Modeling and Predicting the Simultaneous Growth of Escherichia coli O157:H7 and Ground Beef Background Microflora for Various Enrichment Protocols

Abstract

The simultaneous growth of Escherichia coli O157:H7 (O157) and the ground beef background microflora (BM) was described in order to characterize the effects of enrichment factors on the growth of these organisms. The different enrichment factors studied were basal medium (Trypticase soy broth and E. coli broth), the presence of novobiocin in the broth, and the incubation temperature (37°C or 40°C). BM and O157 kinetics were simultaneously fitted by using a competitive growth model. The simple competition between the two microfloras implied that O157 growth stopped as soon as the maximal bacterial density in the BM was reached. The present study shows that the enrichment protocol factors had little impact on the simultaneous growth of BM and O157. The selective factors (i.e., bile salts and novobiocin) and the higher incubation temperature (40°C) did not inhibit BM growth, and incubation at 40°C only slightly improved O157 growth. The results also emphasize that when the level of O157 contamination in ground beef is low, the 6-h enrichment step recommended in the immunomagnetic separation protocol (ISO EN 16654) is not sufficient to detect O157 by screening methods. In this case, prior enrichment for approximately 10 h appears to be the optimal duration for enrichment. However, more experiments must be carried out with ground beef packaged in different ways in order to confirm the results obtained in the present study for non-vacuum- and non-modified-atmosphere-packed ground beef.

Escherichia coli O157:H7 and other virulent Shiga toxin-producing E. coli serotypes, such as O26, O55, O91, O103, O111, O128, and O145, are an emerging cause of food-borne illness and have become a public health priority (13). In human infections, O157 and the other serotypes mentioned above may cause bloody diarrhea and, more precisely, may lead to hemolytic-uremic syndrome and to thrombotic and thrombocytopenic purpura in children and adults, respectively (33, 57). Illness is often linked to the consumption of contaminated or undercooked ground beef (6, 14, 15, 45, 46, 47, 49, 50, 51, 54). For instance, in Scotland 512 persons were infected by E. coli O157:H7 after consumption of beef, and 34 cases of hemolytic-uremic syndrome and 17 deaths were reported (18). Moreover, in samples of food responsible for diseases, the levels of contamination with E. coli O157:H7 are sometimes very low. For instance, a level of 4 CFU in 25 g of food caused a food-borne disease in a Japanese school in 1996 (53). Thus, very sensitive detection methods are required.

Different methods of screening for E. coli O157:H7 are now available, including genetic methods like PCR (9, 36, 38) and immunological methods like immunomagnetic separation (IMS) and enzyme-linked immunosorbent assays (7, 16, 37). These methods require an enrichment step to improve their limits of detection (8, 25, 39, 58). Cui et al. (20) reported that the PCR detection limit decreases from 3 log₁₀ CFU g⁻¹ with no prior enrichment to 0 log₁₀ CFU g⁻¹ with a 6-h enrichment step at 37°C. For IMS combined with a detection method (IMS-DM), the detection limit decreases from 4 log₁₀ CFU g⁻¹ with no prior enrichment to 1 log₁₀ CFU g⁻¹ with a 6-h enrichment step at 37°C (28).

Several enrichment protocols are currently used for growth of E. coli O157:H7, and they are characterized by different factors. The main enrichment protocol factors are the enrichment broth (basal medium), the addition of antibiotics, the temperature, and the length of the enrichment period. In previous studies, Trypticase soy broth (TSB) and E. coli broth (EC) have frequently been used as basal media, often with one antibiotic added (novobiocin in most studies) (1, 20, 22, 24, 41, 58). For incubation, the enrichment broth is most often incubated at 37°C for 16 to 24 h (1, 22). Nevertheless, some authors have reported an optimal growth temperature for E. coli O157:H7 of around 40°C (31, 44), and in recent studies workers have also used this temperature for incubation of enrichment cultures (1, 2, 32). However, there have been few studies in which the workers tried to compare the efficacies of different enrichment protocols, and the few results obtained have been different in different studies (21, 39, 42, 58).

Furthermore, the enrichment step may be difficult to control because of the presence of background microflora (BM), which may influence the growth of E. coli O157:H7. Some authors have described antagonistic activity of background microflora against E. coli O157:H7 (23, 52, 55, 60). It must be emphasized that in order to optimize the enrichment conditions, it is essential to understand the interaction between the two floras during simultaneous growth. However, so far there has been little information concerning this.

