Advertisement

The effect of mastitis management input and implementation of mastitis management on udder health, milk quality, and antimicrobial consumption in dairy herds

  • M. Stevens
    Affiliations
    M-team and Mastitis and Milk Quality Research Unit, Department of Reproduction, Obstetrics and Herd Health, Faculty of Veterinary Medicine, Ghent University, 9820 Merelbeke, Belgium
    Search for articles by this author
  • S. Piepers
    Affiliations
    M-team and Mastitis and Milk Quality Research Unit, Department of Reproduction, Obstetrics and Herd Health, Faculty of Veterinary Medicine, Ghent University, 9820 Merelbeke, Belgium
    Search for articles by this author
  • S. De Vliegher
    Correspondence
    Corresponding author
    Affiliations
    M-team and Mastitis and Milk Quality Research Unit, Department of Reproduction, Obstetrics and Herd Health, Faculty of Veterinary Medicine, Ghent University, 9820 Merelbeke, Belgium
    Search for articles by this author
Open ArchivePublished:January 25, 2019DOI:https://doi.org/10.3168/jds.2018-15237

      ABSTRACT

      The main objective of this study was to evaluate evolutions in herd-level antimicrobial consumption (AMC) and in udder health and milk quality parameters between herds that received mastitis management input on a regular basis (actively advised by the first author; referred to as intervention herds) and herds that did not (referred to as control herds). Strikingly, herds in the intervention group had a significantly higher prevalence of new intramammary infections compared with those in the control group. No significant differences were observed in the percentage of chronically infected cows, the bulk milk somatic cell count, and the bacterial and coliform count between the intervention and control herds, nor did the herd-level AMC differ between them. Furthermore, the level of mastitis management applied in each herd was assessed and scored [mastitis management score (MMS); higher is better], as was the level of implementation of different recommended mastitis management practices over time, expressed as the mastitis management implementation score (MMIS; higher is better). A large variation was observed in MMS and MMIS in the intervention herds (median = 16 and range = 12 to 22; median = 13 and range = −5 to 31, respectively) and the control herds (median = 16 and range = 9 to 22; median = 9 and range = −13 to 22, respectively). Also, intervention herds in which the herd veterinarian attended each herd visit executed by the first author had a higher MMS and MMIS (median = 20 and 24, respectively) compared with herds in which the veterinarian sometimes (median = 16 and 17, respectively) or never (median = 16.5 and 7.5, respectively) attended the herd visits. Further, the association between MMS or MMIS on one hand and udder health, milk quality, and the herd-level AMC over time on the other was studied using the data of both groups of herds. Better mastitis management was associated with a reduction in the consumption of antimicrobials that are critically important for human health over time and with lower bacterial counts and bulk milk somatic cell count. Better mastitis management can be helpful in obtaining better milk quality and more responsible use of critically important antimicrobials on dairy farms.

      Key words

      INTRODUCTION

      The concern about imprudent use of veterinary antimicrobials is increasing as there is some evidence that antimicrobial consumption (AMC) in animals poses a risk to public health (
      • Marshall B.M.
      • Levy S.B.
      Food animals and antimicrobials: Impacts on human health.
      ;
      • Aarestrup F.M.
      The livestock reservoir for antimicrobial resistance: A personal view on changing patterns of risks, effects of interventions and the way forward.
      ;
      • Chang Q.
      • Wang W.
      • Regev-Yochay G.
      • Lipsitch M.
      • Hanage W.P.
      Antibiotics in agriculture and the risk to human health: How worried should we be?.
      ) through the selection for and spread of antimicrobial resistance (
      • Chantziaras I.
      • Boyen F.
      • Callens B.
      • Dewulf J.
      Correlation between veterinary antimicrobial use and antimicrobial resistance in food-producing animals: A report on seven countries.
      ;
      • Callens B.
      • Faes C.
      • Maes D.
      • Catry B.
      • Boyen F.
      • Francoys D.
      • de Jong E.
      • Haesebrouck F.
      • Dewulf J.
      Presence of antimicrobial resistance and antimicrobial use in sows are risk factors for antimicrobial resistance in their offspring.
      ). Lowering AMC in farm animals therefore can be an effective strategy for reducing antimicrobial resistance (
      • Agersø Y.
      • Aarestrup F.M.
      Voluntary ban on cephalosporin use in Danish pig production has effectively reduced extended-spectrum cephalosporinase-producing Escherichia coli in slaughter pigs.
      ), as was recently evidenced by
      • Dorado-García A.
      • Mevius D.J.
      • Jacobs J.J.
      • Van Geijlswijk I.M.
      • Mouton J.W.
      • Wagenaar J.A.
      • Heederik D.J.
      Quantitative assessment of antimicrobial resistance in livestock during the course of a nationwide antimicrobial use reduction in the Netherlands.
      . In recent years, several national and international antimicrobial-reducing policy measures were introduced to heed and reduce the AMC in livestock (
      • AMCRA (Antimicrobial Consumption and Resistance in Animals)
      Visie 2020.
      ;
      • WHO (World Health Organization)
      Global Action Plan on Antimicrobial Resistance.
      ;
      • Kuipers A.
      • Koops W.J.
      • Wemmenhove H.
      Antibiotic use in dairy herds in the Netherlands from 2005 to 2012.
      ;
      • Dupont N.
      • Diness L.H.
      • Fertner M.
      • Kristensen C.S.
      • Stege H.
      Antimicrobial reduction measures applied in Danish pig herds following the introduction of the “Yellow Card” antimicrobial scheme.
      ).
      In the dairy industry, the majority of antimicrobials are used for the prevention and control of mastitis (
      • Mitchell J.M.
      • Griffiths M.W.
      • McEwen S.A.
      • McNab W.B.
      • Yee A.J.
      Antimicrobial drug residues in milk and meat: Causes, concerns, prevalence, regulations, tests, and test performance.
      ;
      • Pol M.
      • Ruegg P.L.
      Treatment practices and quantification of antimicrobial drug usage in conventional and organic dairy farms in Wisconsin.
      ;
      • González S.
      • Steiner A.
      • Gassner B.
      • Regula G.
      Antimicrobial use in Swiss dairy farms: Quantification and evaluation of data quality.
      ;
      • Ivemeyer S.
      • Smolders G.
      • Brinkmann J.
      • Gratzer E.
      • Hansen B.
      • Henriksen B.I.F.
      • Huber J.
      • Leeb C.
      • March S.
      • Mejdell C.
      • Nicholas P.
      • Roderick S.
      • Stoeger E.
      • Vaarst M.
      • Whistance L.K.
      • Winckler C.
      • Walkenhorst M.
      Impact of animal health and welfare planning on medicine use, herd health and production in European organic dairy farms.
      ;
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ). Although mastitis prevention and control programs have been available since 1960 and have been successfully implemented in many herds (
      • Hillerton J.E.
      • Bramley A.J.
      • Staker R.T.
      • McKinnon C.H.
      Patterns of intramammary infection and clinical mastitis over a 5-year period in a closely monitored herd applying mastitis control measures.
