One Health Perspective of Salmonella Serovars in South Africa Using Pooled Prevalence: Systematic Review and Meta-Analysis

Salmonella is a bacterium that is commonly associated with food-borne infections and is regarded as one of the most important pathogens in public health. Salmonella serovars, particularly Typhimurium and Enteritidis, which are widely distributed globally, mainly result in outbreaks commonly linked to the consumption of animal products. This study is a systematic review and meta-analysis of studies reporting the prevalence of Salmonella serovars from one health perspective that included human, environmental, and animal samples in South Africa. PubMed, ScienceDirect, African Journals Online, and Scopus databases were used to conduct extensive searches of articles which were ultimately included or excluded following the Systematic Reviews and Meta-Analysis (PRISMA) guidelines. According to the data obtained in this review, the overall pooled prevalence estimates (PPE) of Salmonella serovars detection were 79.6%, 61.6%, 56.5%, and 43.2% for human, environment, animal, and environment/animal samples in South Africa, respectively. The majority of the studies (50%) used the polymerase chain reaction (PCR) technique for the detection of Salmonella serovars, followed by culture methods (26.7%), while 20% used serotyping. The PPE for nontyphoidal Salmonellae (NTS) was 65.6% and 34.4% for Salmonella Typhimurium and Salmonella Enteritidis, respectively. Our data further shows that 3 serovars, namely, Salmonella Typhimurium, Salmonella Enteriditis, and Salmonella Hadar, have been isolated from animals, humans, and the environment in South Africa. Our results highlight the ongoing spread of Salmonella spp. especially on animals which might end up infecting humans via direct contact with infected animals or eating infected animal products. This calls for deliberate “One Health” epidemiological studies in order to document information on the transmission between humans, animals, and the environment. This will ultimately result in the formulation of a consolidated salmonellosis control policy by the environmental, human, and veterinary health sectors.


Introduction
Salmonella is a serious public health issue that affects both humans and animals [1]. Salmonella strains are grouped as typhoidal and nontyphoidal organisms based on their disease distribution dynamics [2]. Nontyphoidal Salmonella (NTS) is responsible for the most important public health problem worldwide including South Africa, and accounts for food-borne illnesses with an estimated 94 million cases globally [3,4]. According to Backhans and Fellström [5], the majority of Salmonella cases are caused by Salmonella serovar Enteritidis globally, and the major sources are eggs and poultry meat. Salmonellosis is one of the most serious zoonotic illnesses, affecting both humans and animals around the world [6]. In South Africa, Salmonella Typhimurium and Enteritidis are the commonly reported serotypes [7].
Many African countries rely on meat production for their livelihoods, with meat from cattle and poultry serving as a major source of protein in subsistence groups [8]. Various Salmonella serovars have been isolated from the gastrointestinal tracts of animals such as chickens, horses, ducks, cattle, pigs, goats, and sheep [9][10][11][12]. e published literature on human salmonellosis in South Africa is growing [13][14][15] including detection of the Salmonella serovars from the environment [16][17][18].
Humans, on the other hand, can become infected by coming into contact with live animals or being in an environment contaminated with animal feces and then accidently ingesting pathogens [8,19]. e systematic reviews and meta-analyses: a step-by-step guide will be used to quantify and summarize the findings of these studies [20]. Few systematic reviews and meta-analyses studies have been performed in sub-Saharan Africa and Africa in the last ten years on invasive nontyphoidal Salmonella (iNTS) disease and the prevalence of Salmonella [8,[21][22][23]. Such reviews focused on surveillance of iNTS infections in humans, in food animals and meat, and antibiotic resistance. All this evidence has pointed out the existence, perpetuation, and dominance of three Salmonella enterica serovars Enteriditis, Heidelberg, and Typhimurium in different vertebrate hosts (human and animal) and in environmental samples in South Africa. A better understanding of the existing prevalence of Salmonella is required to inform the development and implementation of effective preventative methods.
erefore, the current study was carried out to identify the prevalence gap, analyse and summarize the pooled prevalence of Salmonella serovars isolated from humans, animals (chickens, ducks, cattle, pigs, goats, horses, and sheep), and the environment from published data in South Africa by carrying out a systematic review and meta-analysis through a review of published articles.

Study Design.
is study was conducted to estimate the prevalence of Salmonella species in South Africa using published articles. e study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA 2020) revised guidelines [24], to report the study (Table S1). e article search approach is presented on a flow chart in Figure 1.

Search Strategy for Relevant Studies.
e search was conducted on four databases: PubMed, ScienceDirect, African Journals Online, and Scopus, as described by Ramatla et al. [21]. We searched for all studies published in English from 1980 until June 2021. e keywords that were used in all databases are shown in Table 1. We completed our search on the 24 th of June 2021. Titles and abstracts were scanned, and full-text papers were downloaded and accessed from library resources and online platforms. e studies were chosen based on the inclusion and exclusion criteria, which are further listed in the study.

