Search tips
Search criteria 


Logo of canvetjReference to the Publisher site.Journal Web siteJournal Web siteHow to Submit
Can Vet J. 2010 April; 51(4): 385–390.
PMCID: PMC2839827

Language: English | French

Herd risk factors associated with sero-prevalence of Maedi-Visna in the Manitoba sheep population


Disease associated with Maedi-Visna infection results in substantial economic losses in affected sheep producing areas of the world. A survey was conducted to estimate herd and individual seroprevalence in the province of Manitoba and evaluate risk factors for seropositive herds. Of 2207 sheep sampled from 77 selected sheep flocks, the animal level seroprevalence was 2.47% and herd level seroprevalence was 25.10%. The herd-level factors of presence of clinical skin disease, herd size of > 70, history of musculoskeletal/lameness abnormalities, and the purchase of new stock (> 50) in the last 1 to 5 y, showed significant associations with seropositive herd status. The study documented a remarkable stability of low seroprevalence in the province over a 20-year period in the absence of a systematic disease control program.


Facteurs de risque pour le troupeau associés à la séro-prevalence de maedi-visna parmi la population de moutons du Manitoba. La maladie associée à l’infection par maedi-visna se traduit par d’importantes pertes financières dans les régions de production ovine touchées dans le monde. Un sondage a été réalisé pour faire une estimation de la séroprévalence parmi les animaux individuels et les troupeaux dans la province du Manitoba et évaluer les facteurs de risque pour les troupeaux séropositifs. Parmi les 2207 échantillons prélevés chez des moutons des 77 troupeaux ovins sélectionnés, le taux de séroprévalence chez les animaux était de 2,47 % et le taux de séroprévalence au sein du troupeau était de 25,10 %. Les facteurs de présence de maladies cliniques de la peau au niveau du troupeau, de taille du troupeau de > 70, d’antécédents d’anomalies musculo-squelettiques et de boiterie ainsi que l’achat de nouvelles têtes de bétail (> 50) depuis 1 à 5 ans, ont montré des associations importantes avec le statut séropositif du troupeau. L’étude a documenté la stabilité remarquable du faible taux de séroprévalence dans la province au cours d’une période de 20 ans en l’absence d’un programme de contrôle systématique de la maladie.

(Traduit par Isabelle Vallières)


Maedi-Visna or ovine progressive pneumonia (OPP in the United States) causes substantial economic losses to the sheep industry throughout the world. The non-oncogenic, non-immunosupressive viral agent Maedi-Visna virus (MVV) belongs to the family Retroviridae and subfamily Lentivirinae. The disease is named as a result of the pioneering work in the slow-viruses from ‘maedi’ (Icelandic for dyspnea) after the respiratory form and ‘visna’ (Icelandic for shrinking or failing) a form of the neurological disease which manifested as depression and weight loss when the disease first emerged in Iceland in the 1940’s (1,2). In North America, the disease is characterized by lymphoproliferative pneumonia, meningeal arteritis along with encephalitis, non-suppurative arthritis, lymphocytic mastitis and interstitial nephritis with membranoproliferative glomerulonephritis (35). Severe emaciation, chronic pneumonia, dyspnea in advanced cases, indurative mastitis with agalactia, and arthritis with lameness are the clinical manifestations of disease. It is rare for the disease to be clinically apparent before sheep are 4 y old (57). The caprine arthritis-encephalitis virus (CAEV) of goats is a virus that is closely related to MVV (6). Virology, clinical manifestation and disease transmission of the small ruminant lentiviruses, MVV and CAEV, have been recently reviewed (1,7,8).

Maedi-Visna virus, first recognized in Iceland in 1939 and subsequently eradicated, has been reported in major sheep rearing countries throughout the world except Australia and New Zealand. In 1972, the first Canadian cases were reported in Quebec (9). Subsequently 2 Maedi-Visna positive rams, which had originated from the same Quebec flock, were identified in Nova Scotia (10). A national serosurvey for MVV was conducted during 1988 to 1989 and showed that the provinces of Quebec (40%) and Nova Scotia (27%) had significantly higher individual MVV seroprevalence rates in sheep than the other provinces. In that study, 530 ewes selected from 13 Manitoba flocks had a seroprevalence rate of 2.8% and 7 of 13 flocks (53.8%) tested had at least a single individual test positive (11).

