- Email: [email protected]
Risk factors for neospora caninum-associated abortion storms in dairy herds in The Netherlands (1995 to 1997)
ELSEVIER
RISK FACTORS FOR Neosnora caninum-ASSOCIATED ABORTION STORMS IN DAIRY HERDS IN THE NETHERLANDS (1995 to 1997) C. J. M. Bartels,’ W. Wouda,zaand Y. H. Schukken’ ‘Department of Herd Health and Reproduction Utrecht University, The Netherlands ‘Animal Health Service Drachten, The Netherlands Received for publication: Accepted:
23
hly
1998
9 April
1999
ABSTRACT A 2 to 1 matched case control study design was used to analyze herd level risk factors for Neosoora caninum-associated abortion storms in 47 dairy herds. Data were obtained using a questionnaire regarding the state of affairs at the farms over the 2 years prior to the abortion storm. The questionnaire included 120 variables considered to be potential risk factors for either introduction of infection or recrudescence of chronic infection. The relationship between risk factors and case control pairs was analyzed by conditional logistic regression using a three-steps procedure. In addition, cross sectional serology was used to assess the possible role of concomitant infections. The main factors that were significant in the analysis and that were considered to have potential biological relevance were the presence of dogs, the presence of poultry, and the feeding of moldy maize-silage during summer. For both the presence of dogs and the presence of poultry on the farms, a linear relationship was found between the number of animals and the assessed risk for an abortion storm. These findings suggest a possible role of these species in the transmission of N. caninum. Further evidence for such a role of dogs was the significant association between the presence of dogs and the presence of seropositive cattle in the control herds. The feeding of moldy fodder is considered to be a factor which may induce recrudescence of a latent N. caninum-infection by mycotoxins causing immune suppression. We also found some evidence for a possible influence of management practices around calving and a high prevalence of retained afterbirths. No significant association was found for herd level prevalence of antibodies to bovine viral diarrhea virus, bovine herpesvirus 1, Lentosnira &$Q or Salmonella d&&t. 0 1999 by Elsevier Science Inc.
Key words: Neosnora caninum, bovine, abortion, risk factors, transmission Acknowledgments The authors thank the dairy farmers and their veterinary practitioners for participation; A. R. Moen and M. F. Weber for field work and useful suggestions. We are grateful to J. Part for contributing to the design of the questionnaire, to C. L. J. J. Kruitwagen for help with statistics, and to R.N. Zadoks for valuable comments on the manuscript. aCorrespondence and reprint requests: Animal Health Service, P. 0. Box, 9200 AJ Drachten, The Netherlands. Phone: 31 512 570 700 Fax: 31 512 520 013 E-mail: [email protected] Therlogendogy 52:247-257,1999 Q 1999 by Elsevter Science Inc.
CQQ3-691 XI99&-see front matter PII soo93-691x(9s)oo12s-o
Theriogenology
248 INTRODUCTION
Neosoora caninum is a protozoan parasite that has been associated with abortion in cattle. After the first description of bovine neosporosis-like abortions in 1989 (24), numerous reports have been published on both endemic and epidemic N. caninum abortions in cattle all over the world (for review see: 9). Vertical transmission from cow to calf has been well established (2, 4, 18, 35) and may contribute significantly to the persistence of the infection in the herd, causing an increased rate of abortion (5, 18, 27, 32, 35). However, evidence is accumulating that some form of postnatal transmission may be occurring since a high prevalence of N. caninuminfection has been found in dairy herds (11, 18, 32). The recently reported experimental evidence that dogs can act as definitive hosts of N. caninum (15) may have important implications for the epidemiology of bovine neosporosis. Several reports have described N. caninum-associated abortion storms which had an explosive character (14, 16, 25, 28, 37). Two hypotheses can be put forward to explain such abortion storms. The first hypothesis holds that the abortions are caused by the introduction of the infection into a herd by exposure to infectious oocysts. (3, 14, 16, 28, 37). The second hypothesis asserts that the abortions are caused by recrudescence of the infection in latently infected cows, by factors influencing immune suppression. Particularly viral infections and mycotoxicosis have been mentioned in this respect (26). Abortion storms associated with N. caninum have been recorded repeatedly in The Netherlands during recent years (16, 34, 36). Therefore, a case control study was begun to investigate the factors that might contribute to the occurrence of such abortion storms. In a companion paper we describe the characteristics of N_.caninum-associated abortion storms in 50 dairy herds, as monitored by fetal submissions to the laboratories of the Dutch Animal Health Service during 1995, 1996 and 1997 (32). From that study, it appeared that abortion storms occurred in herds which had a high prevalence of antibodies to s. caninum. Most abortion storms appeared to be induced by factors causing recrudescence of infection, rather than the immediate result of recent introduction. Hence our interest was in finding factors that might explain high seroprevalence and factors that might explain the occurrence of simultaneous abortions in these high seroprevalence herds. In the present study, we used a 2 to 1 matched case control study design to analyze dam from these herds that were obtained by a questionnaire. In addition, cross-sectional serology was used to assess the possible role of concomitant infections. MATERIALS
AND METHODS
Study Design A 2 to 1 matched case control design was used in this study. The sampling unit was a dairy herd. Cases were defined as dairy herds that had had an abortion storm associated with N. caninum during 1995, 1996 and 1997. An abortion storm was defined as a series of abortions affecting at least 12.5% (1 of 8) of the animals at risk within a period of 2 mo. Animals at risk were defined as the heifers and/or cows pregnant between 100 and 260 d (10,
Theriogenology
249
32). An association of an abortion storm with N. caninum was presumed when N. caninum infection was diagnosed in at least 2 fetuses (36), when no other cause of abortion was found in more than 1 fetus, and when antibodies to N. caninum were present in at least 80% of the aborting cows. In 4 herds the diagnosis of N. caninum infection was only made in 1 fetus, but 100% of the aborting cows were seropositive to N. caninum (32). An ELISA was used for detection of antibodies to N. caninurn as described previously (33). The population under study consisted of Dutch dairy herds in which the management would report the instance of multiple abortions to a veterinary practitioner for advice and diagnosis. Subsequently, the veterinary practitioner would encourage the submission of aborted fetuses to the Dutch Animal Health Service and take blood samples for diagnostic evaluation. Case herds were selected from the data base of fetal submissions to the laboratories of the Dutch Animal Health Service. Controls consisted of dairy herds without a history of an abortion storm between 1993 and 1997. Controls were matched with veterinary practitioners who were asked to provide a list of 6 client dairies with management practices similar to those of the case herd. It is therefore assumed that if a control herd had an abortion outbreak due to N_ canimun, it would have been classified as a case herd in our catchment scheme. Three dairies were randomly selected from each list. Two of these were asked to participate in the study while the third was kept in reserve in case a dairy farm had withdrawn from the study. Of the 50 case herds reported in a previous study (32), 3 herds with a cumulative incidence of abortions lower than 12.5% were excluded from our present study since they did not meet the originally defined conditions. In the remaining 47 herds 12.5 to 57% of the animals at risk had aborted within a period of 6 to 65 d. For a more detailed description of the abortion storms we refer to our companion paper (32). Data Collection All participating case and control farms were visited by one of the authors (CJMB) between January and November 1997. A questionnaire regarding the state of affairs at the farm over the past 2 yr prior to the abortion storm or prior to the visit (control herds) was administered on site. The questionnaire had been designed using expert consultation and literature review. Potential risk factors had been selected and assigned to clusters. A total of 120 variables as part of 10 clusters was used in the questionnaire and for further analysis. Seven clusters (A-G) were defined that contained variables thought possibly to be related to the introduction of or the horizontal transmission of infection with N. caninum while 3 clusters (H-J) contained variables that might be related to recrudescence of the infection. The defined clusters with the main variables are represented in Table 1. In addition, serology was used to assess the role of concomitant infections. Blood samples had been taken from 20% of the animals in the herds for detection of antibodies to N. caninum. Samples were taken from all age-groups (divided into 5 age groups: calves 0 to 1 yr, heifers 1 to 2 yr, Parity 1 cows, Parity 2 cows, Parity 3 and older cows) as reported previously (32). These sera were also used for detection of antibodies to bovine herpes virus 1 (BHVl), bovine viral diarrhea virus (BVDV), and Lentosnira hardio by ELISA methods and to Salmonella dublin by serum agglutination tests (0 and H antigen). To save costs, it was decided to confine the testing to 80% of the samples (37 of 47 matched cases and controls)