Study and Inclusion Criteria.
All available studies and data were included based on the following predefined eligibility criteria. (a) All articles published primarily on quantitative prevalence of Salmonella in humans, animals, and environment in South Africa. (b) Articles must clearly state type of samples and methods of diagnosis used, (c) exact numbers of positive samples were clearly stated and lastly, and (d) all published articles were in English language.

Study Exclusion Criteria.
Studies were excluded if they were (a) not undertaken in South Africa, (b) book chapters were also excluded, (c) review articles, (d) a smaller sample size (less than 20), (e) articles not reported in English were also discarded, and (f ) articles not published between January 1980 and August 2021.

Data Extraction and Data Collection.
e titles and abstracts of possible journal articles were scanned and downloaded. To determine eligibility, full versions of possibly relevant articles were obtained. Each article's data, including author names, publication year, location, total number of isolates, and total samples collected, was assembled separately, and entered into a spreadsheet (Microsoft Excel ® 2013) and tables were constructed on an MS-word document. e meta-analysis included only journal papers specific to Salmonella species or serotype or isolate. If the number of positive Salmonella isolates reported exceeds the sample size due to numbers harvested from cultures, the number was recorded at 100% prevalence.
2.6. Meta-Analysis. Meta-analysis was performed using Comprehensive Meta-Analysis Version 3.0 (CMA) program (https://www.meta-analysis.com/). e pooled prevalence was calculated using the random-effects model with a 95% confidence interval (CI). Cochran's q statistic and the I-square (I 2 ) test were used to measure statistical heterogeneity between studies. e I 2 scores of more than 75% were regarded to have a high degree of heterogeneity. Heterogeneity with a P value less than 0.05 was considered statistically significant. Finally, the funnel plot was generated to investigate the impact of minor research on the pooled salmonella prevalence estimated by graphing the prevalence measure against its standard error. e Begg and Mazumdar test was used to determine the asymmetry of the funnel plot.

Overview of the Selected Studies.
is review covered studies from eight provinces of South Africa, namely, Western Cape, Eastern Cape, Northern Cape, North West, KwaZulu-Natal, Gauteng, Limpopo, and Mpumalanga. However, there were no studies reporting on Salmonella prevalence in Free State province. All the articles included in this study were published between 1980 and 24 th June 2021. e majority of the studies were carried out from 2010 onwards. All the collected articles were published in English. About 5530 papers were identified through database searching. Articles published primarily on Salmonella isolated from humans, animals (chicken, ducks, cattle, pigs, goats, horses, and sheep), and the environment. After reviewing study titles and abstracts, 5403 papers were found to be ineligible. Duplication resulted in the removal of 50 articles. About 77 full articles were assessed for eligibility, and 47 were excluded for the following reasons: studies carried outside of South Africa (n � 39), case studies (n � 3), and studies without clear total samples collected (n � 5). Finally, 30 articles were included in this review and metaanalysis. Table 2 summarises the studies that were included in this review.

Characteristics of the Eligible Studies at Were Included.
From the 30 (Table 2) studies that were included in this study, only 604 samples were from humans, 4111 from animals only (chicken, ducks, cows, pigs, goats, horses, and sheep), 1652 from the environment (water, soil, poultry houses, abattoirs, feed mills, swabs from large animal hospitals, macadamia nuts, manure, and swabs from hospitals), and 180298 from the environment/animal, 69 from the environment/human, and 200 from animals/humans. e number of samples for each study ranged from 22 to 180298. e prevalence amongst the overall studies ranges between 43.2% and 95.8%.
Based on the provinces, the Eastern Cape (n � 2749) accounted for most of the samples, followed by KwaZulu-Natal (n � 2492), Gauteng (n � 1623), the North West (n � 691), and lastly by the Western Cape (n � 470). However, two studies collected a total of 180320 from all provinces, of which 180298 were from the animals and environment [30], and 22 pediatric wards (humans) [14].
e procedures used to isolate and identify bacterial species from the eligible studies were serotyping, microbiological culture, polymerase chain reaction (PCR), and MALDI-TOF-MS. A total of 10/30 (31.2%) studies used both culture isolation and serotyping of Salmonella spp. for identification, 5/30 (15.6%) used the culture method only, while 1/ 30 (3.3%) studies used a combination of culture isolation, PCR, and MALDI-TOF-MS for Salmonella identification, and lastly, 15/30 (50%) studies utilized only PCR for Salmonella identification. Fifty-six different Salmonella serotypes were identified in this review. We observed that in 13 studies, some Salmonella isolates were not identified to the species or serotype level. A total of 999 nontyphoidal Salmonella isolates were detected from 14/30 (46.6%) studies, whereas 833/999 (83.4%) were identified as Salmonella Typhimurium and 166/999 (16.6%) as Salmonella Enteritidis.  Table 3). International Journal of Microbiology   Figure 5 shows the presence of S. Typhimurium, S. Enteritidis, and S. Hadar in both animals, humans, and environment. In addition, S. Heidelberg S. Newport, and S. Agona were also detected from both the animals and environmental samples. All the isolates from human, animals, and environment are listed in Table 2.

