- Email: [email protected]
Risk factors for epidemic respiratory disease in Norwegian cattle herds
Preventive Veterinary Medicine 44 (2000) 87±96
Risk factors for epidemic respiratory disease in Norwegian cattle herds Madelaine NorstroÈma,*, Eystein Skjerveb, Jorun Jarpa a
b
National Veterinary Institute, PO Box 8156 Dep., N-0033 Oslo, Norway The Norwegian School of Veterinary Science, PO Box 8146 Dep., N-0033 Oslo, Norway Received 31 March 1999; accepted 11 November 1999
Abstract An epidemic of acute respiratory disease associated with bovine respiratory syncytial virus (BRSV) occurred during the winter and spring of 1995 in two neighbouring veterinary districts in the south-eastern part of Norway. The objective of this study was to describe the time course of the outbreak associated with BRSV in the cattle herds, and to determine the association between selected herd factors and the risk of experiencing a herd outbreak of acute respiratory disease. Data from 431 cattle herds on the dates of disease occurrence, location of the farms, herd size, age profile and production type were collected retrospectively for 1995. The risk of acute respiratory disease occurring in a cattle herd was related to the herd size as well as the type of production, with an expressed interaction between these two variables. From the Cox proportionalhazards model, the risk of a herd outbreak in a mixed herd of 20 animals was estimated to be 1.7times greater than in a dairy herd and 3.3-times greater than a beef herd (reference category) of a comparable size. On increasing the herd size to 50 animals, the risk increased 1.3-fold for a mixed herd, 3.3-fold for a dairy herd, and 2.1-fold for a beef herd, compared to the risk for a corresponding type of herd of 20 animals. # 2000 Elsevier Science B.V. All rights reserved. Keywords: Acute respiratory disease; Bovine respiratory syncytial virus; Dairy cattle; Norway
1. Introduction Respiratory diseases represent a major disease problem in young cattle (especially in intensive beef operations) and cause considerable economic losses world wide. Acute *
Corresponding author. Tel.: 47-22597428; fax: 47-22565966. E-mail address: [email protected] (M. NorstroÈm). 0167-5877/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 8 7 7 ( 9 9 ) 0 0 1 1 3 - 0
88
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
respiratory diseases of cattle may be caused by various viral and bacterial pathogens. Many epidemics of acute respiratory disease in young cattle are associated with infection with bovine respiratory syncytial virus (BRSV) (Stott et al., 1980; Verhoeff and van Nieuwstadt, 1984). When farmed cattle located in an area with a previously non-exposed cattle population are infected with BRSV, the infection often spreads very rapidly from farm to farm. This may lead to the disease becoming endemic and affecting the same herds almost every year (Wellemans, 1990). In western Europe (where BRSV is considered to be endemic), outbreaks occur annually during the cold and wet months of the year; these outbreaks affect calves <6 months old (Bryson et al., 1978; Ploeger et al., 1986). Nevertheless, outbreaks of acute respiratory disease associated with BRSV infection can occur in older cattle if the virus is introduced into a previously non-exposed population (Inaba et al., 1972; édegaard and Krogsrud, 1977; Jacobsson et al., 1989; Elvander, 1996). Reports from BRSV outbreaks suggest that the virus can be transmitted directly by movement of infected animals (Wellemans, 1990; Elvander, 1996), although the possibility of transmission by man (édegaard and Krogsrud, 1977; Wellemans, 1990; Van der Poel et al., 1993) and airborne transmission (Elvander, 1996) cannot be excluded. Increasing herd size increases the risk for several infectious diseases in cattle (Toews et al., 1986; Frank and Kaneene, 1992; Van Donkersgoed et al., 1993) and other species (StaÈrk et al., 1992). The severity of infectious respiratory disease is related to the age distribution in the herds (Verhoeff and van Nieuwstadt, 1984); the probability of occurrence of a herd outbreak is higher in herds with a large proportion of calves and young cattle. (The proportion of susceptible individuals could be assumed to be greater in such herds, the infection thereby being able to establish itself more easily.) Since the first case of acute respiratory disease related to BRSV infection was diagnosed in Norwegian cattle in 1976 (édegaard and Krogsrud, 1977), infected animals have been diagnosed only occasionally (Krogsrud, personal communication). In a serological study (Hyllseth et al., 1987) of 250 Norwegian young calves, among various viruses related to respiratory disease, the overall prevalence of specific antibodies was highest for parainfluenza-3 virus and bovine rotavirus Ð whereas specific antibodies against BRSV were less common. A pronounced geographical variation in the seroprevalence was found (varying from 2 to 50% between the regions investigated). An outbreak of acute respiratory disease associated with BRSV infections involving many cattle herds occurred in Norway in 1995. It was assumed that the infection was introduced in December 1994 with a consignment of beef cattle imported from Denmark (where BRSV was associated with enzootic pneumonia of calves) (Uttenthal et al., 1996), and where the virus is now thought to be endemic (Alban et al., 1999). The primary cases appeared about a week after the imported cattle had been introduced into the herd; the disease then spread rapidly between farms. The affected cattle in the primary case herds seroconverted to BRSV (paired ELISA tests). All the imported animals were tested and found to be positive for antibodies against BRSV Ð some of them with high titres (Krogsrud, personal communication). This outbreak was considered to have had a great economic impact on the affected farms, and it would therefore be of great interest to identify risk factors for the occurrence of acute respiratory disease to obtain possible
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
89