Theriogenology
250
unless the results indicated a complete analysis. A sample of 117 cows which had aborted during the abortion storms was analyzed for the serostatus to BVDV and N. caninum. Table 1. Defined clusters with the main variables used in the conditional logistic regression analysis of risk factors for Neosnora caninum-associated abortion storms
A B C D E F G H
Cluster Contacts with other animal species Contacts with man Contacts with cattle from other herds Nature and origin of fodder Source of drinking water Management around calving General management Stress inducing events during last month
I
Prevalence of other diseases
J
Nutritional abnormalities
Main variables Other farm animals, dogs, cats, rabbits, wild rodents, poultry, wild birds. Location of farmer’s house in relation to animal barns, use of artificial insemination. Communal pasture grounds, cattle shows, export sampling places, market places, purchase of animals. Nature of fodder for dairy cows, dry cows and heifers, respectively, turn-over of roughage, origin of fodder. Origin of drinking water by season, change of water supply. Use and hygiene of calving pen, use of calving pen to hospitalize sick animals, disposal of placenta, disposal of aborted fetuses. Housing system, pasturing management, origin of manure. Manipulations with animals (frequent palpation per rectum, herd vaccination, herd hoof trimming), events during previous month (climatic changes, unrest, prevalence of sick animals, frequent visits by veterinary practitioner) Infectious diseases (bovine viral diarrhea, infectious bovine rhinotracheitis, leptospirosis, salmonellosis, paratuberculosis), vaccination strategy, abomasal displacement, mastitis, retained afterbirth. Protein surplus in ration, fodder changes, mode of storage and conservation of roughage, use of moldy fodder, use of remnantfodder.
Statistical Analysis All variables were screened for unlikely values. The relationship between risk factors and case control pairs was analyzed by conditional logistic regression using Cox proportional hazards analysis (12, 19,23). The generalized linear model was expressed as: Ln (p/,.p) = a + pi Xi + s where “Ln (p/,.p)” stands for the log odds of abortion storm, and “a” for the log odds of abortion storm development for herds with “standard” (X=0) set of independent variable values, “lY’ measures the change in the log odds for a one-unit change in variable X and “a” represents the error term. Significant continuous variables were tested for a dose effect relation using categories, except for the variable “presence of dogs.” For this variable no categories were used since the range was only between 0 and 4.
251
Theriogenology
The model selection process consisted of 3 steps. In the first step, all variables were entered 1 by 1 into the analysis. Variables having a P-value of 0.25 or lower in the likelihood ratio test were included for further analysis. In the second step, variable selection was performed within each of the predefined clusters. Per cluster, a conditional logistic regression analysis with backward selection of variables was used, with the entry level at 0.05 and the removal level at 0.10. In the third step, the variables per cluster left were combined and again a backward stepwise selection was used with the same entry and removal level. Interaction terms were not analyzed in this study. The study size was estimated using a formula designed for case control studies with multiple controls per case (20). Defined values were needed for the level of significance (0.05), the power (>O.SO), the hypothesized relative risk associated with exposure value (RR=3) and for the relative frequency of exposure among controls (0.20). The outcome showed that with a sample size of 47 cases the power was more than 0.80, particularly when it was taken into account that this formula did not incorporate the matching as applied in our study. RESULTS Conditional
Logistic Regression Analysis
After the first step of analysis, 35 variables were significant and were used for further analysis. These variables represented 8 clusters. All variables within the clusters “source of drinking water” and “general management” were excluded from further analysis. After the second step of analysis by cluster, 15 variables remained significant. The variable “disposal of aborted fetuses” had to be. excluded from analysis in the final model because of its instability. The variable “frequent visits by the veterinary practitioner during the last month” was excluded from the final model because missing values severely affected the outcome. In the final modeling phase (Step 3) based on 13 variables, 5 variables were excluded because of nonsignificance. These variables were disposal of placenta (0 = in the manure pit, 1 = on the manure heap, 2 = otherwise); feeding of straw to dry cows during winter (dichotomous); feeding of brewer’s grain to dairy cows during winter (dichotomous); farmhouse connected to animal barn(s) (dichotomous); animals returned from market (dichotomous). The frequency distribution of the 8 variables that remained significant in the final model is represented in Table 2. The estimated coefficients for these variables are presented in Table 3. The final model containing 8 variables had a change in deviance from 103.27 to 36.41, with 9 degrees of freedom. The observed significance level was
of N. caninum Antibodies
in Cattle and Presence of Dogs
There was a significant association between the presence of dogs on control farms and the presence of N. caninum seropositive cattle on these farms. (OR = 4.2, P = 0.009). Eight of
Theriogenology
252
16 control farms where no dog was present had a zero N. caninum seroprevalence in the dairy stock. The other 8 herds without dogs had seroprevalences ranging from 4.8 to 37.5%, with a mean of 14.6%. The farm with the highest seroprevalence (37.5%) was closely neighboring a case farm, where 2 dogs were present. In addition, there was a weak but significant correlation between the number of dogs present on control farms and the seroprevalence of &I. caninum antibodies in the dairy stock (r = 0.23; P = 0.025). No such correlation was found in the case farms.
Table 2.
Frequency distribution of variables significant in the final step of conditional logistic regression analysis for assessment of risk factors for Neosnora caninum associated abortion storms.
Variable 1.
Presence of dogs
(No.1 (0) (1)
(2) (3) (4) (0) (l-10) P 10) pasture grounds for young
2.
Presence of poultry
3.
Use of communal stock Attendance at cattle shows during past 2 years Calving pen used to hospitalize sick animals Prevalence of retained afterbirths during previous year more than 10% Feeding of moldy maize-silage to dairy cows during summer Feeding of remnant-fodder to heifers during summer
4. 5. 6. 7. 8.
Case herds n=47 n (%I 1 (2.1) 24 (51.1) 17 (36.2) 5 (10.1) 0 (0) 12 (25.5) 20 (42.6) 15 (31.9)
Control herds n=94 W) 1: (17.0) 60 (63.8) 16 (17.0) 1 (1.1) :2 30 12
(E.:‘) (31.9) (12.8)
16
(34.0)
15
(15.9)
8 41
(17.0) (87.2)
30 70
(31.9) (74.5)
30
(63.8)
35
(37.2)
24
(51.1)
34
(36.2)
9
(19.1)
6
(6.4)
Serology No significant relationship was found between the seroprevalence on herd level of BVDV, BHVI, Lentosnira m and Salmonella dublin infections and the occurrence of abortion storms in the case herds. Analysis of serostatus to BVDV and N. caninum of 117 cows which had aborted during the abortion storm showed a negative relationship between seropositivity to BVDV and seropositivity to N. caninum (OR=O.42; P=O.O3). Seroprevalence of N. caninum antibodies was significantly higher in case herds compared to control herds as presented in a companion paper (32).
