Veterinary Research Communications, 14 (1990) 279-285 Copyright 6 Kluwer Academic Publishers bv - Printed in the Netherlands
THE USE OF THE ENZYME-LINKED IMMUNOSORBENT ASSAY (ELISA) IN SEROLOGICAL DIAGNOSIS OF MyCOPLASiW4 BOKW IN DAIRY CATTLE I.J. UHAAl, H.P. RIEMANN’, M.C. THURMOND2 AND C.E. FRANTI’ ‘Department of Epidemiology and Preventive Medicine, University of California, Davis, CA 95616, USA ?‘he Veterinary Medicine Teaching and Research Center, Tulare, CA 93274, USA
Uhaa. I.J., Riemann, H.P., Thunnond, M.C. and Franti, C.E., 1990. The use of the enzyme-linked immunosorbent assay (ELI&A) in serological diagnosis of Mycopkwna bovis in dairy cattle. Veterinary Research Communications, 14 (4), 279-285 Between December 1985 and March 1987 an enzyme-linked immunosorbent assay (ELISA) was used with 3774 sera to estimate the prevalence of antibodies to Mycoplasma bwis in sera from three age groups of cattle in four dairies in California and to test for possible associations between the presence of M. t&s antibodies and the age or breed of the cattle and the farm. Unadjusted and adjusted associations were evaluated using the x-square test for associations and multiple logistic regression analysis, respectively. There was a tendency for the proportion of cattle seropositive for M. boric to increase steadily and approximately linearly with age (‘p ~0.05). There was also a statistically significant relationship bbehveen a M. bovi.r seropositive test and being from Farm IV @ c 0.05). Farm IV was the largest of the four dairies and this association may be due to the effect of herd size. These findings confirm the ubiquitous distribution of antibodies to M. bovis in dairy cattle in California and also support previous reports of herd size as an important factor in mycoplasmal mastitis. Kcywordsz age, breed, California, cattle, ELKA, Mycoplasma bovis, serodiagnosis.
INTRODUCTION Mycoplasma mycoides, subspecies mycoides, and Mycoplasma bovis (M. bovis) are considered the most pathogenic species of Mycoplasma in cattle (Gourlay and Howard, 1979). The first isolation of M. bovis in cattle in the United States was from an epidemic of mastitis (Hale et al., 1%2). In California, M. bovis was first diagnosed in 1964 (Fahey and McKelvey, 1%5) and has since been observed quite frequently here and elsewhere (Boughton, 1979; Jasper ef af., 1979; Boothby et al., 1982). M. bovis has been implicated in a variety of bovine diseases, including polyarthritis, synovitis and pneumonia (Hjerpe and Knight, 1972, Stalheim and Page, 1975; Langsford, 19n, Rosendal and Martin, 1986). In young calves it is sometimes involved in pneumonia, especially in those kept under intensive management conditions (Campbell, 1983). MycopZasma may be responsible for the primary lesions, although in some instances there may be concurrent viral infections. Mycoplasmal pneumonia occasionally occurs as an outbreak in calves but the disease may be subclinical or
280
clinically manifest as a mild, chronic pneumonia. Concurrent infection with other bacteria exacerbates the clinical picture (Campbell, 1983). In cows, M. bovis may be the most common cause of mycoplasmal mastitis in some herds (Bar-Moshe, 1%4, Jasper et al., 1974; Jasper, 1977,198l). Mastitis caused by M. bovis may be subclinical, clinically mild or severe, but a common observation is the absence of any systemic reaction (Campbell, 1983). The organism has also been found in the genitalia of apparently healthy cows and may be associated with some cases of salpingitis. Transmission of M. bovis probably occurs through inoculation into a quarter or by external application of infected milk secretion to the teat skin. It may also be transmitted in calves through the consumption of infected milk. New cases may be expected in the herd as long as large numbers of mycoplasmids are shed, The spread within a herd may be dramatic and explosive; at other times it is slow and insidious. The speed of transmission may be related both to management practices and to the strain of the organism. Very few studies to evaluate for possible risk factors associated with M. bovis infection have been reported. Thomas et al. (1982), in an attempt to identify key factors which act as determinants of M. bovis infection, compared bulk milk tank positive and negative herds from a California survey. They used a case-control study design to investigate whether the incidence of clinical outbreaks of mycoplasmal mastitis differed in herds classified differently in the 1981 bulk tank survey with respect to certain key herd factors. The identification of such herd factors and of cow factors, such as age and breed, which are similarly associated, may be helpful in formulating effective control and preventive strategies. Farm associations may help define management (herd size, animal handling, milking instruments and practices) or iatrogenic factors, associated with veterinary treatments, on a particular farm which lead to greater exposure to M. bovis infection. In the present study, our objective was to use an enzyme-linked immunosorbent assay (ELISA) technique to determine the seroprevalence of M. bovis antibodies in apparently healthy cattle sampled from four dairy herds in the Central Valley of California. We also utilized cross-sectional data to examine age, breed and farm associations with a positive antibody response.
