The use of mobility score to predict dairy cow reproductive performance

Transcription

1 The use of mobility score to predict dairy cow reproductive performance by James Alex Willshire BVSc CertCHP MRCVS A dissertation of 8,667 words submitted as part of the requirements in the examination for the RCVS Diploma in Cattle Health and Production. February 2012

2 1 Acknowledgements Thanks must first go to Jon Huxley and Nick Bell for their inputs and enthusiasm through the entire process from inception to the document you see before you without them I never would have stood a chance! This project would not have been possible without the help and enthusiasm of the farmers concerned; Richard and Clare Sainsbury, Andy Mathews, Dan Warman and Liz Birkett and Paul Keeton and Nick Helyer so my heartfelt thanks to them all. Thanks to Jasmeet Kaler for her tireless and patient tuition on the use and interpretation of Stata and Cox Proportional Hazard Analysis and to James Breen for proof reading a draft of this thesis. Finally thanks to my family for their support throughout and to Molly, my faithful companion, for her never ending support, forgiveness and cheerfulness. 2

3 2 Table of Contents 1 Acknowledgements Table of Contents Table of Figures Table of Tables List of abbreviations Abstract Hypothesis Introduction Materials and Methods Herd Details Herd Management Foot Health Reproductive Management Data Capture and Handling Mobility Scores Body Condition Scores Clinical Events Reproductive Events Milk Recording Data Background Cow Information Statistical Analysis Results General Mobility Scores Body Condition Scores Analysis Presence or absence of lameness Number of lame scores Timing of lame scores Chronicity of lame scores Discussion Limitations Suggestions for further work Conclusions References Appendix Appendix

4 16 Appendix Appendix Appendix MS results by Cow Interval for individual herds Associations between predictor variable

5 3 Table of Figures Figure Histogram of Ca-Con for cows conceiving within 200 days Figure Kaplan-Meier Survival Curve and 95% confidence intervals for individual herds (Herd 1 = blue, Herd 2 = red, Herd 3 = green, Herd 4 = orange) Figure Summary of mobility score by herd for study period Figure 10.4 Mobility score by Cow Interval for all herds Figure 10.5 BCS summary by Cow Interval for Herd Figure BCS summary by Cow Interval for Herd Figure BCS summary by Cow Interval for Herd Figure 10.8 BCS summary by Cow Interval for Herd Figure Kaplan-Meier Survival Curve and 95% confidence intervals for cows who ever recorded a lame score (red) vs cows who never recorded a lame score (blue) Figure Kaplan-Meier Survival Curve for cows recording no lame scores (blue), one or two lame scores (red) or three or more lame scores (green) Figure Kaplan-Meier Survival Curve with 95% confidence intervals for cows recording no lame scores (blue), lame scores in the 1, 2 or 3 rd recording (red) or lame scores in the fourth or later recordings (green) Figure Kaplan-Meier Survival Curve with 95% confidence intervals for cows recording no lame scores (blue), two or more consecutive lame scores (red) or one or more lame scores but never consecutively (green) 35 Figure 13 - Body Condition Scoring Figure MS by Cow Interval for Herd Figure MS by Cow Interval for Herd Figure MS by Cow Interval for Herd Figure MS by Cow Interval for Herd Table of Tables Table Summary of literature regarding effects of lameness on calving to conception interval Table Background Herd data (means for the 12mth period 01/07/09-31/06/10 prior to study onset) Table Description of the DairyCo mobility score used in the study (DairyCo 2010) Table Definitions of categories created to summarise mobility score results Table Definitions of the BCS summaries used Table Categories for month of latest calving Table Univariable Analysis for variables defined in Methods and the calving to conception interval Table Distribution of cows suffering one or more lame scores vs cows who were never lame Table 10.3 Ca-Con description Table Multivariable Analysis for EverLame Table Distribution of cows who were never lame vs cows who suffered one or two lame scores vs cows who suffered three or more lame scores

6 Table Multivariable analysis for "LameEvents" Table Distribution of cows who were never lame vs a lame score in 1, 2 or 3 rd recording vs cows a lame score in the fourth or latest recording Table Multivariable analysis for "LameIntervals" Table Distribution of cows who were never lame vs two more consecutive lame scores vs cows who recorded single or multiple lame scores but never consecutively Table Multivariable analysis for "LameConsecutive" Table Lameness Incidence Values Table Lameness Prevalence Values Table Associations between lameness and various reproductive parameters Table Correlations between final model covariables Table Details of correlations between SCC (x000) at first milk recording and cows who were ever lame 59 6

7 5 List of abbreviations BCS Body condition score as defined by Edmonson, Lean et al. (1989) Ca-Con Calving to conception interval (days) CI Confidence Interval DMI Dry Matter Intake HR Hazard Ratio MS Mobility score as defined by Whay, Main et al. (2003) UK United Kingdom US United States VWP Voluntary Waiting Period 7

