A Meta-Analysis and Systematic Review of Listeria monocytogenes Response to Sanitizer Treatments

Listeria monocytogenes is a ubiquitous organism that can be found in food-related environments, and sanitizers commonly prevent and control it. The aim of this study is to perform a meta-analysis of L. monocytogenes response to sanitizer treatments. According to the principle of systematic review, we extracted 896 records on the mean log-reduction of L. monocytogenes from 84 publications as the dataset for this study. We applied a mixed-effects model to describe L. monocytogenes response to sanitizer treatment by considering sanitizer type, matrix type, biofilm status, sanitizer concentration, treatment time, and temperature. Based on the established model, we compared the response of L. monocytogenes under different hypothetical conditions using forest plots. The results showed that environmental factors (i.e., sanitizer concentration, temperature, and treatment time) affected the average log-reduction of L. monocytogenes (p < 0.05). L. monocytogenes generally exhibited strong resistance to citric acid and sodium hypochlorite but had low resistance to electrolyzed water. The planktonic cells of L. monocytogenes were less resistant to peracetic acid and sodium hypochlorite than the adherent and biofilm cells. Additionally, the physical and chemical properties of the contaminated or inoculated matrix or surface also influenced the sanitizer effectiveness. This review may contribute to increasing our knowledge of L. monocytogenes resistance to sanitizers and raising awareness of appropriate safety precautions.


Introduction
Listeria monocytogenes, a Gram-positive bacterium, is widely distributed and persists in contaminated foods and food-related environments [1]. It is also a typical foodborne pathogen that can cause human invasive listeriosis, leading to meningitis, abortion, or even death [2,3]. Many reports on food safety regarding L. monocytogenes exist [4,5]. From 1 January 2017, to 17 July 2018, a total of 1060 cases were reported in South Africa. According to the World Health Organization (WHO), this was the biggest listeriosis outbreak that ever occurred in the world [6]. Thus, applying disinfecting and cleaning practices to reduce the possibility of L. monocytogenes exposure in the food chain is essential.
Researchers have widely studied sanitizers, such as sodium hypochlorite [7], chlorine dioxide water [8], peroxyacetic acid [9], electrolyzed oxidizing water [10], hydrogen peroxide [11], and surfactant detergent [12], and have demonstrated that these sanitizers are highly efficient at reducing and inactivating L. monocytogenes and other pathogens in foods and environments. However, the physiological and ecological behaviors of L. monocytogenes perform differently under adverse conditions, including when sanitizers and other bactericidal agents are present [13][14][15][16][17]. The lack of understanding of L. monocytogenes response to sanitizers may mislead subsequent control measures.
A meta-analysis is a useful statistical tool to systematically analyze a large collection of data from multiple individual studies. The main characteristic of meta-analyses is the use of statistical methods to quantitatively integrate the results of studies. Currently, meta-analysis is increasingly applied in the field of food microbiology, providing quantitative knowledge based on a cross-sectional survey of microbiological characteristics [18][19][20]. Many authors used meta-analysis to evaluate the interventions in inactivating three foodborne pathogens in fresh produce [21], electrolyzed water treatments in reducing foodborne pathogens on different foods [22], and virus inactivation on hard surfaces or in suspension by chemical disinfectants [23]. This method can be further utilized to comprehensively determine the link between internal/external factors and microbiological responses in various cleaning and disinfection scenarios.
Therefore, the aim of this study was to conduct a meta-analysis to compare and predict the effectiveness of various sanitizers and cleansers on L. monocytogenes. The response of L. monocytogenes to food-related sanitizers in different situations would be explained by constructing a mixed-effects model. Additionally, with the constructed model, the overall effectiveness on the reduction of L. monocytogenes was predicted and compared under different hypothetical conditions.