MATERIALS
AND METHODS
The protocols for data gathering, serum collection, serological assay and data storage and analysis were similar to those described previously (Uhaa et al., 1990). Briefly, sera were collected from cattle in three age cohorts (calves, heifers and cows) at intervals of approximately two months. Calves were defined as cattle less than one year old, heifers were between one and two years old and cows were over two years old. The design of the study was to select randomly at least thirty cattle in each of the three cohorts and to blood sample them at each visit to the end of the study or until they left the cohort. When possible we added more cattle to the cohorts to attain our set sample size of thirty, but this could not always be done. The main criterion for including these farms in the study was the expressed willingness of the farmer to participate. Farms I and III are located in the Sacramento valley, while farms II and IV are located in the San Joaquin valley. The herd size varied from approximately 150 on farm I to more than 500 on farm IV.
281
All the sera were assayed using the enzyme-linked immunosorbent assay (ELISA) with an M. bovis antigen. The indirect ELISA method used was a modification (Uhaa et af., 1990) of that described by Behymer er 41. (1985). To discriminate between negative and positive tests, the ELISA results were printed out as a histogram so as to identity visually and place a bound on the apparently negative subpopulation. The discriminating value was placed at the mean plus two standard deviations for this distinct grouping of the negative subpopulation. The primary unit of concern in this study was the individual cow because we intended to estimate the seroprevalence rates as the cattle aged from calves to adults and because we were interested in cow-related phenomena such as age, breed and antibody response.
DATA ANALYSIS Statistical analyses were performed using the Biomedical Data Programs (BMDP) (Dixon and Brown, 1985). Descriptive statistics were initially computed for each independent variable used in this study. This was done to detect possible significant departures from normal distributions and to provide an overall picture of the data. Seroprevalence estimates for M. bovis were calculated by dividing the number of test positive sera by the total number of serum samples for that particular bleeding date and age cohort. Odds ratios were calculated as measures of the strength of the association between age, breed, farm and a positive reaction to M. bovis. Unadjusted odds of these variables and positive M. bovis reaction were tested for statistical significance by use of the x2 test for associations (Daniel, 1987). Multiple logistic regression analysis (Breslow and Day, 1980; Kleinbaum et al., 1987) was used to examine the strength of these relationships and to form predictive models for positive M. bovis reaction using the main effects and interaction terms. Forward selection and backward elimination algorithms based on asymptotic covariance estimates were used to decide which variables and interaction terms to include in the final logistic regression model. F-to-enter and F-to-remove of 0.10 and 0.15 respectively were used. At each step of variable entry an improvement x2 statistic was generated to determine whether there was a significant improvement in fit over the previous step’s model. The selection of the best fitting model was based on the goodness of fit criteria (Lemeshow and Hosmer, 1982), biological plausibility and interpretability. Regression coefficients were used to estimate adjusted odds of a positive M. bovis.
RESULTS The point seroprevalence of positive reactors to M. bovis antigen for each farm, by bleeding date and age group, is presented in Table I. Although there was great variability in seroprevalence estimates the average seroprevalence rates rose fairly steadily from about 28% in March to about 75% in December, 1986. There was a significant tendency for the proportion of cattle seropositive for M. bovis to increase steadily and approximately linearly with age. However, this relationship reversed after adjustment for the possible confounding effects of breed
and farm, with calves being more likely to have an M. bovis seropositive test than heifers or cows (p ~0.05). Unconditional and conditional associations between age, breed, farm and positive reaction to M. bovi~ and 95% confidence intervals for each factor in the final logistic regression model are presented in Table II.
TABLE I Age and farm specific distribution of antibodies to Mycopla~mab&r central valley of California, December 1985 to March 1986
Farm
Bleeding
date
calves No. tested
I
Aug May 22,1986 121986 Oct30,1986 Dee 24,1986 Feb 26,1987
II
Apr Jun Aug Ott Dee
III
Iv
Numbers
antigen
in cattle
from
Heifers
No. positive
No. tested
four
dairies
cows
No. positive
No. tested
0
10
0
30,1986 24,1986 28,1986 10,1986 16,1986
10 10 19 23
0 1: (78.9) 0
35 34 52 64
7 6 48 6
(20.0) (17.6) (92.3) (9.4)
41 54 44 42 36
13 17 39 11 12
Jul24,1986 Ott 10,1986 Dee 12,1986
14 16 50
1 (7.0) 0 1.5 (30.0)
53 39 52
22 (41.5)
0
5” (9.6)
40 79 1%
Dee Mar Jun Sep Dee
26 18 29 1
l(3.8) 0 1 (3.4) 0
140 133 139 135 146
4 (2.9) 14 (10.5) Q(36.7) 60 (44.4) % (65.8)
267 284 276 270 275
are percentages
of test positives
in brackets
102 54 108 103 109
No. positive
6
17,198S 241986 23,1986 29,1986 l&l986
in the
1; (10.8) 3 (2.8) 20 (19.4) l(O.9) (31.7) (315) (88.6) (26.2) (33.3)
1: (7.0) 160 107 165 153 216
(seroprevalence)
A statistically significant association was found between an M. bovis seropositive test and being from farm IV, when adjusted for the possible confounding effects of age and breed (pcO.0001). As there was no sign&ant relationship between breed and an M. bovis seropositive test, both the breed main effect and the associated two-way interaction terms were dropped from the final logistic regression model.