8 6 Abstract This dissertation investigates the impact of lameness, as measured by mobility score two or three (on a zero to three score) on the reproductive performance, as measured by calving to conception, of UK dairy cows. The dataset consists of approximately 14,000 mobility and body condition scores taken from 836 cows taken over an 18 month period. Cows were enrolled at calving and scored every two weeks until they conceived or were 200 days in milk. A Cox Proportional Hazard analysis was used to examine the risk of conception on cows who were ever lame, cows who recorded multiple lame scores, cows who recorded lame scores in the first six weeks of lactation or later, cows who were lame on two or more consecutive occasions compared to cows who were never lame. This was with reference to the effects of herd, parity, period of calving, total number of health events (at a cow level, which included; dystocia, left displaced abomasum, mastitis, milk fever, retained foetal membranes, right displaced abomasum, routine foot trims, twins and vulval discharge), together with milk protein (%),fat (%), yield (L) and somatic cell count at first milk recording. Cows who ever suffered a lame score (median 178 days), cows who recorded three or more lame scores (median 200 days), cows who recorded lame scores the eighth week of lactation or later (medians 189 days) and cows who were lame on two or more consecutive occasions (median 200 days) all took longer to conceive than nonlame cohorts (median 117 days). This suggests that the presence of lameness, its timing and chronicity all have an effect on the subsequent reproductive performance. 7 Hypothesis Lameness (as measured by a score 2 or 3 on a 0-3 mobility scoring system) in cows 14 days calved to conception (or 200 days in milk) negatively influences reproductive performance (as measured by days to conception). Null hypothesis; lameness (as measured by a score 2 or 3 on a 0-3 mobility scoring system) has no effect on subsequent reproductive performance. 8

9 8 Introduction Lameness has been described as the single most common cause of distress in dairy cows (Webster 1986) and levels in UK have been termed unacceptable (FAWC 1997). A recent report suggested that these levels had not changed significantly in the last decade, a situation which needs to be tackled urgently (FAWC 2009). Lameness also has impacts on health and productivity resulting it being a major reason for early culling (Greenough and Weaver 1997; Boettcher, Dekkers et al. 1998). Domesticated cattle are descended from herbivores and are essentially a prey species; as a consequence they have maintained a strong evolutionary pressure to mask pain and weakness (Phillips 2002), this then transfers to the expression of lameness. The current UK lameness situation is difficult to accurately establish as studies often use multiple observers (with the inherent issues of inter-observer variation (Channon, Walker et al. 2009)) and include observations by farmers, who consistently under-report lameness incidence and prevalence (Wells, Trent et al. 1993; Whay, Main et al. 2002; Whay, Main et al. 2003). Despite this the prevalence is believed to be around 20% (see Appendix 1 (p.48) and Appendix 2 (p.51)). Lameness is described as a debilitating condition, associated with tissue damage, pain and discomfort manifesting as an inability to walk normally (O'Callaghan 2002). Eighty eight percent of lameness is associated with disorders of the feet (Russell, Rowlands et al. 1982) of which 92% will be in the hind feet with the lateral hind claw being four times more likely to suffer a lameness causing lesion than the medial (Clarkson, Downham et al. 1996). Sixty five percent of lameness involves the claw (rather than the skin) with sole ulcers ( %) and white line lesions ( %) consistently quoted as being the most common causes (Clarkson, Downham et al. 1996; Barker, Amory et al. 2007). Various visual scoring techniques for gauging the degree of lameness have been developed which include; asymmetry of gait (Wells, Trent et al. 1993), back posture when standing and walking (Sprecher, Hosteler et al. 1997) and length of stride (Whay, Waterman et al. 1997). The system adopted by the UK is the 0-3 DairyCo mobility score as described by Whay et al. (2003) and summarised in Table 9.2 (p.16). Dairy herd fertility in the UK is thought to be declining with first service pregnancy rates falling at a rate of approximately one percent per annum and the calving index of the national herd continuing to rise (Huxley 2009). Much of this decline is attributed to increasing yields, as frequency of standing to be mounted and first service pregnancy rate are negative correlated to yield (Dobson, Walker et al. 2008), however other production diseases which can negatively impact fertility cannot be ignored. Lameness and fertility, as with most production diseases of dairy cows, are multi-factorial and highly interlinked. There is an increasing body of work regarding the negative effects lameness can have on fertility (ArgaezRodriguez, Hird et al. 1997; Melendez, Bartolome et al. 2003; Garbarino, Hernandez et al. 2004), however studies using mobility score as a measure of lameness to investigate the associations with fertility are limited (Sprecher, Hosteler et al. 1997; Hernandez, Garbarino et al. 2005; Bicalho, Vokey et al. 2007; Peake, Biggs et al. 2011). Most studies investigating associations between lameness and fertility find a negative 9

10 association (Lucey, Rowlands et al. 1986; Collick, Ward et al. 1989; Tranter and Morris 1991) however others have found no association (Cobo-Abreu, Martin et al. 1979; Dohoo and Wayne Martin 1984; Peake, Biggs et al. 2011). Table 8.1 summarises the recent literature surrounding the effects of lameness on the calving to conception interval, further effects of lameness on various fertility factors can be seen in Appendix 3 (p.52). Table Summary of literature regarding effects of lameness on calving to conception interval Effect on calving to conception Description Study location Study No effect Lameness (determined by mobility score) (Peake, Biggs et al. had no effect on calving to conception on UK 2011) UK cows using DairyCo scoring system Extended intervals Cows scoring >2 (on a 5-point mobility (Sprecher, Hosteler et US score) al. 1997) 3.4 day extension Lameness diagnosed by both farmer and vet Netherlands (Barkema, Westrik et al. 1994) 8.9 day extension Clinical digital disease in cows culled for (Enting, Kooij et al. Netherlands this reason 1997) 11 day extension Sole ulcer or white line disease days (Lucey, Rowlands et UK in milk al. 1986) 14 day extension UK (Collick, Ward et al. 1989) 20 day extension Cows affected with digital dermatitis Mexico (ArgaezRodriguez, Hird et al. 1997) 28 day extension US (Lee, Ferguson et al. 1989) (Hernandez, 36 day extension Cows with low cumulative mobility scores US Garbarino et al. 2005) 40 day extension Lame cows with claw lesions relative to healthy US (Hernandez 2001) (Hernandez, 50 day extension Cows with high cumulative mobility scores US Garbarino et al. 2005) Previous work has associated lameness in the post-partum period with a higher incidence of ovarian cysts (Melendez, Bartolome et al. 2003), delayed onset of ovarian cyclicity (Garbarino, Hernandez et al. 2004), reduced pregnancy rates (Suriyasathaporn, Nielen et al. 1998; Hernandez 2001; Hultgren, Manske et al. 2004), a requirement for more serves per conception (Collick, Ward et al. 1989; Sprecher, Hosteler et al. 1997) and a lower frequency of standing to be mounted (Sood and Nanda 2006). 10

