Salmonella Behavior in Meat during Cool Storage: A Systematic Review and Meta-Analysis

Simple Summary Salmonella is an important pathogen associated with many foodborne disease outbreaks that can cause serious issues regarding public health, economic conditions, and quality of life, among others. Meat is the main human infection route for this bacterium, making food quality control in all production steps paramount. As Salmonella is a mesophilic bacterium, the cold chain is very important during meat processing, so pathogen behavior studies under cool storage, simulating the industry environment, can provide important data to the food industry. In this context, the aim of the present study was to perform a systematic review and meta-analysis of Salmonella behavior in meat during cool storage. Other conditions were also analyzed, such as meat sources (beef, chicken, pork, poultry, and turkey), fish, shellfish, media broth, package types, storage time, and bacterial inoculation (concentration and inoculation type). Abstract The aim of the present study was to investigate Salmonella behavior in meat stored in cool conditions (between 0 °C and 7.5 °C), by employing a systematic review and meta-analysis. The data were obtained from research articles published in SciELO, PubMed, the Web of Science, and Scopus databases. The results of the retrieved studies were obtained from meat (beef, chicken, pork, poultry, and turkey), fish, shellfish, and broth media samples The data were extracted as sample size (n), initial concentration (Xi), final concentration (Xf), standard deviation (SD), standard error (SE), and microbial behavior effects (reduction or growth). A meta-analysis was carried out using the metaphor package from R software. A total of 654 articles were initially retrieved. After applying the exclusion criteria, 83 articles were selected for the systematic review, and 61 of these were used for the meta-analysis. Most studies were conducted at 0 °C to 4.4 °C storage temperatures under normal atmosphere package conditions. Salmonella Typhimurium, S. Enteritidis, and a cocktail (strain mixture) were inoculated at 5.0 and 6.0 log CFU mL−1. Articles both with and without the addition of antimicrobial compounds were found. Salmonella concentration decreases were observed in most studies, estimated for all study combinations as −0.8429 ± 0.0931 log CFU g−1 (95% CI; −1.0254, −0.6604) (p < 0.001), varying for each subgroup analysis. According to this survey, Salmonella concentration decreases are frequent during cool storage, although concentration increases and no bacterial inactivation were observed in some studies.


Introduction
Salmonella is an important pathogen, responsible for food disease outbreaks and termed salmonellosis. This microorganism can be found in the gastrointestinal tract of

Material and Methods
The systematic review methodology has been registered on OSF (Open Science Framework) platform [20]  behavior (log CFU reduction or growth) in meat (beef, chicken, pork, poultry, and turkey), fish, shellfish, and in broth media were the specified populations. The measured outcome was derived from pathogen concentrations detected after cool storage.

Search Strategies
The research was performed in the SciELO, PubMed, Web of Science, and Scopus databases, using the following terms as a string: "(Salmonella) AND (Meat) AND ((growth) OR (survival) OR (Kinetic)) AND ((cold AND storage) OR (chill) OR (shelf AND life) OR (refrigerat*))". No restriction filters were applied, and the terms were searched in retrieved paper titles, abstracts, and keywords.
The JabRef program (JabRef Team, US) [21] was used to organize the publications and identify duplicate articles. The selected articles included in this review employed storage temperatures between 0 • C and 7.5 • C.

Eligibility Criteria
Selected articles should be published in English or Spanish. Reviews, book chapters, and articles that did not use bacterial inoculation were excluded.
A second selection was performed by reading the titles and abstract, followed by full article reading. The third criterion was the use of an approved microbiological method for pathogen enumeration. Salmonella concentrations should be reported as log CFU per g, mL, or cm 2 . Experiments with results in MPN (most probable number) were excluded. As a fourth inclusion criterion for meta-analysis, the primary study must clearly describe the sample size and the standard deviations or errors for means. Several studies that did not report standard deviations or errors were used only for the systematic review. Experiments using ozone, high pressure, irradiation, and combined antimicrobial agents were excluded from the systematic review and meta-analysis. The selected articles were categorized as (1) no use of antimicrobial compounds and (2) use of antimicrobial compounds.

Data Extraction
Data of interest described in the articles were organized into spreadsheets by one reviewer. The extracted information included authors, year, sample type (meat or culture medium), package condition (normal atmosphere, vacuum, modified atmosphere), antimicrobial type, antimicrobial concentration, forms of antimicrobial application, Salmonella strain (subspecies or cocktail), inoculum concentration (in CFU g −1 , CFU mL −1 , CFU cm −2 ), inoculation type (surface and mixture), time (day) and temperature ( • C) of storage, sample size (n), initial concentration (Xi), final concentration (Xf ), standard deviation (SD), standard error (SE), and effect (reduction or growth) on microorganism behavior. Gimp 2.10.8 (GIMP team) [22] and ImageJ (ImageJ team, US) [23] software were used to plot the data.

