The risk of the introduction of classical swine fever virus at regional level in the European Union: a conceptual framework.

Recent classical swine fever (CSF) epidemics in the European Union (EU) have clearly shown that preventing the introduction of CSF virus (CSFV) deserves high priority. Insight into all the factors contributing to the risk of CSFV introduction is a prerequisite for deciding which preventive actions are cost-effective. The relations between virus introduction and spread, prevention and control, and economic losses have been described using the conceptual framework presented in this paper. A pathway diagram provides insight into all the pathways contributing to the likelihood of CSFV introduction (LVI_CSF) into regions of the EU. A qualitative assessment based on this pathway diagram shows that regions with high pig densities generally have a higher LVI_CSF, although this cannot be attributed to pig density only. The pathway diagram was also used to qualitatively assess the reduction in LVI_CSF achieved by restructuring the pig production sector. Especially integrated chains of industrialised pig farming reduce the LVI_CSF considerably, but are also difficult and costly to implement. Quantitative assessment of the LVI_CSF on the basis of the pathway diagram is needed to support the results of the qualitative assessments described.

Probability of and risk factors for introduction of infectious diseases into Dutch SPF dairy farms: a cohort study.

A 2-year cohort study was conducted to investigate the probability of disease introduction into Dutch dairy farms. The farms were tested regularly for diseases and were visited biannually to collect management data. Ninety-five specific pathogen-free (SPF) dairy farms were selected from a database of bovine herpesvirus type 1 (BHV1)-free farms to study the probability of, and risk factors for, introduction of BHV1, bovine viral diarrhoea virus (BVDV), Salmonella enterica subsp. enterica serotype Dublin (S. dublin), and Leptospira interrogans serovar hardjo (L. hardjo). Although most of the 95 SPF farms had a low risk on introduction of infectious diseases, one disease was introduced into 12 farms and two diseases were introduced into one farm. Three farms experienced an outbreak of BHV1, one farm an outbreak of L. hardjo, two farms BVDV, six farms S. dublin, and one farm both BHV1 and S. dublin. The total incidence rate was 0.09 (0.06-0.12) per herd-year at risk. The results suggest that the "non-outbreak" farms were significantly more closed than the "outbreak" farms. Direct animal contacts with other cattle should be avoided and professional visitors should be instructed to wear protective clothing before handling cattle.

An economic model for on-farm decision support of management to prevent infectious disease introduction into dairy farms.

A more-closed farming system can be a good starting point for eradication of infectious diseases from within a herd. The economic implications of a more-closed farming system will not always be obvious to farmers. The management decisions are related to different parts of the farm and are farm-specific. To support these decisions, a model was developed of the economic consequences of a more-closed system (a simple static and deterministic design was used). The risk factors in the model were based solely on bovine herpesvirus type 1 (BHV1) but losses due to introduction of BVDV, L. hardjo, and S. dublin were added to the model. The model was verified and partly validated and a sensitivity analysis was done. The cost to one 55-cow dairy farm that refrained from purchasing cattle, provided protective clothing to professional visitors and a temporary employee, and built and maintained a double fence around 6 ha of land to prevent over-the-fence contacts was Dfl. 4495 over 5 years. The probability of disease introduction was decreased by 74%. The prevented losses for disease introduction amounted to Dfl. 7033 over 5 years (net benefits of Dfl. 2538 over 5 years).A more-closed system would be still beneficial when a sanitary barrier was used instead of just protective clothing, when the probability of introduction of infectious diseases was decreased, and when odds ratios in the model were replaced by more-conservative relative risks. The benefits became negative when a farm had to build and maintain a double fence around 12 ha instead of 6 ha, when the probability of introduction of all diseases was decreased by 50%, and when the estimations were based solely on BHV1.

Risk factors for introduction of BHV1 into BHV1-free Dutch dairy farms: a case-control study.

