Animal Welfare Program, Faculty of Land and Food Systems, University of British Columbia, Vancouver, BC V6T 1Z4, Canada.
Animal Welfare Program, Faculty of Land and Food Systems, University of British Columbia, Vancouver, BC V6T 1Z4, Canada.
J Dairy Sci. 2020 Jan;103(1):638-648. doi: 10.3168/jds.2019-16426. Epub 2019 Oct 31.
The reliability of locomotion scoring is often low, making it unclear how a single gait score should be interpreted. In addition, differences in assessment frequency between longitudinal studies makes it hard to compare results. Our aims were to evaluate how lameness definition and assessment frequency affect measures of lameness incidence. Six dairy farms in British Columbia, Canada, were enrolled, and 262 cows that were sound at dry-off had their locomotion score (LS) assessed weekly from dry-off to calving, using a 1 to 5 scale. Cows were categorized as remaining sound or becoming lame using 3 different case definitions (LAME1: ≥LS3 at least once; LAME2: ≥2 consecutive scores of LS3, or ≥LS4 at least once; and LAME3: ≥3 consecutive scores of LS3, or ≥LS4 at least once). We analyzed the correspondence between the 3 definitions with percent agreement and weighted κ (linear and quadratic weighting). Comparing LAME1 to LAME3 resulted in lower percent agreement (53%) and κ values (linear κ = 0.50; quadratic κ = 0.64) than comparing LAME2 and LAME3 (85%; linear κ = 0.83; quadratic κ = 0.89), indicating that cows scored LS3 twice were likely to be scored LS3 a third time. We also compared the 3 case definitions against trim records from trimmings occurring 90 d or less before calving (n = 117), and used logistic regression models to determine sensitivity, specificity, and positive and negative predictive value. Using the LAME1 criterion resulted in high sensitivity (horn lesions = 0.90; infectious lesions = 0.92) and low specificity (horn = 0.21; infectious = 0.24). We observed higher specificity for LAME2 (horn = 0.62; infectious = 0.66) and LAME3 (horn = 0.71; infectious = 0.77), but LAME2 had higher sensitivity than LAME3 (horn = 0.89 vs. 0.64; infectious = 0.69 vs. 0.64). When evaluating the effects of assessment frequency, we obtained 3 data sets by keeping every, every other, and every third locomotion assessment, and using LAME2 as a case definition. More cows were categorized as lame when assessment frequency increased. Of the cows that were classified as lame when assessed weekly, 72% of the mildly lame, and 33% of the severely lame were classified as sound when assessed every third week. Our results suggest that a single LS3 score should not be used as a criterion for lameness in longitudinal studies. To correctly identify new cases of lameness, dairy cows should be assessed at least every 2 wk.
跛行评分的可靠性通常较低，因此不清楚如何解释单个步态评分。此外，纵向研究中评估频率的差异使得很难比较结果。我们的目的是评估跛行定义和评估频率如何影响跛行发生率的测量。在加拿大不列颠哥伦比亚省的 6 个奶牛场，招募了 262 头在干奶期时健康的奶牛，从干奶期到分娩，每周使用 1 到 5 的评分标准评估它们的运动评分（LS）。使用 3 种不同的病例定义（跛行 1：至少有一次 LS3 评分≥3；跛行 2：至少有 2 次连续 LS3 评分，或至少有一次 LS4 评分≥3；跛行 3：至少有 3 次连续 LS3 评分，或至少有一次 LS4 评分≥3）将奶牛分为保持健康或跛行。我们使用百分一致性和加权 κ（线性和二次加权）来分析这 3 种定义之间的一致性。与跛行 3 相比，跛行 1 的百分一致性（53%）和 κ 值（线性 κ=0.50；二次 κ=0.64）较低，表明 LS3 评分两次的奶牛更有可能 LS3 评分第三次。我们还将这 3 种病例定义与干奶前 90 天或更短时间发生的修剪记录（n=117）进行了比较，并使用逻辑回归模型来确定敏感性、特异性和阳性预测值和阴性预测值。使用跛行 1 标准可获得高敏感性（角病变=0.90；感染病变=0.92）和低特异性（角病变=0.21；感染病变=0.24）。我们观察到跛行 2（角病变=0.62；感染病变=0.66）和跛行 3（角病变=0.71；感染病变=0.77）的特异性更高，但跛行 2 的敏感性高于跛行 3（角病变=0.89 比 0.64；感染病变=0.69 比 0.64）。当评估评估频率的影响时，我们通过保留每次、每隔一次和每隔三次的运动评估获得了 3 个数据集，并使用跛行 2 作为病例定义。随着评估频率的增加，更多的奶牛被归类为跛行。在每周评估时被归类为跛行的奶牛中，72%的轻度跛行和 33%的严重跛行在每三周评估时被归类为健康。我们的结果表明，在纵向研究中，不应将单个 LS3 评分作为跛行的标准。为了正确识别新的跛行病例，奶牛应至少每 2 周评估一次。