The aim of the present study was to investigate the effects of different enrichment factors on simultaneous growth of the background microflora and E. coli O157:H7 in ground beef. The different factors studied were the basal medium (TSB and EC), the presence of novobiocin in broth (N+ or N−), and the incubation temperature (37°C and 40°C). In order to visualize the potential effects of factors and thus to optimize the enrichment step, growth follow-up and modeling of the simultaneous growth of the background microflora and E. coli O157:H7 were carried out.

MATERIALS AND METHODS

Inoculum preparation.

All four strains used in the present study were stored at −80°C in a glycerol-containing medium. Table 1 shows the sources of these strains. Prior to challenge testing each strain was grown in brain heart infusion (Oxoid, Basingstoke, Hampshire, United Kingdom) at 37°C in order to obtain late-exponential-phase cultures. After 24 h, all the cultures had reached a density of 10⁸ CFU ml⁻¹, and these precultures were used for inoculation of ground beef samples.

TABLE 1.

E. coli O157:H7 strains used in this study

Strain	Origina
1	Food (ground beef)
2	Clinical (HUS)
3	Environmental (feces)
4	Food (milk)

^aStrains were obtained from the National Reference Laboratory of STEC, Marcy-l'Etoile, France. HUS, hemolytic-uremic syndrome.

Microbiological culture media.

In different stages of the study, various microbiological culture media were used. Tryptone medium (Fluka-Biochemika, Switzerland) was used for preparation of serial dilutions throughout the experiments. TSB (Biomérieux, Marcy-l'Etoile, France) and EC (Fluka-Biochemika, Switzerland) were tested with novobiocin (TSB.N+ and EC.N+) (at a concentration of 20 mg liter⁻¹; Sigma, Steinheim, Germany) or without novobiocin (TSB.N− and EC.N−) in the enrichment protocol study. Finally, plate count agar (PCA) (Biomérieux, Marcy-l'Etoile, France) and sorbitol MacConkey agar (Biokar, Beauvais, France) supplemented with a cefixime-tellurite mixture (CT-SMAC) (Biomérieux, Marcy-l'Etoile, France) were used as a nonselective medium for enumeration of the ground beef background microflora and as a selective medium for enumeration and isolation of E. coli O157:H7 (O157) during the enrichment step, respectively.

Ground beef sample inoculation.

Four packages containing two 125-g portions of ground beef (15% fat, non-vacuum- and non-modified-atmosphere-packed packages) were purchased, two from each of two supermarkets in the suburbs of Lyon (France). All eight 125-g portions of ground beef were mixed together, and 25-g portions of meat were aseptically weighed into 32 flasks and stored at 4°C for less than 24 h before the experiments. The precultures of the four E. coli O157:H7 strains were serially diluted in order to obtain a final concentration of 3 log₁₀ CFU ml⁻¹ in fresh tryptone medium, and then each portion was individually inoculated (1 ml) into a series of eight different flasks containing the 25-g portions of ground beef. After manual homogenization, the 32 flasks (8 flasks for each strain) were stored overnight at 4°C. Next, the contents of each of the 32 flasks were transferred into a stomacher bag to which 225 ml of TSB.N−, TSB.N+, EC.N−, or EC.N+ was added according to the growth enrichment protocol (Table 2). Hence, the final concentration was roughly 4 CFU of O157 per ml in each stomacher bag.

TABLE 2.

Enrichment protocols tested in the present study

Basal medium	Novobiocina	Temp (°C)b	Code
TSB	N−	37	TSB.N − .37
	N+	37	TSB.N + .37
	N−	40	TSB.N − .40
	N+	40	TSB.N + .40
EC	N−	37	EC.N − .37
	N+	37	EC.N + .37
	N−	40	EC.N − .40
	N+	40	EC.N + .40

^aN+, novobiocin added; N−, novobiocin not added.

^bThe basal medium and antibiotics were incubated at 37 or 40°C.

Enrichment protocols.

Growth experiments were carried out with the four O157 strains in the presence of BM for different enrichment protocols, which were the result of the combination of two temperatures (37 or 40°C), two basal media (TSB and EC), and the presence or absence of novobiocin (N+ and N−) (Table 2). To do this, the 32 samples obtained at the end of the ground beef sample inoculation were stomached for 1 min and left at the ambient temperature for a regeneration step for 40 min.

Growth monitoring.

