Intra-class correlation attributable to claw-trimmers scoring common hind claw disorders in Dutch dairy herds

Transcription

1 Claw Health in Dairy Cows in the Netherlands Chapter 2 Intra-class correlation attributable to claw-trimmers scoring common hind claw disorders in Dutch dairy herds M. Holzhauer 1, C.J.M. Bartels 2, B.H.P. van den Borne 2, G. van Schaik 2 1 Ruminant Health Department, GD Ltd., P.O. Box 9, 7400 AA Deventer, The Netherlands 2 Epidemiology Department, GD Ltd., P.O. Box 9, 7400 AA Deventer, The Netherlands Preventive Veterinary Medicine 75 (2006),

2 Chapter 12 Abstract Data of a cross-sectional study on prevalence of 7 common bovine-claw disorders were used to calculate the intra-class correlation coefficients (ICCs) for claw trimmers. Fifteen professional claw trimmers were trained in diagnosing claw disorders uniformly. During regular trimming of all dairy cows in a herd, they recorded the presence (yes/no) of hind-claw disorders for digital dermatitis (DD), chronic laminitis (CL), sole ulcer (SU), white-line disease (WLD) and interdigital hyperplasia (HYP). For the claw disorders interdigital dermatitis/ heel-horn erosion (IDHE) and sole haemorrhage (SH), claw trimmers also recorded (using graduated scores 1-3) the extent of the lesion. Complete information of 21,153 animals from 361 herds in The Netherlands was recorded between July 2002 and December To assure objective recording of claw disorders, ideally there would not be any correlation between observations. However, quantified ICCs from claw-trimmers observations ranged from 4.9% for DD to 38.0% for CL, while ICCs for herd ranged from 7.9% for SU to 26.7% for DD. Changes in the cut-off values for the diagnosis of IDHE and SH did not result in any relevant changes of the ICCs from claw-trimmers observations and these remained larger than the ICC explained by the herd. Based on these results, we conclude that for estimating measures of frequencies for different claw-disorders, uniformity in recording certain diagnoses becomes an issue to consider when involving more than one observer. Keywords: intra-class correlation coefficient, recorders agreement, monitoring, dairy cattle, multi-level modelling. 50

3 Claw Health in Dairy Cows in the Netherlands Introduction Sound feet and legs are important to the cow for optimal productivity, health and animal welfare (Brand et al., 1996). Lameness as a result of claw problems causes lower milk production, involuntary culling and treatment costs (e.g. Collick et al., 1989; Sprecher et al., 1997; Warnick et al., 2001; Green et al., 2002) and also decreased fertility (Barkema et al., 1994; Melendez et al., 2003; Hernandez et al., 2005). Above all, lameness is one of the most obvious symptoms of impaired animal welfare (Webster, 1995; Whay et al., 1997), because most claw disorders are long lasting and painful (Alban, 1995). Consumer and government organisations consider typical prevalence of claw disorders as unacceptable from an animal welfare point of view (RDA, 1996; Farm Animal Welfare Council, 1997). Most lameness problems are related to claw disorders (Murray et al., 1996; Weaver, 2000). To avert these effects of poor claw health, monitoring claw disorders is a useful tool for: defining a health problem (Mills et al., 1986; Bargai, 2000), directing management decisions and evaluating the influence and approach of these decisions (Brand et al., 1996). Monitoring of foot health can be based on scoring the cow s locomotion (e.g. Mason and Leaver, 1988; Sprecher et al., 1997), or on recording claw disorders from all dairy cows at the moment of claw trimming all dairy cows in a herd. In the last situation, subclinical disorders also are recorded (e.g. Frankena et al., 1992; Vaarst et al., 1998; Manske et al., 2002a). According to Weaver (1997) and Borsberry et al. (1999) recording of lesions should be an integral part of the claw-trimming routine, and interpretation of recordings by the veterinarian should be part of the basis of modern herd health and production management (Noordhuizen, 2003). Data for monitoring claw disorders mostly are recorded by claw trimmers at the time of regular claw trimming all dairy cows of a herd and need to be accurate. The objective of our study was to quantify the variance explained by the recorders (claw trimmers in this case) when scoring different claw lesions in Dutch dairy herds at the time of regular claw trimming of all the dairy cows in a herd and to relate this to the variance explained by herd as a reflection of herdrelated aspects like housing, feeding, and herd management. 51

