Use of an Enzyme-Linked Immunosorbent Assay to Monitor the Control of Staphylococcus aureus Mastitis T. M. GROVE and G. M. JONES1 Virginia Polytechnic Institute and State University Blacksburg 24061-()315 ABSTRACT

Conventional culture methods were used to evaluate the ability of an ELISA to identify Staphylococcus aureus IMI. The test was 96% accurate; sensitivity was 90%, and specificity was 97%. The test was used to screen preserved milk samples rapidly in 10 cooperator herds. Prevalence of IMI was >10% in 6 herds at the first test. Average prevalence of cows scoring +2 (suspect) and +3 (positive) was 12.6%. Prevalence declined during the 12-mo study. Incidence of new IMI decreased from 7.9% at 6 mo to 3.6% at 12 mo. Rinsing teat cup liners with a 25-ppm iodophor or l00-ppm chlorine solution reduced the presence of S. aureus on the milking machine liners by 97%. Elevated scores were correlated with increases in lactation number. Milk antibody concentrations changed quadratically with increasing sec. The sec increased as milk antibody concentration increased. In 38 dairy herds, bulk tank antibody tests reflected herd prevalence of S. aureus infection. The average prevalence was 15.0% in 87 herds in which all lactating cows were tested. (Key words: mastitis, enzyme-linked immunosorbent assay, Staphylococcus) INTRODUCTION

Mastitis is the most costly disease of dairy cattle. Losses have been estimated at $2 billion/yr (19), of which 70% was attributed to reduced milk yield from subclinical mastitis (5). A single instance of mastitis may cause a 10 to 25% decrease in milk yield (3), but

Received January 25, 1991. tccepted September 3, 1991. Department of Dairy Science. 1992 J Dairy Sci 75:423-434

losses have ranged from 3 to 50% (11). Nickerson and Heald (24) suggested that average loss per lactation for one infected quarter was 727 kg of milk. Staphylococcus aureus infection is a primary cause of contagious mastitis. These infections tend to be chronic subclinical infections and account for more than one-half of the major infections (10). Somatic cell counts tend to vary in cows with S. aureus infection, and antibody concentrations are elevated in an attempt to fight off the invading microorganism (1). After the organism establishes itself in the udder, it becomes extremely difficult to eliminate with antibiotic therapy. Deep-seated abscesses become established and are walled off by antibiotic-impermeable scar tissue (4). Staphylococcus aureus also produces capsules or pseudocapsules, which have been shown to bestow virulence to this microorganism (4). Transmission of S. aureus usually occurs during the milking process, when milk from infected cows contaminates the teat cup liners and transfers the infection to the next animal milked with that unit (9, 22). A practical mastitis control program must be economical, effective under most management conditions, prevent new IMI, and help to eliminate existing IMI (26). A reduction in incidence of IMI can be achieved only by reducing the occurrence of new IMI, by increasing the rate at which IMI are eliminated, or by replacing infected cows with uninfected animals (10). Dodd et al. (10) concluded that, until IMI previously contracted are eliminated, prevention of new IMI will have little effect on the incidence of IMI. Only 40% of IMI can be detected by milkers with a strip cup (10). To identify an infected animal and type of infection, milk samples from individual cows have to be cultured. Hemolysin and coagulase production are considered key characteristics for colony iden-

423

424

GROVE AND JONES

tification of S. aureus from bovine IMI (7). Some cows shed the organism intermittently. Others shed it on every test for several lactations and then shed it intermittently or stop shedding (28). During nonshedding periods, one may not be able to culture the organism from the milk sample even though it is present in the udder. In addition, conventional culture methods are labor intensive, require numerous types of selective media, and require incubation periods that hamper timely identification of infectious organisms. Sears et al. (29) found that culturing cows once detected only 74% of S. aureus infections, and culturing two samples identified only 90% of infections. An ELISA has been developed to detect specific antibodies that bind to a purified antigen fraction of S. aureus (1). This S. aureus milk antibody test (ProStaph, ProScience Corp., Sterling, VA) was 97% accurate in detecting IMI by S. aureus when compared with bacterial culture methods (16) 00 a panel of 30 samples submitted by four universities. The test had 92% sensitivity for positive cultures and 100% specificity for negative cultures. The ProStaph test offers the dairy industIy a fast, accurate diagnostic' tool that can screen large numbers of milk samples for S. aureus IMI (2). Our experiments were designed 1) to compare the ProStaph test with conventional culture methods for correct identification of cows with S. aureus IMI; 2) to evaluate the efficacy of sanitizing teat cup liners with a sanitizing solution (to reduce the numbers of S. aureus on liners as an alternative to milking infected cows last), which might assist in the prevention of new IMI; 3) to determine the usefulness of screening herds with the ProStaph test for identification of cows with S. aureus IMI and then using this information to implement management practices aimed at preventing the spread of S. aureus IMI; 4) to determine the prevalence of S. aureus IMI in more herds; and 5) to define the relationship between S. aureus antibody in herd bulk tank milk and prevalence of S. aureus IMI. We examined changes in sec, milk yield, and IMI rate in 10 herds during 1 yr. We also examined the prevalence of S. aureus infections in COJllJIlC1'Cial dairy herds tested through the DIU program. Journal of Daily ScieIIce VoL 75, No.2, 1992

