©2008 Poultry Science Association, Inc. Validation of a Three-Point Gait-Scoring System for Field Assessment of Walking Ability of Commercial Broilers A. B. Webster,*1 B. D. Fairchild,* T. S. Cummings,† and P. A. Stayer‡ *Department of Poultry Science, University of Georgia, Athens 30602; †College of Veterinary Medicine, PO Box 6100, Mississippi State, MS 39762; and ‡Sanderson Farms, PO Box 988, Laurel, MS 39441 Primary Audience: Researchers, Broiler Producers, Animal Welfare Auditors SUMMARY A 3-point gait-scoring system used to evaluate broiler walking ability in welfare audits of commercial flocks in the United States was compared with the 6-point Kestin system. In 2 university trials, market-age broilers of 2 commercial varieties were gait-scored by 2 observers for each scoring system. Subsamples of birds were rescored and evaluated in latency-to-lie (LTL) tests. Too few birds had significant walking difficulties in these trials to allow for a good comparison of the 2 scoring systems across all the gait score categories, but the data were encouraging despite sampling limitations. There was a significant association between the 2 systems, and both had substantial between-observer agreement. Both scoring systems had significant correlations with LTL, but the variation of LTL was too high to give gait score or LTL much predictive value for each other. In the field observations, 2 teams of observers scored broilers 47 to 61 d of age on 5 commercial farms each. Two pairs of observers on each team scored the same birds, each pair using the 3-point system or the Kestin system. Broilers with walking problems were oversampled to obtain an adequate number in each gait score category. Weighted κ statistics showed substantial between-observer agreement in each system but more so in the 3-point system, suggesting that the application of the 3-point system was more consistent between observers. Spearman correlations between 3-point and Kestin scores for individual birds indicated good correspondence between the 2 systems. The simplicity of the 3-point gait-scoring system appears to facilitate between-observer agreement, making it preferable to more complex systems for use in commercial animal welfare audits. The correspondence between the gait-scoring systems validates the 3-point system in light of the 6-point Kestin system. Key words: broiler, gait score, walking ability, animal welfare audit 2008 J. Appl. Poult. Res. 17:529–539 doi:10.3382/japr.2008-00013 DESCRIPTION OF PROBLEM Modern strains of broilers are susceptible to leg problems compared with unselected strains of chickens. Under some circumstances, inci1 Corresponding author: [email protected] dence of leg disorder can be high, associated with elevated rates of mortality [1–3]. Impaired gait has also been found to be linked to pain or discomfort [4, 5]. Observations such as these JAPR: Field Report 530 have led many people to expect that commercial broiler flocks routinely have large percentages of birds with walking difficulty and have fueled concern that the broiler industry is living with a level of bird welfare that is unacceptably poor. Fortunately, there is growing evidence that environmental factors can have a considerable influence on incidence of leg problems in broilers and that leg abnormalities can be kept to low levels under good management [6, 7]. This evidence has come in part by means of evaluations of walking ability of birds in commercial flocks. Because animal welfare assurance has become an important aspect of the marketing of poultry products to food retailers, the need has arisen for an appropriate method to measure the incidence of leg disorder in commercial broiler flocks. Visual inspection of walking ability (i.e., gait scoring) offers the advantage of allowing noninvasive evaluation of large numbers of birds in a short period of time. A valid and feasible method for gait scoring broilers on commercial farms is needed so that auditors can apply a consistent scoring method and minimize inequities in the assessment of different farms. The method should allow a broiler house to be evaluated in a reasonable length of time. Gaitscoring systems have been developed to assess leg condition in breeder stocks and have been adapted for evaluation of broilers [8, 9]. The systems used by these authors define 6 categories of walking abnormality on an ordinal scale of severity. The differences between categories are subtle enough to make gait scoring perhaps more difficult and slower than necessary in a commercial production environment. Other studies used fewer categories to score walking ability but did not evaluate the systems in a context comparable to the real-time observations necessary for an animal welfare audit or for comprehensive sampling of farms as part of a company animal welfare program [6, 