**3. Results**

### *3.1. Natural Microbiota on Carcass*

Both lactic acid and the ozone interventions significantly reduced (*p* < 0.0001) aerobic plate counts, coliform, and *E. coli* when applied to beef carcasses (Figure 1). Aerobic plate counts on carcasses were significantly reduced on average by 3.26 Log CFU/cm<sup>2</sup> and 3.83 LogCFU/cm<sup>2</sup> after ozone and lactic acid interventions, respectively. Coliform counts on carcasses were significantly reduced on average by 1.42 Log CFU/cm<sup>2</sup> and 1.37 Log CFU/cm<sup>2</sup> after ozone and lactic acid interventions, respectively. Likewise, *E. coli* counts on beef carcasses were significantly reduced by 1.29 LogCFU/cm<sup>2</sup> and 1.35 LogCFU/cm<sup>2</sup> after ozone and lactic acid intervention, respectively. Significant reduction of *E. coli* to undetectable levels was achieved after lactic acid and ozone interventions on beef carcasses. For each microorganism, there were no statistical differences in microbial populations between any of the two interventions.

### *3.2. Natural Microbiota on Trim*

Coliforms and *E. coli* counts on the trim were substantially low when analyzed on a per cm<sup>2</sup> basis. When transformed to Log CFU/cm<sup>2</sup> for statistical analysis, most counts were below 1 CFU/cm2, therefore resulting in negative Log CFU/cm<sup>2</sup> counts, making analysis and visualization more difficult. Thus, an analysis on a per sample (Log CFU/500 cm2) basis was made to assess the effectiveness of the interventions. This conversion was

achieved by multiplying the Log CFU/cm<sup>2</sup> by 500 cm<sup>2</sup> of area sampled, resulting in Log CFU/500 cm<sup>2</sup> which is equivalent to Log CFU/sample. On trim, both lactic acid and the ozone interventions significantly reduced (*p* < 0.003) aerobic plate counts, coliform, and *E. coli* when applied to trim (Figure 2). Moreover, lactic acid greatly reduced (*p* < 0.009) aerobic plate count and coliforms when compared to ozone. Aerobic plate counts on trim were significantly reduced on average by 0.74 Log CFU/sample and 2.08 Log CFU/sample after ozone and lactic acid interventions, respectively. Coliform counts on trim were significantly reduced on average by 0.93 Log CFU/sample and 2.13 Log CFU/sample after ozone and lactic acid interventions, respectively. Moreover, *E. coli* counts on beef trim were significantly reduced on average by 0.67 Log CFU/ sample and 1.08 Log CFU/sample after ozone and lactic acid interventions, respectively.

**Figure 1.** Carcass Aerobic plate count, coliform, and *Escherichia coli* counts (limit of detection < 0.05 CFU/cm2) before and after the application of the interventions (LogCFU/cm2). Horizontal line within the boxplot represents the median. The box upper and lower limit represents the interquartile range, and the bars represent 1.5xInterquartile Range. a,b Box plots with different letters within the same microorganism type represent statistical differences (*p* < 0.05).

Since trim natural microbiota encountered in coliforms and *E. coli* was substantially low, authors decided to inoculate *E. coli* O157:H7 and Salmonella surrogates on the trim and apply the ozone intervention to assess its efficacy. For both trim types, the ozone intervention significantly reduced (*p* < 0.0001) *E. coli* O157:H7 and Salmonella surrogate cocktail counts (Figure 3). Initial inoculation attachment was on average 5.67 Log CFU/cm<sup>2</sup> and 5.52 Log CFU/cm<sup>2</sup> for chuck and foreshank trim, respectively. *E. coli* cocktail attachment was well within target inoculation of 5–6 Log CFU/cm2. On average, counts were reduced by 1.17 Log CFU/cm<sup>2</sup> after the ozone intervention. Reduction between trim types was similar (*p* = 0.18). Consequently, the intervention efficacy is expected to be the same when applied to different trim types.

**Figure 2.** Trim aerobic plate count, coliforms, and *Escherichia coli* counts (limit of detection < 0.05 CFU/cm2) before and after the application of the interventions (Log CFU/sample). Horizontal line within the boxplot represents the median. The box upper and lower limit represents the interquartile range, and the bars represent 1.5xInterquartile Range. a,b Box plots with different letters within the same microorganism type represent statistical differences (*p* < 0.05).