Publication Bias.
e Begg and Mazumdar rank correlation test demonstrated no significant publishing bias for practically all parameters except for the studies conducted in Eastern Cape province, where both the asymmetry of the funnel plots and the P value of 0.045 indicated considerable bias (Table 3 and Figure 6).

Discussions
e data obtained from 30 published studies showed an overall prevalence of 66.3% for Salmonella serovars in South Africa which is higher than the findings of the studies conducted in Africa, the Middle East, North Africa, and Africa which reported the prevalence of Salmonella serovars at 5.7%, 8.8%, and 44.8%, respectively [20,45,46]. e PPE of Salmonella serovars was high in humans with 79.6%, animals with 61.6%, the environment with 56.5%, and the environment/animal with 43.2%. is is higher compared to a similar systematic review and meta-analysis conducted in Iran with 6.89% in animals [47], (6.6%) in humans in the Middle East and North Africa (MENA) [20], and human (8.4%) in sub-Saharan Africa [48]. e difference could be explained by microbiological diagnostic procedures employed, antibiotic resistance prevention and control practices, and differences in Salmonella serovars isolated. Contamination of food products and close contact with livestock animals could be a source of Salmonella transmission to humans [49].
Nontyphoidal Salmonella (NTS) infections induce gastroenteritis in people because the bacteria cause an invasive, extra-intestinal condition that leads to bacteraemia and localized systemic infections known as invasive NTS [50]. In this study, nine hundred and forty-one (941) serovars of S. Enteritidis and S. Typhimurium were identified as NTS. ey were also identified as the most prevalent serotypes with Gauteng province having the highest number of NTS isolates. ere were about 4.9% of NTS identified from the study conducted by Smith et al. [14] from all nine provinces. According to the systematic review conducted in the Middle      East and North Africa, Salmonella Typhimurium and Enteritidis were the most prevalent serotypes [20]. e S. Typhimurium and S. Enteritidis are among the top five most common serotypes reported in the United States [51], and they are associated with salmonellosis [52][53][54]. e current study recorded NTS Salmonella from both humans and the environment, and animals dominated by 86.8% of S. Typhimurium, followed by 13.2% of S. Enteritidis. Our study showed that the prevalence of NTS was higher in animal samples. By contrast with the findings in America [55], reported a high prevalence of NTS from the environmental samples. ese findings quantify the existing NTS status in animal production while also emphasizing the main public health hazard associated with the presence of NTS in the animal production chain, which may eventually affect humans.