10]. We have developed a real-time, 3-point (3PT) gaitscoring system that is currently being used on commercial farms in the United States. The 3PT system seeks to identify broilers as having no impairment of walking ability (score 0), having obvious impairment but still ambulatory (score 1), and having severe impairment and not able to walk without great difficulty (score 2). The 3PT system needs to be validated for reliability and comparability with other scoring systems. Motivation to walk can influence evident walking ability, and even pain-related behavior such as limping can be overridden by attentional mechanisms [11]. Understanding this, some researchers have sought alternate methods of assessing leg weakness. Weeks et al. [12] described the latency-to-lie (LTL) test, wherein broilers are made to stand in shallow water. Being a novel and aversive stimulus, the water is thought to produce avoidance motivation strong enough to induce broilers to stand as long as possible. In LTL tests, birds with good walking ability are able to stand a long time, whereas those with poorer gait scores sit down sooner. Berg and Sanotra [13] modified the LTL test with the intent of making it easier to use in commercial broiler houses. In the present study, the 3PT gait-scoring system was compared with adaptations of the 6-point Kestin gait-scoring system [8, 9] in 2 university trials and in 10 commercial broiler flocks. Subsamples of birds in the university trials were rescored to assess retest reliability and were evaluated in LTL tests to compare the different scoring systems against a common standard. MATERIALS AND METHODS University Trials In each of 2 trials, straight run broiler chicks of 2 varieties were obtained from commercial hatcheries and housed in an environmentally controlled room on the University of Georgia (UGA) Poultry Research Farm in separate floor pens bedded with pine shavings. One hundred fifty Cobb 500 [14] and 200 Ross 708 [15] chicks were placed in trial 1, and 200 chicks of each variety were placed in trial 2. Space allocation was 0.7 ft2 per bird. The photoperiod was 24 h of light from d 1 to 7 and 20L:4D from d 8 to 42 (or d 45, trial 2), when the trials ended. Standard UGA practices were followed for flock management, and UGA-formulated broiler feed and water were provided ad libitum. Body weights were obtained at the beginning and at the end of the trial. Gait-scoring observations were conducted at the end of each trial. A total of 681 broilers Webster et al.: THREE-POINT GAIT-SCORING SYSTEM were observed in the 2 trials. This number is less than the number housed minus mortality, because some unscored broilers escaped past the divider separating unscored and scored birds and were lost among the scored birds. Two observers scored each bird according to the 3PT gait-scoring system, and 2 other observers simultaneously scored the same birds according to a 6-point gait-scoring system. The sex of each broiler was noted. The observers consisted of poultry science professors, technicians, graduate students, and undergraduate students. Although all were familiar with broiler chickens, most had no prior experience with gait scoring and had to be trained. During training, an experienced individual explained the categories of the different gait-scoring systems, which were provided to trainees in written form, and then worked with the trainees to score live birds according to the scoring system they were assigned until they became proficient. In trial 1, all observers used stopwatches to record the time it took to determine the gait score of each bird. Although the original purpose was to compare the length of time it took to score birds using the different systems, the time data prompted us to switch from the 6-point modified Kestin system [9] used in trial 1 to the 6-point Kestin system [8] in trial 2. The observers in trial 1 gravitated toward the Kestin system by overlooking the time criteria built into the modified Kestin system. It was thought that animal welfare auditors and other gait scorers working in commercial situations would have the same tendency, making the modified system impractical for field use. The gait-scoring systems thus used in this study are outlined in Table 1. During gait scoring, the observers stood outside the pen of the birds where they could avoid viewing the broilers from above. A handler in the pen used folding wire catch pens to clear space for birds to walk freely without being impeded by other birds. The handler separated broilers from the flock and encouraged them to walk in the open area in the direction of other birds. This encouragement consisted, as necessary, of approaching the bird from behind, tapping the catch pen near the bird, or gently prodding the bird with a length of polyvinyl chloride tubing or a foot. Although