11 It has been suggested that lameness is more likely to occur in the first 60 days of lactation (Rowlands, Russell et al. 1985; Gröhn, Erb et al. 1990) as a result of claw horn disruption at parturition and the critical changes occurring during the transition period, to include changes to the under-foot and housing environment, feeding management and nutrition (Fourichon, Seegers et al. 1999; Bergsten 2001; van Amstel and Shearer 2001). Lameness is a painful and stressful condition resulting in hyperalgesia and catabolism (Whay, Waterman et al. 1997) which can result in increases in the circulating levels of catecholamines, glucocorticoids and stressinduced prostaglandins from the adrenal glands (Watson and Munro 1984; Nanda, Dobson et al. 1990). Increases in adrenocorticotrophic hormone, cortisol and progesterone have been related to a delay or inhibition of the gonadotrophic releasing hormone (GnRH) and/or luteinising hormone (LH) surge and the alteration of normal follicular activity with the formation of persistent ovarian follicles (Nanda, Dobson et al. 1990; Dobson, Ribadu et al. 2000; Dobson and Smith 2000). It is likely that this pain results in lame cows spending longer lying down leading to a reduced dry matter intake (DMI) with less time spent ruminating (Hassall, Ward et al. 1993; Singh, Ward et al. 1993). As a result these animals may lose more condition and experience a greater energy deficit when compared with non-lame postpartum animals. Excessive body condition loss during early lactation may effect cyclicity via; an inhibitory effect on ovarian follicular growth and development (Lucy, Staples et al. 1992; de Vries and Veerkamp 2000), less frequent pulses of LH and lower serum concentrations of insulin-like growth factor 1 (IGF-1), these act synergistically to promote follicular development (Lucy 2000; Lucy 2001). Dominant follicles in cows experiencing negative energy balance need more time and must reach a larger size to produce oestradiol levels capable of inducing ovulation (Beam and Butler 1999). Lameness can also influence oestrus expression, as although lame cows have a similar duration and incidence (once ovarian cyclicity has restarted (Walker, Smith et al. 2008)) as non-lame cows, lame cows have a lower frequency of standing to be mounted, with a recent study reporting 2.4 compared to 8.0 mounts/hour (Sood and Nanda 2006) together with a lower intensity (Van Eerdenburg, Karthaus et al. 2002) suggesting that the odds of being served are reduced, probably as a result of the pain related stress. In summary, as lame cows may lose more weight and may have a more pronounced negative energy balance it seems less likely they will display oestrus and be served, more likely to suffer concurrent disease which might influence conception, but also if served they are less likely to ovulate and conceive. High yielding cows experience an increase in energy requirements to facilitate the dramatic increases in daily milk yield which peaks at approximately four to eight weeks post-partum, this increased requirement is met through a combination of increased feed consumption and mobilisation of body reserves resulting in a loss of condition, while it is normal for the high yielding animal to lose body condition and enter a period negative energy balance (Grummer 2007), it is the amount of body condition loss and the depth of the negative energy balance which is important. Body condition score has been genetically and phenotypically correlated to fertility performance (Berry, Buckley et al. 2003; Buckley, O Sullivan et al. 2003). Cows who are in low BCS at calving 11

12 or suffer excessive loss during early lactation are less likely to ovulate, have decreased submission rate, pregnancy rates to first service, six week pregnancy rate and an increased calving to conception interval (Roche, Friggens et al. 2009) partly as a result of impaired oocyte competence associated with low BCS (Snijders, Dillon et al. 2000). Excessive BCS at calving can also have a detrimental effect on fertility performance due to impaired DMI resulting in greater fat mobilisation and a more severe BCS loss (Roche, Friggens et al. 2009). Increasingly, evidence is coming to light of negative energy balance resulting in compromised immune systems (Nonnecke, Kimura et al. 2003; Hammon, Evjen et al. 2006; Wagner, Schimek et al. 2008) leaving the cow concerned more likely to suffer concurrent disease and/or clear pre-existing uterine contaminations again reducing the risk of conception. Yield has a negative association to fertility (Royal, Darwash et al. 2000; Butler 2003) previously defined as a decrease in the percentage of cows standing-to-be-mounted (Van Eerdenburg, Karthaus et al. 2002; Roelofs, Van Eerdenburg et al. 2005; Walker, Smith et al. 2008), an increase in the number of silent heats (Harrison, Ford et al. 1990) and a decrease in the risk of pregnancy and first insemination (Suriyasathaporn, Nielen et al. 1998). Yield has also been suggested to increase the risk of lameness, with each additional 100kg of milk produced by 100 days in milk in the preceding lactation the risk of being lame increased by 1.06 (Barkema, Westrik et al. 1994). Milk fever, uterine infection, dystocia, clinical mastitis and retained foetal membranes have all been suggested to negatively influence subsequent fertility (Borsberry and Dobson 1989; Oltenacu, Bendixen et al. 1990a; Suriyasathaporn, Nielen et al. 1998). Management of the herd may also have effects on fertility performance. Housing systems can directly influence fertility as cows kept on hard concrete surfaces show less mounting activity than cows kept on dirt yards (Britt, Scott et al. 1986). Housing may indirectly affect fertility performance by increasing the incidence of other diseases including lameness, with cows housed in cubicles having a higher incidence of lameness than those housed on straw yards in one study (Baggott and Russell 1981). It has also been suggested that lameness incidence may be higher in larger herds (Whitaker, Kelly et al. 2000). Lameness can increase the incidence of other disease, lame cows find it harder to lie down and rise which increases the risk of teat tramps (Rajala-Schultz and Gröhn 1999) which in turn predicts clinical mastitis (Bendixen, Vilson et al. 1988; Oltenacu, Bendixen et al. 1990b; Elbers, Miltenburg et al. 1998). Herd level associations between poor foot health and high incidence of clinical mastitis have been made (Arvidson 2000) with feet being more frequently trimmed in herds with low bulk milk somatic cell count than in herds with high (Ekman 1998). However Hultgren, Manske et al. (2004) failed to find any association between sole ulcer and clinical mastitis or elevated bulk milk somatic cell count. Cows who suffer digital diseases are more likely to suffer metabolic disease, with an odds ratio of 1.60, and an attributable risk of 0.12 reported in one study (Enting, Kooij et al. 1997). The same study also found that three percent of gynaecological disorders and four percent of retained foetal membranes could be attributed to digital disease. 12

