Acknowledgements 2. Introduction 3. Speakers at the symposium 4

Transcription

1 Contents Acknowledgements 2 Introduction 3 Speakers at the symposium 4 Automatic mastitis monitoring: will it be the next successful revolution in automation? 5 C. Kamphuis, J. Jago and G. Mein Detecting clinical mastitis with automatic milking system assistance and with other aids 9 John Morton, Jade Hammer and Kendra Kerrisk Molecular tests for milk performance and application in Australia 13 John Penry, John Morton, Jakob Malmo and Graeme Mein A survey of mastitis pathogens in the south-eastern Australian dairy industry 18 Neil Charman, Rodney Dyson, Andrew Hodge, Natalie Robertson and Sarah Chaplin How to R.E.S.E.T. farmer mindset? Experiences from the Netherlands 23 Jolanda Jansen, Roeland Wessels and Theo Lam The Dairy Focus approach Technote 13 and beyond 28 Rod Dyson Experiences with mastitis control over the past 50 years 30 Jakob Malmo CellCheck a new solution to an old challenge 33 F. McCoy, C. Devitt, S. More, K. Heanue and K. McKenzie Timetable 36

2 Acknowledgements The Countdown project team would like to thank the following organisations and individuals for their assistance in the preparation of the Countdown Symposium In particular we wish to express our gratitude to our co-sponsors: Jonathan Leslie and the team at Boehringer Ingelheim; Jane Parker and the team at Pfizer Animal Health. We also acknowledge the sponsorship support provided by Dairy Australia and University of Melbourne. Thank you to: Boehringer Ingelheim Pfizer Animal Health University of Melbourne Faculty of Veterinary Science Dairy Australia Harris Park Group: Pauline Brightling, Anne Hope, Andrea Thompson, John Craven, Helen Pitman, Narelle Savige, Nina Philadelphoff-Puren Rydges on Swanston SUBStitution Pty Ltd (design and layout): Anne Burgi Printworks (proceedings publication) Camperdown Veterinary Centre Primary Logic Dairy & Agricultural Consulting Dairy Australia 2012 For further information, please contact: Harris Park Group, Level 2, Swann House, 22 William Street, Melbourne, Victoria 3000, Australia Ph or Website: The views expressed in this document are those of the authors and do not necessarily represent the official position or policy of Dairy Australia. SUBS/Printworks/12-07

3 Introduction On behalf of the Countdown team and the University of Melbourne, welcome to the Countdown Symposium This is the third Countdown Symposium and we are pleased to welcome back many advisers who participated at our meetings in 2010 and For those of you who are here for the first time, we are certain that you will enjoy a stimulating day with the aim of putting the science of mastitis control into practice. Four dairying countries are represented at this Symposium through our invited speakers. The scientists, researchers and practitioners speaking today come with many years of experience in their respective countries and fields of interest. We would especially like to acknowledge and celebrate the contribution of Jakob Malmo who, in 2012, has marked 50 years in dairy veterinary practice in Victoria. This year we have three interconnected themes running through the day: New frontiers in mastitis detection; Farmers: mastitis, risk, barriers, motivation, rewards Advisers in the frontline. As with the last two meetings, the organising team has attempted to put together topics which will be of use to all dairying advisers regardless of their professional group. Whether you are a vet, milking machine technician or dairy adviser we hope there will be manys new ideas from today that you will be able to take back to your workplace and use with your dairying clients tomorrow. We are particularly enthused about the communication theme which will make up the middle of the day. So much of what constitutes mastitis control best practice has, at its core, people who understand their respective roles. Effective communication between advisers and farmers creates the right environment for all people to perform to their potential in this domain. This also marks the first year that the Symposium has been associated with the new Australian Mastitis and Milk Quality Steering Group (AMMQSG). This group has been jointly facilitated through Dairy Australia and the Geoffrey Gardiner Dairy Foundation and has been in operation since July This group aims to provide wide representation from all sections of the dairying industry in the area of mastitis and milk quality. It has a strong and well defined leadership role in further refining and facilitating the activity areas in milk quality nationally as described through the Dairy Moving Forward process. Like its predecessor, the Australian Mastitis Advisory Council, the role of AMMQSG is to provide the framework for our continuing conversation and plans around milk quality across all industry levels into the future. Alan Davenport, the current chair of AMMQSG will be chairing the second themed session today. An event of this scale cannot happen without energetic individuals providing support and resourcing. We gratefully acknowledge our co- sponsors, Boehringer Ingelheim and Pfizer. Jonathan Leslie (BI) and his associates have fully supported the organisation of the symposium since its initial planning, continuing their contribution from 2010 and Jane Parker (Pfizer)and her team have come on board this year and we are most grateful for their support. Without the generosity of our two sponsors, we would not have this impressive line-up of guest speakers. Finally, we would like to acknowledge the ongoing partnership of Dairy Australia through their direct support of the day and their investment in Countdown. The team at Harris Park Group have provided the communication and logistics backbone for today. Mastitis control on Australian dairy farms is truly one area of animal performance where the diligence around daily activities makes a difference and where performance is measured each day. The advisers in this room are, through their work with farmers, at the frontline in ensuring we maintain and improve milk quality thereby shoring up farm profitability, sustainability for animals and people and our industry as a whole. Please enjoy and be stimulated by the presentations today. John Penry Graeme Mein Peter Mansell (for the organising team at Countdown and University of Melbourne)

4 Speakers at the symposium Claudia Kampuis (NZ) Originally from the Netherlands, Claudia in now part of the Milk Harvesting and Farm Automation team with DairyNZ which focuses on using automation and information technology to reduce labour and to improve farm productivity. John Morton (AUS) John delivers epidemiological consulting services from his base in Geelong, Victoria. He has worked in private veterinary practice, as a government veterinarian, and as senior lecturer at the University of Queensland s veterinary school. He has published extensively, and is regularly asked to act as a peer reviewer of scientific publications. He is an examiner with the Epidemiology Chapter of the Australian and New Zealand College of Veterinary Scientists. John Penry (AUS) John graduated from the University of Melbourne in John has worked at the Camperdown Veterinary Centre in South- West Victoria since 1991 and has extensive project development and implementation experience across four of the Dairy Australia national programs. He is an associate of the Rural Innovation and Research Group at The University of Melbourne and the current Countdown project leader. Rod Dyson (AUS) Rod has been in dairy cattle veterinary practice since he graduated from the University of Melbourne in He has spent most of his career working in a veterinary dairy practice in the Goulburn Valley and was instrumental in the development of Countdown through most of the project s life. He is the current principal of Dairy Focus, a mastitis consultancy organisation based in Melbourne. Jolanda Jansen (NL) After completing her PhD on communication strategies to improve udder health, Jolanda has emerged as a leading expert on farmer communication strategies. Her research has been presented at many conferences and forums spurring the creation of an informal network of scientists and mastitis advisers interested in social factors relating to disease control. Robert Poole (AUS) Robert has a bachelor of Agricultural Science from the University of Melbourne and a Masters of Business Leadership from RMIT. He is General Manager Shareholder Relations for Murray Goulburn Co-operative and president of the Australian Dairy Products Federation (ADPF) the peak body representing dairy companies. Jakob Malmo (AUS) Jakob works full-time as a cattle specialist in the veterinary practice in Maffra he co-developed. He provides lectures in cattle medicine and production to veterinary students at the University of Melbourne during the last two years of their course and has provided practical instruction to undergraduates at the Rural Veterinary Unit at Maffra for more than 30 years. Finola McCoy (IRE) With keen interest and experience in international models of mastitis control, Finola managed a pilot study evaluating a team based approach to mastitis control when she joined the Teagasc research team in Ireland in Finola is the current leader of the Animal Health Ireland national mastitis control program CellCheck. Pauline Brightling (AUS) Pauline is a veterinarian and educationalist and has spent most of the past decade designing and leading agricultural change management programs, including Dairy Australia s The People in Dairy and Countdown the national industry program for milk quality and mastitis control. 4 Countdown Symposium 2012

5 Automatic mastitis monitoring: will it be the next successful revolution in automation? C. Kamphuis, 1 J. Jago 1 and G. Mein 2 1. DairyNZ Ltd, Hamilton, New Zealand 2. Werribee, Victoria 3030, Australia Future dairy historians may look back on the year 2000 as the dawn of a global revolution in automation and information technologies. Robotic milking systems are just one example of this revolution, with more than 8,000 farms in 25 countries now milking their cows robotically (De Koning, 2010). However, this revolution has had a slow start in Australia and New Zealand. Latest New Zealand figures show that adoption mainly involves technologies that reduce labour during the milking process, like automatic cup removers and teat sprayers (Table 1). Adoption rates of technologies contributing to decision-making processes, like automated mastitis detection, are much lower: only 2% of farmers with herringbone dairies and 4% of farmers with rotary dairies have such systems in place (Table 1). Despite low adoption rates, farmers are interested in investing in automation, including automated mastitis detection systems. This technology ranked third-highest on farmers wish list of technologies they would like to add to their dairies (Table 1). With the revolutionary acceptance of robotic milking in mind, it can be expected that sooner or later electronic sensors, coupled with high-capacity computing and data-mining systems, will replace human visual methods for udder health monitoring. Currently, Australian and New Zealand farmers interested in investing in automated mastitis detection systems can choose from two mainstream technologies to monitor udder health in conventional herringbones and rotary dairies: 1. Electrical conductivity (EC) of milk. Measuring EC of composite milk (i.e. at the cow level) is presently the most commonly adopted mastitis detection technology in New Zealand (J. Jago, unpub. data). The use of composite EC is more likely due to the simplicity and low cost of this technology (using electrodes that are already in place to determine milk level within the milk meter) rather than to its accuracy and reliability for monitoring mastitis. Monitoring EC at the quarter level provides superior accuracy and reliability in detecting mastitis over EC measured at the cow level (Mollenhorst et al., 2010). Monitoring EC at the quarter level has already been applied in robotic milking systems and recently a technique has been developed by Waikato Milking Systems, New Zealand, to also do this in conventional milking systems (Claycomb et al., 2009). Table 1: Technology adoption and technology wish list (% of farmers) by dairy type. Data were collected through a telephone survey among 280 herringbone and 252 rotary farmers in New Zealand (from Cuthbert, 2008). Technology Technology adoption Technology wish list Herringbone Rotary Herringbone Rotary Drafting system Automatic cup remover Mastitis detection system Teat sprayer In-bail feeding system Milk meters Automatic weigh scales Heat detection system Recently, a different technology has been developed by Radian Technology Ltd, New Zealand, which ranks cows according to their relative severity of mastitis signs by analysing complex impedance electrical properties of milk at the cow level. 2. Somatic cell count (SCC) of milk. An in-line system developed by Sensortec, New Zealand, applies principles of the Rapid Mastitis Test to provide an estimation of the SCC at the cow level (Whyte et al., 2004). Using the Sensortec system, sampling and testing of foremilk are automated and an estimate of the SCC is made by using the relationship between milk sample viscosity and actual SCC. The sensor shows a high correlation with laboratory determined SCC of milk, especially at higher SCC levels (Kamphuis et al., 2008). This sensor technology, however, is more expensive than sensor systems measuring EC. In addition, testing involves adding a reagent to the milk and the subsequent sample must be discarded. Light spectroscopy is another technology that can be used to analyse milk components including SCC at the cow level (Arazi, 2009). A New Zealand field study, however, revealed that further development of the sensor algorithm was required to adapt it Countdown Symposium

6 Kamphuis, Jago and Mein to the typical New Zealand cross-bred herd (B. Dela Rue, 2010, unpub. data). On-farm performance Objective and independent performance information Traditionally, mastitis detection systems are evaluated using two performance indicators. First, the sensitivity is calculated as the proportion of all true positives (cows with mastitis) which are detected by the system. Second, the specificity is calculated as the proportion of all true negatives (cows without mastitis) that are correctly not detected by the system. Sherlock et al. (2008) introduced the term number of false alerts per 1,000 cow milkings. This third performance indicator is a more practical measure, referring to the proportion of healthy cows that are falsely identified and that will be checked unnecessarily. This measure may help farmers assess the number of false alerts they think is manageable. Unfortunately, there is a lack of objective information on how well (or badly) mastitis detection systems perform on-farm. A summary of scientific papers assessing the performance of in-line clinical mastitis detection performances is provided by Hogeveen et al. (2010). Sensitivities ranged between 32% and 100% and specificities between 69% and 99.8%. Reported performances are unlikely to be representative of what can be expected on-farm as the models reviewed mainly involved highly selected data, often from a single research farm. Also, reported performances were not comparable due to large variations in study characteristics. Hogeveen et al. (2010) concluded that data used for analyses should be collected from multiple farms to reflect farm practices, that a mastitis detection system should alert for clinical mastitis within a short time-window (24-48 hours) and that detection systems should operate with at least 80% sensitivity and 99% specificity (i.e. <10 false alerts per 1,000 cow milkings). To date, only one scientific study which was validated on a commercial, pasture-based farm, has reported a detection model based on quarter EC measurements. That model performed with 2.3 to 7 false alerts per 1000 cow milkings in combination with a sensitivity between 68% and 88% (Claycomb et al., 2009), depending on the choice of alertthreshold and on the applied definition of clinical mastitis. Kamphuis et al. (2010) validated a detection model on nine commercial farms and reported a large variation in sensitivity (range: 13-71%) at a fixed number of ten false alerts per 1000 milkings. It was unclear though, as to what caused the range in performance between farms and how much of this variation was attributed to the system or to farm management. Farmer experiences and perceptions There is little information on how system performances are perceived by farmers. Cuthbert (2008) reports that satisfaction levels of early adopters of these mastitis detection systems varied largely, with 60% of farmers reporting they are satisfied or very satisfied with their investment. There is no clear explanation why the other 40% were not satisfied, but reasons may include systems not meeting farmers expectations, insufficient training or support from technology suppliers or simply a low detection performance of the system. Shortcomings of these technologies are supported by comments (see below) from three experienced Australian farm advisers on current applications of sensing technologies (Dairy Moving Forward, 2010): People have blind faith in some of these systems, without proper training it s almost a waste having them there. 19 out of 20 farmers have no idea how to use their automated system and what to get out of it. Maintenance is also an issue. I see farms with systems that are not being used because no one has been able to keep them working properly. So, it is clear from anecdotal stories from the field that some of these systems are a help on some farms and a hindrance on others. An example is a New Zealand farmer who is highly positive about his mastitis detection system: Our detection system is based on SCC and is set to alert cows above a certain threshold. At the start of the season this facility is also used to draft cows into the colostrum herd for extra calf milk if required. Normally, the system is set to alert at cups-on so staff can strip the cow and check for mastitis. The cow is then drafted if clinical and requires treatment otherwise she is left in the herd. Non-clinical high SCC cows repeatedly alert, but staff soon learn and remember these animals. We get about 20 alerts per milking with the alert SCC threshold set at 400,000 cells/ml. We tend to find the mastitis earlier than by manual detection alone and so far we have never had to strip the whole herd which is a huge labour saving. On a less optimistic note, two New Zealand farmers monitoring EC at the cow level reported their systems generated too many false alerts which affected milking procedures. One farmer s advice to other farmers interested in investing in such a system would be to look at other technologies available. The second farmer turned the system off and would not advise other farmers to buy it. It has to be noted, however, that many other farmers using EC at the cow level do report acceptable performances and are satisfied with this sensing technique. Improving on-farm performance Recently, a Dutch research group studied different approaches to improve automated detection of clinical mastitis with robotic milking (Kamphuis, 2010; Steeneveld, 2010). The main conclusions were that: a data mining technique, known as decision-tree induction, improves detection performance; detection performance is improved when sensor information is combined; adding non-sensor cow information to a sensor-based mastitis detection model did not improve detection performance; and 6 Countdown Symposium 2012