DISCUSSION Although the ELISA for detecting exposure to Mycophma is a well researched and documented test (Carroll et aZ., 1976; Boothby er al., 1983; Thomas et al., 1987), the sensitivity and specitkity are not precisely known. However, Thomas et al. (1987) found that all of 19 bovine sera obtained from cows with confirmed M. bovis mastitis were positive by ELISA for M. bovis. All 127 negative reference sera obtained from
(60.0) (37.7) (59.8) (56.7) (78.6)
283
the University of California, Davis dairy herd in the same study tested negative by M. bovis ELISA. This herd had been tested repeatedly for the presence of M. bovis antibodies for a number of years without yielding any positive results; These results suggest that the M. bovis ELISA test has a sensitivity and specificity of close to 100% compared to culture and milk testing. TABLE II Crude and adjusted associations between age, farm location and positive antibody reaction to Mycophma bovis in dairy cattle sampled from the central valley of California, December 19@ja
Variablesb
Crude associations
Adjusted associationsd
OR
95% CIC
Coefficient
OR
95% CI
Heifers cows
2.43 1.80
(1.25,4.73) (0.%,3.37)
-2.0028 -1.4697
0.14 0.23
(0.06,0.31) (0.11,0.49)
Farm 3 Farm 4
0.57 11.45
(0.40, 1.23) (6.90,18.99)
-0.8026 2.7661
0.45 15.90
(0.24,0.84) (9.17,27.78)
“n = 856 records having complete age and farm location information bAge is categorized into age (1) calves, age (2) heifers and age (3) cows; Farmid is categorized into 1 = farm I, 3 = farm III and 4 = farm IV ‘Confidence intervals not including 1 indicate statistical significance at the 5% level dAdjusted for the main effects OR = odds ratio Results are for statistically significant variables
Cattle positive for M. bovis were found throughout the four herds studied. The association between being a calf and having antibodies to M. bovis, found after adjustment for confounding effects of breed and farm, may be due to one or a combination of factors, including the presence of colostral antibodies, the involvement of M. bovis in the bovine respiratory diseases complex and/or arthritis and the possibility that the levels of antibodies in heifers and cows may be below the threshold detectable by the ELISA. Another possible explanation may be a cohort effect, whereby the heifers and cows did not experience much exposure while the calves experienced an unusually intense exposure sometime in their lives. Multiple logistic regression analysis revealed a statistically significant association between farms and M. bovis ELISA positive tests. Animals on farm IV, with a herd size greater than 500, were 11 times more likely to have a positive test to M. bovis ELISA than those from the other three farms (herd size 150-300). This supports the association between clinical mycoplasma mastitis and increasing herd size reported by
Thomas and Jasper (1982). These researchers found a 14 to P&fold greater risk of clinical mastitis in dairy herds with more than 350 cows than in smaller herds. They speculated that herd size may be an indirect measure of several factors of management, animal density, animal movement and environment. In larger herds, the probability of having at least one infected cow which may act as a source of infection to other cows is greater than in smaller herds. Furthermore, poor management is more likely to be a problem in larger herds than smaller herds. All these factors, independently or in combination, could put larger herds at greater risk of M. bovis infection. With the present study design, however, we cannot identify which of these factors is important. In diagnosing exposure to M bovis based on cross-sectional testing for antibodies, we determined point prevalence, without knowing exactly what had happened to the animals prior to the test date. Such point prevalence estimates will be influenced by both the persistence of antibodies and the incidence of seroconversion. Lastly, detection of antibodies implies exposure but does not necessarily mean infection. Thus caution must be taken in the interpretation of the results presented in this study and especially in extrapolating the findings to different populations. REFERENCES Bar-Moshe, B., 1964. The isolation of Mjcoplasma from an outbreak of bovine mastitis in Israel. Journal of Veterinary Medicine, 21,97-99 Behymer, D.E., Ruppanner, R, Brook, D., Williams, J.C. and Franti, C.E., 1985. Enzyme immunoassay for surveillance of Q fever.American Journal of Veterinary Research, 46,2314-2417 Boothby, J.T., Jasper, D.E., Lutz, H. and Rollins, M.H., 1982. Gel electrophoresisderived enzyme-linked immunosorbent