Global Prevalence of Yersinia enterocolitica in Cases of Gastroenteritis: A Systematic Review and Meta-Analysis

The prevalence of Yersinia enterocolitica in gastroenteritis is often underestimated. It relates considerably to morbidity and medical expenses around the world. Understanding the cause of gastroenteritis leads to making the appropriate treatment decisions. We systematically searched PubMed, Science Direct, Embase, and Scopus to identify all published studies between Jan. 1, 2000, and Dec. 31, 2019, to assess the prevalence of Y. enterocolitica in gastroenteritis patients. A total of 5039 articles were identified that lead to the extraction of data from 47 of them. The pooled prevalence of Y. enterocolitica in cases of gastroenteritis was estimated as 1.97% (1.32–2.74%) in the culture method and 2.41% (1.07–4.22%) in the molecular method. Among the biotypes of Y. enterocolitica, 1A (62.48%) and 1B (2.14%) had the most and least prevalence, respectively. Serotype O3 Y. enterocolitica with 39.46% had the highest and O5,27 with 0.0% had the least prevalence in gastroenteritis cases. In conclusion, the findings of this systematic review show that Y. enterocolitica is prevalent in gastroenteritis in all age groups. Serotypes O3 and O9 of Y. enterocolitica had the highest prevalence and O5,27 had the least prevalence in diarrheal patients. The prevalence of Y. enterocolitica was similar in both gender and different seasons. It should be noted that to determine the role of the organism, more studies are needed especially in food-borne diseases.


Background
Yersiniosis is caused by Gram-negative bacteria Yersinia enterocolitica (Y. enterocolitica) and Y. pseudotuberculosis. Although Y. enterocolitica is a frequent cause of human infection especially in developed countries of temperate zones, Y. pseudotuberculosis human infection is rare [1]. It mainly caused a gastrointestinal infection in humans. Additionally, Y. enterocolitica can cause other clinical manifestations including mesenteric lymphadenitis, endocarditis, and predominantly infects children [2]. Yersiniosis is the third cause of notifiable bacterial zoonosis in the European Union after campylobacteriosis and salmonellosis [3]. Y. enterocolitica is a psychrotrophic organism that can replicate at temperatures ranging from 0 to 44°C. As such, the organism can replicate in the refrigerator and survives in frozen foods and liquids for long periods. Peritrichous flagella causes the motility of Y. enterocolitica. Motility is temperature dependent, as the bacterium is motile at 25°C but is not motile when it grows at 37°C. Pathogenesis of Y. enterocolitica also depends on temperature. e invasive proteins of Y. enterocolitica produce at environmental temperatures of less than 28°C and under acidic conditions at 37°C. e expression of virulence factors necessary to infection initiates by the gradual increase of temperature within the host [2]. Infections caused by Y. enterocolitica pathogenic strains do not belong to a specific age group, but the clinical manifestation is frequently observed in children and younger adults. Adults can be asymptomatic carriers of infection [4]. Fever, abdominal pain, and diarrhea are the common symptoms of yersiniosis in children [2]. e bacterium was isolated from domestic and wild animals. Pigs are regarded as the reservoir of the pathogen [5], but high titers of anti-Yersinia antibodies in domestic animals, such as cattle, goats, and sheep revealed that there are other possible sources [6]. e main method of human infection is through consumption of contaminated food especially raw or undercooked ones [2] though drinking of contaminated water, close exposure to pet animals, and blood transfusion have also been mentioned [2,7]. Y. enterocolitica had several biotypes and serotypes. Virulent isolates of Y. enterocolitica are attributed to certain biotypes and serotypes. Among the six known biotypes (including 1A, 1B, 2, 3, 4, and 5), 1A is reported as an nonpathogenic biotype in healthy people. Y. enterocolitica serotypes O3, O8, O9, and O5. 27 were isolated from most cases of human yersiniosis [2]. e most serious disease is caused by serotype O8 with extensive ulceration of the gastrointestinal tract and sometimes death of the patients [8].
Patients may defecate Y. enterocolitica for 90 days after the recovery, which shows the importance of early detection of the bacterium in order to prevent transmission and possible outbreak [9]. In order to detect the Y. enterocolitica, a culture method and molecular assays were developed. e conventional culture method is time-consuming and has false-negative results while PCR is not only a sensitive and specific detection method but also is able to identify the pathogenic isolates and further characterization of the isolates [4]. Around the world, there is limited information about the prevalence of yersiniosis due to the clinical presentation of the disease as gastroenteritis so the diagnosis and treatment mainly depend on the clinicians and not on the microbiological culture. e aim of the present study was to estimate the global prevalence of yersiniosis in cases of gastroenteritis. Moreover, the main biotypes and serotypes were determined. e existing data and knowledge were synthesized through a systematic literature review and metaanalysis.