the effort made to get birds to walk was not rigorously standardized, there was no advantage 531 in trying to excessively motivate them, because fear or novelty can cause a chicken to ignore a painful leg condition and run or walk normally [11], which would decrease the numbers of birds recorded in the poorer gait score categories and compromise comparison of the gait-scoring systems. When a broiler without evident leg abnormality refused to walk, the handler left it where it was and went on to other birds. The reluctant bird usually could be induced to walk after other birds had traveled past it. In each trial, up to 10 broilers of each variety representing each category of the 6-point system were identified for subsequent rescoring of their walking ability and for participation in LTL tests. Only those birds were selected for which both observers using each system agreed on the gait score (total, 90 birds). This was done to provide a common basis for subsequent analyses comparing the gait-scoring systems. Broilers with gait abnormalities were oversampled from the population to obtain a balanced distribution of birds across the score categories as much as possible. The gait rescoring and LTL tests were carried out separately. The second gait-scoring test was conducted in the same manner as the first, except that it involved only the identified birds. The LTL tests were done in the same room that housed the broilers, using a procedure similar to that of Berg and Sanotra [13]. For these tests, the birds were removed from their home pen and placed in tubs containing an inch (2.5 cm) of lukewarm water. The time it took for the broiler to first attempt to sit down was recorded, at which time the bird was removed and replaced with another bird. A test was terminated after 600 s if the bird stood that length of time. Three tubs were monitored at 1 time by a single observer. The tubs were surrounded by cardboard walls high enough to prevent test birds from seeing other broilers. In trial 1, gait scoring and LTL tests for Ross 708 broilers took place on d 39 and gait rescoring on d 40. For Cobb 500 broilers, all observations were done on d 41. In trial 2, gait scoring of both broiler varieties was carried out on d 45 and gait rescoring and LTL tests on d 46. Although 2 broiler varieties were used to reinforce the generality of the study, the intent was not to make a direct comparison of the 2 varieties. The age of the parent flocks was not JAPR: Field Report controlled, and the 2 stocks were grown on an identical lighting and nutritional program, which may not have been the optimum for one or the other variety. The walking ability observed may not have reflected the optimum leg condition that could be achieved with either stock. Gait score data will not be presented separately for the broiler varieties. Complete lameness 5 Field Observations Severe impairment of function but still capable of walking 4 No detectable abnormality. Slight defect in walking ability that is difficult to define. An uneven gait. Definite and identifiable defect in gait but with little hindrance of movement. Obvious gait defect, which affects ability to move about (e.g., limp, jerky or unsteady stride, or splaying of a leg). Severe gait defect. Capable of walking with difficulty when driven or strongly motivated. Squats at first opportunity. Bird cannot walk. May shuffle on shanks or hocks with assistance of wings. Bird can walk at least 5 ft with a balanced gait. Bird may appear ungainly but with little effect on function. Bird can walk at least 5 ft but with a clear limp or decidedly awkward gait. Bird will not walk 5 ft. May shuffle on shanks or hocks with assistance of wings. 3-point gait-scoring system 0 None 1 Obvious impairment 2 Severe impairment Kestin gait-scoring system (adapted from Kestin et al. [8]) 0 None 1 Detectable but unidentifiable abnormality 2 Identifiable abnormality that has little effect on overall function 3 Identifiable abnormality that impairs function Criterion Degree of impairment Gait score Table 1. Description of gait-scoring systems used to evaluate walking ability of market-age broilers 532 Observations were made on 10 commercial farms contracted to a broiler complex in Mississippi over 2 d in the month of September. Only 1 house was used on each farm. A total of 1,381 birds was observed. All houses were solid-wall construction and designed for tunnel ventilation. They each held 20,000 to 25,000 broilers on litter. Two teams of observers, comprised of university professors, veterinary medical students from Mississippi State University, and broiler company employees, scored broilers on 5 farms each. The flocks ranged in age from 47 to 61 d of age, and all birds were the Ross 308 variety [15]. Although familiar with broiler chickens, most of the observers had not done