assay of serum from cows resistant to and cows susceptible to challenge exposure with Mycoplasma bovis.American Journal of Veterinaty Research, 43,553-S% Boothby, J.T., Jasper, D.E. and Rollin, M.H., 1983. Characterization of antigens for mycoplasmas of animal origin. American Journal of Veterirtaty Research, 44,433-439 Boughton, E., 1979. Mycoplasma bovis mastitis. Veterinary Bulletin, 49, 377-387 Breslow, N.E. and Day, N.E., 1980. Statistical Methoak in Cancer Research Vol. 1. The Analysis of Case-Control Studies, (International Agency on Cancer, Lyon, France), pp. 338 Campbell, EA., 1983. Animal Health in Australia. Vol. 3. Numtional Deficiencies and Diseases of Livestock, (Australian Government Publishing Service, Canberra), pp. 171 Carroll, E.J., Rollins, M.R and Jasper, D.E., 1976. The immune response of rabbits to 3 strains of Mjcoplasma agalactiae var. bovi.r isolated from mastitic bovine udders. Cornell Veterinarian, 66, 143-151 Daniel, W.W., 1987. Biostatistics: A Foundation for Analysts in the Health Sciences, 4th edn., (John Wiley, New York) Dixon, WJ. and Brown, M.B., 1985. Biomedical Data Programs, (University of California Press, Berkeley, C-9 Fahey, J.L. and McKeIvey, E.M., 196.5. Quantitative determination of serum immunoglobulins in antibody-agar plate. Journal of Immunology, 94,&I-90 Gourlay, RN. and Howard, CJ., 1979. In J.G. Tully and RF. Whitcomb (eds.), The Mycoplasmas, Vol. 2, (Academic Press, New York), pp. 129 Hale, H.H., Hemboldt, C.F., Plastridge, W.N. and Stula, E.F., 1%2. Bovine mastitis caused by a Mjcoplasma species. Cornell Veterinarian, 52,582-591 Hjerpe, CA and Knight, H.D., 1972. Polyarthritis and synovitis associated with Mycoplasma bovimastitidis in feedlot cattle. Journal of the American Veterinary Medical Association, 160,1414-1418 Jasper, D.E., 1977. Mycoplasma and mycoplasma mastitis. Journal of the American Veterinary Medical Association, 170,1162-1172 Jasper, D.E., 1981. Bovine mycoplasmal ma&is. In: C.E. Comellius and B.F. Simpson (eds.), Advances in Veterinary Science, (Academic Press, New York), pp. 121-159 Jasper, D.E., Bushnell, RB. and Dellinger, J.D., 1974. Mycoplasma mastitis, a brief review. California Veterinarian, 28,12-14 Jasper, D.E., Dellinger, J.D. and Hakason, H.D., 1979. Prevalence of mycoplasmal mastitis in California. American Journal of Veterinaty Research, 40,1043-1047
285 Kteinbaum, D.G., Kupper, L.L. and Muller, ICE, 1987. Applied Regrekon and other Multivariate Me&&, (Duxbury Press, North Scivtate, MA, USA) Langsford, E.U., 1977. Mjcoplasma agalactiae sub. sp. bovis in pneumonia and arthritis of bovine. Canadian Journal of Comparative Medicine, 41,89-W Lemeshow, S. and Hosmer, D.W., 1982. Review of the goodness of fit statistic for use in the development of logistic regression models. American Journal of Epidemiology, 115,92-106 Rosen&l, S. and Martin, S.W., 19% The association between serological evidence of mycoplasma infection and respiratory diseases in feedlot cahes. Canadian Journal of Veterbwy Research, 50, 179-183 Stalheim, O.H. and Page, LA., 1975. Naturally occurring experimentally induced mycoplasmal arthritis of cattle. Journal of Clinical Microbiology, 2, X.5-168 Thomas, C-B. and Jasper, D.E., 1982. Mycoplasma mastitis and herd size factor. California Veterinarian, 36,l.5-16 Thomas, C.B., Jasper, DE. and Willebetg, P., 1982. Clinical bovine mycoplasmal mastitis - an epidemiologic study of factors associated with problem herds. Acta Veterinaria Scandinavica, 23, 53-64 Thomas, C.B., Jasper, D.E., Bcothby, J.T. and Dellinger, J.D., 1987. Detection of antibody specific to Mycopiasma bow% and Mycopkwna califonticum by enzyme-linked immunosorbent assay (ELISA). IsraeI Journal of Medical Sciences, 23,X23-728 Uhaa, IJ., Riemann, H.P., Thurmond, MC. and Franti, C.E., 1990. A seroepidemiological study on bluetongue virus in dairy cattle in the Central Valley of California. Veterinary Research Communications, 14,99-112 (Accepted: 27 February 1990)
THE USE OF THE ENZYME-LINKED IMMUNOSORBENT ASSAY (ELISA) IN SEROLOGICAL DIAGNOSIS OF MyCOPLASiW4 BOKW IN DAIRY CATTLE I.J. UHAAl, H.P. RIEMANN’, M.C. THURMOND2 AND C.E. FRANTI’ ‘Department of Epidemiology and Preventive Medicine, University of California, Davis, CA 95616, USA ?‘he Veterinary Medicine Teaching and Research Center, Tulare, CA 93274, USA