      ), mastitis remains a major issue in dairy production. Participation of dairy farmers in workshops or implementation of web-based tools as part of a veterinary herd health management program as well as use of external input (e.g., via a nutritionist or veterinarian, participation in DHI programs, regular pregnancy checks, AI in lactating cows) had a positive effect on the milk quality status of those dairy herds, as reflected by a decreased bulk milk SCC (BMSCC;
      • Cicconi-Hogan K.M.
      • Gamroth M.
      • Richert R.
      • Ruegg P.L.
      • Stiglbauer K.E.
      • Schukken Y.H.
      Associations of risk factors with somatic cell count in bulk tank milk on organic and conventional dairy farms in the United States.
      ;
      • Steeneveld W.
      • Velthuis A.G.
      • Hogeveen H.
      Short communication: Effectiveness of tools provided by a dairy company on udder health in Dutch dairy farms.
      ;
      • Tremetsberger L.
      • Leeb C.
      • Winckler C.
      Animal health and welfare planning improves udder health and cleanliness but not leg health in Austrian dairy herds.
      ). Other studies demonstrated that veterinary herd health management support or participation in small focused farmer groups resulted in a lower AMC on both pig farms (
      • Postma M.
      • Vanderhaeghen W.
      • Sarrazin S.
      • Maes D.
      • Dewulf J.
      Reducing antimicrobial usage in pig production without jeopardizing production parameters.
      ) and (primarily organic) dairy farms (
      • Ivemeyer S.
      • Maeschli A.
      • Walkenhorst M.
      • Klocke P.
      • Heil F.
      • Oser S.
      • Notz C.
      Effects of a two-year dairy herd health management programme on udder health, use of antibiotics and longevity.
      ;
      • Bennedsgaard T.W.
      • Klaas I.C.
      • Vaarst M.
      Reducing use of antimicrobials—Experiences from an intervention study in organic dairy herds in Denmark.
      ;
      • Kuipers A.
      • Koops W.J.
      • Wemmenhove H.
      Antibiotic use in dairy herds in the Netherlands from 2005 to 2012.
      ;
      • Speksnijder D.C.
      • Graveland H.
      • Eijck I.
      • Schepers R.W.M.
      • Heederik D.J.J.
      • Verheij T.J.M.
      • Wagenaar J.A.
      Effect of structural animal health planning on antimicrobial use and animal health variables in conventional dairy farming in the Netherlands.
      ) compared with herds not influenced by such programs. Apart from the last 2 studies conducted in the Netherlands (
      • Kuipers A.
      • Koops W.J.
      • Wemmenhove H.
      Antibiotic use in dairy herds in the Netherlands from 2005 to 2012.
      ;
      • Speksnijder D.C.
      • Graveland H.
      • Eijck I.
      • Schepers R.W.M.
      • Heederik D.J.J.
      • Verheij T.J.M.
      • Wagenaar J.A.
      Effect of structural animal health planning on antimicrobial use and animal health variables in conventional dairy farming in the Netherlands.
      ), none of the abovementioned studies had a control group to compare with, which is obviously a prerequisite for investigating the effect of veterinary herd health management on AMC and udder health on dairy farms in a scientifically sound manner.
      It is well known how animal diseases, including mastitis, can be substantially reduced (
      • LeBlanc S.J.
      • Lissemore K.D.
      • Kelton D.F.
      • Duffield T.F.
      • Leslie K.E.
      Major advances in disease prevention in dairy cattle.
      ). Nevertheless, the challenge is to give appropriate and herd-specific recommendations based on herd-specific risk factors determining a herd's udder health. Moreover, the implementation of the recommended management practices needs to be motivated by the farmer (
      • LeBlanc S.J.
      • Lissemore K.D.
      • Kelton D.F.
      • Duffield T.F.
      • Leslie K.E.
      Major advances in disease prevention in dairy cattle.
      ;
      • Vaarst M.
      • Bennedsgaard T.W.
      • Klaas I.
      • Nissen T.B.
      • Thamsborg S.M.
      • Ostergaard S.
      Development and daily management of an explicit strategy of nonuse of antimicrobial drugs in twelve Danish organic dairy herds.
      ;
      • Green M.J.
      • Leach K.A.
      • Breen J.E.
      • Green L.E.
      • Bradley A.J.
      National intervention study of mastitis control in dairy herds in England and Wales.
      ;
      • Ivemeyer S.
      • Smolders G.
      • Brinkmann J.
      • Gratzer E.
      • Hansen B.
      • Henriksen B.I.F.
      • Huber J.
      • Leeb C.
      • March S.
      • Mejdell C.
      • Nicholas P.
      • Roderick S.
      • Stoeger E.
      • Vaarst M.
      • Whistance L.K.
      • Winckler C.
      • Walkenhorst M.
      Impact of animal health and welfare planning on medicine use, herd health and production in European organic dairy farms.
      ;
      • Postma M.
      • Stark K.D.
      • Sjolund M.
      • Backhans A.
      • Beilage E.G.
      • Losken S.
      • Belloc C.
      • Collineau L.
      • Iten D.
      • Visschers V.
      • Nielsen E.O.
      • Dewulf J.
      Alternatives to the use of antimicrobial agents in pig production: A multi-country expert-ranking of perceived effectiveness, feasibility and return on investment.
      ;
      • Speksnijder D.C.
      • Jaarsma A.D.
      • van der Gugten A.C.
      • Verheij T.J.
      • Wagenaar J.A.
      Determinants associated with veterinary antimicrobial prescribing in farm animals in the Netherlands: A qualitative study.
      ;
      • Tremetsberger L.
      • Winckler C.
      Effectiveness of animal health and welfare planning in dairy herds: A review.
      ). In this regard, the average rate of implementation of recommended mastitis management measures that have been proposed and discussed varies in the literature between 43% (
      • Brinkmann J.
      • March S.
      Animal Health in Organic Dairy Farming—Health State as Well as Development, Application and Evaluation of a Preventive Herd Health Planning Concept.
      ) and 57% (
      • Tremetsberger L.
      • Leeb C.
      • Winckler C.
      Animal health and welfare planning improves udder health and cleanliness but not leg health in Austrian dairy herds.
      ). The latter stresses the need to take the level of implementation into account when the effect of receiving mastitis management input (MMI) on the herd's performances and AMC is investigated.
      The main objective of this study was to describe evolutions in herd-level AMC and differences in udder health and milk quality parameters between herds that received MMI on a regular basis (referred to as intervention herds) and herds that did not (referred to as control herds). Further, the level of mastitis management applied in each herd in the middle of the project (March 2013) was scored, as was the level of implementation of the different recommended mastitis management practices between the start (January 2012) and the end (April 2014) of the project. Subsequently, the association between the herd-level (implementation of) mastitis management (over time) and the herd-level udder health, milk quality, and AMC over time was studied. Also, the level of implementation of recommended mastitis management practices was compared between herds in the intervention group and herds in the control group. Finally, the influence of the involvement of the herd veterinarian in intervention herds in the project on the implementation of recommended mastitis management practices was assessed.