Theriogenology
253
Table 3. Results of conditional logistic regression analysis (backward stepwise method) on 47 case herds with an Neosuora caninum associated abortion storm and 94 matched control herds without such abortion storm. Significant variables
1. 2.
3. 4. 5. 6. 7. 8.
Presence of dogs Presence of pc&ry No of poultry l-10 Noofpoultry> 10 Use of communal pasture grounds for young stock Attendance at cattle shows during previous 2 years Calving pen used to hopitalize sick animals Prevalence of retained afterbirths in previous year > 10% Feeding of moldy maize-silage to dairy cows during summer Feeding of remnant-fodder to heifers during summer
pcoefficient
SE
P-value
Odds ratio
1.645
0.657
5.181
1.529 2.344
0.910 1.056
0.0123 0.0319 0.0928 0.0265
95% Confidence interval 11.43 - 18.8>
4.61 10.42
~0.78 - 27.5> cl.31 - 82.8>
2.176
1.131
0.0545
8.817
10.96 - 81.1>
-2.079
1.011
0.0398
0.125
~0.02 - 0.91>
3.538
1.406
0.0118
34.41
~2.19 - 541.1>
1.545
0.688
0.0248
4.691
cl.22 - 18.1>
1.884
0.821
0.0217
6.584
11.32 - 32.9>
3.684
1.316
0.0051
39.82
~3.02 - 525.3>
DISCUSSION This study was set up to explore herd level risk factors for N. caninum-associated abortion storms. The study was designed as a matched case control study in order to obtain optimal statistical efficiency. All case herds were managed by professional dairy farmers who had contacted their veterinary practitioner and subsequently had submitted fetuses for diagnostic evaluation. Control herds were chosen within the same veterinary practice in order to accommodate risk factors related to the veterinary practice that could not be measured. The questionnaire was composed to accommodate 2 hypothetical modes of origin of a N. caninum-associated abortion storm: 1) introduction of the infection into the herd and 2) recrudescence of the infection in chronically infected animals. When designing the questionnaire, we decided to include a large number of possible factors rather than to focus on a few factors in depth. Therefore, many variables were created and had to be screened separately. As a consequence, it cannot be excluded that some variables were significant by chance alone (significance level P = 0.05). However, in our opinion, the use of a three-steps analysis would have considerably limited the possibility of finding significant variables by chance in the final model. The interviews with the dairy farmers of case and control herds were held during a period of 11 mo in 1997. Seasonal differences may therefore have influenced the answers given, but this bias is considered nondifferential, since the interviews on case and matched control farms were conducted on the same day. However, there may have been some
254
Theriogenology
differential misclassification due to recall bias, as 12 of the abortion storms had occurred in 1995, 19 in 1996 and 16 in 1997. It is emphasized that a case control study does not determine causal relationships. This type of study is mainly used to generate hypotheses. The results from the conditional logistic regression analysis indicate statistical associations between a factor and an outcome, and asses the magnitude of the associated risk. Therefore, variables significant in our final model are not automatically definite risk factors for N. caninum-associated abortion storms. It is essential to assess the biological relevance of statistically significant variables. The presence of dogs on the farm was the most consistent factor in the model, and there was a significant linear relationship between the number of dogs and the assessed risk of an abortion storm. In addition, there was a significant association between the presence of dogs and the presence of seropositive cattle on the control farms. Similar observations have been reported in Canada (I 7). These results show a role of the dog in the transmission of the infection and they further corroborate the recent experimental evidence that dogs may shed N. caninum oocysts (15). Another significant variable was the presence of poultry on the premises. This included various species of domestic fowl like hens, ducks and geese. Again, a significant linear relationship was found between the number of animals and the assessed risk for an abortion storm. These animals could serve as vectors of oocysts, particularly when they are grubbing about freely. In addition, poultry might play a role as an intermediate host by which a dog may become infected (eating dead fowl). Two variables in the cluster “contacts with cattle from other herds” turned out to be significant in the final model. The variable “use of communal pasture grounds” was assessed as a risk for an abortion storm. Mostly this was related to young stock being pastured together with that from other herds on communal grounds away from the farm. It is doubtful that such animal contacts would lead to infection with N. --3 caninum as no evidence of lateral spread has been found by serology in a trial with 25 congenitally infected heifers and 25 noninfected heifers which were housed fed and managed together until calving (2). The other significant variable in this cluster was “attendance at cattle shows”, usually a presentation of young stock at a local fair. However, this variable showed a preventive effect in relation to the abortion storms, which is even more difficult to explain. The hygiene around calving seems to play a role as the variables “use of calving pen to hospitalize sick cows” and “high prevalence of retained afterbirth during previous year” both appeared in the final model. One could speculate that exposure to placenta and uterine effusions might serve as a source of infection as N. caninum tachyzoites have been demonstrated in the placenta (22,34). In a recent study, it was demonstrated that calves can be infected by oral ingestion of culture derived N. caninum tachyzoites added to colostrum (30). However, it is generally believed that tachyzoites will not survive for long in the environment, and a high infection dose would be needed, as has been demonstrated also for Toxonlasma nondii tachyzoites given orally to mice and cats (8). The last 2 significant variables “feeding of moldy maize-silage to dairy cows during the summer” and “feeding of remnant-fodder to heifers during summer” are both related to fodder of possibly inferior quality. Particularly, contamination of fodder by fungi is
Theriogenology
255
considered to be a potential risk factor, because mycotoxins may have been produced. Several mycotoxins have been shown to cause immunosuppression after repeated ingestion at low doses (7, 21). Such immunosuppression may lead to tissue cyst rupture, as has been demonstrated in mice with chronic toxoplasmosis (3 1). An explanation for the seasonal pattern of N. caninum-associated abortion storms, as reported previously (32), could lay in the favorable conditions of temperature and humidity for fungal growth during the summer months. On the other hand, similar conditions of temperature and humidity are probably also favorable for the sporulation of oocysts (13). We did not find any relationship between N. caninum-associated abortion storms and infection by BVDV, BI-IVI, L. hardio and S. dublin, based on the seroprevalence data at herd level for these infections. In particular, BVDV has been suspected to play a role in this respect (26), because infection with BVDV is known to cause immunosuppression (29). However, we found a negative relationship between seropositivity to BVDV and seropositivity to N_. caninum in aborting cows. In addition, we did not find dual infections with other abortifacients in fetuses submitted during the abortion storms (32) as has been reported previously (1). Various factors causing immune suppression may have been involved without being significant in the analysis. Several dairy farmers of case herds had their own explanations for the abortion storm, varying from a period of extremely high temperatures to vaccination of the herd prior to the abortion storm. Also mentioned repeatedly was a disturbed balance between energy and protein supply, due to low quality grass at the end of the pasturing season. However, immune suppression can not be easily assessed, and, therefore, its possible significance as a risk factor for the clinical manifestation of bovine neosporosis can not be demonstrated by statistical methods. The results of this study and of a previous study (32) support the idea that both vertical and postnatal transmission of N. caninum play a role in the epidemiology of the infection in cattle. The vertical transmission route has been well established previously (2,4, 18, 35). The present study provides further evidence that,in particular,dogs play a role in the transmission of N. caninum infection to cattle. REFERENCES 1. Agerholm JS, Willadsen CM, Nielsen TK, Giese SB, Holm E, Jensen L, Agger, JF. Diagnostic studies of abortion in Danish dairy herds. J Vet Med (A) 1997;44:551-558. 2. Anderson ML, Reynolds JP, Rowe JD, Sverlow KW, Packham AE, Barr BC, Conrad PA. Evidence of vertical transmission of Neosoora sp infection in dairy cattle. J Am Vet Med Assoc 1997;210:1169-1172. 3. Barr BC, Bjerkas I, Buxton D, Conrad PA, Dubey JP, Ellis JT, Jenkins MC, Johnston SA, Lindsay DS, Sibley D, Trees AJ, Wouda W. Neosporosis - Report of the International Neospora Workshop. Compend Contin Educ Pratt Vet 1997; 19:s 120-5 126,5 144. 4. Barr BC, Conrad PA, Breitmeyer R, Sverlow K, Anderson ML, Reynolds J, Chauvet AE, Dubey JP, Ardans AA. Congenital Neosnora infection in calves born from cows that had previously aborted Neospora-infected fetuses: four cases (1990-1992). J Am Vet Med Assoc 1993;202:113-117.