gait scoring before the study, so training was conducted first for these individuals. The training consisted of a classroom session to explain the categories of the 2 gait-scoring systems being compared and discuss important aspects of bird catching and handling procedures during gait scoring. This was followed by a session with a commercial flock in which catching and observation procedures were demonstrated and trainees scored birds representing a full range of walking ability according to the system assigned to them (3- or 6-point) until they demonstrated proficiency. In the field observations, 2 observers scored birds using the 3PT system, and 2 others simultaneously scored the same birds using the Kestin system in each commercial flock. The sex of each bird was recorded. The intent was to observe about 20 birds representing each of the 6 Kestin gait score categories for a total of 120 broilers per flock. Catch pens were used to enclose and observe 3 groups of 20 to 30 birds in succession at different locations along the side wall of the house. After each group was gathered, a space of approximately 10 ft was cleared of birds along the wall and out to the first line of drinkers (about 5 ft from the wall). The catch Webster et al.: THREE-POINT GAIT-SCORING SYSTEM pen was opened toward the open space and the birds were encouraged as gently as possible by similar methods as for the university trials to walk out of the pen. Typically, the broilers would leave the enclosure singly or in small groups. The observers stood several feet away from and slightly behind the pen opening so that they could see the birds from the side but not be perceived as an obstacle to broilers attempting to leave the pen. The broilers were gait scored as they left the catch pen, much as would be done in an animal welfare audit. Most of these birds were found to be in the Kestin 0 or 1 category, so each flock was searched in different zones to come up with up to 3 additional pens of 25 to 30 broilers having evident walking abnormalities. These were scored in the same manner as the other birds to make up the numbers of birds in the Kestin 2 to 5 categories as much as possible. Thus, the greater Kestin categories, reflecting greater leg problems, were oversampled in regard to their true incidence. To avoid overly disturbing the flocks, no further effort was made to find lame broilers even if some Kestin categories still lacked the desired number of birds. Data Analysis Various procedures of SAS [16] were employed to carry out data analyses enabling comparison of the 2 gait-scoring systems. The data from the 2 university trials and the 10 commercial flocks were combined, respectively, and analyzed without attempting to partition out trial or farm effects. To evaluate between-observer agreement within each system (i.e., how similarly observers using the same gait-scoring system scored birds), the FREQ procedure of SAS [16] was used to construct frequency tables matching the scores of the 2 observers for every bird and calculate weighted kappa (κ) values and Spearman rank correlations between the 2 sets of scores. The κ value is a measure of the strength of agreement between ratings within a scoring system. Values ranging from 0.4 to 0.6, 0.6 to 0.8, and 0.8 to 1.0 are considered to reflect moderate agreement, substantial agreement, and almost perfect agreement, respectively. The κ value was weighted to compensate for the different number of categories in each gait-scoring system, because it is to be expected, for instance, that a 1-category difference in score between 533 observers would reflect a larger disagreement in the 3PT system than the 6-point system. The same type of analysis was done with the data from the university trials to evaluate withinobserver (i.e., test-retest) agreement using the broilers that were rescored for walking ability. The CORR procedure of SAS [16] was used to calculate Spearman correlations between the LTL times and first gait scores of the selected broilers in the university flocks and to calculate Spearman partial correlations to assess the association remaining between LTL and 1 gait-scoring system after removing the association that could be attributed to the other gait-scoring system. Finally, the 3PT and Kestin systems were compared directly in the university trials and the field observations. The FREQ procedure of SAS [16] was used to produce frequency tables for each of the 4 combinations of observers (i.e., 3PT1-K1, 3PT1-K2, 3PT2-K1, and 3PT2-K2). The output data sets were combined and averaged to produce a composite frequency for each 3PT-Kestin score combination. After rounding the frequencies to the nearest integer, the FREQ procedure was used again to produce the Spearman rank correlation between the ratings of the 2 scoring systems. Kappa values could