4 Chapter 12 Material and methods Data and case definition Between July 2002 and December 2003 information about the presence or absence of 7 claw disorders of hind claws of 21,153 animals in 361 dairy herds was collected in The Netherlands at the time of regular claw trimming of all the dairy cows in a herd. Fifteen professional claw trimmers were selected from two private organisations of professional claw trimmers. They had to record claw disorders of all eligible milking cows and sometimes pre-partum heifers (if already in the milking herd to become acclimated ) at the end of their pregnancy in the herd during trimming. Registered were the disorders: digital dermatitis (DD), interdigital dermatitis/heel-horn erosion (IDHE), sole haemorrhage (SH), chronic laminitis (CL: deformed claw, concave dorsal hoof wall with ripples), sole ulcer (SU), white-line disease (WLD) and interdigital hyperplasia (HYP). Information was collected at hind-leg level and presence of a disorder was reported if the disorder was diagnosed in at least one hind claw. Recordings were performed based on visual inspection, hardness of the horn, smell and pain reactions of the cows and based on the internationally standardized diagnoses as described by Espinasse et al. (1984). Recordings were presented as 0 or 1 (absent and present), except for the diagnosis IDHE and SH; these diagnoses were scored in ordinal numbers (0, 1, 2, 3). For IDHE, the gradations were based on the definitions of Peterse (1980), 0: no signs; 1: slightly affected in axial surface of the bulb; 2: some big grooves in the horn of the bulb and/or sole, not extending to the corium; 3: disappearance of heel horn and/or deep defects in the horn of bulb or sole, extending to the corium. For SH, the graduations were based on the definitions as supposed by Bergsten (2000), 0: no signs; 1: mild (single haemorrhagic spot or superficial haemorrhages of a small area, and/or yellowish discoloration); 2: moderate (moderate haemorrhage on a single spot or superficial haemorrhages of several spots or a large area of the sole (>20%)); 3: severe (profound haemorrhage (extending into the corium) on a single spot or extensive haemorrhagic discoloration of the sole (>50%)). Claw trimmers were trained by the first author in diagnosing claw disorders during 3 meetings. One month and one week before the start of data collection, 52

5 Claw Health in Dairy Cows in the Netherlands the 7 claw disorders were discussed by studying photographs of affected claws with the first author as facilitator. Finally, during the first two months of recording, diagnoses of claw trimmers were compared with the diagnosis of the first author at least once during hoof trimming on location. Reproducibility was assessed with the three claw trimmers who had contributed most to the complete dataset (together they had recorded 1/3 of all observations) and with 1 additional professional trimmer of the Veterinary Faculty of Utrecht University who has been training vets and vet students for the past 20 years. At the time of a regular claw-trimming session in a dairy herd, all 66 eligible cows were trimmed in two groups by 1 claw trimmer each, and scored by all four claw trimmers independent of each other. The information was used to calculate kappa values as a measure of agreement between different observers. Because prevalence sometimes were <10% or >90%, also Kendall s measure of concordance W is given. Selection of herds Claw trimmers were requested to select herds where trimming was done on a regular basis and regardless of previous trimming results. Participation was agreed on by the farmer, so farmers could choose not to participate. Herds could participate only once in the study. Ninety percent of the selected herds (360 of 380) agreed to participate. The reasons for refusals were: time-consuming, absence of (financial) compensation, not convinced about the advantages of the recordings, not willing to cooperate for unclear reasons. Statistical analysis ICCs were used to separate the error variance into a proportion for the correlation of animals within a herd and a proportion for correlation of observations by the same claw trimmer (Schukken et al., 2003). This was done using a hierarchical model (MLWIN 2.0, 2004) by estimating the intercept value for each claw disorder while taking both claw trimmer and herd as random effects into the model (random-intercept model). The errors in this model stemmed from a binomial distribution which complicated the interpretation of the output of such a model. An approximation method of the computation of variance components in a linear mixed model for binary data was based on 53

