Quantification of vertical and horizontal transmission of Neospora caninum infection in Dutch dairy herds

Veterinary Parasitology 148 (2007) 83–92 www.elsevier.com/locate/vetpar

Quantification of vertical and horizontal transmission of Neospora caninum infection in Dutch dairy herds Chris J.M. Bartels a,*, Irene Huinink a, Marten L. Beiboer b, Gerdien van Schaik a, Willem Wouda a, Thomas Dijkstra a, Arjan Stegeman c a Animal Health Service Ltd., P.O. Box 9, 7400 AA, Deventer, The Netherlands Veterinary Software Design, Koekoeksbloem 1, 98801 LW Zuidhorn, The Netherlands c Department of Farm Animal Health, Faculty of Veterinary Medicine, Utrecht University, Yalelaan 7, 3584 CL Utrecht, The Netherlands b

Received 29 January 2007; received in revised form 20 May 2007; accepted 4 June 2007

Abstract Ninety-six of 108 randomly selected Dutch dairy herds had one or more cows with a positive serostatus for N. caninum. In these 96 herds, we have quantified the probabilities of vertical transmission (VT) and horizontal transmission (HT) of N. caninum infection by combining serostatus and pedigree data in 4091 dam-daughter pairs. The probability of animals infected by vertical transmission during pregnancy (Prob(VT)) was calculated as the proportion of seropositive daughters among daughters of seropositive dams. The probability of animals infected by horizontal transmission (Prob(HT)) was the proportion of seropositive daughters among daughters of seronegative dams. These probabilities were calculated after the frequencies of observed damdaughter combinations were corrected for (1) imperfect test-characteristics, (2) underestimation of horizontal transmission in situations that seronegative dams were horizontally infected after the birth of their daughters and (3) overestimation of vertical transmission in situations that seronegative daughters born from seropositive dams were horizontally infected. The incidence rate for horizontal transmission was calculated based on Prob(HT) and the average age of the animals in these herds. Based on the analysis of dam-daughter serology, Prob(VT) was 61.8% (95% CI: 57.5–66.0%) and Prob(HT) was 3.3% (95% CI: 2.7–3.9%). After adjusting the observed frequencies for imperfect test-characteristics, underestimation of horizontal transmission and overestimation of vertical transmission, Prob(VT) decreased to 44.9% (95% CI: 40.0–49.9%) while Prob(HT) increased to 4.5% (95% CI: 3.9–5.2%). Prob(HT) corresponded with an incidence rate for horizontal transmission of 1.4 (95% CI: 1.2–1.7) infections per 100 cow-years at risk. When stratifying herds for the presence of farm dogs, Prob(HT) was higher (5.5% (95% CI: 4.6–6.4%)) in herds with farm dogs than in herds without farm dogs (2.3% (95% CI: 1.5–3.4%)). When stratifying for within-herd seroprevalence, Prob(HT) was higher (10.3% (95% CI: 8.6–12.2%)) in herds with high (10%) within-herd seroprevalence compared with herds with low (<10%) within-herd seroprevalence (2.0% (95% CI: 1.5–2.6%)). Although there was this relation between Prob(HT) and within-herd seroprevalence (crude ORPREV = 5.7 (95% CI: 4.0–7.9)), in herds without farm dogs, this relationship was no longer statistical significant (ORPREVjDOG- = 1.9 (95% CI: 0.7–5.5)). It indicated that the association between seroprevalence and Prob(HT) depended largely on the presence of farm dogs. In addition, when looking for the presence of specific age-groups with significantly higher seroprevalence compared with the rest of the herd, there were 7 herds in which two or more horizontally-infected animals were present in specific age-groups. This was an indication of a recent point-source exposure to N. caninum.

* Corresponding author. Tel.: +31 570660380. E-mail address: [email protected] (C.J.M. Bartels). 0304-4017/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.vetpar.2007.06.004

84

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

These results reiterate the current control strategies to apply strict dog-management measures as well as to minimize within-herd seroprevalence by monitoring serostatus of animals. # 2007 Elsevier B.V. All rights reserved. Keywords: Cattle; Epidemiology; Neospora caninum; Horizontal and vertical transmission of infection

1. Introduction One of the leading causes of bovine abortion is the protozoan parasite Neospora caninum. N. caninum has a heteroxenous life cycle in which cattle are important intermediate hosts and dogs and coyotes are the only recognized definitive hosts hosts (Gondim et al., 2004; McAllister et al., 1998). The parasite has a worldwide prevalence and may cause abortion both after a primary infection and as a result of recrudescence of a persistent infection (Dubey et al., 2006). Apart from abortions, economic losses due to infection with N. caninum are primarily caused by premature culling (Bartels et al., 2006b; Tiwari et al., 2005; Thurmond and Hietala, 1996) and possibly decreased milk production (Bartels et al., 2006b; Romero et al., 2005; Hobson et al., 2002; Hernandez et al., 2001; Thurmond and Hietala, 1997). Current strategies to control neosporosis focus on a reduction of the seroprevalence in cattle and on separating dogs and dog-faeces from cattle to avoid new infections in cattle (Dijkstra et al., 2005; Fro¨ssling et al., 2005). Both vertical and horizontal transmission routes play a role in the infection of cattle. Vertical transmission is responsible for the spread of infection from a persistently-infected dam to her offspring during pregnancy and contributes significantly to the persistence of N. caninum infection in a herd by propagating the infection to successive generations (Bjo¨rkman et al., 1996; Anderson et al., 1997; Schares et al., 1998; Wouda et al., 1998). Reported vertical transmission probabilities range from 41% (Pan et al., 2004) to 95% (Davison et al., 1999b). Despite the efficiency of vertical transmission, it is evident that infection with N. caninum cannot be sustained in cattle herds without horizontal transmission and this was modelled by French et al. (1999). Horizontal transmission occurs when cattle ingest sporulated N. caninum oocysts. There is convincing evidence that horizontal transmission can be associated with N. caninum abortion outbreaks, suggesting a point source exposure (McAllister et al., 1996; Thurmond et al., 1997; Mainar-Jaime et al., 1999; Waldner et al., 1999; Dyer et al., 2000; Dijkstra et al., 2001). A few studies found evidence for ongoing horizontal transmission of N.