In May 1998, a compulsory eradication programme for BHV1 started in the Netherlands. In December 1999 approximately 24% of Dutch dairy farms were certified BHV1-free (Animal Health Service (AHS)). Ninety-three certified BHV1-free dairy farms participated in a cohort study that investigated the probability of introduction of infectious diseases. The probability of introduction of BHV1 was determined from March 1997 until April 1999. Ninety of these farms remained BHV1-free and could be used as control farms. From January 1997 until March 1998, BHV1 was introduced into 41 BHV1-free dairy farms in the Netherlands (case farms). Management data were collected for both cases and controls and were complete for 37 case farms and 82 control farms. For small data sets and for data in which both low and high frequencies were expected in the contingency tables, the asymptotic methods were unreliable. Our data set clearly resembled such a data set; the risk factors were rare events because the BHV1-free farms were closed farms on which few direct animal contacts occurred. Therefore, an exact stratified modelling approach was most suitable for the data. The study showed that dairy farms should prevent cattle from escaping or mingling with other cattle and that professional visitors should always wear protective farm clothing.

Spatial and stochastic simulation to compare two emergency-vaccination strategies with a marker vaccine in the 1997/1998 Dutch Classical Swine Fever epidemic.

Two alternative emergency-vaccination strategies with a marker vaccine that could have been applied in the 1997/1998 Dutch Classical Swine Fever (CSF) epidemic were evaluated in a modified spatial, temporal and stochastic simulation model: InterCSF. In strategy 1, vaccination would be applied only to overcome a shortage in destruction capacities. Destruction of all pigs on vaccinated farms distinguishes this strategy from strategy 2, which assumes intra-Community trade of vaccinated pig meat. InterCSF simulates the spread of CSF between farms through local spread and three contact types. Disease spread is affected by control measures implemented through different mechanisms. Economic results were generated by a separate model that calculated the direct costs (including the vaccination costs) and consequential losses for farmers and related industries subjected to control measures. The comparison (using epidemiological and economic results) between the different emergency-vaccination strategies with an earlier simulated preventive-slaughter scenario led to some general conclusions on the Dutch CSF epidemic. Both emergency-vaccination strategies were hardly more efficient than the non-vaccination scenario. The intra-Community trade strategy (vaccination-strategy 2) was the least costly of all three scenarios.

A concept for a decision support system based on practical experiences from a national disease emergency. The Dutch experience.

In 1997-1998, the Netherlands experienced a large epidemic of classical swine fever (CSF). The magnitude of this epidemic stressed the role information systems could play in supporting the management during an eradication campaign. The enforcement of an eradication strategy can become very complicated, especially with large epidemics, due to time pressure and many different procedures that have to be executed at the same time. The application of comprehensive information systems may result in more control over the process and in a relief of the operational management. After a brief description of the Dutch epidemic the authors provide an overview and the general application of four different types of information systems, classified as decision support systems. The application of these information systems in animal disease control is illustrated by providing concepts for a system architecture for transaction processing, management and executive information support and decision support. The application of a data warehouse as part of this systems architecture is explained. The eradication of CSF from the Netherlands was complicated by several factors. It is important to notice that information systems cannot prevent these factors. However, information systems can support disease control authorities in controlling these factors.

Optimization of dairy heifer management decisions based on production conditions of Pennsylvania.

We used a dynamic programming model to determine optimum rearing decisions of dairy replacements. Heifers were described in the model by age, season, body weight, pregnancy state, and prepubertal growth rate. Prices and parameters were chosen to represent the dairy population of Pennsylvania. We calculated monthly costs and revenues from calf value, feed costs, veterinary costs, semen costs, carcass value, and full-grown heifer value. The model considered a stochastic variation in the onset of puberty, conception, involuntary disposal, and a seasonal variation in the prices of calves, heifers, and feed. Based on a critical prepubertal average daily gain of 0.9 kg/d and a maximum achievable postpubertal growth rate of 1.1 kg/d, the optimum practice resulted in an average age at first calving of 20.5 mo at a body weight of 563 kg. Discounted net returns equaled $107 per heifer per year. The optimum rearing practice was not sensitive to seasonal variation in prices. Nevertheless, the economic results per season of birth varied considerably; the highest income per heifer was obtained from heifers born in December ($142/yr), whereas those born in May yielded the lowest ($100/yr). Sensitivity analyses demonstrated a considerable influence of growth rate restrictions and variation in reproductive performance on both the optimal rearing practices as the expected net returns.