13 A recent review by Madouasse, Huxley et al. (2010) found that the probability of conception before 145 days in milk increased with lower milk production on the second test-day, higher percentage of protein on the second test-day, and higher percentage of lactose on the first test-day. A seasonal aspect has also been suggested to lameness, with a higher incidence and prevalence in the winter than the summer (Clarkson, Downham et al. 1996; Whitaker, Kelly et al. 2000). 13

14 9 Materials and Methods 9.1 Herd Details A convenience sample of four herds were selected for this project based on their geographical proximity to the author s location, willingness to participate, commitment to milk recording and record keeping practices. Herd health history, including the prevalence or incidence of lameness was not part of the selection criteria. All four herds are situated within a 20 mile radius of Salisbury, Wiltshire. Appropriate background data for the herds enrolled are summarised in Table 9.1. Table Background Herd data (means for the 12mth period 01/07/09-31/06/10 prior to study onset) Herd 1 Herd 2 Herd 3 Herd 4 Number of cows in milk Milk Production (kg, 305d) 5,889 8,057 9,171 9,005 Proportion of cows conceived in 200d (%) Calving to conception (d) Calving-first service (d) First service pregnancy rate (%) day pregnancy rate (%) Bulk Tank Somatic Cell Count (cells/ml) 71, , , , Herd Management The four herds enrolled all had all-year round calving patterns with annual rolling mean herd sizes of 138 to 359 cows in milk and 305d yields of 5,889-9,171kg/cow (National Milk Records). The cows were predominately Holstein x Friesians housed through the winter period (November to February) and with access to pasture during the summer (March to October), although one herd was zero grazed (herd 3). Three of the herds are managed as two groups depending on yield and one as a single group (herd 1). All four herds, when housed, use cubicles; one on deep sand (herd 4), one on mats with straw (herd 3) and two on mats with sawdust. All herds dry cows were loose housed on straw. All four herds had loose straw areas available for freshly calved and sick cows Foot Health Routine foot trimming was carried out on all four herds, the policy being to trim cows prior to drying off; all four herds employed a contractor for this task. Cows identified as lame were predominately treated by the herdsman unless the foot trimmer was due in the next 24 hours, lesions identified by the foot trimmer were recorded on National Association of Cattle Foot Trimmer (NACFT) record sheets and those identified by the herdsman were recorded in the diary, both records (foot trimmer and herdsman) were transferred to the on-farm 14

15 recording software. All four herds had endemic digital dermatitis controlled through the routine use of formalin or copper sulphate footbaths and antibiotic footbaths during flare-ups Reproductive Management All four herds undertake artificial insemination (AI) with one herd also running a beef stock bull (herd 2) in the lower yielding group, voluntary waiting periods (VWP) for all herds is 42 days. Management strategies for all herds meant that cows which had not been served by ~70 days were presented at the subsequent routine visit and treated appropriately; no reproductive synchronisation programmes were used. Pregnancy diagnosis was undertaken using ultrasound from 30 days onwards. All heats which were not served to, services and pregnancy diagnoses were recorded using on-farm software. 9.2 Data Capture and Handling Visits for mobility score and body condition score capture were made to each herd fortnightly at a time separate to any other veterinary involvement (therefore two herds were visited every other Wednesday) beginning on the 01/07/2010 through to the 14/01/2012. Cows were enrolled into the study at calving, were scored at the subsequent visit and every visit afterwards until they were confirmed pregnant, scores collected after confirmation of pregnancy were retrospectively removed. Cows were identified in the parlour by freeze brand prior to scoring and scores were captured on a pre-populated form to try and minimise the number of missed cows. Two scoring intervals were defined: Overall Interval. A visit was undertaken every week for the 18mth study period with each farm visited fortnightly, making a total of 81 visits. The Overall Interval number defines the number of the weekly visit e.g. a cow calves on the 01/09/10 and is first scored at the 62 nd Overall Interval. Cow Interval. Each enrolled cow was scored every two weeks from calving to conception or until 200 days in milk making a potential maximum number of scoring visits of 15. The Cow Interval defines the number of the scoring visit at a cow level e.g. a cow calves on the 01/09/10 and is first scored at the 62 nd Overall Interval, at a cow level this was the 1 st Cow Interval, two weeks later (the 64 th Overall Interval) the cow is scored at the 2 nd Cow Interval. A Microsoft Access database was constructed to hold the following captured information: Index table containing all of the enrolled cows background details (9.2.6 Background Cow Information). Mobility scores (9.2.1 Mobility Scores). Body condition scores (9.2.2 Body Condition Scores). Clinical events (9.2.3 Clinical Events). Reproductive events (9.2.4 Reproductive Events). Milk recording data (9.2.5 Milk Recording Data). 15