measures to prevent transmission of the disease and to minimise the economic losses associated with any future epidemics. As far as we know, no other studies of risk factors in herd outbreaks of acute respiratory disease associated with BRSV in cattle have been conducted. The aims of the present study were to investigate the association between selected herd factors and the risk of such acute respiratory disease at the herd level. 2. Material and methods 2.1. Study area and study population Norwegian counties are divided into veterinary districts which may include one or more municipalities. The veterinary districts in this study were within the same veterinary region, administered by one Regional Veterinary Officer (RVO). The study area encompassed two adjacent veterinary districts located in the south-east of Norway where a number of cattle herds were placed under restrictions for acute respiratory disease during the observation period. The consequence of a restriction was that all movement of cattle from and to the restricted herds were prohibited. Data regarding the dates on which herd restrictions had been imposed were collected from the RVO. Pursuant to the Animal Disease Act, the RVO had decided that restrictions should be placed on affected herds during the outbreak in order to prevent further spread of the disease. The District Veterinary Officer (DVO) in each district reported the affected herds and also took decisions regarding introduction and lifting of restrictions. This decision was primarily based upon clinical observations of acute respiratory disease among the cattle in the herds, and on the knowledge that the first herds affected had confirmed diagnoses of BRSV infection. The veterinary visits to the affected farms were voluntary initiated by farmers recognising the disease problem in their cattle herds. All cattle herds in the study area which were registered in the Register of Production Subsidies (RPS) as of 1 January 1995, were included in the study. The RPS contains records of all farmers who apply for production subsidies from the Ministry of Agriculture. The figures recorded in the RPS is used for the official Norwegian Statistics of Agriculture and is estimated to be the census population of herds. The registry contains information on the number of animals of each species and age category present on the farm on the date of application. The age categories for cattle in the registry are: dairy cows, suckling cows, bulls older than 12 months, heifers older than 12 months, male calves younger than 12 months and female calves younger than 12 months. The registry is updated twice a year on 1 January and 1 July. Comparison of the herd sizes given on these two dates showed that they remained relatively constant, and therefore only the information recorded on 1 January 1995 was used in the analyses. 2.2. Observation period The study period was defined as beginning 7 days before restrictions were first applied until the day when all restrictions had been lifted in the area and lasted from 1 January to 31 May.
90
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
2.3. Design The study was a retrospective cohort study with the herd as the observational unit. The outcome event was the imposition of restrictions due to acute respiratory disease. 2.4. Statistical analysis The statistical analyses were performed with the SAS-PC System1 Version 6.12 for Windows (SAS Institute, Cary, NC, USA 1996). The explanatory variables examined in the study were: veterinary district, production type, age profile and herd size. Veterinary district was treated as a categorical variable with two levels. The production type was defined and classified as a categorical variable with the following categories: Dairy herds: Cattle herds with milking cows, calves, heifers older than 12 months, not more than one bull older than 12 months, and no suckling cows. Beef herds: Cattle herds with no milking cows. Mixed herds: Cattle herds with milking cows and more than one bull older than 12 months of age or suckling cows. The age profile was included in the analysis as three separate variables; calf, young and adult, and based on the proportions of each age category in the herd relative to the total number of cattle. The denominators of the variables were defined as follows: calf; calves <1 year, young; heifers and bulls (1±2 years) and adults; parous cows. The herd size was estimated as the sum of all bovine animals in each category present in the herd on the date of the application for subsidies. The survival time was defined as the period from 7 days before application of the first restrictions in each of the two veterinary districts, or until the end of the study period for the herds never diagnosed as infected. The assumption of proportional hazards between the levels of all independent variables was evaluated by the log±log survival plot of the stratified continuous variables (quartiles), or of the various categories of qualitative variables. After introductory descriptive analyses, all the independent variables were available for inclusion in a Cox proportional-hazards model (Cox, 1972). Independent variables with unconditional p-values <0.25 were offered to the multivariate Cox model. Each explanatory variable was forced into the model in a forward stepwiseselection procedure. To determine the best-fitting model, a series of nested hazards models with and without 2-way interaction terms of all the included variables were fitted. The proportionality assumption was evaluated individually for each explanatory variable by including in the model a time-dependent covariate representing the interaction of the explanatory variable with the survival time. A 2-tailed likelihoodratio statistic test (a<0.05) was used to compare all the various models based on a w2-distribution. A goodness-of-fit test developed by Schoenfeld (1982) was used to evaluate the fit of the final multivariate model.
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
91
Table 1 The mean proportion of animals within various age categories in relation to the production types of the 431 cattle herds included in the study of risk factors of acute respiratory disease in cattle during 1 January to 31 May of 1995 in two veterinary districts located in the south-east of Norway Production type
Mixed Dairy Beef
Age category (95% CI) Calf
Young
Adult
0.34 (0.33±0.35) 0.35 (0.33±0.37) 0.38 (0.33±0.42)
0.34 (0.32±0.35) 0.23 (0.21±0.25) 0.45 (0.39±0.50)
0.32 (0.31±0.33) 0.42 (0.40±0.44) 0.18 (0.14±0.22)
3. Results The study population consisted of 431 cattle herds (206 and 225 in Hadeland and Toten veterinary districts, respectively). The distribution of the herds according to production type was: 118 dairy, 107 beef, and 206 mixed herds. The median herd size of the study population was 37 animals (1±219), differing between herds of various production type as follows: dairy: 26 (2±130), beef: 18 (1±203), mixed; 48 (7±219). The mean proportions of the different age categories in relation to production type are shown in Table 1. The cumulative herd-level incidence of acute respiratory disease was 35.0% (95% CI: 26.0±43.5) in dairy herds, 19.6% (95% CI: 12.0± 27.3) in beef herds, and 47.0% (95% CI: 40.2±54.0) in mixed herds. The survival times of the herds varied in relation to herd size (Fig. 1) and production type (Fig. 2). Neither the log±log survival plots nor the goodness of fit test revealed
Fig. 1. Kaplan±Meier curves for the occurrence of acute respiratory disease in relation to herd size categories (quartiles) for 431 cattle herds located in two veterinary districts in the south-east of Norway during the study period of 1 January to 31 May in 1995.