256
Theriogenology
5. Bjiirkman C, Johansson 0, Stenlund S, Holmdahl JM, Uggla A. Neosuora species infection in a herd of dairy cattle. J Am Vet Med Assoc 1996;208: 144 l-l 444. 6. Breslow NE, Day NE. Statistical methods in cancer research, Vol 1. The Analysis of Case-Control Studies. Lyon: IARC Sci Publ, 1980; 248-279. mechanisms of immunosuppresstion. Vet Immunol 7. Corrier DE. Mycotoxines: Immunopathol 1991;30:73-87. 8. Dubey JP. Re-examination of resistance of Toxonlasma gondii tachyzoites and bradyzoites to pepsin and trypsin digestion. Parasitology 1998;116:43-50. 9. Dubey JP, Lindsay DS. A review of Neospora caninum and neosporosis. Vet Parasitol 1996;67:1-59. 10. Fetrow J, McClary D, Harman R, Butcher K, Weaver L, Studer E, Ehrlich J, Etherington W, Guterbock W, Klinborg D, Reneau J, Williamson N. Calculating selected reproductive indices: recommendations of the American Association of Bovine Practitioners. J Dairy Sci 1990;73:778-90. 11. French NP, Davison HC, Clancy D, Begon M, Trees AJ. Modelling of Neosnora species infection in dairy cattle: the importance of horizontal and vertical transmission and differential culling. Thrusfield MV, Goodall EA (eds), Proc Sot Vet Epidemiol Prev Med 1998;113-122. 12. Hosmer DW, Lemeshow S. AppliedLogisticRegression. New York: John Wiley and Sons, Inq1989; 187-215. 13. Lindsay DS, Current WL, Ernst. Sporogony of Isosnora suis Biester, 1934 of swine. J Parasitol 1982;68:861-865. 14. McAllister MM, Huffman EM, Hietala SK, Conrad PA, Anderson ML, Salman MD. Evidence suggesting a point source exposure in an outbreak of bovine abortion due to neosporosis. J Vet Diagn Invest 1996; 8:355-357. 15. McAllister MM, Dubey JP, Lindsay DS, Jolley WR, Wills RA, McGuire AM. Dogs are definitive hosts of Neosuora caninum Int J Parasitol 1998;28:1473-1478. 16. Moen AR, Wouda W, Mu1 GGraat EAM, Werven T van. Increased risk of abortion following Neosuora caninum abortion outbreaks: a retrospective and prospective cohort study in four dairy herds. Theriogenology 1998; 49: 1301-l 309. 17. Pare J, Fecteau G, Fortin M, Marsolais G. Seroepidemiologic study of Neosnora canimnn in dairy herds. J Am Vet Med Assoc 1998;213:1595-1598. 18. Pare J, Thurmond MC, Hietala SK. Congenital Neosnora caninum infection in dairy cattle and associated calthood mortality. Can J Vet Res 1996;60: 133-l 39. 19. Rothman KJ. Modem Epidemiology. Boston/Toronto: Little, Brown and Co, 1986; 297298. 20. Schlesselman JJ. Case-Control Studies, Design, Conduct, Analysis. Oxford: Oxford University Press, 1982; Chapter 6: 150-152. 21. Sharma RP. Immunotoxicity of mycotoxins. J Dairy Sci 1993;78:892-897. 22. Shivaprasad HL, Ely R and Dubey JP. A Neoswra-like protozoon in an aborted bovine placenta. Vet Parasitol 1989; 34: 145-8. 23. SPSS Advanced Statistics 6.1 software package. Users Guide. Chicago 1994; 291-328. 24. Thilsted JP, Dubey JP. Neosporosis-like abortions in a herd of dairy cattle. J Vet Diagn Invest 1989;1:205-209. 25. Thornton RN, Gajadhar A, Evans J. Neosnora abortion epidemic in a dairy herd. N Z Vet J 1994;42:190-191. 26. Thurmond M, Hietala S. Strategies to control Neosnora infection in cattle, Bovine Pratt 1995;29:60-63.
Theriogenology
257
27.
Thurmond M, Hietala S. Effect of congenitally acquired Neosoora caninum infection on risk of abortion and subsequent abortions in dairy cattle. Am J Vet Res 1997;58:13811385. 28. Thurmond M, Hietala S, Blanchard PC. Herd-based diagnosis of Neosoora caninuminduced endemic and epidemic abortion in cows and evidence for congenital and postnatal transmission. J Vet Diagn Invest 1997;9:44-49. 29. Trautwein G. Immune mechanisms in the pathogenesis of viral disease: a review. Vet Microbial 1992;33: 19-34. 30. Uggla A, Stenlund S, Holmdahl OJM, Jakubek EB, Thebo P, Kindahl H, Bjorkman C. Oral Neosnora caninum inoculation of neonatal calves. Int J Parasitol 1998;28:14671472. 31. Venturini MC, Quiroga MA, Risso MA, Lorenzo CD, Omata Y, Venturini L, Godoy H. Mycotoxin T-2 and aflatoxin B 1 as immunosuppressors in mice chronically infected with Toxoplasma pondii. J Comp Path01 1996;115:229-237. 32. Wouda W, Bartels CJM, Moen AR. Characteristics of Neosuora caninum-associated abortion storms in dairy herds in The Netherlands (1995 to 1997). Theriogenology 1999 ; 52:233-245.
33.
34. 35. 36.
37.
Wouda W, Brinkhof J, Maanen C van, Gee ALW de, Moen AR. Serodiagnosis of neosporosis in individual cows and dairy herds, a comparison of three enzyme linked immunosorbent assays. Clin Diagn Lab Immunol 1998;5:711-716. Wouda W, lngh TSGAM van den, Knapen F van, Sluyter FJH, Koeman JP, Dubey JP. Neosnora abortus bij het rund in Nederland. Tijdschr Diergeneeskd 1992; 117:599-602. Wouda W, Moen AR, Schukken YH. Abortion risk in progeny of cows that experienced a Neosoora caninum abortion epidemic. Theriogenology 1998;49: 13 lo- 13 16. Wouda W, Moen AR, Visser IJR, Knapen F van. Bovine fetal neosporosis, a comparison of epizootic abortion cases and different age classes with regard to lesion severity and immunohistochemical identification of organisms in brain, heart and liver. J Vet Diagn Invest 1997;9: 180-l 85. Yaeger MJ, Shawd-Wessels SC, Leslie-Steen P. Neosnora abortion storm in a midwestem dairy. J Vet Diagn Invest 1994;6:506-508.