caninum in cattle herds following a point-source infection (Bjo¨rkman et al., 2003; Dijkstra et al., 2002b). In several other studies, there was a low incidence of seroconversion in endemically-infected herds suggesting a low level of horizontal transmission (Davison et al., 1999b; Fro¨ssling et al., 2005; Hietala and Thurmond, 1999; Schares et al., 1998; Wouda and Brinkhof, 1998). These above-mentioned studies were all based on herds with a history of clinical neosporosis. To our knowledge, no studies have measured vertical and horizontal transmission probabilities based on a random sample of herds. Within-herd seroprevalence and presence of farm dogs are putative risk factors for N. caninum-associated abortions (Bartels et al., 1999; Pare´ et al., 1998; Schares et al., 2004; Wouda et al., 1999) and thus N. caninum infection. As definitive hosts for N. caninum, dogs are known to spread oocysts leading to horizontal transmission in cattle. The relation between withinherd seroprevalence and horizontal transmission is less clear. Biologically, it can be hypothesized that increased within-herd seroprevalence might lead to increased horizontal transmission if cow to cow transmission would be possible. For this reason, we were interested to know if seroprevalence in itself is related to the probability of horizontal transmission. The objective of this study was to quantify the probabilities of vertical and horizontal transmission of N. caninum infection in Dutch dairy herds in general. Additionally, we compared the probability of horizontal transmission between herds with and without farm dogs, and between herds with high versus low withinherd seroprevalence. This was done by combining serological data with pedigree data of seropositive animals from an earlier seroprevalence study. 2. Materials and methods 2.1. Selection of herds As part of a prevalence study (Bartels et al., 2006a), 108 dairy herds were randomly selected from the Dutch dairy-herd population and blood was collected from all female cattle above 3 months of age (11,672 animals). In 96 herds (10,350 animals), 1 or more animals tested

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92 Table 1 Contingency table used for calculating the probability of horizontal and vertical transmission of N. caninum infection Status of daughters

Status of dam Seropositive

Seronegative

Seropositive

a

b

Seronegative

c a+c

d b+d

seropositive and these herds were included in the present study. 2.2. Serological testing of animals Serum samples were tested using the Animal Health Service (AHS) in-house serum ELISA (Wouda and Brinkhof, 1998). An S/P-ratio of >0.5 was defined as positive. The diagnostic sensitivity was 96.9% (95% CI: 94.5–99.4%) and the diagnostic specificity was 97.3% (95% CI: 95.5–99.0%) (Von Blumro¨der et al., 2004). 2.3. Calculation of vertical and horizontal transmission probabilities Pedigree information of cattle was obtained from the Dutch Identification and Registration (I&R, Royal Dutch Dairy Syndicate, Arnhem, The Netherlands). A software program Neospora# (Beiboer, Veterinary Software design, Zuidhorn, The Netherlands, 2002) was used to facilitate tracing dam-daughter relations and linking serological-test results. Data on dam-daughter relations were compiled in 2  2 tables (Table 1). The fraction of animals infected by vertical transmission during pregnancy was calculated as the proportion of seropositive daughters among the daughters of seropositive dams (a/a +c). The

85

fraction of animals infected by horizontal transmission was the proportion of seropositive daughters among the daughters of seronegative dams (b/b + d). As this was a cross-sectional study, the observed frequencies of dam-daughters combinations were adjusted for imperfect test characteristics, for possible underestimation of horizontal transmission (i.e. some positive-tested dams may have been infected after they gave birth to a daughter) and overestimation of vertical transmission (i.e. some positive-tested daughters from positive dams may have been infected after being born as negative daughter). 2.3.1. Adjustment for imperfect test characteristics It is likely that due to imperfect test-characteristics a proportion of animals were wrongly classified as seropositive or seronegative, and as a consequence the routes of transmission were assigned incorrectly. The extent to which this occurred was calculated based on the prevalence of infection in the study population and the point-estimates of sensitivity and specificity of the AHSinhouse test. The predictive value for a positive test (PV+) result was 78.4% (95% CI: 75.9–80.7) and for a negative test result (PV) 99.7% (95% CI: 99.5–99.8). Assignment of vertical-infection status (Dam + daughter+, cell a in Table 1) is the result of 4 different probabilities. The fraction of animals correctly assigned a vertical-transmission status was calculated as the product of the probability of a truly-seropositive daughter (PV+ = 0.784) times a truly-seropositive dam (PV + = 0.784) which amounts to 0.615. The fraction of animals incorrectly assigned to the combination Dam + daughter + (1  0.615a) was divided over three other possible dam-daughter combinations (0.169 for Dam + daughter, 0.169 for Dam  daughter+ and 0.047 for Dam  daughter) according to the algorithms given in Table 2.