      MATERIALS AND METHODS

      Data Collection

      Study Design, Initial Data Set, and Data Handling

      Part of the data used in this study were described previously (
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ), and an overview of the project design is presented in Figure 1. In short, the data originate from a convenience sample of 56 Flemish dairy herds, all participating in the Flemish DHI program, who volunteered to be part of this project. The 56 participating dairy herds were randomly divided at the start of the project into an intervention group (n = 28) that received MMI by the first author and a control group (n = 28) that did not receive MMI during the study period (Figure 1). Udder health, milk quality, and AMC data were available for all 56 herds during the entire study period. Data on the mastitis prevention and control practices implemented on the farm were obtained through a face-to-face interview with the farmer (more information is available in
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ; questionnaire available on request) during the first herd visit in January 2012 (start of the project), the herd visit in March 2013 (middle of the project), and the final herd visit in April 2014 (end of the project; Figure 1). The questions included general herd information and information related to the milking routine, the milking machine, the housing of the lactating cows, treatment strategies of cases of clinical mastitis and subclinical mastitis, dry cow management, culling policy, heifer management, general herd health status, and participation in a veterinary herd health management program. All questions referred to management practices in place during the study period (i.e., 2012–2014). Additionally, milk quality parameters were obtained from 275 randomly selected, nonparticipating Flemish dairy herds, referred to as the external group, that served as an additional control group. The milk quality parameters of this external group were used in this study to position the 56 participating herds compared with the larger population throughout the study period.
      Figure thumbnail gr1
      Figure 1Overview of the study design from January 2012 to April 2014. Δ = difference; ATI = antimicrobial treatment incidence (overall and for antimicrobials that are critically important to human health); IRTM = incidence rate of treated mastitis cases; MMI = mastitis management input; MMS = mastitis management score; MMIS = mastitis management implementation score; BMSCC = bulk milk SCC; CC = coliform count; BC = bacterial count.

