Implications
Body condition scoring is a widely used tool for assessing subcutaneous fat appositions to detect metabolic status and health risks like subacute ketosis or fertility problems in dairy cows. The scoring process contains visual and palpatory assessment and is therefore dependent on subjective impressions of the observer. Owing to its subjectivity, validity of body condition scoring is sometimes questioned. The current study analysed two body condition scoring systems regarding their validity. Further it evaluated the effect of including the cow’s overall appearance into the assessment.
Introduction
Body condition, respectively the amount of body fat reserves, vary through the lactation cycle of the cow in a predictable way (Friggens, Reference Friggens2003). After calving, energy output exceeds energy input, resulting in a negative energy balance. Cows usually react with a mobilization of body fat reserves (Gross et al., Reference Gross, van Dorland, Bruckmaier and Schwarz2011), which comes along with changes in blood plasma metabolites, for example an increase of non-esterified fatty acids and β-hydroxy-butyrate and a decrease of triglycerides (Leiber et al., Reference Leiber, Hochstrasser, Wettstein and Kreuzer2011). Fat mobilization post calving is necessary for productivity, but can also impair health and reproduction. Therefore, monitoring the energetic status and the quantity of the cow’s fat storage is recommended (Bewley and Schutz, Reference Bewley and Schutz2008). Subcutaneous adipose tissue seems to represent the proportional change of total body fat weight within the different physiological phases of the lactation cycle and therefore serves as a good indicator for major changes in total fat reserves (Butler-Hogg et al., Reference Butler-Hogg, Wood and Bines1985).
A tool for monitoring the fat reserves of the cow is the body condition scoring. Tail head, pelvic and lumbar area of the cow were identified as adequate regions for assessing the adipose tissue shifts via body condition scoring (Edmonson et al., Reference Edmonson, Lean, Weaver, Farver and Webster1989; Ferguson et al., Reference Ferguson, Galligan and Thomsen1994). Several scoring systems exist. The decision tree-based system developed by Ferguson et al. (Reference Ferguson, Galligan and Thomsen1994) is widely used (Kristensen et al., Reference Kristensen, Dueholm, Vink, Andersen, Jakobsen, Illum-Nielsen, Petersen and Enevoldsen2006) and most closely approximates an international standard (Bewley and Schutz, Reference Bewley and Schutz2008). The flow chart published by Elanco (1997) containing the Ferguson system directs the user to examine certain anatomical sites of the tail head, pelvic and loin region in a given order which then leads to the appropriate score. The scale of the scoring system reaches from 2 to 5 points with increment units of 0.25. Body condition score (BCS) benchmarks for the particular time of the lactation cycle exist. Roche et al. (Reference Roche, Friggens, Kay, Fisher, Stafford and Berry2009) defined the optimal BCS at calving between 3 and 3.25. Even if several studies showed the usefulness of body condition scoring for assessing the energy balance status of the cow (Wright and Russel, Reference Wright and Russel1984; Butler-Hogg et al., Reference Butler-Hogg, Wood and Bines1985), it also has its limitations. Owing to the assessment, which is based on personal perception, repeatability can be a problem (Ferguson et al., Reference Ferguson, Galligan and Thomsen1994; Kristensen et al., Reference Kristensen, Dueholm, Vink, Andersen, Jakobsen, Illum-Nielsen, Petersen and Enevoldsen2006).
The widely used BCS assessment advices by Edmonson et al. (Reference Edmonson, Lean, Weaver, Farver and Webster1989) and Ferguson et al. (Reference Ferguson, Galligan and Thomsen1994) were both developed with Holstein cows in the United States. Cattle type differences in body fat distribution and body shape have to be considered when these BCS systems are used for other breeds (Wright and Russel, Reference Wright and Russel1984; Rastani et al., Reference Rastani, Andrew, Zinn and Sniffen2001). In mountainous regions like Switzerland, dual purpose cattle are often kept for milk production. If body scoring systems developed for dairy breeds are also used for dual purpose cattle, their application might be inaccurate. Body condition scoring may cause particular problems in breeds which are less homogenous as it is the case in the milk focused dual-purpose Swiss Brown Cattle, which has a varying hybridization of milk-type Brown Swiss and of the more meaty Original Braunvieh. This might require the inclusion of the cows overall appearance into the decision-making process in order to avoid mistakes caused by deviating body fat distribution. On the other hand this overall appearance of the cow might be susceptible for erroneous assessments. Therefore the question might be whether a higher formularization leads to better repeatability.
