Syndromic surveillance system based on near real-time cattle mortality monitoring

Evaluation of a voluntary passive surveillance component in cattle through notification of excess mortality

Passive surveillance can be most effective in the early detection of disease outbreaks given that farmers observe their animals daily. The European Animal Health Law states that unexplained excess mortality should be reported to the veterinary authorities. In the Netherlands, in addition to notifications to the competent authority, Royal GD is commissioned a passive surveillance component that consists of a veterinary helpdesk and postmortem examination for early detection of emerging diseases. The aim of this study was to evaluate this voluntary passive surveillance component through excess mortality in cattle. Weekly on-farm mortality was calculated using the cattle Identification and Registration records. Mortality was assessed on regional level for dairy, veal and other beef cattle using a Generalized Linear Model (GLM) (log-link, negative binomial). We used a cumulative sum of the model residuals to identify periods of excess mortality. The mortality was defined as excessive when above five times the standard error. The analysis was also conducted on herd level, but these models did not converge. We checked for an association between the two passive surveillance components elements and excess mortality. A GLM (log-link, negative binomial) with the number of contacts or submissions per region as the dependent variables and excess mortality per region and year as independent variables was carried out. Overall, the models showed significantly higher use of passive surveillance components in periods of excess mortality compared to non-excess periods. In dairy cattle the odds for contact or submission were between 1.72 (1.59-1.86) and 2.02 (1.82-2.25). For veal calves we found the odds of 2.19 (1.18-4.04) and 2.24 (1.78-2.83) relative to periods without excess mortality. Beef cattle operations, other than veal, showed only an increased odds for postmortem submissions in calves of 3.71 (2.74-5.01), submissions for cattle and contact in general was not increased for this farm type. In conclusion, the voluntary passive surveillance component in the Netherlands is used more often in periods of excess mortality in cattle. The chance of getting a timely response is highest for dairy farms. For veal calf operations the chance of receiving a timely response is more likely for postmortem submissions. A comparison with passive surveillance for excess mortality in other countries was not possible because no literature could be found. However, the method of this study can be used by other countries to evaluate their passive surveillance. This would make comparison of the performance of passive surveillance in different countries possible.

Bovine mortality: the utility of two data sources for the provision of population-level surveillance intelligence

Introduction

The use of existing data to provide surveillance intelligence is widely advocated but often presents considerable challenges. Two data sources could be used as proxies for the mortality experienced by the Scottish cattle population: deaths recorded in the mandatory register [Cattle Tracing System (CTS)] and fallen stock collections by the National Fallen Stock Company (NSFCo) with a nationwide voluntary membership.

Methods

Data for the period 2011-2016 were described and compared to establish their strengths and limitations. Similarities and differences in their temporal, seasonal and spatial patterns were examined overall, at postcode area level and for different age groups. Temporal aberration detection algorithms (TADA) were fitted.

Results

Broadly, similar patterns were observed in the two datasets; however, there were some notable differences. The observed seasonal, annual and spatial patterns match expectations, given knowledge of Scottish cattle production systems. The registry data provide more comprehensive coverage of all areas of Scotland, while collections data provide a more comprehensive measure of the mortality experienced in 0-1-month-old calves.

Discussion

Consequently, estimates of early calf mortality and their impact on the livestock sector made using CTS, or successor registers, will be under-estimates. This may apply to other registry-based systems. Fitted TADA detected points of deviations from expected norms some of which coincided in the two datasets; one with a known external event that caused increased mortality. We have demonstrated that both data sources do have the potential to be utilized to provide measures of mortality in the Scottish cattle population that could inform surveillance activities. While neither is perfect, they are complementary. Each has strengths and weaknesses, so ideally, a system where they are analyzed and interpreted in parallel would optimize the information obtained for surveillance purposes for epidemiologists, risk managers, animal health policy-makers and the wider livestock industry sector. This study provides a foundation on which to build an operational system. Further development will require improvements in the timeliness of data availability and further investment of resources.