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
92
Fig. 2. Kaplan±Meier curves for the occurrence of acute respiratory disease in relation to production type categories for 431 cattle herds located in two veterinary districts in the south-east of Norway during the study period of 1 January to 31 May in 1995.
any violations of the assumption of proportional hazards for the various explanatory variables, and no time-dependency could be detected for any of the explanatory variables. A final multivariable model (Table 2) contained the terms for production type and herd size and their interaction. For example, a dairy herd with 20 animals had a hazard ratio of 3 whereas the hazard ratio for a mixed herd or a beef herd of a corresponding size was 5 and 1.5, respectively. Increasing the herd size to 50 head increased the hazard ratio for a dairy herd to 8, for a mixed herd to 6 and for a beef herd to 3 (Fig. 3).
Table 2 The final multivariable Cox proportional-hazard model for the risk of an outbreak of acute respiratory disease in 431 Norwegian cattle herds during the study period (1 January to 31 May of 1995) Variable
Levels
Production category
Mixed Dairy Beefa
Herd size
±
Production typeherd size
Mixedherd size Dairyherd size Beefherd sizea
a
Reference level.
b
S.E.
P
HR
95% CI
1.49 0.50 0
0.37 0.44 ±
0.001 0.26 ±
4.43 1.65
2.14, 9.13 0.70, 3.93 ±
0.02
0.001
0.001
1.02
1.01, 1.03
ÿ0.013 0.01 0
0.004 0.008 ±
0.004 0.21
0.99 1.01 1
0.98, 1.00 0.99, 1.03 ±
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
93
Fig. 3. Estimated risk based on the final Cox proportional-hazard model of a herd outbreak of acute respiratory disease for each production type in relation to herd size for 431 cattle herds located in two veterinary districts in the south-east of Norway during the study period of 1 January to 31 May in 1995.
4. Discussion Because BRSV infection was demonstrated in affected cattle in the primary case herds, BRSV was most probably the infectious agent associated with the outbreak. Although respiratory diseases are usually encountered as sporadic cases in single animals in the Norwegian cattle population and no specific respiratory virus occurs endemically in Norway, other respiratory viruses might nevertheless have been involved. The risk of acute respiratory disease in herds varied significantly with type of production, although the biological exposure factors associated with this factor are unclear. Veterinarians and animal handlers have been considered to play a role in the transmission of BRSV (Wellemans, 1990). Dairy and mixed herds are regularly visited by a number of people such as milk-tanker drivers, veterinarians, AI-technicians and relief milkers, whereas beef herds are more isolated. The patterns of purchase and human traffic thus might differ between herds according to type of production, and therefore possibly explain the increased risk found in this study in mixed and dairy herds compared to the beef herds. (This outbreak occurred during winter and spring months when all herds were indoors with little risk of direct cow-to-cow transmission between herds.) The risk of acute respiratory disease was highest in the larger herds. The increasing risk related to increase in herd size could be explained by greater indirect contact with other herds due to more frequent human traffic. This is in accordance with findings from other
94
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
studies of infectious diseases (Toews et al., 1986; Frank and Kaneene, 1992; StaÈrk et al., 1992; Van Donkersgoed et al., 1993) and the theory of De Jong et al. (1995). Moreover, any infectious agent will establish itself more easily in a large herd because of the greater degree of animal-to-animal contact. However, the increase in risk associated with increase in herd size was more marked in dairy herds than in the other types of herd. An increase of the size of a dairy herd would probably result in a greater increase in human traffic (such as veterinarians or AI-technicians) than would be the case for a beef or mixed herd. There is obviously a larger proportion of dairy cows in dairy herds than in the mixed herds, and any increase in size in dairy herds will thus lead to a corresponding increase in the number of dairy cows while any increase in size in mixed herds could just as well be related to a higher proportion of beef cattle in the herd. However, our study included only a few large dairy herds (the mixed herds being generally larger). This might explain most of the differences found between the dairy and mixed herds. A further contributory factor could be that in larger dairy herds there is a greater proportion of high-yielding cows such animals having been reported to be more susceptible to infection by BRSV (Elvander, 1996). No relationship was demonstrated between the age profile of the animals in the herds and the risk of acute respiratory disease. A possible explanation could be that all the cattle herds were equally susceptible to infection because no recent outbreaks associated with BRSV had occurred in the area. The variability of the age distribution in the study herds was small, which might be a more-obvious reason for not finding any association. After the primary herd outbreaks in each district, the hazard of outbreaks of acute respiratory disease did not differ between the veterinary districts. This similar progression of the epidemic in the two areas supports the assumption that the epidemic was caused by the same infectious agent. Interpretation of the results of the present study is confounded by lack of data on the incidence of disease in the herds under no restrictions (either due to lack of reporting or to failure to diagnose disease in these herds). However, this lack of sensitivity is likely to be of a conservative type (reducing the power of the study). Different management practices in the different types of herd could have influenced the extent to which herd outbreaks were reported, and thereby also the apparent incidence and, in turn, the results of the study. Moreover, dairy farmers with milk production may be more motivated to contact a veterinarian in the event of a disease outbreak than beef farmers. If so, the hazard ratio would be underestimated for the beef herds and overestimated for the other types of herd. In the case definition, the time for censoring was set equal to the day of notification due to acute respiratory disease. It is to be expected that the onset of acute respiratory disease was days or even up to a week before the defined day of onset of acute respiratory disease. We assumed that the notification percentage was the same over the whole observation period and for all categories of herd. The study material consisted of the census population of the herds in the area registered in the RPS but this registry only includes active applicants. Further analyses have to be performed to determine and evaluate other factors that could possibly explain these results. Neither the effect of neighbourship between the herds, nor the population density in the areas concerned, were considered in the present analysis.
