A framework for assessing confidence in freedom from infection in animal disease control programmes.

In the Surveillance Tool for Outcome-based Comparison of FREEdom from infection (STOC free) project (https://www.stocfree.eu), a data collection tool was constructed to facilitate standardised collection of input data, and a model was developed to allow a standardised and harmonised comparison of the outputs of different control programmes (CPs) for cattle diseases. The STOC free model can be used to evaluate the probability of freedom from infection for herds in CPs and to determine whether these CPs comply with the European Union's pre-defined output-based standards. Bovine viral diarrhoea virus (BVDV) was chosen as the case disease for this project because of the diversity in CPs in the six participating countries. Detailed BVDV CP and risk factor information was collected using the data collection tool. For inclusion of the data in the STOC free model, key aspects and default values were quantified. A Bayesian hidden Markov model was deemed appropriate, and a model was developed for BVDV CPs. The model was tested and validated using real BVDV CP data from partner countries, and corresponding computer code was made publicly available. The STOC free model focuses on herd-level data, although that animal-level data can be included after aggregation to herd level. The STOC free model is applicable to diseases that are endemic, given that it needs the presence of some infection to estimate parameters and enable convergence. In countries where infection-free status has been achieved, a scenario tree model could be a better suited tool. Further work is recommended to generalise the STOC free model to other diseases.

Dans le cadre du projet européen STOC free (Surveillance Tool for Outcome-based Comparison of FREEdom from infection, outil de surveillance permettant de comparer les probabilités d'absence d'infection sur la base des résultats, https://www.stocfree.eu), un outil de recueil des données a été construit pour faciliter une collecte normalisée des données d'entrée ; un modèle a également été élaboré pour permettre une comparaison normalisée et harmonisée des données sur les résultats des différents programmes de contrôle des maladies des bovins. Le modèle STOC free peut être utilisé pour évaluer la probabilité d'absence d'infection au sein des troupeaux dans le cadre des programmes de contrôle et déterminer si ces programmes sont conformes aux normes définies par l'Union européenne en termes de résultats attendus. L'infection par le virus de la diarrhée virale bovine a été choisie comme maladie d'étude pour ce projet en raison de la diversité des programmes de contrôle dans les six pays participants. Les informations relatives aux programmes de contrôle et aux facteurs de risque d'infection ont été recueillies à l'aide de l'outil de collecte des données. Les aspects clés et valeurs par défaut ont été quantifiés en vue d'être inclus dans le modèle STOC free. Un modèle de Markov caché dont les paramètres sont estimés par inférence bayésienne a été considéré comme le plus adapté et développé pour une application aux données issues des programmes de contrôle de la diarrhée virale bovine. Ce modèle a été testé et validé en utilisant des données réelles des programmes de contrôle du virus de la diarrhée virale bovine des pays participants ; le code informatique correspondant a été rendu public. Le modèle STOC free utilise des données au niveau des troupeaux, même si des données au niveau des animaux individuels peuvent être incluses une fois agrégées au niveau du troupeau. Le modèle STOC free s'applique aux maladies endémiques, puisqu'un certain niveau de présence de l'infection est nécessaire pour estimer les paramètres et permettre la convergence. Dans les pays ayant obtenu le statut indemne d'infection, un modèle du type arbre de scénario pourrait être un outil plus adapté. Des travaux supplémentaires sont recommandés pour généraliser le modèle STOC free à d'autres maladies.

Como parte del proyecto europeo STOC free (Surveillance Tool for Outcome-based Comparison of FREEdom from infection, herramienta de vigilancia para comparaciones por resultados respecto a la ausencia de infecciones, https://www.stocfree.eu), se confeccionó una herramienta de obtención de datos para facilitar la recogida normalizada de datos entrantes y se elaboró un modelo que posibilitara una comparación normalizada y armonizada de los resultados (datos salientes) de distintos programas de control de enfermedades bovinas. El modelo STOC free puede servir para calcular la probabilidad de ausencia de infección en los rebaños como parte de los programas de control y para determinar si estos programas se ajustan a las normas predefinidas de resultados de la Unión Europea. Como ejemplo de estudio para el proyecto se eligió el virus de la diarrea viral bovina (virus DVB) por la diversidad que presentaban los correspondientes programas de control de los seis países participantes. Empleando la herramienta de obtención de datos, se reunió información pormenorizada de los programas de control del virus DVB y los factores de riesgo. Para incluir los datos en el modelo STOC free, se cifraron unos aspectos clave y valores predeterminados Juzgando conveniente el empleo de un modelo oculto de Markov cuyos parámetros se estiman por inferencia bayesiana, se elaboró un modelo de esta índole aplicable a los programas de control del virus DVB. Para ensayar y validar el modelo se utilizaron datos reales de los programas de control del virus DVB de los países participantes, tras lo cual se hizo público el correspondiente código informático. El modelo STOC free trabaja con los datos por rebaño, aunque tras la agregación por rebaños pueden incluirse también datos por individuo. Para que este modelo sea aplicable a una enfermedad es preciso que esta sea endémica, pues el modelo requiere la presencia de cierto nivel de infección para calcular los parámetros y determinar convergencias. En aquellos países donde ya esté reconocida la ausencia de infección, sería más apropiado utilizar como herramienta un modelo de árbol de hipótesis. Los autores recomiendan ahondar en esta línea de trabajo para poder extender a otras enfermedades el uso del modelo STOC free.