not be calculated, because the 2 scoring systems had different numbers of categories. Spearman partial correlations adjusting for the effect of sex were calculated in addition to the simple Spearman correlations but are not reported, because the sex effect was found to be negligible. RESULTS AND DISCUSSION University Trials Even though growth rates were high in the university trials, averaging 57 and 63 g/bird per day for Ross 708 and 63 and 67 g/bird per day for Cobb 500 birds in trials 1 and 2, respectively, relatively few broilers evidenced significant walking disability. As a result, almost 90% of the birds were assigned a score of 0 in the 3PT system, and, correspondingly, fewer than 5% of the birds were given Kestin scores of 3 or greater (Table 2). For gait rescoring, the deliberate oversampling of birds with gait abnormalities increased the percentages of birds in the greater score categories somewhat (Table 2), but these categories were still underrepresented for a bal- JAPR: Field Report 534 Table 2. Percentage of birds (mean of observer averages; range in parentheses) in each gait score category for the 3-point and Kestin gait-scoring systems in each test1 Score Study Test Scoring system University trials First gait score 3-point Kestin Gait rescore2 Latency-to-lie 3-point Kestin 3-point Kestin Field observations First gait score 3-point Kestin 0 89.2 (0.6) 73.7 (5.0) 73.9 44.4 299 (28) 341 (38) 63.2 (1.2) 37.0 (0.7) 1 2 9.5 1.2 (1.0) (0.5) 19.6 3.4 (5.7) (1.5) 19.4 6.7 31.7 8.3 176 43 (52) (43) 221 182 (35) (90) 23.5 13.3 (2.2) (0.9) 21.0 11.6 (3.6) (2.2) 3 4 5 — — — 0.3 (0) — 2.2 — 0.7 (0) — 3.9 — 0 (0) — 0 (0) — 9.7 (1.1) 11.5 (1.1) 2.4 (0.7) — 9.4 — 177 (80) — 9.1 (5.0) 1 Latency-to-lie is in seconds to indicate average latencies (SE in parentheses). No measure of variation is indicated, because selection of broilers for rescoring was dependent on agreement of gait scores between observers. The greater-numbered gait score categories were oversampled to maximize their representation in the rescore test. 2 anced comparison of the 2 gait-scoring systems across all levels of walking ability. The numbers of broilers available in the commercial flocks allowed oversampling to achieve better balance of birds across the gait score categories (Table 2), but even so, the relative rarity of birds with serious walking disabilities in the commercial flocks made it not possible to equalize numbers in each category without overly disturbing the flocks. Kappa values indicated substantial betweenobserver agreement within both the 3PT and Kestin gait-scoring systems in the university trials but somewhat more so for the Kestin system (Table 3). Similarly, Spearman rank correlations indicated highly significant positive associations between observers in each scoring system, but the association was stronger between the observers using the Kestin system. The deficient numbers of birds in the greater score categories and the possible effect of specific pairs of observers mitigate against giving too much importance at this point to the apparent difference in between-observer agreement between the 2 gaitscoring systems. Within-observer agreement based on rescoring the gaits of selected birds is a measure of how well individual observers agree with their original gait assessment. Because some time elapsed between the first and second gait-scoring events, the walking performance of the birds itself may have changed, affecting assessment of within-observer reliability. Nonetheless, weighted κ values indicate substantial within-observer agreement, and the strength of agreement was virtually the same for both gait-scoring systems. Likewise, Spearman rank correlations provide evidence of highly significant, positive associations between first and second gait scores, which were equivalent for the 2 scoring systems (Table 3). Within-observer agreement for the Kestin system was close to that reported originally by Kestin et al. [8], even though the second test took place the following day as opposed to later the same day as in Kestin et al. [8]. Garner et al. [9] reported much stronger within-observer reliability with the Kestin system using a shorter test-retest interval and with birds being scored from video clips recorded under standardized circumstances. The results for the LTL tests in the university trials are presented in Table 2. Birds having no gait abnormalities had the longest LTL, as has been shown in other studies [12, 13]. The variability of the measure appeared to increase among birds in the greater gait score categories except for Kestin categories 4 and 5, in which the few birds involved sat immediately. Al- Webster et al.: THREE-POINT GAIT-SCORING SYSTEM 535 Table 3. Between-observer and within-observer agreement and association between latency-to-lie and gait scores for