Search Strategy and Study Selection.
A systematic review was performed in PubMed, Science Direct, Embase, and Scopus to identify all published studies between Jan 1, 2000, andDec 31, 2019, with the search keywords of "gastroenteritis," "Yersinia enterocolitica," and "yersiniosis" and related terms without any language restriction. e searched keywords were extracted from the Medical Subject Headings thesaurus.
e search strategy was presented in the supplementary file. Titles and abstracts of relevant original articles after the removal of duplicates were screened by two independent reviewers (TZ and EA). e bibliographies of the included articles were hand-searched for additional references. Gray literature was searched by using Google Scholar. PRISMA guidelines were used to perform the systematic reviews.
Selection of studies was carried out by the following criteria: primary research studies including original article either published or in press; studies with a cross-sectional design; case group of case-control studies; studies including detection of Y. enterocolitica on the samples based on culture or PCR; patients having the symptoms of gastroenteritis; studies performed in a specified region or country; having a known number of sample size; and studies with available full texts. Studies with confusing text or incomprehensible analyses that did not report the sample size and number or percent of positive cases toward Y. enterocolitica were excluded. Reviews, letters, or editorial articles without original data were also excluded.

Data Extraction and Risk of Bias Assessment.
A standard dedicated data extraction form was designed in Excel software. Two authors (TZ and SMR) extracted data independently. If provided, the following data were extracted from each study: bibliographic characteristics, including first author, year of publication, start and end year of the study, study design (cross-sectional or case-control), and country (income, HDI and WHO region); population characteristics, including the age of the participants (mean ± standard deviation (SD), minimum and maximum), gender, and total number of tested patients; methodological information, including diagnostic method, number of patients positive for Y. enterocolitica in culture and PCR separately, season of sampling, biotypes and some prevalent pathogenic serotypes of isolated Y. enterocolitica, and geographic location (latitude and longitude). We included samples with both Y. enterocolitica and another pathogen detected (e.g., E. coli or viruses).
Data were stratified by the diagnostic method and age. Regarding age, data were stratified into four categories: younger than 6 years, 6 to 18 years old, 18 to 59 years old, and more than 60 years old. As an indicator of development and epidemiological context, income, WHO region, and human development index were used to categorize the data on the basis of the country in which the study was performed. e eligible studies were qualified independently by two authors (TZ and EA) according to the Joanna Briggs Institute [10].

Statistical Analysis.
In the current study, random-effect models were used for estimating pooled prevalence and 95% confidence intervals (95% CI). Metaprop command was used in Stata software. Pooled prevalence was calculated using a Freeman-Tukey double arcsine transformation [11,12]. Heterogeneity among studies was examined by I 2 , Cochran's Q. I2 index ranges between 0 and 100 percent and I 2 ˃ 70% was considered heterogeneous [13,14]. A Forest plot in the random-effects model was applied to show pooled prevalence. Subgroup analysis and metaregression were done to identify the sources of heterogeneity [15]. Univariate metaregression analysis was used for assessing the effect of publication year, human development index, geographical location (longitude/latitude), and quality score on the prevalence of Y. enterocolitica. In a subgroup analysis, we estimated the prevalence of Y. enterocolitica in different age groups, type of diagnostic method, study's type, income, and WHO regions. Publication bias was not examined because the aim of the study is not to determine the association between exposures and outcome [15]. e significance level was considered 0.05 in all analyses. All analyses were done by using STATA 13 (STATA Corp., College Station, Texas). In the metaregression, p value <0.1 was considered as a significant level due to the little range of prevalence of Y. enterocolitica and the rare nature of the organism.

Study Characteristics.
A total of 5039 articles were identified of which 4845 were not duplicates. According to the title and abstract, 202 articles were included and assessed for eligibility by full texts (Figure 1). From these, 49 articles passed the quality assessment and data were extracted from 47 of them. e final extracted data included 25 countries from all WHO regions (eight from the Americas, 17 from Europe, ten from Eastern Mediterranean, five from Africa, and seven from Western Pacific) except for the South-East Asia region. From these 47 studies, the prevalence of Y. enterocolitica by culture diagnosis method in cases of gastroenteritis was estimated as 1.97% (95% CI 1.32-2.74; I 2 � 99.19%; p < 0.09 test for heterogeneity) (Figure 2(a)). However, by the PCR method the estimate of pooled prevalence for Y. enterocolitica was 2.41 (95% CI 1.07-4.22; I 2 � 98.39%; p < 0.00 test for heterogeneity) (Figure 2(b)).
ere was significant heterogeneity among the included studies. Table 1 shows the pooled prevalence of Y. enterocolitica by culture and PCR method according to the countries. e highest prevalence of Y. enterocolitica in culture and PCR method was in Madagascar (16.56%). e lowest prevalence of Y. enterocolitica in culture and PCR method was in Australia (0.00%) and Brazil (0.00%), respectively. Table 2 shows the main characteristics of the included studies.