6 Chapter 12 latent variables (Dohoo et al., 2003). The error terms were assumed to follow a logistic distribution with mean zero and variance 2 /3 (Snijders and Bosker, 1999). The computation of the variance at herd and at claw-trimmer level follows the rules for calculation of the intra-class correlation coefficient (ICC= ); h = 2 h / ( 2 h + 2+ cl + 2 /3) cl = 2 c / ( 2 cl + 2+ h + 2 /3) where h and cl stand for the ICC for herd (within-herd correlation) and claw trimmer (within-claw- trimmer correlation) respectively and where 2 2 h and cl stand for the variance at herd and claw- trimmer level in the data. A of zero means that there is no correlation of animals within the same herd or recorded by the same claw trimmer; a value of 1.0 means that there is no difference between animals of the same herd or recorded by the same claw trimmer. In this last situation, it means that there would be no need to record more than one animal per herd, because all animals within the same herd get the same outcome. The analyses were carried out for two different situations. In the first analysis, presence of IDHE or SH in a cow was defined when the grade was 1. In the second analysis, presence of IDHE or SH in a cow was defined when the grade was 2. In this way, the effect of changing the cut-off values for the diagnosis of disorders that were scored in different grades could be determined. Results Observed herd prevalence and within-herd prevalence for the 7 claw disorders are given in Table 1. 54

7 Claw Health in Dairy Cows in the Netherlands Table 1. Observed herd and within-herd prevalence (%) (minimum, 25 th percentile, median, 75 th percentile, maximum for non-zero prevalence herds) of 7 claw disorders as recorded by 15 trimmers (361 dairy herds and 21,153 dairy cows, years ). Chronic laminitis (CL); Digital dermatitis (DD); Interdigital hyperplasia (HYP); Interdigital dermatitis/heel-horn erosion (IDHE); Sole haemorrhage (SH); Sole ulcer (SU); White-line disease (WLD). CL DD HYP IDHE SH SU WLD Herd prevalence Within-herd prevalence Minimum percentile Median percentile Maximum

8 Chapter 12 The average number of cows trimmed per herd was 57.8 (SD: 21.2, median: 57, range: ). Each claw trimmer recorded the results of a median of 15 herds (range: 2-55). The estimated ICCs for claw trimmers (Table 2) exceeded 20% in case of CL, IDHE and SH were higher than the ICCs for herds. Changing the cut-off value for IDHE and SH did not change these results. 2 Table 2. Results of random-intercept models (variance ( ˆ ) and intra-class correlation coefficients (ICC, %) for claw-trimmer and herd effects for 7 bovine claw disorders. Chronic laminitis (CL); (Digital dermatitis (DD); Interdigital hyperplasia (HYP); Interdigital dermatitis/heel-horn erosion (IDHE); Sole haemorrhage (SH); Sole ulcer (SU); White-line disease (WLD)). Presence of IDHE and SH is defined by graduation 1 and graduation 2 (n=361 dairy herds; 21,153 dairy cows; 15 claw-trimmers; years ). Diagnosis of Claw trimmer level Herd level claw disorders 2 ˆ ICC (%) 2 ˆ ICC (%) CL DD HYP IDHE IDHE SH SH SU WLD Estimated kappa values between 3 participating trimmers (A, B, F) for the different lesions were poor for WLD (< 0.4), moderate for CL, DD, IDHE and SH ( ), sufficient for SU ( ), and good for HYP (> 0.8) (see Table 3). The estimated kappa values between 3 participating trimmers (A, B, F) and an external professional (Z) for the different lesions were typically poor for WLD, IDHE and DD (< 0.40), moderate for SH and sufficient for CL, HYP and SU (see Table 3). The results of W-values lead to similar conclusions. 56