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
95
5. Conclusion In this study of herd outbreaks of acute respiratory disease in Norwegian cattle herds thought to be associated with BRSV, the risk of such outbreaks was associated with production type and herd size. These results suggest that there are underlying risk factors related to the production type and herd size that could further explain the dynamics and transmission routes of the causative agents, presumable BRSV, between cattle herds. Acknowledgements This study was financially supported by the Norwegian Research Council. The authors are grateful to the Norwegian Animal Health Authority for providing the data used in the analyses. References Alban, L., Larsen, L.E., ChrieÂl, M., Tegtmeier, C., Nielsen, T.K., 1999. The occurrence of clinical outbreaks of enzootic pneumonia in calves in 10 Danish dairy herds during the winter 1996±97: Descriptive Results, Proceedings of a meeting held at the University of Bristol, 24±26 March 1999. Society for Veterinary Epidemiology and Preventive Medicine, pp. 118±130. Bryson, D.G., Mc Ferran, J.B., Ball, H.J., Neill, S.D., 1978. Observations on outbreaks of respiratory disease in housed calves. Epidemiological and clinical findings. Vet. Rec. 103, 485±489. Cox, D.R., 1972. Regression models and life tables. J. R. Stat. Soc. B 74, 187±220. De Jong, M.C.M., Diekmann, O., Heesterbeek, J.A.P., 1995. How does transmission depend on population size? In: Mollison, D. (Ed.), Epidemic Models: Their Structure and Relation to Data. Cambridge University Press, Cambridge, pp. 84±94. Elvander, M., 1996. A study of bovine respiratory syncytial virus infections in Swedish cattle. Ph.D. Thesis, 124 pp. Frank, N.A., Kaneene, J.B., 1992. Management risk factors associated with calf diarrhoea in Michigan dairy herds. J. Dairy Sci. 76, 1313±1323. Hyllseth, B., Larsen, H.J., Norheim, K., 1987. Calf virus infections in Norway, a serosurvey. Seventh International Congress of Virology, Edmonton, Canada, p. 321. Inaba, Y., Tanaka, Y., Omori, T., Matumoto, M., 1972. Bovine respiratory syncytial virus: studies on an outbreak in Japan. Jpn. J. Microbiol. 16, 373±383. Jacobsson, S.O., Alenius, S., Ottander, G., PalmeÂr, H., Persson, J., Wennberg, K., Juntti, N., 1989. Bovint respiratoriskt syncytialt virus, BRSV som orsak til pneumonie hos kor. Sven. Vet. Tidn. 41, 641±647 (in Swedish). édegaard, é.A., Krogsrud, J., 1977. A field outbreak caused by bovine respiratory syncytial virus. Acta Vet. Scand. 18, 429±431. Ploeger, H.W., Boon, J.H., Klaassen, C.H.L., van Florent, G., 1986. A sero-epidemiological study of infections with bovine respiratory syncytial virus in first season grazing calves. J. Vet. Med. 33, 311±318. Schoenfeld, D., 1982. Partial residuals for the proportional hazards model. Biometrika 69, 51±55. StaÈrk, K.D.C., Keller, H., Eggenberger, E., 1992. Risk factors for the reinfection of specific pathogen-free pig breeding herds with enzootic pneumonia. Vet. Rec. 131, 532±535. Stott, E.J., Thomas, L.H., Collins, A.P., Crough, S., Jebbett, J., Smith, G., Luther, P.D., Caswell, R., 1980. A survey of virus infections of the respiratory tract of cattle and their association with disease. J. Hyg. 85, 237±247. Toews, D.W., Martin, S.W., Meek, A.H., 1986. Dairy calf management, morbidity in Ontario Holstein herds. Prev. Vet. Med. 4, 159±171.
96
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
Uttenthal, A., Jensen, N.P.B., Blom, J.Y., 1996. Viral ñtiology of enzootic pneumonia in Danish dairy herds: diagnostic tools and epidemiology. Vet. Rec. 139, 114±117. Van der Poel, W.H.M., Brand, A., Kramps, J.A., Middel, W.G.J., Van Oirschot, J.T., 1993. Dynamics of bovine respiratory syncytial virus: a longitudinal epidemiological study in dairy herds. Arch. Virol. 133, 309±321. Van Donkersgoed, J., Ribble, C.S., Boyer, L.G., Townsend, H.G.G., 1993. Epidemiological study of enzootic pneumonia in dairy calves in Saskatchewan. Can. J. Vet. Res. 57, 247±254. Verhoeff, J., van Nieuwstadt, A.P.K.M.I., 1984. BRS virus, PI3 virus, and BHV1 infections of young stock on self containing dairy-farms: epidemiological and clinical findings. Vet. Rec. 114, 288±293. Wellemans, G., 1990. Bovine respiratory syncytial virus. In: Dinter, Z., Morein, B. (Eds.), Virus Infections of Ruminants. Elsevier, Amsterdam, pp. 363±375.
