate the maximum growth rate and time-to-detection. Linear regression with polynomial terms was used to analyze the effect of environmental factors on time-to-detection and maximum growth rate. Logistic regression with polynomial terms was used to analyze the data for percent-growth-positive. Despite the fact that the variance is larger in this extended data set (which includes two temperatures that are further away from the optimum), the inoculum size effect is clearly demonstrated. When inoculum size increased, the percent-growth-positive samples increased and the time-to-detection decreased. When the inoculum was 1000 spores/sample or higher, little additional effect on time-to-detection was observed. Inoculum size might influence results through simple probability or quorum sensing. Our results show that the observed effect of inoculum size from the previous report at a single temperature is not restricted to a specific growth condition, but rather a general phenomenon. The maximum growth rate was independent of inoculum levels, confirming our previous results.

Bacillus megaterium spore germination is influenced by inoculum size.

AIMS: The effect of spore density on the germination (time-to-germination, percent germination) of Bacillus megaterium spores on tryptic soy agar was determined using direct microscopic observation. METHODS AND RESULTS: Inoculum size varied from approximately 10(3) to 10(8) cfu ml(-1) in a medium where pH=7 and the sodium chloride concentration was 0.5% w/v. Inoculum size was measured by global inoculum size (the concentration of spores on a microscope slide) and local inoculum size (the number of spores observed in a given microscope field of observation). Both global and local inoculum sizes had a significant effect on time-to-germination (TTG), but only the global inoculum size influenced the percentage germination of the observed spores. CONCLUSIONS: These results show that higher concentrations of Bacillus megaterium spores encourage more rapid germination and more spores to germinate, indicating that low spore populations do not behave similarly to high spore populations. SIGNIFICANCE AND IMPACT OF THE STUDY: A likely explanation for the inoculum size-dependency of germination would be chemical signalling or quorum sensing between Bacillus spores.

Modelling bacterial spoilage in cold-filled ready to drink beverages by Acinetobacter calcoaceticus and Gluconobacter oxydans.

AIMS: Mathematical models were created which predict the growth of spoilage bacteria in response to various preservation systems. METHODS AND RESULTS: A Box-Behnken design included five variables: pH (2.8, 3.3, 3.8), titratable acidity (0.20%, 0.40%, 0.60%), sugar (8.0, 12.0, 16.0 * Brix), sodium benzoate concentration (100, 225, 350 ppm), and potassium sorbate concentration (100, 225, 350 ppm). Duplicate samples were inoculated with a bacterial cocktail (100 microl 50 ml(-1)) consisting of equal proportions of Acinetobacter calcoaceticus and Gluconobacter oxydans (5 x 10(5) cfu ml(-1) each). Bacteria from the inoculated samples were enumerated on malt extract agar at zero, one, two, four, six, and eight weeks. CONCLUSION: The pH, titratable acidity, sugar content, sodium benzoate, and potassium sorbate levels were all significant factors in predicting the growth of spoilage bacteria. SIGNIFICANCE AND IMPACT OF THE STUDY: This beverage spoilage model can be used to predict microbial stability in new beverage product development and potentially reduce the cost and time involved in microbial challenge testing.

Glove barriers to bacterial cross-contamination between hands to food.

Human hands are an important source of microbial contamination of foods. However, published data on the effectiveness of handwashing and glove use in a foodservice setting are limited. Bacterial transfer through foodservice quality gloves was quantified using nalidixic acid-resistant Enterobacter aerogenes (a nonpathogenic surrogate with attachment characteristics similar to Salmonella). Five transfer rates were determined: chicken to bare hand, chicken to hand through gloves, bare hand to lettuce, hand to lettuce through gloves (with low inoculum on hands), and hand to lettuce through gloves (with high inoculum on hands). At least 30 observations were made for each percent transfer rate using 30 individual volunteers. The logarithm of percent transfer data were then fit to distributions: chicken to bare hand, normal (0.71, 0.42); chicken to hand through gloves, gamma (5.91, 0.40, -5.00); bare hand to lettuce, logistic (1.16, 0.30); hand to lettuce through gloves (low inoculum), normal (0.35, 0.88); hand to lettuce through gloves (high inoculum), normal (-2.52, 0.61). A 0.01% transfer was observed from food to hands and from hands to food when subjects wore gloves and a 10% transfer was observed without a glove barrier. These results indicate that gloves are permeable to bacteria although transfer from hands to food through a glove barrier was less than without a glove barrier. Our results indicate that gloves may reduce both bacterial transfer from food to the hands of foodservice workers and in subsequent transfer from hands back to food.