9 Claw Health in Dairy Cows in the Netherlands Table 3. Agreement in recording 7 bovine claw disorders between 3 participating trimmers (A,B,F) and an external professional trimmer of the Veterinary Faculty of Utrecht (Z) expressed in Cohen s kappa and Kendall s W values on 66 cows in one herd, Chronic laminitis (CL); Digital dermatitis (DD); Interdigital hyperplasia (HYP); Interdigital dermatitis/heel erosion (IDHE); Sole haemorrhage (SH); Sole ulcer (SU); White-line disease (WLD). (# of cows) Cohens s kappa CL (3) DD (19) HYP (5) IDHE (61) SH (56) SU (7) WLD (20) A-B A-F B-F A-Z B-Z F-Z Kendall s W A,B, F A,B, F,Z Discussion In the present study, uniform diagnoses of claw disorders by a number of professional claw trimmers was achieved for some claw disorders (DD, HYP and SU) but was much less uniform for CL, IDHE, SH and WLD. The magnitudes of ICCs were not expected, because claw trimmers were extensively trained and followed closely for correct recording of claw disorders. Particularly for CL, IDHE and SH, this study showed that presence or absence of the disorder was more dependent on the observer than on herd-related aspects like housing, feeding, pasturing, foot bath management etc. This led us to conclude that recorded data on these claw disorders are more difficult to interpret in studies on disease frequency than other claw disorders. The variances for different claw disorders explained by herd were in agreement with earlier studies and confirmed that herd-level factors are important for the prevalence of hoof lesions (Manske et al., 2002b; Somers et al., 2003; Sogstad et al., 2005). 57

10 Chapter 12 Representativeness of the data Regular claw trimming of all cows in a dairy herd in The Netherlands is performed by professional claw trimmers who are either privately working or are employed by a commercial farm- supporting organisation. Based on information of the Dutch board of claw trimmers and the commercial farm-supporting organisations, there are about 100 private and 200 organisation-employed claw trimmers. The claw trimmers involved in this study were proposed by their employer as being representative for all their employees and they performed claw trimming for almost 100% of their professional activities. The herds included in this study were located in the Northern part of The Netherlands. There was no general information about the background of the herds normally trimmed by claw trimmers. However, there was no reason to believe that these herds were different from a cross-section of all Dutch dairy herds, because the average herd size in this study (57.8 cows/milking herd) was comparable with the average herd size in the Netherlands (56.3 cows/ milking herd; CBS, 2003). In addition, the percentage of herds with all year housed cows (12%) and herds with slatted floors (91%) was not different for herds in this study compared to the total Dutch dairy-herd population (Braam and Swierstra, 1997; Somers et al., 2003; CBS, 2003). Estimations of reproducibility The kappa values we found were comparable to the results of Manske et al. (2002b), who scored lesions by 3 different observers and estimated the best agreement for SU ( = 0.86) and moderate agreement for SH ( = 0.4). Other studies used one or a limited number of persons (e.g. Frankena et al., 1992; Livesey et al., 2003; Somers et al., 2003; Donovan et al., 2004). In the study of Hedges et al. (2000) and Webster (2002), congruity in diagnosing different lesions was promoted by making photographs of the lesions and by discussing the results at regular intervals throughout the trial. Variation in diagnosing claw lesions can be expected because scoring claw lesions is subjective, just as scoring lameness (Greenough, 1987) and the quality of the information depends on the perception and skill of the observer (Whay, 2002). Whay (2002) also suggested that scoring requires optimal conditions such as a suitable area with sufficient light to view all animals. However, this is not always available under 58

11 Claw Health in Dairy Cows in the Netherlands the circumstances in which claw trimmers have to work (the first author s personal observation). In poor circumstances for scoring, claw trimmers should demand better light conditions or refuse to make a diagnosis. In our study, agreement in diagnosis of lesions scored by different observers was quantified for a small group of trimmers towards the end of the study (see Table 3). From the results of this study, it can be concluded that to improve inter-observer repeatability in scoring hoof lesions, perhaps a claw-trimming session with all participating claw trimmers is best organised before the start of a study and can replace the class-room session. An alternative training of scoring would be at the slaughterhouse or/and on slaughterhouse material as done in the study of Manske et al. (2002b). Conclusion Conclusion When recording claw disorders for monitoring herd claw health, we found that the ICC attributable to the person recording these disorders is particularly high for IDHE, SH and CL. For prevalence studies, we suggest gathering information with a small number of well trained claw trimmers. For monitoring studies within and between herds, it is necessary to take the ICCs between claw trimmers into account. Acknowledgements We thank the participating Dutch dairy farmers and hoof trimmers for their cooperation. This study was funded by the Dairy Commodity Board (Rijswijk, The Netherlands). We also thank W. Swart for assistance with the statistical analyses and J. Verhoeff for his critical comment on this manuscript. 59