Risk factors for epidemic respiratory disease in Norwegian cattle herds Madelaine NorstroÈma,*, Eystein Skjerveb, Jorun Jarpa a
b
National Veterinary Institute, PO Box 8156 Dep., N-0033 Oslo, Norway The Norwegian School of Veterinary Science, PO Box 8146 Dep., N-0033 Oslo, Norway Received 31 March 1999; accepted 11 November 1999
Abstract An epidemic of acute respiratory disease associated with bovine respiratory syncytial virus (BRSV) occurred during the winter and spring of 1995 in two neighbouring veterinary districts in the south-eastern part of Norway. The objective of this study was to describe the time course of the outbreak associated with BRSV in the cattle herds, and to determine the association between selected herd factors and the risk of experiencing a herd outbreak of acute respiratory disease. Data from 431 cattle herds on the dates of disease occurrence, location of the farms, herd size, age profile and production type were collected retrospectively for 1995. The risk of acute respiratory disease occurring in a cattle herd was related to the herd size as well as the type of production, with an expressed interaction between these two variables. From the Cox proportionalhazards model, the risk of a herd outbreak in a mixed herd of 20 animals was estimated to be 1.7times greater than in a dairy herd and 3.3-times greater than a beef herd (reference category) of a comparable size. On increasing the herd size to 50 animals, the risk increased 1.3-fold for a mixed herd, 3.3-fold for a dairy herd, and 2.1-fold for a beef herd, compared to the risk for a corresponding type of herd of 20 animals. # 2000 Elsevier Science B.V. All rights reserved. Keywords: Acute respiratory disease; Bovine respiratory syncytial virus; Dairy cattle; Norway
1. Introduction Respiratory diseases represent a major disease problem in young cattle (especially in intensive beef operations) and cause considerable economic losses world wide. Acute *
Corresponding author. Tel.: 47-22597428; fax: 47-22565966. E-mail address: [email protected] (M. NorstroÈm). 0167-5877/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 7 - 5 8 7 7 ( 9 9 ) 0 0 1 1 3 - 0
88
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
respiratory diseases of cattle may be caused by various viral and bacterial pathogens. Many epidemics of acute respiratory disease in young cattle are associated with infection with bovine respiratory syncytial virus (BRSV) (Stott et al., 1980; Verhoeff and van Nieuwstadt, 1984). When farmed cattle located in an area with a previously non-exposed cattle population are infected with BRSV, the infection often spreads very rapidly from farm to farm. This may lead to the disease becoming endemic and affecting the same herds almost every year (Wellemans, 1990). In western Europe (where BRSV is considered to be endemic), outbreaks occur annually during the cold and wet months of the year; these outbreaks affect calves <6 months old (Bryson et al., 1978; Ploeger et al., 1986). Nevertheless, outbreaks of acute respiratory disease associated with BRSV infection can occur in older cattle if the virus is introduced into a previously non-exposed population (Inaba et al., 1972; édegaard and Krogsrud, 1977; Jacobsson et al., 1989; Elvander, 1996). Reports from BRSV outbreaks suggest that the virus can be transmitted directly by movement of infected animals (Wellemans, 1990; Elvander, 1996), although the possibility of transmission by man (édegaard and Krogsrud, 1977; Wellemans, 1990; Van der Poel et al., 1993) and airborne transmission (Elvander, 1996) cannot be excluded. Increasing herd size increases the risk for several infectious diseases in cattle (Toews et al., 1986; Frank and Kaneene, 1992; Van Donkersgoed et al., 1993) and other species (StaÈrk et al., 1992). The severity of infectious respiratory disease is related to the age distribution in the herds (Verhoeff and van Nieuwstadt, 1984); the probability of occurrence of a herd outbreak is higher in herds with a large proportion of calves and young cattle. (The proportion of susceptible individuals could be assumed to be greater in such herds, the infection thereby being able to establish itself more easily.) Since the first case of acute respiratory disease related to BRSV infection was diagnosed in Norwegian cattle in 1976 (édegaard and Krogsrud, 1977), infected animals have been diagnosed only occasionally (Krogsrud, personal communication). In a serological study (Hyllseth et al., 1987) of 250 Norwegian young calves, among various viruses related to respiratory disease, the overall prevalence of specific antibodies was highest for parainfluenza-3 virus and bovine rotavirus Ð whereas specific antibodies against BRSV were less common. A pronounced geographical variation in the seroprevalence was found (varying from 2 to 50% between the regions investigated). An outbreak of acute respiratory disease associated with BRSV infections involving many cattle herds occurred in Norway in 1995. It was assumed that the infection was introduced in December 1994 with a consignment of beef cattle imported from Denmark (where BRSV was associated with enzootic pneumonia of calves) (Uttenthal et al., 1996), and where the virus is now thought to be endemic (Alban et al., 1999). The primary cases appeared about a week after the imported cattle had been introduced into the herd; the disease then spread rapidly between farms. The affected cattle in the primary case herds seroconverted to BRSV (paired ELISA tests). All the imported animals were tested and found to be positive for antibodies against BRSV Ð some of them with high titres (Krogsrud, personal communication). This outbreak was considered to have had a great economic impact on the affected farms, and it would therefore be of great interest to identify risk factors for the occurrence of acute respiratory disease to obtain possible
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
89