MATERIALS AND METHODS

The ProStaph I (ProScience Corp., Sterling, VA) was used to screen individual cow DHIA milk samples from 10 commercial dairy herds in January 1989 to determine the prevalence of infection in each herd and to identify specific infected cows. Herd sizes ranged from 55 to 320 cows. Prior to initiating the study, herd visits were conducted to evaluate milking management practices and equipment, dry cow treatment, and cow environment. Cooperating herds were selected that used the following management practices: individual singleservice towels for drying teats after washing, effective postmilking teat dipping, and dry cow antibiotic therapy in all quarters. The most effective means of controlling S. aureus mastitis is to minimize or to eliminate conditions contributing to the spread of infection and those conditions that allow bacteria to penetrate the teat canal. Therefore, the following recommendations were presented to the 10 cooperator herds after the initial visits. Teats should be washed with a sanitizing solution before milking. A low volume of water should be used, and common cloth or sponge should not be used. Teats should be dried with a single-service paper or cloth towel An effective teat dip should be used, and at least the bottom one-half to two-thirds of each teat should be covered. All quarters should be treated with a commercial dry cow product at drying off. Fresh cows and heifers should not be milked with the same teat cup or claw unit used to milk cows with mastitis. Cows with S. aureus IMI should be segregated and milked last, or milking units should be backflushed or sanitized with 25 ppm of iodophor or 100 ppm of chlorine after milking any infected cow. Cows with established S. aureus infections should be considered for culling, especially those with low milk yields, high days open, or feet and leg problems. Cows with S. aureus IMI should be placed on a "do not breed" list and should be culled when milk yield declines. Herds were revisited after the second ProStaph analysis of all milking cows. Procedures were discussed with the dairy farmer for isolating infected cows at milking from the remainder of the herd. Management reply cards were sent to each dairy farm on two occasions

CONlROUlNG STAPHYWCOCCUS AUREUS MASTITIS

dming the study to determine what procedures had been put into practice. COmparison of S. Bureus Milk Antibody Test with Culture Results

A triaI compared the ProStaph test with culture for identification of infected cows. One hundred cows were sampled in five cooperator herds (20 cows per herd). Animals that had been identified previously by the ProStaph test as infected with S. aUTeus accounted for up to 10 of the samples from each herd The remainder of the samples (to make 20 samples per farm) were from randomly selected cows with ProStaph scores of 0 or +1. An aseptic composite milk sample (teats were predipped. dried with paper towels, and cleansed with alcohol on cotton swabs) and a preserved (potassium dichromate) composite milk sample were collected from each cow. Aseptic samples (.050 ml) were cultmed on blood agar to isolate colonies, and S. aUTeus was identified by morphology and hemolysis (7). The 4-h tube coagulase and catalase tests were used to differentiate S. aUTeus from other staphylococcal species. Preserved samples were analyzed using the ELISA ProStaph test (1). An MR 600 microplate reader (Dynatech Lab, Inc., supplied by ProScience Corp.) was used to obtain optical density readings. The optical densities of the samples were compared with those of positive and negative controls. Scores of 0, +1, +2, or +3 were assigned to each sample according to its relationship to the controls. A 0 score had an optical density less than the negative control. A +2 score was 85 to 100% of the optical density for the positive control, and a +3 score exceeded the optical density of the positive control Effectiveness of sanlUzlng Teat Cup Liners

Teat cup liners were rinsed with a sanitizing solution after cows with a ProStaph +3 score were milked. Liners were swabbed on 34 units from four different herds. Sterile cotton swabs were swirled four times around the liner and were moved from bottom to top at the same time. Units were dipped in or rinsed out with a 25-ppm iodine or lOO-ppm cblorine solution