Statistical Analyses
Effect size was determined by the raw mean difference between the initial and final concentrations, as all the primary studies were reported on a log CFU scale. Each treatment was considered an individual observation.
Considering a primary study j, the effect size θ is the difference (RawDiff) in the sample means log reduction (R) or growth (G). Equation (1) was used for experiments without the addition of antimicrobial compounds, and Equations (2) and (3) were used for experiments with the addition of antimicrobial compounds. where Xi, Xci, and Xti are the means of the initial concentration, initial concentration in the control samples, and initial concentration in the treated samples, respectively, while Xf, Xcf, and Xtf are the means of the final concentration, final concentration in the control samples, and final concentration in the treated samples, respectively. The variance of mean log reduction or growth (VarRawDiff) was estimated as: where Sdf and Sdi are the final and initial standard errors, respectively, nf and ni are the numbers of final and initial samples (repetitions), respectively. When the article only provided the standard error (SE), this was transformed into SD through Equation (5).
The systematic review and meta-analyses results were separated into groups from 0 • C to 4.4 • C and 5.0 • C to 7.5 • C, as many scientific reports describe that Salmonella cannot survive at temperatures below 5 • C. The metafor R software package [24] was used to fit meta-analytic random-effects models.

Systematic Review
A total of 654 articles published between 1985 and 2019 were selected following an electronic database search (Table 1). A total of 155 duplicate articles were found. After initial title, abstract, and full article reading, 83 studies [12][13][14][15][16][17][18][19] were selected for the systematic review, and 61 of them were used in the meta-analysis (Figure 1), as the remaining articles did not present concentration means alongside standard errors or deviations. Several experiments employed normal, vacuum, and modified atmosphere package systems, also employing antimicrobial agents such as natural compounds, organic acids, seasoning, marination, and industrial sanitizers.
Data from 363 treatments were extracted from 83 selected articles in the systematic review. Most selected studies stored samples in a normal atmosphere between 0 • C and 4 • C for 11 to 35 days and applied a cocktail strain inoculation ( Table 2). Articles with antimicrobial compound additions were also selected, and most cases used inoculum levels from 5.0 to 6.0 log CFU mL −1 of Salmonella.

Identification of studies via databases and registers
* experiments without antimicrobial compound addition; ** experiments with antimicrobial compound addition: control and antimicrobial treatments (N = natural antimicrobial, S = synthetic antimicrobial).

Meta-Analyses Results
A total of 61 articles were selected for the meta-analysis, while 22 studies were excluded because they did not present means alongside standard deviations or standard errors.
Salmonella behavior was tested under many environmental conditions, such as different sample types (beef, chicken, pork, fish, and turkey), package conditions (normal, vacuum, and modified atmosphere packaging), inoculum concentrations, and binomial time versus temperature. Concentration decreases were observed in most studies, but a pathogen concentration increase or no changes were also noted.
Decreased Salmonella concentrations were observed in most treatments, independent of package condition and antimicrobial compound addition ( Table 2). Decreased between 0.1 and 2 log CFU g −1 or mL −1 were more frequently reported, although some increases of less than 1 log CFU g −1 were also observed.
A funnel plot graphic was prepared to verify potential publication bias between the results ( Figure 2). pathogen concentration increase or no changes were also noted.
Decreased Salmonella concentrations were observed in most treatments, independent of package condition and antimicrobial compound addition (Table 2). Decreased between 0.1 and 2 log CFU g −1 or mL −1 were more frequently reported, although some increases of less than 1 log CFU g −1 were also observed.
A funnel plot graphic was prepared to verify potential publication bias between the results ( Figure 2).     Another subgroup consisted of effects analyses as a function of two storage temperature intervals, from 0 °C to 4.4 °C and 5 °C to 7.5 °C. These temperature intervals were selected because temperatures lower than 5 °C are a challenge for microorganisms, and a maximum temperature at 7 °C is recommended for refrigerated storage. Data from 226 treatments extracted from 48 studies were analyzed for the group stored at 0 °C to 4.4 Another subgroup consisted of effects analyses as a function of two storage temperature intervals, from 0 °C to 4.4 °C and 5 °C to 7.5 °C. These temperature intervals were selected because temperatures lower than 5 °C are a challenge for microorganisms, and a maximum temperature at 7 °C is recommended for refrigerated storage. Data from 226 treatments extracted from 48 studies were analyzed for the group stored at 0 °C to 4.4 Another subgroup consisted of effects analyses as a function of two storage temperature intervals, from 0 • C to 4.4 • C and 5 • C to 7.5 • C. These temperature intervals were selected because temperatures lower than 5 • C are a challenge for microorganisms, and a maximum temperature at 7 • C is recommended for refrigerated storage. Data from 226 treatments extracted from 48 studies were analyzed for the group stored at 0 • C to 4.4 • C, while data from 66 treatments extracted from 17 articles were analyzed for the group stored at 5 • C to 7.5 • C (Figure 8).
Animals 2022, 12, x 1 °C, while data from 66 treatments extracted from 17 articles were analyzed for the g stored at 5 °C to 7.5 °C (Figure 8). Subgroup analyses were also performed on Salmonella concentration results function of the already referenced storage temperature intervals (Table 3).  Subgroup analyses were also performed on Salmonella concentration results as a function of the already referenced storage temperature intervals (Table 3).