At each sampling time (every 2 h for each enrichment protocol), the numbers of viable cells of BM and O157 were determined by plating 1-ml portions of appropriate dilutions of samples onto PCA and 0.1-ml portions onto CT-SMAC, respectively. After incubation, the colonies on PCA and CT-SMAC plates were counted in order to obtain BM and O157 growth data for each enrichment protocol.

Growth modeling. (i) Modeling of individual BM and O157 growth kinetics.

Two growth models were used to separately fit individually observed BM and O157 growth kinetics for the eight protocols tested. The first model, which has three parameters, is a very simple model with two phases (5, 43) (model 1). It describes only the exponential and stationary phases and does not take into account any lag phase:

(1)

where y(t) is the bacterial density (in log₁₀ CFU ml⁻¹) at time t (in h), y₀ is the initial bacterial density (in log₁₀ CFU ml⁻¹), μ_max is the maximum specific growth rate (in h⁻¹), and t_max is the time at which the stationary phase begins (i.e., the time at which the maximum bacterial density is reached).

The second model, which has four parameters, incorporates a third phase, the lag phase (λ) (in h) (model 2). It has been given different names by various authors (11, 43). Curves described by this model have an abrupt transition (breakpoint) between the lag and exponential phases:

(2)

(ii) Global modeling of simultaneous growth of BM and O157.

A global model for the simultaneous growth of BM and O157 is defined simply by using model 2 for each flora (model 3). It is defined by eight parameters (y_0O157, λ_O157, μ_maxO157, t_maxO157, y_0BM, λ_BM, μ_maxBM, and t_maxBM), which may be equivalently estimated by two individual fittings of model 2 with BM and O157 growth kinetics or by global fitting of model 3. A partial model nested in model 3 is also defined in the same way, but a common value for t_max is assumed for the two floras (model 4). The last model thus has only seven parameters (y_0O157, λ_O157, μ_maxO157, y_0BM, λ_BM, μ_maxBM, and t_max) and is based on the hypothesis that there is competition between the two floras studied. More precisely, in model 4 the growth of each microflora is assumed to stop as soon as one microflora reaches its maximum density in the enrichment basal medium. This phenomenon has been described previously and is commonly called the Jameson effect (10, 12, 17, 30, 40, 48).

Statistical methods.

Fitting of models to the BM and O157 data was performed by nonlinear regression (4) by using the least-squares criterion. Estimates for parameters were obtained by minimizing the residual sum of squares (RSS):

(3)

where N is the number of data points, y_i is the observed data value, and ŷ_i is the fitted value. Nonlinear regression was computed with the NonlinearLeastSquares function of R-Software (35). The precision of each parameter estimate was also reported in terms of asymptotic marginal confidence intervals. Comparisons of nested models (model 2 nested in model 1 and model 4 nested in model 3) were performed using an F test (4):

(4)

where N is the size of the data set, p_f is the number of parameters of the full model, p_p is the number of parameters of the partial model, RSS_f is the residual sum of squares of the full model fit, and RSS_p is the residual sum of squares of the partial model fit. The observed F value (F_obs) must be compared with a theoretical F value with v₁ = p_f − p_p, v₂ = N − p_f, and degrees of freedom.

An analysis of variance was carried out in order to compare the maximum levels of O157 reached at the end of the different enrichment protocols. To do this, the maximum level (y_max) was calculated for each of the four strains tested and each protocol by determining the average of the bacterial counts measured after 12 h. A mixed analysis of variance model was considered, in which the basal medium, the addition of antibiotic (novobiocin), and the temperature were fixed factors and the strain was a random factor. The level of significance for each factor and each interaction was P < 0.05.

RESULTS

Individual fits of BM and O157 growth kinetics.

For each enrichment protocol studied, almost the same growth kinetics were obtained for all strains, which were characterized by three phases, a short lag phase followed by an exponential phase and a stationary phase (Fig. 1). For each enrichment protocol, model 2 was thus globally fitted to data obtained for the four strains. The use of a nonnull lag phase was justified by the fact that model 2 fits the growth data significantly better than model 1 (P < 0.05, as determined by an F test) in all the cases except one protocol (P = 0.12 for protocol TSB.N+.37, as determined by an F test).

FIG. 1.

Simultaneous growth kinetics of the four E. coli O157:H7 strains and background microflora obtained for the eight protocols and theoretical curves obtained by fitting of model 4. ○, strain 1; ▵, strain 2; +, strain 3; ×, strain 4. Gray symbols and curves show the data for the background flora, and black symbols and curves show the data for O157.