      Udder Health

      All 56 herds participated in the Flemish DHI program, and therefore individual composite SCC data were available on a 4- to 6-wk basis. Using these data, the average monthly herd-level prevalence of chronically infected cows (chronic IMI) and newly infected cows (new IMI) was calculated between March 2013 and April 2014, further referred to as study period 2 (Figure 1), as previously described (
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ). For each herd, a maximum of 11 measurements was used (Table 1). An ln transformation of the prevalence of chronic IMI and of the prevalence of new IMI was performed to normalize the data.
      Table 1Overview of the type of data used to calculate udder health parameters, milk quality parameters, and antimicrobial consumption, and the reference period used to evaluate the effect of the independent variables mastitis management score (MMS) and mastitis management implementation score (MMIS)
      Independent variable
      Y = the association with the outcome variable was studied; N = the association with the outcome variable was not studied.
      Outcome variablePeriodMMSMMIS
      Udder health
       Prevalence of new IMI (%)April 2013
      Dairy Herd Improvement data for March 2013 were not available; therefore, the baseline was April 2013.
      YN
       Prevalence of chronic IMI (%)April 2013
      Dairy Herd Improvement data for March 2013 were not available; therefore, the baseline was April 2013.
      YN
       Δ IRTM
      Difference in herd-level incidence rate of treated mastitis cases.
      January 2012 to February 2013 (study period 1)YN
      Milk quality
       Bulk milk SCCMarch 2013YN
       Coliform countMarch 2013YN
       Bacterial countMarch 2013YN
      Antimicrobial consumption
       Δ total ATI
      Difference in total antimicrobial treatment incidence.
      January 2012 to February 2013 (study period 1)YY
       Δ critically important ATI
      Difference in antimicrobial treatment incidence of critically important antimicrobials.
      January 2012 to February 2013 (study period 1)YY
      1 Y = the association with the outcome variable was studied; N = the association with the outcome variable was not studied.
      2 Dairy Herd Improvement data for March 2013 were not available; therefore, the baseline was April 2013.
      3 Difference in herd-level incidence rate of treated mastitis cases.
      4 Difference in total antimicrobial treatment incidence.
      5 Difference in antimicrobial treatment incidence of critically important antimicrobials.
      The incidence rate of treated mastitis (IRTM) cases was based on treatment records for all mastitis cases, both clinical and subclinical, that were recorded by the participating farmers, either on paper or using an online software program. The herd-level IRTM was calculated by dividing the number of treated cows by the cow-days at risk and expressed as treated cases per 10,000 cow-days at risk as before (
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ). Treatments of the same cow within 2 wk from a previous case were not considered new cases and therefore were excluded from the analysis (
      • Verbeke J.
      • Piepers S.
      • Supre K.
      • De Vliegher S.
      Pathogen-specific incidence rate of clinical mastitis in Flemish dairy herds, severity, and association with herd hygiene.
      ). For every herd, the difference in IRTM was calculated as the difference between the herd-level IRTM obtained in study period 2 and the herd-level IRTM obtained in the period January 2012 to February 2013, further referred as study period 1 (Table 1; Figure 1).

      Milk Quality

      Bulk milk SCC records measured at approximately 1-wk intervals from study period 2 were available through the Milk Control Center Flanders (Lier, Belgium), which executes the regulatory milk quality screening program in Flanders. For each herd, the geometric mean BMSCC per month of study period 2 was calculated based on the weekly measurements. Both the bulk milk bacterial count (BC) and coliform count (CC) records per herd measured at 2-wk intervals during study period 2 were retrieved from the Milk Control Center Flanders as well. Bacterial counts were expressed as the number of individual BC per milliliter of milk. Coliform counts were expressed as colony-forming units per milliliter of milk (cfu/mL). Per herd, the monthly geometric mean BC and CC were calculated based on the measurements taken every 2 wk (Table 1). In 1 intervention herd, the CC was not available during study period 2. An ln transformation of the geometric mean BMSCC was performed to normalize the data before statistical analysis. The geometric mean BC and CC were normalized by an inverse log10 transformation and a log10 transformation, respectively. Additionally, milk quality parameters (BMSCC, BC, and CC) were obtained from 275 randomly selected, nonparticipating Flemish dairy herds that were not aware of the project, referred to as the external group.

      AMC

      Antimicrobial consumption data were retrieved by “garbage can audits” (
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ). The AMC at the herd level was quantified by the antimicrobial treatment incidence (ATI) for both study period 1 and study period 2 (Figure 1). Therefore, the total amount of active substances used was divided by the defined daily dose animal multiplied by the herd-level cow-days. The total amount of active substances was calculated per study period (i.e., 1 and 2) by multiplying the volume per receptacle by the total number of used receptacles and by the concentration of the drug. The ATI was defined as the number of defined daily dose animal used per 1,000 cows per day. Fluoroquinolones and third- and fourth-generation cephalosporins were classified as critically important for human health based on the World Health Organization classification. The difference in overall ATI and the difference in ATI of critically important antimicrobials was calculated per herd by the difference in herd-level overall ATI and ATI of the critically important antimicrobials, respectively, obtained in study period 2 and the herd-level overall ATI and ATI of the critically important antimicrobials obtained in study period 1 (Table 1; Figure 1).

      MMI

      The 56 participating dairy herds were randomly divided into an intervention group (n = 28) that received MMI and a control group (n = 28) that did not receive MMI. All herds in the intervention group were intensively followed and advised by the first author according to the strategy and principles described by
      • Barkema H.W.
      • De Vliegher S.
      • Piepers S.
      • Zadoks R.N.
      Herd level approach to high bulk milk somatic cell count problems in dairy cattle.
      . Briefly, each 6 wk from January 2012 until March 2014 the intervention herds received advice concerning prevention and control of udder health problems at both the herd and animal levels. The first herd visit included a general herd tour to gain insight into the general dairy management, with a focus on mastitis management. Milking routine procedures were evaluated by the first author by scoring the teat cleanliness before cluster attachment, measuring the interval between cleaning teats and attachment of the milking cluster, counting the times a milking cluster was kicked off, and so on. Also, the milking machine was evaluated based on the teat end score, congestion of the teats, automatic cluster removal, and so on. Based on an SCC and IRTM data analysis and results of bacteriological culturing of cows with clinical or subclinical mastitis, a farm-specific prevention and control mastitis plan was formulated together with the farmer. During each 6-wk herd visit, housing was inspected and critical points in the management were discussed. Additionally, an evaluation was made of the incidence and cure rate of clinical mastitis cases, the cure rate and rate of IMI during the dry period, and the average herd milk SCC and the average SCC of first-lactation cows. Also, all lactating animals with an elevated SCC (i.e., first-lactation cows ≥150,000 cells/mL; multiparous cows ≥250,000 cells/mL) were listed and for each animal the most appropriate solution [wait until the next DHI; sample for bacteriological culture potentially followed by an antimicrobial treatment based on the results; dry-off (early) or apply a systemic antimicrobial treatment around the onset of the dry period] was suggested and discussed with the farmer. Decisions were based on the literature and driven by factors associated with the cure rate of subclinical mastitis (
      • Sol J.
      • Sampimon O.C.
      • Snoep J.J.
      • Schukken Y.H.
      Factors associated with bacteriological cure during lactation after therapy for subclinical mastitis caused by Staphylococcus aureus.
      ). Herd veterinarians of the herds in the intervention group were motivated to be actively involved in the project and in the herd visits by contacting and inviting them to attend and give their input during the herd visits. Based on their presence during herd visits, intervention herds were stratified into “veterinarian present at each herd visit,” “veterinarian present at some herd visits,” and “veterinarian not present at any herd visit.”
      Dairy farmers of the intervention herds and their herd veterinarians were invited to 3 group meetings organized during the study and shortly thereafter to (1) explain the aims of the study and present the study design, (2) discuss the progress of the study and the first preliminary results, and (3) present and discuss the final results of the study. They then received a summary with herd-specific results, as was promised at the onset of the project. Separate and much more general meetings (n = 3) were organized for the farmers of the control herds and their respective veterinarians to keep them informed about the progress of the study and to keep them motivated because the control herds' data on udder health, milk quality, and AMC were essential for future analyses as part of the study. For both the intervention and control herds, the importance of a reduced and more responsible use of AMC in livestock for human health was discussed in each meeting.