Exact Bayesian inference of epidemiological parameters from mortality data: application to African swine fever virus

Pathogens such as African swine fever virus (ASFV) are an increasing threat to global livestock production with implications for economic well-being and food security. Quantification of epidemiological parameters, such as transmission rates and latent and infectious periods, is critical to inform efficient disease control. Parameter estimation for livestock disease systems is often reliant upon transmission experiments, which provide valuable insights in the epidemiology of disease but which may also be unrepresentative of at-risk populations and incur economic and animal welfare costs. Routinely collected mortality data are a potential source of readily available and representative information regarding disease transmission early in outbreaks. We develop methodology to conduct exact Bayesian parameter inference from mortality data using reversible jump Markov chain Monte Carlo incorporating multiple routes of transmission (e.g. within-farm secondary and background transmission from external sources). We use this methodology to infer epidemiological parameters for ASFV using data from outbreaks on nine farms in the Russian Federation. This approach improves inference on transmission rates in comparison with previous methods based on approximate Bayesian computation, allows better estimation of time of introduction and could readily be applied to other outbreaks or pathogens.

Improving the Utility of Voluntary Ovine Fallen Stock Collection and Laboratory Diagnostic Submission Data for Animal Health Surveillance Purposes: A Development Cycle

There are calls from policy-makers and industry to use existing data sources to contribute to livestock surveillance systems, especially for syndromic surveillance. However, the practical implications of attempting to use such data sources are challenging; development often requires incremental steps in an iterative cycle. In this study the utility of business operational data from a voluntary fallen stock collection service was investigated, to determine if they could be used as a proxy for the mortality experienced by the British sheep population. Retrospectively, Scottish ovine fallen stock collection data (2011-2014) were transformed into meaningful units for analysis, temporal and spatial patterns were described, time-series methods and a temporal aberration detection algorithm applied. Distinct annual and spatial trends plus seasonal patterns were observed in the three age groups investigated. The algorithm produced an alarm at the point of an historic known departure from normal (April 2013) for two age groups, across Scotland as a whole and in specific postcode areas. The analysis was then extended. Initially, to determine if similar methods could be applied to ovine fallen stock collections from England and Wales for the same time period. Additionally, Scottish contemporaneous laboratory diagnostic submission data were analyzed to see if they could provide further insight for interpretation of statistical alarms. Collaboration was required between the primary data holders, those with industry sector knowledge, plus veterinary, epidemiological and statistical expertise, in order to turn data and analytical outcomes into potentially useful information. A number of limitations were identified and recommendations were made as to how some could be addressed in order to facilitate use of these data as surveillance "intelligence." e.g., improvements to data collection and provision. A recent update of the fallen stock collections data has enabled a longer temporal period to be analyzed, with evidence of changes made in line with the recommendations. Further development will be required before a functional system can be implemented. However, there is potential for use of these data as: a proxy measure for mortality in the sheep population; complementary components in a future surveillance system, and to inform the design of additional surveillance system components.

Quantifying calf mortality on dairy farms: Challenges and solutions

In the Netherlands, the mortality rate of ear-tagged calves <1 yr is one of the indicators that is continuously monitored in census data and is defined as the number of deceased calves relative to the number of calf-days-at-risk. In 2017, yearly calf mortality rates were published in the lay press and resulted in discussions about the calculation of this parameter among stakeholders because the same parameter appeared to be calculated in many different ways by different organizations. These diverse definitions of calf mortality answered different aims such as early detection of deviations, monitoring trends, or providing insight into herd-specific results, but were difficult to understand by stakeholders. The aim of this study was to evaluate several definitions of calf mortality for scientific validity, usefulness for policymakers, and comprehensibility by farmers. Based on expert consultations, 10 definitions for calf mortality were evaluated that assessed different age categories, time periods, and denominators. Differences in definitions appeared to have a large effect on the magnitude of mortality. For example, with the original mortality parameter, the mortality rate was 16.5% per year. When the first year of life was subdivided into 3 age categories, the mortality rate was 3.3, 4.5, and 3.1% for postnatal calves (≤14 d), preweaned calves (15-55 d), and weaned calves (56 d-1 yr), respectively. Although it was logical that these mortality rates were lower than the original, the sum of the 3 separate mortality rates was also lower than the original mortality rate. The reason was that the number of calves present in a herd and the risk of mortality are not randomly distributed over a calf's first year of life and the conditional nature of mortality rates when calculated for different age categories. Ultimately, 4 parameters to monitor calf mortality in Dutch dairy herds were chosen based on scientific value, usefulness for monitoring of trends, and comprehensibility by farmers: perinatal calf mortality risk (i.e., mortality before, during, or shortly after the moment of birth up to the moment of ear-tagging), postnatal calf mortality risk (≤14 d), preweaned calf mortality rate (15-55 d), and weaned calf mortality rate (56 d-1 yr). Slight differences in definitions of parameters can have a major effect on results, and many factors have to be taken into account when defining an important health indicator such as mortality. Our evaluation resulted in a more thorough understanding of the definitions of the selected parameters and agreement by the stakeholders to use these key indicators to monitor calf mortality.