Table 2 Algorithms used to convert observed frequencies of dam-daughter combinations into frequencies corrected for imperfect test characteristics. This was done using the positive (PV+) and negative predictive (PV) terms Observed

Corrected Dam + daughter+

Dam + daughter

Dam  daughter+

Dam  daughter

Dam + daughter+ (cell a )

a  PV+  PV+

a  PV+  (1  PV+)

a  (1  PV+)  PV+

Dam + daughter (cell b)

b  PV+  (1  PV)

b  PV+  PV

Dam  daughter+ (cell c)

c  (1  PV)  PV+

Dam  daughter (cell d)

d  (1  PV)  (1  PV)

c  (1  PV)  (1  PV+) d  (1  PV+)  PV

b  (1  PV)  (1  PV+) c  PV  PV+

a  (1  PV+)  (1  PV+) b  (1  PV)  PV

a

a

References to cell a–d relate to Table 1.

d  PV  (1  PV)

c  PV  (1  PV+) d  PV  PV

86

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

Table 3 Conversion from observed to corrected frequencies of 4091 dam-daughter combinations. This was based on the algorithms in Table 2 and the pointestimates for predictive values for positive (PV+ = 0.752) and negative (PV = 0.997) test results Observed

Dam + daughter+ (cell a) Dam + daughter (cell b) Dam  daughter+ (cell c) Dam  daughter (cell d)

Observed frequencies

325 201 117 3448

Corrected frequencies

Corrected Dam + daughter+

Dam + daughter

200 0.5 0.3 0.03

55 157 0.08 10.3

201

223

The corrected numbers of dam-daughter pairs were summed (Table 3) and subsequently used for the additional adjustments (Table 4, line b). 2.3.2. Adjustment for horizontal infection in dams Horizontal transmission tended to be underestimated when using cross-sectional data. This happened when a seronegative dam had been infected after its daughter was born. Romero and Frankena (2003) provided a method to adjust for this. They assumed that the probability of horizontal transmission in dams occurred at the same probability as in daughters. Of seronegative dams, the probability of daughters that was infected horizontally was calculated as (b/b + d) (referring to Table 1) and then the frequencies of a, c and (a + c) (Table 1) were reduced according to this proportion while the subtracted numbers were added to b, d and (b + d), respectively (Table 4, line c).

Dam  daughter+ 55 0.1 91.5 10.3

Dam  daughter 15.2 43.3 25.2 3427

157

3510

2.3.3. Adjustment for horizontal infection in daughters from a seropositive dam Overestimation of vertical transmission may have occurred when a seronegative daughter born from a seropositive dam had been infected horizontally. Adjustment for this kind of overestimation was done using the following algorithm: a0 ¼ ða þ cÞ ProbðVTÞ þ ða þ cÞ  ð1  ProbðVTÞÞ ProbðHTÞ

(1)

where a0 is equal to a after the abovementioned adjustment for horizontal infection in dams. It is the sum of the number of vertically-infected daughters born from seropositive dams ((a + c)Prob(VT)) plus the number of seropositive dams with daughters (originally not infected by vertical transmission) that were infected by horizontal transmission ((a + c)(1  Prob(VT)(Prob(HT)). Conse-

Table 4 Frequencies of dam-daughter pairs and probabilities of vertical (Prob(VT) and horizontal (Prob(HT) transmission of 4091 dam-daughter pairs in 96 Dutch dairy herds with 1 or more seropositive animal to N. caninum. Starting with the observed dam-daughter frequencies (a), the effects of three adjustments are illustrated: (1) adjusting for imperfect test characteristics (b); (2) adjustment for postnatal infection in dams after birth of daughters (c) and (3) adjustment for postnatal infection in seronegative daughters from seropositive dams (d). The vertical transmission probability (Prob(VT)) is given as the proportion of seropositive daughters among daughters from seropositive dams and the horizontal transmission probability (Prob(HT)) as the proportion of seropositive daughters among daughters from seronegative dams Status of daughters

Status of dams Seropositive

Seronegative

Prob(VT)% (95% CI)

Prob(HT)% (95% CI)

a

Observed dam-daughter combinations

Seropositive Seronegative

325 201

117 3448

61.8 (57.5–66.0)

3.3 (2.7–3.9)

b

Dam-daughter combinations adjusted for imperfect test characteristics

Seropositive Seronegative

201 223

157 3511

47.4 (42.7–52.4)

4.3 (3.6–5.0)

c

Dam-daughter combinations adjusted for postnatal infection of dams after birth of daughter

Seropositive

192

166

47.4 (42.6–52.3)

4.5 (3.9–5.2)

Seronegative

213

3521

Dam-daughter combinations adjusted for postnatal infection of daughters born from seropositive dams after birth

Seropositive

182

166

44.9 (40.0–49.9)

4.5 (3.9–5.2)

Seronegative

223

3521

d

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

quently, we assumed that after failure of vertical transmission the probability of horizontal transmission in daughters of seropositive dams was the same as that probability in daughters of seronegative dams. In this algorithm (a + c) and Prob(HT) are known (see adjustment of postnatal infections in dams after giving birth). Moreover, since Prob(VT) equals a/(a + c), a can be solved by rewriting Eq. (1) as: a¼

a0  ða þ cÞ ProbðHTÞ 1  ProbðHTÞ

(2)

The result of this adjustment on the frequencies of dam-daughter combinations is given in Table 4, line d. 2.4. Rate of horizontal infection The incidence rate of horizontal infection (IR(HT)) was calculated based on the adjusted Prob(HT) and the average age of animals. Based on the formula (Dohoo et al., 2003): ProbðHTÞ ¼ 1  expðIRðHTÞTÞ