**Figure 3.** *Escherichia coli* surrogate attachment levels and after intervention counts (limit of detection < 4 CFU/cm2) on LogCFU/cm<sup>2</sup> basis. Horizontal line within the boxplot represents the median. The box upper and lower limit represents the interquartile range, and the bars represent 1.5xInterquartile Range. a,b Box plots with different letters represent statistical differences (*p* < 0.05).

In the beef processing plant, the use of the ozone intervention was implemented on 11 October 2019. Chi-square analysis comparing the year prior (1.06%, 102/9,609) to implementation of Biosafe ozone intervention and the year after (0.26%, 25/9,439) implementation indicates statistical difference (*p* < 0.0001) in the percentage of presumptive positive rates of *E. coli* O157:H7 in trim per year. A month-by-month comparison can be

observed in Figure 4. The year before implementation of the ozone intervention presented a 4.1 times greater incidence of presumptive *E. coli* O157:H7 than the year after implementation, indicating a potential 75.5% reduction of presumptive *E. coli* O157:H7 presence in trim.

**Figure 4.** In-plant monthly Presumptive positive rate of *E. coli* O157:H7 in beef trim before and after implementation of the ozone intervention (N = 19,048). I Error bars represent 95% confidence intervals of the monthly incidence.

### **4. Discussion**

The ozone intervention in carcasses significantly reduced indicator microorganisms studied in the commercial beef processing plant environment. This reduction was equivalent in magnitude to the reduction observed by using a final lactic acid carcass wash. The processing plant that allowed this study to be conducted, used 82 ◦C (180 ◦F) hot carcass wash prior to the lactic acid wash as their usual final harvest intervention before the carcasses entered the hot box. For this study, they left the hot water wash on and switched the lactic acid spray with the aqueous ozone treatment to evaluate the effect of ozone compared to that achieved with the use of lactic acid. Consequently, it can be observed that the multiple hurdle approach of using ozone after a hot water wash has equivalent reduction of APC, coliforms, and *E. coli* compared to using lactic acid after a hot water wash. Minimal sampling requirements to demonstrate process control in beef slaughter operations published by the FSIS require one generic *E. coli* sample for every 300 head of cattle harvested. A negative result is the acceptable outcome, but if in 13 subsequent generic *E. coli* tests there are more than three samples between 1 and 100 CFU/cm2, the commercial processing plant fails the performance standards [17]. In this study, *E. coli* cell count was below the detection limit (<0.05 CFU/cm2) after both final carcass interventions. Thus, the facility passed the performance standards and can demonstrate appropriate process control while using lactic acid or ozone interventions.

Ozone in an aqueous solution has been used in the past as a possible antimicrobial intervention in beef. Some studies have reported no significant reduction compared to a 28 ◦C water wash, whereas others have observed a significant reduction of 1.46 LogCFU/cm<sup>2</sup> of *E. coli* O157:H7 compared to 0.60 LogCFU/cm<sup>2</sup> reduction of water spray chill and a reduction of APC of 0.99 LogCFU/cm<sup>2</sup> [10,11]. In this study, a reduction of APC of 3.26 LogCFU/cm<sup>2</sup> was observed after hot water wash and ozone treatment. A multiple hurdle approach in the commercial plant environment is followed to more effectively eliminate pathogen presence in beef products [18,19]. Therefore, different interventions can act synergistically and more effectively to reduce the microbial load of beef in a commercial

processing plant. Moreover, the recent development of an enhanced ozone technology and techniques to increase ozone half-life and reactivity in aqueous solution may increase the efficacy of ozone interventions in beef as observed in this study.

When comparing the ozone intervention against the lactic acid intervention in beef trim, we assessed the individual effect that the intervention has on trim. It is worth noting that the analysis in trim was done on a per-sample basis instead of a per-cm<sup>2</sup> basis due to substantially low coliform and *E. coli* presence in commercial samples. In this trim study, lactic acid further reduced APC and coliform counts compared to the aqueous ozone treatment. However, similar reductions were observed in generic *E. coli* when comparing both treatments. Lactic acid has been known to have a residual effect in the reduction of microbial load, where significant reductions in indicator microorganisms can be seen even after 12 days of treatment [20]. Contrastingly, ozone interventions have not ye<sup>t</sup> been observed to have a residual effect in beef, since it is unstable and breaks down into oxygen shortly after generation and reaction with organic materials. Further research must be conducted to assess differences in shelf-life effects that ozone interventions may have in beef over extended storage times.