      Mastitis Management Score

      Based on the questionnaire completed in the middle of the project (March 2013; Figure 1), the level of mastitis management was assessed by combining several variables into a new variable, further referred to as the herd-level mastitis management score (MMS). It was the additive of the 26 variables described in Table 2 that were all binary in nature (0 = no; 1 = yes). Some variables were assessed according to a simplified Likert scale including the answers “yes, always,” “mostly,” and “no, never.” Those variables were recoded into a binary variable. If the farmer declared that a certain management practice was performed mostly yet not consistently, we still assigned a score of 1 (yes).
      Table 2Overview of the different questionnaires (January 2012,
      Questionnaire used for calculation of the MMIS.
      March 2013,
      Questionnaire used for calculation of the MMS.
      and April 2014
      Questionnaire used for calculation of the MMIS.
      ) used in this study, the specific recommended management measures used to calculate the mastitis management score (MMS) and mastitis management implementation score (MMIS), and the number of herds in the intervention (n = 28) and control (n = 28) groups implementing each of the individual measures
      ItemInterventionControl
      January 2012March 2013April 2014January 2012March 2013April 2014
      Milking procedures
       One towel per cow for cleaning the teats
      Herds equipped with an automatic milking system received a score of 1.
      162020192019
       Disinfection of teats before cluster attachment by foaming or use of disinfectant towels
      Herds equipped with an automatic milking system received a score of 1.
      81717121514
       Time between preparation and cluster attachment is at least 60 s
      Herds equipped with an automatic milking system received a score of 1.
      182526192727
       Prestripping
      Herds equipped with an automatic milking system received a score of 1.
      162017222222
       Wearing gloves during milking
      Herds equipped with an automatic milking system received a score of 1.
      102012171819
       Disinfecting hands or gloves after milking infected cows
      Herds equipped with an automatic milking system received a score of 1.
      81316578
       Disinfecting teats with dip or spray after milking252425262727
       Disinfecting milking clusters with hot water or steam after milking infected animals51010775
       Keep cows standing after milking
      Herds equipped with an automatic milking system received a score of 0.
      799101213
      Milking machine
      During the study period, in the intervention group as well as in the control group, 1 herd changed from a conventional milking parlor to an automatic milking system. At the end of the study, 5 herds in the intervention group and 6 herds in the control group had an automatic milking system.
       Replacement of teat liners after 2,500 milkings for rubber teat liners or 5,000 milkings for silicon teat liners4211281411
       Automatic cluster removal27
      Dash indicates that parameters were not registered or evaluated in the March 2013 questionnaire.
      272727
      Dry cow management
       Feeding dry cow minerals252422262527
       Having a separate calving pen
      The large difference between the herds having a separate calving pen might be partially due to a nuance in the question in January 2012 and April 2014 questionnaires versus the question in the March 2013 questionnaire (“having a separate calving pen that is different from the nursery” vs. “having a separate calving pen”).
      121974279
       Internal teat sealer for each cow82422152018
      Hygiene
       Clean udders
      Udder hygiene was scored for at least 20 randomly selected lactating cows during herd visits in January 2012 and April 2014 as described by Schreiner and Ruegg (2003). The proportion of cows with an udder hygiene score of 3 or 4 was calculated for each visit. Herds with an average proportion of >50% over the 2 visits were categorized as dirty; other herds were categorized as clean. Score 1 = clean; score 0 = dirty.
      12202218
       Clipping udders at least twice a year142019141813
       Clipping tails at least twice a year132521152319
       Cleaning cubicles at least twice a day252827262827
       Using disinfectant in the cubicles182625191822
       Cleaning alleyways at least twice a day202424232625
      Other
       Veterinary herd health management in place other than by first author71015151516
       Bacteriological culturing of each clinical mastitis case4228183
       Bacteriological culturing for subclinical mastitis at least once a year122623858
       Correct fly control in heifers on pasture151616151611
       Not feeding waste milk to heifer calves101822101618
       Monitoring and eradicating bovine viral diarrhea virus132526121923
      1 Questionnaire used for calculation of the MMIS.
      2 Questionnaire used for calculation of the MMS.
      3 Herds equipped with an automatic milking system received a score of 1.
      4 Herds equipped with an automatic milking system received a score of 0.
      5 During the study period, in the intervention group as well as in the control group, 1 herd changed from a conventional milking parlor to an automatic milking system. At the end of the study, 5 herds in the intervention group and 6 herds in the control group had an automatic milking system.
      6 Dash indicates that parameters were not registered or evaluated in the March 2013 questionnaire.
      7 The large difference between the herds having a separate calving pen might be partially due to a nuance in the question in January 2012 and April 2014 questionnaires versus the question in the March 2013 questionnaire (“having a separate calving pen that is different from the nursery” vs. “having a separate calving pen”).
      8 Udder hygiene was scored for at least 20 randomly selected lactating cows during herd visits in January 2012 and April 2014 as described by
      • Schreiner D.A.
      • Ruegg P.L.
      Relationship between udder and leg hygiene scores and subclinical mastitis.
      . The proportion of cows with an udder hygiene score of 3 or 4 was calculated for each visit. Herds with an average proportion of >50% over the 2 visits were categorized as dirty; other herds were categorized as clean. Score 1 = clean; score 0 = dirty.
      Table 3Overview of the questionnaire scoring system used to calculate the mastitis management score (MMS) and mastitis management implementation score (MMIS)
      VariableQuestionnaireScore
      January 2012March 2013April 2014
      MMSYes1
      No0
      MMISYesNo−2
      NoNo−1
      YesYes+1
      NoYes+2

      Mastitis Management Implementation Score

      To assess the level of implementation of the different recommended mastitis management practices over time, a herd-level mastitis management implementation score (MMIS) was calculated based on the difference between the score for each individual variable obtained using the data from the final questionnaire (April 2014; i.e., the end of study period 2) and the score obtained using the first questionnaire (January 2012; i.e., the start of study period 1). If on a specific farm a measure was implemented in January 2012 but not anymore in April 2014, a score of −2 was assigned to that specific measure. If in a specific herd a measure was implemented in January 2012 and was still implemented in April 2014, a score of +1 was assigned. If in a specific herd a measure was not applied in January 2012 but was applied in April 2014, a score of +2 was assigned to that specific measure. If in a specific herd a measure was not applied in January 2012 and was still not implemented in April 2014, a score of −1 was assigned. The MMIS eventually is the additive of the scores of all 26 measures (Table 2). The higher the absolute MMIS, the more (recommended) mastitis management practices were implemented during the study period.

      Statistical Analyses

      Udder Health

      To determine the association between the MMI and the MMS on one hand and herd-level udder health on the other, several separate linear (mixed) regression models were fit. First, 2 separate linear mixed regression models were built with the prevalence of chronic IMI and the prevalence of new IMI per month during study period 2 (Figure 1) as continuous outcome variables and MMI (0 = control group; 1 = intervention group) and measurements (maximum of 11 measurements per herd in study period 2) as categorical independent variables. In those models, herd was included as a random effect to correct for multiple measurements within herds using an autoregressive correlation structure (PROC MIXED; SAS version 9.4; SAS Institute Inc., Cary, NC). The interaction term between the variables herd-level MMI and measurement was included in the model and withheld when significant (PROC MIXED; SAS version 9.4; SAS Institute Inc.). Similar models with the prevalence of chronic IMI and the prevalence of new IMI per month during study period 2 (Figure 1) as continuous outcome variables and the MMS as continuous independent variable were built. Statistical significance was assessed at P < 0.05.
      Second, 3 separate univariable linear regression models were built with difference in IRTM between study periods 2 and 1 as the continuous outcome variable and the variable MMI as the categorical independent variable (0 = control group; 1 = intervention group) and the variables MMS and MMIS as continuous independent variables (SPSS Statistics 22.0; IBM, New York, NY). Statistical significance was assessed at P < 0.05.

      Milk Quality

      To assess the association between MMI and MMS on one hand and herd-level milk quality during study period 2 on the other, 3 separate linear mixed regression models were fit with BMSCC, BC, and CC as continuous outcome variables, respectively, and MMI (0 = control group;1 = intervention group) and measurement (maximum of 12 measurements per herd, collected in study period 2) as categorical independent variables. Herd was included as a random effect as well, and an autoregressive correlation structure was included. The interaction term between the variables herd-level MMI and measurement was included in the model and withheld when significant (PROC MIXED; SAS version 9.4; SAS Institute Inc.). A similar model was built with MMS as a continuous independent variable. Statistical significance was assessed at P < 0.05.

      AMC

      Three separate univariable linear regression models were fit with difference in total ATI and difference in ATI of critically important antimicrobials (i.e., difference in ATI between study periods 2 and 1) per herd as continuous variables, respectively; MMI (0 = control group; 1 = intervention group) as the categorical independent variable; and MMS and MMIS as continuous independent variables (SPSS Statistics 22.0; IBM). To visualize the association between the difference in total ATI and difference in ATI of critically important antimicrobials, and the MMS and MMIS, a simulation was performed by adding different MMS and MMIS values varying between the minimum and maximum value in the data set to the obtained regression model. The data were graphically represented (Figure 2).
      Figure thumbnail gr2
      Figure 2Simulation based on data obtained from expressing the difference in antimicrobial treatment incidence (ATI) between study period 2 and study period 1 (total ATI and ATI that are critically important for human health; defined daily doses animal/1,000 cow-days; see ) for (a) different mastitis management scores (MMS; varying between 9 and 22) and (b) different mastitis management implementation scores (MMIS; varying between −13 and 31).