The aim of this study was to analyse how a formularized strict hierarchic system – or to say a more standardized and guided scoring system with less opportunities for subjective flexibility – performs compared to the common and widely used body condition scoring system with its strong influence of the assessors’ perception of the cows’ overall appearance. Both approaches were validated by ultrasonic back fat thickness (BFT) measurements. A further aim was to identify particularly problematic body regions for the BCS by comparing the two approaches.
Material, methods and animals
Farms, animals and housing
The data used in this experiment were collected within 6 weeks on 40 organic and low input farms visited once during November and December 2009. All dairy cows included in the study were Swiss Brown Cattle with a varying hybridization of Brown Swiss resulting in a milk-type dual purpose breed. The average size of the herds was 25 animals (minimum 12, maximum 66, ±11.1). In 31 farms the animals were kept in a free stall barn, nine farms had a tie-stall housing system. The average milk yield per year was 6739 kg (minimum 4797 kg, maximum 8359 kg, ±716). All farms except two were organic (as defined in EC, 1991 and 1999; BioSuisse, 2014). All animal-related procedures were in compliance with the Swiss animal welfare act and the animal welfare ordinance (Authorization Nr. 75′541). Lactating and dry cows were measured. For the measurement the cows in the free stall barn needed to be locked in head gates at the feed manger. In the study, 1164 animals were included. After excluding missing values, technical measuring errors or cows of other breeds, 1111 animals were left for statistical analysis.
Methods of assessing the body condition
Two methods were used to assess or measure the body condition of the cow. The thereby generated scores are called independent BCS (iBCS) and dependent BCS (dBCS). Both systems are based on the hierarchically constructed body condition scoring system developed by Ferguson et al. (Reference Ferguson, Galligan and Thomsen1994), published as a flowchart by Elanco (1997) and adapted to breed-independent scoring by Ivemeyer et al. (Reference Ivemeyer, Klocke and Spengler Neff2006) (Figure 1). The system has a scale from 2 to 5 with increment units of 0.25. The method involves visual assessment as well as palpation of specific body parts of the loin, pelvic and tail head region. The flow chart directs the assessor to view the defined body regions in a fixed order. The assessment grades for the respective body regions appear as verbal codes, which are defined and thus guide the scoring process in an unambiguous way (Figure 1). The first step is to decide whether the pelvic area or rather the line from the hooks to the trochanter region to the pin bones is angular (V-shaped) or blunt (U-shaped). This decision step determines if the animals are classified ⩽3.00 or ⩾3.25 and therefore predefines which body regions have to be assessed next. When animals are classified ⩽3.00, the hooks are the next region to be assessed. If they are categorized to be round, the assessing process finishes at this point with a given score for the cow of 3.00. If the hooks show an angular form, the pin bone region has to be assessed next and so on (see Figure 1). If the animals are classified ⩾3.25, the ligaments have to be assessed subsequently. When they are both clearly visible, the animal is scored with a 3.25 and the assessment process finishes. If not, the assessing process continues as shown in Figure 1.
Body condition scoring system of Ivemeyer et al. (Reference Ivemeyer, Klocke and Spengler Neff2006), modified from the system developed by Ferguson et al. (Reference Ferguson, Galligan and Thomsen1994) and Elanco (1997).
Although the two scoring systems are based on the same scheme, their assessment process differed. For the iBCS, all body regions mentioned in the flowchart (Figure 1) were assessed independently, without generating a score but generating an unambiguous verbal code for each body region. Subsequently, the final iBCS was generated outside the barn by applying the verbal codes a posteriori within the decision tree structure (Figure 1) using Microsoft Excel. This method was developed to find out if a very strict use of the decision tree without consideration of the cows’ overall appearance can lead to improved BCS assessments.