Search Strategy
To determine our research question, the PICO (population, intervention, control/ comparison, and outcome) approach used in the evaluation of evidence-based clinical questions was employed [24]. The population was specified as the system of solutions, foods, and food contact surfaces, as well as the bacterial status of plankton, adhesion, and biofilm. The intervention was conducting by the disinfecting and cleaning treatments on the population. The comparison was the used sanitizers and cleansers, and the outcome was the log reductions of L. monocytogenes after disinfecting and cleaning treatments.
The full search query that we used for our literature search in the Web of Science Core Collection was as follows: TS = (detergent OR cleaner OR cleaning agent OR disinfectant OR sanitizer) AND TS = (resistance OR tolerance OR survival OR growth OR inactivation) AND TS = listeria. Additionally, the published dates were from 1 January 1985 to 31 December 2021. We found a total of 1200 results (on 11 April 2022).

Selection Criteria and Data Extraction
We performed a preliminary screening based on the title and abstract. We excluded studies with partial topics, such as literature reviews and studies with research objects other than L. monocytogenes (such as L. innocua or L. grayi). We excluded studies wherein the authors used disinfection methods such as nanostructures, ultraviolet lamps, electron beams, and antibacterial coating. We also excluded studies wherein the authors used biological extracts (such as essential oils). We excluded studies wherein the authors examined Leuconostoc, Enterococcus, Lactobacillus, and bacteriophages. In the full-text screening step, we excluded studies wherein the authors used a combination of multiple disinfection or cleaning methods, unclear sanitizer substances, or L. monocytogenes subjected to pressure or stress adaptation. We selected studies that were performed in a colony unit log CFU and studies that contained concentration, time, and temperature data. To stabilize the parameter estimation of the regression models, we reduced the sanitizer concentration range via a log transformation.
The screening rules and process are shown in Figure 1. A total of 84 papers were chosen for the meta-analysis after the final screening. Then, we assembled the experimental settings and observed data in Microsoft Excel (Microsoft, Redmond, WA, USA) using information from the selected research. The screening rules and process are shown in Figure 1. A total of 84 papers were chosen for the meta-analysis after the final screening. Then, we assembled the experimental settings and observed data in Microsoft Excel (Microsoft, Redmond, WA) using information from the selected research.

Estimation of summary effects
Because most studies use the log CFU scale and because it is an intuitively comprehensible parameter, we selected the mean log reduction of L. monocytogenes that attained the following treatment with chemical sanitizers as the total effect. Because some studies did not directly provide the log reduction values, we used the following equation (1) for conversion: R (log CFU/ sample) refers to the reduction of bacteria after the sanitizer treatment; Nb ((log CFU/ sample) refers to the number of bacteria before the sanitizer treatment, and Na ((log CFU/ sample) refers to the number of bacteria after the sanitizer treatment.
A high level of heterogeneity was anticipated given the variations in testing characteristics, matrices, and analytical techniques found in the studies used in this report. A mixed-effects model is used when estimating the summary effect sizes to account for the high expected heterogeneity levels [25][26][27]. The constructed model is shown in equation (2): where β0 is the intercept of the fixed effect, β1 is the mean effect of the increment in the 10-base logarithm of the sanitizer concentration (Con, %), β2 is the mean effect of a