425

(depending on parlor setup and the sanitizer being used on each farm). Milker units were allowed to drain for 5 to lOs. A second swab was used after the sanitizer rinse. Swabs were placed into screw10% ProStaph scores +2 and +3). Cows were retested in subsequent months based on the following guidelines. The July test group included all cows scoring +2 and all cows that had calved since the last DmA test date (fresh cows). In August, animals scoring 0 and +1 on a previous test were retested, and all fresh cows were tested. All cows in high incidence herds (>10%) were retested in September, and all fresh cows were tested in all herds. October and November tests were conducted only on all fresh cows. In December, all cows in alllO herds were retested for a third complete herd test. Results were reported to the dairy farms as soon as they were available. Only data from the three complete herd tests were used in defining incidence of new IMI. Reladonshlp Between Bulk Tank Antibody Ratios and Herd Infection Rates

The ProStaph test was conducted by the DID laboratory on bulk. tank milk samples from 38 Virginia dairy herds followed by ProStaph testing all lactating cows at the subsequent DIn test. The relationship between herd Jomnal of Dairy Science Vol. 75, No.2, 1992

426

GROVE AND JONES

prevalence of S. aureus antibody and bulk tank milk antibody ratio (to positive control) and herd average Dill sec score was determined. Statistical Analysis

General linear modeling was used for the analysis of the ProStaph score, milk yield, and see in the cooperator field study (27). The DID records were accessed by telephone to obtain DIM, lactation number, and milk yield. Animals less than 30 DIM and those producing less than 13.6 kg/d were excluded from the data set to eliminate the possibility of falsepositive reactions to the ELISA antigen (2). The prevalence of IMI and the incidence of new IMI were determined for all herds and reported as means and standard deviations. The model for the analysis of the ProStaph score was

where Yijkl Hi Tj b i to b4

= ProStaph score for COWijld; = effect of herd; = effect of month of test; = regression of ProStaph score

bs

=

b6 to bg

=

on milk yield (M), see (S), and lactation number (L) for COWijkI; regression of ProStaph score on the product of M and DIM (0) for cowijkI; and quadratic regression of ProStaph score on milk yield M, S, and L for cowijld'

A similar model was used to analyze changes in milk yield and sec for COWijkJ, and this model also included the ProStaph score and the interaction of the ProStaph score and month of test. The interactions between the ProStaph score x sec and ProStaph score x milk yield were analyzed when milk yield and see, respectively, were the dependent variables. The Dill laboratory accidentally discarded milk samples from some herds before an aliquot was obtained for the ProStaph test. These herds were 1 mo off the testing schedule for Journal of Dairy Science Vol. 75, No.2. 1992

the remainder of the study. Month codes were developed to put all herds on the same basis for comparison. Month code 1 grouped all herds tested for the first time in June or July. Cows retested in August or September were assigned to month code 2, and the third month code grouped the data from the December or January test. RESULTS

Comparison Between Culture and ProStaph Results

Results comparing conventional culture methods and the ELISA ProStaph test for 97 milk samples are summarized in Table 1. The ProStaph test was 96% accurate in predicting culture results. The ProStaph test correctly identified 18 of the 20 cows infected with S. aureus, a sensitivity of 90%. The ProStaph test also identified 75 of the 77 cows without S. aureus infections, a 97% specificity. Two milk samples were S. aureus antibody negative by the ProStaph test but were culture positive; two other samples were culture negative and antibody positive. These discrepancies could have been the result of an early IMI or the leakage of blood antibodies into the milk from another S. aureus IMI. Three milk samples were contaminated and were disregarded in the final results. Sterilization of Teat Cup LIners

Staphylococcus aureus were reduced on teat cup liners by sanitization (fable 2). Rinsing

TABLE 1. Comparison of milk antibody test (ProSlapb) scores with culture resulls 1 in !be identification of StaphylococCILS aureus isolated from 97 milk samples from five berds. 2

EUSA ProStaph

S. aureus culture

Total

+

+

18 2 20

Total

2

20

75

n

77

97

IPositive (+) or negative (-) culture results. 2A CCIU'8C)' = (18 + 75)/97 =96%; sensitivity = 18/20 = 90%; specificity 7Sm 97%.

