Highly pathogenic avian influenza (HPAI) viruses continue to threaten global poultry production and public health. To examine how methodological approaches have been used to assess this risk, we conducted a systematic review and critical appraisal of peer-reviewed studies assessing avian influenza virus (AIV) incursion risk in domestic poultry, using PubMed and Web of Science databases. Quantitative modelling approaches ranged from statistical, data-driven methods to mechanistic frameworks and operated at different spatial and temporal scales, using farms or geographical areas as the epidemiological unit of interest. Across studies, analytical choices were largely shaped by data availability, which constrained achievable resolution and influenced the questions models could address. Most applications focused on area-level analyses, often describing historical spatial risk patterns rather than supporting prospective inference. Statistical models frequently relied on proxy variables linked to past incursions, limiting biological interpretability and robustness. Recent improvements in wild bird surveillance, including citizen-science initiatives, have enabled more explicit representation of wild bird dynamics, although persistent gaps in poultry population data continue to restrict model development in many settings. In settings where farm-level data are limited, mechanistic approaches are commonly applied, relying on assumptions regarding underlying disease transmission dynamics. A consistent challenge across studies was limited model validation, reflecting sparse outbreak data and spatial and temporal clustering of outbreaks. Interpretation was further complicated by scale mismatches and risks of ecological and atomistic fallacies when extrapolating between farm-level and area-level outputs. Overall, the review highlights the need for closer alignment between data availability, modelling approaches and the intended decision-making contexts.

Abstract The European Commission requested the assessment of the capacity of the surveillance provisions of the World Organization for Animal Health (WOAH) to detect bovine spongiform encephalopathy (BSE) cases (C‐, H‐ and L‐type) in the European Union (EU) and to propose if any current EU surveillance provisions should be kept. The WOAH provisions stipulate that BSE surveillance should target animals on the BSE clinical spectrum and require the implementation of standardised clinical protocols for selecting animals for testing. Based on expert judgement of retrospective clinical data, between 5% and 49% of the 55 BSE cases detected in the EU & UK since 2015 would have been selected for testing by the WOAH surveillance. Assuming the current EU surveillance, a back‐calculation model predicted a very low number of BSE cases detected for the period 2025–2029: 0.0196 for C‐BSE, 10.94 for H‐BSE and 9.13 for L‐BSE. As no classical BSE (C‐BSE) cases are expected, the application of the WOAH provisions would primarily impact the detection of atypical BSE (H‐BSE and L‐BSE), with an estimated detection over the next 5 years of between zero and five H‐BSE cases and between zero and four L‐BSE cases. Supplementing the WOAH provisions with systematic testing of fallen stock (with or without emergency slaughter) with increased age threshold of 60 or 72 months would achieve a detection capability close to current EU levels. This approach would enable the documentation of the effectiveness of BSE control measures, provide scientific assurance that the decline in C‐BSE is sustained and allow atypical BSE trends to be monitored reliably while maintaining the sensitivity required to detect any re‐emergence of C‐BSE. Finally, this would contribute to risk mitigation by triggering immediate statutory control measures when cases are detected. Modifications of EU surveillance requirements should consider any potential adjustments to other available BSE control measures.

Previous research efforts on highly pathogenic H5N1 avian influenza (HPAI) suggest that different avian species exhibit a varied severity of clinical signs after infection. Waterfowl, such as ducks or geese, can be asymptomatic and act as silent carriers of H5N1, making detection harder and increasing the risk of further transmission, potentially leading to significant economic losses. For backyard hobby farmers, passive reporting is a common HPAI detection strategy. We aim to develop a computational, mechanistic model to quantify the effectiveness of this strategy by simulating the spread of H5N1 in a mixed-species, small-population backyard flock. Quantities such as detection time and undetected burden of infection in various scenarios are compared. Our results indicate that the presence of ducks can lead to a higher risk of an outbreak and a higher burden of infection. If most ducks within a flock are resistant to H5N1, detection can be significantly delayed. We find that within-flock infection dynamics can heavily depend on the species composition in backyard farms. Ducks, in particular, can pose a higher risk of transmission within a flock or between flocks. Our findings can help inform surveillance and intervention strategies at the flock and local levels.