Modeling the survival of Escherichia coli O157:H7 in apple cider using probability distribution functions for quantitative risk assessment.

Outbreaks of foodborne illness from apple cider have prompted research on the survival of Escherichia coli O157:H7 in this food. Published results vary widely, potentially due to differences in E. coli O157:H7 strains, enumeration media, and other experimental considerations. We developed probability distribution functions for the change in concentration of E. coli O157:H7 (log CFU/day) in cider using data from scientific publications for use in a quantitative risk assessment. Six storage conditions (refrigeration [4 to 5 degrees C]; temperature abuse [6 to 10 degrees C]; room temperature [20 to 25 degrees C]; refrigerated with 0.1% sodium benzoate, 0.1% potassium sorbate, or both) were modeled. E. coli survival rate data for all three unpreserved cider storage conditions were highly peaked, and these data were fit to logistic distributions: ideal refrigeration, logistic (-0.061, 0.13); temperature abuse, logistic (-0.0982, 0.23); room temperature, logistic (-0.1, 0.29) and uniform (-4.3, -1.8), to model the very small chance of extremely high log CFU reductions. There were fewer published studies on refrigerated, preserved cider, and these smaller data sets were modeled with beta (4.27, 2.37) x 2.2 - 1.6, normal (-0.2, 0.13), and gamma (1.45, 0.6) distributions, respectively. Simulations were run to show the effect of storage on E. coli O157:H7 during the shelf life of apple cider. Under every storage condition, with and without preservatives, there was an overall decline in E. coli O157:H7 populations in cider, although a small fraction of the time a slight increase was seen.

Comparison of logistic regression and linear regression in modeling percentage data.

Percentage is widely used to describe different results in food microbiology, e.g., probability of microbial growth, percent inactivated, and percent of positive samples. Four sets of percentage data, percent-growth-positive, germination extent, probability for one cell to grow, and maximum fraction of positive tubes, were obtained from our own experiments and the literature. These data were modeled using linear and logistic regression. Five methods were used to compare the goodness of fit of the two models: percentage of predictions closer to observations, range of the differences (predicted value minus observed value), deviation of the model, linear regression between the observed and predicted values, and bias and accuracy factors. Logistic regression was a better predictor of at least 78% of the observations in all four data sets. In all cases, the deviation of logistic models was much smaller. The linear correlation between observations and logistic predictions was always stronger. Validation (accomplished using part of one data set) also demonstrated that the logistic model was more accurate in predicting new data points. Bias and accuracy factors were found to be less informative when evaluating models developed for percentage data, since neither of these indices can compare predictions at zero. Model simplification for the logistic model was demonstrated with one data set. The simplified model was as powerful in making predictions as the full linear model, and it also gave clearer insight in determining the key experimental factors.

Modeling the growth boundary of Staphylococcus aureus for risk assessment purposes.

Knowing the precise boundary for growth of Staphylococcus aureus is critical for food safety risk assessment, especially in the formulation of safe, shelf-stable foods with intermediate relative humidity (RH) values. To date, most studies and resulting models have led to the presumption that S. aureus is osmotolerant. However, most studies and resulting models have focused on growth kinetics using NaCl as the humectant. In this study, glycerol was used to investigate the effects of a glass-forming nonionic humectant to avoid specific metabolic aspects of membrane ion transport. The experiments were designed to produce a growth boundary model as a tool for risk assessment. The statistical effects and interactions of RH (84 to 95% adjusted by glycerol), initial pH (4.5 to 7.0 adjusted by HC1), and potassium sorbate (0, 500, or 1,000 ppm) or calcium propionate (0, 500, or 1,000 ppm) on the aerobic growth of a five-strain S. aureus cocktail in brain heart infusion broth were explored. Inoculated broths were distributed into microtiter plates and incubated at 37 degrees C over appropriate saturated salt slurries to maintain RH. Growth was monitored by turbidity during a 24-week period. Toxin production was explored by enterotoxin assay. The 1,280 generated data points were analyzed by SAS LIFEREG procedures, which showed all studied parameters significantly affected the growth responses of S. aureus with interactions between RH and pH. The resulting growth/no growth boundary is presented.