For the dBCS, which was in line with common practice, the score was given directly in the barn. The decision tree structure was applied, but the overall appearance of the cow was used to correct the scores during the decision-making process. During the scoring process an ongoing reconciliation between the cow’s overall appearance and the scoring process on the base of the flow chart was applied. The ‘overall appearance’ of the cow consists of all body regions listed in the flow chart. The system apparently has a greater subjective impact due to the permitted influence of the overall impression and the resulting range of modification. The score therefore is more dependent on the assessment of the observer and that is why we called it dBCS. This method was developed based on Elanco (1997) and Ivemeyer et al. (Reference Ivemeyer, Klocke and Spengler Neff2006).
Body condition scoring and measuring of BFT (as described below) were conducted by two trained persons. In advance of the study, three preliminary tests – calculations of κ-coefficients based on Grouven et al. (Reference Grouven, Bender, Ziegler and Lange2007) – had been conducted to evaluate inter-observer reliability of body condition scoring. κ-coefficients of 0.55, 0.67 and 0.75 demonstrated a quite good inter-observer agreement. The body region-specific grades for the iBCS were defined and BFT was measured in a first independent step. Later on the same day both assessors applied the dBCS system together without being able to see the grades from the iBCS and the BFT values.
Measuring the BFT
The validity of iBCS and dBCS was tested regarding their ability to predict BFT which was measured by ultrasound applying the method of Staufenbiel (Reference Staufenbiel1992). The examination point was located in the sacral region on an imaginary line between the pin and the hook at the right side of the cow one hand’s width (about 10 cm) away from the pin. The back fat measurement includes the measurement of the skin which is 5 to 6 mm thick in the sacral area. The fat layer is located between the skin and the profound fascia. The profound fascia separates the back fat from the gluteal muscle (Schröder and Staufenbiel, Reference Schröder and Staufenbiel2006). For the measurement the Tringa Linear ultrasound system (Esaote, Genova, Italy) was used. As contact agent 80% ethanol was applied to the unshaven coat.
Statistical analysis
The BFT values were transformed (ln) before calculating the linear regression models to achieve a Gaussian distribution. In the regression models the BFT-values represented the dependent variable whereas the dBCS and iBCS in each case served as independent variables. The linear regressions were computed with the statistical software R 2.15 (R Core Team, 2012). The Gaussian distribution of the residues of the fitted values was tested by a Residual v. Fitted and a QQ Plot. Boxplots showing BFT in dependency on dBCS and iBCS were generated with SigmaPlot 10.0 (Systat Software Inc).
Identification of relevant body regions
To investigate the relevant body regions influencing the discrepancies between the two BCS systems, iBCS and dBCS were compared for each individual animal and the respective deviant decision point or body region was identified (Table 1). The body region (or decision point), at which the decision tree paths (according to Figure 1) for iBCS and dBCS separated, was tagged for each individual cow. For example, the hook bones were identified as a deviation point, when a cow showed an iBCS of 2.5 and a dBCS of 3 or reverse. For iBCS 2.25 and dBCS 2.75 or vice versa, the pin bone region would have been tagged as the deviation point. The number of deviations was set in relation to the number of animals which passed the respective decision point based on iBCS data.
Deviations between two different body condition scores of dairy cattle at different decision points in the flow chart
1 Arrows indicate that assigning the respective score at the given decision point leads automatically to the assessment of the next decision point in the indicated numerical direction.
2 dependent body condition score (dBCS). Overall appearance of the cow was used to correct the scores during the decision making process.
3 independent body condition score (iBCS). This system is characterized by a very strict use of the decision tree structure without consideration of the cows’ overall appearance.
4 These body regions occur twice in the flowchart, but are only assessed once, either in category ⩽3.00 or ⩾3.25.
Results
The results of the linear regression of BFT on dBCS and iBCS, respectively, are displayed in Table 2. The dBCS was able to explain the BFT better; 67.3% of the variance of the BFT were explained by this scoring system (Table 2). By contrast, the iBCS could only explain 47.3% of the BFT variance. Although the accuracy to predict BFT differed between dBCS and iBCS, the boxplots (Figures 2 and 3) reveal that there exists a certain coherence for both BCS approaches with the BFT.