RISK FACTORS FOR Neosnora caninum-ASSOCIATED ABORTION STORMS IN DAIRY HERDS IN THE NETHERLANDS (1995 to 1997) C. J. M. Bartels,’ W. Wouda,zaand Y. H. Schukken’ ‘Department of Herd Health and Reproduction Utrecht University, The Netherlands ‘Animal Health Service Drachten, The Netherlands Received for publication: Accepted:
23
hly
1998
9 April
1999
ABSTRACT A 2 to 1 matched case control study design was used to analyze herd level risk factors for Neosoora caninum-associated abortion storms in 47 dairy herds. Data were obtained using a questionnaire regarding the state of affairs at the farms over the 2 years prior to the abortion storm. The questionnaire included 120 variables considered to be potential risk factors for either introduction of infection or recrudescence of chronic infection. The relationship between risk factors and case control pairs was analyzed by conditional logistic regression using a three-steps procedure. In addition, cross sectional serology was used to assess the possible role of concomitant infections. The main factors that were significant in the analysis and that were considered to have potential biological relevance were the presence of dogs, the presence of poultry, and the feeding of moldy maize-silage during summer. For both the presence of dogs and the presence of poultry on the farms, a linear relationship was found between the number of animals and the assessed risk for an abortion storm. These findings suggest a possible role of these species in the transmission of N. caninum. Further evidence for such a role of dogs was the significant association between the presence of dogs and the presence of seropositive cattle in the control herds. The feeding of moldy fodder is considered to be a factor which may induce recrudescence of a latent N. caninum-infection by mycotoxins causing immune suppression. We also found some evidence for a possible influence of management practices around calving and a high prevalence of retained afterbirths. No significant association was found for herd level prevalence of antibodies to bovine viral diarrhea virus, bovine herpesvirus 1, Lentosnira &$Q or Salmonella d&&t. 0 1999 by Elsevier Science Inc.
Key words: Neosnora caninum, bovine, abortion, risk factors, transmission Acknowledgments The authors thank the dairy farmers and their veterinary practitioners for participation; A. R. Moen and M. F. Weber for field work and useful suggestions. We are grateful to J. Part for contributing to the design of the questionnaire, to C. L. J. J. Kruitwagen for help with statistics, and to R.N. Zadoks for valuable comments on the manuscript. aCorrespondence and reprint requests: Animal Health Service, P. 0. Box, 9200 AJ Drachten, The Netherlands. Phone: 31 512 570 700 Fax: 31 512 520 013 E-mail: [email protected] Therlogendogy 52:247-257,1999 Q 1999 by Elsevter Science Inc.
CQQ3-691 XI99&-see front matter PII soo93-691x(9s)oo12s-o
Theriogenology
248 INTRODUCTION
Neosoora caninum is a protozoan parasite that has been associated with abortion in cattle. After the first description of bovine neosporosis-like abortions in 1989 (24), numerous reports have been published on both endemic and epidemic N. caninum abortions in cattle all over the world (for review see: 9). Vertical transmission from cow to calf has been well established (2, 4, 18, 35) and may contribute significantly to the persistence of the infection in the herd, causing an increased rate of abortion (5, 18, 27, 32, 35). However, evidence is accumulating that some form of postnatal transmission may be occurring since a high prevalence of N. caninuminfection has been found in dairy herds (11, 18, 32). The recently reported experimental evidence that dogs can act as definitive hosts of N. caninum (15) may have important implications for the epidemiology of bovine neosporosis. Several reports have described N. caninum-associated abortion storms which had an explosive character (14, 16, 25, 28, 37). Two hypotheses can be put forward to explain such abortion storms. The first hypothesis holds that the abortions are caused by the introduction of the infection into a herd by exposure to infectious oocysts. (3, 14, 16, 28, 37). The second hypothesis asserts that the abortions are caused by recrudescence of the infection in latently infected cows, by factors influencing immune suppression. Particularly viral infections and mycotoxicosis have been mentioned in this respect (26). Abortion storms associated with N. caninum have been recorded repeatedly in The Netherlands during recent years (16, 34, 36). Therefore, a case control study was begun to investigate the factors that might contribute to the occurrence of such abortion storms. In a companion paper we describe the characteristics of N_.caninum-associated abortion storms in 50 dairy herds, as monitored by fetal submissions to the laboratories of the Dutch Animal Health Service during 1995, 1996 and 1997 (32). From that study, it appeared that abortion storms occurred in herds which had a high prevalence of antibodies to s. caninum. Most abortion storms appeared to be induced by factors causing recrudescence of infection, rather than the immediate result of recent introduction. Hence our interest was in finding factors that might explain high seroprevalence and factors that might explain the occurrence of simultaneous abortions in these high seroprevalence herds. In the present study, we used a 2 to 1 matched case control study design to analyze dam from these herds that were obtained by a questionnaire. In addition, cross-sectional serology was used to assess the possible role of concomitant infections. MATERIALS
AND METHODS
Study Design A 2 to 1 matched case control design was used in this study. The sampling unit was a dairy herd. Cases were defined as dairy herds that had had an abortion storm associated with N. caninum during 1995, 1996 and 1997. An abortion storm was defined as a series of abortions affecting at least 12.5% (1 of 8) of the animals at risk within a period of 2 mo. Animals at risk were defined as the heifers and/or cows pregnant between 100 and 260 d (10,
Theriogenology
249
32). An association of an abortion storm with N. caninum was presumed when N. caninum infection was diagnosed in at least 2 fetuses (36), when no other cause of abortion was found in more than 1 fetus, and when antibodies to N. caninum were present in at least 80% of the aborting cows. In 4 herds the diagnosis of N. caninum infection was only made in 1 fetus, but 100% of the aborting cows were seropositive to N. caninum (32). An ELISA was used for detection of antibodies to N. caninurn as described previously (33). The population under study consisted of Dutch dairy herds in which the management would report the instance of multiple abortions to a veterinary practitioner for advice and diagnosis. Subsequently, the veterinary practitioner would encourage the submission of aborted fetuses to the Dutch Animal Health Service and take blood samples for diagnostic evaluation. Case herds were selected from the data base of fetal submissions to the laboratories of the Dutch Animal Health Service. Controls consisted of dairy herds without a history of an abortion storm between 1993 and 1997. Controls were matched with veterinary practitioners who were asked to provide a list of 6 client dairies with management practices similar to those of the case herd. It is therefore assumed that if a control herd had an abortion outbreak due to N_ canimun, it would have been classified as a case herd in our catchment scheme. Three dairies were randomly selected from each list. Two of these were asked to participate in the study while the third was kept in reserve in case a dairy farm had withdrawn from the study. Of the 50 case herds reported in a previous study (32), 3 herds with a cumulative incidence of abortions lower than 12.5% were excluded from our present study since they did not meet the originally defined conditions. In the remaining 47 herds 12.5 to 57% of the animals at risk had aborted within a period of 6 to 65 d. For a more detailed description of the abortion storms we refer to our companion paper (32). Data Collection All participating case and control farms were visited by one of the authors (CJMB) between January and November 1997. A questionnaire regarding the state of affairs at the farm over the past 2 yr prior to the abortion storm or prior to the visit (control herds) was administered on site. The questionnaire had been designed using expert consultation and literature review. Potential risk factors had been selected and assigned to clusters. A total of 120 variables as part of 10 clusters was used in the questionnaire and for further analysis. Seven clusters (A-G) were defined that contained variables thought possibly to be related to the introduction of or the horizontal transmission of infection with N. caninum while 3 clusters (H-J) contained variables that might be related to recrudescence of the infection. The defined clusters with the main variables are represented in Table 1. In addition, serology was used to assess the role of concomitant infections. Blood samples had been taken from 20% of the animals in the herds for detection of antibodies to N. caninum. Samples were taken from all age-groups (divided into 5 age groups: calves 0 to 1 yr, heifers 1 to 2 yr, Parity 1 cows, Parity 2 cows, Parity 3 and older cows) as reported previously (32). These sera were also used for detection of antibodies to bovine herpes virus 1 (BHVl), bovine viral diarrhea virus (BVDV), and Lentosnira hardio by ELISA methods and to Salmonella dublin by serum agglutination tests (0 and H antigen). To save costs, it was decided to confine the testing to 80% of the samples (37 of 47 matched cases and controls)