12 Chapter 12 References Alban, L Lameness in Danish dairy cows: frequency and possible risk factors. Prev. Vet. Med. 22, Bargai, U Lameness in a dairy herd: an epidemiological model. Comput. Bargai, Cont. Educ. U Pract. Lameness 22, in a dairy herd: an epidemiological model. Comput. Cont. Educ. Pract. 22, Barkema, H.W., Westrik, J.D., Keulen, K.A.S. van, Schukken, Y.H. and Brand, A The effects of lameness on reproductive performance, milk production and culling in Dutch dairy farms. Prev. Vet. Med. 20, Bergsten, C Descriptive nomenclature and scoring of foot lesions at hoof trimmer work shop report. Part II. Proc. of the 11 th International Symposium on Lameness in Ruminants, Parma, Italy, Sept. 7-10, pp Borsberry, S., Logue, D., Laven, R., Watson, C.L., Ossent, P. and Livesey, C Report of lameness workshop (Harper Adams) Lameness in the dairy cow. Cattle Practice 7, Braam, C.R. and Swierstra, D Developments in design of concrete floors in dairy cattle houses; impact of environmental demands. Proc. Concrete for a Sustainable Agriculture. Stavanger, Norway. pp.: Brand, A., Noordhuizen, J.P.T.M. and Schukken, Y.H Herd health and production management in dairy practice. Wageningen Pers, Chapter 7 part 1. CBS Agricultural Census Statistics Netherlands, ed. Elsevier, Voorburg, The Netherlands. Collick, D.W., Ward, W.R. and Dobson, H Associations between types of lameness and fertility. Vet. Rec. 125, Dohoo, I.R., Martin, S.W. and Stryhn, H Veterinary Epidemiological Research, Atlantic Veterinary College Inc., Charlotte town, Prince Edward Island, Canada. Donovan, G.A., Risco, C.A., DeChant Temple, G.M., Tran, T.Q. and Horn, H.H. van Influence of transition diets on occurence of subclinical laminitis in Holstein dairy cows. J. Dairy Sci. 87, Espinasse, J., Savey, M., Thorley, C.M., Toussaint Raven, E. and Weaver, A.D Atlas en couleur des affections du pied des bovins et des ovins. Maisons- Alfort, France: Editions du Point Vétérinaire. 48 pp. Farm Animal Welfare Council Report on Welfare of Dairy Cattle. FAWC. Tolworth, United Kingdom. Frankena, K., Keulen K.A.S. van, Noordhuizen, J.P.T.M., Noordhuizen-Stassen, E.N., Dundelach, J., De Jong, D.J. and Saedt, I Cross sectional study into prevalence and risk indicators of digital haemorrhages in female dairy calves. 60