Generic *E. coli* has historically been used by processing plants to verify process control. The hazard analysis and critical control points system final rule of 1996 required generic *E. coli* testing [21]. *E. coli* presence is important to assess in beef because it is an indicator of fecal contamination as it is commonly found in the cattle gastrointestinal tract and hides. The gastrointestinal tract of cattle is also a possible reservoir of foodborne pathogens such as Salmonella and *E. coli* O157:H7 [17]. Therefore, if *E. coli* is found in beef, the risk of having Salmonella or pathogenic *E. coli* presence is likely to increase. In the trim sampled, over 90% of the trim had < 1 CFU/cm<sup>2</sup> of *E. coli*. Thus, to further validate the efficacy of the ozone treatment, the authors decided to conduct a Salmonella and *E. coli* O157:H7 surrogate inoculation study on the trim inside a commercial beef processing plant, to take into account the effects of commercial processing operations and actual equipment.

In the surrogate inoculation trial, ozone intervention significantly reduced the concentration of the *E. coli* cocktail. Foreshank and chuck trim were chosen as the "worst case scenario" for this section as, historically, these are the two types of trim that the commercial beef processing plant had more frequently found presumptive *E. coli* O157:H7 presence. These surrogates have been previously seen to mimic *E. coli* O157:H7 and Salmonella resistance to antimicrobial treatments when used as a cocktail in validation trials [13–16,22]. In some cases, reporting a slight increase in the magnitude of survival of the surrogate compared to Salmonella or *E. coli* O157:H7 for a relatively higher margin of safety. Thus, it can be inferred that the survival of the pathogens would be less than the one encountered with the surrogates. The surrogates are more on the conservative end of possible reduction since some of these strains might be slightly more resistant to an antimicrobial intervention than the actual pathogens [13,16]. In this context, the ozone intervention can significantly reduce *E. coli* O157:H7 and Salmonella average concentration by at least 1.17 LogCFU/cm2, with further reductions potentially possible if subsequent sequential applications are considered and surface contact is enhanced. Furthermore, the antimicrobial intervention may cause sublethal injuries in cells that may hinder their ability to grow in selective media. Even though the samples were kept at refrigerating temperatures for approximately 24 h prior to processing in BPW while being shipped to the laboratory, bacteria may have not completely recovered from the intervention. However current sampling and quantification protocols used by the North American beef industry for *E. coli* follow quantification in selective media.

Historical data shared by the plant indicates a significant improvement since the implementation of the ozone intervention in the commercial facility. The year before ozone implementation, 102 lots of trim resulted in presumptive positive for *E. coli* O157:H7. After a year of ozone implementation, the plant observed a 75.5% reduction in positives, having only 25 presumptive positive lots. The improvement translates into a significant economic gain as substantially fewer lots of trim had to be disposed of or rerouted to fully cooked products at lower values. Ozone is known to have antimicrobial properties through direct oxidation of the cell wall resulting in cell lysis; however, it can also considerably damage DNA and produce reactions with oxygen radical by-products during its breaking down process [8]. Current methods for *E. coli* O157:H7 detection in beef, have screening procedures that use quantitative PCR for detection of a particular gene encoded in the DNA of the pathogen of interest [23]. In the multiple hurdle intervention setting, bacteria have been affected by a series of antimicrobial interventions, such as hot carcass washes, organic acid washes, carcass trimming, steam vacuuming, among others. By the time carcasses reach the chilling rooms, they have potentially undergone at least 2–4 antimicrobial interventions possibly reducing bacterial loads below detection limits, as it can be observed in coliform and *E. coli* counts in carcasses after interventions evaluated in this study. At that point, an ozone intervention may be able to further reduce bacterial concentration through cell lysis or other mechanisms; such as DNA damaging that has been reported [24,25] and ozone could have accessibility due to the synergistic effect on the bacterial membrane, that may be weakened from the prior antimicrobials used in the facility When cells undergo such damage, their proliferation becomes hindered under stressful conditions, such as refrigeration storage and distribution, enhancing beef safety in the value chain. Ozone's capacity for DNA degradation may be causing mutations in the bacterial genome rendering bacteria harmless and target genes of the real-time PCR screening procedures undetectable [24]. More research is needed to confirm cell damage and viability after the application of sequential ozone treatments, but these findings provide evidence that the aqueous ozone intervention evaluated in this study may play a significant role in controlling pathogen contamination in beef carcasses and trim.