Theriogenology
250
unless the results indicated a complete analysis. A sample of 117 cows which had aborted during the abortion storms was analyzed for the serostatus to BVDV and N. caninum. Table 1. Defined clusters with the main variables used in the conditional logistic regression analysis of risk factors for Neosnora caninum-associated abortion storms
A B C D E F G H
Cluster Contacts with other animal species Contacts with man Contacts with cattle from other herds Nature and origin of fodder Source of drinking water Management around calving General management Stress inducing events during last month
I
Prevalence of other diseases
J
Nutritional abnormalities
Main variables Other farm animals, dogs, cats, rabbits, wild rodents, poultry, wild birds. Location of farmer’s house in relation to animal barns, use of artificial insemination. Communal pasture grounds, cattle shows, export sampling places, market places, purchase of animals. Nature of fodder for dairy cows, dry cows and heifers, respectively, turn-over of roughage, origin of fodder. Origin of drinking water by season, change of water supply. Use and hygiene of calving pen, use of calving pen to hospitalize sick animals, disposal of placenta, disposal of aborted fetuses. Housing system, pasturing management, origin of manure. Manipulations with animals (frequent palpation per rectum, herd vaccination, herd hoof trimming), events during previous month (climatic changes, unrest, prevalence of sick animals, frequent visits by veterinary practitioner) Infectious diseases (bovine viral diarrhea, infectious bovine rhinotracheitis, leptospirosis, salmonellosis, paratuberculosis), vaccination strategy, abomasal displacement, mastitis, retained afterbirth. Protein surplus in ration, fodder changes, mode of storage and conservation of roughage, use of moldy fodder, use of remnantfodder.
Statistical Analysis All variables were screened for unlikely values. The relationship between risk factors and case control pairs was analyzed by conditional logistic regression using Cox proportional hazards analysis (12, 19,23). The generalized linear model was expressed as: Ln (p/,.p) = a + pi Xi + s where “Ln (p/,.p)” stands for the log odds of abortion storm, and “a” for the log odds of abortion storm development for herds with “standard” (X=0) set of independent variable values, “lY’ measures the change in the log odds for a one-unit change in variable X and “a” represents the error term. Significant continuous variables were tested for a dose effect relation using categories, except for the variable “presence of dogs.” For this variable no categories were used since the range was only between 0 and 4.
251
Theriogenology
The model selection process consisted of 3 steps. In the first step, all variables were entered 1 by 1 into the analysis. Variables having a P-value of 0.25 or lower in the likelihood ratio test were included for further analysis. In the second step, variable selection was performed within each of the predefined clusters. Per cluster, a conditional logistic regression analysis with backward selection of variables was used, with the entry level at 0.05 and the removal level at 0.10. In the third step, the variables per cluster left were combined and again a backward stepwise selection was used with the same entry and removal level. Interaction terms were not analyzed in this study. The study size was estimated using a formula designed for case control studies with multiple controls per case (20). Defined values were needed for the level of significance (0.05), the power (>O.SO), the hypothesized relative risk associated with exposure value (RR=3) and for the relative frequency of exposure among controls (0.20). The outcome showed that with a sample size of 47 cases the power was more than 0.80, particularly when it was taken into account that this formula did not incorporate the matching as applied in our study. RESULTS Conditional
Logistic Regression Analysis
After the first step of analysis, 35 variables were significant and were used for further analysis. These variables represented 8 clusters. All variables within the clusters “source of drinking water” and “general management” were excluded from further analysis. After the second step of analysis by cluster, 15 variables remained significant. The variable “disposal of aborted fetuses” had to be. excluded from analysis in the final model because of its instability. The variable “frequent visits by the veterinary practitioner during the last month” was excluded from the final model because missing values severely affected the outcome. In the final modeling phase (Step 3) based on 13 variables, 5 variables were excluded because of nonsignificance. These variables were disposal of placenta (0 = in the manure pit, 1 = on the manure heap, 2 = otherwise); feeding of straw to dry cows during winter (dichotomous); feeding of brewer’s grain to dairy cows during winter (dichotomous); farmhouse connected to animal barn(s) (dichotomous); animals returned from market (dichotomous). The frequency distribution of the 8 variables that remained significant in the final model is represented in Table 2. The estimated coefficients for these variables are presented in Table 3. The final model containing 8 variables had a change in deviance from 103.27 to 36.41, with 9 degrees of freedom. The observed significance level was
of N. caninum Antibodies
in Cattle and Presence of Dogs
There was a significant association between the presence of dogs on control farms and the presence of N. caninum seropositive cattle on these farms. (OR = 4.2, P = 0.009). Eight of
Theriogenology
252
16 control farms where no dog was present had a zero N. caninum seroprevalence in the dairy stock. The other 8 herds without dogs had seroprevalences ranging from 4.8 to 37.5%, with a mean of 14.6%. The farm with the highest seroprevalence (37.5%) was closely neighboring a case farm, where 2 dogs were present. In addition, there was a weak but significant correlation between the number of dogs present on control farms and the seroprevalence of &I. caninum antibodies in the dairy stock (r = 0.23; P = 0.025). No such correlation was found in the case farms.
Table 2.
Frequency distribution of variables significant in the final step of conditional logistic regression analysis for assessment of risk factors for Neosnora caninum associated abortion storms.
Variable 1.
Presence of dogs
(No.1 (0) (1)
(2) (3) (4) (0) (l-10) P 10) pasture grounds for young
2.
Presence of poultry
3.
Use of communal stock Attendance at cattle shows during past 2 years Calving pen used to hospitalize sick animals Prevalence of retained afterbirths during previous year more than 10% Feeding of moldy maize-silage to dairy cows during summer Feeding of remnant-fodder to heifers during summer
4. 5. 6. 7. 8.
Case herds n=47 n (%I 1 (2.1) 24 (51.1) 17 (36.2) 5 (10.1) 0 (0) 12 (25.5) 20 (42.6) 15 (31.9)
Control herds n=94 W) 1: (17.0) 60 (63.8) 16 (17.0) 1 (1.1) :2 30 12
(E.:‘) (31.9) (12.8)
16
(34.0)
15
(15.9)
8 41
(17.0) (87.2)
30 70
(31.9) (74.5)
30
(63.8)
35
(37.2)
24
(51.1)
34
(36.2)
9
(19.1)
6
(6.4)
Serology No significant relationship was found between the seroprevalence on herd level of BVDV, BHVI, Lentosnira m and Salmonella dublin infections and the occurrence of abortion storms in the case herds. Analysis of serostatus to BVDV and N. caninum of 117 cows which had aborted during the abortion storm showed a negative relationship between seropositivity to BVDV and seropositivity to N. caninum (OR=O.42; P=O.O3). Seroprevalence of N. caninum antibodies was significantly higher in case herds compared to control herds as presented in a companion paper (32).