Back fat thickness (BFT) distribution within the categories of the dependent body condition score (dBCS) in Swiss Brown Cattle (n=1111).
Back fat thickness (BFT) distribution within the categories of the independent body condition score (iBCS) in Swiss Brown Cattle (n=1111).
Linear regression of two different body condition scores on back fat thickness in Swiss Brown Cattle (n=1111)
1 dependent body condition score (dBCS). Overall appearance of the cow was used to correct the scores during the decision making process.
2 independent body condition score (iBCS). This system is characterized by a very strict use of the decision tree structure without consideration of the cows’ overall appearance.
Table 1 shows the absolute and relative numbers of cows with deviating iBCS and dBCS scores at the different points of the decision tree. A total of 232 animals were assessed differently when it came to decide whether the pelvis line appears as a U or V. The majority of these animals had a higher iBCS than dBCS, indicating that cows having received the score ‘U’ at the pelvis line, were downgraded by their overall appearance in the dBCS. Also the shape of the hook bones was difficult to assess resulting in 19% deviating cases. Here, dBCS was mostly higher than iBCS. The highest discrepancy between iBCS and dBCS was found in the pin bone region when scoring cows at ⩽2.75. The decision point was passed by 429 animals; in more than 60% dBCS and iBCS differed. In most of these cases the dBCS score was higher than the iBCS score, indicating that a low score given at the pin bones was corrected upwards in dBCS by the overall appearance of the cow. Also assessing the ligaments caused considerable deviations, but here the direction of deviation was not as clear as in the other problematic regions. Table 1 shows that 33% of the animals received deviating iBCS and dBCS at the decision point for score 4.0 (trochanter still visible or not). But due to the low number of animals passing this decision point, the meaningfulness of the result is questionable. In the entire data set, only 31.3% of the animals were classified with the same dBCS and iBCS (Figure 4).
Distribution of the individual differences of the dependent and independent body condition score (dBCS and iBCS) in Swiss Brown Cattle (n=1111).
Discussion
Significance of the overall appearance of the cow
The more open and common dBCS system revealed clearly closer matches with the BFT than the iBCS system which strictly followed the paths of the decision tree. The dBCS method included the overall cow’s appearance. Thus, this approach was apparently more open to subjective impressions of the assessor. On the other hand it might be argued that the assessment of a particular part of the body is not less subjective than that of a whole animal. The iBCS and the dBCS system of the current study might be characterized as a more particularistic (the former) v a more general (the latter) approach. It can be concluded that, given a certain influence of subjective impressions on the decision process, the result gets improved if subjectivity is connected to the general impression rather than only to the particular scorings of separated body regions. For animal behaviour studies, also Wemelsfelder et al. (Reference Wemelsfelder, Hunter, Mendl and Lawrence2001) concluded that a qualitative (and thus ‘subjective’) assessment of animals’ behaviour achieves a higher validity if it is directed to the whole animal rather than to separate traits.
According to Elanco (1997) not all cows fit into the established Ferguson system. In these cases they also recommend an adjustment of the final BCS based on observations of all designated body regions. This approach coincides with the results of the present study. If the overall impression of the cow is included in the assessment process, the decision tree-based system provides valid information about the body reserves. It can be concluded that taking into account the overall appearance of the cow is a necessary step in the assessments of the body condition.
Relevant body regions
Through the analysis of the discrepancies between dBCS and iBCS it was found that specific body regions were particularly difficult to assess and therefore their suitability for a body condition scoring system is questionable. Most deviating results were found in the pin bone region, which was passed by 429 animals, namely all cows receiving iBCS scores of ⩽2.75. Within the deviating scores, more than half of the animals received an iBCS of ⩽2.25 (data not shown) and therefore have to be characterized as thin animals, whereas their dBCS was higher. Why the pin bone region was particularly difficult to assess might have different reasons. One cause might be the fact that the assessor has to differentiate between three levels of fat apposition and it is difficult to distinguish between those levels and to clearly assign the amount of fat apposition to the adequate category. Jilg and Weinberg (Reference Jilg and Weinberg1998) also stated that among others the pin bone region is particularly difficult to assess due to the muscles in this area. Their observations were made on the dual purpose breed Fleckvieh, hence this characteristic might vary between different cattle types.