In the UK and Ireland, the European badger is the main wildlife reservoir for Mycobacterium bovis (M. bovis), the causal agent of bovine tuberculosis (bTB). The ability to diagnose M. bovis infection in badgers is critical to understanding the epidemiology of the infection in this species and for informing control strategies. In this study we determined the sensitivity and specificity of a lateral flow assay (Dual Path Platform (DPP) VetTB assay) to identify infected live badgers using two blood sample types: fresh whole blood (suitable for immediate testing in the field without further processing) and serum (which can be stored frozen for batch testing). Two measures were used for the interpretation of test results: qualitative visual interpretation and quantitative measurement using an optical reader for a range of cut-offs. To overcome the absence of a gold standard comparison test, we used Bayesian latent class methods, applied to results from different sub-populations. Regardless of sample type, the highest sensitivity and specificity of the DPP under qualitative interpretation were obtained using Band 1 (MPB83 antigen) results. Median estimates (95% CIs) of sensitivity and specificity were 79.9% (66.1-91.4%) and 93.3% (90.7-95.7%), respectively for whole blood and 53.0% (43.0-63.7%) and 96.3% (94.7-97.7%), respectively for serum. Band 2 (ESAT-6/CFP-10), when interpreted on its own, had median sensitivity estimates of 21.4% (12.0-32.4%) for whole blood, and 6.8% (3.3-11.9%) for serum. When using Band 1 results from the optical reader, the estimate of sensitivity for whole blood was higher than for serum across the whole range of cut-offs, though with a concomitant reduction in specificity. This study provides reliable estimates of test characteristics for the DPP when applied to whole blood and serum. The results support the use of the DPP test in a field application to identify infected live badgers using whole blood samples.

Accurate estimation of diagnostic test performance is crucial for epidemiological studies and disease control programs. Bayesian latent class models (BLCMs) provide a robust statistical approach to estimate these parameters in the absence of a gold standard test. This study aimed to establish a proof of concept for interlaboratory diagnostic test evaluation and to develop metrics for model fit validation, using serological detection of bovine viral diarrhoea as a case study. A total of 485 samples were collected from France, the Netherlands, Sweden and the United Kingdom and tested in four laboratories using six commercial ELISA kits. We initially fit a 6-test-4-population Hui-Walter model with both minimally informative and strong priors, as well as covariance terms. Model fit was assessed through four novel posterior predictive metrics, targeting the multinomial response frequency (LPmf), test-specific positivity (LPtp), pairwise crude agreement (LPag) and population-specific re-estimation of sensitivity and specificity (LRse/LRsp). BLCM results showed that almost all tests exhibited high sensitivity and specificity (>95 %). In addition, the model fit metrics identified one test breaching the assumption of constant test performance across populations which was therefore removed from the final model. This highlights the importance of robust model validation strategies to ensure reliable estimates. Our findings demonstrate that the joint evaluation of diagnostic tests across laboratories using BLCMs is both feasible and effective, providing robust accuracy estimates while reducing the burden on individual laboratories. As this approach does not require characterized samples, it is readily adaptable for evaluating diagnostics for emerging diseases without established gold standards.

High pathogenicity avian influenza viruses (HPAIVs) cause high morbidity and mortality in poultry species. HPAIV prevalence means high numbers of infected wild birds could lead to spill over events for farmed poultry. How these pathogens survive in the environment is important for disease maintenance and potential dissemination. We evaluated the temperature-associated survival kinetics for five clade 2.3.4.4 H5Nx HPAIVs (UK field strains between 2014 and 2021) incubated at up to three temperatures for up to ten weeks. The selected temperatures represented northern European winter (4 °C) and summer (20 °C); and a southern European summer temperature (30 °C). For each clade 2.3.4.4 HPAIV, the time in days to reduce the viral infectivity by 90% at temperature T was established (DT), showing that a lower incubation temperature prolonged virus survival (stability), where DT ranged from days to weeks. The fastest loss of viral infectivity was observed at 30 °C. Extrapolation of the graphical DT plots to the x-axis intercept provided the corresponding time to extinction for viral decay. Statistical tests of the difference between the DT values and extinction times of each clade 2.3.4.4 strain at each temperature indicated that the majority displayed different survival kinetics from the other strains at 4 °C and 20 °C.