      MMI

      Several separate univariable linear regression models were fit with MMS and MMIS as continuous outcome variables, respectively, and MMI (0 = control group; 1 = intervention group) as the categorical independent variable (SPSS Statistics 22.0; IBM). Also, the associations between the degree of presence of the veterinarian at herd visits (0 = present at each herd visit, n = 5; 1 = present at some herd visits, n = 9; 2 = not present at any of the herd visits, n = 14) in the herds assigned to the intervention group on one hand and the MMS and MMIS on the other were investigated by fitting univariable linear regression models (SPSS Statistics 22.0; IBM). Normal probability plots of standardized residuals, plots of standardized residuals versus the dependent variables, and plots of standardized residuals versus predicted values were generated to check whether the assumptions of normality, linearity, and homogeneity of variance had been fulfilled. No problems were detected.

      RESULTS

      Udder Health

      The median prevalence of chronic infections per month during study period 2 was 8.6 and 8.4% in the intervention and control groups, respectively (Table 4; P = 0.435). The prevalence of new IMI was significantly higher in the herds in the intervention group compared with the herds in the control group (7.1 and 6.8%, respectively; P = 0.012). The difference in IRTM between study periods 2 and 1 did not differ between both groups (P = 0.36).
      Table 4Descriptive statistics of udder health and milk quality parameters and antimicrobial consumption of the herds in the intervention (n = 28) and control (n = 28) groups
      ItemMinimumPercentileMaximum
      25th50th75th
      Udder health
       Prevalence of new IMI
      Average monthly prevalence of new IMI during study period 2 (see Figure 1).
      (%)
        All herds (n = 56)2.35.77.08.217.1
        Intervention3.85.87.18.517.1
        Control2.35.66.88.010.5
       Prevalence of chronic IMI
      Average monthly prevalence of chronic IMI during study period 2 (see Figure 1).
      (%)
        All herds2.96.68.610.830.1
        Intervention3.67.38.611.818.1
        Control2.95.98.410.330.1
       Δ IRTM
      Δ IRTM (treated cases of clinical or subclinical mastitis per 10,000 cow-days at risk) = difference in incidence rate of treated mastitis (IRTM) cases between study period 2 and study period 1 (see Figure 1). IRTM data missing on 1 herd in the intervention group and 3 herds in the control group.
        All herds−9.3−3.5−2.20.613.1
        Intervention−9.3−3.2−1.90.16.0
        Control−6.7−3.7−2.21.213.1
      Milk quality
       Bulk milk SCC
      Geometric mean bulk milk SCC during study period 2 (see Figure 1).
      (cells/mL)
        All herds46,281141,133176,796231,192532,000
        Intervention65,011143,141177,092231,411511,842
        Control46,281139,872176,644229,749532,000
       Coliform count
      Geometric mean (coli) during study period 2 (see Figure 1). Coliform count was not registered in 1 intervention herd.
      (cfu/mL)
        All herds1.01.05.331.0912.5
        Intervention1.01.05.628.0600.9
        Control1.01.04.632.7912.5
       Bacterial count
      Geometric mean bacterial count during study period 2 (see Figure 1).
      (individual count/mL)
        All herds0.004.005,4777,937226,575
        Intervention0.004.005,4778,214226,575
        Control0.004.005,5677,75065,422
      Antimicrobial consumption
       Δ Total ATI
      ΔTotal ATI (defined daily dose animal/1,000 cow-days) = difference in total antimicrobial treatment incidence (ATI) between study period 2 and study period 1 (see Figure 1).
        All herds−26.2−6.0−2.71.115.9
        Intervention−26.2−6.2−2.61.315.9
        Control−10.4−5.5−2.70.511.6
       Δ ATI critically important
      ΔATI critically important (defined daily dose animal/1,000 cow-days) = difference in ATI of antimicrobials that are critically important for human health between study period 2 and study period 1 (see Figure 1).
        All herds−15.1−4.7−0.80.516.4
        Intervention−12.2−5.4−2.30.516.4
        Control−15.1−2.3−0.80.57.0
      1 Average monthly prevalence of new IMI during study period 2 (see Figure 1).
      2 Average monthly prevalence of chronic IMI during study period 2 (see Figure 1).
      3 Δ IRTM (treated cases of clinical or subclinical mastitis per 10,000 cow-days at risk) = difference in incidence rate of treated mastitis (IRTM) cases between study period 2 and study period 1 (see Figure 1). IRTM data missing on 1 herd in the intervention group and 3 herds in the control group.
      4 Geometric mean bulk milk SCC during study period 2 (see Figure 1).
      5 Geometric mean (coli) during study period 2 (see Figure 1). Coliform count was not registered in 1 intervention herd.
      6 Geometric mean bacterial count during study period 2 (see Figure 1).
      7 ΔTotal ATI (defined daily dose animal/1,000 cow-days) = difference in total antimicrobial treatment incidence (ATI) between study period 2 and study period 1 (see Figure 1).
      8 ΔATI critically important (defined daily dose animal/1,000 cow-days) = difference in ATI of antimicrobials that are critically important for human health between study period 2 and study period 1 (see Figure 1).
      Further, the MMS was not associated with the prevalence of new or chronic IMI or with difference in IRTM between study periods 2 and 1 (Table 5). Still, the level of association between MMS on the prevalence of chronic IMI depended on the time of measurement (Table 5; P = 0.009). Also, no association was found between difference in IRTM between study periods 2 and 1 and MMIS.
      Table 5Statistical associations between mastitis management score (MMS) or mastitis management implementation score (MMIS) on one hand and repeated measurements of udder health parameters, milk quality parameters, and antimicrobial consumption on the other hand
      ItemMMSMMIS
      β
      Estimate.
      SE
      Standard error of the variance estimate of the parameter.
      P-valueβSEP-value
      Udder health
       Prevalence of new IMI
      Ln-transformed average monthly prevalence of new IMI as measured in study period 2 (see Figure 1).
      (%)
        Intercept1.80.2<0.001
        Score0.010.010.31
        Measurement
      Categorical variables of 11 measurements from April 2013 until February 2014 (see Figure 1).
      0.49
       Prevalence of chronic IMI
      Ln-transformed average monthly prevalence of chronic IMI as measured in study period 2 (see Figure 1).
      (%)
        Intercept1.90.5<0.001
        Score0.010.030.12
        Measurement
      Categorical variables of 11 measurements from April 2013 until February 2014 (see Figure 1).
      0.01
        Measurement × score0.009
       Δ IRTM
      Δ IRTM = difference in incidence rate of treated mastitis cases between study period 2 and study period 1 (see Figure 1; no repeated measurements).
        Intercept−0.43.00.91−1.20.80.15
        Score−0.10.20.72−0.020.10.77
      Milk quality
       Bulk milk SCC
      Ln-transformed geometric mean bulk milk SCC as measured in study period 2 (see Figure 1).
        Intercept5.70.2<0.001
        Score−0.030.010.003
        Measurement
      Categorical variables of 12 measurements from March 2013 until February 2014 (see Figure 1).
      <0.001
       Coliform count
      Log10-transformed geometric mean coliform count as measured in study period 2 (see Figure 1).
        Intercept0.50.30.06
        Score0.000.010.85
        Measurement
      Categorical variables of 12 measurements from March 2013 until February 2014 (see Figure 1).
      <0.001
       Bacterial count
      Inverse log10 geometric mean bacterial count as measured in study period 2 (see Figure 1).
        Intercept0.90.1<0.001
        Score0.020.010.008
        Measurement
      Categorical variables of 12 measurements from March 2013 until February 2014 (see Figure 1).
      0.08
      Antimicrobial consumption
       Δ Total ATI
      ΔTotal ATI (defined daily dose animal/1,000 cow-days) = difference in total antimicrobial treatment incidence (ATI) between study periods 2 and 1 (see Figure 1; no repeated measurements).
        Intercept1.115.250.83−1.051.440.47
        Score−0.180.310.57−0.080.100.46
       Δ ATI critically important
      ΔATI critically important (defined daily dose animal/1,000 cow-days) = difference in ATI of antimicrobials that are critically important for human health between study period 2 and study period 1 (see Figure 1; no repeated measurements).
        Intercept5.853.670.122.161.000.83
        Score−0.450.220.047−0.160.070.029
      1 Estimate.
      2 Standard error of the variance estimate of the parameter.
      3 Ln-transformed average monthly prevalence of new IMI as measured in study period 2 (see Figure 1).
      4 Categorical variables of 11 measurements from April 2013 until February 2014 (see Figure 1).
      5 Ln-transformed average monthly prevalence of chronic IMI as measured in study period 2 (see Figure 1).
      6 Δ IRTM = difference in incidence rate of treated mastitis cases between study period 2 and study period 1 (see Figure 1; no repeated measurements).
      7 Ln-transformed geometric mean bulk milk SCC as measured in study period 2 (see Figure 1).
      8 Categorical variables of 12 measurements from March 2013 until February 2014 (see Figure 1).
      9 Log10-transformed geometric mean coliform count as measured in study period 2 (see Figure 1).
      10 Inverse log10 geometric mean bacterial count as measured in study period 2 (see Figure 1).
      11 ΔTotal ATI (defined daily dose animal/1,000 cow-days) = difference in total antimicrobial treatment incidence (ATI) between study periods 2 and 1 (see Figure 1; no repeated measurements).
      12 ΔATI critically important (defined daily dose animal/1,000 cow-days) = difference in ATI of antimicrobials that are critically important for human health between study period 2 and study period 1 (see Figure 1; no repeated measurements).