A framework for evaluating health system surveillance sensitivity to support public health decision-making for malaria elimination: a case study from Indonesia.

BACKGROUND: The effectiveness of a surveillance system to detect infections in the population is paramount when confirming elimination. Estimating the sensitivity of a surveillance system requires identifying key steps in the care-seeking cascade, from initial infection to confirmed diagnosis, and quantifying the probability of appropriate action at each stage. Using malaria as an example, a framework was developed to estimate the sensitivity of key components of the malaria surveillance cascade. METHODS: Parameters to quantify the sensitivity of the surveillance system were derived from monthly malaria case data over a period of 36 months and semi-quantitative surveys in 46 health facilities on Java Island, Indonesia. Parameters were informed by the collected empirical data and estimated by modelling the flow of an infected individual through the system using a Bayesian framework. A model-driven health system survey was designed to collect empirical data to inform parameter estimates in the surveillance cascade. RESULTS: Heterogeneity across health facilities was observed in the estimated probability of care-seeking (range = 0.01-0.21, mean ± sd = 0.09 ± 0.05) and testing for malaria (range = 0.00-1.00, mean ± sd = 0.16 ± 0.29). Care-seeking was higher at facilities regularly providing antimalarial drugs (Odds Ratio [OR] = 2.98, 95% Credible Intervals [CI]: 1.54-3.16). Predictably, the availability of functioning microscopy equipment was associated with increased odds of being tested for malaria (OR = 7.33, 95% CI = 20.61). CONCLUSIONS: The methods for estimating facility-level malaria surveillance sensitivity presented here can help provide a benchmark for what constitutes a strong system. The proposed approach also enables programs to identify components of the health system that can be improved to strengthen surveillance and support public-health decision-making.

A description and qualitative comparison of the elements of heterogeneous bovine viral diarrhea control programs that influence confidence of freedom.

For endemic infections in cattle that are not regulated at the European Union level, such as bovine viral diarrhea virus (BVDV), European Member States have implemented control or eradication programs (CEP) tailored to their specific situations. Different methods are used to assign infection-free status in CEP; therefore, the confidence of freedom associated with the "free" status generated by different CEP are difficult to compare, creating problems for the safe trade of cattle between territories. Safe trade would be facilitated with an output-based framework that enables a transparent and standardized comparison of confidence of freedom for CEP across herds, regions, or countries. The current paper represents the first step toward development of such a framework by seeking to describe and qualitatively compare elements of CEP that contribute to confidence of freedom. For this work, BVDV was used as a case study. We qualitatively compared heterogeneous BVDV CEP in 6 European countries: Germany, France, Ireland, the Netherlands, Sweden, and Scotland. Information about BVDV CEP that were in place in 2017 and factors influencing the risk of introduction and transmission of BVDV (the context) were collected using an existing tool, with modifications to collect information about aspects of control and context. For the 6 participating countries, we ranked all individual elements of the CEP and their contexts that could influence the probability that cattle from a herd categorized as BVDV-free are truly free from infection. Many differences in the context and design of BVDV CEP were found. As examples, CEP were either mandatory or voluntary, resulting in variation in risks from neighboring herds, and risk factors such as cattle density and the number of imported cattle varied greatly between territories. Differences were also found in both testing protocols and definitions of freedom from disease. The observed heterogeneity in both the context and CEP design will create difficulties when comparing different CEP in terms of confidence of freedom from infection. These results highlight the need for a standardized practical methodology to objectively and quantitatively determine confidence of freedom resulting from different CEP around the world.

Estimating the probability of freedom from bovine brucellosis in the Galapagos Islands.