BACKGROUND:Consumption of pork and pork products is a major source of human infection with Salmonella. Salmonella is typically subclinical in pigs, making it difficult to identify infected pigs. Therefore, effective surveillance of Salmonella in pigs critically relies on good knowledge on how well the diagnostic tests used perform. A test that has been used in several countries for Salmonella monitoring is serological testing of meat juice using an ELISA (MJ ELISA) to detect antibodies against Salmonella. This MJ ELISA data could be used to estimate infection prevalence and trends. However, as the MJ ELISA output is a sample-to-positive (S/P) ratio, which is a continuous outcome rather than a binary (positive/negative) result, the interpretation of this data depends upon a chosen cut-off. AIM:To apply Bayesian latent class models (BLCMs) to estimate diagnostic accuracy of the MJ ELISA test values in the absence of a gold standard without needing to apply a cut-off. METHODS AND RESULTS:BLCMs were fitted to data from a UK abattoir survey carried out in 2006 in order to estimate the diagnostic accuracy of MJ ELISA with respect to the prevalence of active Salmonella infection. This survey consisted of a MJ ELISA applied in parallel with the bacteriological testing of caecal contents, carcass swabs and lymph nodes (n = 625). A BLCM was also fitted to the same data but with dichotomisation of the MJ ELISA results, in order to compare with the model using continuous outcomes. Estimates were obtained for sensitivity and specificity of the ELISA over a range of S/P values and for the bacteriological tests and were found to be similar between the models using continuous and dichotomous ELISA outcomes. CONCLUSION:The Bayesian method without specifying a cut-off does allow prevalence to be inferred without specifying a cut-off for the ELISA. The study results will be useful for estimating infection prevalence from serological surveillance data.

Methodologies for source attribution (SA) of foodborne illnesses comprise a rapidly expanding suite of techniques for estimating the most important source or sources of human infection. Recently, the increasing availability of whole genome sequencing (WGS) data for a wide range of bacterial strains has led to the development of novel SA methods. These techniques utilize the unique features of bacterial genomes adapted to different host types and hence offer increased resolution of the outputs. Comparative studies of different SA techniques reliant on WGS data are currently lacking. Here, we critically assessed and compared the outputs of three SA methods: a supervised classification random forest machine learning algorithm (RandomForest), an Accessory genes-Based Source Attribution method (AB_SA), and a Bayesian frequency matching method (Bayesian). Each technique was applied to the WGS data of a panel of 902 reservoir host and human monophasic and biphasic Salmonella enterica subsp. enterica serovar Typhimurium isolates sampled in the British Isles (BI) and Denmark from 2012 to 2016. Additionally, for RandomForest and Bayesian, we explored whether utilization of accessory genome features as model inputs improved attribution accuracy of these methods over using the core genome derived features only. Results indicated that this was the case for RandomForest, but for Bayesian the overall attribution estimates varied little regardless of the inclusion or not of the accessory genome features. All three methods attributed the vast majority of human isolates to the Pigs primary source class, which was expected given the known high relative prevalence rates in pigs, and hence routes of infection into the human population, of monophasic and biphasic S. Typhimurium in the BI and Denmark. The accuracy of AB_SA was lower than of RandomForest when attributing the primary source classes to the 120 animal test set isolates with known primary sources. A major advantage of both AB_SA and Bayesian was a much faster execution time as compared to RandomForest. Overall, the SA method comparison presented in this study describes the strengths and weaknesses of each of the three methods applied to attributing potential monophasic and biphasic S. Typhimurium animal sources to human infections that could be valuable when deciding which SA methodology would be the most applicable to foodborne disease outbreak scenarios involving monophasic and biphasic S. Typhimurium.

Atypical scrapie (AS) is a transmissible spongiform encephalopathy (TSE) that affects sheep and goats. Low within-flock incidence suggests that AS is not transmissible between animals, and testing of all animals that exit positive flocks for two years following detection (i.e., intensified monitoring) used to be carried out in the EU to provide data to test this. This intensified monitoring stopped in 2021 but continues in Great Britain (GB). The aim of this study was to predict the number of AS cases missed if this monitoring were also stopped in GB, using a combination of statistical and transmission modelling. The number of AS cases estimated to be missed if the intensified monitoring was stopped was low relative to the number of AS cases detected in other active surveillance streams (e.g., fallen stock and abattoir surveys), at approximately 1 case every 3 years (0.34 per year, 95% CI: 0.18-0.54) compared to 10 per year (95% CI: 4-17) in the active surveillance stream. This suggests that stopping the intensive monitoring of AS would have relatively little impact on AS surveillance and on the power of the available AS data to infer whether AS is contagious.