Subgroup Analysis.
e type of study did not change the pooled prevalence of Y. enterocolitica, as in the culture method, the pooled prevalence in the cross-sectional and case-control studies is 2.20 and 1.22, respectively (random test for heterogeneity p < 0.12) (Figure 3(a)). e pooled prevalence of Y. enterocolitica by the PCR method in the cross-sectional and case-control studies is 2.28 and 4.44, respectively (p < 0.05) (Figure 3(b)). e pooled prevalence of Y. enterocolitica was decreased by the increase in the income of the countries (p < 0.001). e pooled prevalence of Y. enterocolitica in low-and high-income countries was  (7), Case report (11), Short paper (4) Confused text/incomprehensible data and duplicate data (1) Non-available full-text (8) Non-gastroenteritis problems and non-human samples including foods or animal samples (120) 202 full text articles assessed for eligibility 4845 records a er duplicates removed 4845 of records screened 4643 of records excluded       (Figure 4(a)) and 16.56 and 0.36 in PCR method (p < 0.02) (Figure 4(b)), respectively. e source of heterogeneity of included studies is income. According to age, the prevalence was not significantly different in younger than 6 years, 6-18 years, and 18-59 years (1.75%; 0.96-2.54; p < 0.95 for culture and 1.84%; 0. 49-3.19; p < 0.66 for PCR; I 2 � 0.0%) ( Figure 5(a)). By gender of participants and season of sampling, the prevalence was similar (p < 0.98 and p < 0.89, respectively) ( Figure 5(a)). According to the biotype of Y. enterocolitica isolates, 1A (62.48%; 95% CI 27.56-91.77) and 1B (2.14%; 95% CI 0.04-6.14) had the most and least prevalence, respectively. Among the investigated serotypes of Y. enterocolitica isolates, O3 with 39.46% had the highest and O5,27 with 0.0% had the least prevalence ( Figure 5(b)).