Evaluating control strategies for outbreaks in BHV1-free areas using stochastic and spatial simulation.

Several countries within the EU have successfully eradicated bovine herpesvirus type I (BHV1), while others are still making efforts to eradicate the virus. Reintroduction of the virus into BHV1-free areas can lead to major outbreaks - thereby causing severe economic losses. To give decision-makers more insight into the risk and economic consequences of BHV1 reintroduction and into the effectiveness of various control strategies, we developed the simulation model InterIBR. InterIBR is a dynamic model that takes into account risk and uncertainty and the geographic location of individual farms. Simulation of a BHV1-outbreak in the Netherlands starts with introduction of the virus on a predefined farm type, after which both within-farm and between-farm transmission are simulated. Monitoring and control measures are implemented to simulate detection of the infection and subsequent control. Economic consequences included in this study are related to losses due to infection and costs of control. In the simulated basic control strategy, dairy farms are monitored by monthly bulk-milk tests and miscellaneous farms are monitored by half-yearly serological tests. After detection, movement-control measures apply, animal contacts are traced and neighbour farms are put on surveillance. Given current assumptions on transmission dynamics, we conclude that a strategy with either rapid removal or vaccination of infected cattle does not reduce the number of infected farms compared to this basic strategy - but will cost more to control. Farm type with first introduction of BHV1 has a considerable impact on the expected number of secondarily infected farms and total costs. To limit the number of infected farms and total costs due to outbreaks, we suggest intensifying the monitoring program on farms with a high frequency of cattle trade, and monthly bulk-milk testing on dairy farms.

The effectiveness of routine serological surveillance: case study of the 1997 epidemic of classical swine fever in The Netherlands.

The authors describe the value of routine serological surveillance in detecting the introduction of classical swine fever virus into a disease-free population. The first investigation concerned the question of whether the epidemic of classical swine fever (CSF), which occurred from 1997 to 1998 in the Netherlands, could have been detected using the existing monitoring system for notifiable diseases. The investigation used data from the CSF epidemic of 1997/1998 and from the existing monitoring system. Secondly, the probability of detecting a case of CSF using routine serological surveillance was modelled both for multiplier herds and for finishing herds, and then for different herd size categories. The first investigation concluded that the probability of detecting the epidemic at the current level of routine serological surveillance is very low. The second investigation concluded that even employing a sampling scheme of sixty blood samples per month, the probability of detecting an outbreak of CSF within forty days of the introduction of the virus, is less than 40%.

Monte Carlo simulation of virus introduction into the Netherlands.

In order to improve the understanding of the risk of introducing classical swine fever (CSF) and foot-and-mouth disease (FMD) into the Netherlands, a Monte Carlo simulation model was developed. The model, VIRiS (Virus Introduction Risk Simulation model) describes virus introduction into the Netherlands from outbreaks in other European countries. VIRiS is aimed at supporting decision makers involved in disease prevention. The model is based on historical and experimental data, supplemented with expert judgement, and provides the expected number, location and cause of primary outbreaks in the Netherlands. The paper gives a detailed description of the design and behaviour of VIRiS. The default outcomes of VIRiS show that in the current situation, the western and northern regions of the Netherlands are most prone to outbreaks of CSF and FMD. Most outbreaks originate from the countries neighbouring the Netherlands and the countries of southern Europe. Several alternative prevention strategies were evaluated using a combination of the VIRiS model and models describing the spread and economic consequences of outbreaks. A considerable financial window is available for measures aimed at speeding up the detection of epidemics in countries from which a Dutch outbreak may originate. Complete elimination of the risk associated with the risk factor 'returning trucks' reduces the annual losses due to FMD and CSF epidemics by approximately US$ 9 million. The approach is general and could also be applied to other diseases and countries.