Global fits of simultaneous growth of BM and O157.

Figure 1 shows that for each of the eight protocols studied, the parameters t_maxO157 and t_maxBM seemed to be very similar. Hence, BM and O157 data were globally fitted by using model 4 with a common t_max for the two microfloras. No significant difference was found between model 3 (full model) and model 4 (partial model) in terms of the RSS for any of the protocols (P > 0.05, as determined by an F test). Figure 1 shows that growth data are well described by model 4, indicating that O157 growth stopped as soon as BM growth reached the maximal level. Moreover, it is worth noting that the maximal level of BM growth was almost constant from one protocol to another (8.62 to 8.86 log₁₀ CFU ml⁻¹). The maximal levels for O157 varied from 7.28 to 8.12 log₁₀ CFU ml⁻¹ according to the protocol tested. It is worth noting that at this stage of growth, the difference between the two floras was between 0.6 and 1.5 log₁₀ CFU ml⁻¹ depending on the protocol considered; consequently, the O157 counts were always in the minority compared with the background microflora counted, with relative ratios from 4 to 30.

Impact of enrichment factors on BM and O157 growth parameters.

As shown in Fig. 2a and b, the initial level of O157 (y_0O157) was not greatly influenced by the enrichment protocol tested, whereas the initial level of BM (y_0BM) appeared to decrease when EC was used. Indeed, a decrease of roughly 1 log₁₀ was observed for the protocols with EC compared to those with TSB. Figures 2c and d show that the different protocols had little impact on the O157 lag time (λ_O157), while the BM lag time (λ_BM) seemed to increase when novobiocin was added to the basal medium, especially for protocols with TSB. Addition of novobiocin to the basal medium had little effect on the O157 and BM maximum specific growth rates (μ_maxO157 and μ_maxBM) (Fig. 2e and f). Nevertheless, when no antibiotic was added, lower values of μ_maxBM were observed for protocols with TSB than for protocols with EC. The maximum specific growth rates of both microfloras were influenced by the temperature, and on the whole, 40°C resulted in higher μ_max values. Although the t_max values were lower for TSB.N− protocols than for the other protocols, this parameter was almost constant regardless of the protocol, varying between 10.0 and 11.7 h (Fig. 2g).

FIG. 2.

Plots of the estimated growth parameters for model 4 and the corresponding 95% asymptotic confidence intervals. (a) y_0O157; (b) y_0BM; (c) λ_O157; (d) λ_BM; (e) μ_maxO157; (f) μ_maxBM; (g) t_max. Gray indicates data obtained at 37°C, and black indicates data obtained at 40°C. The results obtained for EC and TSB are indicated by dotted and solid lines, respectively. N+ and N− indicate protocols with and without novobiocin, respectively.

Impact of enrichment factors on the maximum O157 level.

Only two main effects appeared to be significant in the analysis of variance: the temperature (P < 0.01) and the strain (P < 0.05). However, two first-order interactions involving these two factors also had significant effects on the maximal O157 level: the interaction between the addition of novobiocin and the strain (P < 0.05) and the interaction between the basal medium and the temperature (P < 0.05).

As shown in Fig. 3, there was variability between the four strains of O157, but none of the four strains grew better in all the protocols. The final O157 levels were generally higher at 40°C than at 37°C, but this was not always the case for every strain and it was not generally the case for the EC.N− protocols. When the antibiotic effect was considered, the addition of novobiocin seemed to increase the variability among strains. On the whole, the differences between the various protocols were small compared to the variability between strains. Indeed, regardless of the protocol and the strain considered, the maximal level of O157 was between 6.5 and 8.5 log₁₀ CFU ml⁻¹.

FIG. 3.

Plots of y_maxO157 for the eight protocols. ○, strain 1; ▵, strain 2; +, strain 3; ×, strain 4.

DISCUSSION

Our results indicate that there is a simple competitive interaction between E. coli O157:H7 and the prevailing background microflora during the enrichment step for ground beef. In fact, O157 growth stopped as soon as the level of BM reached the maximal value, which was almost constant regardless of the enrichment protocol considered. Our model was limited because it did not address other types of interactions which may take place during enrichment, but it seemed to be sufficient to describe the simultaneous growth of the two floras based on the observed data.