Systematic Review
Storage temperatures between 0 • C and 4 • C (77.97%) were more commonly applied compared to temperatures between 5 • C and 7.5 • C (22.03%), probably because the retrieved studies aimed to observe Salmonella behavior at low temperatures.
In studies conducted at 0 • C to 4 • C, Salmonella Typhimurium was the most inoculated strain (32.23%), and the most frequent inoculum concentrations were 5, 6, and more than 6 log CFU mL −1 .
As a result, decreases lower than 1 log CFU g −1 in the control (18.73%) and treated (11.29%) groups at 0 • C to 4 • C storage were the most frequent. A reduction of more than 2 log CFU g −1 was also observed at both temperatures. However, about 19.28% of the analyzed treatments presented growth rates of 5.23% and 3.58% at 0 • C to 4 • C, respectively.
At temperature intervals from 0 • C to 4 • C, the control group treatments presented 18.73% decreases less than 1 log CFU g −1 and 21.48% decreases between 1 log CFU g −1 to 2 log CFU g −1 in pathogen level. From 5 • C to 7.5 • C, the main results were 5.78% reduction <1 log CFU g −1 and 3.03% reductions both >1 log CFU g −1 and >2 log CFU g −1 .
Most articles analyzed bacterial behavior under a normal atmosphere and demonstrated pathogen reduction. Experiments with vacuum packing and MAP also resulted in Salmonella decreases in most cases, although growth-rate values were noted in all conditions.
Although most selected studies reported decreases in Salmonella concentrations, some indicated increased pathogen concentrations even under cool storage. One study, for example, observed a reduction between 1.4 and 1.9 log CFU g −1 in Salmonella concentrations in ground beef packed under a vacuum and in a modified atmosphere at 3 • C for 12 days [17]. In another study, 0.17 and 0.97 log CFU Salmonella -concentration decreases were detected in modified-atmosphere packed (MAP) beef stored at 7.5 • C for 12 days, although a 1.69 log CFU g −1 pathogen reduction was observed in vacuum-packed beef in the same study [42].
A decrease in Salmonella concentrations was observed in chicken samples containing 2% acetic acid at 2 • C, 6 • C, and 8 • C for 9 days, with reduction values of 0.7, 0.9, and 0.9 log CFU/g, respectively [61]. However, in another study, Salmonella Typhimurium growth was observed in vacuum-packed minced chicken treated with olive oil and stored at 2 • C for 60 days [43].
Increasing Salmonella concentrations in vacuum-packed whole shrimp with potassium sorbate, sodium benzoate, sodium diacetate addition and control were observed at 4 • C for 7 days, ranging from 0.93 to 1.84 log CFU g −1 [96]. Edwards et al. [40] also reported decreases between 0.01 and 0.23 log CFU g −1 in shrimp inoculated with S. Typhimurium, S. Enteritidis, and S. Infantis stored at 4 • C for 2 days. The authors, however, also observed growth of approximately 0.10 and 0.13 of the same Salmonella strains in the same storage conditions.
Regarding challenge tests with antimicrobial compounds, Salmonella concentration decreases were observed with malic and acetic acid addition in Mueller Hilton broth stored at 4 • C for 21 days [16]. In another study, decreases of 1.6 and 0.37 log CFU g −1 in Salmonella Typhimurium concentrations were detected in BHI broth stored at 4 • C and 7 • C, respectively, for 7 days [69]. Silva et al. [88] reported a pathogen decrease of over 6.0 log CFU g −1 using chitosan coating in inoculated beef stored at 4 • C for 3 days. Salmonella concentrations were also reduced by approximately 1.8 and 1.6 log cycles following the addition of 0.3% carvacrol and thymol essential oils in beef stored at 4 • C for 7 days [68].
A study comprising lactic acid or acidified sodium chlorite addition in minced beef stored at 5 • C for 14 days reported a 0.031 to 0.264 log-cycle reduction in Salmonella cocktail concentrations, although an increase was observed in the control treatment of 0.8 log CFU g −1 bacterium concentrations [50]. A Salmonella Typhimurium concentration reduction higher than 3 log CFU g −1 was reported for minced meat stored at 4 • C for 7 days [67]. Nisiotou et al. [74] also indicated a decrease in Salmonella Typhimurium concentrations in modified-atmosphere-packaged beef, with or without marination, stored at 5 • C for 19 days of 0.6 and 2.90 log CFU g −1 in control and treated samples, respectively.

