Материал: Bovine Viral Diarrhea Virus Diagnosis, Management, and Control

valence studies are ranked according to prevalence levels there is a tendency to different plateaus. Especially, a large number of studies have revealed prevalence from 60–70% of antibody carriers. At the low scale there are almost as many studies revealing prevalence in the range of 10–20% as there are from 20–60%.

Turning to the prevalence of virus-positive and PI animals it is evident that many studies find prevalence from 0.75 to around 2%. However, another group of studies has revealed prevalence in the range of 0.1–0.2% (Table 2.2). This is also reflected in the prevalence at the herd level where some studies have found 40–50% of the herds to host PI animals (Table 2.2) or being suspicious of hosting such animals (Table 2.3) whereas other studies have found only 10–15% of the herds to host (or being suspicious of hosting) PI animals (Tables 2.2 and 2.3). This could lead one to speculate that areas can be divided into high-and low-prevalence areas.

Looking at the regional differences, many European countries were found to have high prevalence. However, generalizations are not possible because some countries (e.g., Norway, Finland, and Austria) show low prevalence. It is noteworthy that prevalence studies in North America have revealed a very low prevalence of PI animals although the prevalence of antibody carriers in nonvaccinated herds in some studies has been relatively high. The reason for these differences can only be speculated upon. Variation in vaccination programs and management styles may result in differences. It can also be speculated that the mortality among PI animals may vary or that the outcome of fetal infection is different (i.e., the ratio of abortion versus birth of PI animals may vary). Furthermore, it would be reasonable to assume that some of the differences are due to variations in demographic circumstances (see later section, “Risk Factors for Occurrence of BVDV Infections”). In U.S. and Canadian studies it is also characteristic that PI animals can occur in high numbers in some herds.

Follow-up studies in some areas have shown that the infection can stay endemic in an area but with a pronounced change in infection status among the underlying herds (Houe et al., 1997). A study in Estonia showed a large decrease in herds suspected to have PI animals from 1993 to 2000 (Viltrop et al., 2002). The change was believed to be due to decreased trade of breeding animals. Further, it was proposed that the reduction of BVDV in these herds was facilitated by the facts that most large herds were closed and the density of cattle farms was low.

In Sweden, self-clearance has been fostered by controlling introduction of new animals (Lindberg and Alenius, 1999). However, without such specific interventions, BVDV infections will usually maintain themselves at high levels. In the Nordic countries, eradication programs were initiated in the beginning of the 1990s. In these countries the prevalence of herds with PI animals has been significantly reduced and is now approaching zero (Alenius et al., 1997; Bitsch et al., 2000; Valle et al., 2000a; Houe, 2001).

EPIDEMIOLOGICAL STUDIES ON

OCCURRENCE OF DIFFERENT GENOTYPES

The variation of BVDV strains is described in detail in Chapter 3. On most occasions, the variation in BVDV is described at a qualitative level, but quantification of the strains in different regions and their relationship to risk factors is largely unknown. Among 96 field isolates collected in Germany between 1992 and 1996, 11 (11%) were identified as BVDV genotype 2 (Wolfmeyer et al., 1997). Another study of 61 field isolates collected between 1960 and 2000 in Northern Germany identified 2 isolates as genotype 2 (Tajima et al., 2001). It was also indicated that the virus population had been relatively stable over time. Among 28 field isolates from Belgium, only 1 belonged to genotype 1a and the remaining could be divided into genotypes 1b and 2 (Couvreur et al., 2002). Other isolations of BVD genotype 2 in Europe include findings from Slovakia (Vilcek et al., 2002) and United Kingdom where the first definitive genotype 2 isolate was obtained in 2002 (Drew et al., 2002).

In the U.S., several strains gathered over the last many years have belonged to genotypes 1a, 1b, and 2 (Ridpath et al., 2000). Among a sample of 203 BVDV isolates from lots of pooled fetal bovine serum, 51 (25.1%) were identified as genotype 1a, 64 (31.5%) as genotype 1b, 65 (32%) as genotype 2, and 23 (11%) as a mixture of isolates (Bolin and Ridpath, 1998). Of the 105 BVDV isolates from diagnostic laboratory accessions, 61% belonged to genotype 1 and 39% to genotype 2 (Fulton et al., 2000). In South America, analysis of 17 Brazilian isolates identified 13 isolates (76%) as genotype 1 (4 as 1a and 9 as 1b) and 4 isolates (24%) as genotype 2 (Flores et al., 2000). In summary, BVDV genotype 2 seems much more prevalent in North America than in Europe. There are little or no data on prevalence in the African or Asian continents and only very limited data from South America thus far.