16 9.2.1 Mobility Scores Routine mobility scores did not form part of the routine foot health management any of the herds, although two undertook quarterly mobility scores as part of a milk buyer contractual agreement. All four herds understood the observational nature of the study and scores were not reported back to the herd managers. All four herds were mobility scored using the DairyCo method as described by Whay, Main et al. (2003) on a fortnightly basis as summarised in Table 9.2. All scores were undertaken by the author. Table Description of the DairyCo mobility score used in the study (DairyCo 2010) Score Category of score Description of cow behaviour 0 Good mobility Walks with and even weight bearing and rhythm on all four feet, with a flat back. Long, fluid strides possible. 1 Imperfect mobility Steps uneven (rhythm or weight bearing) or strides shortened; affected limbs or limbs not immediately identifiable. 2 Impaired mobility Uneven weight bearing on a limb that is immediately identifiable and/or obviously shortened strides (usually with an arch to the centre of the back). 3 Severely impaired mobility Unable to walk as fast as a brisk human pace (cannot keep up with the healthy herd) and signs of score 2. Cows were scored on exit from the parlour after morning or evening milking on flat, level concrete for at least 15 unhurried strides whilst observed from the side and behind. Cows were not sorted or footbathed on the day of scoring. Once captured, scores were transferred to the Microsoft Access database. To provide a binary input to the model mobility scores were reclassified from the 0-3 scale described in Table 9.2. Mobility scores 2 and 3 were assigned the value 1 ( lame ) and scores 0 and 1 the value 0 ( not lame ). Four categorical variables were created to try and explore the interactions of lameness (as defined by mobility score 2 or 3) and the calving to conception interval, these were; whether a cow was ever lame (EverLame), the number of lame scores (LameEvents), whether the lame scores occurred in the first, second or third scoring or later (LameInterval) and whether two or more lame scores occurred consecutively or not (LameConsecutive). These are defined in Table 9.3. Table Definitions of categories created to summarise mobility score results Category Definition 0 Cows who were never lame EverLame 1 Cows who were ever lame 1 Cows who were never lame LameEvents 2 Cows who had one or two lame scores at any time 3 Cows who had three of more lame scores at any time 1 Cows who were never lame LameInterval 2 Cows who were lame during the first, second or third assessment 3 Cows who were lame during or after the fourth assessment 16

17 LameConsecutive 1 Cows who were never lame 2 Cows who were lame on two or more consecutive recordings 3 Cows who were lame but were never lame consecutively at any point Body Condition Scores Routine body condition scoring did not form part of the routine herd management and results were not communicated back to the herd manager. Scores were captured at the time of mobility scoring using a 0-5 scale (with 0.25 points) as described by (Edmonson, Lean et al. 1989) see Appendix 4 (p.55). Once captured, scores were transferred to the Microsoft Access database. Up to 15 body condition scores for each cow were available these were reclassified from the scale (with 0.25 intervals) to binary categories as described in Table 9.4 Table Definitions of the BCS summaries used Category Definition 2 BCS of two or less at first recording Body condition score at first milk 3 BCS of three at first recording recording BCS of four at first recording (there were no cows 4 calving in with BCS greater than four) Cows who lost greater than or equal to one BCS by the 1 Body condition score loss of 1.0 by the sixth recording 6 th measurement (~80 days in milk) Cows who lost less than one, stayed the same or gained 0 BCS by the sixth recording Clinical Events The following clinical events were captured by the respective herdsman into on-farm record keeping software: Dystocia. Lameness. Left displaced abomasum. Mastitis. Milk Fever. Retained foetal membranes. Right displaced abomasum. Routine foot trims. Twins. Vulval discharge. Definitions were agreed prior to onset of the study. On a monthly basis clinical events were imported into the Access database from the on-farm software via Interherd. The total number of clinical events occurring before conception or 200 days in milk (whichever came first) was calculated for input into the model. 17

18 9.2.4 Reproductive Events Heats (not served to), services and pregnancy diagnosis results were captured by the farmer into diaries then transferred to the on-farm recording software. These reproductive events were downloaded into Interherd on a monthly basis and transferred to the Access database. Calving to conception intervals were calculated. Calving to conception intervals of less than 42 days were censored in line with the farm voluntary waiting periods and cows who failed to conceive within 200 days were right censored Milk Recording Data All four herds were enrolled in monthly milk recording to include as a minimum; somatic cell count, milk fat, milk protein and yield. Milk recording information for cows less than or equal to five days was discarded as described in Madouasse, Huxley et al. (2010). On a monthly basis milk recording data were imported into the Access database from the on-farm software or milk recording company via Interherd. Up to 15 milk recordings were available for each enrolled cow; the yield (kg, continuous), milk protein and fat (%, continuous), somatic cell count (cell/ml, categorised as 200 or >200) and days in milk at first recording (continuous) were identified for addition to the model Background Cow Information Baseline information for each enrolled cow was captured, which compromised; herd, date of birth, parity and latest calving date. The month of the latest calving was categorised as summarised in Table 9.5. Table Categories for month of latest calving Month of calving Category January-March 1 April-June 2 July-September 3 October-December 4 Herd, parity and month of calving category were identified for addition to the model. 9.3 Statistical Analysis Kaplan-Meier survival curves were generated for the effect of herd, parity, latest calving quarter, the presence or absence of lame scores (Figure 10.9), the number of lame scores (Figure 10.10), the timing (Figure 10.11) and chronicity of the lame scores (Figure 10.12). The median time to conception was then compared for the herds, the presence or absence of lame scores (Table 10.2), the number of lame scores (Table 10.5), the timing (Table 10.7) and chronicity of the lame scores (Table 10.9). Univariable analysis between calving to conception and the variables identified above was undertaken to identify correlations (Table 10.1), any predictors with a p<0.1 were carried through to the multivariable 18