the 3-point and Kestin gait-scoring systems1 Study Measure Agreement 3-point system Kestin system University trials Between-observer agreement2 Weighted κ 0.63 (0.19) 0.62 (P < 0.0001) 0.67 (0.22) 0.68 (P < 0.0001) −0.41 (P < 0.0001) −0.14 (P > 0.05) 0.78 (0.06) 0.84 (P < 0.0001) 0.72 (0.11) 0.71 (P < 0.0001) 0.68 (0.18) 0.69 (P < 0.0001) −0.43 (P < 0.0001) −0.18 (P > 0.05) 0.65 (0.05) 0.74 (P < 0.0001) Spearman correlation Within-observer agreement Weighted κ Spearman correlation Latency-to-lie Spearman correlation Spearman correlation (partial)3 Field observations Between-observer agreement Weighted κ Spearman correlation In parentheses: 95% confidence interval for weighted κ values and probability level for statistical significance for Spearman correlations. 2 Between-observer and within-observer agreement based, respectively, on comparison between observers of first gait scores for all broilers observed and between first and second gait score within observers for rescored broilers. 3 Partial correlation adjusting for the association of the alternate gait-scoring system with latency-to-lie. 1 though the average LTL for broilers scored as 2 in the 3PT system was not zero, the SE was large enough to overlap with zero. The numbers of birds scored in the Kestin system as 4 or 5 or the 3PT system as 2 were small, being 5 and 6 birds, respectively, and a small difference in assignment of birds to gait score categories could produce a large apparent difference in average LTL. Spearman rank correlations of LTL with the 3PT and Kestin gait scores were highly significant but did not indicate a strong association of LTL with either gait-scoring system (Table 3). Spearman partial correlation of LTL with 3PT gait score after adjusting for LTL association with Kestin gait score was low and not statistically significant (Table 3). Likewise, the comparable Spearman partial correlation of LTL with Kestin gait score was also low and not significant. These results indicate that the 2 gait-scoring systems had overlapping associations with LTL so that when the association of 1 with LTL was accounted for, LTL had little remaining association with the other gait-scoring system. Thus, it appears that the 3PT and Kestin gait-scoring systems were similar measures, although not very precise measures, of the underlying factor affecting willingness or ability of the bird to remain standing. The Spearman rank correlation directly comparing the 3PT and Kestin scores from the university trials was highly significant but not particularly strong (r = 0.53, P < 0.0001). Field Observations The field observations allowed a better assessment of between-observer agreement within systems and direct agreement between the 2 gait-scoring systems, because it was possible to sample from large numbers of birds on multiple farms and to take the samples in the same environment and with the methods used in commercial animal welfare audits. Figures 1 and 2 depict the between-observer agreement for the 3PT and Kestin systems, respectively. Weighted κ values indicated substantial between-observer agreement within the 3PT system. Betweenobserver agreement was also substantial for the Kestin system but at a lower level (Table 3). Similarly, Spearman rank correlations between scores for observers within gait-scoring systems were high but more so for the 3PT system. The challenges of observing broilers in commercial environments may have made it easier to apply the 3PT system. Between-observer agreement for the Kestin system as indicated by the Spear- JAPR: Field Report 536 Figure 1. Between-observer agreement: 3-Point gait-scoring system—field observations. man correlation (r = 0.74) was lower than that reported by Garner et al. [9] for observations made under more controlled circumstances (r = 0.89). Figure 3 shows the association between the 3PT and Kestin systems in the field observations. The Spearman rank correlation comparing the 3PT and Kestin gait-scoring systems indicates a good association between the 2 (r = 0.76, P < 0.0001). General Discussion Although the relative infrequency of gait abnormalities found in this study limited our ability to evaluate the gait-scoring systems with equal precision across the full spectrum of score categories, the results allow some general conclusions. Overall, the 3PT and Kestin gait-scoring systems showed good observer reliability, both having substantial between-observer and within-observer agreement. Good observer reliability is essential for an auditing tool. For the more comprehensive data set derived from the field observations, the distribution of 3PT scores across the Kestin scores was not perfectly balanced in that birds scored as Kestin 2 were most frequently given a 3PT score of 0, and broilers scored as Kestin 4 were most frequently scored 3PT 1. Some of this