Bovine brucellosis is a worldwide zoonotic disease that still burdens several countries in the Mediterranean, Asia, Africa and Latin America. Although the disease is present in Ecuador, the Galapagos Islands seem to be free from the disease based on a survey conducted in 1997 where all tested animals showed negative results. This study aimed at estimating the probability of freedom from brucellosis in this Ecuadorian province in 2014. A survey was implemented on the three main cattle-producing islands of the province: Santa Cruz, Isabela and San Cristóbal. Thirty-three cattle farms and 410 cattle were tested for brucellosis using the Rose Bengal test and indirect ELISA. All animals showed negative results for both tests. Probability of freedom was estimated at 98%, 91% and 88% for Santa Cruz, Isabela and San Cristóbal, respectively, considering a herd-level design seroprevalence of 20% and animal-level design seroprevalence of 15%, and assuming a perfect specificity of the survey. The negative results found in 1997 and present surveys suggest that the Galapagos Islands are free from bovine brucellosis.

Evaluation of herd-level sampling strategies for control of Salmonella in Swedish cattle.

Based on Swedish legislation, all herds where Salmonella of any serotype is detected are put under restrictions, and measures aiming at eradication are required. Costs for sampling and control have increased in recent years and the aim of this study was to investigate the efficiency of different sampling strategies. We also compiled test results from recent surveillance activities and used these to complement and compare with calculated results. Sensitivities and specificities at group and herd level were calculated for different test strategies. A scenario-tree modeling approach was used to account for the hierarchy of animals within herds, and different relative risk of salmonella in different age groups. Negative and positive predictive values (NPV and PPV), and probability of freedom from Salmonella were calculated to compare the added value of different sampling strategies. Results showed that more fecal samples than serological samples per group are needed to reach a group sensitivity >0.50. This also means that serological testing leads to a higher NPV. For example, with 10 negative test-results from a group of 25 animals in a herd with a suspicion of Salmonella, the NPV based on serology was 0.75 and based on culture was 0.56. For the PPV, testing based on culture from fecal sampling was superior, as specificity of such testing was close to perfect. By changing the threshold for considering a group positive, from 1 test-positive animal to 2, the PPV of serological results could be increased without substantial loss in NPV. The herd sensitivity based on (1) bulk milk sampling, (2) fecal sampling of all animals, and (3) bulk milk sampling and individual sera from 20 animals within each age group was 0.53, 0.88, and 0.95, respectively. In low-prevalence regions, this sensitivity was enough to verify a high probability of freedom (>0.99), as the probability of infection in such Swedish regions has been shown to be 0.01. For herds with a higher prior probability of infection, repeated sampling (2-9 sampling occasions) was needed to reach the same level of confidence. Analysis of surveillance data indicated that boot swabs can be used to replace the standard fecal sampling presently used in Sweden. It was also confirmed that the individual specificity of the tests used for serological testing of Swedish calves is high (0.99). The results can form a basis for fit-for-purpose testing strategies (e.g., surveillance or prepurchase testing).

Review state-of-the-art of output-based methodological approaches for substantiating freedom from infection.

A wide variety of control and surveillance programmes that are designed and implemented based on country-specific conditions exists for infectious cattle diseases that are not regulated. This heterogeneity renders difficult the comparison of probabilities of freedom from infection estimated from collected surveillance data. The objectives of this review were to outline the methodological and epidemiological considerations for the estimation of probabilities of freedom from infection from surveillance information and review state-of-the-art methods estimating the probabilities of freedom from infection from heterogeneous surveillance data. Substantiating freedom from infection consists in quantifying the evidence of absence from the absence of evidence. The quantification usually consists in estimating the probability of observing no positive test result, in a given sample, assuming that the infection is present at a chosen (low) prevalence, called the design prevalence. The usual surveillance outputs are the sensitivity of surveillance and the probability of freedom from infection. A variety of factors influencing the choice of a method are presented; disease prevalence context, performance of the tests used, risk factors of infection, structure of the surveillance programme and frequency of testing. The existing methods for estimating the probability of freedom from infection are scenario trees, Bayesian belief networks, simulation methods, Bayesian prevalence estimation methods and the STOC free model. Scenario trees analysis is the current reference method for proving freedom from infection and is widely used in countries that claim freedom. Bayesian belief networks and simulation methods are considered extensions of scenario trees. They can be applied to more complex surveillance schemes and represent complex infection dynamics. Bayesian prevalence estimation methods and the STOC free model allow freedom from infection estimation at the herd-level from longitudinal surveillance data, considering risk factor information and the structure of the population. Comparison of surveillance outputs from heterogeneous surveillance programmes for estimating the probability of freedom from infection is a difficult task. This paper is a 'guide towards substantiating freedom from infection' that describes both all assumptions-limitations and available methods that can be applied in different settings.