measures to prevent transmission of the disease and to minimise the economic losses associated with any future epidemics. As far as we know, no other studies of risk factors in herd outbreaks of acute respiratory disease associated with BRSV in cattle have been conducted. The aims of the present study were to investigate the association between selected herd factors and the risk of such acute respiratory disease at the herd level. 2. Material and methods 2.1. Study area and study population Norwegian counties are divided into veterinary districts which may include one or more municipalities. The veterinary districts in this study were within the same veterinary region, administered by one Regional Veterinary Officer (RVO). The study area encompassed two adjacent veterinary districts located in the south-east of Norway where a number of cattle herds were placed under restrictions for acute respiratory disease during the observation period. The consequence of a restriction was that all movement of cattle from and to the restricted herds were prohibited. Data regarding the dates on which herd restrictions had been imposed were collected from the RVO. Pursuant to the Animal Disease Act, the RVO had decided that restrictions should be placed on affected herds during the outbreak in order to prevent further spread of the disease. The District Veterinary Officer (DVO) in each district reported the affected herds and also took decisions regarding introduction and lifting of restrictions. This decision was primarily based upon clinical observations of acute respiratory disease among the cattle in the herds, and on the knowledge that the first herds affected had confirmed diagnoses of BRSV infection. The veterinary visits to the affected farms were voluntary initiated by farmers recognising the disease problem in their cattle herds. All cattle herds in the study area which were registered in the Register of Production Subsidies (RPS) as of 1 January 1995, were included in the study. The RPS contains records of all farmers who apply for production subsidies from the Ministry of Agriculture. The figures recorded in the RPS is used for the official Norwegian Statistics of Agriculture and is estimated to be the census population of herds. The registry contains information on the number of animals of each species and age category present on the farm on the date of application. The age categories for cattle in the registry are: dairy cows, suckling cows, bulls older than 12 months, heifers older than 12 months, male calves younger than 12 months and female calves younger than 12 months. The registry is updated twice a year on 1 January and 1 July. Comparison of the herd sizes given on these two dates showed that they remained relatively constant, and therefore only the information recorded on 1 January 1995 was used in the analyses. 2.2. Observation period The study period was defined as beginning 7 days before restrictions were first applied until the day when all restrictions had been lifted in the area and lasted from 1 January to 31 May.
90
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
2.3. Design The study was a retrospective cohort study with the herd as the observational unit. The outcome event was the imposition of restrictions due to acute respiratory disease. 2.4. Statistical analysis The statistical analyses were performed with the SAS-PC System1 Version 6.12 for Windows (SAS Institute, Cary, NC, USA 1996). The explanatory variables examined in the study were: veterinary district, production type, age profile and herd size. Veterinary district was treated as a categorical variable with two levels. The production type was defined and classified as a categorical variable with the following categories: Dairy herds: Cattle herds with milking cows, calves, heifers older than 12 months, not more than one bull older than 12 months, and no suckling cows. Beef herds: Cattle herds with no milking cows. Mixed herds: Cattle herds with milking cows and more than one bull older than 12 months of age or suckling cows. The age profile was included in the analysis as three separate variables; calf, young and adult, and based on the proportions of each age category in the herd relative to the total number of cattle. The denominators of the variables were defined as follows: calf; calves <1 year, young; heifers and bulls (1±2 years) and adults; parous cows. The herd size was estimated as the sum of all bovine animals in each category present in the herd on the date of the application for subsidies. The survival time was defined as the period from 7 days before application of the first restrictions in each of the two veterinary districts, or until the end of the study period for the herds never diagnosed as infected. The assumption of proportional hazards between the levels of all independent variables was evaluated by the log±log survival plot of the stratified continuous variables (quartiles), or of the various categories of qualitative variables. After introductory descriptive analyses, all the independent variables were available for inclusion in a Cox proportional-hazards model (Cox, 1972). Independent variables with unconditional p-values <0.25 were offered to the multivariate Cox model. Each explanatory variable was forced into the model in a forward stepwiseselection procedure. To determine the best-fitting model, a series of nested hazards models with and without 2-way interaction terms of all the included variables were fitted. The proportionality assumption was evaluated individually for each explanatory variable by including in the model a time-dependent covariate representing the interaction of the explanatory variable with the survival time. A 2-tailed likelihoodratio statistic test (a<0.05) was used to compare all the various models based on a w2-distribution. A goodness-of-fit test developed by Schoenfeld (1982) was used to evaluate the fit of the final multivariate model.
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
91
Table 1 The mean proportion of animals within various age categories in relation to the production types of the 431 cattle herds included in the study of risk factors of acute respiratory disease in cattle during 1 January to 31 May of 1995 in two veterinary districts located in the south-east of Norway Production type
Mixed Dairy Beef
Age category (95% CI) Calf
Young
Adult
0.34 (0.33±0.35) 0.35 (0.33±0.37) 0.38 (0.33±0.42)
0.34 (0.32±0.35) 0.23 (0.21±0.25) 0.45 (0.39±0.50)
0.32 (0.31±0.33) 0.42 (0.40±0.44) 0.18 (0.14±0.22)
3. Results The study population consisted of 431 cattle herds (206 and 225 in Hadeland and Toten veterinary districts, respectively). The distribution of the herds according to production type was: 118 dairy, 107 beef, and 206 mixed herds. The median herd size of the study population was 37 animals (1±219), differing between herds of various production type as follows: dairy: 26 (2±130), beef: 18 (1±203), mixed; 48 (7±219). The mean proportions of the different age categories in relation to production type are shown in Table 1. The cumulative herd-level incidence of acute respiratory disease was 35.0% (95% CI: 26.0±43.5) in dairy herds, 19.6% (95% CI: 12.0± 27.3) in beef herds, and 47.0% (95% CI: 40.2±54.0) in mixed herds. The survival times of the herds varied in relation to herd size (Fig. 1) and production type (Fig. 2). Neither the log±log survival plots nor the goodness of fit test revealed
Fig. 1. Kaplan±Meier curves for the occurrence of acute respiratory disease in relation to herd size categories (quartiles) for 431 cattle herds located in two veterinary districts in the south-east of Norway during the study period of 1 January to 31 May in 1995.