Modeling the effect of an outbreak of bovine herpesvirus type 1 on herd-level milk production of Dutch dairy farms.

One of the impacts of disease is its effect on milk production. In the present study the effect of an outbreak of bovine herpesvirus type 1 on milk production at the herd level of certified bovine herpesvirus type 1-free dairy farms was modeled. The objective was to study several linear models to quantify the effects of a bovine herpesvirus type 1 outbreak on milk production accounting for the repeated measurements and incorporating our assumptions about the most likely duration of effects of this virus. Because milk production is measured at regular intervals, the data consisted of repeated measurements at the herd and cow levels. A marginal model, a subject-specific random-effect model, and a transition model were developed. The effect of a bovine herpesvirus type 1 outbreak was statistically significant in the random-effect model, and this model fitted the investigated farms best. However, a transition model might be a better model for generalizing the results to the whole population of Dutch dairy farms. The effect of a bovine herpesvirus type 1 outbreak on milk production derived from the random-effect model amounted to, on average, a loss of 0.92 kg of milk per cow per day during a period of 9 wk. The milk production loss varied from almost none to 2 kg of milk per cow per day. This reduction resulted in an average loss of Dfl 372 (Dfl1 = $US $0.50) with lower and upper confidence limits of, respectively, Dfl 12 and Dfl 730 per bovine herpesvirus type 1 outbreak.

Relationships between dairy cow mastitis and fertility management and farm performance.

A field study was carried out on 38 dairy farms in the Netherlands to determine the relationship between mastitis and fertility management with 305-day milk production and gross margin. Questionnaires were used to get insight into the farmers' management. Out of 150 variables related to mastitis and fertility management, and technical and economic results, 44 variables were selected based on correlation of > or = 0.25 or < or = -0.25 with milk production and/or gross margin. These variables were used in two separate partial least squares (PLS) analyses. PLS has the advantage that it can handle a large number of variables in relation to the number of cases. The PLS-model of 305-day milk production had R2 = 0.54 and showed a positive relation between 305-day milk production and awareness of the farmer regarding bulk somatic-cell count (BSCC), the goal level of the farmer for BSCC, and hygiene of the milking parlour. Fertility was negatively related to 305-day milk production, in spite of a relatively good fertility management on high-producing farms. R2 = 0.46 for gross margin. The aspiration level of fertility did not seem to affect gross margin, but awareness of BSCC and calving interval (CI) had positive effects. Such awareness seemed a general parameter for good economic results, because it was correlated with different aspects of gross margin.

Repeatability in ranking of dairy farms on technical and economic performance over years.

Repeatability of farm average 305-day milk production and gross margin per 100 kg of milk was evaluated for 39 farms. Ranking of gross margin, its underlying factors (i.e. milk price, returns from cull cows and calves, costs of concentrates, and costs of roughage purchases per 100 kg of milk), and 305-day milk production was not completely random over the four years of the study. The coefficient of concordance ranged between 0.55 and 0.82. The costs of roughage purchased had the lowest concordance over time, and 305-day milk production had the highest concordance. For each year and each farm, the difference between average gross margin and farm-specific gross margin was calculated. The standard deviations (SD) of these values was calculated for each farm, and showed differences between farms in variability in gross margin over years (the farm-year-specific SD varied between farms from 0.56 to 5.73). All the underlying factors showed a deviation over years. So, variability of gross margin can be due to changes in all underlying factors. The impact on gross margin of purchased roughage was not of major importance because its absolute impact on the gross margin is small. We concluded that milk-production data over one year is a reliable indicator for the typical farm milk production. Because gross margin fluctuates considerably over time, however it is preferred to base economic research on data from more than one year.

A model to estimate the financial consequences of classical swine fever outbreaks: principles and outcomes.