7 Automatic mastitis monitoring non-sensor cow information was not useful for discriminating between true positive and false positive clinical mastitis alerts. Another way to improve detection performance is to develop systems that better meet farmers expectations. To research these expectations, Mollenhorst et al. (2012) conducted an adaptive conjoint analysis to define farmers preferences for mastitis alert lists when milking robotically. According to that study, farmers prefer mastitis detection systems that are adaptable to individual preferences, that systems should be evaluated at high levels of specificity (>99%) and that time-windows should be kept short (preferably 24 h or less). Steeneveld et al. (2011) suggested the development of treatment protocols for farmers with robotic milking systems to deal better with the imperfect detection results of current systems. This Dutch research involved robotic milking systems only, but these approaches should be applicable to improve on-farm performance of mastitis monitoring systems in conventional milking systems also. Future steps towards a better understanding Towards uniform evaluation and presentation of detection performance It is not surprising that so little independent information on evaluating on-farm performances is available. Evaluating a system is a costly procedure and guidelines describing how to do this are lacking. Therefore, the first step towards a better understanding on how mastitis detection systems perform on-farm is to evaluate these systems in the field using a uniform methodology and to report performance in a similar and understandable manner. Kamphuis et al. (2011) proposed a framework which describes gold standards, evaluation protocols and performance targets for three outcome-focused performance requirements to identify cows: 1. Clinical mastitis, detected promptly and accurately. A mastitis detection system should operate with at least 80% sensitivity and 99% specificity (i.e. <10 false alerts per 1000 cow Table 2: Influence of time-window (TW, length in hours) on detection performance for an automated clinical mastitis (CM) detection model (from Kamphuis et al., 2010). TW (h) before CM TW (h) after CM observation observation Sensitivity (%) at a fixed specificity of 99% < milkings). It is known that detection performance can increase considerably with increasing lengths of time-window (Table 2). However, the time-window used for evaluation should be practical and is likely to be around 48 h (Claycomb et al., 2009; Hogeveen et al., 2010). Analyses should be based on at least 20 cases of clinical mastitis per farm, collected from at least three farms each, where the observation of clots in two out of three consecutive milkings on in-line filter screens are used to define clinical events (Claycomb et al., 2009; Mein and Rasmussen, 2008). 2. High SCC cows, detected for managing the bulk milk SCC. This involves ranking cows based on their total SCC contribution to the vat and excluding cows one by one in order to decrease bulk milk SCC. A mastitis detection system should not exclude more than twice the percentage of the herd as indicated by herd test results to decrease the bulk milk SCC by a certain percentage. Figure 1 displays an example (Kamphuis et al., 2011) where cows are ranked according to their total SCC contribution using herd test data and when two mastitis detection systems are used (one SCC-based, one EC-based). If the goal is to decrease bulk milk SCC by 25%, herd test data indicate that the top 2% of the herd (ie, cows that add the most somatic cells into the farm milk supply) would need to be excluded. The SCC-based system also indicates that a similar percentage of cows would have to be excluded to decrease bulk milk SCC by 25%. However, when the EC-based system was used, Figure 1 indicates that the top 9% of cows would have to be excluded. % decrease in bulk milk SCC Figure 1: Effect on bulk milk somatic cell count (SCC) after excluding milk from cows with the highest (SCC) contribution to the bulk tank, according to SCC values measured at herd test days (gold standard, black line) and two sensing system using either electrical conductivity (grey line) or SCC technology (dotted line); represents a 25% decrease in bulk milk SCC (adapted from: Kamphuis et al., 2011). % of the ranked cows excluded from the bulk tank. Countdown Symposium

8 Kamphuis, Jago and Mein 3. Subclinical mastitis, detected for use in decisions regarding use of dry cow therapy. To evaluate the ability of a sensing system to detect cows which are infected (clinically or sub-clinically), the gold standard should be based on true infection status. This implies aseptic sampling and culturing of a random or stratified sample of cows in any test herds. Towards understanding the people-side of mastitis detection technology A second step towards a better understanding on how mastitis detection systems perform on-farm is to gain insight into the people-side of this technology. Preliminary results from a New Zealand survey currently being conducted among early adopters suggest large variations in investment reasons, including saving labour, finding clinical mastitis, or no specific reason ( it was a package deal ). Most farmers were quite satisfied, but when asked what they now wished they knew before investing, their responses include I expected a better association with SCC, promised capabilities did not happen, and I wish I knew when and how to alter the alert-threshold. Surveyed farmers advice to other farmers who have just invested in a system included responses such as: Master the basics of the software, then concentrate on details, Ensure good maintenance as these systems rely on fully-working sensors and Be persistent and ask suppliers when things are not working. Conclusions and recommendations The investment in automated mastitis detection systems has been perceived as a help by some early adopters and as a hindrance by others. In order that these systems are beneficial on all farms, is it essential to gain a better understanding of their current performance, farmers expectations regarding performance and where these systems require improvement. To make the adoption of mastitis detection systems the next successful revolution in automation, it is essential that: 1. Results produced by the sensing system are accurate. It is human nature to quickly learn to ignore alerts or alarms if farmers find that most are false. On the flip side, if only a proportion of true positive events are flagged, then reliance on the data for detection or alerting is rapidly decreased. 2. Cow ID is accurate and reliable. Cow identification is an essential component of any data monitoring system. It must be reliable and the frequency of lost electronic cow identification devices minimised. 3. There is agreement on evaluation criteria. These criteria must include agreed practical reference standards and test protocols that should be used to evaluate performance and how performance should be reported to the industry. These guidelines should be used to evaluate current mastitis sensing technologies for their ability to monitor udder health. 4. The industry capacity to support retailers and users. Suppliers currently provide training to their clients but are limited by financial resources, location and availability of appropriately skilled trainers. There is a role for consultants and industry extension personnel to provide advice and training for farmers after the initial installation period. References Arazi A Automated daily analysis of milk components and automated cow behaviour: developing new applications in the dairy farm, ICAR Technical Series 13, pp Claycomb R.W., P.T. Johnstone, G.A. Mein, R. Sherlock An automated in-line clinical mastitis detection system using measurement of conductivity from foremilk of individual quarters. New Zealand Veterinary Journal 57(4): Cuthbert, S DairyNZ milking practices and technology use survey, pp 17-18, 21 and 25. Report prepared for DairyNZ. Dairy Moving Forward, Report of Expert Group on Information and Automation Technologies, Animal Performance Strategy 4, 2009, Australia. De Koning, C.J.A.M Automatic milking common practice on dairy farms. Proceedings of the first North American Conference on Precision Dairy Management, Toronto, Canada, pp Hogeveen, H., C. Kamphuis, W. Steeneveld, H. Mollenhorst Sensors and clinical mastitis the quest for the perfect alert. Sensors. 10: Kamphuis, C., R. Sherlock, J. Jago, G. Mein, and H. Hogeveen Automatic detection of clinical mastitis is improved by in-line monitoring of somatic cell count. Journal of Dairy Science. 91(12): Kamphuis, C Making sense of sensor data: detecting clinical mastitis in automatic milking systems. PhD thesis. Faculty of Veterinary Medicine, Utrecht University, the Netherlands. Kamphuis, C., H. Mollenhorst, J.A.P. Heesterbeek, H. Hogeveen Detection of clinical mastitis with sensor data from automatic milking systems is improved by using decision-tree induction. Journal of Dairy Science. 93: Kamphuis, C., Dela Rue B., Mein G. and Jago J Practical evaluation of automatic In-line mastitis sensors. Proceedings of the 3rd International on Mastitis and Milk Quality, National Mastitis Council, St. Louis, USA, pp Symposium on Mastitis and Milk Quality, National Mastitis Council, St Louis, USA. Mein G.A, Rusmussen MD Performance evaluation of systems for automated monitoring of udder health: would the real gold standard please stand up? Mastitis control From science to practice, pp , Wageningen Academic Publishers, the Netherlands. Mollenhorst, H., P.P.J. van der Tol, H. Hogeveen Somatic cell count assessment at the quarter or the cow milking level. Journal of Dairy Science. 93: Mollenhorst, H., L.J. Rijkaart, H. Hogeveen Mastitis alert preferences of farmers milking with automatic milking systems. Journal of Dairy Science. 95: Sherlock, R., H. Hogeveen, G. Mein, M.D. Rasmussen Performance evaluation of systems for automated monitoring of udder health: analytical issues and guidelines. Mastitis control From science to practice, pp T.J.G.M. Lam (ed.). Wageningen Academic Publishers, Wageningen, the Netherlands. Steeneveld, W Decision support for mastitis on farms with an automatic milking system. PhD Thesis. Faculty of Veterinary Medicine, Utrecht University, Utrecht, the Netherlands. Steeneveld, W., C. Kamphuis, H. Mollenhorst, T. Van Werven, and H. Hogeveen The role of sensor measurements in treating mastitis on farms with automatic milking systems. Proceedings of the International conference on udder health and communication, pp , H. Hogeveen and T.J.G.M Lam (eds.), October 2011, Utrecht, the Netherlands. Whyte, D., R. Orchard, P. Cross, T. Frietsch, R. Claycomb, G. Mein An online somatic cell count sensor. Automatic milking: a better understanding, pp A. Meijering, H. Hogeveen, C.J.A.M. de Koning (eds.), Wageningen Academic Publishers, the Netherlands. 8 Countdown Symposium 2012

9 Detecting clinical mastitis with automatic milking system assistance and with other aids John Morton, 1 Jade Hammer 2 and Kendra Kerrisk 3 1. Jemora Pty Ltd, PO Box 2277, Geelong, Victoria 3220, Australia 2. Main Street Veterinary Clinic, 325 Main Street, Bairnsdale, Victoria 3875, Australia 3. University of Sydney, Private Mailbag 4003, Narellan, NSW 2570, Australia Automatic milking systems (AMS) in dairy herds routinely collect and store a wide range of data. Some data are at the quarter rather than cow level. Milk from each quarter can be analysed separately and variables such as duration of milking can vary between quarters. This is in contrast to traditionally milked herds, where milk from all quarters is pooled, and where all four cups are removed simultaneously. Further, because AMS cows can choose when they will be milked, the intervals between milkings vary considerably more than in traditionally milked herds, facilitating study of the effects of this factor on clinical mastitis risk. Because some of the variables routinely collected in AMS are known or likely to be predictive of clinical mastitis occurrence, we conducted a study that assessed predictive models for clinical mastitis based on these risk factors and indicators. Full details of the study have been published (Hammer et al. 2012). Using data from the AMS research dairy herd at the Elizabeth Macarthur Agricultural Institute, Camden, Australia, we studied 245 clinical mastitis episodes (quarter-milkings at which a new case of clinical mastitis occurred as diagnosed by farm staff) and 2,450 quarter-milkings at which no new clinical mastitis episode occurred within the previous 30 days). (A quarter-milking occurs when a given quarter is milked once.) Quarters that were milked infrequently, and that had low yield, fast peak milk flow rates, blood in milk and/or elevated milk conductivity were at increased risk of clinical mastitis. Quarters were also at increased risk between days 10 and 29 of lactation, and during higher parity lactations. Two models to predict which quarter-milkings had new cases of clinical mastitis were assessed: we assessed the ability of the models to identify which quarters were affected by a new case of clinical mastitis and at which milkings these cases first occurred. (No attempt was made to predict new infections; the study focussed on new clinical cases.) The first model consisted of lactation number, days in milk, and for days 0-7 prior, number of milkings, average yield, average of peak flow rates during milkings, number of times blood concentration in quarter milk exceeded 300 ppm, and number of times absolute electrical conductivity of quarter milk exceeded 5,000 μs/cm. The second model included these same variables and the cow s two most recent somatic cell counts. For each quarter-milking, the models generated a probability of a new case of clinical mastitis, based on the status of quarter-milking for all of the variables listed above (Table 1). For example, the model predicted a very low probability of clinical mastitis (0.003) for quarter-milkings in 1 st lactation cows that were at least 100 days in milk and for days 0-7 prior had at least 14 milkings, average yields of at least 4 L/quarter/milking, averages of peak flow rates during milkings from 1 to 2 L/ minute, and no milkings at which blood exceeded 300 ppm or absolute electrical conductivity exceeded 5000 μs/cm (Table 1). The model predictions were then compared to actual clinical mastitis occurrences as diagnosed by farm staff. At any particular probability threshold, the model sensitivity was calculated using quarter-milkings with new cases of clinical mastitis; sensitivity was the proportion of these that had Table 1: Statuses for seven variables for two quarter-milkings, and probabilities of clinical mastitis as predicted by model. Variable Example 1 Example 2 Lactation number 1 10 Days in milk to <20 For days 0-7 prior: Number of milkings 14 <7 Average yield (L/qtr/milking) 4 <1 Average of peak flow rates during milkings (L/min) 1 to <2 2 Number of times blood exceeded 300 ppm 0 3 Number of times absolute electrical conductivity exceeded 5000 μs/cm 0 4 Probability of clinical mastitis as predicted by model Countdown Symposium