(3)

where T is the average age of an animal, IR(HT) can be calculated by converting algorithm (3) into:   1  ProbðHTÞ IRðHTÞ ¼ ln (4) T assuming that the rate of infection is constant during an animal’s lifetime. 2.5. Stratification The study population was divided into herds with (N = 66) and without (N = 30) farm dogs, and herds with high prevalence (10% within-herd prevalence, N = 35) and with low prevalence (<10% within-herd prevalence, N = 61). Crude odds ratios for farm-dog presence (ORDOG) and seroprevalence (ORPREV) were calculated to quantify the effect of these explanatory variables. In addition, to assess the effect of farm-dog presence on the relation between seroprevalence and horizontal transmission, stratum-specific odds ratios (ORPREVjDOG) and the Mantel-Haensel summary odds ratio (ORMH_PREV) were calculated. 2.6. High seroprevalence age groups For each herd, the number of horizontally-infected animals was counted. Clusters of horizontally-infected animals were determined according to the method of Dijkstra et al. (2001). These researchers demonstrated

87

that the seroprevalence in specific age-groups was significantly higher compared with the other animals in the herd, indicating a point-source exposure to N. caninum. They also demonstrated that such exposures were found during a limited period of common housing and feeding of animals (Dijkstra et al., 2002b). We made use of this method by looking at a clustered presence of seropositive animals within time periods of 9 months. The time-window of 9-months was used because (pregnant) young stock is traditionally grouped together in specific age-groups in which the age difference varies between 6 and 9 months. 2.7. Statistical analyses Calculations for Prob(VT) and Prob(HT) were done in Excel (Microsoft1 Excel, 2002). Ninety-five percent confidence intervals for Prob(VT) and Prob(HT) were calculated using the exact binomial distribution (STATA/SE 8.2, 2004). Calculation of OR, stratified OR and MH summary OR was done as described by Dohoo et al. (2003) using Win Episcope 2.0 (Thrusfield et al., 2001). We considered P  0.05 to indicate statistical significance. 3. Results The average age of animals was 3.2 (S.D. 2.3) years. Overall seroprevalence in the 96 seropositive study herds was 11.3% (1173/10,350), and within-herd seroprevalence ranged from 0.5 to 49.1%. Of 4091 dam-daughter pairs, the serostatus of both dam and daughter was known (Table 4) (29% of all possible dam-daughters pairs in the sampled population). Based on the unadjusted analysis of dam-daughter serology, Prob(VT) was 61.8% (95% CI: 57.5–66.0%) and Prob(HT) was 3.3% (95% CI: 2.7–3.9%). When adjusting the observed frequencies for the predictive values of positive and negative test results, underestimation of horizontal and overestimation of vertical transmission, Prob(VT) decreased significantly to 44.9% (95% CI: 40.0–49.9%) while the Prob(HT) increased to 4.5% (95% CI: 3.9–5.2%) (Table 4). Based on this result and the average age of 3.2 years, the incidence rate for horizontal transmission was calculated as 1.4 (95% CI: 1.2–1.7) infections per 100 cowyears at risk. In herds with farm dogs, Prob(HT) was higher (5.5% (95% CI: 4.6–6.4%)) compared with herds without farm dogs (2.3% (95% CI: 1.5–3.4%)) and the crude ORDOG was 2.5 (95% CI: 1.7–3.9). Consequently, IR(HT) was higher in herd with farm dogs (1.8 (95% CI: 1.5–2.2)

88

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

Table 5 Frequencies of dam-daughter pairs, probabilities (Prob(HT) and incidence rates for horizontal transmission of 4091 dam-daughter pairs in 96 Dutch dairy herds with 1 or more animals seropositive for N. caninum. Frequencies of dam-daughter pairs are given after adjustment for imperfect test characteristics and horizontal infection of dams and daughters Status of daughters Herds with dog(s) present (N = 2924 pairs in 66 herds) Herds with no dog(s) present (N = 1167 pairs in 30 herds) High within-herd seroprevalence (N = 1486 pairs in 35 herds) Low within-herd seroprevalence (N = 2605 pairs in 61 herds)

Status of dams Seropositive

Seropositive

160

141

Seronegative

185

2439

Seropositive

20

26

Seronegative

38

1083

Seropositive

136

124

Seronegative

146

1080

Seropositive

32

49

Seronegative

78

2445

infections per 100 cow-years at risk) compared with herds without farm dogs (0.7 (95% CI: 0.5–0.11) infections per 100 cow-years at risk). A similar effect was seen for seroprevalence. In herds with high within-herd seroprevalence, Prob(HT) was higher (10.3% (95% CI: 8.6–12.2%) compared with herds with low seroprevalence (2.0% (95% CI: 1.5– 2.6%)). This resulted in a higher IR(HT) in high prevalence herds (3.4 (95% CI: 2.8–4.1) infections per 100 cow-years at risk) compared with low prevalence herds (0.6 (95% CI: 0.5–0.8) infections per 100 cowyears at risk) (Table 5). The crude ORPREV was 5.7 (95% CI: 4.0–7.9). However, the stratified ORPREVjDOG+ was 6.6 (95% CI: 4.2–9.8) while ORPREVjDOG- was 1.9 (95% CI: 0.7–5.5). The Mantel-Haensel summary ORMH_PREV

Prob(HT)% (95% CI)

Seronegative

Incidence rate per 100 cow-years at risk (95% CI)

5.7 (4.8–6.7)

1.8 (1.5–2.2)

2.3 (1.5–3.4)

0.7 (0.5–1.1)