Uhaa. I.J., Riemann, H.P., Thunnond, M.C. and Franti, C.E., 1990. The use of the enzyme-linked immunosorbent assay (ELI&A) in serological diagnosis of Mycopkwna bovis in dairy cattle. Veterinary Research Communications, 14 (4), 279-285 Between December 1985 and March 1987 an enzyme-linked immunosorbent assay (ELISA) was used with 3774 sera to estimate the prevalence of antibodies to Mycoplasma bwis in sera from three age groups of cattle in four dairies in California and to test for possible associations between the presence of M. t&s antibodies and the age or breed of the cattle and the farm. Unadjusted and adjusted associations were evaluated using the x-square test for associations and multiple logistic regression analysis, respectively. There was a tendency for the proportion of cattle seropositive for M. boric to increase steadily and approximately linearly with age (‘p ~0.05). There was also a statistically significant relationship bbehveen a M. bovi.r seropositive test and being from Farm IV @ c 0.05). Farm IV was the largest of the four dairies and this association may be due to the effect of herd size. These findings confirm the ubiquitous distribution of antibodies to M. bovis in dairy cattle in California and also support previous reports of herd size as an important factor in mycoplasmal mastitis. Kcywordsz age, breed, California, cattle, ELKA, Mycoplasma bovis, serodiagnosis.
INTRODUCTION Mycoplasma mycoides, subspecies mycoides, and Mycoplasma bovis (M. bovis) are considered the most pathogenic species of Mycoplasma in cattle (Gourlay and Howard, 1979). The first isolation of M. bovis in cattle in the United States was from an epidemic of mastitis (Hale et al., 1%2). In California, M. bovis was first diagnosed in 1964 (Fahey and McKelvey, 1%5) and has since been observed quite frequently here and elsewhere (Boughton, 1979; Jasper ef af., 1979; Boothby et al., 1982). M. bovis has been implicated in a variety of bovine diseases, including polyarthritis, synovitis and pneumonia (Hjerpe and Knight, 1972, Stalheim and Page, 1975; Langsford, 19n, Rosendal and Martin, 1986). In young calves it is sometimes involved in pneumonia, especially in those kept under intensive management conditions (Campbell, 1983). MycopZasma may be responsible for the primary lesions, although in some instances there may be concurrent viral infections. Mycoplasmal pneumonia occasionally occurs as an outbreak in calves but the disease may be subclinical or
280
clinically manifest as a mild, chronic pneumonia. Concurrent infection with other bacteria exacerbates the clinical picture (Campbell, 1983). In cows, M. bovis may be the most common cause of mycoplasmal mastitis in some herds (Bar-Moshe, 1%4, Jasper et al., 1974; Jasper, 1977,198l). Mastitis caused by M. bovis may be subclinical, clinically mild or severe, but a common observation is the absence of any systemic reaction (Campbell, 1983). The organism has also been found in the genitalia of apparently healthy cows and may be associated with some cases of salpingitis. Transmission of M. bovis probably occurs through inoculation into a quarter or by external application of infected milk secretion to the teat skin. It may also be transmitted in calves through the consumption of infected milk. New cases may be expected in the herd as long as large numbers of mycoplasmids are shed, The spread within a herd may be dramatic and explosive; at other times it is slow and insidious. The speed of transmission may be related both to management practices and to the strain of the organism. Very few studies to evaluate for possible risk factors associated with M. bovis infection have been reported. Thomas et al. (1982), in an attempt to identify key factors which act as determinants of M. bovis infection, compared bulk milk tank positive and negative herds from a California survey. They used a case-control study design to investigate whether the incidence of clinical outbreaks of mycoplasmal mastitis differed in herds classified differently in the 1981 bulk tank survey with respect to certain key herd factors. The identification of such herd factors and of cow factors, such as age and breed, which are similarly associated, may be helpful in formulating effective control and preventive strategies. Farm associations may help define management (herd size, animal handling, milking instruments and practices) or iatrogenic factors, associated with veterinary treatments, on a particular farm which lead to greater exposure to M. bovis infection. In the present study, our objective was to use an enzyme-linked immunosorbent assay (ELISA) technique to determine the seroprevalence of M. bovis antibodies in apparently healthy cattle sampled from four dairy herds in the Central Valley of California. We also utilized cross-sectional data to examine age, breed and farm associations with a positive antibody response.