Estimation of Summary Effects
Because most studies use the log CFU scale and because it is an intuitively comprehensible parameter, we selected the mean log reduction of L. monocytogenes that attained the following treatment with chemical sanitizers as the total effect. Because some studies did not directly provide the log reduction values, we used the following Equation (1) for conversion: R (log CFU/ sample) refers to the reduction of bacteria after the sanitizer treatment; N b ((log CFU/ sample) refers to the number of bacteria before the sanitizer treatment, and N a ((log CFU/ sample) refers to the number of bacteria after the sanitizer treatment.
A high level of heterogeneity was anticipated given the variations in testing characteristics, matrices, and analytical techniques found in the studies used in this report. A mixed-effects model is used when estimating the summary effect sizes to account for the high expected heterogeneity levels [25][26][27]. The constructed model is shown in Equation (2): where β 0 is the intercept of the fixed effect, β 1 is the mean effect of the increment in the 10-base logarithm of the sanitizer concentration (Con, %), β 2 is the mean effect of a unit increment in time (Time, min) of the sanitizer, and β 3 is the mean effect of a unit increment in the temperature (Temp, • C) of the sanitizer. The parameters β 0 , β 1 , β 2 , and β 3 can be viewed as continuous variables describing the sanitizer's ability to disinfect. This is because the log-reduction R for a sanitizer increases with the intercept β 0 . Similar to this, for a given concentration, temperature, and time, a sanitizer with higher slopes β 1 , β 2 , and β 3 will result in a higher log-reduction R. Due to the diversity of the data structure, considering the influence of the bacteria status m and different studies n on the results is necessary, so we combined the two variables into an interaction variable (mn). We assumed that this interaction would change depending on the random effects ϑ mn used in Equation (2). Additionally, we assumed a normal distribution for the errors or residuals ε lmn , with a mean of zero and variance s 2 . When interaction effects were examined according to cell status or food contact, matrix p and study n were considered [21,22]. The letter l refers to a certain sanitizer.

Statistic Analysis
We used the "lme" function in the nlme package (version 3.1-158, https://cran.rproject.org/web/packages/nlme/index.html, accessed on 18 June 2022) of R (version 4.2.1, https://cran.r-project.org/, accessed on 18 June 2022) to fit the mixed-effects model, and we assigned weights to each primary study in the meta-analyses, typically using the standard errors of the effect sizes as precision measures. However, obtaining the standard error of the log reduction for each study for the current meta-analysis was impossible. Therefore, we modified the precision to mean that it is equivalent to the N sample size employed in each study. For each of the meta-analysis models, we also calculated goodness of fit measures such as the Akaike information criterion (AIC) and Bayesian information criterion (BIC). In addition, for the regression models, we calculated a measure of between-study heterogeneity, I 2 [28][29][30]. To further analyze specific scenarios, we subgrouped the data according to the biofilm status and food contact. We used the metafor package (version 3.4-0, https://cran.r-project.org/web/packages/metafor/index.html, accessed on 18 June 2022) of R to create forest plots under hypothetical conditions. We defined the statistical significance as p < 0.05.

Characteristics of the Extracted Information
Based on the 84 selected studies, we extracted and formatted 896 data items for further analysis (see Table S1). The dataset was cross-examined by all authors, including the records of sanitizer type, sanitizer concentration, treatment time and temperature, strain types, biofilm status, matrix, bacterial log reduction (mean, standard deviation, and replications), and reference records.
The included papers concern the effect of 36 sanitizers on L. monocytogenes response, mainly processed the other treatments on solid surfaces, including food and abiotic contact surfaces.

Effect size estimation by sanitizer
We quantitatively measured the tolerance of L. monocytogenes under the treatment of different sanitizers using the mixed-effect model of equation (2). As listed in Table 1, we successfully derived the intercept values and coefficients for six sanitizers. Under the treatment of most sanitizers, the sanitizer concentration, temperature, and treatment time directly affected the log reduction of L. monocytogenes (P < 0.05). Sanitizers in high concentrations with a high temperature and extended treatment time will help to eliminate L. monocytogenes viable cells. We could not estimate the model parameters of some sanitizers very well mainly due to the lack of observations or because the sanitizer was either too weak or too efficient.

