Significant differences in lameness and limb-related abnormalities between Baseline and Final examinations of 149 working horses belonging to 131 owners from Jaipur, India.

(Participatory intervention group n = 83 horses belonging to 73 people; Control group n = 66 horses belonging to 58 people). The group with the largest relative improvement between Baseline and Final has the text depicting the magnitude of change in bold.1 See Table 1 for scoring methods2 For continuous or ordinal data = mean of horses (for whole horse measures) or limbs (for limb-specific measures) affected. For binary measures or those aggregated into two groups for analysis = percentage of horses/ limbs affected3 Relative improvement: the group which showed the most positive clinical change and/or the least negative clinical change (except for 'Shod' and 'Active response when approached' where no clinical value was applied)4 Calculated as Year 1—Year 3, i.e. a positive value = clinical improvement; a negative value = deterioration (except for 'Shod' and 'Active response when approached' where no clinical value was applied)5 1st-order MQL estimation (otherwise, less-biased 2nd-order PQL estimation was employed in all binary and multinomial response models)Significant differences in lameness and limb-related abnormalities between Baseline and Final examinations of 149 working horses belonging to 131 owners from Jaipur, India.(参与干预组：73名主人共饲养83匹马；对照组：58名主人共饲养66匹马)。基线检测与最终检测间相对提升幅度最大的组别，其变化幅度的描述文本将以加粗格式呈现。1 评分方法详见表1；2 针对连续型或有序型数据：取受影响马匹（全马维度指标）或受影响四肢（四肢特异性指标）的均值；针对二分类指标或经聚合为两组用于分析的指标：取受影响马匹/四肢的占比；3 相对提升：指临床变化最正向且/或临床负向变化最少的组别（“是否装蹄”与“主动回应靠近”两项无临床价值，予以除外）；4 差值计算方式为第1年数值减去第3年数值：正值代表临床症状改善，负值代表病情恶化（“是否装蹄”与“主动回应靠近”两项除外）；5 一阶边际拟似然（MQL, Marginal Quasi-likelihood）估计；其余二项及多项响应模型均采用偏差更小的二阶边际拟似然（PQL, Penalized Quasi-likelihood）估计。本数据集涵盖印度斋浦尔地区131位主人饲养的149匹役马，其基线与最终检测结果显示，跛行与四肢相关异常存在显著差异。