10.3 (8.6–12.2)

3.4 (2.8–4.1)

2.0 (1.5–2.6)

0.6 (0.5–0.8)

was 5.6 (95% CI: 3.9–8.2) with a significant Chi-square value for the Breslow-Day statistic (P-value = 0.03). This indicated that the strength of the relation between seroprevalence and horizontal transmission depends largely on the presence of farm dogs(s). Fifty-seven (59%) out of 96 seropositive herds had at least 1 horizontally-infected animal and 27 herds had 2 horizontally-infected animals. In seven herds, specific age-groups with significantly higher seroprevalence compared with the rest of the herd were present (Table 6). In Fig. 1, the situation in herd 7 with 128 animals and 33.3% seroprevalence of N. caninum infection is illustrated. A cluster of horizontally infected animals was born between April and December 2000. Sixteen out of the 18 animals born during this period

Table 6 Descriptive information on seven herds with specific seropositive age-groups indicative for point-source infection with N. caninum from a random sample of 108 Dutch dairy herds (sampled in 2003) Farm

#Animals sampled

#Seropos. animals/#animals excl. cluster

#Seropos. animals/#animals in cluster

P-value

#Horizontally infected animals in cluster

Period of birth-dates for cluster

1 2 3

89 67 55

3/83 22/60 19/46

2/6 7/7 8/9

0.03 <0.01 <0.01

2 2 3

November 2000 November 1999–February 2000 August 2000–January 2001

4

170

10/126

3/10 4/12 5/14 3/8

0.02 <0.01 <0.01 0.03

2 2 2 2

November–December 1999 July 2000–January 2001 July–November 2001 April–June 2002

5 6 7

146 127 128

20/130 21/107 27/110

14/16 10/20 16/18

<0.01 <0.01 <0.01

6 7 7

April 1998–January 1999 May–December 2000 April–December 2000

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

89

Fig. 1. Overview of serological testing for N. caninum in herd 7 (N = 128 animals) and assigned infection route of seropositive animals.

tested seropositive. Six of these had an unknown infection route, three were assigned a vertical and seven a horizontal infection route. Two dogs had been purchased on this farm, the first at the end of 1995 and the second at the beginning of 2001. 4. Discussion In the present study, we quantified the probability of vertical and horizontal transmission of N. caninum infection in Dutch dairy herds. We applied adjustments to the observed frequencies of dam-daughter combinations to account for imperfect test characteristics, underestimation of horizontal and overestimation of vertical transmission. Additionally, we defined a time at risk, allowing for the conversion of probabilities into rates. For vertical transmission the time at risk was one pregnancy, while for horizontal transmission, we converted Prob(HT) into IR(HT) by using the average age of animals as the time at risk. In epidemiological terms, the incidence rate is an important parameter and it allows direct comparison between studies. For this reason, it is preferably used in simulation models to evaluate different disease control strategies. The unadjusted probability of vertical transmission (Prob(VT)) in the observed dam-daughter relations was 61.8%. This percentage decreased significantly to 44.9% when dam-daughter frequencies were corrected. This drop in Prob(VT) was mainly caused by the effect of test specificity in a population with moderate seroprevalence, leading to relatively low predictive

value for positive test results. The adjustment related to overestimation of vertical transmission had much less effect on the final outcome of Prob(VT). When applying adjustments on Prob(HT), this probability increased from 3.3 to 4.5%. Again, the adjustment for imperfect test characteristics had a greater effect on Prob(HT) than adjusting for underestimation of horizontal transmission. This probability corresponded with an incidence rate for horizontal transmission of 1.4 per 100 cow-years at risk in a random group of seropositive Dutch dairy herds. Previously, a limited number of prospective studies on dairies with N. caninum-associated abortion problems have been conducted, providing incidence rates of horizontal transmission. These varied between less than 1% per year (Hietala and Thurmond, 1999) to an overall estimate of 1.9 horizontal infections per 100 heifer-years at risk (Davison et al., 1999b) and 8.5 horizontal infections per 100 cow-years at risk (Pare´ et al., 1997). Comparison with our incidence rate needs caution because these three studies were based on herd(s) situations with N. caninum-associated abortion problems. There have also been studies looking into vertical and horizontal transmission probabilities based on cross-sectional data (Dijkstra et al., 2001; Pan et al., 2004; Romero and Frankena, 2003). However, none or limited adjustments such as described in our study were carried out. Therefore, the presented probabilities for vertical transmission will most likely be overestimated and for horizontal transmission will be underestimated.