MATERIALS
AND METHODS
The protocols for data gathering, serum collection, serological assay and data storage and analysis were similar to those described previously (Uhaa et al., 1990). Briefly, sera were collected from cattle in three age cohorts (calves, heifers and cows) at intervals of approximately two months. Calves were defined as cattle less than one year old, heifers were between one and two years old and cows were over two years old. The design of the study was to select randomly at least thirty cattle in each of the three cohorts and to blood sample them at each visit to the end of the study or until they left the cohort. When possible we added more cattle to the cohorts to attain our set sample size of thirty, but this could not always be done. The main criterion for including these farms in the study was the expressed willingness of the farmer to participate. Farms I and III are located in the Sacramento valley, while farms II and IV are located in the San Joaquin valley. The herd size varied from approximately 150 on farm I to more than 500 on farm IV.
281
All the sera were assayed using the enzyme-linked immunosorbent assay (ELISA) with an M. bovis antigen. The indirect ELISA method used was a modification (Uhaa et af., 1990) of that described by Behymer er 41. (1985). To discriminate between negative and positive tests, the ELISA results were printed out as a histogram so as to identity visually and place a bound on the apparently negative subpopulation. The discriminating value was placed at the mean plus two standard deviations for this distinct grouping of the negative subpopulation. The primary unit of concern in this study was the individual cow because we intended to estimate the seroprevalence rates as the cattle aged from calves to adults and because we were interested in cow-related phenomena such as age, breed and antibody response.
DATA ANALYSIS Statistical analyses were performed using the Biomedical Data Programs (BMDP) (Dixon and Brown, 1985). Descriptive statistics were initially computed for each independent variable used in this study. This was done to detect possible significant departures from normal distributions and to provide an overall picture of the data. Seroprevalence estimates for M. bovis were calculated by dividing the number of test positive sera by the total number of serum samples for that particular bleeding date and age cohort. Odds ratios were calculated as measures of the strength of the association between age, breed, farm and a positive reaction to M. bovis. Unadjusted odds of these variables and positive M. bovis reaction were tested for statistical significance by use of the x2 test for associations (Daniel, 1987). Multiple logistic regression analysis (Breslow and Day, 1980; Kleinbaum et al., 1987) was used to examine the strength of these relationships and to form predictive models for positive M. bovis reaction using the main effects and interaction terms. Forward selection and backward elimination algorithms based on asymptotic covariance estimates were used to decide which variables and interaction terms to include in the final logistic regression model. F-to-enter and F-to-remove of 0.10 and 0.15 respectively were used. At each step of variable entry an improvement x2 statistic was generated to determine whether there was a significant improvement in fit over the previous step’s model. The selection of the best fitting model was based on the goodness of fit criteria (Lemeshow and Hosmer, 1982), biological plausibility and interpretability. Regression coefficients were used to estimate adjusted odds of a positive M. bovis.
RESULTS The point seroprevalence of positive reactors to M. bovis antigen for each farm, by bleeding date and age group, is presented in Table I. Although there was great variability in seroprevalence estimates the average seroprevalence rates rose fairly steadily from about 28% in March to about 75% in December, 1986. There was a significant tendency for the proportion of cattle seropositive for M. bovis to increase steadily and approximately linearly with age. However, this relationship reversed after adjustment for the possible confounding effects of breed
and farm, with calves being more likely to have an M. bovis seropositive test than heifers or cows (p ~0.05). Unconditional and conditional associations between age, breed, farm and positive reaction to M. bovi~ and 95% confidence intervals for each factor in the final logistic regression model are presented in Table II.
TABLE I Age and farm specific distribution of antibodies to Mycopla~mab&r central valley of California, December 1985 to March 1986
Farm
Bleeding
date
calves No. tested
I
Aug May 22,1986 121986 Oct30,1986 Dee 24,1986 Feb 26,1987
II
Apr Jun Aug Ott Dee
III
Iv
Numbers
antigen
in cattle
from
Heifers
No. positive
No. tested
four
dairies
cows
No. positive
No. tested
0
10
0
30,1986 24,1986 28,1986 10,1986 16,1986
10 10 19 23
0 1: (78.9) 0
35 34 52 64
7 6 48 6
(20.0) (17.6) (92.3) (9.4)
41 54 44 42 36
13 17 39 11 12
Jul24,1986 Ott 10,1986 Dee 12,1986
14 16 50
1 (7.0) 0 1.5 (30.0)
53 39 52
22 (41.5)
0
5” (9.6)
40 79 1%
Dee Mar Jun Sep Dee
26 18 29 1
l(3.8) 0 1 (3.4) 0
140 133 139 135 146
4 (2.9) 14 (10.5) Q(36.7) 60 (44.4) % (65.8)
267 284 276 270 275
are percentages
of test positives
in brackets
102 54 108 103 109
No. positive
6
17,198S 241986 23,1986 29,1986 l&l986
in the
1; (10.8) 3 (2.8) 20 (19.4) l(O.9) (31.7) (315) (88.6) (26.2) (33.3)
1: (7.0) 160 107 165 153 216
(seroprevalence)
A statistically significant association was found between an M. bovis seropositive test and being from farm IV, when adjusted for the possible confounding effects of age and breed (pcO.0001). As there was no sign&ant relationship between breed and an M. bovis seropositive test, both the breed main effect and the associated two-way interaction terms were dropped from the final logistic regression model.
