Accuracy of pregnancy specific protein-B test for early pregnancy diagnosis in dairy cattle

The objective was to evaluate the accuracy of an ELISA for pregnancy specific protein B (PSP-B) for early pregnancy diagnosis in dairy cattle. Blood from lactating (>100 d postpartum) dairy cows (n = 738), was collected on Days 28, 30, and 35 (Day 0 = estrus), analyzed with an ELISA for PSP-B, and the cows designated as pregnant, probable, unlikely, or non-pregnant. Immediately after blood collection, transrectal ultrasonography (TRUS) was done for pregnancy diagnosis, and the results used as a criterion standard test for comparison with PSP-B. At Day 28, 46.3% were diagnosed by TRUS as pregnant. The PSP-B sensitivity was 93.9% on Day 28 and similar on Days 30 and 35. The PSP-B specificity, positive predictive value, negative predictive value, and accuracy were all >94% on Day 28 and similar on Days 30 and 35. However, the accuracy was significantly less compared to TRUS (P < 0.01). The percentage of all samples from cows that were probably pregnant or unlikely to be pregnant was 5.6%. At Days 28, 30, and 35, percentages of uncertain samples were 8.5, 4.8, and 3.3%, respectively (P < 0.01), and Kappa values were 0.92, 0.92, and 0.95. False negative and false positive results were attributed to low concentrations of PSP-B in pregnant animals and to persistence of pregnant concentrations of PSP-B in females with pregnancy loss, respectively. In conclusion, PSP-B ELISA was a sensitive, specific, and accurate test for pregnancy diagnosis (relative to TRUS) at Days 28, 30, and 35 after breeding.