imbalance may be due to Figure 2. Between-observer agreement: Kestin gait-scoring system—field observations. Webster et al.: THREE-POINT GAIT-SCORING SYSTEM 537 Figure 3. Scoring agreement between the 3-Point and Kestin systems—field observations. individual observer differences in interpretation of severity of gait abnormality, such as was seen between observers using the Kestin system in regard to birds scored as 1, 2, or 3. It should be noted that the 3PT-Kestin between-system agreement was as strong as the between-observer agreement within the Kestin system. The criterion in the 3PT system that a broiler be able to walk 5 ft might influence an observer to interpret walking ability differently from an observer using the Kestin system for birds with the more severe walking abnormalities. Because many broilers with serious leg problems can travel 5 ft if sufficiently motivated, it is important when using the 3PT gait-scoring system not to drive the birds. Modest efforts to get the birds to walk, such as described earlier in this report, are sufficient. If a bird has an abnormal gait and must be prompted several times to travel 5 ft, it should be scored as 2 using the 3PT system even if it ultimately goes the distance. This may require some judgment, because it is not unusual to find broilers without obvious gait defects that are reluctant to walk and sit down repeatedly before traveling 5 ft. As a tool for assessing bird well-being, a gait-scoring system should separate birds that are likely to experience discomfort or pain from those that are not. Broilers scored as Kestin 3 have been found to show evidence of discomfort or pain [4, 5], and this score has come to be viewed as the threshold at which animal welfare becomes an issue in terms of leg problems. In the present study on commercial farms, broilers given Kestin 3 scores were preponderantly scored as 3PT 1, with little imbalance of scores between the 2 systems. As such, the 3PT system enabled identification of lame birds with similar sensitivity to the Kestin system, and the 3PT 1 category provided the needed separation of sound birds from those with walking problems. With the 3PT system, flock leg health would be reflected by the percentage of broilers with a score of 0. The remaining percentage would represent birds having diminished welfare to some degree. If this were the only objective of a simplified gait-scoring system, a 2-point system might be preferable to one with 3 categories. However, we believe that a 3PT system better serves the objectives of animal welfare audits and other assessments of the quality of flock management for the following reason. Broilers classified as 2 are crippled. Because these birds should be killed when found, a rigorous culling program would keep the percentage of broilers in this category low. If all is well, birds in the 2 category constitute an insignificant proportion of the flock; however, retention of the 2 category (and the 3PT system) would allow an auditor to document the result of an inadequate culling program or some other factor seriously affecting leg health. There was little imbalance of scores between observers using the 3PT system on commercial farms. The stronger between-observer agreement for the 3PT system may indicate that the JAPR: Field Report 538 simplicity of the 3PT system promotes consistency of scoring in commercial environments where there are limitations in the control of light and bird behavior. Good training is essential for the proper application of a gait-scoring system, including a relatively simple one such as the 3PT system. Butterworth et al. [17] demonstrated that the background or culture of an assessor affects gait-scoring accuracy and that proper training improves scoring accuracy. As animal welfare auditing becomes established in the commercial poultry industry, flock assessments will be done in many cases by professional auditors and others who may not come from backgrounds giving them understanding of poultry welfare. The body type of a commercial broiler produces a stride that may appear ungainly compared with chickens that have not been bred for meat production. Interindividual variation of body conformation in a flock, particularly when sexes are mixed, leads to a range of walking styles even among birds that have good leg health. These broilers must be distinguished from those with impaired welfare caused by a leg or spinal abnormality that affects ability to walk. The welfare impairment may be due to pain or decreased ability to access resources. The critical distinction to be made using the 3PT gait-scoring system, and others, is whether or not a gait is sufficiently abnormal to negatively affect function (i.e., score 0 vs. score 1 in the 3PT system). Gait-scoring training must ensure that auditors are able to make this distinction