10 Morton, Hammer and Kerrisk 1.00 Sensitivity/Specificty Specificity (final model with SCC) Sensitivity (final model with SCC) Specificity (final model) Sensitivity (Final model) Threshold Figure 1: Sensitivity and specificity of two models of risk factors and indicators for quarters becoming affected with clinical mastitis during lactation in pasture-fed dairy cows managed in an automatic milking system at various probability thresholds; final model (solid lines) consisted of lactation number, days in milk, and for days 0-7 prior, number of milkings, average yield, average of peak flow rates during milkings, number of times blood exceeded 300 ppm, and number of times absolute electrical conductivity exceeded 5000 μs/cm and second model ( final model with SCC ; dashed lines) included these same variables and the cow s two most recent somatic cell counts (Reproduced from Hammer et al. (2012) with permission). predicted probabilities above the threshold. Model specificity was calculated using quarter-milkings without clinical mastitis; specificity was the proportion of these that had predicted probabilities below the threshold. Both models had high specificity but low sensitivity (Figure 1). It is possible to increase sensitivity by using a lower threshold but this would result in very low specificity. In most situations, this would result in very low positive predictive values. This would mean that most quarter-milkings with predicted probabilities above these lower thresholds would, in fact, not have clinical mastitis. We conclude that these models are not adequate for assisting in the detection of clinical mastitis in herds with automatic milking systems. More complex models may have acceptable performance but very high specificity is required to minimise the frequency of false positive alerts. (Alerts would occur for quarter-milkings whose predicted probability is above a preset threshold). Requirements for acceptable sensitivity and specificity for automatic detection of abnormal milk have been published (>70% and >99%, respectively; ISO, 2007) and Sherlock et al. (2008) proposed a third measure: the number of false alerts per 1,000 milkings. One limitation of this latter measure is that it is affected by the prevalence of clinical mastitis in the herd (a factor that is not inherent in the predictive model), thus potentially affecting comparisons of models between studies and herds. However, this measure is affected much more by specificity than prevalence of clinical mastitis (Table 2). In fact, unless specificity is at least 99.5%, the number of false alerts (Table 2) would probably be unacceptable to many herd managers. Even higher specificity would be required if quarterlevel alerts are generated. Detecting clinical mastitis with and without other aids This work aimed to explore the probability that a quarter affected by clinical mastitis would be detected under various clinical mastitis detection strategies. A simple deterministic model was developed, and probabilities of detection assessed under each of the following strategies: Clinical mastitis cases detected based on monitoring by milking staff without routine handstripping of all quarters As above but aided by in-line filters Relying solely on handstripping of all quarters following detection of clots on the filter in the milk delivery line at the end of a herd milking Relying solely on routine handstripping of all quarters once weekly A proportion of clinical mastitis cases were assumed to remain clinical until the signs were detected; all remaining cases were assumed to resolve spontaneously if not detected and treated after a specified duration. Probabilities were estimated for detection a) at the first milking after onset of clinical signs, b) by the second milking after onset of clinical signs, and c) before the clinical signs resolve spontaneously. Table 2: Numbers of false alerts per 1000 milkings for various specificities and prevalences of clinical mastitis. Sensitivity was set at 80%. Specificity 95% 99% 99.5% Prevalence of clinical mastitis 0% 10% 0% 10% 0% 10% Number of false alerts per 1,000 milkings Countdown Symposium 2012

11 Detecting clinical mastitis with automatic milking system assistance Table 3: Input variables and values used in a simple model of the probability that a quarter affected by clinical mastitis would be detected under various clinical mastitis detection strategies. Variable Value Basis for value Probabilities for various detection activities Probability that a clinically affected quarter will be detected at a single milking based on visual appraisal; udder palpation and/or handstripping only used for suspect quarters 0.5 Expert opinion 1 As above combined with use of in-line filters 0.6 As above combined with Hoyle and Dodd 1970 Probability that clots will be detected on the delivery line filter given that one clinically affected quarter was milked into vat (400 cow herd) 0.7 Expert opinion Probability that a clinically affected quarter will be detected at a single milking by hand stripping each quarter: Handstripping all quarters at the milking after clots were detected on the herd filter 0.8 Expert opinion 1 and Dodd et al Routine handstripping, i.e. handstripping all quarters even though no clots had been detected recently on the herd filter 0.7 Expert opinion and Dodd et al Other inputs Typical clinical mastitis cases: Proportion of clinical cases that remain clinical until the signs are detected 0.2 Plausible value For other cases, duration of clinical signs in the absence of treatment (days) 6.5 Nash et al Mild clinical mastitis cases: Proportion of clinical cases that remain clinical until the signs are detected 0.13 Chamings 1984 For other cases, duration of clinical signs in the absence of treatment (days) 3.2 Chamings Expert opinion for milking staff who are doing a good job in detecting clinical cases 2. Although the text is unclear, cases in this study were probably treated. If so, this value of 6.5 days is probably too low. Input values were based on expert opinion (Drs Graeme Mein, Rod Dyson, Peter Younis, John Penry and Jakob Malmo), published research and plausible values. For simplicity, each clinical mastitis case was assumed to have the same duration for a particular scenario (either 3.2 or 6.5 days; Table 3), probabilities were assumed to not change with repeat applications of the same detection activity for the same clinical mastitis case (i.e. all applications were considered to be independent of each other), and the specificity of each detection activity was assumed to be 100%. Twice-daily milking was assumed. Input variables and values used are shown in Table 3. Results of modelling are shown in Table 4. For all strategies, predicted probabilities of detection at the first milking were low, and predicted probabilities of detection by the second milking were, at best, only modest. Under these assumptions and input values: important proportions of clinical mastitis cases are not detected by the second milking and so will be milked into the vat under the first three strategies, virtually all typical and mild cases would be detected before clinical signs resolve spontaneously in-line filters result in marginal improvements in predicted probabilities of detection at the first milking and by the second milking relying solely on routine handstripping of all quarters once weekly will result in delayed detection and non-detection of many cases. These results are highly sensitive to some of the input values, and these input values will vary by herd. For example, the Table 4: Predicted probabilities that a clinical mastitis case would be detected under various clinical mastitis detection strategies. Strategy Clinical mastitis cases detected based on monitoring by milking staff without routine handstripping of all quarters At first milking 1 By second milking 2 Before clinical signs resolve spontaneously Typical Mild As above but aided by in-line filters Relying solely on handstripping of all quarters following detection of clots on the herd filter at the end of milking (assuming one clinical quarter was milked into vat (400-cow herd) Relying solely on routine handstripping of all quarters once weekly At first milking after onset of clinical signs 2. By second milking after onset of clinical signs Countdown Symposium

12 Morton, Hammer and Kerrisk experts considered that the probability of detection of a clinically affected quarter at a single milking by hand stripping of each quarter depends on many factors including: whether milking staff are rigorous or going through the motions milking staff fatigue assessing too many quarters at a single milking assessing quarters too quickly attitude to risk of missing cases the extent of desire by milking staff to not detect cases poor lighting in the dairy stripping milk onto the concrete platform. This model includes a number of simplifying assumptions, and results may change if some of these complexities are addressed. In particular, the model would be improved by allowing the duration of clinical signs to vary between cases, and by assuming applications of the same detection activity for the same clinical mastitis case are dependent. For example, if a clinically affected quarter is not detected at a particular milking, the probability of detection by the same method at the next milking would be reduced. It is also desirable to assess combinations of strategies, e.g. clinical mastitis cases detected based on monitoring by milking staff combined with handstripping of all quarters if clots are detected on the herd filter. However, the validity of the current model is uncertain due to limited knowledge about input values relevant for commercial dairy herds, and this lack of knowledge would also affect interpretation of more complex modelling. References Chamings RJ (1984) The effect of not treating mild cases of clinical mastitis in a dairy herd. Veterinary Record 115: Dodd FH, Westgarth DR, Neave FK, Kingwill RG (1969) Mastitis the strategy of control. Journal of Dairy Science 52: Hammer JF, Morton JM, Kerrisk KL (2012) Quarter-milking-, quarter-, udderand lactation-level risk factors and indicators for clinical mastitis during lactation in pasture-fed dairy cows managed in an automatic milking system. Australian Veterinary Journal 90: Hoyle JB, Dodd FH (1970) The detection of clinical mastitis with in-line filters. Journal of Dairy Research 37: ISO (2007) ISO/DIS 20966: Automatic milking installations Requirements and testing. International Organisation for Standardisation, Geneva, Switzerland. Nash DL, Rogers GW, Cooper JB, Hargrove GL, Keown JF (2002) Relationships among severity and duration of clinical mastitis and sire transmitting abilities for somatic cell score, udder type traits, productive life, and protein yield. Journal of Dairy Science 85: Sherlock R, Hogeveen H, Mein G, Rasmussen MD (2008) Performance evaluation of systems for automated monitoring of udder health: Analytical issues and guidelines.; In: Mastitis Control From Science to Practice. Lam TJGM, Editor, Wageningen Academic Publishers, Wageningen, The Netherlands, pp Countdown Symposium 2012

13 Molecular tests for milk performance and application in Australia John Penry, 1 John Morton, 2 Jakob Malmo 3 and Graeme Mein 4 1. Primary Logic Pty Ltd, PO Box 219, Camperdown, Victoria 3260, Australia 2. Jemora Pty Ltd, PO Box 2277, Geelong, Victoria 3220, Australia 3. Maffra Veterinary Clinic, Maffra, Victoria 3860, Australia 4. Werribee, Victoria 3030, Australia Molecular test technologies, such as polymerase chain reaction (PCR), are not new in either human or veterinary medicine diagnostics. However, the application of these technologies to the assessment of milk for the presence of mastitis pathogens is relatively recent. In the mid 2000s, the University of Melbourne Veterinary Science Faculty developed a PCR for Streptococcus agalactiae; this test had a limited commercial life. Subsequently, the Livestock Teaching Unit, University of Sydney, developed a separate molecular test for Strep. agalactiae and Mycoplasma spp. using loop-mediated isothermal amplification (LAMP). This test is in the research and validation phase. In early 2011, a new milk molecular test, the PathoProof PCR, became commercially available in Australia, through Dairy Technical Services. Because the validity of various possible applications of molecular testing in milk were undefined, in July 2011, Countdown Downunder, through the generous support of the Geoffrey Gardiner Dairy Foundation, commenced a research project to assess the performance, application and interpretation of these tests. This research will continue until September This paper presents interim findings to May Further communication on the overall results of this research will be available to farmers and advisers during 2012/13 through Countdown. The partner in this research has been Dairy Technical Services with valued project assistance from the University of Sydney, Fonterra, Murray Goulburn Coop, Pfizer and Gribbles Pathology. This paper focuses largely on the PathoProof PCR milk molecular test, as it has been the primary testing subject for this research project. However, it is highly likely that the principles of test application determined through this work will apply to other milk pathogen molecular tests (for example, LAMP) if they become commercially available. Finding the conceptual fit for this test Countdown has developed the Herd Mastitis Dynamics Chart and used this in adviser and farmer training. This chart was an attempt to represent the movement of animals between non-infected and infected groups in any herd, regardless of size, calving system, management and location. In this chart, at any point in time, each animal has only one of two possible mastitis states they are either non-infected, where no mastitis pathogens are present in any quarter, or they are infected with one or more mastitis pathogens in at least one quarter. Where an animal is infected, the individual cow cell count will generally be greater than 250,000 cells/ml. This chart (Figure 1) can be used to describe some of the factors which can lead to new infections such as poor performance in teat disinfection, milking machine function, milking management and environmental control. The factors which determine cures or animals going from the infected to the clean group are largely limited to lactation treatments and dry cow treatments. In general, there are far more factors that can effectively influence or drive the new infection rate than there are factors that increase rate of cures. Identifying major factors influencing spread in the herd is vital for tailored, herd-specific, mastitis control. Countdown Technote 13 (extract in Figure 2) outlines the steps an advisory team should undertake when investigating a mastitis problem herd. The extract in Figure 2 details the start of this process. When describing the presenting problem, a number of tests can be applied to provide information about the nature and extent of new and more chronic infections. These include: a) bulk milk cell count (BMCC) data; b) individual cow cell count (ICCC) data; and c) standard milk culture of samples from individual cows. Some advisers also employ other ancillary tests such as the Rapid Mastitis Test, milk electrical conductivity and data from other in-line mastitis sensing technologies to assess infection status. Identifying the predominant mastitis pathogens is a key step in this, as this knowledge assists in determining the mechanisms of spread, Countdown Symposium

14 Penry et al. Figure 1: Countdown Herd Mastitis Dynamics Chart. and hence, the major factors influencing spread. The questions this research has attempted to answer are, in the Australian context: 1. Where does milk molecular testing fit among these other tests? 2) How should such tests be interpreted both during a mastitis investigation and where mastitis risk mitigation work is being undertaken (Countdown MAX service model)? PCR test outline The PCR is an example of one type of molecular test for detecting mastitis pathogens. The test differs from standard milk culture as it is designed to identify target strands of DNA uniquely associated with each organism. Recall that DNA is double-stranded, with the strands crosslinked to each other along their entire length by links between Figure 2: Extract from Countdown Technote 13 Mastitis Investigation Flowchart. bases (adenine linking to thymine, and guanine linking to cytosine). The polymerase chain reaction synthesises a very large number of copies of a specified sequence (or section) of the DNA (target DNA) in a sample. The target DNA sequence is defined by short sequences either side of it; primers are necessary to bind to these short sequences. A PCR reaction cycle consists of three steps: 1. Denaturation: The sample is heated, denaturing the DNA from double to single strands. 2. Annealing: The sample is cooled, allowing the primers to bind to short sequences either side of the target DNA sequence. One primer binds to each DNA strand. 3. Elongation: Using DNA polymerase, on each of the single strands, the complementary strand is partially added, commencing at the primer and continuing for a variable distance along the strand. Thus, the single strands become double strands for this distance. This three-step cycle is repeated; the number of double stranded products doubles with each cycle. By the third cycle, some of the double-stranded products represent the target DNA only between the short sequences either side. With further cycles, these become the predominant product in the mixture. After 30 cycles, the original DNA had been amplified a billion-fold. The PathoProof PCR is a real time PCR. With real time PCR assays, a positive reaction occurs when a fluorescent signal is detected. A positive result can be expressed as the 14 Countdown Symposium 2012