The emergence and epidemic of classical Bovine Spongiform Encephalopathy (cBSE) represents one of the most important and unique episodes in disease control of a zoonotic disease due to its novelty and its impact. Since its detection in 1986 in the United Kingdom, it has also been detected in 25 countries. The novel nature of its infectious agent and the discovery of its zoonotic potential (causing the new variant of Creutzfeldt-Jakob disease in humans) caught the food industry, policy makers, scientific community and consumers off guard, with concerns over massive human exposure and health impact. Thirty-five years later, and following the feed bans of mammalian protein to livestock in 1996, the epidemic is now in its final stages, with expectations of occasional cases emerging until the year 2026. In the last six years, two cBSE cases from animals Born After the Reinforced feed Ban (BARB) have been identified in Scotland and England, delaying their application for BSE Negligible status. This paper provides a current and historical analysis of the cBSE epidemic situation in Great Britain and review the policies implemented, its impact and the possible factors explaining the occurrence of new cases. The analysis and review reinforce the hypothesis that cBSE BARB cases occurrence may not be spontaneous, yet there remains much uncertainty of their aetiology. To date, 181,122 cBSE cases have been detected in Great Britain, of which 178 are BARB cases; and 178 human cases of the new variant of Creutzfeldt-Jakob have been diagnosed. The disease triggered major policy responses in the country, and worldwide, that have transformed the industry and our approach to animal health. Almost all its impact originated from societal reactions to the disease, from disposal of animals and products, to reduction of the national herd and its production efficiency, losses through trade restrictions and reduction in market prices and consumers' confidence, hardening of cleaning and control procedures in farms and hospitals, generation of heavy government investment plans through numerous support, surveillance and research schemes, and political and societal changes. BSE is an example of major system shock to a food industry, but which experience has resulted in better traceability systems of animals, increased capacity to develop robust diagnostic methods, numerous lessons learnt on policy coordination, implementation and communication, increased society awareness on food systems and overall improved the country's preparedness to future epidemics.

Peste des petits ruminants (PPR) is an acute infectious disease of small ruminants targeted for global eradication by 2030. The Global Strategy for Control and Eradication (GSCE) recommends mass vaccination targeting 70% coverage of small ruminant populations in PPR-endemic regions. These small ruminant populations are diverse with heterogeneous mixing patterns that may influence PPR virus (PPRV) transmission dynamics. This paper evaluates the impact of heterogeneous mixing on (i) PPRV transmission and (ii) the likelihood of different vaccination strategies achieving PPRV elimination, including the GSCE recommended strategy. We develop models simulating heterogeneous transmission between hosts, including a metapopulation model of PPRV transmission between villages in lowland Ethiopia fitted to serological data. Our results demonstrate that although heterogeneous mixing of small ruminant populations increases the instability of PPRV transmission—increasing the chance of fadeout in the absence of intervention—a vaccination coverage of 70% may be insufficient to achieve elimination if high-risk populations are not targeted. Transmission may persist despite very high vaccination coverage (>90% small ruminants) if vaccination is biased towards more accessible but lower-risk populations such as sedentary small ruminant flocks. These results highlight the importance of characterizing small ruminant mobility patterns and identifying high-risk populations for vaccination and support a move towards targeted, risk-based vaccination programmes in the next phase of the PPRV eradication programme. Our modelling approach also illustrates a general framework for incorporating heterogeneous mixing patterns into models of directly transmitted infectious diseases where detailed contact data are limited. This study improves understanding of PPRV transmission and elimination in heterogeneous small ruminant populations and should be used to inform and optimize the design of PPRV vaccination programmes.