Effect Size Estimation by Sanitizer
We quantitatively measured the tolerance of L. monocytogenes under the treatment of different sanitizers using the mixed-effect model of Equation (2). As listed in Table 1, we successfully derived the intercept values and coefficients for six sanitizers. Under the treatment of most sanitizers, the sanitizer concentration, temperature, and treatment time directly affected the log reduction of L. monocytogenes (p < 0.05). Sanitizers in high concentrations with a high temperature and extended treatment time will help to eliminate L. monocytogenes viable cells. We could not estimate the model parameters of some sanitizers very well mainly due to the lack of observations or because the sanitizer was either too weak or too efficient.
Pathogens tend to be less resistant when higher intercepts (β 0 ) are obtained [21]. The comparison between the estimated intercepts (β 0 ) indicated that L. monocytogenes was more resistant to the CA and SH treatments. In other words, CH, EW, CDS, and PAA may be more effective at inactivating L. monocytogenes. These agents (CH, EW, CDS) are chlorine-based sanitizers, which are commonly used to eliminate microorganisms in food-processing environments and water. Chlorine affects bacterial membranes and changes their permeability, causing intracellular materials to change and resulting in fatal DNA damage [31,32]. However, chlorine-based sanitizers may be more unacceptable to consumers when used in food than edible organic acids such as CA. Therefore, current studies are developing new strategies to combine edible acid with other physical treatments to enhance antimicrobial efficiency against foodborne pathogens [33].

Effect Size Estimation by L. monocytogenes Biofilm Status
According to the common status of L. monocytogenes in the food chain, we further evaluated the sanitizer effects on planktonic, adherent, and biofilm cells using the mixedeffects models. As the estimated intercepts (β 0 ) listed in Table 2, we treated the planktonic, adherent, and biofilm cells with a total of six, six, and five different sanitizers, respectively. By comparing the sanitizers' intercept (β 0 ) values, we found that the planktonic cells had a low tolerance to the investigated sanitizer (PAA, SH) compared with the adherent and biofilm cells. Similarly, the adherent cells were more resistant than the planktonic cells in the AA sanitizer. From the above description, the planktonic cells were less resistant than the adherent and biofilm cells to most sanitizers, with some exceptions, such as HP. We did not find a consistent sanitizer resistance pattern when we compared the adherent cells with the biofilm cells. The fitted intercepts (β 0 ) presented in Table 2 also suggest that chlorine-based sanitizers (CDS, EW) appeared to be a kind of effective sanitizer against L. monocytogenes regardless of the status. For the biofilm cells, only SH and EW passed the significance test (p < 0.05). The original data showed that the HP and LA sanitizer had 14 and 9 data points, respectively, which may lead to the low significance of the intercept (β 0 ). In other words, the amount of data for biofilm disinfection research was not enough.
To horizontally compare the effect of different sanitizers under the same conditions, the fitted meta-analysis models were solved for a hypothetical treatment with 1% sanitizer and exposure time of 15 min at 25 • C. In Figure 3, these predicted values are depicted as forest plots. Under the assumed treatment conditions, we found that AA caused the lowest log reduction in the planktonic cells (−1.09, (−4.39,2.22)) and the adherent cells (0.87, (−0.09,1.83)). In the planktonic cells, the response to different sanitizers fluctuated widely, with the highest being 13.18. The sanitizers (SH, PAA, CDS, EW) at this hypothetical concentration could likely eliminate the planktonic cells. For the biofilm bacteria, we obtained the lowest log reduction with HP (−1.54, (−4.80,1.72)). EW resulted in the highest log reduction in the three statuses. Foods 2022, 11, x FOR PEER REVIEW 7 of significance of the intercept (β0). In other words, the amount of data for biofilm disinfe tion research was not enough.
To horizontally compare the effect of different sanitizers under the same condition the fitted meta-analysis models were solved for a hypothetical treatment with 1% san tizer and exposure time of 15 min at 25 °C. In Figure 3, these predicted values are d picted as forest plots. Under the assumed treatment conditions, we found that AA cause the lowest log reduction in the planktonic cells (−1.09, (−4.39,2.22)) and the adherent cel (0.87, (−0.09,1.83)). In the planktonic cells, the response to different sanitizers fluctuate widely, with the highest being 13.18. The sanitizers (SH, PAA, CDS, EW) at this hypo thetical concentration could likely eliminate the planktonic cells. For the biofilm bacteri we obtained the lowest log reduction with HP (−1.54, (−4.80,1.72)). EW resulted in th highest log reduction in the three statuses. We divided the bacteria into three statuses that are commonly found in the foo industry (plankton, adhesion, and biofilm), which allowed us to more thoroughly un derstand the resistance of L. monocytogenes to sanitizers. Previous studies on the respons of bacteria to sanitizers focused more on specific disinfectants [34], different bacteria [21 and environmental conditions [22]. Our study demonstrated L. monocytogenes' high sen sitivity to EW regardless of its status, which is consistent with the research performed b George et al. [22]. The use of EW to inactivate microorganisms and suppress their d velopment to decrease postharvest loss and increase the shelf life of fresh fruits and veg etables during storage is the subject of an increasing number of studies. In addition, EW We divided the bacteria into three statuses that are commonly found in the food industry (plankton, adhesion, and biofilm), which allowed us to more thoroughly understand the resistance of L. monocytogenes to sanitizers. Previous studies on the response of bacteria to sanitizers focused more on specific disinfectants [34], different bacteria [21], and environmental conditions [22]. Our study demonstrated L. monocytogenes' high sensitivity to EW regardless of its status, which is consistent with the research performed by George et al. [22]. The use of EW to inactivate microorganisms and suppress their development to decrease postharvest loss and increase the shelf life of fresh fruits and vegetables during storage is the subject of an increasing number of studies. In addition, EW has also been applied to animal food and food contact surfaces [10]. In our metamodel, we considered the EW concentration (the available chlorine concentration, ACC) as well as the influence of environmental factors (treatment time and temperature), and for EW, the ORP (oxidation reduction potential) characteristics also had an impact on microorganism suppression. To be consistent with other studies exploring chemical disinfectants, we did not consider ORP characteristics. The high ORP properties of EW alter electron flow in microbial cells, disrupt cell membranes, oxidizes enzymes in cells, and lead to microbial cell death [35]. This seemed to imply the sensitivity of L. monocytogenes to EW. Thus, other characteristics of EW should be quantitatively studied in future research. Commonly used sanitizers are generally accepted to be efficient against L. monocytogenes in suspension; nevertheless, cells attached to surfaces may be more resistant to sanitizers than cells in suspension [30]. This statement is consistent with the effect of PAA, SH, and AA on L. monocytogenes. However, we were unable to find evidence supporting this assertion for all sanitizers. Several authors contend that a biofilm that has developed over a long period of time is more resistant to antimicrobial agents [36,37], whereas others have claimed that the biofilm's age does not increase its susceptibility to disinfectants [38]. To understand this problem, supplementing the inactivation data of different biofilm periods is necessary, because the existing inactivation data and inactivation parameters are lacking.