INCIDENCE OF BVDV INFECTIONS

The incidence of an event (disease, infection, or any other condition) is a measure of new cases within a given time period. The incidence is usually calculated relative to the population at risk. Two oftenused measures are the incidence risk and the inci-

dence rate. The incidence risk (Irisk) can be defined as follows (Toft et al., 2004):

Number of new cases in the study periood

Irisk = Number of animals at risk at the start of the study period

The incidence risk can be interpreted as the probability that a susceptible animal (i.e., animal at risk) will experience the event of interest in a given time period. Thus the incidence risk can be interpreted at the individual animal level. If the incidence risk has been determined for a time period, the incidence risk for n time periods of the same length, where we assume that the incidence risk is constant throughout all individual periods, can be determined as follows:

Irisk = 1 − (1 − Irisk( p))n

The incidence rate (Irate) on the other hand is a measure of the speed (rate) at which healthy animals

becomes diseased or infected. It is expressed per time unit (animal time at risk) and can mathematically be defined as:

The animal time at risk is the sum of the time at risk for each animal until it either becomes sick, is

withdrawn from the study or the study ends. The Irate is suitable for studies in dynamic populations where

animals enter and leave the study during the study period, because it is a measure per animal time and

not per animal, as is the case for the Irisk. If animals enter and leave the study population uniformly

throughout the study period (i.e., on average they enter and leave in the middle of the study period) the time at risk can be estimated as follows:

Number of animals at risk at start + Number of

Time at risk = animals at risk at end * time period. 2

Under the assumption that the incidence risk or incidence rate is constant during the study period,

the following relationship exists between the two measures:

Irisk = 1 − e−Irate *t

EPIDEMIOLOGICAL STUDIES FOR

ESTIMATING INCIDENCE

There are few studies with direct determination of incidence of infection based on repeat sampling. It is, therefore, relevant to determine the incidence indirectly from prevalence measures (see next section, “Prevalence and Incidence”). The incidence of infection within herds or group of animals will depend on whether only transiently infected animals are present or both PI animals and transiently infected animals are present. The incidence of infection when only transiently infected animals are present varies from case to case. In a dairy herd where the last PI animal was removed in April 1987, 7 of 41 animals born in 1988 and 7 of 54 animals born in 1989 had seroconverted by December 1990 (Barber and Nettleton, 1993). As we do not know the birth dates of calves, the exact incidence rate cannot be calculated. However, assuming an average time at risk of 2 years for each animal, the total time at risk will be 95 animals of 2 years—i.e., a total of 190 animal years at risk. A total of 14 seroconversions in 190 animal years at risk gives an incidence rate of 14/190 = 0.073 cases of seroconversions per animal year corresponding to an annual incidence risk of 0.067 (6.7%) using the shown formula for conversion between incidence risk and incidence rate.

In a long-term epidemiological study in a dairy herd where the virus was spreading slowly over more than 2 years and where there was no direct contact with PI animals, a SIR (susceptible/ infectious/removed) model was used to determine the basic reproduction rate R0 (the number of secondary cases produced by one infectious animal during the whole period in which it is infectious). Among groups of freely mixing cattle with no contact with PI animals, R0 was estimated as 3.3 (CI: 2.6–4.1) (Moerman et al., 1993). There is no consensus on whether the transient infection will be an ongoing process or will be self-limiting and affect only a few animals (1992a). In some experimental studies, susceptible animals in direct contact with transiently infected animals did not get infected (Niskanen et al., 2000; Niskanen et al., 2002).

When PI animals are present in a herd there will be a continuous high infection pressure. Among 67 antibody-negative animals from herds with PI ani-

mals, retesting 6 months later showed that 65 had seroconverted (Houe and Meyling, 1991). This corresponds to an incidence risk of 65/67 = 0.97 for a 6-month period and an annual incidence risk of: 1 (1 0.97)2 = 0.999 i.e., practically a 100% risk of infection. The study by Moerman et al. (1993) also showed a 100% risk of seroconversion among groups of cattle having lived among PI animals. Among 22 cows that have had only a brief accidental contact with a PI animal, only 12 had become positive 4 months later (Moerman et al., 1993). Therefore the incidence risk naturally depends on the length of exposure to PI animals. Among 22 heifers brought in contact with a PI animal, all seroconverted within 5 months, with the fastest seroconversion among heifers with the closest contact to the PI animal (Wentink et al., 1991). It is important to stress that in production systems in which subgroups of the herd are segregated from PI animals, these subgroups of animals can stay uninfected for long periods (Taylor et al., 1997). In addition to variation in the production system, there are considerable differences in the pasturing conditions among countries. In Switzerland, animals are often pastured in the Alps where incidence risk during the summer for replacement animals has been calculated as 45% (Braun et al., 1998). On pastures with PI animals the incidence varied between 33–100% as compared to 6–22% on pastures without PI animals.

