Measuring claw conformation in cattle: Assessing the agreement between manual and digital measurement

Abstract

Five measurements of claw conformation (toe angle, claw height, claw width, toe length and abaxial groove length) taken directly from the hoof were compared with the measurements taken from digital images of the same claws. Concordance correlation coefficients and limits-of-agreement analysis showed that, for four of the five measures (claw height, claw width, toe length and abaxial groove length), agreement was too poor for digital and manual measures to be used interchangeably. For all four of these measures, Liao’s modified concordance correlation coefficient (mCCC) was ≤0.4, indicating poor concordance despite Pearson’s correlation being >0.6 in all cases. The worst concordance was seen for toe length (mCCC = 0.13). Limits-of-agreement analysis showed that, for all four measures, there was a large variation in the difference between the manual and digital methods, even when the effect of mean on difference was accounted for, with the 95% limits-of-agreement for the four measures being further away from the mean difference than 10% of the mean in all four cases. The only one of the five measures with an acceptable concordance between digital and manual measurement was toe angle (mCCC = 0.81). Nevertheless, the limits-of-agreement analysis showed that there was a systematic bias with, on average, the manual measure of toe angle, being 2.1° smaller than the digital. The 95% limits-of-agreement for toe angle were ±3.4°, probably at the upper limit of what is acceptable. However, the lack of data on the variability of individual measurements of claw conformation means that it is unclear how this variability compares to measurement of toe angle in the same animal using the same or a different manual technique.