As part of the measures to reduce the prevalence of Salmonella in poultry in the UK, National Control Programmes (NCPs) have been implemented. These involve regular statutory testing of poultry holdings to monitor and estimate the prevalence of Salmonella in the national flock population and to control Salmonella on holdings with positive flocks, especially those serovars most identified with human illness: Salmonella Enteritidis (SE) and S. Typhimurium (ST). It is very important to ensure that the level of testing is appropriate so that it is sufficiently effective to identify positive flocks and to monitor prevalence, but also efficient in the use of resources. The aim of this study was to estimate the sensitivity of both the Operator and Competent Authority (CA) Official sampling used to detect infected flocks, and to also estimate the true proportion of infected holdings of commercial laying flocks in GB each year of the NCP, along with the trend of any changes in prevalence for both SE/ST and non-SE/ST. A Bayesian model was developed to estimate the sensitivity of both Operator and CA Official sampling from the NCP data 2009-2018, and to estimate the true prevalence of infected holdings. The model estimate for the prevalence of infected holdings for the first complete year of the NCP was 3.9% (95% Credible Interval (CI) 2.8-6.2%) for non-SE/ST and 0.8% (95% CI: 0.4%-1.5%) for SE/ST. Prevalence had reduced to 1.6% (non-SE/ST) (95% CI 1.0%-2.5%) and 0.2% (SE/ST) (95% CI 0.1%-0.4%) in 2018. Results indicated a very low sensitivity of Operator sampling (~9%), but a much higher sensitivity of CA Official sampling (~44%). The true prevalence of Salmonella infected holdings in the UK had a mean average reduction of 10.6% (95% CI: 6.3%-15.1%) per annum (non-SE/ST) and 15.9% (95% CI: 6.0%-19.8%) annual reduction for SE/ST. This has shown the effectiveness of the NCP for Salmonella in commercial laying flocks, with reductions in Salmonella overall more or less equal to the target reduction for regulated serovars of 10% per annum. The true prevalence of SE/ST was estimated to be below the final target of less than 2% in every year and was below 0.5% at the end of the 10 year period.

Background: Mobility restrictions prevent the spread of infections to disease-free areas, and early in the coronavirus disease 2019 (COVID-19) pandemic, most countries imposed severe restrictions on mobility as soon as it was clear that containment of local outbreaks was insufficient to control spread. These restrictions have adverse impacts on the economy and other aspects of human health, and it is important to quantify their impact for evaluating their future value. Methods: Here we develop Scotland Coronavirus transmission Model (SCoVMod), a model for COVID-19 in Scotland, which presents unusual challenges because of its diverse geography and population conditions. Our fitted model captures spatio-temporal patterns of mortality in the first phase of the epidemic to a fine geographical scale. Results: We find that lockdown restrictions reduced transmission rates down to an estimated 12\% of its pre-lockdown rate. We show that, while the timing of COVID-19 restrictions influences the role of the transmission rate on the number of COVID-related deaths, early reduction in long distance movements does not. However, poor health associated with deprivation has a considerable association with mortality; the Council Area (CA) with the greatest health-related deprivation was found to have a mortality rate 2.45 times greater than the CA with the lowest health-related deprivation considering all deaths occurring outside of carehomes. Conclusions: We find that in even an early epidemic with poor case ascertainment, a useful spatially explicit model can be fit with meaningful parameters based on the spatio-temporal distribution of death counts. Our simple approach is useful to strategically examine trade-offs between travel related restrictions and physical distancing, and the effect of deprivation-related factors on outcomes.

Background: Mobility restrictions prevent the spread of infections to disease-free areas, and early in the coronavirus disease 2019 (COVID-19) pandemic, most countries imposed severe restrictions on mobility as soon as it was clear that containment of local outbreaks was insufficient to control spread. These restrictions have adverse impacts on the economy and other aspects of human health, and it is important to quantify their impact for evaluating their future value. Methods: Here we develop Scotland Coronavirus transmission Model (SCoVMod), a model for COVID-19 in Scotland, which presents unusual challenges because of its diverse geography and population conditions. Our fitted model captures spatio-temporal patterns of mortality in the first phase of the epidemic to a fine geographical scale. Results: We find that lockdown restrictions reduced transmission rates down to an estimated 12\% of its pre-lockdown rate. We show that, while the timing of COVID-19 restrictions influences the role of the transmission rate on the number of COVID-related deaths, early reduction in long distance movements does not. However, poor health associated with deprivation has a considerable association with mortality; the Council Area (CA) with the greatest health-related deprivation was found to have a mortality rate 2.45 times greater than the CA with the lowest health-related deprivation considering all deaths occurring outside of carehomes. Conclusions: We find that in even an early epidemic with poor case ascertainment, a useful spatially explicit model can be fit with meaningful parameters based on the spatio-temporal distribution of death counts. Our simple approach is useful to strategically examine trade-offs between travel related restrictions and physical distancing, and the effect of deprivation-related factors on outcomes.