With increasing quality and affordability of laboratory diagnostic tests, it may be possible to eradicate certain animal diseases. The policy question of a disease eradication program includes consideration of test performance, current disease prevalence, producer awareness, societal cost-benefit, and industry commitment. The purpose of this study was to estimate current animal and herd level seroprevalence of Maedi-Visna in the Manitoba sheep population, and to identify herd-level predictors of the disease as information prerequisites in considering geographic-based disease control program options.

Materials and methods

The sample size for the study was designed to generate an animal level MVV prevalence at a precision level of 5%. To calculate the sample size required to obtain this level of precision, the Aussie Vet on-line calculator for prevalence estimation with an imperfect test (12) was used to calculate the number of samples required, taking into account the published sensitivity of 98.6% and specificity of 96.9% of the enzyme-linked immuosorbent assay (cELISA) test used in this study (6). Although the Aussie Vet calculator is set up for herd-level analysis, it can also be used to identify the sample size required at the animal level. Expected disease prevalence was set at 20%, and was estimated based upon previous provincial serological surveys taken in Canada over the past 25 y: Quebec 1983, 37.3% (13); Quebec 1999, 32% (14); Ontario 1988, 20.9% (15); Alberta 2002, 13% (16). The sample size required for the study was then modified to take into account the potential effect of within flock clustering of MVV (17). In this calculation, it was assumed that an average of 30 animals would be tested per flock, and that the intra-class correlation (ICC) was 0.3. The intra-class correlation is a measure of the degree to which observations of disease cases are clustered within a herd; the higher the ICC, the larger the sample that is required to generate a prevalence estimate of desired precision, since additional animals tested within a herd are assumed to contribute less information than initially tested animals. These calculations estimated that a sample of 80 flocks and 2400 individual animals would be required for the study. The sample size calculations undertaken were focused on ensuring an adequate sample size for MV prevalence estimation and not on risk factor modeling, a secondary objective of the study.

Herds were selected for testing based on a stratified random sample. The sampling frame for the survey was a list of sheep-producing premises obtained from the 270 members of the Manitoba Sheep Association. Based on the owners’ postal code, premises were allocated to 1 of 12 agricultural districts in the province (18) using the 2006 Postal Code Conversion file (19). Agricultural districts used in this study are shown in Figure 1. A random list of premises was generated within an agricultural district and the target number of premises for inclusion was proportional to the number of herds within the district. If the initial randomly selected producer declined participation in the study, the next premises in the list was chosen. Each of the participating premises represented a single flock. Thirty animals (both rams and ewes) 24 mo or older were then randomly chosen within each flock for testing. If there were < 30 eligible animals in a flock, all eligible animals were tested.

Figure 1
Map of the 12 agricultural districts in the province of Manitoba as described by Statistics Canada (18). The percentage of flocks tested (pie chart) indicates the number of flocks sampled divided by the number of flocks within that agricultural district ...

A farmer questionnaire was developed to collect flock-level information including flock size and production characteristics, management practices, veterinary contact, and flock level animal health syndromes (observed in the last 5 y) potentially associated with MVV prevalence. Questionnaires were sent to each farmer to be completed prior to the farm visit for blood collection and were marked with a flock identity number which was used to track sample results and maintain confidentiality (questionnaire available from corresponding author).

Each farm was visited in the winter of 2007–2008 and blood was collected in 9 mL vaccutainer tubes by jugular venipuncture using standard techniques. Information related to flock identity and individual eartag number were recorded on a project laboratory submission form and submitted to the Manitoba provincial veterinary laboratory. Completion and accuracy of the production questionnaire was assured at the time of sample collection.

Samples were centrifuged and sera were decanted in 2 mL plastic serum tubes, similarly identified with flocks’ identity numbers and given sample numbers. These tubes were stored at −20°C until laboratory tests were run, usually within 5 working days.

Serum was tested using a commercially available CAEV antibody cELISA test kit (VMRD, Pullman, Washington, USA), which identifies antibodies on the surface envelope of small ruminant lentiviruses. This test was recently validated in sheep infected with North American MVV strains to have a sensitivity of 98.6% and specificity of 96.9% when used with sheep sera and an OD cut-off value of greater than 20.9% inhibition indicating an individual sample is positive (6).

Statistical estimates at both the animal and herd level were determined using the Survey module in STATA 9.0 (StataCorp LP, College Station, Texas, USA) (20). This module allows the front-end declaration of a complex sample design and associated sample weight variables which are then used to calculate appropriate parameter estimates and standard errors that take into account the effect of sample design. In this study, herds were defined as the primary sampling unit, with Agricultural Region a stratification variable. Sampling weights for both the animal and herd level prevalence were calculated in inverse proportion to the probability of being selected as per the method described by Dohoo et al (17). The survey version of the proportion command in STATA, with confidence intervals set at 95%, was used to calculate MVV prevalence estimates at both the animal and flock levels.