19 analysis. Associations between the predictor variables were also explored using pairwise correlations and chi square test (Appendix 5, p.56). A Cox proportional hazard regression 1 was used to investigate factors associated with time to conception with outcome defined as the calving to conception interval. Since variables created for mobility (as defined in Table 9.3; EverLame, LameEvents, LameInterval, LameConsecutive) were correlated they could not be added together, four separate models were built. 1 The Cox model produces a hazard ratio which is measure of relative risk, a HR of 1.0 means that the risk is no different between the two groups (e.g. HR for the risk of pregnancy in cows who have white ears is 1.0 compared to cows who have black, that is to say cows with white ears have equal chance of conception as those with black); a HR of 0.5 means that the risk in the group exposed to a variable is 50% higher than those in a group not exposed (e.g. HR of abortion for cows treated with exogenous prostaglandin at day 100 of pregnancy is 0.5 compared to cows who are not treated, that is to say that cows treated with prostaglandin at day 100 of pregnancy are 50% more likely to abort than those who aren t). 19

20 10 Results A total of 84 visits were made during the course of the project, with each farm receiving 42 visits over a 565 day period (~18 months). A total of 19,238 MS and BCS observations were made on 836 cows of which 5,180 (26.9%) were excluded as they were captured after conception leaving 14,058 eligible. Eight thousand and ninety four (57.4%, 308 cows) observations were censored for a failure to conceive within 200 days as described in the methods and a further 132 observations (15 cows) were censored as a result of Ca-Con of <42 days leaving a total 5,964 observations on 513 cows with Ca-Con intervals 42 and < 200 days. Over the total study period on eligible cows 120 MS and BCS scores were missed due to a failure to identify the cow concerned. The mean number of mobility scores per cow was General Figure 10.1 illustrates a histogram of the calving to conception interval for all cows who conceived within 200 days in milk (cows who conceived later than 200 days in milk were censored, see Reproductive Events ). The data is left skewed and the Anderson-Darling test for normality confirms that the data is not normally distributed (p<0.005). Figure Histogram of Ca-Con for cows conceiving within 200 days The effect of Herd was strongly associated with the Ca-Con throughout the analysis, as demonstrated by Figure

21 Kaplan-Meier survival estimates analysis time Figure Kaplan-Meier 95% CI Survival Curve 95% and CI 95% confidence intervals 95% CIfor individual 95% herds CI (Herd 1 = blue, Herd 2 = red, Herd 3 = green, herd_id Herd 4 = = 1orange). herd_id = 2 herd_id = 3 herd_id = Mobility Scores The mean prevalence of lameness across all herds for the entire study period was 16% (defined as cows scoring 2 or 3) with a range of 7-32% by herd (Figure 10.3). Forty five percent of cows were lame on at least one visit, the median number of lame scores for each lame cow was 2 (inter-quartile range 1-5) with the maximum number of lame records being % 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Herd Figure Summary of mobility score by herd for study period 21

22 Number The number of cows in each scoring group for all herds across the entire study period is summarised in Figure 10.4 (MS results by Cow Interval for individual herds can be seen in Appendix 5, p.56) Cow Interval Figure 10.4 Mobility score by Cow Interval for all herds 10.3 Body Condition Scores A total of 9,543 individual cow body condition scores were collected with an additional 76 cows being missed/absent from a scoring visit. Body condition scores for each of the herds at a Cow Interval level are summarised below in Figure 10.5, Figure 10.6, Figure 10.7 and Figure

23 Number Number Cow Interval 2 >2 Figure 10.5 BCS summary by Cow Interval for Herd Cow Interval 2 >2 Figure BCS summary by Cow Interval for Herd 2 23

24 Number Number Cow Interval 2 >2 Figure BCS summary by Cow Interval for Herd Cow Interval 2 >2 Figure 10.8 BCS summary by Cow Interval for Herd Analysis Univariable analysis (Table 10.1) demonstrated strong associations (p<0.05) between Ca-Con and herd, days in milk at first milk recording, milk protein at first milk recording, loss of 1.0 BCS at 6 th recording, ever recording a lame score (EverLame), recording one or two lame scores, recording three or more lame scores (LameEvent), recording a lame score in the first three scores after calving, recording a lame score in the four or 24

25 more scores after calving (LameInterval), recording two or more consecutive lame scores or recording lame scores but never consecutively (LameConsecutive). Calving in the Oct-Dec quarter and SCC of >200,000cells/ml at first milk recording were correlated to Ca-Con (p<0.1). All 12 variables were carried through to the multivariable analysis. Table Univariable Analysis for variables defined in Methods and the calving to conception interval Variable Name No of observations (%) Hazard Ratio p-value 95% Confidence Interval Herd_ID Herd (16) 1.0 Ref Ref Ref Herd (44) 0.34 < Herd (22) 0.25 < Herd (18) 0.56 < Parity (30) 1.0 Ref Ref Ref (19) (15) (36) Latest Calving Quarter Jan-Mar 184 (22) 1.0 Ref Ref Ref Apr-Jun 142 (17) Jul-Sep 296 (35) Oct-Dec 214 (26) Total Health Events 436 (100) Milk Recording Variables Days in milk at FIRST milk recording 807 (100) Milk Protein % at FIRST milk (100) recording Milk Fat % at FIRST milk recording 742 (100) Milk Yield (L) at FIRST milk (100) recording SCC (x000) at FIRST milk recording (85) 1.0 Ref Ref Ref > (15) BCS Variables BCS at FIRST measurement (73) 1.0 Ref Ref Ref (26) (1)