Bacille Calmette-Guerin (BCG) is a potential tool in the control of Mycobacterium bovis in European badgers (Meles meles). A five year Test and Vaccinate or Remove (TVR) research intervention project commenced in 2014 using two BCG strains (BCG Copenhagen 1331 (Years 1-3/ BadgerBCG) and BCG Sofia SL2222 (Years 4-5). Badgers were recaptured around 9 weeks after the Year 5 vaccination and then again a year later.The Dual-Path Platform (DPP) Vet TB assay was used to detect serological evidence of M. bovis infection. Of the 48 badgers, 47 had increased Line 1 readings (MPB83 antigen) between the Year 5 vaccination and subsequent recapture. The number of BCG Sofia vaccinations influenced whether a badger tested positive to the recapture DPP VetTB assay Line 1 (p < 0.001) while the number of BadgerBCG vaccinations did not significantly affect recapture Line 1 results (p = 0.59). Line 1 relative light units (RLU) were more pronounced in tests run with sera than whole blood. The results from an in_house MPB83 ELISA results indicated that the WB DPP VetTB assay may not detect lower MPB83 IgG levels as well as the serum DPP VetTB assay.Changes in interferon gamma assay (IFN-c) results were seen in 2019 with significantly increased CFP-10 and PPDB readings. Unlike BadgerBCG, BCG Sofia induces an immune response to MPB83 (the immune dominant antigen in M. bovis badger infection) that then affects the use of immunodiagnostic tests. The use of the DPP VetTB assay in recaptured BCG Sofia vaccinated badgers within the same trapping season is precluded and caution should be used in badgers vaccinated with BCG Sofia in previous years. The results suggest that the DPP VetTB assay can be used with confidence in badgers vaccinated with BadgerBCG as a single or repeated doses.Crown Copyright CO 2022 Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

The incidence of bovine tuberculosis (TB, caused by Mycobacterium bovis) in cattle has been associated with TB in badgers (Meles meles) in parts of England. The aim was to identify badger-associated M. bovis reservoirs in the Edge Area, between the High- and Low-Risk Areas for cattle TB. Data from badger TB surveys were sparse. Therefore, a definition for a local M. bovis reservoir potentially shared by cattle and badgers was developed using cattle TB surveillance data. The performance of the definition was estimated through Latent Class Analysis using badger TB survey data. Spatial units (25 km(2)) in the Edge Area were classified as having a reservoir if they had (i) at least one TB incident in at least three of the previous 7 years, (ii) at least one TB incident in a cattle herd confirmed by post-mortem tests as due to M. bovis infection and not attributable to cattle movements in the previous 2 years and (iii) more confirmed TB incidents than un-confirmed in the previous 2 years. Approximately 20% of the Edge Area was classified as having a local M. bovis reservoir using the cattle-based definition. Assuming 15% TB prevalence in Edge Area badgers, sensitivity for the local M. bovis reservoir definition varied from 25.7% [95% credible interval (CrI): 10.7%-85.1%] to 64.8% (95% CrI: 48.1%-88.0%). Specificity was 91.9% (CrI: 83.6%-97.4%). Over 90% of the local reservoir was in stable endemic TB areas identified through previous work and its spatial distribution was largely consistent with local veterinary knowledge. Uncertainty in the reservoir spatial distribution was explored through its recalculation in spatial units shifted in different directions. We recommend that the definition is re-evaluated as further data on badger infection with M. bovis become available.

The purpose of the study was to apply a Bayesian source attribution model to England and Wales based data on Salmonella Typhimurium (ST) and monophasic variants (MST), using different subtyping approaches based on sequence data. The data consisted of laboratory confirmed human cases and mainly livestock samples collected from surveillance or monitoring schemes. Three different subtyping methods were used, 7-loci Multi-Locus Sequence Typing (MLST), Core-genome MLST, and Single Nucleotide Polymorphism distance, with the impact of varying the genetic distance over which isolates would be grouped together being varied for the latter two approaches. A Bayesian frequency matching method, known as the modified Hald method, was applied to the data from each of the subtyping approaches. Pigs were found to be the main contributor to human infection for ST/MST, with approximately 60% of human cases attributed to them, followed by other mammals (mostly horses) and cattle. It was found that the use of different clustering methods based on sequence data had minimal impact on the estimates of source attribution. However, there was an impact of genetic distance over which isolates were grouped: grouping isolates which were relatively closely related increased uncertainty but tended to have a better model fit.