A systematic review of aberration detection algorithms used in public health surveillance

The algorithms used for detecting anomalies have evolved substantially over the last decade to take advantage of advances in informatics and to accommodate changes in surveillance data. We identified 145 studies since 2007 that evaluated statistical methods used to detect aberrations in public health surveillance data. For each study, we classified the analytic methods and reviewed the evaluation metrics. We also summarized the practical usage of the detection algorithms in public health surveillance systems worldwide. Traditional methods (e.g., control charts, linear regressions) were the focus of most evaluation studies and continue to be used commonly in practice. There was, however, an increase in the number of studies using forecasting methods and studies applying machine learning methods, hidden Markov models, and Bayesian framework to multivariate datasets. Evaluation studies demonstrated improved accuracy with more sophisticated methods, but these methods do not appear to be used widely in public health practice.

Changing surveillance objectives during the different phases of an emerging vector-borne disease outbreak: The Schmallenberg virus example

In the late summer of 2011, a sudden rise in incidence of fever, drop in milk production and diarrhoea was observed in dairy cows in the eastern region of the Netherlands and in north-western Germany. In the autumn of 2011, a novel orthobunyavirus was identified by metagenomic analyses in samples from acutely diseased cows on a farm near the German city of Schmallenberg, and was thereafter named Schmallenberg virus (SBV). Due to the novelty of the virus, there was an immediate need for knowledge regarding the epidemiological characteristics of SBV-infections to inform surveillance and control strategies. A rapid assessment of the spread and impact of an emerging disease supports decision-makers on allocation of resources. This paper reviews the disease mitigation activities during and after the SBV epidemic in the Netherlands, to illustrate the phases in surveillance when a new (vector-borne) pathogen emerges in a country or region. Immediate and short-term disease mitigation activities that were initiated after SBV was identified are discussed in detail, as well as ways to enhance future surveillance (e.g. by syndromic surveillance) and preparedness for similar disease outbreaks. By doing so, lessons learnt from the SBV epidemic will also improve surveillance for other emerging diseases in cattle.

Characterization of the Temporal Trends in the Rate of Cattle Carcass Condemnations in the US and Dynamic Modeling of the Condemnation Reasons in California With a Seasonal Component

Based on the 2016 National Cattlemen's Beef Association statistics, the cattle inventory in the US reached 93.5 million head, from which 30.5 million were commercial slaughter in 2016. California ranked fourth among all the US states that raise cattle and calves, with 5.15 million head and approximately 1.18 million slaughtered animals per year. Approximately 0.5% of cattle carcasses in the US are condemned each year, which has an important economic impact on cattle producers.In this study, we first described and compared the temporal trends of cattle carcass condemnations in all the US states from Jan-2005 to Dec-2014. Then, we focused on the condemnation reasons with a seasonal component in California and used dynamic harmonic regression (DHR) models both to model (from Jan-2005 to Dec-2011) and predict (from Jan-2012 to Dec-2014) the carcass condemnations rate in different time horizons (3 to 12 months).Data consisted of daily reports of 35 condemnation reasons per cattle type reported in 684 federally inspected slaughterhouses in the US from Jan-2005 to Dec-2014 and the monthly slaughtered animals per cattle type per states. Almost 1.5 million carcasses were condemned in the US during the 10 year study period (Jan 2005-Dec 2014), and around 40% were associated with three condemnation reasons: malignant lymphoma, septicemia and pneumonia. In California, emaciation, eosinophilic myositis and malignant lymphoma were the only condemnation reasons presenting seasonality and, therefore, the only ones selected to be modeled using DHRs. The DHR models for Jan-2005 to Dec-2011 were able to correctly model the dynamics of the emaciation, malignant lymphoma and eosinophilic myositis condemnation rates with coefficient of determination (Rt2) of 0.98, 0.87 and 0.78, respectively. The DHR models for Jan-2012 to Dec-2014 were able to predict the rate of condemned carcasses 3 month ahead of time with mean relative prediction error of 33, 11, and 38%, respectively. The systematic analysis of carcass condemnations and slaughter data in a more real-time fashion could be used to identify changes in carcass condemnation trends and more timely support the implementation of prevention and mitigation strategies that reduce the number of carcass condemnations in the US.