90

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

Looking at the data of one of these studies based on Dutch dairy herds with N. caninum-associated abortion problems (Dijkstra et al., 2001), the presented Prob(VT) and Prob(HT) were 68.8% and 22.8%, respectively. When carrying out the adjustments described in our study, Prob(VT) and Prob(HT) were 42.0% and 30.0%, respectively. When assuming the same average age of animals as in our study, the incidence rate for horizontal transmission in these abortion-problem herds was 11.1 infections per 100 cow-years at risk, or eight times higher as in the random herds studied in our study. Thus, comparison between studies shows that differences in vertical and horizontal transmission probabilities are related to what is known about the herd situation. In addition, it is plausible that vertical and horizontal transmission probabilities vary between countries because of variation in the presence of definitive hosts, conditions for oocyst sporulation and virulence in parasite strains. Both farm-dog presence and high within-herd seroprevalence had a strong statistical relation with Prob(HT): the ORDOG for presence of farm dogs and ORPREV for high within-herd seroprevalence were 2.5 and 5.7, respectively. A similar effect of the relation between within-herd seroprevalence and horizontal transmission was found by Romero and Frankena (2003). This relationship was explained by an increased exposure to environmental sources of infection, including infected placentas, amniotic fluid or water and food contaminated with N. caninum oocysts from dogs. We were able to underscore the importance of farm dogs on the occurrence of horizontal transmission by stratification for farm-dog presence. The association between within-herd seroprevalence and Prob(HT) increased when farms had dogs while, on farms without dogs, this relation became weaker (ORPREVjDOG = 1.9) and was no longer statistically significant. This illustrates that the effect of within-herd seroprevalence on Prob(HT) depended for a large part on the presence of farm dogs. In biological terms, it reinforced the important role farm dogs play in spreading the infection. In an environment of high within-herd seroprevalence, dogs have a greater chance to become infected and the probability of horizontal transmission increases accordingly. When no farm dogs were present, the probability of horizontal transmission tended to increase with an increasing seroprevalence. This indicated that cattle may acquire new infections by sources other than farm dogs. In the Dutch situation, where dogs are the only known definitive hosts, stray dogs most likely act as additional sources of oocysts. This was earlier hypothesized by Schares et al. (2004) in

a study comparable to the Dutch situation. Another more speculative hypothesis is the uptake of infectious stages other than sporulated oocysts such as tachyzoites. In a study by Uggla et al. (1998), neonatal calves were infected orally by colostrum spiked with Nc-SweB1 tachyzoites. In an experiment in which heifers were challenged orally with tachyzoites, one out of eight animals seroconverted (Weston et al., 2005). Possible horizontal transmission in cattle with oral lesions was suggested when these cattle ingest food or water contaminated with tachyzoites such as from placenta or vaginal discharge from cattle calving or aborting due to N. caninum. In 7 out of 96 herds evidence was found for a pointsource infection based on the existence of a specific agegroup of animals in which the seroprevalence for N. caninum was higher compared with the rest of the herd. Simultaneously, it indicated that point-source infections did not necessarily lead to increased clinical neosporosis but could occur without abortion problems. Previously Dijkstra et al. (2002a) have described a similar finding in one herd in which more than half of the animals seroconverted without any signs of abortion. Whatever the unknown mechanisms of horizontal transmission, our findings emphasize the appropriateness of current control strategies to reduce seroprevalence by testing animals and subsequently deciding to cull seropositive animals and/or their offspring. In addition, strict dog-management measures are necessary. It is particularly important to prevent dogs (both dogs living on the premises and stray dogs) from being present at calving and to prevent contamination of feed and drinking water with dog faeces (Dijkstra et al., 2005). When using unadjusted cross-sectional data, probabilities of vertical transmission tend to be overestimated and probabilities of horizontal transmission tend to be underestimated. A cohort study in which both dam and daughter were bled at regular intervals would be a preferred study design. In such a study design, more accurate probabilities for vertical and horizontal transmission could be calculated because it would allow for correction biases such as variation in antibody titers by age and throughout pregnancy (Davison et al., 1999a; Maley et al., 2001; Stenlund et al., 1999). In addition, a cohort study would allow for accurate calculation of ‘animal time at risk’ as denominator for the horizontal-transmission rate. However, as cohort studies require long study periods and subsequently more financial means, these are applied scarcely and often involving a limited number of herds. By adjusting

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

cross-sectional data according the methods described here, we believe that cross-sectional studies can provide reliable estimates for vertical and horizontal transmission probabilities. In conclusion, the adjusted probabilities of vertical and horizontal transmission in Dutch cattle were 44.9% and 4.5%, respectively. The incidence rate for horizontal transmission was 1.4 infections per 100 cow-years at risk. The adjustment for imperfect test characteristics had a major impact on the estimated probabilities. The observed relation between withinherd seroprevalence and Prob(HT) was largely dependent on the presence of farm dogs. Additionally, in 7% of dairy herds we found a specific age-group with high seroprevalence indicating a point-source infection. Such point-source infections apparently may occur without obvious clinical signs. This means that the within-herd seroprevalence can increase unnoticed to levels that pose a risk for an abortion epidemic. Therefore, current control strategies based on containing seroprevalence and managing dogs on farm remain important. Acknowledgements Serological data of the participating dairy herds were obtained from previous studies, which were funded by the Dairy Commodity Board (Rijswijk, The Netherlands) and the European Union (Research Project QLK2-CT2001-0150 ‘‘Diagnosis and epidemiology of Neospora caninum associated bovine abortions’’). The authors thank Dr. Maarten Eysker and Ir. Wim Swart for their valuable contribution to this manuscript and participating farmers for allowing the use of herd data, the NRS (the Dutch Cattle Improvement Organisation, Arnhem, The Netherlands) for data supply. References Anderson, M.L., Reynolds, J.P., Rowe, J.D., Sverlow, K.W., Packham, A.E., Barr, B.C., Conrad, P.A., 1997. Evidence of vertical transmission of Neospora sp. infection in dairy cattle. J. Am. Vet. Med. Assoc. 210, 1169–1172. Bartels, C.J.M., Arnaiz-Seco, I., Ruiz-Santa-Quitera, J.A., Bjo¨rkman, C., Fro¨ssling, J., Von Blumro¨der, D., Conraths, F.J., Schares, G., van Maanen, C., Wouda, W., Ortega-Mora, L.M., 2006a. Supranational comparison of Neospora caninum seroprevalences in cattle in Germany, The Netherlands, Spain and Sweden. Vet. Parasitol. 137, 17–27. Bartels, C.J.M., van Schaik, G., Veldhuisen, J.P., van den Borne, B.H.P., Wouda, W., Dijkstra, Th., 2006b. Effect of Neospora caninum serostatus on culling, reproductive performance and milk production in Dutch dairy herds with and without a history of Neospora caninum-associated epidemics. Prev. Vet. Med. 77, 186–198.