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
92
Fig. 2. Kaplan±Meier curves for the occurrence of acute respiratory disease in relation to production type categories for 431 cattle herds located in two veterinary districts in the south-east of Norway during the study period of 1 January to 31 May in 1995.
any violations of the assumption of proportional hazards for the various explanatory variables, and no time-dependency could be detected for any of the explanatory variables. A final multivariable model (Table 2) contained the terms for production type and herd size and their interaction. For example, a dairy herd with 20 animals had a hazard ratio of 3 whereas the hazard ratio for a mixed herd or a beef herd of a corresponding size was 5 and 1.5, respectively. Increasing the herd size to 50 head increased the hazard ratio for a dairy herd to 8, for a mixed herd to 6 and for a beef herd to 3 (Fig. 3).
Table 2 The final multivariable Cox proportional-hazard model for the risk of an outbreak of acute respiratory disease in 431 Norwegian cattle herds during the study period (1 January to 31 May of 1995) Variable
Levels
Production category
Mixed Dairy Beefa
Herd size
±
Production typeherd size
Mixedherd size Dairyherd size Beefherd sizea
a
Reference level.
b
S.E.
P
HR
95% CI
1.49 0.50 0
0.37 0.44 ±
0.001 0.26 ±
4.43 1.65
2.14, 9.13 0.70, 3.93 ±
0.02
0.001
0.001
1.02
1.01, 1.03
ÿ0.013 0.01 0
0.004 0.008 ±
0.004 0.21
0.99 1.01 1
0.98, 1.00 0.99, 1.03 ±
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
93
Fig. 3. Estimated risk based on the final Cox proportional-hazard model of a herd outbreak of acute respiratory disease for each production type in relation to herd size for 431 cattle herds located in two veterinary districts in the south-east of Norway during the study period of 1 January to 31 May in 1995.
4. Discussion Because BRSV infection was demonstrated in affected cattle in the primary case herds, BRSV was most probably the infectious agent associated with the outbreak. Although respiratory diseases are usually encountered as sporadic cases in single animals in the Norwegian cattle population and no specific respiratory virus occurs endemically in Norway, other respiratory viruses might nevertheless have been involved. The risk of acute respiratory disease in herds varied significantly with type of production, although the biological exposure factors associated with this factor are unclear. Veterinarians and animal handlers have been considered to play a role in the transmission of BRSV (Wellemans, 1990). Dairy and mixed herds are regularly visited by a number of people such as milk-tanker drivers, veterinarians, AI-technicians and relief milkers, whereas beef herds are more isolated. The patterns of purchase and human traffic thus might differ between herds according to type of production, and therefore possibly explain the increased risk found in this study in mixed and dairy herds compared to the beef herds. (This outbreak occurred during winter and spring months when all herds were indoors with little risk of direct cow-to-cow transmission between herds.) The risk of acute respiratory disease was highest in the larger herds. The increasing risk related to increase in herd size could be explained by greater indirect contact with other herds due to more frequent human traffic. This is in accordance with findings from other
94
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
studies of infectious diseases (Toews et al., 1986; Frank and Kaneene, 1992; StaÈrk et al., 1992; Van Donkersgoed et al., 1993) and the theory of De Jong et al. (1995). Moreover, any infectious agent will establish itself more easily in a large herd because of the greater degree of animal-to-animal contact. However, the increase in risk associated with increase in herd size was more marked in dairy herds than in the other types of herd. An increase of the size of a dairy herd would probably result in a greater increase in human traffic (such as veterinarians or AI-technicians) than would be the case for a beef or mixed herd. There is obviously a larger proportion of dairy cows in dairy herds than in the mixed herds, and any increase in size in dairy herds will thus lead to a corresponding increase in the number of dairy cows while any increase in size in mixed herds could just as well be related to a higher proportion of beef cattle in the herd. However, our study included only a few large dairy herds (the mixed herds being generally larger). This might explain most of the differences found between the dairy and mixed herds. A further contributory factor could be that in larger dairy herds there is a greater proportion of high-yielding cows such animals having been reported to be more susceptible to infection by BRSV (Elvander, 1996). No relationship was demonstrated between the age profile of the animals in the herds and the risk of acute respiratory disease. A possible explanation could be that all the cattle herds were equally susceptible to infection because no recent outbreaks associated with BRSV had occurred in the area. The variability of the age distribution in the study herds was small, which might be a more-obvious reason for not finding any association. After the primary herd outbreaks in each district, the hazard of outbreaks of acute respiratory disease did not differ between the veterinary districts. This similar progression of the epidemic in the two areas supports the assumption that the epidemic was caused by the same infectious agent. Interpretation of the results of the present study is confounded by lack of data on the incidence of disease in the herds under no restrictions (either due to lack of reporting or to failure to diagnose disease in these herds). However, this lack of sensitivity is likely to be of a conservative type (reducing the power of the study). Different management practices in the different types of herd could have influenced the extent to which herd outbreaks were reported, and thereby also the apparent incidence and, in turn, the results of the study. Moreover, dairy farmers with milk production may be more motivated to contact a veterinarian in the event of a disease outbreak than beef farmers. If so, the hazard ratio would be underestimated for the beef herds and overestimated for the other types of herd. In the case definition, the time for censoring was set equal to the day of notification due to acute respiratory disease. It is to be expected that the onset of acute respiratory disease was days or even up to a week before the defined day of onset of acute respiratory disease. We assumed that the notification percentage was the same over the whole observation period and for all categories of herd. The study material consisted of the census population of the herds in the area registered in the RPS but this registry only includes active applicants. Further analyses have to be performed to determine and evaluate other factors that could possibly explain these results. Neither the effect of neighbourship between the herds, nor the population density in the areas concerned, were considered in the present analysis.