To model the relationship between flock level predictor variables and positive flock MVV sero-status, the survey version of negative binomial regression was implemented in STATA, with flocks designated as MVV positive if 1 or more animals in the flock tested positive for MVV. This analysis produced risk ratios and associated confidence intervals corrected for sample design. For this analysis, Agricultural Districts were truncated into 3 broad regions. Multi-level modeling was not considered for this study because of the relatively small sample size and because of the limited amount of information collected at the animal level. Multi-variable analysis was also not undertaken because of the small sample size (n = 77 herds) and because of the broad descriptive focus of the study (estimating MV prevalence, describing herd level factors associated with MV).


A total of 2207 animal samples were collected from 77 flocks in the 12 agriculture regions. Figure 1 shows the proportion of flocks sampled in each region. Forty-three animals were seropositive for MVV and 17 flocks out of 77 had at least 1 animal test positive for MVV. After adjusting for the complex sampling design, the individual animal level prevalence of MVV for Manitoba was calculated as 2.41% (95% CI: 1.13% to 3.81%) and the provincial herd level prevalence was calculated as 25.10% (95% CI: 13.9% to 36.27%).

Flock-level predictors of MVV prevalence are reported in Table 1. As shown, large flock size (rate ratio = 2.94) and muculoskeletal/lameness abnormalities (rate ratio = 2.56) were positively and significantly associated with positive MVV flock status. Flocks exhibiting skin lesions within the last 5 y were significantly less likely to be MVV positive (rate ratio = 0.25). Purchasing more than 50 new sheep in the last 5 y (rate ratio = 2.31) and unthriftiness (rate ratio = 1.94) had positive associations with MVV flock status, but did not quite achieve statistical significance. No statistically significant relationships were observed for any of the other disease syndromes, for type of sheep operation (purebred, commercial, mixed) or for geographic region (Table 2).

Table 1
Flock level predictors of Maedi-Visna in sheep, Manitoba 2008, negative binomial regression analysis
Table 2
Distribution of herds by model predictors and herd level MV status


The results of this study are similar to those of a previous study undertaken in Manitoba in 1988 (11). In that study, animal level prevalence was estimated at 2.8% and flock level prevalence was estimated to be 53.8%. This compares to prevalence estimates observed in this study of 2.41% and 25.1%, respectively. Precise comparisons of prevalence rates over time are difficult since diagnostic techniques have evolved and because sampling designs vary between studies. The 1988 Manitoba study results, for example, were based on only 530 animals in 13 flocks and it is conceivable that if confidence intervals had been reported for these prevalence results that they would overlap the prevalence values in the current study. This apparent agreement between studies suggests a remarkable stability of the MVV disease prevalence in the Manitoba sheep population, especially considering that there have been at least 5 generations of sheep in the 20-year time period between studies.

A more recent study in Alberta, which estimated MV prevalence using paired histology and serology from culled ewes, reported animal level MVV prevalence rates of 26.8% and 13.0% using the 2 techniques (16). The Alberta estimate is significantly higher than the results reported in this study, which was also the case in the 1988 national study, suggesting relative stability in prevalence of MVV in that population (11.8% individual animal prevalence in 1988) (11). The cELISA used in the recent Alberta study was similar to the one used in this study. However, our study used the recommendation for percent inhibition cutoff of a test positive of 20.9% as reported in the literature (6) not the 35% inhibition as recommended by the manufacturer for the testing of CAEV in goat sera. This may explain some of the lack of agreement between serology and histology in the Alberta study. The Alberta study was also directed at an older cohort of sheep than the present study which would be expected to inflate the apparent prevalence.

The stability of low-level population infection rates over time is consistent with recent longitudinal research on MVV transmission in Spain. Extensive rearing has been demonstrated to interfere with efficient MVV transmission within herds (21). This series of studies suggest that maternal transmission of the MV virus via naturally fed colostrum is only about 20% effective and is not sufficient to maintain MVV in a population over time if lateral transmission is prevented (22,23). Lambs infected in the first month of life have a high potential to infect seronegative lambs if housed in close proximity during the neo-natal period (23). This peak of inter-flock transmission is thought to reflect a period of high viremia in newly infected neo-natal sheep. Animals that have been infected for some time consistently have low viremia and the virus is closely associated with blood monocyte-macrophage cell lines (1,2,6).