Metaregression.
According to Figures 6(a) and 6(b), by the increase of publication year, the prevalence did not have any significant change (p < 0.51 for culture and p < 0.38 for PCR). Countries with higher HDI had a lower prevalence of Y. enterocolitica (p < 0.39 for culture and p < 0.01 for PCR) (Figures 6(c) and 6(d)). Longitude had not any significant effect on the prevalence of Y. enterocolitica (p < 0.7 for culture and p < 0.24 for PCR) (Figures 6(e) and 6(f )). e prevalence of Y. enterocolitica increased slightly with increasing latitude but was not statistically significant (p < 0.12) in the culture method; in contrast, its prevalence was decreased with the increasing latitude in the PCR method (p < 0.01) (Figures 6(g) and 6(h)). Metaregression for quality assessment and prevalence was carried out and no relation was observed (p < 0.74 for culture and p < 0.33 for PCR).
Yersiniosis had a global prevalence and is a reportable disease in some countries, such as Denmark, Norway, and 38 states of USA [59,60]. According to PCR detection, Africa and Western Pacific [3,9,18,23,27,37] had the most, and the Americas [20, 30-32, 41, 51, 52, 55] had the least prevalence of Y. enterocolitica. In the present study, the highest prevalence of Y. enterocolitica in culture and PCR method was in Madagascar (16.56%). e lowest prevalence of Y. enterocolitica in culture and PCR method was in Australia (0.00%) and Brazil (0.00%), respectively, in the current study.  reported that the prevalence of Y. enterocolitica is 16.56% in Madagascar and Assiss et al. (2014) reported it is 0.0% in Brazil. In the United States (US), 0.33 per 100000 individuals were infected by Yersinia during 1996 to 2012 in the general population according to Food-borne Diseases Active Surveillance Network, 2012 [61]. In Denmark, Y. enterocolitica was reported as a common cause of bacterial diarrheal disease with 4.9 cases per 100,000 inhabitants in 2016 [62]. Among developed countries, food-borne yersiniosis was higher in most European countries than US [63,64]. e prevalence of Y. enterocolitica was higher in gastroenteritis patients than in the general healthy population. In the present study, income was the origin of heterogeneity among included studies. As, in the low-income countries, Y. enterocolitica was more prevalent than high-income ones.
is can be related to considering hygiene principles. Human yersiniosis is commonly caused by Y. enterocolitica [59]. Yersiniosis caused self-limiting diarrhea that sometimes may be bloody in children younger than four years old. However, fever and abdominal pain accompanied by diarrhea and/or vomiting were reported in older children and adults [9]. e clinical presentation of gastrointestinal disease can be different based on the age and immune status of the host [2]. Diagnosis of yersiniosis is done by isolation of the microbe from human feces or blood or following removal of the appendix, mistakenly [59], although the culture of the bacterium is not a usual procedure for gastrointestinal patients in most hospitals that may lead to underestimates of yersiniosis [59].
Age was not a significant factor regarding gastroenteritis caused by Y. enterocolitica in the current study. Some studies reported that younger children are more susceptible to diarrhea caused by Y. enterocolitica [9,27,39,40]. Al Jarousha  Hawash et al. International Journal of Microbiology 7  Okwori. et al. Stephen et al.   International Journal of Microbiology et al. reported higher isolation of Y. enterocolitica from diarrheic children with the age of one to six years than children less than one year and more than 6 years [48]. Y. enterocolitica had different biotypes and serotypes. e insignificant effect of age may be due to infection of children, adults, and the elderly with different serotypes that may not necessarily create immunity to other serotypes [65]. Furthermore, limited studies were performed on older ages. Gender difference was not seen in the current study. Men and women did not show different symptoms in yersiniosis [40,49]. A seasonal variation was not seen in the present study. Some studies reported more cases during the cooler season [46,66], but according to the report of the European Centre for Disease Prevention and Control, no seasonal pattern was observed for yersiniosis for a period of three years [67]. Some other studies did not also report a significant difference between seasons [9,47,68], which may support the hypothesis that the infection is transmitted via food items that are consumed consistently throughout the year, such as meat and meat products [65].

12
International Journal of Microbiology from children, whereas Y. enterocolitica serotype O9 was frequently isolated from adults (≥40 years of age). Exposure of children to Y. enterocolitica O3 may conceivably provide some immunity against acute infections due to the same serotype during their life, but not necessarily from other serotypes [65]. According to HDI, the prevalence of International Journal of Microbiology 13 Y. enterocolitica was increased with the decrease of HDI that can be related to a higher level of hygienic standards in these countries. In the current study, latitude had a different effect on the prevalence of Y. enterocolitica in culture and molecular diagnosis. Y. enterocolitica is a psychrotrophic bacterium and can replicate in cooler climates [59]. A study on seroprevalence of Y. enterocolitica in wild boars showed that the prevalence was higher in cold climates [70]. Similar results were seen in pigs [71]. e viable organisms were detected in the culture method, but in PCR, the not viable ones were also detected which may be the reason for the higher prevalence of Y. enterocolitica in the temperate zone in the molecular diagnosis compared to culture. e range of prevalence was narrow in the current study which may be the reason for different observations in culture and PCR method, although, in culture, it was not significant.

Strengths and Limitations.
is was the first systematic review and meta-analysis to gain a global prevalence of Y. enterocolitica in gastroenteritis patients. We considered both the culture and PCR isolation of the organism. ere was high heterogeneity among the studies especially due to income but mostly reduced by the application of subgroup analysis and metaregression. Additionally, this study has some limitations that must be acknowledged: first, in some analyses, the number of included studies was low, especially in the older ages (e.g., >60 years); second, there were not sufficient related studies for assessing risk factors; third, the age of participants was not reported clearly in some included studies. Forth, the transmission method of the organism was not reported in the studies. However, estimating the global prevalence of Y. enterocolitica is challenging as most of the studies were performed in hospitalized patients with gastrointestinal symptoms. We encourage further studies, especially in the western Pacific and southeast WHO regions to produce and share local data about yersiniosis. An update of our study should be done due to the availability of additional data.

Conclusion
In conclusion, the findings of this systematic review show that Y. enterocolitica is prevalent in gastroenteritis in all age groups. Y. enterocolitica was not prevalent in high-income countries and countries with higher HDI values. Serotypes O3 and O9 of Y. enterocolitica had the highest prevalence and O5,27 had the least prevalence in diarrheal patients. e prevalence of Y. enterocolitica was similar in both gender and different seasons. It should be noted that to determine the role of the organism, more studies are needed especially in food-borne diseases.