26 BCS loss of 1.0 by 6 th measurement 150 (20) MS Variables EverLame No 457 (55) 1.0 Ref Ref Ref Yes 379 (45) 0.67 < LameEvents Never Lame 457 (55) 1.0 Ref Ref Ref One or two LAME events 204 (24) Three or more LAME events 175 (21) 0.54 < LameIntervals Never Lame 457 (55) 1.0 Ref Ref Ref LAME during the 1, 2 or 3rd measurement 245 (29) LAME during or after the 4th measurement 134 (16) 0.60 < LameConsecutive Never LAME 457 (55) 1.0 Ref Ref Ref LAME on TWO or more consecutive recordings 196 (23) 0.52 < LAME but were never lame consecutively at any point 183 (22) Presence or absence of lameness Three hundred and seventy nine cows (45%) suffered one or more lame scores (as defined by an MS of 2 or 3) (Table 10.2). Cows who ever recorded a lame score (MS 2 or 3) had a mean Ca-Con of 150 days (median 178 days) whereas cows who never recorded a lame score had a mean Ca-Con of 130 days (median 117 days) (Table 10.3), the strong association between Ca-Con and ever recording a lame score was demonstrated by the Kaplan- Meier survival graph (Figure 10.9) and confirmed by multivariable analysis (p<0.001,table 10.4). Cows that were lame had a 32% lower risk of conception (HR=0.68; 95% CI= ) by 200 days compared to cows who were never lame (Table 10.4). 26

27 Table Distribution of cows suffering one or more lame scores vs cows who were never lame Variable Never Lame Ever Lame Cows 457 (100) 379 (100) Herd (24) 26 (7) (35) 208 (55) 3 95 (21) 90 (24) 4 92 (20) 55 (15) Parity (39) 71 (19) 2 93 (20) 66 (17) 3 67 (15) 60 (16) (26) 183 (48) Latest Calving Quarter Jan-Mar 74 (16) 110 (29) Apr-Jun 75 (16) 67 (18) Jul-Sep 183 (40) 113 (30) Oct-Dec 125 (27) 89 (23) SCC (x000) at first milk recording (89) 300 (79) > (11) 78 (21) Total Health Events (50) 171 (45) (34) 144 (38) 2 72 (16) 64 (17) BCS at FIRST measurement (70) 290 (73) (29) 85 (22) 4 3 (1) 4 (1) BCS loss of 1.0 by 6 th measurement Yes 76 (19) 74 (21) No 319 (81) 275 (79) Median Ca-Con (first and third quartiles) 117 (78-200) 178 (98-200) 27

28 Kaplan-Meier survival estimates analysis time Figure Kaplan-Meier Survival 95% Curve CI and 95% confidence intervals 95% CI for cows who ever recorded a lame score (red) vs cows who never recorded a lame everlame score (blue) = 0 everlame = 1 Table 10.3 Ca-Con description n Mean SE Standard Mean Deviation Minimum Q1 Median Q3 Maximum Never Lame Ever Lame Table Multivariable Analysis for EverLame Variable Name Hazard Ratio p-value 95% Confidence Interval Herd_ID Herd Ref Ref Ref Herd < Herd < Herd < Milk Recording Variables Days in milk at FIRST milk recording Milk Protein % at FIRST milk recording SCC (x000) at FIRST milk recording BCS Variables BCS loss of 1.0 by 6 th measurement

29 MS Variables EverLame 0.68 < Number of lame scores Two hundred and four cows (24%) suffered one or two lame scores through their study period, 175 cows (21%) suffered three or more lame scores and 457 cows (55%) never suffered a lame score. Cows who suffered one or two lame scores had median Ca-Con intervals of 150 days which was not statistically significant (p=0.098), whereas those who suffered three or more lame scores had a median of 200 days which was statistically significant (p<0.001); cows who never suffered a lame score had a median Ca-Con of 117 days. Kaplan-Meier survival graphs (Figure 10.10) suggest that three or more lame scores recorded might be significant, this was confirmed by multivariable analysis (p<0.001, Table 10.6) where recording three or more lame scores had a significant effect on the Ca-Con interval. Cows who suffered three or more lame scores had a 49% lower chance (HR=0.51, 95% CI= ) of conception by 200 days compared to cows who were never lame. Table Distribution of cows who were never lame vs cows who suffered one or two lame scores vs cows who suffered three or more lame scores Variable Never Lame One or two lame scores Three or more lame recorded scores recorded Cows 457 (100) 204 (100) 175 (100) Herd (24) 24 (12) 2 (1) (35) 78 (38) 130 (74) 3 95 (21) 62 (30) 28 (16) 4 92 (20) 40 (17) 15 (9) Parity (39) 62 (30) 9 (5) 2 93 (20) 48 (24) 18 (10) 3 67 (15) 25 (12) 35 (20) (26) 69 (34) 113 (65) Latest Calving Quarter Jan-Mar 74 (16) 58 (28) 52 (30) Apr-Jun 75 (16) 24 (12) 43 (25) Jul-Sep 183 (40) 66 (32) 47 (27) Oct-Dec 125 (27) 56 (27) 33 (19) SCC (x000) at first milk recording (89) 164 (81) 136 (78) > (11) 39 (19) 39 (22) Total Health Events 29