The present study showed that the enrichment protocol factors had little impact on the simultaneous growth of BM and O157. Indeed, the selective factors (i.e., EC and novobiocin) had only slight effects at the beginning of BM growth (on y_0BM and λ_BM) but did not inhibit BM growth. The higher incubation temperature (40°C) slightly improved O157 growth, but like the other selective factors, did not reduce BM growth. The observed values for the O157 maximum specific growth rate, between 1.51 and 1.86 h⁻¹, were not very variable according to the protocol tested and were close to values reported previously for growth of E. coli O157:H7 in ground beef without any enrichment protocol. Tamplin et al. (56) reported μ_maxO157 values of 2.45 and 1.79 h⁻¹ at 40°C and 35°C, respectively, in sterile irradiated ground beef samples, and Walls and Scott (61) reported a μ_maxO157 of 1.89 h⁻¹ at 35°C in a raw ground beef sample. A comparison between these previously reported results and our results emphasized that the enrichment protocol factors, especially the selective factors, had little effect on the simultaneous growth of BM and O157. Nevertheless, other conclusions might be drawn for untested conditions, such as vacuum-packed ground beef, in which the interaction between BM and O157 might be different because of a qualitatively different background microflora (60).

The optimal length of incubation could be determined from the t_max, because of the relative stability of the maximal level of target bacteria after t_max. Our results (Fig. 2g) show that the t_max values ranged from 10.02 to 11.7 h, depending on the protocol considered. Thus, an incubation time of around 12 h could be recommended based on these results. Nevertheless, it is worth noting that t_max is dependent on the initial BM level, which may vary from one ground beef sample to another. Hence, if the initial level of BM is higher than that in our experiment, the t_max will be lower. For example, the parameters of model 4 obtained for the TSB.N−.37 protocol allowed prediction of a t_max of 8 h if y_0BM was fixed around the higher BM level observed in French food industries (5 log₁₀ CFU ml⁻¹) instead of the 4 log₁₀ CFU ml⁻¹ in our study. On the contrary, if y_0BM is fixed around the lowest level observed in French food industries (2 log₁₀ CFU ml⁻¹), model 4 with the same enrichment protocol predicts a t_max of around 14 h. The optimal length of incubation may also depend on other ground beef characteristics (the kind of packaging, the physiological state of the BM, and the O157 flora).

For the final level of O157 (y_0O157), no marked differences between the enrichment protocols tested were observed (Fig. 3). It is worth noting that the final level obtained at the end of the enrichment step must be higher than the detection threshold of the screening method used for detection. Immunological methods (enzyme-linked immunosorbent assay, radioimmunoassay, immunochromatography, etc.), often followed by isolation of the target bacteria if there are positive results, are the main methods used for detection of E. coli O157:H7. Currently, the international reference method for isolation of E. coli O157:H7 is the IMS method (ISO EN 16654). The IMS protocol consists of a 6-h enrichment step at 42°C, followed by an immunoconcentration step (using immunomagnetic beads coated with anti-O157 antibody) and finally by a detection step (using plating medium [i.e., sorbitol MacConkey agar]). With no prior enrichment step at least 3.5 to 4.4 log₁₀ CFU of E. coli O157:H7 g⁻¹ of sample is required for detection of the bacteria after immunomagnetic separation combined with detection methods (26, 27, 28). Genetic methods, like PCR, are also used, but to a lesser extent. The PCR detection threshold with no prior enrichment step that has been reported previously is 3 log₁₀ CFU g⁻¹ (20, 63).

Figure 3 shows that all the protocols tested allowed, after 24 h, O157 to reach levels higher than PCR and IMS-DM detection thresholds. However, for the enrichment protocols with novobiocin and incubation at 37°C, the y_maxO157 values obtained after 6 h of incubation scarcely reached the IMS-DM detection threshold. Nevertheless, in the IMS method (ISO EN 16654), incubation for 6 h is recommended for isolation of E. coli O157:H7.