=

=

427

eONfROllJNG STAPHYLOCOCCUS AUREUS MASTITIS TABLE 2. Effects of sanitizing teat cup liners on the recovery of Staphylococcus aureus. Treatment leat cup liners

n

Culture positive for S. aureus

Presanitize Postsanitize

34 34

18 I

Gram-positive bacilli

TABLE 3. Average lactation number, milk yield, and see by milk antibody test (ProStaph) score from the second to third herd test Lactation number

Milk

0 +1 +2 +3

2.1 2.9 3.2 3.5

28.0 27.7 27.4 30.0

teat cup liners with a sanitizing solution of 25 ppm of iodophor or of 100 ppm of chlorine reduced the number of S. aureus colonies recovered from the liners by 97% compared with presanitized liners. There were 18 culturepositive samples on teat cup liners after milking 34 ProStaph-positive cows, but S. aureus was found on only 1 unit after sanitizing. Milk remained in the liner of this culture-positive unit. Gram-positive bacilli and coliforms, mainly Escherichia coli, were common isolates from postsanitized liners. There were no detectable differences between sanitizing with iodophor or chlorine solutions. The ContrOl of S. aureus Infections In Cooperator Herds

Score

(kg)

(no.) 10 9

see

ProStaph score

2.3 3.6 4.3 4.9

relationship (P > .05) to ProStaph score. The mean sec increased as the ProStaph score increased from 0 to +3 (Table 3). A curvilinear relationship (P < .01) existed between ProStaph score and sec (Figure 1). The relationship between ProStaph score and sec was highly significant (P < .01), which indicated that the ProStaph score was associated with increased sec. At the initial screening of the 10 herds, the prevalence of S. aureus averaged 8.7% for a ProStaph score of +3 and 3.9% for a ProStaph score of +2 (fable 4). Prevalence was less than 10% in 5 herds (ProStaph scores of +2 and +3). The average prevalence of infection declined slightly from 12.6 to 11.9% by the second complete herd test. At the second test, 6 herds had high prevalence (>10% IMI), and 4 herds were classified as low prevalence «10% IMl). Incidence of new IMI between the first and second complete herd tests averaged 7.9%. New heifers entering the herd between these tests were considered to have new IMI only if they had ProStaph scores of +2 or +3 when more than 30 DIM. Older cows were considered to have new IMI postpartum if the ProStaph score had been 0 or +1 before drying

ProStaph score, milk yield, and sec differed among herds (P < .01). The DIM and lactation number were highly associated (P < .01) with variations in milk yield, as were the interaction between ProStaph score and test month. Some of this variation (P < .05) in milk yield was explained by the month in which the ProStaph test was conducted. Daily milk yield was higher (P < .05) in June or July (month code 1) and in August or September (month code 2) than it was in December or January (month code 3). ProStaph score alone off. was not associated (P > .05) with milk yield Control recommendations had been made to variations, but a ProStaph score of 0 was asso- each dairy farmer but were implemented on a ciated with the lowest mean milk yield, and a voluntary basis. After the second complete score of +3 was associated with the highest herd test, additional control procedures were mean milk yield. This may be attributed to incorporated into the milking program for six confounding of lactation number, because a herds to help prevent the spread of IMI via positive ProStaph score had the highest mean surface of the teat cup liners (hygiene group). lactation number as well as the highest mean Managers for three of these herds visually milk yield. marked (leg band, colored tape, etc.) infected Elevated ProStaph scores were highly asso- animals or posted a list of cow numbers in the ciated (P < .01) with increases in lactation parlor and manually sanitized the teat cup number. The DIM and milk yield showed no liners after milking infected animals. For 2

Journal of Dairy Science Vol. 75, No.2, 1992

428

GROVE AND 10NES

3r-----------------.

o

2

3

4

5

sonTIC au

e

7

e

9

W

COUNT

Figure 1. Estimated milk antibody test (ProStaph) score for an average herd at various sec levels (P < .01).

herds, dairy fanners also visually identified cows with S. aureus IMI but separated them at milking and milked them last. The dairy fanner for 1 herd segregated infected cows into an S. aureus group that was housed separately and milked last. Reply cards indicated that all dairy fanners culled some of the cows identified as infected with S. aureus during the study. Average prevalence of S. aureus IMI dropped significantly (P < .01) to 8.2% by the third complete herd test (Table 4). The incidence of new IMI averaged 3.6% from the second to third herd test. Over this period. chronically infected cows (cows with ProStaph score of +3 on two or more tests) averaged 6.9 ± 4.8%. At the third herd test, less than 5% of the cows in 5 herds were infected (ProStaph scores of +2 or +3), and less than 10% of the cows in another herd were infected. Two of the 4 herds with greater than 10% prevalence had reduced prevalence from previous levels (23.8 to 12.2% and 21.4 to 11.9%). The number of

