Multilayer network analysis unravels haulage vehicles as a hidden threat to the British swine industry

Abstract

When assessing the role of live animal trade networks in the spread of infectious diseases in livestock, attention has focused mainly on direct movements of animals between premises, whereas the role of haulage vehicles used during transport, an indirect route for disease transmission, has largely been ignored. Here, we have assessed the impact of sharing haulage vehicles from livestock transport service providers on the connectivity between farms as well as on the spread of swine infectious diseases in Great Britain (GB). Using all pig movement records between April 2012 and March 2014 in GB, we built a series of directed and weighted static multiplex networks consisting of two layers of identical nodes, where nodes (farms) are linked either by (a) the direct movement of pigs and (b) the shared use of haulage vehicles. The haulage contact definition integrates the date of the move and the duration Δs that lorries are left contaminated by pathogens, hence accounting for the temporal aspect of contact events. For increasing Δs, descriptive network analyses were performed to assess the role of haulage on network connectivity. We then explored how viruses may spread throughout the GB pig sector by computing the reproduction number R. Our results showed that sharing haulage vehicles increases the number of contacts between farms by >50% and represents an important driver of disease transmission. In particular, sharing haulage vehicles, even if Δs < 1 day, will limit the benefit of the standstill regulation, increase the number of premises that could be infected in an outbreak, and more easily raise R above 1. This work confirms that sharing haulage vehicles has significant potential for spreading infectious diseases within the pig sector. The cleansing and disinfection process of haulage vehicles is therefore a critical control point for disease transmission risk mitigation.

Impact of truck contamination and information sharing on foot-and-mouth disease spreading in beef cattle production systems

“…For example, the median [interquartile range] number of infected producers is 7 [5–10] vs. 72 [62–82] in scenarios DI_B (direct contact only) and D&IN_B (direct and indirect contact) respectively. This finding is consistent with recent studies [ 19 , 41 , 43 , 44 ], which all highlighted the substantial effect of indirect contacts on the ability of farms to potentially spread diseases. Focusing on dairy farms [ 45 ], applied network analysis techniques and showed that indirect contacts through on-farm visits by veterinarians produced a more connected network compared to direct contact only.…”

“…The duration a truck or a loading/unloading area of premises remaining contaminated, i.e., the contamination period h, is affected by not only the ability of the pathogen to survive in fomites, but also by environmental factors such as temperature, and by the frequency of the disinfection operations [41]. In this study, we assume the contamination period a constant number of days, i.e., 14 days, similar to the work in [2,41,43]. The sensitivity analyses on the contamination period and indirect contact transmission probability show their significant influence on the epidemic size.…”

Modeling the transmission and vaccination strategy for porcine reproductive and respiratory syndrome virus

“…The challenges of model fitting could be attributed to the lack of information about other contact networks that connect farms indirectly, such the routes taken by feed deliveries and dead picks, which could explain in a greater detail the spatial dispersion of PRRSV. Recent work reinforced the relevance of the indirect contacts among farms, in which sharing haulage vehicles has significant potential for spreading infectious diseases within the pig sector in Great Britain (Porphyre et al, 2020). Therefore, futures including other indirect contact networks into our model framework, would also likely to improve our forecast accuracy.…”

“…farm visitors, other animals, insects, and contaminated transport-tools) (Pitkin, Deen and Dee, 2009; Pitkin, Deen, Otake et al, 2009; Reiner et al, 2009; Lowe et al, 2014). In future work it would be pivotal to include indirect contact networks such as feed deliveries and mortality management through dead pickups (Porphyre et al, 2020), which would allow PRRSV to spread throughout multilayer networks, for example. This will likely shed light on other routes involved in PRRSV spread (Dee et al, 2020).…”

“…In addition, the model was calibrated on the incidence across all farm types for the temporal model calibration, while we decided to fit the spatial component to infected sow farms only, largely because PRRSV outbreaks records in sow farms were more consistent allowing for a more confident assessment of model sensitivity. Transmission events not explained by between-farm movements or re-breaks at sow farms were classified in the model as local transmitted, thus future studies will include other movement flows such as feed truck deliveries in order to explain in more details this route of transmission (Porphyre et al, 2020). For the assessment of vaccination strategies, we assumed that vaccinated farms acquire life-long immunity, future studies should also consider the heterologous host responses to vaccination (Ezanno et al, 2020).…”