A model is presented aimed at a financial analysis of a Classical Swine Fever outbreak. Financial consequences are calculated for affected parties, including governments (EU and national), farms, and related industries in the production chain. The model can be used to calculate the losses of a real outbreak as well as of a simulated one. In this article, the model is applied to the 1997/1998 outbreak of Classical Swine Fever in the Netherlands. Results show that total financial consequences of the outbreak are US $2.3 billion. Consequential losses for farmers and related industries are US $423 million and US $596 million respectively. Budgetary consequences for governments include less than 50% of the total losses calculated by the model. The model can be adapted easily to suit other diseases and countries.

Spatial and stochastic simulation to evaluate the impact of events and control measures on the 1997-1998 classical swine fever epidemic in The Netherlands. I. Description of simulation model.

The simulation model InterCSF was developed to simulate the Dutch Classical Swine Fever (CSF) epidemic of 1997-98 as closely as possible. InterCSF is a spatial, temporal and stochastic simulation model. The outcomes of the various replications give an estimate of the variation in size and duration of possible CSF-epidemics. InterCSF simulates disease spread from an infected farm to other farms through three contact types (animals, vehicles, persons) and through local spread up to a specified distance. The main disease-control mechanisms that influence the disease spread in InterCSF are diagnosis of the infected farms, depopulation of infected farms, movement-control areas, tracing, and pre-emptive slaughter. InterCSF was developed using InterSpread as the basis. InterSpread was developed for foot-and-mouth disease (FMD). This paper describes the process of modifying InterSpread into InterCSF. This involved changing the assumptions and mechanisms for disease spread from FMD to CSF. In addition, CSF-specific control measures based on the standard European Union (EU) regulations were included, as well as additional control measures that were applied during the Dutch epidemic. To adapt InterCSF as closely as possible to the Dutch 1997/98 epidemic, data from the real epidemic were analysed. Both disease spread and disease-control parameters were thus specifically based on the real epidemic. In general, InterSpread turned out to be a flexible tool that could be adapted to simulate another disease with relative ease. The most difficult were the modifications necessary to mimic the real epidemic as closely as possible. The model was well able to simulate an epidemic with a similar pattern over time for number of detected farms as the real outbreak; but the absolute numbers were (despite many relevant modifications) not exactly the same--but were within an acceptable range. Furthermore, the development of InterCSF provided the researchers with a better insight into the existing knowledge gaps. In part II (see the final paper in this issue), InterCSF was used to compare various control strategies as applied to this epidemic.

Spatial and stochastic simulation to evaluate the impact of events and control measures on the 1997-1998 classical swine fever epidemic in The Netherlands. II. Comparison of control strategies.

Using the spatial, temporal and stochastic simulation model InterCSF, several alternative pre-emptive slaughter strategies that could have been applied in the Dutch Classical Swine Fever (CSF) epidemic of 1997-1998 were evaluated. Furthermore, effects of changes in some disease-spread and disease-control parameters were studied. InterCSF simulates the spread of CSF between farms through local spread and contacts (animals, transport and persons). Disease spread is affected by control measures implemented through different mechanisms (e.g. depopulation of infected farms, pre-emptive slaughter, movement control). The starting point for the evaluation of strategies was a simulated basic scenario, which mimicked the real epidemic. Strategies were compared using epidemiological as well as economic results. Economic results were generated by a separate model (EpiLoss) that calculated the direct losses and consequential losses for farmers and related industries. The comparison of the different alternatives to the basic scenario led to some general conclusions on the Dutch CSF-epidemic. Pre-emptive slaughter seemed to be an effective strategy to reduce the size of an epidemic, if started at an early stage. Economically, pre-emptive slaughter was not as expensive as expected; the resulting smaller size of the epidemic, combined with less welfare slaughter, led to much lower overall losses. Furthermore, although large movement control areas seemed effective in reducing the size of the epidemic, the total losses were relatively high because of subsequent welfare slaughter. If infection probabilities could be reduced, for example by improved biosecurity, the resulting epidemics would be much smaller.

Use of AI technician scores for body condition, uterine tone and uterine discharge in a model with disease and milk production parameters to predict pregnancy risk at first AI in Holstein dairy cows.