Output-based assessment of herd-level freedom from infection in endemic situations: Application of a Bayesian Hidden Markov model.

Countries have implemented control programmes (CPs) for cattle diseases such as bovine viral diarrhoea virus (BVDV) that are tailored to each country-specific situation. Practical methods are needed to assess the output of these CPs in terms of the confidence of freedom from infection that is achieved. As part of the STOC free project, a Bayesian Hidden Markov model was developed, called STOC free model, to estimate the probability of infection at herd-level. In the current study, the STOC free model was applied to BVDV field data in four study regions, from CPs based on ear notch samples. The aim of this study was to estimate the probability of herd-level freedom from BVDV in regions that are not (yet) free. We additionally evaluated the sensitivity of the parameter estimates and predicted probabilities of freedom to the prior distributions for the different model parameters. First, default priors were used in the model to enable comparison of model outputs between study regions. Thereafter, country-specific priors based on expert opinion or historical data were used in the model, to study the influence of the priors on the results and to obtain country-specific estimates. The STOC free model calculates a posterior value for the model parameters (e.g. herd-level test sensitivity and specificity, probability of introduction of infection) and a predicted probability of infection. The probability of freedom from infection was computed as one minus the probability of infection. For dairy herds that were considered free from infection within their own CP, the predicted probabilities of freedom were very high for all study regions ranging from 0.98 to 1.00, regardless of the use of default or country-specific priors. The priors did have more influence on two of the model parameters, herd-level sensitivity and the probability of remaining infected, due to the low prevalence and incidence of BVDV in the study regions. The advantage of STOC free model compared to scenario tree modelling, the reference method, is that actual data from the CP can be used and estimates are easily updated when new data becomes available.

Capacity of a Bayesian model to detect infected herds using disease dynamics and risk factor information from surveillance programmes: A simulation study.

Control programmes against non-regulated infectious diseases of farm animals are widely implemented. Different control programmes have different definitions of "freedom from infection" which can lead to difficulties when trading animals between countries. When a disease is still present, in order to identify herds that are safe to trade with, estimating herd-level probabilities of being infected when classified "free from infection" using field data is of major interest. Our objective was to evaluate the capacity of a Bayesian Hidden Markov Model, which computes a herd-level probability of being infected, to detect infected herds compared to using test results only. Herd-level risk factors, infection dynamics and associated test results were simulated in a population of herds, for a wide range of realistic infection contexts and test characteristics. The model was used to predict the infection status of each herd from longitudinal data: a simulated risk factor and a simulated test result. Two different indexes were used to categorize herds from the probability of being infected into a herd predicted status. The model predictive performances were evaluated using the simulated herd status as the gold standard. The model detected more infected herds than a single final test in 85 % of the scenarios which converged. The proportion of infected herds additionally detected by the model, compared to test results alone, varied depending on the context. It was higher in a context of a low herd test sensitivity. On average, around 20 %, for high test sensitivity scenarios, and 40 %, for low test sensitivity scenarios, of infected herds that were undetected by the test were accurately classified as infected by the model. Model convergence did not occur for 39 % of the scenarios, mainly in association with low herd test sensitivity. Detection of additional newly infected herds was always associated with an increased number of false positive herds (except for one scenario). The number of false positive herds was lower for scenarios with low herd test sensitivity and moderate to high incidence and prevalence. These results highlight the benefit of the model, in particular for control programmes with infection present at an endemic level in a population and reliance on test(s) of low sensitivity.

Risk-based surveillance of chronic wasting disease in semi-domestic reindeer.