15 Molecular tests for milk Table 1: A comparison of the attributes of PCR and standard milk culture (adapted from Bradley et al. 2011). Bacteria identified Sampling PCR - At present, PathoProof detects 12 organisms - PathoProof can also detect the penicillin resistance gene - PCR can detect dead and live bacteria - PCR can be carried out on milk samples treated with preservative, allowing normal milk recording samples to be used for mastitis testing Standard milk culture (bacteriology) - Bacteriology has potential to identify a wide range of bacteria, including more exotic bugs - It does not detect Mycoplasma species- a rare cause of mastitis in Australia - A separate test must be carried out to identify penicillin resistance - Bacteriology can only identify live bacteria - Bacteriology using milk samples treated with preservatives is uninformative - No growths can result when bacteria die between collection and plating. Interpretation Contamination - Because PCR is a new technology in Australia, there is limited knowledge around interpreting results; Countdown is currently completing research in this area - Vets and farmers need to be educated in interpretation of results - There is currently no good measure for identifying contaminants from the udder and teat skin identified through PCR - Bacteriology has been used for many years in Australia, providing extensive experience in interpretation - Most farmers and vets know how to interpret results - Bacteriology allows easier identification of contamination Time PCR testing takes less than 4 hours Bacteriology takes hours Cost About $45 per milk sample About $15-$20 per milk sample cycle threshold (Ct). Ct is the number of cycles required for the fluorescent signal to become strong enough to be distinguishable from background values. The PathoProof PCR being evaluated as part of this research is available commercially in two testing formats. The first format tests for the presence of DNA from four organisms: Strep. agalactiae, Staphylococcus aureus, Streptococcus uberis and Mycoplasma bovis. A second format tests for these organisms and another eight organisms, along with the gene that confers penicillin resistance in Staph. aureus. Of the organisms targetted using this PCR, only Strep. agalactiae and Mycoplasma bovis originate only from inside infected quarters. All other organisms can replicate outside the udder, so unless aseptic milk samples are collected, they can originate from teat skin and the milking plant. Milk samples can be from individual quarters, individual cow samples (milk pooled from all four quarters from the same cow), samples from groups of cows pooled, or bulk vat samples. The presence of milk preservative, such as bronopol, does not interfere with the test function and contamination of the sample also does not interfere with the conduct of the test. The key differences between PCR and standard milk culture are summarised in Table 1. The Countdown research project includes the following components: A review of literature describing the diagnostic validity of PCRs for detecting mastitis pathogens in milk from dairy cows Assessment of no growth standard culture samples Dilution studies and other methodologies to assess PCR performance in bulk milk Research to estimate prevalences of pathogens Using pooled herd test-sampled milks A review of the literature The purpose of the review of the literature describing the diagnostic validity of PCRs for detecting mastitis pathogens in milk from dairy cows was to summarise and critically evaluate scientific evidence about the diagnostic validity of polymerase chain reaction (PCR) assay results for detecting mastitis pathogens in milk from dairy cows. The review particularly focused on the PathoProof PCR because this assay is being offered commercially in Australia, but results for other PCRs were also reviewed for comparison. Lower limits for diagnostic sensitivity and specificity that are tolerable under various scenarios were also explored. More than 40 papers were assessed and much of this work had been undertaken in Europe. Of the seven publications from which diagnostic sensitivities and specificities for the PathoProof PCR were summarised (see References), only three had been peer-reviewed. Of the five publications from which diagnostic sensitivities and specificities results for other PCRs were summarised, four had been peer-reviewed Disregarding results known to be based on small numbers of samples, relative diagnostic sensitivities of the PathoProof PCR were: Staph. aureus 87% and 94% (2 results); Strep. agalactiae 90% (1 result); Strep. dysgalactiae 89% and 100% (2 results); Strep. uberis 88% and 100% (2 results); Corynebacterium bovis 77% and 80% (2 results); coagulase negative Staphylococcus spp. 80% (1 result). In one study, diagnostic sensitivities of the PathoProof PCR for Strep agalactiae (from latent class models) were estimated as 96%, 92%, 87% and 74%, respectively, at PCR Ct value cutoffs of 39, 37, 34, and 32. No results allowed calculation of relative diagnostic sensitivity for the PathoProof PCR for Mycoplasma bovis. Disregarding results known to be based on small numbers of samples, relative diagnostic specificities of the Countdown Symposium

16 Penry et al. PathoProof PCR were: Staph. aureus 92% to 99% (5 results); Strep. agalactiae 97% to 100% (5 results); Strep. dysgalactiae 90% to 99% (4 results); Strep. uberis 80% to 97% (4 results); C. bovis 69% to 92% (4 results); coagulase negative Staphylococcus spp. 62% to 86% (3 results). Diagnostic specificities of the PathoProof PCR for Strep. agalactiae (from latent class models) were estimated as 97%, lower than estimates for culture from the same study. No results allowed calculation of relative diagnostic specificity for the PathoProof PCR for Mycoplasma bovis. Interim conclusions from the review are: Diagnostic sensitivity must be above 99% when false negative test results are expensive and the particular organism is quite likely to be present in the milk sample, but lower sensitivities are tolerable when the particular organism is less likely to be present in the milk sample. High to very high diagnostic specificities are required (99% to 99.99%) when false positive test results are expensive. Diagnostic sensitivities of the PathoProof PCR for Staph. aureus, Strep. agalactiae, Strep. dysgalactiae and Strep. uberis are probably at least moderately high at high Ct cut-offs. However, diagnostic sensitivities of this PCR should not be assumed to be 100% and negative results should not be considered to be unequivocal proof that the organism was not present. When false negative test results are expensive and the particular organism is quite likely to be present in the milk sample, additional strategies to improve overall confidence about the absence of particular organisms are required. (Such strategies are probably also required if culture were used instead of the PathoProof PCR.) Diagnostic specificities of the PathoProof PCR for Staph. aureus, Strep. agalactiae, Strep. dysgalactiae and Strep. uberis are probably moderately high. However, diagnostic specificities of this PCR should not be assumed to be 100% and positive results should not be considered to be unequivocal proof that the organism was present. If the prior probability that the particular organism is present in the milk sample is low, a substantial proportion of PCR positives are likely to be false positives. Further studies assessing the diagnostic sensitivities and specificities of the PathoProof PCR with improved study designs are required. Studies to assess the diagnostic sensitivity and specificity of the PathoProof PCR for Mycoplasma bovis are required. In the studies reviewed, PCRs were almost always assessed at quarter and cow levels yet potentially, practical applications under Australian conditions may be more likely at group (i.e. pools of cows) or vat (i.e. herd) level. There is a need for information about the interrelationships in diagnostic sensitivities and specificities at these latter levels. Assessment of no growth standard culture samples As standard culture does not identify Mycoplasma bovis, it is possible that some clinical mastitis cases where there is no growth on standard culture are due to Mycoplasma bovis. The aim of this component was to estimate the prevalence of PCRpositives for Mycoplasma bovis in milk samples from clinical mastitis cases where there was no growth on standard culture. Fifty-five milk samples from clinical mastitis cases where there was no growth on standard culture were tested using the PCR. No PCR-positives for Mycoplasma bovis were identified. Assuming no prior knowledge about this question, this result indicates that we can be 95% sure that the true proportion of such cases that are Mycoplasma bovis PCR positive is no more than 6%. There were also no Strep agalactiae positives on PCR. Dilution studies and other methodologies to assess performance in bulk milk As previously indicated, much of the peer-reviewed research on the diagnostic validity of the PCR was undertaken quarter samples or individual cow samples. The Countdown research group identified the need for further information about potential test performance where samples from groups of cows pooled, or bulk vat samples were used. Dilution studies were performed using individual cow samples positive for Strep agalactiae and Mycoplasma bovis on the PathoProof PCR Six individual cow samples positive to either bacteria on standard or selective culture were diluted down with milk negative on the PathoProof PCR for these pathogens at dilution rates of 1/10, 1/100, 1/500 and 1/1000. These pathogens were chosen as it seems most likely that bulk vat testing would be most useful for these, based on the ecology of the bacteria as described previously. For both pathogens, all individual cow samples were positive on PCR when tested undiluted and then at all serial dilutions that followed. These results suggest that PCR on bulk vat samples may have reasonable diagnostic sensitivity for detecting these pathogens in herds with low prevalences of these infections. This is the first step in building a picture of test performance on bulk vat samples. Diagnostic sensitivity (the performance of the test in infected animals) and specificity (the performance of the test in noninfected animals) of the PCR for detecting Strep agalactiae in bulk vat test performance will also be estimated using PCR and BMCC data from bulk vats two herd data sets: 240 herds selected at random across the three dairying regions of Victoria and 219 herds selected because they had elevated BMCCs. The PCR test was used only once for each herd. The results of this analysis were not completed at the time of writing. 16 Countdown Symposium 2012

17 Molecular tests for milk Research to estimate prevalences of pathogens Utilising these same 240 and 219 herd bulk vat samples, prevalences of both Strep. agalactiae and Mycoplasma bovis will be estimated at the herd level. Apparent prevalences of Strep. agalactiae and Mycoplasma bovis in the 240 samples selected at random were 14% and 0.8%, respectively. Apparent prevalence of Strep. agalactiae in the elevated BMCC herds was higher than that seen in the randomly selected herd set as would be expected. Based on the literature review, the specificity of the Strep. agalactiae PCR is probably not 100%. So it is possible that some of these Strep. agalactiae PCR-positive results are false positives and further work will be conducted to estimate the true herd-level prevalence of Strep. agalactiae. The results for Mycoplasma bovis suggest that the diagnostic specificity of the Mycoplasma bovis PCR at vat level is high. Using pooled herd test-sampled milks Milk molecular tests can be performed using milk containing preservative as used when samples for ICCCs are taken during herd testing. The research group was interested in exploring the use of the PCR test with pooled milk samples from high ICCC cows using samples obtained as the milk samples from herd testing were being run through the cell counter. A testing protocol was derived which involved pooling two lots of ten high ICCC cows (> 400,000 cells/ml) along with one pool of five low ICCC cows from 20 herds. The herd test sample derived pools were tested with the PCR and milk samples were then collected aseptically from the same cows. The aseptic samples were then tested with both standard culture and PCR. Analyses of this large data set across 20 herds will be completed in August If it is shown that pooled herd test derived samples play a role in assessing high ICCC cows in a useful way, this will create an added test option for advisers and herd managers. Next steps Based on the research work undertaken by Countdown thus far, it appears there is a role for the milk molecular tests, and the PathoProof PCR more specifically. While the results of all the research components have yet to be finalised across all individual work areas, evidence is emerging that the test can be useful for the detection of Strep. agalactiae and Mycoplasma bovis at the individual cow, pooled cow sample and bulk vat level. This would be both as a surveillance screening test and as a test to be employed as a result of an investigation. In July and August, all research components will be finalised and the Countdown team will develop a decision-making flowchart for the use of PCR on milk samples of all types and under typical scenarios. This will be made available to advisers later in Acknowledgements The Countdown research team thanks the following colleagues for their assistance and support through this project: Dr Karensa Delany (Geoffrey Gardiner Dairy Foundation); Assoc. Prof John House and Assoc Prof Paul Sheehy (University of Sydney); Dean Baylis and Dr Graeme Richardson (DTS); Dr Pauline Brightling, Dr John Craven, Dr Anne Hope, Helen Pitman and Narelle Savige (Harris Park Group); Meaghan Johnston (Murray Goulburn); Lisa Archer (Fonterra). The team especially thanks the members of the Project Reference Group: Dr Margaret Deighton (RMIT), Dr Zoe Vogels (The Vet Group) and Craig Gallpen (Dairy farmer, NSW). Finally, we thank the farmers who have donated their time and supported this research through the provision of milk samples. References Anonymous (2012) Wisconsin Veterinary Diagnostic Laboratory. wvdl.wisc.edu/diagnosticaids.asp; viewed 19 th June Bexiga R, Koskinen MT, Holopainen J, Carneiro C, Pereira H, Ellis KA, Vilela CL (2011) Diagnosis of intramammary infection in samples yielding negative results or minor pathogens in conventional bacterial culturing. Journal of Dairy Research 78: Bradley A, Bartlett B, Biggs A, (2011). Mastitis testing : PCR or bacteriology, UK Farmers Weekly. Brightling PB, Mein GA, Hope AF, Malmo J, Ryan DP. Using ICCCs. In: Countdown Downunder Technotes for Mastitis Control, Dairy Research and Development Corporation, Melbourne, Australia, 2000; including Technote Update Pack, February Friendship C, Kelton D, van de Water D, Slavic D, Koskinen M (2010) Field evaluation of the PathoProof mastitis PCR assay for the detection of Staphylococcus aureus infected cows using DHI samples. NMC Annual Meeting Proceedings (2010), pp Hames D, Hooper N (2005) Biochemistry, BIOS Instant Notes, Taylor & Francis Group, New York, pp Katholm J, Bennedsgaard TW (2011) Survey of bulk tank milk from all Danish dairy herds in 2009 and 2010 with real-time PCR. NMC Annual Meeting Proceedings (2011). Koskinen MT, Wellenberg GJ, Sampimon OC, Holopainen J, Rothkamp A, Salmikivi L, van Haeringen WA, Lam TJGM, Pyörälä S (2010) Field comparison of real-time polymerase chain reaction and bacterial culture for identification of bovine mastitis bacteria. Journal of Dairy Science 93: Mahmmod Saadeldien Y, Klaas Christine I, Toft N, Katholm J (2012) Estimation the diagnostic performance of PathoProof Mastitis PCR and bacterial culturing for the detection of Streptococcus agalactiae in milk of Danish dairy cows using latent class models. Poster presented at the 27th Congress of the World Association for Buiatrics, Lisbon, Portugal; 4th to 8th June, Taponen S, Salmikivi L, Simojoki H, Koskinen MT, Pyörälä S (2009) Real-time polymerase chain reaction-based identification of bacteria in milk samples from bovine clinical mastitis with no growth in conventional culturing. Journal of Dairy Science 92: Wellenberg GJ, Sampimon OC, Rothkamp A, van Haeringen WA, Lam TJGM (2010) Detection of mastitis pathogens by real-time PCR in clinical and subclinical mastitis samples. Mastitis Research into Practice: Proceedings of the 5th IDF Mastitis Conference 2010, pp Countdown Symposium

18 A survey of mastitis pathogens in the south-eastern Australian dairy industry Neil Charman, 1 Rodney Dyson, 2 Andrew Hodge, 1 Natalie Robertson 1 and Sarah Chaplin 2 1. Pfizer Animal Health Australia, Parkville, Victoria, Australia 2. Dairy Focus, Kyabram, Victoria, Australia Introduction Mastitis represents a significant cost to the Australian dairy industry. Dairy Australia has estimated that more than $40 million is lost to Australian dairy farmers each year through poor udder health, and mastitis is the major cause of that loss. Pfizer Animal Health in conjunction with Dairy Focus conducted a survey of mastitis pathogens causing both clinical and subclinical mastitis across dairy farms in south-eastern Australia. Thirteen veterinary practices enrolled 65 farms to conduct this survey. Samples have been collected from 2,986 cases of clinical mastitis and 1,038 cases of subclinical mastitis. The survey commenced in February 2011 and concluded in March All samples were submitted for culture and sensitivity to Gribbles Veterinary Pathology in Clayton, Victoria. Materials and Methods Inclusion criteria for enrolled farms Farms were enrolled into this survey based on an assessment by the veterinary practice responsible for mastitis management on each property. The farms met the following enrolment criteria. Herd size: >250 cows in milking herd Mean Bulk Milk Cell Count (BMCC) in previous year of 100, ,000 somatic cells/ml Located within a reasonable commuting distance of the participating veterinary practice Ability to deliver milk samples to the participating veterinary practice within 48 hours of collection Undertook herd testing on a regular basis with a minimum of four herds tests per year and maintained an electronic herd health recording system Met acceptable animal welfare standards Definition of clinical mastitis For the purposes of this survey, and in accordance with the definition defined by Countdown Downunder, a case of clinical mastitis was defined as meeting the following conditions: One or more udder quarters inflamed, as evidenced by heat, redness and/or swelling; AND/OR Abnormal appearance (clots, blood, discolouration) in more than the first three strips of milk from one or more teats Study design A veterinarian from each veterinary practice, in consultation with the study investigator, selected up to six suitable dairy farms fitting the property inclusion criteria with effort made to ensure representative farms from each major production system were included in this survey. From the start of each calendar month, each dairy farm collected a milk sample from the first ten cases of clinical mastitis from cows that met the above Animal Inclusion Criteria (where available). In addition, on a single occasion on many of the enrolled farms, a further samples were collected from a selected group of cows that displayed a high somatic cell count in their milk in mid-to-late lactation. The individual cow was the experimental unit for this study. Approval to conduct this study was granted by the CSL/Pfizer Animal Ethics Committee at the meeting held on 11 October 2010 (approval number 854-0). Laboratory analysis Gribbles Veterinary Pathology (Clayton, Victoria, Australia) were responsible for all laboratory analyses. This included milk sample culture (including screening for Mycoplasma) and antibiotic sensitivity testing. All antibiotic sensitivity testing was conducted according to the Clinical and Laboratory Standards Institute Performance Standards for Antimicrobial Disk and Dilution Susceptibility Tests for Bacteria Isolated From Animals; Approved Standard Third Edition. CLSI document M31-A3. Wayne, P.A.: Clinical and Laboratory Standards Institute; Data summary and analysis A Biometrics Representative from Pfizer Animal Health was responsible for all data summaries and analyses. The key outcome variable for this survey was the overall number 18 Countdown Symposium 2012