cows with ProStaph scores of +2 and +3 (16.7 to 15.7% and 16.3 to 16.1%) was essentially unchanged for the remaining 2 herds. Nine of the herds averaged less than 5% incidence of new IMI between the second and third screenings. The 10 herds had an average culling rate of 26.0 ± 18.2% (range 5.1 to 57.1%) of the cows with +3 scores. The overall culling rate for these herds averaged 35.1 ± 8.7%. An average of 5.4 ± 3.2% of the herd was culled for S. aureus antibody. The cows culled were lower producers. Their DIn 305-d mature equivalent milk averaged 685 ± 1610 kg below herd average. The cows with S. aureus scores of +3 that were not culled had a 305-d mature equivalent milk yield that was 14 ± 1311 kg above the herd average. The 6 herds in the hygiene group had culled a greater percentage of infected cows with +3 scores (25% by the second test before a hygiene program and 36% by the third test). For the 4 herds in the control group, 10 and 11% of the cows with +3 scores were culled by the second and third tests, respectively. Herds from which a greater percentage of cows with +3 scores were culled had lower rates of new IMI (r = -.66, P < .05). The correlation between prevalence of cows with +3 scores at the third herd test and the percentage of cows with +3 scores culled was -.64 (P < .05). Culling of cows with a +3 score had a significant effect on the prevalence and incidence of new IMI. Many of these cows (71 %) had milk yield below herd average. In the 6 herds in the hygiene group, the prevalence of IMI (ProStaph scores of +2 and +3) decreased from 7.6% at the first test to 4.9% by the third test. Prevalence in the 4 control group herds averaged 20.1, 16.2, and 13.0% at the three complete herd tests. At the

TABLE 4. Prevalence and incidence of SUlplrylococcus aureus IMI and average l-yr period.

see

in 10 cooperator berds over a

IDitial

IMI

screening (Ian) X

St>

Second test (lun or lul)

Third test (Dec or Ian)

X

X

SD

12 7.0 3.6

.9 5.1 2.8

SD

(% of cows on study) ProStaph score

+2 +3

3.9 8.7

2.6 8.1

Incidence of new IMl Ioumal of Dairy Science VoL 75. No.2, 1992

4.4 7.5 7.9

2.8 5.0 4.6

CONfROLLING STAPHYLOCOCCUS AUREUS MASTITIS

second and third herd tests, the average incidence of new IMI was greater in the herds that did not implement control practices (9.8 and 5.4%) compared with the herds that segregated cows or sanitized milking units after milking cows infected with S. aureus (6.6 and 2.3%). At the second test, spontaneous or treatment cure rates (from scores of +3 to either 0 or +1) were 38 and 32% in the control and hygiene herds. By the third test, cures rates were 39% in the control herds and 58% in the hygiene herds. Although all 10 dairy farmers used such recommended practices as teat dipping, dry cow therapy, and washing and drying teats with single-service towels, the 6 hygiene dairy farmers were willing to implement other procedures for controlling mastitis, including hygiene and culling; and they had higher cure rates. However, these cure rates should be interpreted with caution. Some cows with +3 scores followed by 0 or +1 scores were later found to have +3 scores again. Whether these cows were cured and reinfected or remained infected is not known. There were 300 primiparous cows tested during the 1st or 2nd mo postpartum. Infections (+2 or +3 scores) were found in 15 (5%), but most of these retested 0 or +1. The control and hygiene groups averaged 6.2 and 5.4% for the period preceding the second test and 3.3 and 4.5% preceding the third test. The correlation coefficient between herd prevalence for a ProStaph +3 score and percentage of chronically infected cows or percentage of incidence of new IMI was .68 (P < .05) and .12 (P > .05), respectively. The 4 herds with 12% ProStaph +3 or higher scores had 15 to 27% of cows with chronic ProStaph +3 scores. The 5 herds with the lowest Dill see had ProStaph scores that averaged 6.3 ± 3.8% +3, 8.1 ± 7.8% chronic +3, and 7.0 ± 4.5% incidence of new +3 IMI. The 5 herds with higher Dill sec had an average prevalence of 17.5 ± 13.8% of +3 IMI, 13.0 ± 10.7% of chronic +3 IMI, and 8.7 ± 5.0% of new +3 IMI. Infection Status of ProStaphPositive Cows In Milk Less Than 30 d

The possibility exists for false-positive ProStaph results among cows less than 30 DIM.

