Biosecurity measures reducing Salmonella spp. and hepatitis E virus prevalence in pig farms-a systematic review and meta-analysis

Abstract

Salmonella spp. and hepatitis E virus (HEV) are significant foodborne zoonotic pathogens that impact the health of livestock, farmers, and the general public. This study aimed to identify biosecurity measures (BSMs) against these pathogens on swine farms in Europe, the United States, and Canada. Overall, 1,529 articles from three scientific databases were screened manually and with the artificial intelligence (AI) tool ASReview. We identified 54 BSMs from 32 articles, primarily focused on Salmonella spp. control. Amongst the extracted BSMs, only five measures for Salmonella spp. control, namely, 'acidification of feed', 'acidification of drinking water', 'rodent control', 'all-in and all-out production', and 'disinfection' had sufficient observations to conduct a meta-analysis. Of these five, acidification and rodent control were found to be protective measures, that is, their summary odds ratios in the corresponding meta-analyses were lower than 1, indicating lower odds of Salmonella spp. presence on farms which implemented these BSM compared to farms which did not implement them (odds ratio [OR] around 0.25). All-in and all-out production showed a non-significant protective effect (OR = 0.71), while disinfection showed a statistically non-significant lack of association between disinfection and the presence of Salmonella spp. on the farm (OR = 1.03). For HEV, no meta-analysis could be performed. According to multiple articles, two BSMs were significantly associated with a lower risk of HEV presence, namely, disinfecting vehicles (OR = 0.30) and quarantining pigs before introducing them on the farm (OR = 0.48). A risk of bias assessment for each included article revealed a high risk in the majority of the articles, mainly due to selection and performance bias. This emphasises the lack of standardised, high-quality study designs and robust empirical evidence linking BSM implementation to pathogen reduction. The limited data available for meta-analysis, coupled with the high risk of bias (RoB) in the literature, highlights the urgent need for more substantial evidence on the effectiveness of BSMs in mitigating the transmission and spread of zoonotic pathogens, such as Salmonella spp. and HEV on pig farms.

Keywords: HEV; Salmonella spp.; biosecurity; interventions; meta-analysis; risk reduction; swine herds; zoonoses.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram showing the workflow of the systematic review. Source: Adapted from reference (71). 35 articles made it to the final list, but BSMs were only extracted and analysed from 32 articles as one article reported confidence intervals for pathogen being not transformable into confidence intervals (CIs) for an OR and only 2 articles with 3 unique BSMs were identified for PEC and were, therefore, excluded from further analysis.

Figure 2

Forest plot of random-effects meta-analysis (with restricted maximum likelihood estimator of the amount of heterogeneity) for the biosecurity measure (BSM) acidification of feed for Salmonella spp. at the fattening stage. Each observation is encoded with internal BIOPIGEE designations for the article ID (Art Id) and the observation ID (Obs Id). Where available, the number of Salmonella spp. positive (‘Salm+’) and negative (‘Salm−’) samples for the BSM and the control condition, as stated in the corresponding publication, are provided. Each square indicates the odds ratio (OR) for one observation, with the size of the square indicating the weight with which that observation contributed to the summary value of the meta-analysis. The whiskers on the squares indicate the 95% confidence intervals (CIs) for each observation. The diamond at the bottom showed the summary value of the meta-analysis, indicating a summary OR of 0.26 with a 95% CI of 0.17–0.39.

Figure 3

Forest plot of random-effects meta-analysis (with restricted maximum likelihood estimator of the amount of heterogeneity) for the biosecurity measure (BSM) water acidification for Salmonella spp. at the fattening stage. Each observation is encoded with internal BIOPIGEE designations for the article ID (Art Id) and the observation ID (Obs Id). Where available, the number of Salmonella spp. positive (‘Salm+’) and negative (‘Salm−’) samples for the BSM and the control condition, as stated in the corresponding publication, are provided. Each square indicates the odds ratio (OR) for one observation, with the size of the square indicating the weight with which that observation contributed to the summary value of the meta-analysis. The whiskers on the squares indicate the 95% confidence intervals (CIs) for each observation. The diamond at the bottom indicated the summary value of the meta-analysis, indicating a summary OR of 0.25 with a 95% CI of 0.17–0.36.

Figure 4

Forest plot of random-effects meta-analysis (with restricted maximum likelihood estimator of the amount of heterogeneity) for the biosecurity measure (BSM) rodent control for Salmonella spp. ignoring the production stage. Each observation is encoded with internal BIOPIGEE designations for the article ID (Art Id), the observation ID (Obs Id), and the production stage. Where available, the number of Salmonella spp. positive (‘Salm+’) and negative (‘Salm−’) samples for the BSM and the control condition, as stated in the corresponding publication, are provided. Each square indicates the odds ratio (OR) for one observation, with the size of the square indicating the weight with which that observation contributed to the summary value of the meta-analysis. The whiskers on the squares indicate the 95% confidence intervals (CI) for each observation. The diamond at the bottom indicated the summary value of the meta-analysis, indicating a summary OR of 0.22 with a 95% CI of 0.10–0.45.

Figure 5

Contour-enhanced funnel plot for observations included in the meta-analysis on the mitigating effect of rodent control on Salmonella spp. infection in pig production, ignoring the stage. Each point is an observation from the literature. The shaded areas help identify whether observations lie above or under statistical significance. The funnel with dotted lines indicates how the observations are scattered around the summary OR of 0.22.

Figure 6

Diagnostic plot for sensitivity analysis for the meta-analysis on the mitigating effect of rodent control on Salmonella spp. infection ignoring the stage in pig production.

Figure 7

Contour-enhanced funnel plot for observations included in the meta-analysis on the mitigating effect of feed acidification on Salmonella spp. Infection at the fattening stage in pig production. Each point is an observation from the literature. The shaded areas help identify whether observations lie above or under statistical significance. The funnel with dotted lines indicates how the observations are scattered around the summary OR of 0.26.

Figure 8

Contour-enhanced funnel plot for observations included in the meta-analysis on the mitigating effect of water acidification on Salmonella spp. infection at the fattening stage in pig production. Each point is an observation from the literature. The shaded areas help identify whether observations lie above or under statistical significance. The funnel with dotted lines indicates how the observations are scattered around the summary OR of 0.25.

Figure 9

Diagnostic plot for sensitivity analysis for the meta-analysis on the mitigating effect of feed acidification on Salmonella spp. infection at the fattening stage in pig production.

Figure 10

Diagnostic plot for sensitivity analysis for the meta-analysis on the mitigating effect of water acidification on Salmonella spp. infection at the fattening stage in pig production.

Figure 11

Risk of bias ratings per bias domain for all studies in the final selection stage (n = 32). The numbers shown are percentages of articles for each rating by domain.