19 A survey of mastitis pathogens Table 1: Summary of overall clinical mastitis culture results. CULTURE Total n % Arcanobacterium (Actinomyces) pyogenes Corynebacterium bovis Candida species/yeasts Escherichia coli Klebsiella species Listeria monocytogenes Nocardia species Pasteurella species Pseudomonas aeruginosa Salmonella species Serratia species Staphylococcus aureus Other Staphylococcus spp Streptococcus agalactiae Streptococcus dysgalactiae Streptococcus uberis Other* Contamination No growth Total *Other includes: Bacillus spp, Citrobacter spp, Gram negative bacillus, Hafnia alvei, Kluyvera intermedia, Kocuria kristinae, Pantoea sp, Prototheca sp, Raoultella ornitholytica and percentage of positive culture results for each mastitis pathogen cultured from both clinical and subclinical mastitis samples. Secondary outcome variables included the number and percentage of positive culture results for each pathogenic bacteria recorded for each phase of lactation, geographical region and parity (clinical mastitis samples only). The antibiotic sensitivity percentage for each bacterial species will be reported in a subsequent presentation after the data analysis has been completed. Results Clinical mastitis submissions The culture data has been considered in the following manner for this report. The positive culture of a single organism at a single time-point for an individual cow was identified as the experimental unit for analysis purposes. If multiple pathogens were identified at a single time-point for an individual cow then these were treated as multiple results. Specimens submitted from the same animal at different time-points were analysed as separate submissions. Primary outcome The key outcome variable for this study was the overall number and percentage of positive culture results for each mastitis pathogen. These are listed and summarised in Table 1. Overall, 39.3% of samples were either contaminated (16.1%) or produced no growth (23.2%). This will assist advisers to communicate meaningful interpretation of results to farms, and to be able to identify when collection techniques on farm need attention. An alternative way of viewing the data is to exclude the 39.3% of contaminated and no growth samples. When viewed in this light, 54.3% of samples cultured Streptococcus uberis, 14.8% cultured Staph aureus, 11.7% Escherichia coli and 8.9% cultured Streptococcus dysgalactiae. Secondary outcomes 1. Stage of lactation The recording of clinical mastitis was linked to the stage of lactation (Table 2). Stage of lactation has been summarised into the first month post calving, 2-6 months post calving, 6-10 months post calving and greater than 10 months (305 days) post calving. Clinical mastitis was recorded in the first month of lactation for 23.1% of all cases submitted in this survey. The percentage of clinical cases submitted during the first month of lactation was similar across all regions except Tasmania, where 53.5% of clinical cases submitted originated from cows during the first month after calving. Overall, 46.9% of clinical mastitis submissions were sampled from cows calved between two and six months. The Western District of Victoria submitted almost half (49.2%) of all clinical submissions during this 2-6 month stage of lactation. In general, clinical mastitis submission rates reduced from cows during the late stages of lactation (7-10 months post calving). However, 25.7% of all clinical mastitis submissions from the Northern district of Victoria were collected during this 7-10 month stage of lactation. The stage of lactation during which samples were collected appears to have had little influence on the relative frequencies which the major pathogens (pathogens with an overall incidence recorded > 5%) were isolated during this study (Figure 1). Streptococcus uberis remained the dominant pathogen at all stages of lactation and was isolated at a rate greater than 30% from all submissions during all phases of lactation. No clear trends of increasing or decreasing isolation rates during the course of lactation were evident Table 2: Stage of lactation overall submission summary by region. STAGE OF LACTATION Gippsland Northern Victoria Tasmania Western District Total n % N % n % n % n % First month months >6 months and <10 months ( 305 d) >10 months (305 days) Total Countdown Symposium

20 Charman et al. Pathogen Isolated % s. uberis S. aureus E. coli S. dysgalactiae Pathoegn Isolated % Serratia spp Candida spp/yeasts A. pyogenes C. bovis Nocardia spp Klebsiella spp Pasteurella spp 0 First month 2-6 months 7-10 months >10 months Stage of Lactation Figure 1: Major pathogens isolated during lactation. for Staphylococcus aureus, Escherichia coli or Streptococcus dysgalactiae. Minor pathogens (pathogens with an overall incidence recorded 0.5% but < 5%) cultured in this study included Serratia spp, Candida spp/yeast, Arcanobacterium pyogenes, Corynebacterium bovis, Nocardia spp, Klebsiella spp and Pasteurella spp. Corynebacterium bovis is the only minor pathogen that displays a trend of an increasing isolation rate as the length of lactation increases (Figure 2). Nocardia spp was consistently isolated during all stages of lactation at isolation rates varying between 1.1% and 1.7% of all individual culture results. 2. Geographical region The overall number and percentage of positive culture results for each mastitis pathogen by region are listed and summarised in Table 3. The major pathogens isolated in this survey are represented graphically by region in Figure 3. The Western District of Victoria First month 2-6 months 7-10 months >10 months Stage of Lactation Figure 2: Minor pathogens isolated during lactation. recorded the highest incidence of Streptococcus uberis at 36.8%, while the Gippsland region isolated Streptococcus uberis from 27.5% of cultures. Staphylococcus aureus was recorded at a rate of 12.7% of cultures submitted from Gippsland, while only 4.5% of cultures recorded this pathogen from Northern Victoria. Escherichia coli was the second most likely pathogen to be isolated from mastitis samples submitted from Northern Victoria at an isolation rate of 9.7%. The isolation of minor pathogens by region is represented graphically in Figure 4. Corynebacterium bovis was isolated at a rate of around 1.5% for the mainland states yet only isolated from a single submission in Tasmania. Nocardia spp was isolated on 22 occasions from samples submitted from the Western District of Victoria giving a 2% incidence for this pathogen. Klebsiella spp was isolated from around 1% of cultures when the samples were collected from Gippsland, Northern Victoria or Tasmania. Table 3: Mastitis culture results for each pathogen by region. Culture Gippsland Northern Victoria Tasmania Western District Total n % n % n % n % n % Arcanobacterium (Actinomyces) pyogenes Corynebacterium bovis Candida species/yeasts Escherichia coli Klebsiella species Listeria monocytogenes Nocardia species Pasteurella species Pseudomonas aeruginosa Salmonella species Serratia species Staphylococcus aureus Other staphs Streptococcus agalactiae Streptococcus dysgalactiae Streptococcus uberis Other Contamination No growth Total Countdown Symposium 2012

21 A survey of mastitis pathogens Pathogen Isolated % Gippsland Northern Victoria Tasmania Western Victoria S. uberis S. aureus E. coli S. dysgalactiae Pathogen Isolated % A. pyogenes C. bovis Candida spp/yeasts Klebsiella spp Nocardia spp Pasteurella spp Gippsland Northern Victoria Tasmania Western Victoria Serratia spp Major Pathogen Minor Pathogens Figure 3: Major pathogens isolated by region. Figure 4: Minor pathogens isolated by region. 3. Clinical mastitis by parity The overall number and percentage of submissions were related to parity of the affected cow and are summarised below in Figure 5. Clinical mastitis samples were most likely to be collected from cows in their third lactation (20.1% of all samples). More than half the clinical samples received for culture were collected from 14% 8% 17% 9% 15% Figure 5: Clinical mastitis submissions. 21% Parity 16% cows on their 2 nd 4 th lactation (53.3%) and 15% of all clinical samples were submitted from heifers during their first lactation. The frequency with which each major mastitis pathogen was isolated from cows of different parity was generally consistent for the four major pathogens (Figure 6). Streptococcus uberis was isolated most frequently from heifers in their first lactation (40.4%). Cows in their third lactation had the lowest isolation rate for Streptococcus uberis (27.5%), however this isolation rate was still far higher than any of the other major pathogens. Staphylococcus aureus was consistently isolated from cows of all parities within a range of % of culture results. Escherichia coli and Streptococcus dysgalactiae were consistently isolated from cows of all parities at a rate of % of cultures. Subclinical mastitis submissions The summary of subclinical mastitis culture results by region is represented in Table 4. The recording of a no growth result for the subclinical mastitis samples (17.4%) was broadly similar to the rate calculated for the clinical mastitis samples (23.2%). However, the frequency with which a contaminated sample Table 4: Summary of overall subclinical mastitis culture results. Culture Gippsland Northern Victoria Tasmania Western District Total n % n % n % n % n % Corynebacterium bovis Candida species/yeasts Escherichia coli Nocardia species Pseudomonas aeruginosa Serratia species Staphylococcus aureus Other staphs Streptococcus agalactiae Streptococcus dysgalactiae Streptococcus uberis Other* Contamination No growth Total * Other includes: Bacillus spp, Citrobacter spp, Gram negative bacillus, Hafnia alvei, Kluyvera intermedia, Kocuria kristinae, Pantoea sp, Prototheca sp, Raoultella ornitholytica Countdown Symposium

22 Charman et al. Pathogen isolated % S. uberis S. aureus E. coli S. dysgalactiae Parity Figure 6: Major pathogens isolated by parity. was submitted from cows with subclinical mastitis (40.7%) was far higher than that recorded for clinical mastitis submissions (16.1%). Staphylococcus aureus was the organism most frequently isolated from subclinical mastitis samples (17.5%, Figure 7), followed by Streptococcus uberis with 141 positive cultures recorded (13.1%). Escherichia coli was only cultured on 10 occasions (0.9%) from the samples submitted from high ICCC cows. Acknowledgements Pfizer Animal Health and Dairy Focus gratefully acknowledge the assistance given by Gribbles Veterinary Pathology and the participating veterinary practices, dairy farmers and herd improvement organisations involved in this survey Pathogen Isolated % S. aureus s. uberis S. dysgalactiae C. bovis E. coli Other Staphs Nocardia spp Candida/yeasts Other Pseudomonas Serratia S. agalactiae Figure 7: Subclinical mastitis pathogens isolated. 22 Countdown Symposium 2012

23 How to R.E.S.E.T. farmer mindset? Experiences from the Netherlands Jolanda Jansen, 1 Roeland Wessels 2 and Theo Lam 3 1. St Anna Advies; Wageningen UR Livestock Research, The Netherlands 2. St Anna Advies, The Netherlands 3. GD Animal Health Service, Faculty of Veterinary Medicine, Utrecht University, The Netherlands From a historical perspective, agricultural extension specialists, researchers and veterinarians assumed that agriculture was an activity executed by an individual farmer, based primarily on rational, technical and economic considerations. 1,2 Although such rational choices still play a role in farm management, we have learned that farmers decision-making about mastitis management based on these considerations is not always clear and understandable. 3 Why some farmers do not implement effective mastitis management practices, even though it would benefit their results, is not always known, 4 but it is often assumed that, besides these deliberate rational considerations, other farmer mindset factors play a role. 1,3-14 This paper describes the R.E.S.E.T model that can be used as a framework for changing a farmer s mindset and improving udder health. Farmer mindset Farmer mindset comprises a variety of social psychology constructs such as the farmer s personality, attitudes, beliefs, values, intentions, skills, knowledge, perceived norms and perceived self-efficacy. For example, see the Theory of Planned Behavior and the Health Belief Model, which are both frequently used to explain people s health behaviour All these factors, and probably more, make up the human factor which, for the sake of convenience, is termed mindset. Research at the Dutch Udder Health Centre (UGCN) in the Netherlands has shown that mastitis problems can be explained to a certain extent by farmer mindset and behaviour, and that mindset explains a substantial part in these models. 24,25 In this study, from 2004, elements of farmer mindset explain 17% of the variance in clinical mastitis incidence and 47% of the variance in bulk milk somatic cell count (BMSCC), while farmers self-reported behaviour explains, respectively, 12% and 14% of the variance of these parameters. The results of several UGCN studies show that two factors of farmer mindset seem to be important behavioural determinants for mastitis prevention: the perceived threat (i.e. Do I have a problem? ) and the perceived efficacy of preventive measures (i.e. Can I solve the problem easily? ). 26,27 Interestingly, these factors are also known to be indispensable in motivating people to work on their own health and are included in the so-called Health Belief Model, see Figure ,28,29 It is important for veterinarians and other herd health advisers to acknowledge these mindset factors and to make sure they have a complete understanding of farmers perceptions on benefits and barriers of preventive measures when advising them. Perceived susceptibility Belief in a personal threat Perceived severity Cue to action Preventive behavior Perceived benefits Belief in the effectiveness of a preventive behavior Perceived barriers Figure 1: The Health Belief Model, 18 adapted by Koelen and Van den Ban. 30 Countdown Symposium

