The Accuracy and Repeatability of Sow Body Condition Scoring 1

Citation data:

The Professional Animal Scientist, ISSN: 1080-7446, Vol: 25, Issue: 4, Page: 415-425

Publication Year:
2009
Usage 16
Abstract Views 11
Downloads 3
Link-outs 2
Captures 16
Readers 16
Citations 2
Citation Indexes 2
Repository URL:
https://lib.dr.iastate.edu/ans_pubs/395
DOI:
10.15232/s1080-7446(15)30736-1
Author(s):
Fitzgerald, R. F.; Stalder, K. J.; Dixon, P. M.; Johnson, A. K.; Karriker, L. A.; Jones, G. F.
Publisher(s):
American Registry of Professional Animal Scientists
Tags:
Agricultural and Biological Sciences; body condition score; interobserver agreement; intraobserver agreement; sows
article description
The objective of this study was to estimate observer accuracy and repeatability of body condition scoring of sows when scorers had different levels of prior experience. Participants (n = 17) for this study were identified as having no (n = 7), some (n = 4), and extensive (n = 4) prior experience evaluating conformation or body condition in livestock species or having served as instructors (TR; n = 2) during the training sessions. Twenty-five of a total 150 sows were used in the participant training session, and the remaining sows (n = 125) were used during the independent scoring process. Sows used in the scoring process were objectively categorized into BCS on a 5- and 9-point scale (BCS 5 and BCS 9, respectively) using last-rib backfat estimates. Participant BCS 5 and BCS 9 deviation evaluations from last-rib backfat estimates revealed a tendency to overestimate BCS in some sows and underestimate BCS in others. Repeatability, a measure of variance attributed to individual participants between rounds of scoring, was the largest contributor (70.6%) to the total test variability. Reproducibility, a measure of variance attributed to BCS 9 assigned by multiple participants, accounted for only 29.4% of the total test variation. Much more opportunity exists to reduce repeatability variance as compared with variance associated with participant bias. Standard errors of the difference were lower on the BCS 9 scale, and participants were within scores of 0.50 between their first- and second-round values. Participants were consistently within 0.65 scores of BCS assigned by backfat measurements. Averaged across all participants, the ultrasonic trait of last-rib backfat yielded the greatest correlation (0.58) with BCS 9, followed by 10th-rib backfat (0.51), 10th-rib loin eye area (0.47), and last-rib loin eye area (0.43). Similar trends were observed for the BCS 5 scale. Therefore, practice assigning independent BCS should lower repeatability variance, but training to calibrate specific participants may influence total test variance only to a relatively small degree.