429

For 12 animals at less than 30 d postpartum, the ProStaph test was compared with culture. Milk samples from 2 of these animals scored ProStaph +3 and cultured positive for S. aureus. Eight animals produced milk with ProStaph +1 scores and were negative for S. aureus. The remaining 2 cows were both ProStaph- and culture negative. Nine primiparous animals within 30 d postpartum scored ProStaph +2 or +3 on their first test Three remained ProStaph +2 or +3 on the second test, and the others decreased to +1. The change probably was not in response to antIbiotic therapy because the dairy farmers apparently did not treat cows based only on ProStaph test results. The change may have been the result of spontaneous recovery or false-positive test results. Thirty-eight multiparous animals scored ProStaph +2 or +3 within 30 d of parturition, 18 cows (47%) continued to have antibody scores of +3 on the next test, and 17 cows decreased to scores of 0 or +1. The 23 heifers and cows that changed to scores of 0 or +1 were significant because they increased the prevalence of IMI by 34% in the 10 herds at each of the three complete herd test periods. A herd screening of all cows to determine herd prevalence should not include cows less than 30 DIM RelationshIp Between ProStaph

SCore and

see

There were 102 milk samples with ProStaph scores of +2. Of these, 47% had see of less than 3 x lOS or see linear scores of 4 or less. There were 261 milk samples with ProStaph scores of +3; 36% of these had low see. Also from these 10 herds, 13% of the cows had see higher than 5 x lOS but ProStaph scores of 0 or +1. Repeatability 01 ProStaph +2 Scores

Forty-eight cows with ProStaph scores of +2 (suspects) were retested. The subsequent ProStaph score decreased in 54%, remained +2 for 25%, and increased to +3 for 21%. We recommended retesting any cow with a ProStaph score of +2 before decisions were made regarding culling any of these cows based on S. aureus or adding these cows to a separate S. Joumal of Daily Science Vol. 75, No.2, 1992

430

GROVE AND IONES

TABLE 5. Multiple regression analysis of herd Staphylococcus aureus prevalence to bulk tank S. aureus antibody ratio and Dill see parameters. S. aureus +3 1

S. aureus +2, +3 b}

Scores

r

Bulk tank antibody ratio Dill Herd average see score see Scores, % 4 5 6 7 to 9

.87 .54

.82 .27

.41 .30 .41 .46

.14

r

bl

.82 .52

.78 .29

.34

.45

.24

.34

-.09

.39 .38

.18 -.15 .03 -.23

Ir = Linear correlation coefficient, bl = standard partial regression coefficient.

aureus group. We recommended backflushing the milking unit with disinfectant after milking infected cows as soon as the scores were detected. Tests Conducted by DHIA on Producer Samples

In May 1990, DHIA offered free ProStaph tests on bulk tank milk samples. Dairy fanners could follow up and have all lactating cows tested for a lower fee. We had access to bulk tank ProStaph, complete herd ProStaph tests, and herd sec score for 38 herds. The multiple regression model that included bulk tank milk antibody ratio (to a positive control), herd average sec score, and percentage of cows in the herd with sec scores of 4, 5, 6, and 7 to 9 accounted for 80% of the variability in percentage of cows in the herd positively infected with S. aureus (ProStaph score of +3) and 72% of the variability for cows with ProStaph scores of +2 and +3 (P < .01). Bulk tank antibody ratio alone accoWlted for 76 and 67%, respectively. Standard partial regression coefficients indicated that bulk tank milk antibody ratio had the greatest association with herd S. aureus prevalence (Table 5). The correlation between percentage of herd S. aureus prevalence and bulk tank antibody ratio was .87 and .82, respectively, for the percentage of scores +3 alone and the percentage of +2 and +3 scores combined. Correlations to Dill sec parameters were lower. The percentage of herd prevalence (ProStaph score of +3) could be predicted by Y = 3.50 + 17.25 X, where X was the bulk tank antibody concentration. The 38 herds were categorized according to an average Dill sec score 30% prevalence) were found in 10% of the bulk tank samples. Virginia DHIA tested all lactating cows in 87 Virginia dairy herds, which included 7546

431

CONTROLLING STAPHYLOCOCCUS ItUREUS MAS1TTIS

TABLE 6. Bulk tank Staphylococcus aureus antibody ratio and herd prevalence of IMI for 38 dairy herds with low or high sec. Herd average DHI SCC score

No. herds SCC Score (± SD) Bulk tank antibody ratio (± SO) Percentage of herds wilh

.99 Prevalence (percentage of cows with ProStaph +3) Percentage of herds