Effect Size Estimation by Matrix
The effect of sanitizers on L. monocytogenes has been observed by numerous studies, mostly focusing on three systems: liquid solutions [39,40], foods [41,42], and food contact surfaces [43][44][45]. According to the results of this meta-analysis, L. monocytogenes may differ in terms of their inactivation depending on the matrix. Different matrices with various physical and chemical characteristics may change the treatment efficiency. Based on this, we conducted models by considering different contact surfaces and examining the log reduction on food and food contact surfaces under hypothetical conditions. Nevertheless, the discontinuity of the available data hampered our work on data analysis via a matrix. Some fixed-effect levels were not sufficient to build a model, so some of the surfaces presented in Table 3 did not fully contain the parameters of the three fixed effects (Con, Temp, Time). By comparing the intercept (β 0 ) results, we found that the two sanitizers (SH, EW) were more effective when used on stainless steel surfaces as opposed to other surfaces, as shown in Table 3. L. monocytogenes was less resistant to EW when on stainless steel than when on salmon fillets and lettuce (Table 3). Additionally, we again used the forest plot to compare the predicted log reduction of L. monocytogenes in different matrices under a hypothetical treatment with 0.01% SH or EW. Figure 4 illustrates that, when EW was the sanitizer used, the mean log reductions varied from 0.64 for the salmon fillet to 6.54 for stainless steel, whereas the log reduction varied from 0.50 to 1.69 for SH (Figure 4). L. monocytogenes had different resistance levels to different sanitizers on the same surface: the EW resistance of L. monocytogenes was lower on stainless steel, and the SH resistance of L. monocytogenes was higher on stainless steel. Because bacteria naturally have a predisposition to attach to surfaces as a survival mechanism, researchers have described the colonization of solid surfaces by bacteria as a fundamental and natural bacterial approach in a range of environments [46]. Researchers have frequently discovered L. monocytogenes in biofilm forms in food-processing facilities [47]. We found that identical sanitizing processes produced a different log reduction depending on the matrix type. We previously verified the effectiveness of EW (Tables 1 and 2 and Figure 3). L. monocytogenes attaches to these substances differently depending on the matrix, which makes sanitizing L. monocytogenes on various matrices harder or easier. Salmon tissues are tough for disinfectants to reach and can harbor L. monocytogenes. Microorganisms on fresh produce cannot be eliminated by sanitizers because of the structure of the plant surface, including its nooks, crannies, and minute fissures, as well as the hydrophobic qualities of the waxy cuticle [48,49]. Additionally, chlorine reacts with organic matter on vegetables and reduces the effectiveness of chlorine [50].
have frequently discovered L. monocytogenes in biofilm forms in food-processing facilit [47]. We found that identical sanitizing processes produced a different log reduct depending on the matrix type. We previously verified the effectiveness of EW (Table  and 2 and Figure 3). L. monocytogenes attaches to these substances differently depend on the matrix, which makes sanitizing L. monocytogenes on various matrices harder easier. Salmon tissues are tough for disinfectants to reach and can harbor L. monocy genes. Microorganisms on fresh produce cannot be eliminated by sanitizers because of structure of the plant surface, including its nooks, crannies, and minute fissures, as w as the hydrophobic qualities of the waxy cuticle [48,49]. Additionally, chlorine rea with organic matter on vegetables and reduces the effectiveness of chlorine [50].