DISCUSSION Although the ELISA for detecting exposure to Mycophma is a well researched and documented test (Carroll et aZ., 1976; Boothby er al., 1983; Thomas et al., 1987), the sensitivity and specitkity are not precisely known. However, Thomas et al. (1987) found that all of 19 bovine sera obtained from cows with confirmed M. bovis mastitis were positive by ELISA for M. bovis. All 127 negative reference sera obtained from
(60.0) (37.7) (59.8) (56.7) (78.6)
283
the University of California, Davis dairy herd in the same study tested negative by M. bovis ELISA. This herd had been tested repeatedly for the presence of M. bovis antibodies for a number of years without yielding any positive results; These results suggest that the M. bovis ELISA test has a sensitivity and specificity of close to 100% compared to culture and milk testing. TABLE II Crude and adjusted associations between age, farm location and positive antibody reaction to Mycophma bovis in dairy cattle sampled from the central valley of California, December 19@ja
Variablesb
Crude associations
Adjusted associationsd
OR
95% CIC
Coefficient
OR
95% CI
Heifers cows
2.43 1.80
(1.25,4.73) (0.%,3.37)
-2.0028 -1.4697
0.14 0.23
(0.06,0.31) (0.11,0.49)
Farm 3 Farm 4
0.57 11.45
(0.40, 1.23) (6.90,18.99)
-0.8026 2.7661
0.45 15.90
(0.24,0.84) (9.17,27.78)
“n = 856 records having complete age and farm location information bAge is categorized into age (1) calves, age (2) heifers and age (3) cows; Farmid is categorized into 1 = farm I, 3 = farm III and 4 = farm IV ‘Confidence intervals not including 1 indicate statistical significance at the 5% level dAdjusted for the main effects OR = odds ratio Results are for statistically significant variables
Cattle positive for M. bovis were found throughout the four herds studied. The association between being a calf and having antibodies to M. bovis, found after adjustment for confounding effects of breed and farm, may be due to one or a combination of factors, including the presence of colostral antibodies, the involvement of M. bovis in the bovine respiratory diseases complex and/or arthritis and the possibility that the levels of antibodies in heifers and cows may be below the threshold detectable by the ELISA. Another possible explanation may be a cohort effect, whereby the heifers and cows did not experience much exposure while the calves experienced an unusually intense exposure sometime in their lives. Multiple logistic regression analysis revealed a statistically significant association between farms and M. bovis ELISA positive tests. Animals on farm IV, with a herd size greater than 500, were 11 times more likely to have a positive test to M. bovis ELISA than those from the other three farms (herd size 150-300). This supports the association between clinical mycoplasma mastitis and increasing herd size reported by
Thomas and Jasper (1982). These researchers found a 14 to P&fold greater risk of clinical mastitis in dairy herds with more than 350 cows than in smaller herds. They speculated that herd size may be an indirect measure of several factors of management, animal density, animal movement and environment. In larger herds, the probability of having at least one infected cow which may act as a source of infection to other cows is greater than in smaller herds. Furthermore, poor management is more likely to be a problem in larger herds than smaller herds. All these factors, independently or in combination, could put larger herds at greater risk of M. bovis infection. With the present study design, however, we cannot identify which of these factors is important. In diagnosing exposure to M bovis based on cross-sectional testing for antibodies, we determined point prevalence, without knowing exactly what had happened to the animals prior to the test date. Such point prevalence estimates will be influenced by both the persistence of antibodies and the incidence of seroconversion. Lastly, detection of antibodies implies exposure but does not necessarily mean infection. Thus caution must be taken in the interpretation of the results presented in this study and especially in extrapolating the findings to different populations. REFERENCES Bar-Moshe, B., 1964. The isolation of Mjcoplasma from an outbreak of bovine mastitis in Israel. Journal of