      Milk Quality

      The geometric mean of the BMSCC varied throughout the year (Figure 3a), as did the CC (Figure 3b) and BC (Figure 3c). Further, the BMSCC (P = 0.73), BC (P = 0.52), and CC (P = 0.97) did not differ between intervention and control herds. The better the mastitis management (higher MMS), the lower the BMSCC and the higher the BC (Table 5).
      Figure thumbnail gr3
      Figure 3Change over time of the geometric mean of the (a) bulk milk SCC (BMSCC), (b) coliform count, and (c) bacterial count in the dairy herds in the intervention, control, and external groups.

      AMC

      The herds belonging to the intervention group had a numerically larger absolute difference in ATI of critically important antimicrobials compared with the 28 herds in the control group (Table 4), although the difference was not statistically significant (P = 0.31). The MMS and MMIS, however, were both positively associated with a decrease in AMC of antimicrobials that are critically important for human health (P ≤ 0.05; Figure 2; Table 5), indicating that better mastitis management was associated with reduced usage of critically important antimicrobials.

      Mastitis Management

      The MMS and MMIS strongly varied in both the herds in the intervention group (median = 16 and range = 12 to 22; median = 13 and range = −5 to 31, respectively) and the herds in the control group (median = 16 and range = 9 to 22; median = 9 and range = −13 to 22, respectively; Table 6; Figure 4). Also, the level of implementation of the different recommended mastitis management practices strongly varied between study period 1 and study period 2 (Table 2). The MMS and MMIS of herds in the intervention group were 1.89 and 5.29 units higher, respectively, compared with herds in the control group (P = 0.018 and 0.034, respectively). Of the 10 herds with the lowest MMS, 7 herds belonged to the control group. Considering the 10 herds with the highest MMS, 8 herds belonged to the intervention group. Only 3 of the 10 herds with the lowest MMIS belonged to the intervention group. Considering the 10 herds with the highest MMIS, only 1 herd belonged to the control group. Among intervention herds, herds in which the vet attended each herd visit performed by the first author had a higher MMS and MMIS (median = 20 and 24, respectively) compared with herds in which the veterinarian only sometimes attended the herd visits (median = 16 and 17, respectively) or never attended of the herd visits (median = 16.5 and 7.5, respectively; P = 0.047 and 0.002, respectively).
      Table 6Descriptive statistics of the mastitis management score (MMS) and mastitis management implementation score (MMIS) of the herds in the intervention and control groups
      ItemNo. of herdsMinimumPercentileMaximum
      25th50th75th
      MMS
       All herds56915161922
       Intervention281216162022
       Control28914161722
      MMIS
       All herds56−135101731
       Intervention28−55132131
       Control28−13291322
      Figure thumbnail gr4
      Figure 4Box plot of the mastitis management score (MMS) based on the reference questionnaire (January 2012), the second questionnaire (March 2013), and the mastitis management implementation score (MMIS) of the intervention group and the control group [difference between the third questionnaire (April 2014) and the first questionnaire (January 2012); see ]. The lower whisker represents the minimum, the upper whisker represents the maximum; the box shows the lower limit (25th percentile) and upper limit (75th percentile) and the midline is the median. Dots represent outliers.

      DISCUSSION

      To our knowledge, this is the first observational study investigating the effect of MMI, level of mastitis management, and level of implementation of mastitis management over time on udder health, milk quality, and AMC in commercial dairy herds. Although it is far from easy to motivate farmers from commercial dairy herds to participate as control herds because they are not receiving any MMI while being obligated to collect data on udder health, milk quality, and AMC, we successfully convinced 28 farmers to do so. The latter approach allowed us to straightforwardly unravel the effect of MMI on the udder health, milk quality, and AMC in dairy herds in a clinical trial-like approach.
      The mastitis management of the different herds was evaluated via the MMS and MMIS. Both scores were calculated based on a simple and arbitrary system, and the same weight was assigned to each measure. Although it is well known from the literature (
      • Dufour S.
      • Frechette A.
      • Barkema H.W.
      • Mussell A.
      • Soholl D.T.
      Invited review: Effect of udder health management practices on herd somatic cell count.
      ) that not all measures carry the same risk, assigning a different score to the different measures would have been arbitrary as well and maybe even less straightforward—even more so because the importance of a certain mastitis management measure even depends on a herd's mastitis (main) etiology, as some measures are more efficacious in the control of environmental mastitis, whereas others are more efficacious in the control of contagious mastitis (
      • Hogeveen H.
      • Huijps K.
      • Lam T.
      Economic aspects of mastitis: New developments.
      ).
      Despite the systematic MMI for the herds in the intervention group, no differences in udder health, milk quality, and AMC were identified compared with control herds, except for the somewhat surprising higher prevalence of new IMI in intervention herds (discussed later). Multiple explanations for the lack of differences can be given. First, dairy farmers of the intervention and control groups were not selected randomly and rather were recruited by volunteer selection, although certain selection criteria had to be met (
      • Stevens M.
      • Piepers S.
      • Supre K.
      • Dewulf J.
      • De Vliegher S.
      Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
      ). The farmers of these 56 herds were likely more motivated to improve udder health and reduce AMC than the average dairy farmer in Flanders at that time. This approach also resulted in a selection of better managed dairy herds, as can be derived from the lower BMSCC, BC, and CC in herds of the intervention and control groups compared with the external group. A second explanation for the unexpected low effect of the MMI might be that herds in the control group could not be prevented from improving their mastitis management and treatment behavior during the study, which might have affected the estimations of the true effect of MMI on the herds' udder health status and AMC. On average but to a lesser extent than the intervention herds, control herds did improve their mastitis management during the study. Clearly, this narrowed the gap between the control and intervention herds and likely resulted in an underestimation of the true effect of MMI on udder health. Third, to keep the farmers of the control herds motivated, several more general meetings were organized for them, and this might eventually have had an influence on their behavior toward mastitis management. Fourth, the MMS varied strongly between intervention herds, and not all herds of the intervention group actually improved their mastitis management during the study. In fact, some intervention herds even decreased their mastitis management between the 2 study periods, as evidenced by the negative MMIS. Of course, one should also take into account that the intervention itself and the proposed adaptations in the management might not have been equally adequate in all herds. Fifth, herds belonging to the intervention group received MMI starting with the first herd visit in January 2012. Some herds in the intervention group had already started improving their mastitis management during the first study year, which is reflected by the substantial increase in MMS by the end of study period 1. Because the herd-level IRTM, total AMC, and consumption of the critically important antimicrobials were calculated on an annual basis [difference between study period 2 (i.e., March 2013–February 2014) and study period 1 (i.e., January 2012–February 2013)], the differences between the intervention and control groups might have been underestimated. Still, we should acknowledge that this hypothesis, if true, would be in contrast with the findings of
      • Kuipers A.
      • Koops W.J.
      • Wemmenhove H.
      Antibiotic use in dairy herds in the Netherlands from 2005 to 2012.
      , who concluded that the effect of changes in the level of implementation on the herd's performances and AMC should always be considered on a longer term (
      • Kuipers A.
      • Koops W.J.
      • Wemmenhove H.
      Antibiotic use in dairy herds in the Netherlands from 2005 to 2012.
      ). Sixth, the establishment of a national center of expertise on AMC and resistance in animals in Belgium during the project (
      • AMCRA (Antimicrobial Consumption and Resistance in Animals)
      Visie 2020.
      ) may have influenced to some extent the management and AMC of all herds because it evolved into several nationwide awareness campaigns targeting farmers and veterinarians along with other projects that were initiated during the time our study was executed. The nationwide awareness campaigns in combination with the selection of motivated farmers for both intervention and control groups might also have contributed to the lack of difference between the intervention and control groups for AMC more specifically.
      It is also worth mentioning that the IRTM comprised all treated mastitis cases, including both clinical and subclinical mastitis as well as new and repeated clinical or subclinical mastitis cases. Still, the MMI and MMIS did not include management practices related to treatment protocols of clinical or subclinical mastitis, which could hypothetically be a reason for the lack of association between MMI and MMIS and the difference in IRTM. Differences in treatment strategy clearly may affect the cure rates and thus the incidence of repeat cases.
      Although the reduction in critically important antimicrobials was somewhat larger in the herds of the intervention group than in the herds in the control group, the difference was not statistically different (P = 0.245). Further scrutinizing of the data revealed a large influence of 1 herd belonging the intervention group that encountered a substantial increase in the use of critically important intramammary tubes because of a Klebsiella spp. outbreak of mastitis in the second year of the study. Rerunning this analysis without the data of this herd at least partly confirmed the latter hypothesis (P = 0.085; data not shown).
      Surprisingly, the average prevalence of new IMI per month was significantly higher in herds of the intervention group compared with herds of the control group. A more detailed analysis of the data revealed that part of the high percentage of new IMI in the intervention group could be attributed to 1 herd in particular. It is worth mentioning that the increase in percentage of new IMI in that particular herd coincided with a shift from conventional milking to automatic milking during the study period, although causality cannot be proven (
      • Hovinen M.
      • Rasmussen M.D.
      • Pyorala S.
      Udder health of cows changing from tie stalls or free stalls with conventional milking to free stalls with either conventional or automatic milking.
      ). The prevalence of new IMI in this herd (17.1%) was far higher than the prevalence of new IMI of all other herds in the intervention group (maximum = 11.0%). An analysis without the data of the particular herd resulted in an increase in P-value from 0.012 to 0.049.
      As expected based on the findings of other studies, the BMSCC and CC varied substantially over time (
      • Piepers S.
      • Zrimšek P.
      • Passchyn P.
      • De Vliegher S.
      Manageable risk factors associated with bacterial and coliform counts in unpasteurized bulk milk in Flemish dairy herds.
      ;
      • Testa F.
      • Marano G.
      • Ambrogi F.
      • Boracchi P.
      • Casula A.
      • Biganzoli E.
      • Moroni P.
      Study of the association of atmospheric temperature and relative humidity with bulk tank milk somatic cell count in dairy herds using generalized additive mixed models.
      ). The MMS was negatively associated with the total BC. Hence, better mastitis management was associated with a lower BC, which is comprehensible because the BC acts as an indicator of overall hygiene in herds (
      • Piepers S.
      • Zrimšek P.
      • Passchyn P.
      • De Vliegher S.
      Manageable risk factors associated with bacterial and coliform counts in unpasteurized bulk milk in Flemish dairy herds.
      ). As the MMS contained 6 management practices related to hygiene, this association was expected. Also, the inverse association between the MMS and the BMSCC in study period 2 suggests that a reduction in BMSCC can be achieved by implementation of the well-known standard mastitis management practices based on the National Mastitis Council 10-point prevention and control plan.
      Better management went along with a decrease in consumption of antimicrobials that are critically important for human health. This is in accordance with the findings of
      • Speksnijder D.C.
      • Graveland H.
      • Eijck I.
      • Schepers R.W.M.
      • Heederik D.J.J.
      • Verheij T.J.M.
      • Wagenaar J.A.
      Effect of structural animal health planning on antimicrobial use and animal health variables in conventional dairy farming in the Netherlands.
      and was visualized in Figure 2 by simulation. The fact that the association between MMIS and difference in ATI of critically important antimicrobials is even more pronounced (i.e., lower P-value) than the association between MMS and difference in ATI of critically important antimicrobials suggests that the latter association was probably also driven by the change in MMS between the start of study period 1 and the start of study period 2. Remarkably, with an MMS or MMIS below 13, the total AMC decreased further but the consumption of critically important antimicrobials for human health increased. The latter finding suggests that poor management is probably countered in some herds by the increased use of critically important antimicrobials. As the goal of appropriate use is not only to reduce total AMC but also to have more responsible use of antimicrobials in general, the increase in antimicrobials that are critically important for human health in those herds with inferior mastitis management is an important finding.
      As MMS was associated with the presence of the herd veterinarian during the herd visits and as the MMS was associated with the BMSCC, BC, and the use of critically important antimicrobials, we hypothesize that the participation of herd veterinarians in udder health management can contribute to better milk quality and lower consumption of critically important antimicrobials. Still, it could also be that the herd managers involving their herd veterinarian more often are better managers and see the added value of cooperating with advisors, such as the veterinarian.