Antibody levels in bulk tank milk were taken at 1- year intervals among dairy herds in Sweden. It was found that 5 of 43 herds and 7 of 91 herds with an initial low level showed an increase in antibody level 1 year later, indicating that these herds had experienced a new BVDV infection during the year (Niskanen, 1993; Niskanen et al., 1995). The corresponding annual herd incidence risk can be calculated as 12% and 8%, respectively, for these herds. Among 90 dairy herds in Northern Ireland, with an initial low bulk tank milk antibody level, 12 had a substantial increase after 1 year, corresponding to an annual incidence risk of 13.3% (Graham et al., 2001). If herds with a smaller increase in antibody level were also included, the annual incidence risk could have been as high as 47.7%, but it was not known whether some of these moderate increases in antibody level were due to purchase of antibodypositive animals. In Denmark, repeated serological examination of individual dairy cattle in 9 and 24 herds revealed herd annual incidence risks of 52% and 26%, respectively (Houe and Palfi, 1993; Houe et al., 1997). Based on the annual tested antibody level in bulk tank milk, the average seroconversion

risk in Norwegian herds was estimated as 12% in 1993 (i.e., at the start of the eradication program) declining to 2% in 1997 (Valle et al., 2000a).

The incidence of birth of PI animals will naturally follow the incidence of transient infection in susceptible pregnant cows. Based on the incidence of transient infection and age distribution of cattle at the time of conception in a high-prevalence area, the theoretical incidence risk of fetal infection in the first 3 months of pregnancy was calculated as 3.3% (Houe and Meyling, 1991). Because of fetal death and abortions, the incidence of PI animals born would be somewhat lower. Although the incidence risk may be assumed to be relatively constant in a larger population (i.e., the infection is endemic), there is often high variation over time in the individual herd.

Some possible reasons for this variation are differences among BVDV strains (e.g., differences in viral shed or differences in PI/abortion ratios) and subsequent introduction of other BVDV that are antigenically divergent from the first BVDV. In addition PI animals often occur in clusters. In 22 dairy herds with a total of 129 PI animals, the incidence of PI animals born was closely related to the time of birth of the first PI animal (i.e., oldest identified PI animal) in the herd (Houe, 1992a). Thus 26.3% of the PI animals were born within the first 2 months after the oldest PI animal, whereas no PI animals were born from 2 months until 6 months after the oldest PI animal. Hereafter from 6–14 months after birth of the oldest PI animal, a larger group consisting of 52.7% of all identified PI animals was born (note that these percentages are compared to other PI animals and therefore not incidence risks, but a measure of relative incidence of PI animals born).

This pattern of occurrence of PI animals seems to reflect two periods of transient infection among cows: an initial period with transient infections of short duration (and there are probably no PI animals around) and a second period of transient infections of longer duration caused by continuous presence of PI animals. It is not possible to calculate the levels of incidence risk from these figures because the number of susceptibles at different periods was not known. Taken together, these studies indicate that at the herd level a common scenario concerning incidence could be the following:

•An initial transient infection period of relatively low infection pressure and variable duration (sometimes it stops and sometimes it will continue)

•After the birth of PI animals from 6 months later the incidence increase to high levels staying there as long as PI animals are present.

PREVALENCE AND INCIDENCE

Most epidemiological studies on the occurrence of BVDV infection have focused on determining the prevalence from cross-sectional or repeated crosssectional studies. Studies on incidence are usually more expensive because it is necessary to take samples at more than one time point. However, under certain circumstances, the incidence can be calculated from data on antibody prevalence. The assumption is that the prevalence at a certain age is a measure of the incidence of infection throughout the cow’s life until that age (Houe and Meyling, 1991; Toft et al., 2003).

We, therefore, need a formula to sum up the Irisk over the animal’s life. As calculation of incidence over se-

quential periods is dependent on the previous periods, it is easier to calculate the probability of avoiding infection over several periods and then finally subtract that probability from 1. The probability of avoiding

infection in a given time period is 1 Irisk(i) and the total probability of avoiding infection during n time

periods is simply the product of each time period:

n

Probability of avoiding infection=∏(1 − Irisk(i))

i = l

where ∏ denotes the products of all time periods. Hereafter the total incidence simply becomes the

following:

n

Irisk total = 1 − ∏(1 − Irisk(i)) i = l

As the prevalence at a given age is a measure of total incidence risk and assuming a constant incidence over different time periods e.g., a constant annual in-

cidence risk (Irisk(p)) the formula can be written as follows:

Age specific prevalence = 1 − (1 − Irsk( p) )n

where n is the number of time periods, each of the same length p—i.e., in this case the age in years. This methodology was used by Houe and Meyling (1991) where age specific prevalence of 0.48, 0.65, 0.75, 0.85, 0.95, and 0.96 for 1-year age groups from 1–7 years was entered on the left side of the formula together with age, whereafter the annual incidence risks were calculated as 0.35, 0.34, 0.32, 0.34, 0.42, and 0.39 for each age group, respectively. Thus it can be seen that the incidence was fairly con-

stant at around 0.34 for most age groups. It is an important assumption that the population is in an endemic situation with a constant infection pressure in the population as a whole. Similar increases in agespecific prevalence of antibody carriers by increasing age has also been seen in other studies (Harkness et al., 1978; Braun et al., 1998; Luzzago et al., 1999; McGowan and Murray, 1999).