Reindeer pastoralism is a widespread practise across Fennoscandia and Russia. An outbreak of chronic wasting disease (CWD) among wild reindeer (Rangifer tarandus) poses a severe threat to the semi-domestic reindeer herding culture. Establishing surveillance is therefore key, but current models for surveillance of CWD are designed for wild cervids and rely on samples obtained from recreational hunters. Targeting animal groups with a higher infection probability is often used for more efficient disease surveillance. CWD has a long incubation period of 2-3 years, and the animals show clinical signs in the later stages of the infection i.e. 1-4 months prior to death. The semi-domestic reindeer are free-ranging most of the year, but during slaughtering in late fall, herders stress the animals in penned areas. This allows removal of animals with deviant behaviour or physical appearance, and such removals are likely to include animals in the clinical stages of CWD if the population is infected. In Norway, the semi-domestic reindeer in Filefjell is adjacent to a previously CWD infected wild population. We developed a risk-based surveillance method for this semi-domestic setting to establish the probability of freedom from infection over time, or enable early disease detection and mitigation. The surveillance scheme with a scenario tree using three risk categories (sample category, demographic group, and deviations in behaviour or physical appearance) was more effective and less invasive as compared to the surveillance method developed for wild reindeer. We also simulated how variation in susceptibility, incubation period and time for onset of clinical signs (linked to variation in the prion protein gene, PRNP) would potentially affect surveillance. Surveillance for CWD was mandatory within EU-member states with reindeer (2018-2020). The diversity of management systems and epidemiological settings will require the development of a set of surveillance systems suitable for each different context. Our surveillance model is designed for a population with a high risk of CWD introduction requiring massive sampling, while at the same time aiming to limit adverse effects to the populations in areas of surveillance.

The impact of pooling samples on surveillance sensitivity for the megalocytivirus Infectious spleen and kidney necrosis virus.

Movements of large volumes and species varieties make the ornamental fish industry a high-risk pathway for the transfer of aquatic pathogens to new geographical regions and naïve hosts, potentially resulting in emergency disease events. Infectious spleen and kidney necrosis virus (genus Megalocytivirus) is considered exotic to Australia despite documented incursions since 2003. There are current import controls requiring freedom from infection for entry to Australia. The objective was to evaluate the effect of tissue pooling strategies for qPCR testing using a SYBR® assay for freedom from ISKNV at 2% expected prevalence with 95% confidence. Tissue homogenates from apparently healthy imported ornamental fish were tested as individuals and in pools of 5 and 10. Analytical sensitivity of the qPCR assay was reduced by two orders of magnitude when the nucleic acid extraction process was accounted for by spiking the plasmid in fish tissues and compared with molecular grade water. Diagnostic sensitivity of the assay was substantially reduced when testing tissues in pools compared with individual testing. For Population 1 (66% positive for ISKNV with moderate viral loads), surveillance sensitivity was only achieved using individual testing. For Population 2 (100% positive ISKNV with high viral loads), surveillance sensitivity was achieved using 260 fish in pools of 10 for a total of 26 tests or 200 fish in pools of 5 for 40 tests. Surveillance sensitivity could be maximized even when there was a reduction in pooled diagnostic sensitivity compared with diagnostic sensitivity for individual fish by increasing the sample size. Pooled sensitivity was influenced by the prevalence and variable virus load among fish with subclinical infections. Pooled testing is highly effective when the prevalence is >10% which should be informed by prior knowledge or pooling can be used for a screening test to rapidly identify populations with high prevalence.

The use of serosurveys following emergency vaccination, to recover the status of "foot-and-mouth disease free where vaccination is not practised".

To eliminate incursions of foot-and-mouth disease (FMD) quickly, a combination of measures, including emergency vaccination, can help block the spread of infection. For the earliest recovery of the FMD-free status for trade, without the slaughter of uninfected vaccinated animals, a serosurvey for antibodies to FMD virus non-structural proteins (NSP) must be used to substantiate absence of occult virus infections. Areas of doubt over requirements for post-vaccination serosurveillance and its feasibility include the required and achievable confidence, the amount of sampling necessary, and the appropriate responses to and consequences of different seropositive findings. This derives largely from uncertainty over the extent of localised pockets of virus infection that may remain within vaccinated populations and the circumstances that permit this. The question therefore remains whether tests are sufficiently sensitive and specific to detect and eliminate infected animals, without excessive culling of uninfected animals, before vaccinated animals mix with non-vaccinated livestock when movement restrictions are lifted. It is recommended to change the rationale for serosurveillance after emergency vaccination. Only when emergency vaccination is used in limited outbreaks is it possible to test and cull comprehensively, an approach compatible with a three-month minimum period to recover the FMD-free status. In other situations, where emergency vaccination is used, such as dealing with large outbreaks in animal-dense regions and where the onset of vaccination has been delayed, post-vaccination serosurveys should be targeted and focus on providing an assurance to detect higher levels of infection, in case of inadequate control measures. As this provides less assurance of absence of infection, the approach would be compatible with a six-month waiting period for free-status recovery and should be complemented by other methods to provide evidence that vaccination and control measures have been effectively implemented, as these are the best guarantee against continuing virus transmission.