24 Jansen, Wessels and Lam The R.E.S.E.T approach The causes of variation in mastitis incidence on herd and cow level are not yet fully understood. However, this does not restrain the dairy sector from implementing policies to reduce mastitis. For a mastitis control program, it is important to influence elements of farmer mindset in order to change farmers management practices to improve udder health. Farmers can be persuaded in many ways to change their behaviour regarding udder health ,31-33 It is most important to note that farmers are different. They have various learning styles and prefer customised communication. 32,33 A combination of actions and communication strategies are therefore needed to reach and change as many farmers as possible, because one size fits all approach does not apply for effective communication. To make life easier, we can discuss some effective intervention strategies by following the R.E.S.E.T model (see Figure 2), that was adapted from van Woerkum et al., 34 and Leeuwis 1 and was described earlier in a different form regarding mastitis by Lam et al. 33 The model shows five main instruments that need to be addressed when a change in behaviour of people is required: The R of Regulations, the E of Education, the S of Social pressure, the E of Economic incentives and the T of Tools. As some people are more influenced by negative stimuli and some more by positive stimuli or social pressure, it is the combination of all that makes a program or campaign effective. 26 Communication can be used as an instrument on its own via Education, but is actually the glue between all instruments together. Simply said, behaviour can be changed in either a voluntary or a compulsory manner. Compulsory behavioural change is facilitated by coercion such as regulations and restrictive provisions 34, 35 (Van Woerkum et al., 1999). It is well known that compulsory behavioural change will probably only last as long as the coercion exists. Therefore, voluntary behavioural change is preferable. Voluntary behavioural change can be reached by internal or external motivation. Internal motivation is the most difficult to influence via a disease control program, as it relates to age, generation, lifestyle, education and character. External motivation is better suited, but mostly underestimated. The R of Regulations Regulation by law forces people to behave in a preferred way: if you don t, you can end up with a fine, in jail, or even worse off in some countries. This mechanism works via coercion. It has nothing to do with voluntary behaviour. This may not apply to udder health directly, but legal limits for cell count, use of antibiotics and other drugs, and environmental regulations may influence farm management. These regulations need to be taken into account: a change in legislation has a direct BEHAVIOURAL CHANGE Compulsory Voluntary Internally motivated Externally motivated Coercion Circumstances Norms & Values Rewards & Fines Provisions & Means Regulation Education Social pressure Economic incentive Tools Figure 2: The R.E.S.E.T model: Behavioural change by a combination of strategies (adapted from Van Woerkum et al., 34 see also Leeuwis 1 ). 24 Countdown Symposium 2012

25 How to R.E.S.E.T. farmer mindset effect when there is proper surveillance. However, this measure can only be implemented by authorities. It should be noted that these measures can work in a counterproductive way. For example, when you want a farmer to renovate his barn, this is sometimes prohibited by local authorities or it can take years to actually get all the necessary permissions. This can be one of the barriers within farmer mindset as presented in Figure 1. The E of Education Education is one of the most-used intervention instruments. The effect of it is important, but sometimes overestimated. Education does not mean that you can send farmers a book with information about udder health, and then hope they will learn from it. 31 Education should start early, even within agricultural schools and vet schools. Education can be more effective if the information is offered and adapted to the different learning styles of farmers. 33 Using study groups as a method for education is quite effective for a specific group of farmers with a specific learning style. About 13% of farmers will participate in study groups if actively offered them. 31, 33, 36 This means that a large part of the farming community are not reached. Different education and communication strategies are therefore needed as study groups focus more on rational behavioural change. Peripheral communication strategies may educate people without them even being aware of it, as was shown by the milking gloves campaign of the UGCN. Milking glove use increased from 21% to 42% within one year of a short peripheral campaign focused on their use. Moreover, the opinion and knowledge of farmers about milking gloves changed in their favour, without actually addressing the benefits of gloves in the campaign itself. 31 The S of Social pressure Social pressure influences farmers norms and values, and can also have a long-term effect on internal motivation. However, this will mostly influence the circumstances on a farm and within a family, which makes a farmer change his behaviour. Humans need social cohesion to be successful; it is one of the most powerful tools that can be used in intervention strategies. The success of study groups, where farmers influence each other, is mostly based on this principle. Social pressure influences people s frame of reference. What is normal for you on your farm? If an important and highly respected person within a farmer s social network has a different frame of reference, a farmer may want to copy that in order to comply with the norms. Everybody wants to be unique, but nobody wants to be too different. Veterinarians and other herd health advisers are important in increasing social pressure and setting a frame of reference about what is normal and what is not. The more people within a farmer s social network who put such a pressure on the farmer, the harder it is not to comply. That s why it is important that all farm advisers, and even all vets within the same practice, work together and send the same message. They also need to take into account others who may influence the farmer, such as family members, peers or staff. If the social pressure is high enough, the other tools of the R.E.S.E.T model may have no or less effect. The E of Economic incentives External motivation can be accomplished by financial stimuli such as bonuses and penalties related to bulk milk somatic cell count (BMSCC). 37,38 Currently in the Netherlands a penalty is imposed for a geometric mean BMSCC above 400,000 cells/ ml. This policy is effective in reducing the number of herds with a BMSCC above this threshold level. Penalties are always extensively discussed, as it is known that they have an effect as soon as their levels are adjusted. However, this does not solve all udder health problems, because serious clinical mastitis problems may occur in herds with a low BMSCC. 39,40 In addition, the milk of individual cows with clinical or subclinical mastitis can be withhold from the bulk and thus are not represented in the BMSCC. These penalties will change people s behaviour, not just through some sort of coercion, but mostly because they set a social norm. They tell you when you are not doing well. Unfortunately, these norms sometimes have the opposite effect. As long as a farmer does not reach this norm, he may think he is okay. Therefore, it should be better if at the same time another statement is being made by rewarding farmers that do well. The premiums paid by many Australian milk processors for milk <250,000 BMSCC is a good example of external motivation in a positive way. A good quality milk can be rewarded economically, but rewards also apply through the farmer s sense of pride and status (social pressure) for milestones such as best quality milk, best reduction in cell count, best udder health, etc. Incentives can work in a counterproductive fashion. For example, some practices and veterinary pharmaceutical companies in the Netherlands give a quantity discount on dry cow therapy with antibiotics: the more tubes you buy, the cheaper. But one should keep in mind the symbolism of such a communication when you want farmers to reduce the amount of antibiotics they use, which is currently a hot issue in the Netherlands. Finally, economic incentives can work to help implement control measures by showing farmers the economic benefits of the measures or by making certain control options much cheaper, such as bacteriology on milk samples. However, it should be taken into account that most farmers do not behave in an economically rational way as was studied by Huijps et al. 38, and Asseldonk et al. 44 The T of Tools Tools, such as technical provisions, means and methods, can stimulate farmers to perform a certain way. They can make the desired behaviour much easier to perform. For example, consider the possibility for easy milk sampling or the fact that the design of the milking parlour is optimal to treat mastitis Countdown Symposium

26 Jansen, Wessels and Lam cows as soon as you notice them. Tools can also be software to analyse the udder health problems. Tools only work if they are used properly and in combination with the other intervention instruments. Educating farmers that they need to take milk samples is of no use if the nearest vet who can analyse such samples is difficult to reach or remote from the farm. Tools can also help farmers to behave well unconsciously. For example, gloves can be attached to treatment tubes to make sure they are reminded to wear them when they apply dry cow therapy. Scientists are more and more aware of the effect of automatic, unconscious behaviour in daily life. With our growing capacity to analyse people s brains, we get a better picture of what happens within the unconscious brain. The previously mentioned peripheral campaigns can subconsciously influence people. Rational approaches via study groups may not be enough to make farmers use some tools. 31 Concluding remarks Elements of farmer mindset are important determining factors in mastitis control, including the perceived threat (i.e. Do I have a problem? ) and the perceived efficacy of preventive measures (i.e. Can I solve the problem easily? ). These issues can be addressed in communication strategies using the R.E.S.E.T model as framework and can be used even as a guide to evaluate the communication strategies applied by veterinary practices and practitioners themselves. To be effective, a disease program should do more than simply distribute technical information about best management practices to dairy farmers. Prevention of complex diseases, such as mastitis, requires customised communication strategies as well as an integrated approach between various stakeholders and different scientific disciplines. Such programs need to be supported by a combination of several policy measures to change farm management on the long term, because farmers are part of, and are influenced by, a wide institutional context. References 1. Leeuwis C. Communication for Rural Innovation. Rethinking Agricultural Extension. Third edition. Third edition edn. Blackwell Science Ltd, Oxford, Burton RJF. Reconceptualising the behavioural approach in agricultural studies: a socio-psychological perspective. Journal of Rural Studies 2004;20: Vaarst M, Paarup-Laursen B, Houe H, Fossing C, Andersen HJ. Farmers choice of medical treatment of mastitis in Danish dairy herds based on qualitative research interviews. Journal of Dairy Science 2002;85: Barkema HW, Van der Ploeg JD, Schukken YH, et al. Management style and its association with bulk milk somatic cell count and incidence rate of clinical mastitis. Journal of Dairy Science 1999;82: Seabrook MF. The Psychological Interaction between the Stockman and His Animals and Its Influence on Performance of Pigs and Dairy-Cows. Veterinary Record 1984;115: Van der Ploeg JD. Bedrijfsstijlen als socio-technische netwerken. De virtuele boer. first edn. Van Gorcum & Cromp B.V., Assen, 1999: Beaudeau F, Van der Ploeg JD, Boileau B, Seegers H, Noordhuizen JPTM. Relationships between culling criteria in dairy herds and farmers management styles. Preventive Veterinary Medicine 1996;25: Andersen HJ, Enevoldsen C. Towards a Better Understanding of the Farmer s Management Practices the Power of Combining Qualitative and Quantitative Data. In: Andersen HJ, editor. Radgivning, Bev aegelse mellem data og dialog. Mejeriforeningen, Arhus, 2004: Reneau JK. Milk Quality Mind Set. Oregon, Ohio, April 30- May Tarabla H, Dodd K. Associations between farmers personal characteristics, management practices and farm performance. The British Veterinary Journal 1990;146: Barnouin J, Chassagne M, Bazin S, Boichard D. Management practices from questionnaire surveys in herds with very low somatic cell score through a national mastitis program in France. Journal of Dairy Science 2004;87: Wenz JR, Jensen SM, Lombard JE, Wagner BA, Dinsmore RP. Herd management practices and their association with bulk milk somatic cell count on United States dairy operations. Journal of Dairy Science 2007;90: Dohoo IR, Martin SW, Meek AH. Disease, production and culling in Holstein-Friesian cows VI. Effects of management on disease rates. Preventive Veterinary Medicine 1984;3: Nyman AK, Ekman T, Emanuelson U, et al. Risk factors associated with the incidence of veterinary-treated clinical mastitis in Swedish dairy herds with a high milk yield and a low prevalence of subclinical mastitis. Preventive Veterinary Medicine 2007;78: Ajzen I, Madden TJ. Prediction of goal-directed behavior: attitudes, intentions and perceived behavioral control. Journal of Experimental Social Psychology 1986;22: Ajzen I. The Theory of Planned Behavior. Organizational Behavior and Human Decision Processes 1991;50: Fishbein M, Yzer MC. Using Theory to Design Effective Health Behavior Interventions. Communication Theory 2003;13: Janz N, Becker MH. The health belief model: A decade later. Health Education Quarterly 1984;11: Sun X, Guo Y, Wang S, Sun J. Predicting Iron-Fortified Soy Sauce Consumption Intention: Application of the Theory of Planned Behavior and Health Belief Model. Journal of Nutrition Education and Behavior 2006;38: Garcia K, Mann T. From I Wish to I Will : social-cognitive predictors of behavioral intentions. J Health Psychol 2003;8: Armitage CJ, Conner M. Efficacy of the Theory of Planned Behaviour: a meta-analytic review. British Journal of Social Psychology 2001;40: Painter JE, Borba CP, Hynes M, Mays D, Glanz K. The use of theory in health behavior research from 2000 to 2005: a systematic review. Ann Behav Med 2008;35: Noar SM, Chabot M, Zimmerman RS. Applying health behavior theory to multiple behavior change: Considerations and approaches. Preventive Medicine 2008;46: Jansen J, Van den Borne BHP, Renes RJ, et al. Explaining mastitis incidence in Dutch dairy farming: the influence of farmers attitudes and behaviour. Preventive Veterinary Medicine 2009;92: Jansen J, Van Schaik G, Renes RJ, Lam TJGM. The effect of a national mastitis control program on the attitudes, knowledge and behavior of farmers in the Netherlands. Journal of Dairy Science 2010;93: Jansen J, Lam TJGM. The role of communication in improving udder health. Veterinary Clinics of North America Food Animal Practice 2012;28: Jansen J. Mastitis and farmer mindset. Towards effective communication strategies to improve udder health management on Dutch dairy farms. Communication and Innovation Studies. Wageningen University, Wageningen, Rogers RW. Cognitive and physiological processes in fear appeals and attitude change: a revised theory of protection motivation. In: Cacioppo JT, Petty RE, editors. Social psychology: a source book. Guilford Press, New York, 1983: Griffin RJ, Dunwoody S, Neuwirth K. Proposed model of the relationship of risk information seeking and processing to the development of preventive behaviors. Environmental Research Section A 1999;80:S230-S Koelen MA, Van den Ban AW. Health education and health promotion. Wageningen Academic Publishers, Wageningen, Jansen J, Renes RJ, Lam TJGM. Evaluation of two communication strategies to improve udder health management. Journal of Dairy Science 2010;93: Countdown Symposium 2012

27 How to R.E.S.E.T. farmer mindset 32. Jansen J, Steuten CDM, Renes RJ, Aarts N, Lam TJGM. Debunking the myth of the hard-to-reach farmer: effective communication on udder health. Journal of Dairy Science 2010;93: Lam TJGM, Jansen J, Van den Borne BHP, Renes RJ, Hogeveen H. What veterinarians need to know about communication to optimise their role as advisor on udder health in dairy herds. New Zealand Veterinary Journal 2011;59: Van Woerkum C, Kuiper D, Bos E. Communicatie en innovatie. Een inleiding. Samsom, Alphen aan de Rijn, Van Woerkum C, Van Meegeren Pe. Basisboek communicatie en verandering. Uitgeverij Boom, Amsterdam, Meesters AJM, Jansen J, Van Veersen J, Lam TJGM. Study groups for udder health improvement led by practitioners experiences from the Netherlands Valeeva NI, Lam TJGM, Hogeveen H. Motivation of Dairy Farmers to Improve Mastitis Management. Journal of Dairy Science 2007;90: Huijps K, Hogeveen H, Antonides G, et al. Economic behavior of dairy farmers regarding mastitis management. In: Hillerton JE, editor. Mastitis research into practice: Proceedings of the 5th IDF mastitis conference, Christchurch New Zealand Vetlearn, Wellington, New Zealand, 2010: Barkema HW, Schukken YH, Lam TJGM, et al. Management practices associated with low, medium and high somatic cell counts in bulk milk. Journal of Dairy Science 1998;81: Barkema HW, Schukken YH, Lam TJGM, et al. Incidence of clinical mastitis in dairy herds grouped in three categories by mulk milk somatic cell counts. Journal of Dairy Science 1998;81: Huijps K, Lam TJGM, Hogeveen H. Costs of mastitis: facts and perception. J Dairy Res 2008;75: Huijps K, Hogeveen H, Lam TJ, Huirne RB. Preferences of cost factors for mastitis management among Dutch dairy farmers using adaptive conjoint analysis. Prev Vet Med 2009;92: Huijps K, Hogeveen H, Lam TJGM, Lansink AGJMO. Costs and efficacy of management measures to improve udder health on Dutch dairy farms. Journal of Dairy Science 2010;93: van Asseldonk MAPM, Renes RJ, Lam TJGM, Hogeveen H. Awareness and perceived value of economic information in controlling somatic cell count. Veterinary Record 2010;166: Countdown Symposium