Moreover, it must be emphasized that y_maxO157 obviously depends on the initial O157 level in ground beef (y_0O157). The initial O157 inoculum (0.6 log₁₀ CFU ml⁻¹) was relatively large in our study because of experimental limits. However, Crowley et al. (19) reported sporadic counts of E. coli O157:H7 in ground beef products, and the average count was 0.91 log₁₀ CFU g⁻¹. By taking into account the 10-fold dilution carried out at the beginning of the enrichment protocol (i.e., 25 g of sample transferred into 225 ml of basal medium), the concentration corresponded to a y_0O157 value of around −0.09 log₁₀ CFU ml⁻¹. Predictions made by model 4 with this realistic value for y_0O157 gave final O157 levels ranging from 6.3 to 7.7 log₁₀ CFU ml⁻¹ for the eight protocols tested. When the levels of O157 obtained after 6 h of enrichment were determined, the predicted values varied between 2.9 and 3.7 log₁₀ CFU ml⁻¹. In this case, based on data close to reality, the PCR detection threshold would thus be reached regardless of the protocol and the length of enrichment, whereas for all protocols the IMS-DM threshold of detection would be not reached after the recommended incubation period (6 h).

Moreover, other data have been reported, and Shinagawa (53) found that a level of 4 CFU in 25 g of food caused a food-borne disease in a Japanese school in 1996. When the 10-fold dilution carried out at the beginning of the enrichment protocol was taken into account, this corresponded to a y_0O157 of around −1.8 log₁₀ CFU ml⁻¹. Predictions made by model 4 with this realistic value for y_0O157 gave final O157 levels ranging from 1.2 to 2 log₁₀ CFU ml⁻¹ and from 4.6 to 5.6 log₁₀ CFU ml⁻¹ after 6 and 24 h of enrichment incubation, respectively. In this case, in which the initial level of contamination by E. coli O157:H7 was low, the predictions showed that the y_maxO157 values obtained after 6 h of enrichment would be lower than the PCR and IMS-DM detection thresholds. Hence, some false-negative results could be obtained after the IMS analysis, for which only 6 h of incubation is recommended (ISO EN 16654). For an enrichment with 6 h of incubation before detection by PCR, some false-negative results could be also obtained (20, 34).

In addition, healthy exponential-phase O157 cells were used in our study, but in reality the cells could be starved, desiccated, stressed, or in the stationary phase (59). Natural O157 contaminants may behave in a different way. However, Whiting and Bagi (62) reported that exponential-phase cells of Listeria monocytogenes had the shortest lag phase compared to cells grown under unfavorable conditions. The lag phase was also extended when the inoculum size was very small (3). Hence, the last predictions obtained with model 4 (based on fitting our experiment data) could overestimate the O157 level obtained after 6 or 24 h. It is worth noting that when the background microflora level was higher than that in our experiment, the IMS-DM and PCR threshold of detection could scarcely be reached even after 24 h of enrichment.

The length of the enrichment step is currently being reduced more and more in order to obtain more rapid methods for isolation of E. coli O157:H7. However, the last predictions showed that 6 h of enrichment is not sufficient to reach the detection thresholds of numerous screening methods. In order to evaluate the optimal length of incubation, the IMS-DM and PCR detection limits after a 6-h enrichment step were estimated by using model 4. Depending on the protocol, the predicted values varied from 2 to 9 CFU ml⁻¹ for PCR and from 18 to 93 CFU ml⁻¹ for IMS-DM. These values are consistent with those obtained previously after 6 h of enrichment at 37°C (20, 28, 29). Moreover, if the length of incubation varies from 6 to 10 h, the predicted PCR and IMS-DM detection limits are decreased by roughly 3 log₁₀ and vary from −3 to −4 log₁₀ CFU ml⁻¹ for PCR and from −2 to −3 log₁₀ CFU ml⁻¹ for IMS-DM. Hence with prior enrichment for 10 h, samples with a low initial level of O157 (−1.8 log₁₀ CFU ml⁻¹) would be detected. Since the t_max values range from 10.02 to 11.7 h depending on the protocol considered, it does not seem useful to further increase the incubation temperature, and 10 to 12 h appears to be the optimal length of enrichment for ground beef.

Model 4 allowed us to predict the simultaneous growth of E. coli O157:H7 and background microflora in other conditions which were not tested (i.e., low initial O157 level and high initial BM level) and thus to propose recommendations concerning the optimal incubation period. In the present study we obtained results and drew conclusions for non-vacuum- and non-modified-atmosphere-packed ground beef, but more experiments should be carried out with ground beef in various packaging. Indeed, vacuum- or modified-atmosphere-packed ground beef might contain other background microfloras, and consequently the interactions between E. coli O157:H7 and the background microflora could be other types of interactions. For instance, in vacuum-packed ground beef, the Lactobacillus flora in the background microflora is more important. Thus, antimicrobial metabolites, acids, and bacteriocins produced by the Lactobacillus flora may involve another type of interaction with E. coli O157:H7.