According to Ferguson et al. (Reference Ferguson, Galligan and Thomsen1994), a 0.25-increment accuracy can only be reached between the scores 2.5 and 4.0, while on the upper or lower end of the scale the system loses accuracy and only 0.5 increments are possible. However, BCS below 2.5 and above 4.0 may rather be cases where animal welfare is generally impaired. Such a diagnosis might even not need an accuracy of 0.25 increments, which is however necessary for differentiated body condition curves in normal healthy cows (cf. Vasseur et al., Reference Vasseur, Gibbons, Rushen and de Passillé2013). The weakness of the system for thinner cows may be partly explained by the difficulty to assess the pin bone region correctly. However, also the pelvis line, the hook bones and the ligaments frequently received different iBCS and dBCS assessments in the current study. Thus, the problem is not restricted to thin cows.
Another point which has to be analysed is the possibility of a malfunction of the hierarchic order when it comes to assess the problematic body regions. An erroneous score in one body region would not be too dramatic if it would be corrected by other regions in an open system like that of Edmonson et al. (Reference Edmonson, Lean, Weaver, Farver and Webster1989) or by the whole animals’ appearance like in the current study (dBCS system). By contrast, in a very strict system like iBCS such a regional assessment error cannot be corrected, but may even cause larger errors, because it leads the assessor to the next assessment step within the wrong body region.
Applicability of the Ferguson system to different breeds and ages
Even though in this trial no comparison between different breeds regarding BCS systems took place, some studies indicate that breed related differences might have played a role for the weak performance of the iBCS system. The hierarchic body condition scoring system of Ferguson et al. (Reference Ferguson, Galligan and Thomsen1994) is based on Holstein regularities. If the subcutaneous fat distribution varies between different cattle types and breeds (Truscott et al., Reference Truscott, Wood and Denny1983; Wright and Russel, Reference Wright and Russel1984; Jilg and Weinberg, Reference Jilg and Weinberg1998; Rastani et al., Reference Rastani, Andrew, Zinn and Sniffen2001), it is quite reasonable to assume that hierarchically structured body condition scoring systems developed for Holstein cows entail the risk of an erroneous assessment when applied strictly without consideration of the cows overall appearance. Depending on their amount of hybridization with Brown Swiss, the cattle used in this study were to a greater or lesser extent dairy-type. Therefore the animals were heterogeneous regarding their exterior, musculature and probably also regarding their way of fat storage. Also Mösenfechtel et al. (Reference Mösenfechtel, Eigenmann, Wanner and Rüsch2000) observed that Swiss Brown Cattle vary widely regarding their musculature and that it is therefore more difficult to assess their subcutaneous fat amount than the fat appositions on Holstein cows.
Another point which exacerbates a consistent scoring of the animals is the age-related fat modification. Dairy cows within the first two lactations are still in process of growth; within the first lactation temporarily fat becomes even mobilized for the buildup of protein (Klawuhn and Staufenbiel, Reference Klawuhn and Staufenbiel1997). Hence it could be that the breed and age-related differences in fat storage mechanisms as well as the heterogeneity within the breed aggravated the assessing process in this study. The dBCS system with its more open approach was obviously more capable to mitigate these difficulties.
Conclusion
It can be concluded that the dBCS method which includes the overall appearance of the cow into the flowchart system, performs with a much higher accuracy than the strictly decision tree-based iBCS method. This implies that the flowchart system, if it includes a holistic view of the whole animals’ appearance gives good results even if different cattle types are considered. By contrast, the strictly applied decision tree bears no possibility to correct erroneous decisions, but it is rather at risk to even enhance these errors. Additionally this system can barely account for differences in breeds and individual types. Thus, a formularization of the body condition scoring does not lead to a higher accuracy of the values.
Acknowledgements
The authors gratefully acknowledge financial support from the European Community under the 7th framework project LowInputBreeds, FP7 project No. KBBE 222 632. Special thanks to all the dairy producers who took part in the study.