International Journal of Microbiology
Traditional microbiological methods such as culture isolation using the spread plate technique are still considered the gold standard for diagnostic tests since they efficiently allow the identification of different bacterial species [56]. Generally, the International Organization for Standardization (ISO-6579, 2002) recommends classical microbiological culture isolation for the identification of Salmonella spp. (ISO, 2002). About 41.5% of the studies included in this review used culture-based methods. e culturing method is time-consuming, labourintensive, and sometimes has low sensitivity which makes it unsuitable for regular examination of large numbers of samples as opined by different researchers [11,12,17,28,37,57]. Due to its low sensitivity, some studies combine the culturing method with other sensitive molecular techniques such as PCR [58]. e combination of traditional culture isolation and molecular methods accurately identify bacterial pathogens [58]. Moreover, research has also shown that using molecular methods (PCR) reveals more identification of bacterial pathogens than traditional culture-based methods [56,58]. In this study, we observed that PCR appears to be the most utilized diagnostic method in the detection of Salmonella serovars/isolates with a prevalence of 72.1%, as employed in about 15 studies with over 4219 samples screened. However, research should also not exclude one or the other technique, as a polyphasic approach has proven ideal to investigate bacterial pathogens in microbiological practice [59]. All the studies included in this study were carried out in diverse ways, with some examining a high number of samples and others using various diagnostic approaches. e data on publication bias analysis leads us to conclude that numerous factors, including sample size and different diagnostic approaches could be responsible for the substantial heterogeneity between the findings.
is systematic review and meta-analysis comprised studies from different provinces, of which most of the studies (n � 26, 86.7%) were conducted between 2010 and 2021. en followed by the period of the year 2000 to 2010 with 10% of the studies which screened 187547. From year period 1980-1990 and 1990-2000, PPE were not calculated because there was only one eligible study for each period with 501 samples screened [9,41]. is low number of studies might be due to lack of facilities and funds for conducting research at that time or salmonellosis was a neglected disease in that period in South Africa. In terms of diagnostic methods, these two studies employed serotyping for detecting Salmonella serovars/isolates. e Free State province was not represented in the data sets due to unavailable published data on the prevalence of Salmonella spp. Gauteng appeared to be a province with high number of published articles which may be connected with the availability of resources to conduct research.
Our analysis indicates an increase in the number of samples tested from 1980 to 2021. ese findings could be attributed to higher consciousness by researchers and the health sector about zoonotic diseases which included Salmonella serovars as well as available advanced equipment to    According to the World Health Organization (WHO), food safety, zoonotic disease control, laboratory services, neglected tropical diseases, environmental health, and antimicrobial resistance are among the areas of work where a "One Health" approach is particularly pertinent (https:// www.euro.who.int/en/home). erefore, the presence of zoonotic isolates, namely, S. Typhimurium, S. Enteritidis, and S. Hadar in both animals and humans, and the environment should be taken into consideration in South Africa. e two most commonly reported serotypes of nontyphoidal Salmonella are S. Typhimurium and S. Enteritidis. NTS is a common cause of invasive bacterial disease and is linked to death. Transmission of Salmonella isolates includes animals, animal products, water, and infected humans [14]. Furthermore, S. Heidelberg, S. Newport, and S. Agona were also detected from both the animals and the environmental samples. is indicates that there is the possibility of ongoing circulation of the serovars.

Significance/Strengths of the Study.
is study has a considerable number of strengths including: (1) the present meta-analysis provides the estimates for the prevalence of Salmonella on animals, human, and the environment in South Africa and revealed that there are some provinces with few/no studies conducted. (2) is study selected highquality peer-reviewed studies to give an overview of unbiased data on the prevalence of Salmonella species isolated from humans, the environment, and animals in South Africa. (3) ere are no human studies published in the North West, Limpopo, Gauteng, Mpumalanga, and Northern Cape provinces. (4) Additionally, this study demonstrated that Salmonella serovars infect both humans and animals and have been detected in the environmental samples. is observation suggests the need for a consolidated "One Health" approach from the ecological, human, and animal health sectors.

Limitations of the Study
Several limitations have been identified which include the following: (1) the majority of the studies included samples between 1980 and 2021; (2) the number of articles from some provinces (Western Cape, Eastern Cape, Northern Cape, North West, KwaZulu-Natal, Gauteng, Limpopo, and Mpumalanga) was unusually high, which may have influenced the overall estimate, whilst provinces such as Free State, Limpopo, Mpumalanga, and Northern Cape were underrepresented; (3) some studies reported the total number of isolates from cultures without showing the number of individual serotypes; (4) the studies included in our analyses showed a high level of heterogeneity, hence readers should exercise caution when interpreting the pooled analysis and subgroup; (5) our bias assessment revealed the overall risk of bias from sample size and different diagnostic methods used and sample selection; (6) few reports from humans were observed whereby only four studies were undertaken in Gauteng, Eastern Cape, Kwa-Zulu-Natal, and from all the provinces; and (7) PPE of Salmonella in animal/human, and environment/human were not calculated because there are single reports on each.

Conclusion
Our systematic review and meta-analysis showed the prevalence of Salmonella serovars from animals, humans, and the environment in South Africa. is study highlights the huge knowledge gap on salmonellosis in this country.
ere are significant gaps in surveillance and a lack of published studies on the prevalence of Salmonella spp. in some provinces like Free State province. e results demonstrated that a high prevalence of Salmonella serovars is noticeable in animals rather than in humans and the environment. ese emphasize the main public health hazard associated with the presence of Salmonella serovars, especially NTS in the animal production chain, which may eventually affect humans. Several studies had methodological data gaps, which cast doubt on their validity and made comparisons difficult. e fact that Salmonella serovars infect both humans and animals and have been detected in environmental samples means there is a need for a consolidated "One Health" approach from the ecological, human, and animal health sectors in terms of epidemiological, therapeutic, and policy formulation research.
Data Availability e datasets generated and analysed will be available on request to the corresponding author.

Conflicts of Interest
e authors state that there are no conflicts of interest.

Authors' Contributions
All authors made substantial contributions to the conception and design of the study, acquisition of data, or analysis and interpretation of data, and took part in drafting the manuscript or revising it. All authors had read and approved the final version of the manuscript and agreed to submit it to the current journal.

Acknowledgments
is review received no external funding.