28 The Dairy Focus approach Technote 13 and beyond Rod Dyson Dairy Focus, 1012 Henderson Road, Tongala, Victoria 3620, Australia Evaluation and research into the effectiveness and ongoing impact of the Countdown Downunder Farmer Short Course in achieving on-farm change was largely informed by the Insights report (Insight to the dairy industry s capacity to manage mastitis Nettle et al, 2005) commissioned and conducted by Countdown Downunder in association with the Rural Innovation & Change Centre (RICC) at the University of Melbourne (now known as the Rural Innovation and Research Group). One of the key findings of that research was that farmers who developed a mastitis management plan as part of completing the Farmer Short Course had not continued to evolve, develop and adapt that plan to changing circumstances and neither had the advisory services working with them. The research team identified core changes in advisory services that would significantly build the capacity of Australian dairy farms to develop and implement action plans on farms, including a specific need to change the culture of service provision by veterinarians towards a more strategic herd-level management basis. Furthermore, the research showed that dairy farming clients were looking for this type of service, and would pay for such a service that would meet their needs. This finding was consistent with other independent industry reviews. As part of a Review of Australia s Rural Veterinary Services in 2002, Peter Frawley stated: veterinary practices need to develop and promote the services they can offer to improve productivity in animal production recognising that the range of services will have to be more innovative. To assess the opportunity that might be offered to the dairy industry by these types of services, Dairy Australia commissioned the Achieving Sustainable Improvement (ASI) project, also as a collaborative project between Countdown Downunder and RICC. The ASI project further defined the opportunity for a pro-active planning & review approach, and developed a prototype service model around a process, now called the MAX process, by which this could be achieved. The MAX process operates in a repeatable cycle of assessment, planning, review and replanning. The process proved to be highly successful in helping farmers to achieve their goals in terms of both risk management and progressive improvement, but is highly dependent on the advisory population being able to take up the opportunity and successfully integrate it into their business structure, and this has proved harder to achieve. The Dairy Focus approach is built on the belief that a significant proportion of today s dairy farming community is either seeking or amenable to such a pro-active risk management program, and the business has invested heavily in the development of a systematic risk management approach to mastitis control. That said, the most common entry point for a farm into the Dairy Focus ongoing mastitis risk management program (Dairy Focus Mastitis Control System) is via the desire to resolve an existing mastitis problem. It is here that the Countdown Downunder approach and tools defined in Technote 13 still provide a sound and solid methodology to work through a mastitis problem. However, it is during this process that Dairy Focus introduces the farm to the concept of thinking about mastitis in terms of risk and risk management. Dairy Focus has developed a suite of mastitis risk assessment tools and measures to assess mastitis risk on farms, using the parameters and measures (triggers) defined in the Countdown Downunder Farm Guidelines & Technotes, but adding the concept of a mastitis risk score to many of these parameters. Most of the Dairy Focus clients who successfully resolve mastitis issues, subsequently graduate into the Dairy Focus Mastitis Control System ongoing risk management program they firmly believe that the risk management approach will prevent them going back to the dark place they were in! This has resulted in a client demographic where, currently, about half the Dairy Focus clients have a mastitis problem they want solved, and the other half don t want a mastitis problem! Significantly, a recent trend has been towards enquiries from 28 Countdown Symposium 2012

29 The Dairy Focus approach farmers who say I don t have a mastitis problem, but the reason for their call is because I don t want a problem! It is now clear that substantial further development over and above the concepts of the original Countdown MAX platform has been necessary to achieve the current workable system. Some key requirements have been as follows: A fundamental requirement has been the need to couch the thinking and communication by the adviser in terms of risk and risk management. Without this, the adviser will remain a problem solver, and Technote 13 will continue to be the bible for this approach. A customer relationship database is essential to program and co-ordinate communication and activities. Dairy Focus has invested in highly customisable software that allows the use and integration of dairy specific fields A different framework in which to place the discussion about risk was necessary adaption of the traditional Countdown stage of lactation framework was necessary to cut through blockage points and traditionally held views. Dairy Focus uses a wheel of five key mastitis risk areas into which it incorporates the traditional Countdown guidelines this has proved highly successful. A planning tool was developed to allow the adviser and farmer to easily develop the farm specific program for the coming 12 months. Ownership of the program is generated from the fact that the farm gets to choose its own program it is the farm s program, not the adviser s. This tool also needed to work in a manner that ensures the program that is developed will deliver the farm s goals. A diary system (Dairy Focus Mastitis Diary) was developed in which the farms mastitis control events and mastitis key performance indicators (KPIs) are constantly monitored. Software was developed to download and monitor each farms mastitis KPIs, and alerts were developed to ensure key mastitis control events occurred when and how they are supposed to, and that any corrective action was timely and appropriate. On-farm staff training and update packages have been developed to complement existing industry packages, and are incorporated into a farm s yearly planner. This systematic approach has been highly successful in assisting farms to meet their farm goals, and the dropout rate once farms have enrolled in the ongoing risk management program has been extremely low. There have also been a number of key learnings from this Dairy Focus approach to risk management in mastitis control: The concept is not for everyone in both the advisory and farmer populations! But those who embrace the concept are enthusiastic in their involvement. This is a very different business model to traditional businesses such as veterinary practices. To be successful, the business needs to truly embrace the concept. It has become clear that the activities and communication have to be driven by the adviser. Many of our clients openly say that this is one of the things they like most about our system, because they are just too busy to remember otherwise! The sense of a complete package has been important covering all aspects of mastitis control, and incorporating both staff and management training. There has been a clear benefit from the use of objective measures of mastitis risk wherever possible subjective measures are often viewed as suspect! For a number of years, Technote 13 has provided the framework and foundation to advisers on which to base a mastitis investigation. At Dairy Focus, we believe there is a market for a pro-active risk management approach to mastitis control that takes the adviser and the farmer beyond problem solving and Technote 13, into the world of consultancy and ongoing risk management. We believe that this approach offers substantial additional opportunities for farmers in mastitis control, and also in terms of career paths for advisers and advisory businesses who can embrace the concept. References Brightling P, Hope AF, Thompson A, Dyson R. Countdown Downunder Building industry capacity to control mastitis and manage milk quality, Dairy Australia, November Dyson R. Countdown MAX making plans work, Australian Cattle Vets, 2007 Nettle RA, Hope AF, Thompson A, Smolenaars F, Brightling P. Insight to the dairy industry s capacity to manage mastitis, Dairy Australia, September Penry JF, Paine M, Brightling PB. (2009) Achieving sustainable improvement. Final report to Dairy Australia. Dairy Australia, Melbourne, Australia. Countdown Symposium

30 Experiences with mastitis control over the past 50 years Jakob Malmo Maffra Veterinary Centre, Maffra, Victoria 3860, Australia I have been asked to present my experiences with mastitis control over the past 50 years an interesting, and at times exciting, journey. My early interest in mastitis control was stimulated by publication of the results of a co-ordinated approach to controlling bovine mastitis developed in the UK. This program began in the 1960s with the development of the Five Point Plan from research at the National Institute for Research in Dairying (NIRD) at Reading. As the name suggests, the Plan comprised five key points: the identification and treatment of clinical cases, routine antibiotic dry cow therapy for every cow at drying-off, post-milking teat disinfection, culling of persistently infected cows and regular maintenance of the milking machine. Uptake of the plan in the UK resulted in rapid progress in control of clinical and sub-clinical mastitis. This came about primarily through better control of contagious mastitis pathogens (i.e. those adapted for survival within and around the host mammary gland in particular, Staph aureus and Strep agalactiae). The incidence rate of clinical mastitis fell from more than 150 cases per 100 cows per year to just 40 cases per 100 cows per year between 1967 and At the same time, the national bulk milk somatic cell count reduced from more than 600,000 cells/ml to 400,000 cells/ml. My interest and involvement in mastitis control was further stimulated when Professor Doug Blood, Dean of the newly re-opened University of Melbourne Veterinary School, visited Maffra in the mid- 60s to speak to a farmer meeting about mastitis control programs, in particular the five-point program. A number of farmers who attended this meeting were looking for support to introduce this program, or a version of it, onto their farms. The manager of the Maffra Herd Improvement Organisation (HIO) and I got together and developed a program whereby technicians from the HIO would visit farms to collect aseptic milk samples and subject these samples to a Rapid Mastitis Test (RMT). Those samples that were positive to the RMT were cultured onto sheep blood agar plates. Individual quarters were cultured and each plate was divided into four quadrants allowing four samples to be tested per plate. A quarter was classified as infected if it had a positive RMT reaction (an indication of inflammation) and a major mastitis pathogen was obtained on culture. Staph. aureus was identified based on the morphology of the colonies on the culture plates and on the production of either an α or β hemolysin as indicated by the haemolysis on the sheep blood agar plates. Strep. agalactiae was confirmed using the CAMP test. Cows classified as infected in one or more quarter were selected for dry cow treatment, but as discussed below we were initially hampered by the lack of a registered dry cow product. When we began this project, no dry cow product was available in Australia in the first year of the project we used Orbenin lactating cow (provided by Beechams for the purpose of the project) as our dry cow treatment as this was the most suitable preparation that we could come up with. The second year came around and there was still no suitable dry cow product available so we decided to have one manufactured for use in these herds based on advice from several veterinarians working in the field. We decided on a product containing oxytetracycline in a long-acting aluminium monostearate base. The name given to this product was somewhat prophetic FiniLac. Veterinarians in other areas heard about the product and decided to use it on some of their farms. From their point of view, this was unfortunate, but very fortunate from our point of view. The cows in these other areas calved much earlier than cows in the Macalister Irrigation District (they started to calve before we started to dry off). Fortunately, only a relatively small number of cows had been treated with this dry cow product when it was found that when they calved many of them had teats that were rock hard and from which milk could never be expressed. The most likely explanation was that the oxytetracycline was in sufficiently high concentration to cause severe irritation to the teat and this caused a severe reaction which precluded further milking so, in retrospect, the name Finilac was appropriate. We heard of these reactions before our farmers were to start using this product so naturally was not used in our herds. This episode was referred to as the Finilac fiasco. We maintained the program for several years and were able to get a good indication of the pathogens present in dairy herds in the Macalister Irrigation District. It also helped to further develop enthusiasm for the five-point plan among our farmers. Throughout this time we received support from Prof Blood and Dr Roger Morris at the Veterinary school. We tried to further refine the mastitis five-point program by adding modifications such as back flashing of the cluster between milking cows long milk tubes were separated from the cluster and a hose attached to this so that around three-quarters of a litre of cold water was flushed through the teatcups between cows. This approach was fine in small herds being milked in herringbone sheds, but very quickly became too labour intensive to persist with. A few farms used a T junction device that fitted in the 30 Countdown Symposium 2012

31 50 years of mastitis control long milk tube and contained a valve to allow the water hose to be attached. This enabled back flashing without the need to disconnect the long milk tube from the cluster. While preparing for this presentation, I found in my library a booklet published by the Wonthaggi Rotary Club in This booklet detailed a farmers conference held at Archie s Creek Hall over two days in It involved speakers including Graham Mein, Lloyd Fell, Bill Thompson (mastitis expert with the Victorian Department of Agriculture) and a number of practitioners, including myself. By the time of this meeting the five-point program had become a little more detailed and was published as follows: 1. a. Annual testing of milking machines b. Attention to milking technique 2. a. Teats and udder washed in running water b. Back flashing of cups between cows (if practicable) 3. Post-milking teat dip 4. Treatment a. Clinical cases treated as they occur with a quick release antibiotic b. Dry cow treatment ALL quarters of ALL cows treated at drying off with a slow release antibiotic 5. a. Identification of infected animals b. Segregation in the milking line (if practicable) c. Culling of chronic cases that do not respond to treatment Briefly looking through this booklet, I believe that much of the information available then has not changed dramatically in the intervening 40 years. Over time, the five-point program was further modified and include developments such as the 5 x 5 program I cannot remember the five sub-points for each of the five-point program. Basically such programs were modifications and complications on the original five-point program, which was aimed at reducing the spread of infection and reducing the number of infected cows in the herd. The next major change in mastitis control was the introduction of individual cow cell counting as a means of assessing infection status of all cows in the herd, and of monitoring infection change over time. In the mid- 70s, Dr Bruce Anderson, a practitioner in Kyabram, was working with the local HIO to offer an individual cow cell counting service. This service was performed on milk samples submitted for herd testing and carried out by an automated cell counting machine. Bruce was of the view that this was a useful service, but it is fair to say that there were many doubters who questioned how the information could be interpreted and used by the farmer. In an attempt to find information on how these individual cow cell counts might be used, the Victorian Mastitis Research Group undertook a major study in 12 commercial herds in the Maffra and Kyabram district to investigate mastitis and cell count status of individual cows (Victorian Mastitis Research Group 1992). A composite (milk meter) sample from each cow was individual cow cell counted at least three times during lactation as part of the normal milk recording services for each farm. Before drying off, each quarter of each cow was sampled aseptically for bacterial culture on three successive days. Subclinical mastitis was defined as present in the quarter when a major pathogen (Staph aureus, Strep. species, E. coli or Pseudomonas sp.) was isolated from at least two of the three milk samples. This definition was based on the paper of Neaves (1975) in which it was stated: the evidence provided indicates that an examination for bacteria only in two successive quarter samples (with confirmation when the two disagree) is likely to give an assessment which is less than 1% false positive and false negative quarters when compared with an assessment based on the combined bacteriological and indirect tests of 6 or more milk samples. Complete bacteriological results were obtained from 1,101 cows, of which 333 were found to be infected. This study indicated a peak ICCC of 250,000 cells / ml was an appropriate threshold to detect cows infected with subclinical mastitis. The false negative and false positive rates for the individual cow cell count threshold of 250,000 cells/ml as applied to the population of 1,101 cows was 24.3% (false negative) and 25.9% (false positive). (Table 1) The study showed that the lactational average ICCC for cows sub-clinically infected with major mastitis pathogens was 3.8 times higher than uninfected cows or cows infected with minor pathogens. This was a very labour-intensive study that involved staff from the local Department of Agriculture, practitioners and laboratory workers; it would there be a very difficult study to implement now. Individual cow cell counting was then introduced by virtually all herd test organisations and since that time has formed one of the very important planks of mastitis control on many dairy farms. Many dairy farmers tell me that the main reason they now herd test is to receive individual cow cell counts that can help in their mastitis control. I think that the single most important factor driving farmers to improve their cell count was the dairy companies introduction of a premium payment system. Under this system, milk provided below a factory-specific cell count threshold would Table 1: Accuracy of mastitis diagnosis at drying-off using various ICCC thresholds (Victorian Mastitis Research Group 1992). Peak ICCC threshold ( 000 cells/ml) Test sensitivity Cows with mastitis* above the ICCC threshold (%) Test specificity Uninfected cows* below the ICCC threshold (%) * The study included 768 uninfected cows and 333 cows with mastitis. Countdown Symposium