In this study, herd size (> 70 animals) was significantly associated with MVV-positive herd status which is consistent with previous studies in Canada (14) and elsewhere (7) that have examined the relationship of herd size to the risk of MVV infection. The positive association of introduction of new animals (> 50 animals/y) in the last 1 to 5 y with flock level MVV status observed in this study reflects the infective nature of the disease. Unthriftiness/inappetence and musculoskeletal/lameness abnormalities in a herd are clinical features of progressing MVV and it was not surprising that these were positively associated (although not statistically significant) with a herd being positive on serology. The inverse association of skin disease and MVV has not been previously reported to be associated with MVV infection, is paradoxical, and no biologic explanation for this finding is known to us.

Clinical MVV disease in Canada has been characterized by chronic pneumonia, and may result in indurative mastitis (14,16,23). Indurative mastitis (mammary syndrome) was not identified as being associated with MVV in this study. Clinically, the disease is usually apparent only in sheep 4 y and older. Although the age demographics of ewes in Manitoba or in this study was not established, it is probable that only a small proportion of infected flocks would be old enough to manifest overt MVV disease. Similarly, in a low prevalence flock, a disease manifestation of clinical agalactia may not be recognized by the producer and affected ewes may subsequently be culled for reasons of poor production. Palpation of the udder of the individual sheep is not the regular management practice in sheep operations in Manitoba. In the Mediterranean basin it has been estimated that only 25% to 30% of seropositive animals develop clinical signs of agalactia (24).

This study has a number of limitations which need to be taken into account when interpreting the results. First, the relatively small sample size (77 flocks, 2207 animals) may have made it difficult to identify true flock factors associated with low farm-level infection with MVV. A number of predictors, which almost achieved statistical significance (unthriftiness and the number of new sheep purchased), should be treated seriously since their lack of statistical significance may be an artifact of small sample size. Secondly, the flock level information used to model farm-level infection with MVV was self-reported by producers and was not empirically validated. Accurate collection of this information was predicated upon the assumption that producers would be able to accurately remember and report any and all health issues they had observed in their flocks over the past 5 y. It is possible that there may be significant inaccuracies in this data which act to hide the relationship between important flock level factors and MVV infection. In addition detailed production records were not generally available in a format suitable for data analysis and minor production differences were not available for statistical analysis.

The results of this study suggest that regional eradication of MVV should be logistically possible for provinces such as Manitoba, which combine extensive management systems with pre-existing low prevalence rates. Epidemiologically MVV parallels the transmission of brucellosis in that there is a limited temporal period of high infectivity, there exists an excellent individual animal test and the majority of individually infected animals seroconvert prior to the next high risk temporal period. Programs of herd test and removal combined with animal movement restrictions and regional disease eradication have been very successful for brucellosis eradication. Industry based quality assurance programs may be a reasonable route to exploit in the development of disease control programs directed at increasing individual farm profitability.

Government veterinary authorities also have a public interest in controlling or eradicating small animal lentiviruses, as these viruses are the cause of significant poor animal welfare in agriculture production systems (7).


We are grateful to Lynn Herrmann-Hoesing, Washington State University, Pullman, Washington for providing our laboratory with Maedi-Visna positive and negative sera. We also thank Sharon Niebel for personal assistance in sample handling and Deanne Wasylyshen, Veterinary Diagnostic Services, Provincial Veterinary Laboratory for providing the serological testing. CVJ


This paper was made possible, in part, by The Manitoba Sheep Association and the Career Gateway Program, Manitoba Civil Service Commission.

Use of this article is limited to a single copy for personal study. Anyone interested in obtaining reprints should contact the CVMA office ( gro.vmca-amvc@nothguorbh) for additional copies or permission to use this material elsewhere.