Meta-Analysis
The funnel plot graph (Figure 2) presents the dispersion noted for the selected data, with the most observed outcomes varying between approximately −3 and 1 log CFU g −1 , demonstrating result variability and the absence of publication bias.
The second subgroup comprised antimicrobial compound addition, categorized into groups with (152 treatments/37 articles) and without (142 treatments/55 articles) antimicrobial compound addition compounds or controls. Both studies demonstrated a significant effect (p < 0.001), but antimicrobial addition was reduced by −1.2041 ± 0.1559 log CFU g −1 (95% CI: −1.5096, −0.8986), while in the control group or in treatments without antimicrobial compounds, a −0.4526 ± log CFU g −1 (95% CI: −0.6282, −0.2769) was observed. This demonstrated the importance of studying antimicrobial compound effects, mainly those of natural compounds, against Salmonella in meat.
A decrease in Salmonella concentrations was observed in both temperature intervals, albeit with a difference in p values and higher pathogen concentration decreases at lower temperatures.
Concerning all subgroups, a higher Salmonella concentration reduction was observed for fish, under both normal and MAP atmospheres, with antimicrobial compound addition and stored between 0 • C and 4.4 • C.
As expected, decreases were noted in all analyses except for treatments using broth samples and MAP packages at 5 • C to 7.5 • C (interval 2), with treatments analyzed at 0 • C to 4 • C (interval 1) presenting more significant Salmonella concentration decreases. At 0 • C to 4 • C, only vacuum-packaged pork samples with up to 4 log CFU mL −1 pathogen concentration treatments presented no significant statistical effect, and, at 5 • C to 7.5 • C, these effects were observed in less than half of the studies.
The more significant effects in both temperature intervals were observed for studies employing fish samples, with −1.93 and 1.34 log CFU g −1 decreases at intervals 1 and 2, respectively. When considering studies with K > 60, beef samples and antimicrobial compound addition treatments exhibited more significant effects, with −1.24 and −1.25 log CFU g −1 values, respectively.
Thus, differential statistical effects were observed between treatments according to storage temperature intervals. For example, beef samples presented a −1.2 log CFU g −1 (p < 0.001) decrease at interval 1 and a −0.59 log CFU g −1 (p < 0.05) decrease at interval 2. In addition, this same condition was observed in many other results. Only studies on fish presented the same statistical effect at both temperature intervals (p < 0.001). It is important to note that chicken samples did not present significant effects at either temperature interval.
Experiments without antimicrobial compound addition and under vacuum packaging and MAP packaging are less effective in controlling Salmonella growth at interval 2. Decreases in Salmonella concentrations were lower for treatments using the S. Typhimurium and S. Enteritidis mixture inoculation type, up to 4 log CFU g −1 for 10 days, compared to interval 1 treatments.

Conclusions
Salmonella concentration decreases were observed in meat under cool storage, which were higher at lower storage temperatures. According to this meta-analysis, other factors also contribute to Salmonella concentration decreases during refrigerated storage time, such as beef and fish samples, normal atmosphere and MAP, antimicrobial compound addition, and storage from 0 • C to 4.4 • C.
Cool storage was effective for Salmonella growth control, where decreases in meat between −0.15 and −1.24 log CFU g −1 were observed at colder temperatures, demonstrating the importance of cold chains for both the industry and customers.
Although most experiments reported Salmonella decreases during cool storage, the pathogen was not eliminated in the samples. Thus, the risk of salmonellosis transmitted by meat remains and should be considered.