RISK FACTORS FOR OCCURRENCE OF BVDV INFECTIONS

A risk factor can be defined as a factor that is associated with increased probability of a given event or outcome (infection, disease, or reduced productivity). Sometimes risk factors are used for all factors that have a statistical association with the outcome, but often they are used only for factors where a certain degree of causality can be anticipated. On the other hand, the term etiology is used for the immediate cause of the disease. The combination of etiology and risk factors defines the disease determinants, which are often described according to their association with the host, the infectious agent, or the environment. As the host and agent are described in detail in other chapters, this chapter will focus on the environmental determinants, which in this case are environmental risk factors.

Risk factors can be viewed as either the risk of infection or the risk of disease. For example, the use of common pasture will increase the risk of introducing the infection, whereas high humidity in the calf barn will increase the risk that calves develop pneumonia following infection. This section covers risk factors for the occurrence of infection, and the next section covers the effect of infection on disease and production (i.e., the risk of damages following infection).

When the outcome of risk factor studies is infection, disease, or other dichotomous variables, the prevalence of a given outcome (e.g., infection) in an exposed group can be compared to the prevalence in a nonexposed group by setting up the data in a 2 2 table:

c + d = n

The risk in the exposed group (i.e., prevalence among exposed: a/(a + b)) as compared to the risk in the unexposed group (i.e., prevalence among nonexposed: c/(c + d)) is measured as the relative risk (RR):

= a / (a + b)

RR

c / (c + d)

The RR is a measure of the relative importance or association of the risk factor with infection. If we want to measure the additional importance of the risk factor we simply subtract the prevalence in nonexposed group from the prevalence in the exposed group obtaining the so called attributable risk (AR):

Another important measure is the odds ratio (OR). Rather than establishing the relation between prevalence in exposed and nonexposed group as for RR, the OR establishes the relationship between the odds in the exposed versus nonexposed group:

OR = a / b = a × d c / d b × c

Note that if a and c are small (low prevalence of infection), the OR is a good approximation of the RR. Usually the RR is preferable to OR, but as the RR is not a valid measure in case-control studies it becomes an important measure.

Risk factor studies on BVDV infections have often been performed as observational studies where the main types are cross-sectional, cohort or casecontrol studies. In cross-sectional studies we sample without regard to infection or exposure status, in cohort studies we sample according to risk factor status (only among noninfected), and in case-control studies we sample according to infection status. In case-control studies where we have sampled according to infection, we are not allowed to establish the prevalence measures of infection. Instead, prevalence measures or odds of risk factor relative to infection status can be calculated. Especially the odds for risk factor status are useful as the OR for risk factors ends up being the same formula as for OR for infection status:

Still, one should be more careful interpreting causality from OR obtained in case-control studies as it

measures the ORs of exposure. Thus, if infected animals in a case-control study have an OR of 2.0 of having been on the pasture, this is not the same as to say that pastured animals have an OR of 2.0 of being infected, although we intuitively feel that in practice it will almost be the same. There are several other measures on the importance of risk factors. For more detail, the reader is referred to epidemiological textbooks. Among the risk factor studies on BVDV the OR seems to be the most frequently used measure of effect.

The presence of PI animals in the vicinity of susceptible animals poses the highest proven risk of seroconversion in the population. In a cross-sectional study, the mean antibody prevalence among cattle from herds with PI animals was 87% as compared to 43% in herds without PI animals (Houe and Meyling, 1991). The RR of being infected according to the risk factor of coming from a herd with PI animals was not stated but it can be obtained simply as the relationship between the two prevalences (87/43 = 2.0) or to illustrate the use of the RR formula directly from the distribution of antibody carriers in the two herd categories:

Antibody Status of

Individual Animals

Hereafter the RR is calculated as follows:

= a / (a + b) = 1083 / 1249 = RR 2.0

c / (c + d) 572 / 1321

This may be an underestimation of the effect of PI animals because many of the animals in herds without PI animals at the time of cross-sectional study may have been exposed to PI animals previously.

The following description of other risk factors will often be a reflection of the increased risk of direct or indirect contact with PI animals and to some extent with acutely infected animals or other sources. Some may claim that the presence of PI animals should be viewed as the direct etiology and not a risk factor, and often the distinction between