Theriogenology
253
Table 3. Results of conditional logistic regression analysis (backward stepwise method) on 47 case herds with an Neosuora caninum associated abortion storm and 94 matched control herds without such abortion storm. Significant variables
1. 2.
3. 4. 5. 6. 7. 8.
Presence of dogs Presence of pc&ry No of poultry l-10 Noofpoultry> 10 Use of communal pasture grounds for young stock Attendance at cattle shows during previous 2 years Calving pen used to hopitalize sick animals Prevalence of retained afterbirths in previous year > 10% Feeding of moldy maize-silage to dairy cows during summer Feeding of remnant-fodder to heifers during summer
pcoefficient
SE
P-value
Odds ratio
1.645
0.657
5.181
1.529 2.344
0.910 1.056
0.0123 0.0319 0.0928 0.0265
95% Confidence interval 11.43 - 18.8>
4.61 10.42
~0.78 - 27.5> cl.31 - 82.8>
2.176
1.131
0.0545
8.817
10.96 - 81.1>
-2.079
1.011
0.0398
0.125
~0.02 - 0.91>
3.538
1.406
0.0118
34.41
~2.19 - 541.1>
1.545
0.688
0.0248
4.691
cl.22 - 18.1>
1.884
0.821
0.0217
6.584
11.32 - 32.9>
3.684
1.316
0.0051
39.82
~3.02 - 525.3>
DISCUSSION This study was set up to explore herd level risk factors for N. caninum-associated abortion storms. The study was designed as a matched case control study in order to obtain optimal statistical efficiency. All case herds were managed by professional dairy farmers who had contacted their veterinary practitioner and subsequently had submitted fetuses for diagnostic evaluation. Control herds were chosen within the same veterinary practice in order to accommodate risk factors related to the veterinary practice that could not be measured. The questionnaire was composed to accommodate 2 hypothetical modes of origin of a N. caninum-associated abortion storm: 1) introduction of the infection into the herd and 2) recrudescence of the infection in chronically infected animals. When designing the questionnaire, we decided to include a large number of possible factors rather than to focus on a few factors in depth. Therefore, many variables were created and had to be screened separately. As a consequence, it cannot be excluded that some variables were significant by chance alone (significance level P = 0.05). However, in our opinion, the use of a three-steps analysis would have considerably limited the possibility of finding significant variables by chance in the final model. The interviews with the dairy farmers of case and control herds were held during a period of 11 mo in 1997. Seasonal differences may therefore have influenced the answers given, but this bias is considered nondifferential, since the interviews on case and matched control farms were conducted on the same day. However, there may have been some
254
Theriogenology
differential misclassification due to recall bias, as 12 of the abortion storms had occurred in 1995, 19 in 1996 and 16 in 1997. It is emphasized that a case control study does not determine causal relationships. This type of study is mainly used to generate hypotheses. The results from the conditional logistic regression analysis indicate statistical associations between a factor and an outcome, and asses the magnitude of the associated risk. Therefore, variables significant in our final model are not automatically definite risk factors for N. caninum-associated abortion storms. It is essential to assess the biological relevance of statistically significant variables. The presence of dogs on the farm was the most consistent factor in the model, and there was a significant linear relationship between the number of dogs and the assessed risk of an abortion storm. In addition, there was a significant association between the presence of dogs and the presence of seropositive cattle on the control farms. Similar observations have been reported in Canada (I 7). These results show a role of the dog in the transmission of the infection and they further corroborate the recent experimental evidence that dogs may shed N. caninum oocysts (15). Another significant variable was the presence of poultry on the premises. This included various species of domestic fowl like hens, ducks and geese. Again, a significant linear relationship was found between the number of animals and the assessed risk for an abortion storm. These animals could serve as vectors of oocysts, particularly when they are grubbing about freely. In addition, poultry might play a role as an intermediate host by which a dog may become infected (eating dead fowl). Two variables in the cluster “contacts with cattle from other herds” turned out to be significant in the final model. The variable “use of communal pasture grounds” was assessed as a risk for an abortion storm. Mostly this was related to young stock being pastured together with that from other herds on communal grounds away from the farm. It is doubtful that such animal contacts would lead to infection with N. --3 caninum as no evidence of lateral spread has been found by serology in a trial with 25 congenitally infected heifers and 25 noninfected heifers which were housed fed and managed together until calving (2). The other significant variable in this cluster was “attendance at cattle shows”, usually a presentation of young stock at a local fair. However, this variable showed a preventive effect in relation to the abortion storms, which is even more difficult to explain. The hygiene around calving seems to play a role as the variables “use of calving pen to hospitalize sick cows” and “high prevalence of retained afterbirth during previous year” both appeared in the final model. One could speculate that exposure to placenta and uterine effusions might serve as a source of infection as N. caninum tachyzoites have been demonstrated in the placenta (22,34). In a recent study, it was demonstrated that calves can be infected by oral ingestion of culture derived N. caninum tachyzoites added to colostrum (30). However, it is generally believed that tachyzoites will not survive for long in the environment, and a high infection dose would be needed, as has been demonstrated also for Toxonlasma nondii tachyzoites given orally to mice and cats (8). The last 2 significant variables “feeding of moldy maize-silage to dairy cows during the summer” and “feeding of remnant-fodder to heifers during summer” are both related to fodder of possibly inferior quality. Particularly, contamination of fodder by fungi is
Theriogenology
255
considered to be a potential risk factor, because mycotoxins may have been produced. Several mycotoxins have been shown to cause immunosuppression after repeated ingestion at low doses (7, 21). Such immunosuppression may lead to tissue cyst rupture, as has been demonstrated in mice with chronic toxoplasmosis (3 1). An explanation for the seasonal pattern of N. caninum-associated abortion storms, as reported previously (32), could lay in the favorable conditions of temperature and humidity for fungal growth during the summer months. On the other hand, similar conditions of temperature and humidity are probably also favorable for the sporulation of oocysts (13). We did not find any relationship between N. caninum-associated abortion storms and infection by BVDV, BI-IVI, L. hardio and S. dublin, based on the seroprevalence data at herd level for these infections. In particular, BVDV has been suspected to play a role in this respect (26), because infection with BVDV is known to cause immunosuppression (29). However, we found a negative relationship between seropositivity to BVDV and seropositivity to N_. caninum in aborting cows. In addition, we did not find dual infections with other abortifacients in fetuses submitted during the abortion storms (32) as has been reported previously (1). Various factors causing immune suppression may have been involved without being significant in the analysis. Several dairy farmers of case herds had their own explanations for the abortion storm, varying from a period of extremely high temperatures to vaccination of the herd prior to the abortion storm. Also mentioned repeatedly was a disturbed balance between energy and protein supply, due to low quality grass at the end of the pasturing season. However, immune suppression can not be easily assessed, and, therefore, its possible significance as a risk factor for the clinical manifestation of bovine neosporosis can not be demonstrated by statistical methods. The results of this study and of a previous study (32) support the idea that both vertical and postnatal transmission of N. caninum play a role in the epidemiology of the infection in cattle. The vertical transmission route has been well established previously (2,4, 18, 35). The present study provides further evidence that,in particular,dogs play a role in the transmission of N. caninum infection to cattle. REFERENCES 1. Agerholm JS, Willadsen CM, Nielsen TK, Giese SB, Holm E, Jensen L, Agger, JF. Diagnostic studies of abortion in Danish dairy herds. J Vet Med (A) 1997;44:551-558. 2. Anderson ML, Reynolds JP, Rowe JD, Sverlow KW, Packham AE, Barr BC, Conrad PA. Evidence of vertical transmission of Neosoora sp infection in dairy cattle. J Am Vet Med Assoc 1997;210:1169-1172. 3. Barr BC, Bjerkas I, Buxton D, Conrad PA, Dubey JP, Ellis JT, Jenkins MC, Johnston SA, Lindsay DS, Sibley D, Trees AJ, Wouda W. Neosporosis - Report of the International Neospora Workshop. Compend Contin Educ Pratt Vet 1997; 19:s 120-5 126,5 144. 4. Barr BC, Conrad PA, Breitmeyer R, Sverlow K, Anderson ML, Reynolds J, Chauvet AE, Dubey JP, Ardans AA. Congenital Neosnora infection in calves born from cows that had previously aborted Neospora-infected fetuses: four cases (1990-1992). J Am Vet Med Assoc 1993;202:113-117.