Technicians recorded body condition score (BCS) and several parameters related to estrus and/or metritis for 1694 first insemination cows on 23 farms. Additional variables for modeling the adjusted odds ratios (OR) for pregnancy were data on disease prior to or within 21 days of AI and test day milk yields. Significant predictors for pregnancy were farm, year and season, BCS, uterine tone, contaminated insemination gun after AI, fat-protein corrected kilograms milk (FPCM), days in milk (DIM), and diseases. Vaginal mucus, ease of cervical passage, and lameness were not significant predictors for pregnancy. Pregnancy risk at AI increased with increasing DIM, reaching a near optimum after 82 days. Lack of uterine tone was associated with a lowered pregnancy risk (OR = 0.69) as was contaminated insemination gun (OR = 0.67), first-parity lactation, FPCM >33 kg (OR = 0.71), BCS 2.5 at AI (OR = 0.65), clinical mastitis (OR = 0.53), cystic ovarian disease (OR = 0.53), and metritis (OR = 0.74). It was concluded that data on BCS and uterine findings, as collected by AI technicians, are significant predictors of AI outcome. Dairy producers and veterinarians should jointly examine the potential costs and value of such AI technician-based data to improve herd fertility.

Quantifying characteristics of information-technology applications based on expert knowledge for detection of oestrus and mastitis in dairy cows.

Expert opinions were elicited about the characteristics at the commercial-farm level of on-line information technology (IT) applications that are able to detect oestrus and mastitis in dairy cows. Since actual data of these characteristics are not available, judgmental data provided an alternative means to interpret the implications of research results for commercial farms. Applications included were activity measurement, milk-production measurement, electrical conductivity of quarter milk, automated concentrate feeders and milk-temperature measurement. Sensitivity and specificity of detection of oestrus (OD), clinical-mastitis (CMD) and subclinical-mastitis (SCMD) were ascertained. Conjoint-analysis was used to assess the effect of each application indirectly by decomposing the evaluated overall detection characteristics of a predefined number of IT combinations. The individual experts were consistent in evaluating the alternatives, but there was variation in estimates among experts. Estimations of the main effects of the applications and important first-order interactions were incorporated into the detection models. Implementation of all applications under study resulted in overall sensitivities and specificities of 82% and 90%, 73% and 87%, 58% and 82% for OD, CMD and SCMD, respectively. Further research is necessary that should take into account costs and benefits of the different detection systems based on the current status of farm performance (e.g. OD and mastitis incidence) and farm structure (e.g. farm size, years in operation of the milking parlour and parlour layout). Research to do this is currently in progress.

A descriptive study of visits by animal health specialists in pig farming: type, frequency, and herd-health management factors.

This research was carried out to analyse the visits specialists of the Dutch Animal Health Service made to growing and fattening pig farms. The type and frequency of the visits and identified herd-health management factors that did not meet accepted standards were investigated. In total 373 visit reports were studied. The majority of the visits (n = 306 of 373) were made to investigate the cause of health, welfare, and performance problems ('problem-solving visits'). Respiratory disorders were the main reason for requesting a specialist to assess farm conditions and management (n = 156). In the other 67 of 373 visit reports the specialists screened for herd-health management factors that did not meet standards for the prevention of disease ('screening visits'). For both types of visits, the main factors detected were abrupt changes in feeding regimens (e.g. changes in feed type, feed composition or feed supplier) (37%), inadequate measures to prevent introduction of pathogens by people and trucks (83%), and incorrect adjustment of the ventilation system (58-60%). The specialists focusing on housing-climate management, identified the majority of factors in an equal number irrespective of whether the visit was a problem-solving visit or a screening visit. This implies that even on farms that appear not to have health or performance problems, factors that relate to disease are present and may cause problems sooner or later. Although veterinary practitioners and other farm advisors assist farmers in their management to optimize herd health, the findings of the research suggest that advisors could provide additional support in situations where environmental and managerial factors play a role in pig health and performance. The knowledge of advisors about integrated herd-health management can be broadened by means of textbooks, courses, or computer programs.