30 (50) 96 (47) 75 (43) (34) 75 (37) 69 (39) 2 72 (16) 33 (16) 31 (18) BCS at FIRST measurement (70) 152 (75) 138 (79) (29) 51 (25) 34 (19) 4 3 (1) 1 (0) 3 (2) BCS loss of 1.0 by 6 th measurement Yes 76 (19) 34 (19) 40 (24) No 319 (81) 147 (81) 128 (76) Median Ca-Con (first and third quartiles) 117 (78-200) 150 (85-200) 200 ( ) Kaplan-Meier survival estimates analysis time Figure Kaplan-Meier 95% CI Survival Curve for cows 95% recording CI no lame scores (blue), 95% CI one or two lame scores (red) or three or more lame lameevents scores (green) = 1 lameevents = 2 lameevents = 3 Table Multivariable analysis for "LameEvents" Variable Name Hazard Ratio p-value 95% Confidence Interval Herd_ID Herd Ref Ref Ref Herd < Herd < Herd < Milk Recording Variables 30

31 Days in milk at FIRST milk recording Milk Protein % at FIRST milk recording SCC (x000) at FIRST milk recording BCS Variables BCS loss of 1.0 by 6 th measurement MS Variables LameEvents Never Lame 1.0 Ref Ref Ref One or more LAME events Three or more LAME events 0.51 < Timing of lame scores Two hundred and forty five cows (29%) suffered lame scores in the 1 st, 2 nd or 3 rd recording after their enrolment/calving, whereas only 134 (16%) suffered lame scores in the 4 th or later recordings, 457 cows never suffered a lame score (55%). Cows suffering lame scores in the 1 st, 2 nd or 3 rd recording had a median Ca-Con of 163 days, whereas those recording lame scores later in lactation (4 th or later) had median Ca-Con of 189 days and cows who never suffered a lame event had median Ca-Con of 117 days (Table 10.7). Multivariable analysis (Table 10.8) showed both lame scores in the 1 st, 2 nd or 3 rd recordings and in 4 th or later to have a significant effect on Ca-Con (p<0.05, p<0.001). Cows who were lame in the 1 st, 2 nd or 3 rd recording after enrolment were 20% (HR=0.8, 95% CI= ) and cows who were lame at the 4 th recording or later were 46% (HR=54, 95% CI= ) less likely to conceive by 200 days than cows who were never lame, however the 95% CI for the cows who were lame in the 1 st, 2 nd or 3 rd recording includes 1.0 suggesting that there may be no difference in risk of conception from cows who were never lame, even with a p-value of <0.05. Table Distribution of cows who were never lame vs a lame score in 1, 2 or 3 rd recording vs cows a lame score in the fourth or latest recording Variable Never Lame Lame score in the 1, 2 or Lame score in the 4 th or 3 rd recording later recording Cows 457 (100) 245 (100) 134 (100) Herd (24) 13 (5) 13 (10) (35) 143 (58) 65 (49) 3 95 (21) 59 (24) 31 (23) 4 92 (20) 30 (12) 25 (19) Parity (39) 40 (16) 31 (23) 2 93 (20) 41 (17) 25 (19) 3 67 (15) 42 (17) 18 (13) 31

32 (26) 122 (50) 60 (45) Latest Calving Quarter Jan-Mar 74 (16) 80 (33) 30 (22) Apr-Jun 75 (16) 44 (18) 23 (17) Jul-Sep 183 (40) 62 (25) 51 (38) Oct-Dec 125 (27) 59 (24) 30 (22) SCC (x000) at first milk recording (89) 191 (78) 109 (81) > (11) 53 (22) 25 (19) Total Health Events (50) 102 (42) 69 (51) (34) 101 (41) 43 (32) 2 72 (16) 42 (17) 22 (16) BCS at FIRST measurement (70) 195 (79) 95 (71) (29) 48 (20) 37 (28) 4 3 (1) 2 (1) 2 (1) BCS loss of 1.0 by 6 th measurement Yes 76 (19) 43 (20) 31 (24) No 319 (81) 176 (80) 99 (76) Median Ca-Con (first and third quartiles) 117 (78-200) 163 (88-200) 189 ( ) Kaplan-Meier survival estimates analysis time Figure Kaplan-Meier 95% CI Survival Curve with 95% confidence CI intervals for cows 95% recording CI no lame scores (blue), lame scores in the 1, lameintervals 2 or 3 rd recording = 1(red) or lame lameintervals scores the fourth = 2 or later recordings lameintervals (green) = 3 32

33 Table Multivariable analysis for "LameIntervals" Variable Name Hazard Ratio p-value 95% Confidence Interval Herd_ID Herd Ref Ref Ref Herd < Herd < Herd < Milk Recording Variables Days in milk at FIRST milk recording Milk Protein % at FIRST milk recording SCC (x000) at FIRST milk recording BCS Variables BCS loss of 1.0 by 6 th measurement MS Variables LameIntervals Never Lame 1.0 Ref Ref Ref LAME during the 1, 2 or 3rd measurement LAME during or after the 4th measurement 0.54 < Chronicity of lame scores One hundred and ninety six cows (23%) recorded two more lame scores on consecutive occasions and 183 cows (22%) of cows recorded lame scores, but never consecutively (Table 10.9), 457 (55%) cows did not suffer a lame score. Cows suffering two or more consecutive lame scores had a median Ca-Con of 200 days, whereas those recording single or multiple lame scores but never consecutively had median Ca-Con of 141 days and cows who never suffered a lame event had median Ca-Con of 117 days (Table 10.9). Kaplan-Meier survival curves suggest that recording consecutive lame scores might have a more significant effect on Ca-Con (p<0.001) than non-consecutive lame scores (p=0.0165) (Figure 10.12) which was confirmed by multivariable analysis (Table 10.10). Cows who were lame on two or more consecutive recordings were 48% (HR=0.52, 95% CI= ) less likely to conceive by 200 days than cows who were never lame. 33