Veterinary Medicine, 21,97-99 Behymer, D.E., Ruppanner, R, Brook, D., Williams, J.C. and Franti, C.E., 1985. Enzyme immunoassay for surveillance of Q fever.American Journal of Veterinary Research, 46,2314-2417 Boothby, J.T., Jasper, D.E., Lutz, H. and Rollins, M.H., 1982. Gel electrophoresisderived enzyme-linked immunosorbent assay of serum from cows resistant to and cows susceptible to challenge exposure with Mycoplasma bovis.American Journal of Veterinaty Research, 43,553-S% Boothby, J.T., Jasper, D.E. and Rollin, M.H., 1983. Characterization of antigens for mycoplasmas of animal origin. American Journal of Veterirtaty Research, 44,433-439 Boughton, E., 1979. Mycoplasma bovis mastitis. Veterinary Bulletin, 49, 377-387 Breslow, N.E. and Day, N.E., 1980. Statistical Methoak in Cancer Research Vol. 1. The Analysis of Case-Control Studies, (International Agency on Cancer, Lyon, France), pp. 338 Campbell, EA., 1983. Animal Health in Australia. Vol. 3. Numtional Deficiencies and Diseases of Livestock, (Australian Government Publishing Service, Canberra), pp. 171 Carroll, E.J., Rollins, M.R and Jasper, D.E., 1976. The immune response of rabbits to 3 strains of Mjcoplasma agalactiae var. bovi.r isolated from mastitic bovine udders. Cornell Veterinarian, 66, 143-151 Daniel, W.W., 1987. Biostatistics: A Foundation for Analysts in the Health Sciences, 4th edn., (John Wiley, New York) Dixon, WJ. and Brown, M.B., 1985. Biomedical Data Programs, (University of California Press, Berkeley, C-9 Fahey, J.L. and McKeIvey, E.M., 196.5. Quantitative determination of serum immunoglobulins in antibody-agar plate. Journal of Immunology, 94,&I-90 Gourlay, RN. and Howard, CJ., 1979. In J.G. Tully and RF. Whitcomb (eds.), The Mycoplasmas, Vol. 2, (Academic Press, New York), pp. 129 Hale, H.H., Hemboldt, C.F., Plastridge, W.N. and Stula, E.F., 1%2. Bovine mastitis caused by a Mjcoplasma species. Cornell Veterinarian, 52,582-591 Hjerpe, CA and Knight, H.D., 1972. Polyarthritis and synovitis associated with Mycoplasma bovimastitidis in feedlot cattle. Journal of the American Veterinary Medical Association, 160,1414-1418 Jasper, D.E., 1977. Mycoplasma and mycoplasma mastitis. Journal of the American Veterinary Medical Association, 170,1162-1172 Jasper, D.E., 1981. Bovine mycoplasmal ma&is. In: C.E. Comellius and B.F. Simpson (eds.), Advances in Veterinary Science, (Academic Press, New York), pp. 121-159 Jasper, D.E., Bushnell, RB. and Dellinger, J.D., 1974. Mycoplasma mastitis, a brief review. California Veterinarian, 28,12-14 Jasper, D.E., Dellinger, J.D. and Hakason, H.D., 1979. Prevalence of mycoplasmal mastitis in California. American Journal of Veterinaty Research, 40,1043-1047
285 Kteinbaum, D.G., Kupper, L.L. and Muller, ICE, 1987. Applied Regrekon and other Multivariate Me&&, (Duxbury Press, North Scivtate, MA, USA) Langsford, E.U., 1977. Mjcoplasma agalactiae sub. sp. bovis in pneumonia and arthritis of bovine. Canadian Journal of Comparative Medicine, 41,89-W Lemeshow, S. and Hosmer, D.W., 1982. Review of the goodness of fit statistic for use in the development of logistic regression models. American Journal of Epidemiology, 115,92-106 Rosen&l, S. and Martin, S.W., 19% The association between serological evidence of mycoplasma infection and respiratory diseases in feedlot cahes. Canadian Journal of Veterbwy Research, 50, 179-183 Stalheim, O.H. and Page, LA., 1975. Naturally occurring experimentally induced mycoplasmal arthritis of cattle. Journal of Clinical Microbiology, 2, X.5-168 Thomas, C-B. and Jasper, D.E., 1982. Mycoplasma mastitis and herd size factor. California Veterinarian, 36,l.5-16 Thomas, C.B., Jasper, DE. and Willebetg, P., 1982. Clinical bovine mycoplasmal mastitis - an epidemiologic study of factors associated with problem herds. Acta Veterinaria Scandinavica, 23, 53-64 Thomas, C.B., Jasper, D.E., Bcothby, J.T. and Dellinger, J.D., 1987. Detection of antibody specific to Mycopiasma bow% and Mycopkwna califonticum by enzyme-linked immunosorbent assay (ELISA). IsraeI Journal of Medical Sciences, 23,X23-728 Uhaa, IJ., Riemann, H.P., Thurmond, MC. and Franti, C.E., 1990. A seroepidemiological study on bluetongue virus in dairy cattle in the Central Valley of California. Veterinary Research Communications, 14,99-112 (Accepted: 27 February 1990)