13 Claw Health in Dairy Cows in the Netherlands Prev. Vet. Med. 14: Green, L.E., Hedges, V.J., Schukken, Y.H., Blowey, R.W. and Packington, A.J The impact of clinical lameness on the milk yield of dairy cows. J. Dairy Sci. 85, Greenough, P.R An illustrated compendium of bovine lameness. Part 2. Mod. Vet. Pract. 68, Hedges, V.J., Blowey, R.D., Packington, A.J., O'Gallaghan, C.J. and Green, L.E A longitudinal field trial of the effect of biotin on lameness in dairy cows. J. Dairy Sci. 84, Hernandez, J.A., Garbarino, E.J., Shearer, J.K., Risco, C.A. and Thatcher, W.W Use of a locomotion scoring system for diagnosis of lameness, and effect of lameness, on reproductive performance in postpartum Holstein cows. Proc. of a SVEPM meeting held at Nairn, Iverness, Scotland, April, 2005, pp: Livesey, C.T., Metcalf, J.A. and Laven, C.A Effect of concentrate composition and cubicle bedding on the development of hoof haemorrhages in Holstein heifers after calving. Vet. Rec. 152, Manske, T., Hultgren, J. and Bergsten, C. 2002a. The effect of claw trimming on the hoof health of Swedish dairy cattle. Prev. Vet. Med. 54, Manske, T., Hultgren, J. and Bergsten, C. 2002b. Prevalence and interrelationships of hoof lesions and lameness in Swedish dairy cows. Prev. Vet. Med. 54, Mason, F.J. and Leaver, J.D The influence of concentrate amount on locomotion and clinical lameness in dairy cattle. Anim. Prod. 47, Melendez, P., Bartolome, J., Archbald, L.F. and Donovan, A The association between lameness, ovarian cysts and fertility in lactating dairy cows. Theriogenol. 59, Mills, L.L., Leach, D.H., Smart, M.E. and Greenough, P.R A system for the recording of clinical data as an aid in the diagnosis of bovine digital disease. Can. Vet. J. 27, MLWIN User s Guide to MLWIN, Version 2.0, Centre for Multilevel modelling, Institute of Education, 20 Bedford Way, London WC 1H OAL, UK. Murray, R.D., Downham, D.Y., Clarkson, M.J., Faull, W.B., Hughes, J.W., Manson, F.J., Merrit, J.B., Russel, W.B., Sutherst, J.E. and Ward, W.R Epidemiology of lameness in dairy cattle: description and analysis of foot lesions. Vet. Rec. 138, Noordhuizen, J.P.T.M Veterinary Herd Health and Production Advisory Programmes for Dairy Farms: Development and Prospects. Cattle Practice. 11,

14 Chapter 12 Peterse, D.J Judgement of bovine claws by the occurrence of sole lesions. Ph.D. Diss. State Univ. Utrecht, The Netherlands. RDA (Dutch Advisory Board of Animal Welfare & Health Affairs) Inventory of targets for housing and management of cattle from an age of 6 month onwards (in Dutch). The Hague, The Netherlands. Schukken, Y.H., Grohn, Y.T., McDermott, B. and McDermott, J.J Analysis of correlated discrete observations: background, examples and solutions. Prev. Vet. Med. 59, Snijders T. and Bosker, R Multi Level Analysis: An introduction to Basic and Advanced Multilevel Modelling. SAGE Publications, London, U.K. Sogstad, A.M., Fjeldaas, T., Østerås, O. and Forshell, K. P Prevalence of claw lesions in Norwegian cattle housed in tie stalls and free stalls. Prev. Vet. Med. 70, Somers, J.G.J.C., Frankena, K., Noordhuizen-Stassen, E.N. and Metz, J.H.M Prevalence of claw disorders in Dutch dairy cows exposed to several floor systems. J. Dairy Sci. 86, Sprecher, D.J., Hostetler, D.E. and Kaneene, J.B A lameness scoring system that uses posture and gait to predict dairy cattle reproductive performance. Theriogenol. 47, Vaarst, M., Hindhedde, J. and Enevoldsen, C Sole disorders in conventionally managed and organic dairy herds using different housing systems. J. Dairy Res. 65, Warnick, L.D., Janssen, D., Guard, C.L. and Gröhn, Y.Y The effect of lameness on milk production in dairy cows. J. Dairy Sci. 84, Weaver, A.D Claw trimming: what farmers think! Cattle practice 5, Weaver, A.D Lameness. In: Andrews AH. (Eds) The Health of Dairy Cattle. Blackwell Science, Oxford, United Kingdom, pp.: Webster, A.J.F Animal Welfare: a Cool Eye towards Eden: a Constructive Approach to the Man s Domination over Animals. Blackwell Science, Oxford, UK. Webster, A.J.F Effects of housing practices on the development of foot lesions in dairy heifers in early lactation. Vet. Rec. 151: Whay, H.R., Waterman, A.E. and Webster, J.F Associations between locomotion, claw lesions and nociceptive threshold in dairy heifers during the peri-partum period. Vet. J. 154, Whay, H.R Locomotion scoring and lameness detection in dairy cattle. In Practice, Sept. 2002: