Predicting end-of-test semen quality in bulls prior to performance testing.

The first objective of this study was to determine if serum concentrations of specific hormones (testosterone, progesterone and androstenedione) in bulls at the start of performance testing could predict semen quality at the end-of-test when used in a multivariate model. The second objective was to evaluate other clinical measurements (breed, age, body weight, hip height and scrotal circumference) for predicting end-of-test semen quality. End-of-test semen quality was related to steroid concentrations and several pre-testing measurements, including age, body weight, hip height and scrotal circumference (SC). Combining the 3 steroid concentrations into a predictive test had a sensitivity of 0.6 and specificity of 0.5 at its most accurate point. The repeatability of the test result was extremely low (r(2) = 0.16; P < 0.05). In multivariate analyses, breed and start-of-test SC remained significant predictors of end-of-test semen quality (P < 0.05) while the other variables were nonsignificant (P > 0.1), suggesting that start-of-test SC was the most accurate predictor of end-of-test semen quality. Removing bulls at the start-of-test that had scrotal measurements of less than 20 cm, 24 cm, 28 cm or 32 cm resulted in sensitivities and specificities of 0.19, 0.94; 0.41, 0.81; 0.64, 0.56; and 0.94, 0.12, respectively. No cut-point had both adequate sensitivity and specificity. Because clinical tests were correlated, combining the tests to improve accuracy was not justified.