1. Pépin M, Vitu C, Russo P, Mornex JF, Peterhans E. Maedi-visna virus infection in sheep: A review. Vet Res. 1998;29:341–367. [PubMed]
2. Thormar H. Maedi-Visna virus and its relationship to human immunodeficiency virus. AIDS Rev. 2005;7:233–245. [PubMed]
3. Pritchard GC, Dawson M. Maedi-visna. In: Martin WB, Aitken ID, editors. Diseases of Sheep. 3rd ed. Oxford: Blackwell Sci; 2000. pp. 187–191.
4. Angelopoulou K, Brellou GD, Vlemmas I. Detection of Maedi-visna virus in the kidneys of naturally infected sheep. J Comp Pathol. 2006;134:329–335. [PubMed]
5. Narayan O, Zink M, Gorrell S, et al. The lentiviruses of sheep and goats. In: Levy JA, editor. The Retroviridae. New York: Plenum Press; 1993. pp. 229–256.
6. Herrmann LM, Cheevers WP, Marshall KL, et al. Detection of serum antibodies to ovine progressive pneumonia virus in sheep using a caprine arthritis-encephalitis virus competitive-inhibition enzyme-linked immunosorbent assay. Clin Diag Lab Immunol. 2003;10:862–865. [PMC free article] [PubMed]
7. Peterhans E, Greenland T, Badiola J, et al. Routes of transmission and consequences of small ruminant lentiviruses (SRLVs) infection and eradication schemes. Vet Res. 2004;35:257–274. [PubMed]
8. Straub OC. Maedi-visna virus infection in sheep. History and present knowledge. Comp Immunol Microbiol Infect Dis. 2004;27:1–5. [PubMed]
9. Bellavance RD, Turgeon J, Phaneuf B, Sauvageau R. Pneumonie inter-stitielle et progressive du mouton. Can Vet J. 1974;15:293–297. [PMC free article] [PubMed]
10. Stevenson RG. Maedi-visna infection in rams in Nova Scotia. Can Vet J. 1978;19:159–163. [PMC free article] [PubMed]
11. Simard C, Morley RS. Seroprevelence of Maedi-visna virus in Canadian sheep. Can J Vet Res. 1991;55:269–273. [PMC free article] [PubMed]
12. Humphry RW, Cameron A, Gunn GJ. A practical approach to calculate sample size for herd prevalence surveys. [Last accessed February 17, 2010];Prev Vet Med. 2004 65:173–188. [interactive Webtool] Available from [PubMed]
13. Lamontagne L, Roy R, Girard A, Samagh BS. Seroepidemiological survey of Maedi-visna virus infection in sheep and goat flocks in Quebec. Can J Comp Med. 1983;47:309–315. [PMC free article] [PubMed]
14. Arsenault J, Dubreuil P, Girard C, Simard C, Bélanger D. Maedi-visna impact in Quebec sheep flocks (Canada) Prev Vet Med. 2003;59:125–137. [PubMed]
15. Campbell JR, Menzies PI, Waltner-Toews D, Walton JS, Buckrell BC, Thorsen J. The seroprevalence of Maedi-visna in Ontario sheep flocks and its relationship to flock demographics and management practices. Can Vet J. 1994;35:39–44. [PMC free article] [PubMed]
16. Fournier D, Campbell JR, Middleton DM. Prevalence of Maedi-visna infection in culled ewes in Alberta. Can Vet J. 2006;47:460–466. [PMC free article] [PubMed]
17. Dohoo I, Martin W, Stryhn H. Veterinary Epidemiologic Research. AVC Inc; Charlottetown, Canada: 2003. Sampling; pp. 27–47.
18. Statistics Canada. Census Agricultural regions boundary files for the 2006 census of agriculture — reference guide. Minister of Industry. 2007. [Last accessed February 17, 2010]. [monograph on the internet] Available from
19. Statistics Canada. Postal code conversion file. Minister of Industry. 2007. [Last accessed February 17, 2010]. [monograph on the internet] Available from
20. STATA Release 9 Manual. Survey Data. StataCorp LP; College Station, Texas, USA: p. 221.
21. Leginagoikoa I, Juste RA, Barandika J, et al. Extensive rearing hinders Maedi-visna virus (MVV) infection in sheep. Vet Res. 2006;37:767–778. [PubMed]
22. Alvarez V, Arranz J, Daltabuit-Test M, et al. Relative contribution of colostrum from Maedi-visna virus (MVV) infected ewes to MVV-seroprevalence in lambs. Res Vet Sci. 2005;78:237–243. [PubMed]
23. Alvarez V, Daltabuit-Test M, Arranz J, et al. PCR detection of colostrum associated Maedi-visna virus (MVV) infection and relationship with ELISA-antibody status in lambs. Res Vet Sci. 2006;80:226–234. [PubMed]
24. Christodoulopoulos G. Maedi-visna: Clinical review and short reference on the disease status in Mediterranean countries. Small Rum Res. 2006;62:47–53.

Articles from The Canadian Veterinary Journal are provided here courtesy of Canadian Veterinary Medical Association