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
95
5. Conclusion In this study of herd outbreaks of acute respiratory disease in Norwegian cattle herds thought to be associated with BRSV, the risk of such outbreaks was associated with production type and herd size. These results suggest that there are underlying risk factors related to the production type and herd size that could further explain the dynamics and transmission routes of the causative agents, presumable BRSV, between cattle herds. Acknowledgements This study was financially supported by the Norwegian Research Council. The authors are grateful to the Norwegian Animal Health Authority for providing the data used in the analyses. References Alban, L., Larsen, L.E., ChrieÂl, M., Tegtmeier, C., Nielsen, T.K., 1999. The occurrence of clinical outbreaks of enzootic pneumonia in calves in 10 Danish dairy herds during the winter 1996±97: Descriptive Results, Proceedings of a meeting held at the University of Bristol, 24±26 March 1999. Society for Veterinary Epidemiology and Preventive Medicine, pp. 118±130. Bryson, D.G., Mc Ferran, J.B., Ball, H.J., Neill, S.D., 1978. Observations on outbreaks of respiratory disease in housed calves. Epidemiological and clinical findings. Vet. Rec. 103, 485±489. Cox, D.R., 1972. Regression models and life tables. J. R. Stat. Soc. B 74, 187±220. De Jong, M.C.M., Diekmann, O., Heesterbeek, J.A.P., 1995. How does transmission depend on population size? In: Mollison, D. (Ed.), Epidemic Models: Their Structure and Relation to Data. Cambridge University Press, Cambridge, pp. 84±94. Elvander, M., 1996. A study of bovine respiratory syncytial virus infections in Swedish cattle. Ph.D. Thesis, 124 pp. Frank, N.A., Kaneene, J.B., 1992. Management risk factors associated with calf diarrhoea in Michigan dairy herds. J. Dairy Sci. 76, 1313±1323. Hyllseth, B., Larsen, H.J., Norheim, K., 1987. Calf virus infections in Norway, a serosurvey. Seventh International Congress of Virology, Edmonton, Canada, p. 321. Inaba, Y., Tanaka, Y., Omori, T., Matumoto, M., 1972. Bovine respiratory syncytial virus: studies on an outbreak in Japan. Jpn. J. Microbiol. 16, 373±383. Jacobsson, S.O., Alenius, S., Ottander, G., PalmeÂr, H., Persson, J., Wennberg, K., Juntti, N., 1989. Bovint respiratoriskt syncytialt virus, BRSV som orsak til pneumonie hos kor. Sven. Vet. Tidn. 41, 641±647 (in Swedish). édegaard, é.A., Krogsrud, J., 1977. A field outbreak caused by bovine respiratory syncytial virus. Acta Vet. Scand. 18, 429±431. Ploeger, H.W., Boon, J.H., Klaassen, C.H.L., van Florent, G., 1986. A sero-epidemiological study of infections with bovine respiratory syncytial virus in first season grazing calves. J. Vet. Med. 33, 311±318. Schoenfeld, D., 1982. Partial residuals for the proportional hazards model. Biometrika 69, 51±55. StaÈrk, K.D.C., Keller, H., Eggenberger, E., 1992. Risk factors for the reinfection of specific pathogen-free pig breeding herds with enzootic pneumonia. Vet. Rec. 131, 532±535. Stott, E.J., Thomas, L.H., Collins, A.P., Crough, S., Jebbett, J., Smith, G., Luther, P.D., Caswell, R., 1980. A survey of virus infections of the respiratory tract of cattle and their association with disease. J. Hyg. 85, 237±247. Toews, D.W., Martin, S.W., Meek, A.H., 1986. Dairy calf management, morbidity in Ontario Holstein herds. Prev. Vet. Med. 4, 159±171.
96
M. NorstroÈm et al. / Preventive Veterinary Medicine 44 (2000) 87±96
Uttenthal, A., Jensen, N.P.B., Blom, J.Y., 1996. Viral ñtiology of enzootic pneumonia in Danish dairy herds: diagnostic tools and epidemiology. Vet. Rec. 139, 114±117. Van der Poel, W.H.M., Brand, A., Kramps, J.A., Middel, W.G.J., Van Oirschot, J.T., 1993. Dynamics of bovine respiratory syncytial virus: a longitudinal epidemiological study in dairy herds. Arch. Virol. 133, 309±321. Van Donkersgoed, J., Ribble, C.S., Boyer, L.G., Townsend, H.G.G., 1993. Epidemiological study of enzootic pneumonia in dairy calves in Saskatchewan. Can. J. Vet. Res. 57, 247±254. Verhoeff, J., van Nieuwstadt, A.P.K.M.I., 1984. BRS virus, PI3 virus, and BHV1 infections of young stock on self containing dairy-farms: epidemiological and clinical findings. Vet. Rec. 114, 288±293. Wellemans, G., 1990. Bovine respiratory syncytial virus. In: Dinter, Z., Morein, B. (Eds.), Virus Infections of Ruminants. Elsevier, Amsterdam, pp. 363±375.