91

Bartels, C.J.M., Wouda, W., Schukken, Y.H., 1999. Risk factors for Neospora caninum-associated abortion storms in dairy herds in the Netherlands (1995 to 1997). Theriogenology 52, 247–257. Bjo¨rkman, C., McAllister, M.M., Fro¨ssling, J., Na¨slund, K., Leung, F., Uggla, A., 2003. Application of the Neospora caninum IgG avidity ELISA in assessment of chronic reproductive losses after an outbreak of neosporosis in a herd of beef cattle. J. Vet. Diagn. Invest. 15, 3–7. Bjo¨rkman, C., Johansson, O., Stenlund, S., Holmdahl, O.J.M., Uggla, A., 1996. Neospora species infection in a herd of dairy cattle. J. Am. Vet. Med. Assoc. 208, 1441–1444. Davison, H.C., French, N.P., Trees, A.J., 1999a. Herd-specific and age-specific seroprevalence of Neospora caninum in 14 British dairy herds. Vet. Rec. 144, 547–550. Davison, H.C., Otter, A., Trees, A.J., 1999b. Estimation of vertical and horizontal transmission parameters of Neospora caninum infections in dairy cattle. Int. J. Parasitol. 29, 1683–1689. Dijkstra, Th., Bartels, C.J.M., Wouda, W., 2005. Control of bovine neosporosis: experiences from the Netherlands. In: Abstract from the 20th International Conference of the World Association for the Advancement of Veterinary Parasitology (WAAVP), Christchurch, New Zealand, p. 191. Dijkstra, T., Barkema, H.W., Bjo¨rkman, C., Wouda, W., 2002a. A high rate of seroconversion for Neospora caninum in a dairy herd without an obvious increased incidence of abortions. Vet. Parasitol. 109, 203–211. Dijkstra, Th., Barkema, H.W., Hesselink, J.W., Wouda, W., 2002b. Point source exposure of cattle to Neospora caninum consistent with periods of common housing and feeding and related to the introduction of a dog. Vet. Parasitol. 105, 89–98. Dijkstra, Th., Barkema, H.W., Eysker, M., Wouda, W., 2001. Evidence of post-natal transmission of Neospora caninum in Dutch dairy herds. Int. J. Parasitol. 31, 209–215. Dohoo, I., Martin, W., Stryhn, H., 2003. Veterinary Epidemiologic Research, Chapter 4: Measures of Disease Frequency. AVC Inc., Charlottetown, Prince Edward Island, Canada. Dubey, J.P., Buxton, D., Wouda, W., 2006. Pathogenesis of bovine neosporosis. J. Comp. Path. 134, 267–289. Dyer, R.M., Jenkins, M.C., Kwok, O.C.H., Douglas, L.W., Dubey, J.P., 2000. Serologic survey of Neospora caninum infection in a closed dairy cattle herd in Maryland: risk of serologic reactivity by production groups. Vet. Parasitol. 90, 171–181. French, N.P., Clancy, D., Davison, H.C., Trees, A.J., 1999. Mathematical models of Neospora caninum infection in dairy cattle: transmission and options for control. Int. J. Parasitol. 29, 1691– 1704. Fro¨ssling, J., Uggla, A., Bjo¨rkman, C., 2005. Prevalence and transmission of Neospora caninum within infected Swedish dairy herds. Vet. Parasitol. 128, 209–218. Gondim, L.F.P., McAllister, M.M., Pitt, W.C., Zemlicka, D.E., 2004. Coyotes (Canis latrans) are definitive hosts of Neospora caninum. Int. J. Parasitol. 34, 159–161. Hernandez, J., Risco, C., Donovan, A., 2001. Association between exposure to Neospora caninum and milk production in dairy cows. J. Am. Vet. Med. Assoc. 219, 632–635. Hietala, S.K., Thurmond, M.C., 1999. Postnatal Neospora caninum transmission and transient serologic responses in two dairies. Int. J. Parasitol. 29, 1669–1676. Hobson, J.C., Duffield, T.F., Kelton, D., Lissemore, K., Hietala, S.K., Leslie, K.E., McEwen, B., Cramer, G., Peregrine, A.S., 2002. Neospora caninum serostatus and milk production of Holstein cattle. J. Am. Vet. Med. Assoc. 221, 1160–1164.