Inferring within-herd transmission parameters for African swine fever virus using mortality data from outbreaks in the Russian Federation

Mortality data are routinely collected for many livestock and poultry species, and they are often used for epidemiological purposes, including estimating transmission parameters. In this study, we infer transmission rates for African swine fever virus (ASFV), an important transboundary disease of swine, using mortality data collected from nine pig herds in the Russian Federation with confirmed outbreaks of ASFV. Parameters in a stochastic model for the transmission of ASFV within a herd were estimated using approximate Bayesian computation. Estimates for the basic reproduction number varied amongst herds, ranging from 4.4 to 17.3. This was primarily a consequence of differences in transmission rate (range: 0.7-2.2), but also differences in the mean infectious period (range: 4.5-8.3 days). We also found differences amongst herds in the mean latent period (range: 5.8-9.7 days). Furthermore, our results suggest that ASFV could be circulating in a herd for several weeks before a substantial increase in mortality is observed in a herd, limiting the usefulness of mortality data as a means of early detection of an outbreak. However, our results also show that mortality data are a potential source of data from which to infer transmission parameters, at least for diseases which cause high mortality.

Animal health syndromic surveillance: a systematic literature review of the progress in the last 5 years (2011-2016)

This review presents the current initiatives and potential for development in the field of animal health surveillance (AHSyS), 5 years on from its advent to the front of the veterinary public health scene. A systematic review approach was used to document the ongoing AHSyS initiatives (active systems and those in pilot phase) and recent methodological developments. Clinical data from practitioners and laboratory data remain the main data sources for AHSyS. However, although not currently integrated into prospectively running initiatives, production data, mortality data, abattoir data, and new media sources (such as Internet searches) have been the objective of an increasing number of publications seeking to develop and validate new AHSyS indicators. Some limitations inherent to AHSyS such as reporting sustainability and the lack of classification standards continue to hinder the development of automated syndromic analysis and interpretation. In an era of ubiquitous electronic collection of animal health data, surveillance experts are increasingly interested in running multivariate systems (which concurrently monitor several data streams) as they are inferentially more accurate than univariate systems. Thus, Bayesian methodologies, which are much more apt to discover the interplay among multiple syndromic data sources, are foreseen to play a big part in the future of AHSyS. It has become clear that early detection of outbreaks may not be the principal expected benefit of AHSyS. As more systems will enter an active prospective phase, following the intensive development stage of the last 5 years, the study envisions AHSyS, in particular for livestock, to significantly contribute to future international-, national-, and local-level animal health intelligence, going beyond the detection and monitoring of disease events by contributing solid situation awareness of animal welfare and health at various stages along the food-producing chain, and an understanding of the risk management involving actors in this value chain.

Surveillance of cattle health in the Netherlands: Monitoring trends and developments using routinely collected cattle census data

Since 2002, a national cattle health surveillance system (CHSS) is in place that consists of several surveillance components. The CHSS combines enhanced passive reporting, diagnostic and post-mortem examinations, random surveys for prevalence estimation of endemic diseases and quarterly data analysis. The aim of the data-analysis component, which is called the Trend Analysis Surveillance Component (TASC), is to monitor trends and developments in cattle health using routine census data. The challenges that were faced during the development of TASC and the merits of this surveillance component are discussed, which might be of help to those who want to develop a monitoring and surveillance system that includes data analysis. When TASC was developed, there were process-oriented challenges and analytical related issues that had to be solved. Process-oriented challenges involved data availability, confidentiality, quality, uniformity and economic value of the data. Analytical issues involved data validation, aggregation and modeling. Eventually, the results had to provide information on cattle health that was intuitive to the stakeholders and that could support decision making. Within TASC, both quarterly analysis on census data and, on demand, additional in-depth analysis are performed. The key monitoring indicators that are analyzed as part of TASC all relate to cattle health and involve parameters such as mortality, fertility, udder health and antimicrobial usage. Population-Averaged Generalized Estimating Equations, with the appropriate distribution (i.e. Gaussian, Poisson, Negative Binomial or Binomial) and link function (independent, log or logit), are used for analysis. Both trends in time and associations between cattle health indicators and potential confounders are monitored, discussed and reported to the stakeholders on a quarterly level. The flexibility of the in-depth analyses provides the possibility to conduct additional analyses when anomalies in trends of cattle health occur or when developments in the cattle industry need further investigation. In addition, part of the budget for the in-depth analysis can also be used to improve the models or add new data sources. The TASC provides insight in cattle health parameters, it visualizes trends in time, can be used to support or nuance signals that are detected in one of the other surveillance components and can provide warnings or initiate changes in policy when unfavorable trends occur.