256
Theriogenology
5. Bjiirkman C, Johansson 0, Stenlund S, Holmdahl JM, Uggla A. Neosuora species infection in a herd of dairy cattle. J Am Vet Med Assoc 1996;208: 144 l-l 444. 6. Breslow NE, Day NE. Statistical methods in cancer research, Vol 1. The Analysis of Case-Control Studies. Lyon: IARC Sci Publ, 1980; 248-279. mechanisms of immunosuppresstion. Vet Immunol 7. Corrier DE. Mycotoxines: Immunopathol 1991;30:73-87. 8. Dubey JP. Re-examination of resistance of Toxonlasma gondii tachyzoites and bradyzoites to pepsin and trypsin digestion. Parasitology 1998;116:43-50. 9. Dubey JP, Lindsay DS. A review of Neospora caninum and neosporosis. Vet Parasitol 1996;67:1-59. 10. Fetrow J, McClary D, Harman R, Butcher K, Weaver L, Studer E, Ehrlich J, Etherington W, Guterbock W, Klinborg D, Reneau J, Williamson N. Calculating selected reproductive indices: recommendations of the American Association of Bovine Practitioners. J Dairy Sci 1990;73:778-90. 11. French NP, Davison HC, Clancy D, Begon M, Trees AJ. Modelling of Neosnora species infection in dairy cattle: the importance of horizontal and vertical transmission and differential culling. Thrusfield MV, Goodall EA (eds), Proc Sot Vet Epidemiol Prev Med 1998;113-122. 12. Hosmer DW, Lemeshow S. AppliedLogisticRegression. New York: John Wiley and Sons, Inq1989; 187-215. 13. Lindsay DS, Current WL, Ernst. Sporogony of Isosnora suis Biester, 1934 of swine. J Parasitol 1982;68:861-865. 14. McAllister MM, Huffman EM, Hietala SK, Conrad PA, Anderson ML, Salman MD. Evidence suggesting a point source exposure in an outbreak of bovine abortion due to neosporosis. J Vet Diagn Invest 1996; 8:355-357. 15. McAllister MM, Dubey JP, Lindsay DS, Jolley WR, Wills RA, McGuire AM. Dogs are definitive hosts of Neosuora caninum Int J Parasitol 1998;28:1473-1478. 16. Moen AR, Wouda W, Mu1 GGraat EAM, Werven T van. Increased risk of abortion following Neosuora caninum abortion outbreaks: a retrospective and prospective cohort study in four dairy herds. Theriogenology 1998; 49: 1301-l 309. 17. Pare J, Fecteau G, Fortin M, Marsolais G. Seroepidemiologic study of Neosnora canimnn in dairy herds. J Am Vet Med Assoc 1998;213:1595-1598. 18. Pare J, Thurmond MC, Hietala SK. Congenital Neosnora caninum infection in dairy cattle and associated calthood mortality. Can J Vet Res 1996;60: 133-l 39. 19. Rothman KJ. Modem Epidemiology. Boston/Toronto: Little, Brown and Co, 1986; 297298. 20. Schlesselman JJ. Case-Control Studies, Design, Conduct, Analysis. Oxford: Oxford University Press, 1982; Chapter 6: 150-152. 21. Sharma RP. Immunotoxicity of mycotoxins. J Dairy Sci 1993;78:892-897. 22. Shivaprasad HL, Ely R and Dubey JP. A Neoswra-like protozoon in an aborted bovine placenta. Vet Parasitol 1989; 34: 145-8. 23. SPSS Advanced Statistics 6.1 software package. Users Guide. Chicago 1994; 291-328. 24. Thilsted JP, Dubey JP. Neosporosis-like abortions in a herd of dairy cattle. J Vet Diagn Invest 1989;1:205-209. 25. Thornton RN, Gajadhar A, Evans J. Neosnora abortion epidemic in a dairy herd. N Z Vet J 1994;42:190-191. 26. Thurmond M, Hietala S. Strategies to control Neosnora infection in cattle, Bovine Pratt 1995;29:60-63.
Theriogenology
257
27.
Thurmond M, Hietala S. Effect of congenitally acquired Neosoora caninum infection on risk of abortion and subsequent abortions in dairy cattle. Am J Vet Res 1997;58:13811385. 28. Thurmond M, Hietala S, Blanchard PC. Herd-based diagnosis of Neosoora caninuminduced endemic and epidemic abortion in cows and evidence for congenital and postnatal transmission. J Vet Diagn Invest 1997;9:44-49. 29. Trautwein G. Immune mechanisms in the pathogenesis of viral disease: a review. Vet Microbial 1992;33: 19-34. 30. Uggla A, Stenlund S, Holmdahl OJM, Jakubek EB, Thebo P, Kindahl H, Bjorkman C. Oral Neosnora caninum inoculation of neonatal calves. Int J Parasitol 1998;28:14671472. 31. Venturini MC, Quiroga MA, Risso MA, Lorenzo CD, Omata Y, Venturini L, Godoy H. Mycotoxin T-2 and aflatoxin B 1 as immunosuppressors in mice chronically infected with Toxoplasma pondii. J Comp Path01 1996;115:229-237. 32. Wouda W, Bartels CJM, Moen AR. Characteristics of Neosuora caninum-associated abortion storms in dairy herds in The Netherlands (1995 to 1997). Theriogenology 1999 ; 52:233-245.
33.
34. 35. 36.
37.
Wouda W, Brinkhof J, Maanen C van, Gee ALW de, Moen AR. Serodiagnosis of neosporosis in individual cows and dairy herds, a comparison of three enzyme linked immunosorbent assays. Clin Diagn Lab Immunol 1998;5:711-716. Wouda W, lngh TSGAM van den, Knapen F van, Sluyter FJH, Koeman JP, Dubey JP. Neosnora abortus bij het rund in Nederland. Tijdschr Diergeneeskd 1992; 117:599-602. Wouda W, Moen AR, Schukken YH. Abortion risk in progeny of cows that experienced a Neosoora caninum abortion epidemic. Theriogenology 1998;49: 13 lo- 13 16. Wouda W, Moen AR, Visser IJR, Knapen F van. Bovine fetal neosporosis, a comparison of epizootic abortion cases and different age classes with regard to lesion severity and immunohistochemical identification of organisms in brain, heart and liver. J Vet Diagn Invest 1997;9: 180-l 85. Yaeger MJ, Shawd-Wessels SC, Leslie-Steen P. Neosnora abortion storm in a midwestem dairy. J Vet Diagn Invest 1994;6:506-508.