      CONCLUSIONS

      Implementation of (recommended) management measures concerning udder health is higher in herds that receive intensive MMI compared with herds without this input. Moreover, within the herds that receive such MMI, compliance is better if the herd veterinarian is attending at least some of the herd health visits. Further, good or improved mastitis management is associated with decreased use of antimicrobials that are critically important for human health and with better milk quality. In conclusion, good mastitis management can help lower the consumption of antimicrobials that are critically important for human health without negative implications for milk quality and udder health.

      ACKNOWLEDGMENTS

      All farmers who cooperated are gratefully acknowledged. We extend special thanks to all M-team UGent colleagues for helping with the herd visits. Thanks also go to the Milk Control Centre Flanders (Lier, Belgium), the Farmers Union (Boerenbond, Leuven, Belgium), the Milk Quality Label (IKM-QFL, Brussels, Belgium), and the Belgian Confederation of the Dairy Industry (BCZ, Leuven, Belgium) for partially financing this study.

      REFERENCES

        • Aarestrup F.M.
        The livestock reservoir for antimicrobial resistance: A personal view on changing patterns of risks, effects of interventions and the way forward.
        Philos. Trans. R. Soc. Lond. B Biol. Sci. 2015; 370 (25918442): 20140085
        • Agersø Y.
        • Aarestrup F.M.
        Voluntary ban on cephalosporin use in Danish pig production has effectively reduced extended-spectrum cephalosporinase-producing Escherichia coli in slaughter pigs.
        J. Antimicrob. Chemother. 2013; 68 (23129728): 569-572
        • AMCRA (Antimicrobial Consumption and Resistance in Animals)
        Visie 2020.
        https://www.amcra.be/en/amcra-vision-2020/
        Date: 2014
        Date accessed: December 23, 2018
        • Barkema H.W.
        • De Vliegher S.
        • Piepers S.
        • Zadoks R.N.
        Herd level approach to high bulk milk somatic cell count problems in dairy cattle.
        Vet. Q. 2013; 33 (23706026): 82-93
        • Bennedsgaard T.W.
        • Klaas I.C.
        • Vaarst M.
        Reducing use of antimicrobials—Experiences from an intervention study in organic dairy herds in Denmark.
        Livest. Sci. 2010; 131: 183-192
        • Brinkmann J.
        • March S.
        Animal Health in Organic Dairy Farming—Health State as Well as Development, Application and Evaluation of a Preventive Herd Health Planning Concept.
        Georg-August-Universität Göttingen, Göttingen, Germany2010
        • Callens B.
        • Faes C.
        • Maes D.
        • Catry B.
        • Boyen F.
        • Francoys D.
        • de Jong E.
        • Haesebrouck F.
        • Dewulf J.
        Presence of antimicrobial resistance and antimicrobial use in sows are risk factors for antimicrobial resistance in their offspring.
        Microb. Drug Resist. 2015; 21 (25098762): 50-58
        • Chang Q.
        • Wang W.
        • Regev-Yochay G.
        • Lipsitch M.
        • Hanage W.P.
        Antibiotics in agriculture and the risk to human health: How worried should we be?.
        Evol. Appl. 2015; 8 (25861382): 240-247
        • Chantziaras I.
        • Boyen F.
        • Callens B.
        • Dewulf J.
        Correlation between veterinary antimicrobial use and antimicrobial resistance in food-producing animals: A report on seven countries.
        J. Antimicrob. Chemother. 2014; 69 (24216767): 827-834
        • Cicconi-Hogan K.M.
        • Gamroth M.
        • Richert R.
        • Ruegg P.L.
        • Stiglbauer K.E.
        • Schukken Y.H.
        Associations of risk factors with somatic cell count in bulk tank milk on organic and conventional dairy farms in the United States.
        J. Dairy Sci. 2013; 96 (23548286): 3689-3702
        • Dorado-García A.
        • Mevius D.J.
        • Jacobs J.J.
        • Van Geijlswijk I.M.
        • Mouton J.W.
        • Wagenaar J.A.
        • Heederik D.J.
        Quantitative assessment of antimicrobial resistance in livestock during the course of a nationwide antimicrobial use reduction in the Netherlands.
        J. Antimicrob. Chemother. 2016; 71 (27585970): 3607-3619
        • Dufour S.
        • Frechette A.
        • Barkema H.W.
        • Mussell A.
        • Soholl D.T.
        Invited review: Effect of udder health management practices on herd somatic cell count.
        J. Dairy Sci. 2011; 94 (21257026): 563-579
        • Dupont N.
        • Diness L.H.
        • Fertner M.
        • Kristensen C.S.
        • Stege H.
        Antimicrobial reduction measures applied in Danish pig herds following the introduction of the “Yellow Card” antimicrobial scheme.
        Prev. Vet. Med. 2017; 138 (28237240): 9-16
        • González S.
        • Steiner A.
        • Gassner B.
        • Regula G.
        Antimicrobial use in Swiss dairy farms: Quantification and evaluation of data quality.
        Prev. Vet. Med. 2010; 95 (20381180): 50-63
        • Green M.J.
        • Leach K.A.
        • Breen J.E.
        • Green L.E.
        • Bradley A.J.
        National intervention study of mastitis control in dairy herds in England and Wales.
        Vet. Rec. 2007; 160 (17337605): 287-293
        • Hillerton J.E.
        • Bramley A.J.
        • Staker R.T.
        • McKinnon C.H.
        Patterns of intramammary infection and clinical mastitis over a 5-year period in a closely monitored herd applying mastitis control measures.
        J. Dairy Res. 1995; 62 (7738244): 39-50
        • Hogeveen H.
        • Huijps K.
        • Lam T.
        Economic aspects of mastitis: New developments.
        N. Z. Vet. J. 2011; 59 (21328153): 16-23
        • Hovinen M.
        • Rasmussen M.D.
        • Pyorala S.
        Udder health of cows changing from tie stalls or free stalls with conventional milking to free stalls with either conventional or automatic milking.
        J. Dairy Sci. 2009; 92 (19620651): 3696-3703
        • Ivemeyer S.
        • Maeschli A.
        • Walkenhorst M.
        • Klocke P.
        • Heil F.
        • Oser S.
        • Notz C.
        Effects of a two-year dairy herd health management programme on udder health, use of antibiotics and longevity.
        Schweiz. Arch. Tierheilkd. 2008; 150 (18821509): 499-505
        • Ivemeyer S.
        • Smolders G.
        • Brinkmann J.
        • Gratzer E.
        • Hansen B.
        • Henriksen B.I.F.
        • Huber J.
        • Leeb C.
        • March S.
        • Mejdell C.
        • Nicholas P.
        • Roderick S.
        • Stoeger E.
        • Vaarst M.
        • Whistance L.K.
        • Winckler C.
        • Walkenhorst M.
        Impact of animal health and welfare planning on medicine use, herd health and production in European organic dairy farms.
        Livest. Sci. 2012; 145: 63-72
        • Kuipers A.
        • Koops W.J.
        • Wemmenhove H.
        Antibiotic use in dairy herds in the Netherlands from 2005 to 2012.
        J. Dairy Sci. 2016; 99 (26709178): 1632-1648
        • LeBlanc S.J.
        • Lissemore K.D.
        • Kelton D.F.
        • Duffield T.F.
        • Leslie K.E.
        Major advances in disease prevention in dairy cattle.
        J. Dairy Sci. 2006; 89 (16537959): 1267-1279
        • Marshall B.M.
        • Levy S.B.
        Food animals and antimicrobials: Impacts on human health.
        Clin. Microbiol. Rev. 2011; 24 (21976606): 718-733
        • Mitchell J.M.
        • Griffiths M.W.
        • McEwen S.A.
        • McNab W.B.
        • Yee A.J.
        Antimicrobial drug residues in milk and meat: Causes, concerns, prevalence, regulations, tests, and test performance.
        J. Food Prot. 1998; 61 (9709262): 742-756
        • Piepers S.
        • Zrimšek P.
        • Passchyn P.
        • De Vliegher S.
        Manageable risk factors associated with bacterial and coliform counts in unpasteurized bulk milk in Flemish dairy herds.
        J. Dairy Sci. 2014; 97 (24704236): 3409-3419
        • Pol M.
        • Ruegg P.L.
        Treatment practices and quantification of antimicrobial drug usage in conventional and organic dairy farms in Wisconsin.
        J. Dairy Sci. 2007; 90 (17183093): 249-261
        • Postma M.
        • Stark K.D.
        • Sjolund M.
        • Backhans A.
        • Beilage E.G.
        • Losken S.
        • Belloc C.
        • Collineau L.
        • Iten D.
        • Visschers V.
        • Nielsen E.O.
        • Dewulf J.
        Alternatives to the use of antimicrobial agents in pig production: A multi-country expert-ranking of perceived effectiveness, feasibility and return on investment.
        Prev. Vet. Med. 2015; 118 (25650306): 457-466
        • Postma M.
        • Vanderhaeghen W.
        • Sarrazin S.
        • Maes D.
        • Dewulf J.
        Reducing antimicrobial usage in pig production without jeopardizing production parameters.
        Zoonoses Public Health. 2017; 64 (27362766): 63-74
        • Schreiner D.A.
        • Ruegg P.L.
        Relationship between udder and leg hygiene scores and subclinical mastitis.
        J. Dairy Sci. 2003; 86: 3460-3465
        • Sol J.
        • Sampimon O.C.
        • Snoep J.J.
        • Schukken Y.H.
        Factors associated with bacteriological cure during lactation after therapy for subclinical mastitis caused by Staphylococcus aureus.
        J. Dairy Sci. 1997; 80 (9406071): 2803-2808
        • Speksnijder D.C.
        • Graveland H.
        • Eijck I.
        • Schepers R.W.M.
        • Heederik D.J.J.
        • Verheij T.J.M.
        • Wagenaar J.A.
        Effect of structural animal health planning on antimicrobial use and animal health variables in conventional dairy farming in the Netherlands.
        J. Dairy Sci. 2017; 100 (28390724): 4903-4913
        • Speksnijder D.C.
        • Jaarsma A.D.
        • van der Gugten A.C.
        • Verheij T.J.
        • Wagenaar J.A.
        Determinants associated with veterinary antimicrobial prescribing in farm animals in the Netherlands: A qualitative study.
        Zoonoses Public Health. 2015; 62 (25421456): 39-51
        • Steeneveld W.
        • Velthuis A.G.
        • Hogeveen H.
        Short communication: Effectiveness of tools provided by a dairy company on udder health in Dutch dairy farms.
        J. Dairy Sci. 2014; 97 (24418274): 1529-1534
        • Stevens M.
        • Piepers S.
        • Supre K.
        • Dewulf J.
        • De Vliegher S.
        Quantification of antimicrobial consumption in adult cattle on dairy herds in Flanders, Belgium, and associations with udder health, milk quality, and production performance.
        J. Dairy Sci. 2016; 99 (26778315): 2118-2130
        • Testa F.
        • Marano G.
        • Ambrogi F.
        • Boracchi P.
        • Casula A.
        • Biganzoli E.
        • Moroni P.
        Study of the association of atmospheric temperature and relative humidity with bulk tank milk somatic cell count in dairy herds using generalized additive mixed models.
        Res. Vet. Sci. 2017; 114 (28987958): 511-517
        • Tremetsberger L.
        • Leeb C.
        • Winckler C.
        Animal health and welfare planning improves udder health and cleanliness but not leg health in Austrian dairy herds.
        J. Dairy Sci. 2015; 98 (26233459): 6801-6811
        • Tremetsberger L.
        • Winckler C.
        Effectiveness of animal health and welfare planning in dairy herds: A review.
        Anim. Welf. 2015; 24: 55-67
        • Vaarst M.
        • Bennedsgaard T.W.
        • Klaas I.
        • Nissen T.B.
        • Thamsborg S.M.
        • Ostergaard S.
        Development and daily management of an explicit strategy of nonuse of antimicrobial drugs in twelve Danish organic dairy herds.
        J. Dairy Sci. 2006; 89 (16606756): 1842-1853
        • Verbeke J.
        • Piepers S.
        • Supre K.
        • De Vliegher S.
        Pathogen-specific incidence rate of clinical mastitis in Flemish dairy herds, severity, and association with herd hygiene.
        J. Dairy Sci. 2014; 97 (25218745): 6926-6934
        • WHO (World Health Organization)
        Global Action Plan on Antimicrobial Resistance.
        World Health Organization, Geneva, Switzerland2015