92

C.J.M. Bartels et al. / Veterinary Parasitology 148 (2007) 83–92

Mainar-Jaime, R.C., Thurmond, M.C., Berzal-Herranz, B., Hietala, S.K., 1999. Seroprevalence of Neospora caninum and abortion in dairy cows in northern Spain. Vet. Rec. 145, 72–75. Maley, S.W., Buxton, D., Thomson, K.M., Schriefer, C.E.S., Innes, E.A., 2001. Serological analysis of calves experimentally infected with Neospora caninum: a 1-year study. Vet. Parasitol. 96, 1–9. McAllister, M.M., Dubey, J.P., Lindsay, D.S., Jolley, W.R., Wills, R.A., McGuire, A.M., 1998. Dogs are definitive hosts of Neospora caninum. Int. J. Parasitol. 28, 1473–1478. McAllister, M., Huffman, E.M., Hietala, S.K., Conrad, P.A., Anderson, M.L., Salman, M.D., 1996. Evidence suggesting a point source exposure in an outbreak of bovine abortion due to neosporosis. J. Vet. Diagn. Invest. 8, 355–357. Pan, Y.G., Jansen, B., Duffield, T.F., Hietala, S., Kelton, D., Lin, C.Y., Peregrine, A.S., 2004. Genetic susceptibility to Neospora caninum infection in Holstein cattle in Ontario. J. Dairy Sci. 87, 3967–3975. Pare´, J., Fecteau, F.G., Fortin, M., Marsolais, G., 1998. Seroepidemiologic study of Neospora caninum in dairy herds. J. Am. Vet. Med. Assoc. 213, 1595–1598. Pare´, J., Thurmond, M.C., Hietala, S.K., 1997. Neospora caninum antibodies in cows during pregnancy as a predictor of congenital infection and abortion. J. Parasitol. 83, 82–87. Romero, J.J., Van Breda, S., Vargas, B., Dolz, G., Frankena, K., 2005. Effect of neosporosis on productive and reproductive performance of dairy cattle in Costa Rica. Theriogenology 64, 1928–1939. Romero, J.J., Frankena, K., 2003. The effect of the dam-calf relationship on serostatus to Neospora caninum on 20 Costa Rican dairy farms. Vet. Parasitol. 114, 159–171. Schares, G., Ba¨rwald, A., Staubach, C., Ziller, M., Kloss, D., Schroder, R., Labohm, R., Drager, K., Fasen, W., Hess, R.G., Conraths, F.J., 2004. Potential risk factors for bovine Neospora caninum infection in Germany are not under the control of the farmers. Parasitology 129, 301–309. Schares, G., Peters, M., Wurm, R., Ba¨rwald, A., Conraths, F.J., 1998. The efficiency of vertical transmission of Neospora caninum in dairy cattle analysed by serological techniques. Vet. Parasitol. 80, 87–98. Stenlund, S., Kindahl, H., Magnusson, U., Uggla, U., Bjo¨rkman, C., 1999. Serum antibody profile and reproductive performance during two consecutive pregnancies of cows naturally infected with Neospora caninum. Vet. Parasitol. 85, 227–234. Thrusfield, M., Ortega, C., de Blas, I., Noordhuizen, J.P., Frankena, K., 2001. Win Episcope 2.0: improved epidemiological software for veterinary medicine. Vet. Rec. 148, 567–572.

Thurmond, M.C., Hietala, S.K., Blanchard, P.C., 1997. Herd-based diagnosis of Neospora caninum-induced endemic and epidemic abortion in cows and evidence for congenital and postnatal transmission. J. Vet. Diagn. Invest. 9, 44–49. Thurmond, M.C., Hietala, S.K., 1997. Effect of Neospora caninum infection on milk production in first-lactation dairy cows. J. Am. Vet. Med. Assoc. 210, 672–674. Thurmond, M.C., Hietala, S.K., 1996. Culling associated with Neospora caninum infection in dairy cows. Am. J. Vet. Res. 57, 1559– 1562. Tiwari, A., VanLeeuwen, J.A., Dohoo, I.R., Stryhn, H., Keefe, G.P., Haddad, J.P., 2005. Effects of seropositivity for bovine leukemia virus, bovine viral diarrhea virus, Mycobacterium avium subspecies paratuberculosis, and Neospora caninum on culling in dairy cattle in four Canadian provinces. Vet. Microbiol. 109, 147–158. Uggla, A., Stenlund, S., Holmdahl, O.J.M., Jakubek, E.B., Thebo, T., Kindahl, H., Bjo¨rkman, C., 1998. Oral Neospora caninum inoculation of neonatal calves. Int. J. Parasitol. 28, 1467–1472. Von Blumro¨der, D., Schares, G., Norton, R., Williams, D.J., EstebanRedondo, I., Wright, S., Bjo¨rkman, C., Fro¨ssling, J., Risco-Castillo, Fernandez-Garcia, A., Ortega-Mora, L.M., Sager, H., Hemphill, A., van Maanen, C., Wouda, W., Conraths, F.J., 2004. Comparison and standardisation of serological methods for the diagnosis of Neospora caninum infection in bovines. Vet. Parasitol. 120, 11–22. Waldner, C.L., Janzen, E.D., Henderson, J., Haines, D.J., 1999. Outbreak of abortion associated with Neospora caninium infection in a beef herd. J. Am. Vet. Med. Ass. 215, 1485–1489. Weston, J.F., Williamson, N.B., Pomroy, W.E., Olson, W., 2005. Friesian/Jersey crossbred heifers that were previously consistently seronegative to Neospora caninum were challenged at 70 days gestation. In: 20th International Conference of the World Association for the Advancement of Veterinary Parasitology (WAAVP), Christchurch, New Zealand. Wouda, W., Dijkstra, Th., Kramer, A.M.H., Maanen, C., van, Brinkhof, J.M.A., 1999. Seroepidemiological evidence for a relationship between Neospora caninum infections in dogs and cattle. Int. J. Parasitol. 29, 1677–1682. Wouda, W., Brinkhof, J., van Maanen, C., de Gee, A.L.W., Moen, A.R., 1998. Serodiagnosis of neosporosis in individual cows and dairy herds: a comparative study of three enzyme-linked immunosorbent assays. Clin. Diagn. Lab. Immunol. 5, 711–716. Wouda, W., Moen, A.R., Schukken, Y.H., 1998. Abortion risk in progeny of cows after a Neospora caninum epidemic. Theriogenology 49, 1311–1316.