Conclusions
Based on the meta-analysis approach, we quantitatively described and compa the impact of multiple factors on L. monocytogenes reduction using mixed-effects mod with 896 extracted data. The hierarchical analysis supported the idea that the saniti type and concentration, as well as the treatment temperature and time, could primar affect the survival of L. monocytogenes. L. monocytogenes generally exhibited higher sistance to citric acid and sodium hypochlorite. Meanwhile, L. monocytogenes in the herent and biofilm status were more resistant to the investigated sanitizers. Besides, physical and chemical properties of the foods and food contact surfaces might influe the sanitizer's effectiveness. However, the research heterogeneity in strains and con tions also restricted the applicability of the established models and further inference the trends of L. monocytogenes resistance. Therefore, future empirical study should s temically determine L. monocytogenes resistance to sanitizer by considering the str variability and comprehensive environmental effects.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Table  The extracted data for meta-analysis. References

Conclusions
Based on the meta-analysis approach, we quantitatively described and compared the impact of multiple factors on L. monocytogenes reduction using mixed-effects models with 896 extracted data. The hierarchical analysis supported the idea that the sanitizer type and concentration, as well as the treatment temperature and time, could primarily affect the survival of L. monocytogenes. L. monocytogenes generally exhibited higher resistance to citric acid and sodium hypochlorite. Meanwhile, L. monocytogenes in the adherent and biofilm status were more resistant to the investigated sanitizers. Besides, the physical and chemical properties of the foods and food contact surfaces might influence the sanitizer's effectiveness. However, the research heterogeneity in strains and conditions also restricted the applicability of the established models and further inference on the trends of L. monocytogenes resistance. Therefore, future empirical study should systemically determine L. monocytogenes resistance to sanitizer by considering the strain variability and comprehensive environmental effects.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/foods12010154/s1, Table S1: The extracted data for meta-analysis. References  are cited in the supplementary materials.

Conflicts of Interest:
The authors declare no conflict of interest.