32 Malmo Table 2: The milk quality payments applicable to a 700-cow seasonal supply herd (producing 4.1 million litres of milk) over a range of bulk milk cell counts. Milk quality band Cell count range for band Premium 1 <250,000 cells/ml $96,419 Premium 2 250, ,000 cells/ml $40,135 Base 350, ,000 cells/ml 0 Productivity payment if all milk for the year is in this range receive a premium payment and milk of very poor quality would receive a substantial discount. We had talked mastitis control programs for many years, but the implementation of the premium payment schemes meant that farmers could see an immediate financial penalty if their bulk milk cell count rose above a certain threshold level (in many factories, a 10 day average of greater than 250,000 cells/ml). Premium payments for milk that is below the factory specific cell count threshold is still a significant driver for farmers to keep their cell counts down. To estimate the present-day value of these milk quality premium incentives, I obtained a milk income estimate for for a 700-cow seasonal supply herd (producing 4.1 million litres of milk) and supplying the Murray Goulburn Cooperative. Table 2 shows the milk quality payments applicable to this farm over a range of bulk milk cell counts (note that other factors such as the thermoduric and bactoscan bacteriological tests can also move milk to a lower milk quality band). For this farm, supplying milk that is always in the premium 1 band results in an additional payment of more than $56,000 for the year when compared with supplying milk that is in the premium 2 band. For a perhaps more average size farm milking 350 cows, this still represents an additional income of $28,000. Despite all of the above activity, cell counts on many farms were still at an unacceptably high level. In response to this, the Countdown Downunder program was developed. This program set about consolidating the available knowledge on mastitis control into the Countdown Downunder Technotes and Countdown Downunder Farm Guidelines for Mastitis Control and developing an extension strategy that ensured that clear, consistent messages would be delivered to dairy farmers Australia wide on mastitis control issues. The goals of Countdown Downunder (Countdown) were initially set at achieving farm annual average cell counts (BMCC) below 250,000 cells/ml for 90% of milk supply and below 400,000 cells/ml for the remainder. Countdown was launched in 1998 to improve mastitis control on Australian dairy farms and to keep the count down. Since the program s launch, a large number of Australian dairy farmers, and farm workers, have been through Countdown Downunder training programs. The Countdown information is continually being updated as the need arises as evidenced last year by the release of the publication on control of Strep uberis mastitis. Current research Finally, and to bring us up to the present time, I would like to discuss two major research projects which have been funded by the Gardiner Foundation. Money for major mastitis research projects has been limited for a long time and we are certainly indebted to the Gardiner Foundation for giving us the opportunity to investigate two major areas of concern in the area of mastitis. Over the past 18 months, molecular testing in the form of a real-time PCR test has become available to dairy farmers in Australia. While the available evidence is that these tests are both specific and sensitive, many of us are unsure as to how we should interpret the results of these tests. With this in mind, a project to investigate the possible uses of molecular mastitis testing for dairy farmers and milk processors is under way. Our research is examining the role of the new molecular tests for mastitis pathogens in bulk milk samples, in pooled samples taken from high cell count cows at the time of herd testing, as well as in testing individual cow milk samples. The collection of the necessary milk samples is nearly complete, and the project will then move into the data analysis stage. The second major project deals with reducing Strep. uberis mastitis risk by improving pre-milking hygiene. Our review last year of Strep. uberis and our work during the wet conditions clearly identified the need to know more about pre-milking teat preparation. The research project involves a trial involving five dairy farms, and will examine whether improved pre-milking hygiene reduces new mastitis infections. The pre-milking hygiene will involve initially washing the udders of treatment cows, applying a pre-milking teat disinfectant to these cows and then drying them before teat cup application. The milking clusters will be disinfected prior to them being placed on these treatment cows. The project will run over the first 60 days of the calving period a period of high risk of clinical mastitis. The aim is to determine whether or not this pre-milking teat disinfection brings about a significant reduction in the number of cases of clinical mastitis over the treatment period. The cost benefit of this procedure will then need to be examined. In summary, my recollections of my 50-year trip down mastitis memory lane has shown me how far we have come, but also shows me that we still have a fair way to go. It seems to me that our mastitis-causing organisms are formidable opponents and that we need to remain vigilant for changes as well as maintain a reasonable research platform if we are to get on top of this opponent. References Machine milking and mastitis. Wonthaggi Rotary Club Farmers Conference held at Archie s Creek Hall Neave, F.K. Diagnosis of mastitis by bacteriological methods alone in Proceedings of the IDF Seminar on Mastitis Control 1975, page 34. Victorian Mastitis Research Group Individual cow cell counts, milk production and selective dry cow therapy. Mastitis and Milk Quality Workshop, Australian Association of Cattle Veterinarians pp Countdown Symposium 2012

33 CellCheck a new solution to an old challenge F. McCoy, 1 C. Devitt, 2 S. More, 3 K. Heanue 4 and K. McKenzie 5 1. Animal Health Ireland, Carrick-on-Shannon, Co. Leitrim, Ireland 2. Sandymount, Dublin, Ireland 3. Centre for Veterinary Epidemiology and Risk Analysis, University College Dublin, Ireland 4. Rural Economy Research Centre, Teagasc Mellows Campus, Athenry, Co. Galway, Ireland 5. Motiveworks, Dublin, Ireland In Ireland, regulatory animal health issues (such as TB, Brucellosis and FMD) have traditionally been the responsibility of the government, with little or no coordination of nonregulatory animal diseases. However, the establishment of Animal Health Ireland (AHI) in 2009, is helping to address this gap. AHI is a not-for-profit, partnership-based organisation providing national leadership and coordination of non-regulatory animal health issues. Its mission is to improve the profitability, sustainability and competitiveness of Irish livestock producers and related industries, through superior animal health. AHI is a partnership between producers, processors, service providers and government, and is co-funded by these stakeholders. It provides benefits to livestock producers and processors by providing the knowledge, education and coordination required to establish effective control strategies, both on-farm and nationally. One of the first AHI activities, in 2010, was to identify priority disease areas through a policy Delphi study with experts and farmer surveys. Seven diseases were prioritised, and a Technical Working Group (TWG) was established in each area. These TWGs are made up of industry experts in that field, from all disciplines. Their remit is to bring together the most up-todate science and research in their area of expertise, to provide clear, consistent and independent information for the industry. Mastitis was identified by farmers and industry experts as one of the priority diseases. Thus, CellCheck was initiated at the end of 2010, as the AHI-led, national udder health program. Although Ireland is a relatively small dairy producer in global terms, accounting for less than 1% of world dairy production, the Irish dairy industry has a global reach, with 80% of its dairy produce being exported. Over the past two decades, Ireland has become one of the world s leading producers of infant formula, now supplying in excess of 15% of the international market. Figure 1: AHI is a partnership organisation, representing all industry bodies. In 2011, the value of exported dairy products and ingredients reached 2.7 billion, a 17% increase on the previous year. The Irish government s strategy for the agri-food sector, Food Harvest 2020, highlights a promising future for the dairy sector in Ireland (Department of Agriculture, Food and the Marine, 2010). As an EU member state, the milk production of individual dairy farmers in Ireland, and the Irish dairy industry as a whole, is subject to and limited by an EU milk quota system. The expected abolition of this quota system in 2015 will offer Irish farmers an opportunity for expansion. However, it will also result in Irish dairy farmers facing a less-regulated global trading environment with more price volatility than before and an ever-increasing demand for higher standards of milk quality. Superior animal health and milk quality has an important role to play in ensuring competiveness, meeting consumer demand and improving profitability. Data in 2010 from Irish milk recording herds (34% of dairy herds in Ireland), shows that more than 80% of herds have an annual average herd SCC in excess of 200,000 cells/ml. Using this information as a proxy for the national herd, it is clear that there are opportunities to improve udder health nationally. Countdown Symposium

34 McCoy et al. Figure 2: Net farm profit, for a 40 hectare, 94-cow farm in Ireland, at various BMSCC ranges. A coordinated approach to improving udder health, such as CellCheck could provide many benefits for the Irish dairy industry, and individual farmers. The development of CellCheck CellCheck has been shaped by both national and international research and experience. Previous research in Ireland, such as the uromilk pilot mastitis control program, has helped identify some of the obstacles to improving udder health, that exist at farm level. uromilk focussed on facilitating farmers to create multidisciplinary teams with their local service providers (vets, farm advisers, milking machine technicians and/or co-op milk quality advisers), to work together to improve udder health. The main objectives of this pilot study were to identify the drivers and constraints to improving milk quality, from the perspective of participating farmers and other team members, and to determine if a team-based approach to mastitis control could be implementable and effective on Irish farms. Learnings from uromilk identified inconsistent advice, a lack of a coordinated or joined up approach to mastitis control and a normalisation of high herd SCC as some of the obstacles to change at farm level. Other findings, such the motivators to improve milk quality and challenges to implementation of a program such as this on a wider scale have helped guide the development of CellCheck. Close collaboration with Dairy Australia has been invaluable in the development of CellCheck. Access to intellectual property from the Countdown Downunder program, and the expertise of experienced individuals such as Pauline Brightling and John Penry has enabled AHI Ireland to deliver to the industry in a time and resource-efficient manner. CellCheck has convened two important working groups. First, the Technical Working Group (TWG), which is a group of industry experts from various disciplines with skills and expertise in the area of udder health and mastitis control. Their remit is to bring together the known and agreed science around mastitis control, to provide clear and consistent information for the industry. Second, the Industry Consultation Group (ICG), which is a diverse group of individuals with an intimate working knowledge of the Irish dairy industry. They represent the industry groups with the ability to influence change at both farm and industry level. The role of the ICG is to provide industry support, guidance and expertise to assist the development and delivery of the CellCheck program. The CellCheck objectives are: setting goals building awareness establishing best practice building capacity evaluating change. The first year of CellCheck has focused on the agreement of clear, consistent, science-based, information on mastitis control. The TWG reviewed the Countdown Farm Guidelines for Mastitis Control, updating and adapting it as necessary for the Irish industry. This was published in February 2012 as the CellCheck Farm Guidelines for Mastitis Control, and is available to all farmers and their service providers. Since March 2011, a communications strategy, which includes monthly articles in farming media, co-op and client newsletters, etc, has helped to build awareness about the CellCheck program and the value of improved udder health and provides key practical recommendations. A pilot series of farmer workshops has been delivered, the objective of which was to deliver best science and practice information around mastitis control to farmers, and to encourage the uptake of key best practices in everyday milking routines. Engaged local service providers will be the frontline in delivery of these workshops to their farmers. Economic research has also been completed, which looks at the impact of mastitis on farm profitability, as indicated by various bulk tank SCC ranges. This is important to highlight to farmers the value of improving mastitis control, which is often underestimated, or based purely on penalties incurred or cases treated. The economic work continues, now looking at the cost of mastitis to the processing industry. This information will be important in informing discussions within the ICG, around issues such as milk payment policies and motivating change at a higher industry level. 34 Countdown Symposium 2012

35 CellCheck a new solution Building the awareness of service providers of the CellCheck program, the Farm Guidelines for Mastitis Control and their enthusiasm for milk quality, has commenced with multidisciplinary Service Provider Seminars. These seminars, which have aimed to achieve active learning in an enjoyable environment, have been open to all disciplines-veterinary practitioners, milking machine technicians, farm advisors, coop milk quality advisors, commercial sales people etc. Almost 400 service providers have participated to date, and for many, this was their first opportunity to meet people from other disciplines. More than 300 of these participants have expressed an interest in participating in delivery of Farmer Workshops to their clients in the near future. What have we learnt so far? Social science research has been an important component of the CellCheck program. To date, the social science aspect of CellCheck has included two key components. The first was a process called terrain mapping obtaining a clear understanding of the landscape of the Irish dairy industry. This work with members of the ICG has helped identify drivers and constraints at an industry level to improving milk quality. Results from this process showed that within the industry, while there is agreement on the value of enhanced milk quality, reaching a consensus on how to achieve this is contestable. The industry bodies represented in the ICG see themselves with varying levels of responsibility in the process of change. This presents a challenge for understanding the roles they can play in achieving change. The process has also identified the need to clearly communicate the objectives and scope of the program, in order to manage expectations at all levels within the industry. For example, providing solutions to some challenges that have been identified, such as milk payment policies, are beyond the scope of CellCheck. However, the program can facilitate industry discussion and collaboration on these issues, in order to explore solutions. With 12 different milk processors and another 20 milk buyers in Ireland, encouraging collaboration and ensuring that farmers receive consistent quality signals is not easy! This emphasises the importance of the partnership structure of AHI, and the value of the Industry Consultation Group. Through the ICG, CellCheck can encourage and facilitate an otherwise fragmented industry to discuss and progress issues such as collating national SCC data and agreeing on common industry targets. The conscious decision made to underpin all AHI work by science, including the social sciences, has also been important for enabling stakeholders to reach a consensus and make informed decisions for a greater industry good, rather than primarily for personal gain. The second social science component of CellCheck involved the use of a farmer survey as part of the pilot farmer workshops, to determine the extent to which key best practices were being implemented on the farm. Figure 3: CellCheck Farm Guidelines for Mastitis Control front cover. A key finding from this survey is that a significant gap exists between the routine practices and behaviours that people report and the standard to which those practices are done. For example, almost all farmers that participated in the farmer workshops reported that they carried out teat disinfection after every milking. However, data collected showed that approximately 50% of surveyed farmers used less than half the recommended amount of teat disinfectant. This would indicate that the quality of the teat disinfection and teat coverage achieved is poor. Supporting results from uromilk and other Irish research, this suggests that though practices may be implemented on the farm, it is necessary to review the quality of implementation and farmers understanding of why such practices are being implemented. Mastitis is control is achievable it is not just an inevitable part of farming. The solutions lie in understanding that mastitis is multifactorial in nature, and implementing existing science and knowledge, rather than waiting patiently for a silver bullet. The key strengths of CellCheck are its multidisciplinary and collaborative nature, involving all relevant industry bodies in both the development and delivery of the program. While the science behind the CellCheck program is not new, the approach to disseminating and encouraging adoption of this science is. Countdown Symposium