Predictive value of palpation per rectum for detection of the CL in Zebu cattle as evaluated by progesterone concentrations and ultrasonography

The main objective of this study was to calculate the predictive value of the positive and negative diagnosis of the presence of a corpus luteum by palpation per rectum in Zebu cattle. Gyr breed (n=16) heifers were palpated by a veterinarian every 3 d during March and April, and every 5 days during May. The presence or absence of a CL was recorded for 442 examinations. A blood sample was obtained after each examination, and a functional CL was considered to be present if plasma progesterone concentrations were above 1 ng/ml. Progesterone results were used as the reference to calculate the sensitivity and specificity of palpation per rectum for the diagnosis of a functional CL on different months. Predictive values for the positive and negative diagnoses were calculated using different hypothetical rates of the prevalence of CL. To evaluate the morphological basis of an erroneous diagnosis, ultrasonographic scanning of the ovaries was performed in 264 cases and the physical presence or absence of a CL was registered. The calculations were repeated using the ultrasonographic findings as the reference to evaluate the sensitivity and specificity of the diagnosis of a structural CL by per rectum palpation. Several heifers were not cyclic in March but began to cycle during the study. Thus, the prevalence of both functional and structural CL increased with time. As a result, the proportion of palpation diagnoses classified as true positives increased in May, while the proportion of true negatives decreased. Although the values of sensitivity and specificity remained constant from month to month, the predictive values changed markedly in May, when the predictive value of the positive test increased while the predictive value of the negative test decreased. These changes were due to differences in the prevalence of a CL and not to differences in the palpable characteristics of CLs in different months. The prevalence of a structural CL was always higher than that of a functional CL. The Cohen's Kappa test for concordance revealed better correspondence between palpation per rectum and ultrasound (k=0.82) than between progesterone and either ultrasound (k=0.68) or palpation (k=0.66). These results show that a large proportion of the errors imputed to deficient palpation when progesterone concentrations are used as a reference are in fact the result of lack of correspondence between the physical and functional presence of a CL.