accurately and with good interauditor reliability so that audits done using the 3PT system at diverse locations and times have comparable meaning. This should include teaching the various types of walking abnormalities that typically occur and how to conduct a gait-scoring assessment. A standard training program would help achieve these objectives. To this end, gait-scoring educational material based on the 3PT system is being developed under the auspices of the American Association of Avian Pathologists for use in training prospective poultry welfare auditors and others [18]. Although video clips are very helpful to illustrate different categories of walking ability, there is no substitute for live bird observations to learn how to gait score. Training should include sessions in flocks of live birds led by an experienced trainer, in which trainees conduct gait-scoring assessments as would be done on commercial farms until they reach a required standard of proficiency. It is important that at least 2 individuals conduct a farm gait-scoring audit, one to handle the birds and the other to record gait scores. The observer must view the broilers from the side, because, as noted by Garner et al. [9], it can be difficult to judge gaits from above, as would be done if the handler was also the observer. CONCLUSIONS AND APPLICATIONS 1. The 3PT and Kestin (6-point) gait-scoring systems had good intersystem agreement when used on commercial farms. 2. Both the 3PT and Kestin gait-scoring systems were found to have substantial observer reliability, but between-observer agreement was somewhat greater for the 3PT system on commercial farms. 3. The simplicity of the 3PT system may promote observer reliability for gait scoring commercial poultry flocks. REFERENCES AND NOTES 1. Sanotra, G. S., J. Damkjer Lund, and K. S. Vestergaard. 2002. Influence of light-dark schedules and stocking density on behaviour, risk of leg problems and occurrence of chronic fear in broilers. Br. Poult. Sci. 43:344–354. 2. Goliomytis, M., E. Panopoulou, and E. Rogdakis. 2003. Growth curves for body weight and major component parts, feed consumption, and mortality of male broiler chickens raised to maturity. Poult. Sci. 82:1061–1068. 3. Havenstein, G. B., P. R. Ferket, and M. A. Qureshi. 2003. Growth, livability, and feed conversion of 1957 versus 2001 broilers when fed representative 1957 and 2001 broiler diets. Poult. Sci. 82:1500–1508. 4. McGeown, D., T. C. Danbury, A. E. Waterman-Pearson, and S. C. Kestin. 1999. Effect of carprofen on lameness in broiler chickens. Vet. Rec. 144:668–671. 5. Danbury, T. C., C. A. Weeks, J. P. Chambers, A. E. Waterman-Pearson, and S. C. Kestin. 2000. Self-selection of the analgesic drug carprofen by lame broiler chickens. Vet. Rec. 146:307–311. 6. Dawkins, M. S., C. A. Donnelly, and T. A. Jones. 2004. Chicken welfare is influenced more by housing conditions than by stocking density. Nature 427:342–344. 7. Cummings, T. S., P. A. Stayer, and A. B. Webster. 2005. Field assessment of walking ability in commercial broilers. Poult. Sci. 84(Suppl. 1):S114. (Abstr.) 8. Kestin, S. C., T. G. Knowles, A. E. Tinch, and N. G. Gregory. 1992. Prevalence of leg weakness in broiler chickens and its relationship with genotype. Vet. Rec. 131:190– 194. 9. Garner, J. P., C. Falcone, P. Wakenell, M. Martin, and J. A. Mench. 2002. Reliability and validity of a modified gait Webster et al.: THREE-POINT GAIT-SCORING SYSTEM scoring system and its use in assessing tibial dyschondroplasia in broilers. Br. Poult. Sci. 43:355–363. 10.Yalcin, S., P. Settar, and O. Dicle. 1998. Influence of dietary protein and sex on walking ability and bone parameters of broilers. Br. Poult. Sci. 39:251–256. 11.Gentle, M. J. 2001. Attentional shifts alter pain perception in the chicken. Anim. Welf. 10:S187–S194. 12.Weeks, C. A., T. G. Knowles, R. G. Gordon, A. E. Kerr, S. T. Peyton, and N. T. Tilbrook. 2002. New method for objectively assessing lameness in broiler chickens. Vet. Rec. 151:762–764. 13.Berg, C., and G. S. Sanotra. 2003. Can a modified latency-to-lie test be used to validate gait-scoring results in commercial broiler flocks. Anim. Welf. 12:655–659. 14.Cobb-Vantress Inc., Siloam Springs, AR 539 15.Aviagen, Huntsville, AL. 16.SAS Institute. 2002–2003. Version 9.1. SAS Institute Inc., Cary, NC. 17.Butterworth, A., T. G. Knowles, P. Whittington, L. Matthews, A. Rogers, and C. S. Bagshaw. 2007. Validation of broiler chicken scoring training in Thailand, Brazil and New Zealand. Anim. Welf. 16:177–179. 18.Cummings, T. S., B. Webster, K. Jones, and M. Cooper. 2008. Gait scoring of commercial broilers. Abstract 6606. Convention Notes CD. 145th AVMA Annual Convention, New Orleans, LA. Acknowledgments This research was supported by a grant from the US Poultry & Egg Association (Tucker, GA).
© Copyright 2026 Paperzz