Materials and Methods

Animals—Dairy cattle from 16 dairy herds in the central region of the state of New York that were regularly provided service by the Cornell Ambulatory and Production Medicine Clinic were used in the study. Cows and heifers from the 16 farms were used in phase 1, and cows in 4 herds that were a subset of the original 16 herds were used in phase 2.

Description of herds—Holsteins constituted the major breed, although small numbers of other breeds were also included. Herd size ranged from 30 to 3,000 lactating cows, although most of the data were collected from the larger herds. Artificial insemination was used as the main breeding method on these herds, and only those cattle with complete records, including insemination dates, were included in the study. Estrus or ovulation synchronization programs that involved the use of PGF_2α and GnRH analogues were used to manage breeding programs on most of the farms.^3,21,22 Depending on the strategy of a particular herd, the herd veterinarian was requested to perform PPR on selected cattle at intervals of 1 to 4 weeks. The interval from last insemination to first PPR varied among herds from approximately 35 to 60 days. In most herds, a second PPR was performed as a confirmation of pregnancy or identification of embryonic-fetal loss; this PPR was performed approximately 100 days after insemination.

Phase 1—On the same day, PPR was performed on and blood samples collected from cows and heifers examined to determine pregnancy status during regularly scheduled herd visits. Phase 1 was conducted between October 2006 and February 2007.

Sample collection and PPR—During each herd visit, a veterinarian conducted reproductive examinations on cattle selected by the owner or manager. Reproductive examinations were performed by 1 of 5 veterinarians with experience (3 to 30 years) in diagnosing pregnancy by PPR. The technique or techniques used to determine pregnancy status (identification of an amniotic vesicle, allantochorionic membranes, fetus, or placentomes) differed among the 5 veterinarians and stages of gestation.^1,4 After each PPR, the animal was classified as pregnant, not pregnant, or reexamination required. When an animal was classified as not pregnant by PPR, it was generally suggested to resynchronize estrus via hormone administration.

At the time of PPR, blood samples for evaluation by use of a commercial pregnancy test^b (PSPB ELISA) were collected from coccygeal vessels into evacuated glass tubes,^c as suggested by the manufacturer of the test.^b Blood samples were then labeled, packaged, and sent by an overnight courier to a commercial laboratory, which used the ELISA to determine PSPB concentrations. Results for the PSPB ELISA were transmitted electronically to the veterinary clinic 4 or 5 days after samples were submitted. The PSPB ELISA results were provided as an OD value for each sample with a diagnosis of pregnant, not pregnant, PRX, or NPRX. The results were tabulated in an electronic database,^d and cattle with incomplete records were excluded from the analyzed data set.

Data analysis—Results for PPR were obtained from paper records for each herd visit or extracted from a computer record-keeping program^e maintained for the herds. Other data collected for analysis were herd, animal identification number, lactation number, PPR date, parturition date, number of days in lactation, insemination dates, and anticipated calving date. Agreement between PPR and PSPB ELISA results was evaluated by calculating N.^f For each weekly submission or batch of samples sent to the commercial laboratory, we received an electronic report that included classification as not pregnant, NPRX, PRX or pregnant; observed OD; and the 3 OD cutoff values separating the status classes. The cutoff values were different for each batch, but we observed that the ratios of the high and low cutoff values to the middle cutoff value were constant. Therefore, for statistical analyses of OD, values were adjusted in accordance with the following equation:

View Expanded

where OD_ij is the OD of animal i in batch j, C2_j is the middle cutoff value for batch j, and C2_X is the middle cutoff value for an arbitrary reference batch.

Descriptive statistics and data plots were performed by use of commercially available software.^g For cows classified pregnant on the basis of results for the PSPB ELISA, differences in adjusted OD for cows classified not pregnant and pregnant by use of PPR were analyzed^h; a mixed linear model was used, with herd and the herd × PPR diagnosis interaction term as random effects. For this analysis, only cattle within 75 days after insemination were included. Examination of plots of the residuals revealed that there were no important departures from model assumptions.

Phase 2—The second phase was conducted in June 2007 on cattle in 4 herds. Resynchronization protocols were commonly used in these herds, which allowed the collection of blood samples 1 week prior to PPR.

Only lactating cows were enrolled in this portion of the study. Artificial insemination was the only breeding method used in these 4 herds. A record-keeping database^e was used for each of the 4 herds. Estrus and ovulation synchronization and resynchronization programs by use of PGF_2α and GnRH analogues were used to manage breeding. Additional cows were inseminated on the basis of detection of estrus. Examination of the reproductive tract via PPR was used to determine pregnancy by each herd veterinarian; PPR was performed once per week. For 3 of the 4 herds, the first PPR was performed 35 to 41 days after insemination, whereas for the fourth herd, PPR was performed approximately 42 to 48 days after insemination. A second PPR was performed for confirmation of pregnancy approximately 100 days after insemination. Three experienced (3 to 30 years) veterinarians performed the PPR in these 4 herds. The pregnancy rate^e for the 4 herds varied from 14% to 20% during the study period.

Sample collection—Blood samples were collected from cows 1 week before their first scheduled PPR and processed similarly to those in phase 1. One week later, a herd veterinarian performed PPR on these cows as part of the scheduled herd visit. Results of PPR were recorded in the farm computer records. When a cow was classified not pregnant during PPR, it was usually administered PGF_2α as part of a resynchronization protocol. Results of the PSPB ELISA were transmitted electronically to the veterinary clinic prior to the PPR; however, to reduce bias during PPR, the veterinarians performing the PPR were not aware of the PSPB ELISA results.

Data analysis—Data for the PPR and PSPB ELISA were recorded in a manner similar to that in phase 1. Data were tabulated in an electronic database,^d and cattle with an incomplete data set were excluded from the analysis.

Data were analyzed as for phase 1, except C2_X differed from that in phase 1. Consequently, the adjusted OD could not be directly compared between phases 1 and 2.

Results

Phase 1—Initially, 1,104 cattle were included in the study; however, 127 were removed because of incomplete records, and 1 was removed because it was classified to be reexamined by PPR. Thus, 976 cattle were available for analyses. Of these, 88 (9%) were heifers, 351 (36%) were cows in their first lactation, and 537 (55%) were cows in their second or greater lactation. Median number of days in lactation was 151 (range, 77 to 846 days), and the median number of days after the last insemination was 39 (range, 35 to 221 days).

By use of PPR, 414 (42%) cattle were classified not pregnant, and 562 (58%) cattle were classified pregnant (Table 1). By use of the PSPB ELISA, 355 (36%) cattle were classified not pregnant, 571 (59%) were classified pregnant, 30 (3%) were classified NPRX, and 20 (2%) were classified PRX. Agreement between the 2 methods was k = 0.82 (95% confidence interval, 0.76 to 0.88). The middle cutoff value for classifying cattle on the basis of the ELISA OD ranged from 0.116 to 0.181 for 20 batches of samples; 0.148 was selected as the value used to standardize OD measurements (ie, C2_X). A comparison of adjusted PSPB ELISA OD with number of days after the last insemination in cattle classified pregnant by both the PSPB ELISA and PPR was plotted (Figure 1). Cattle for which there were discrepant results were evaluated further.

View Full Size

Graph of the adjusted OD of the PSPB ELISA on the basis of the number of days after the last insemination for 541 dairy cattle classified pregnant by use of the PSPB ELISA and PPR during phase 1.

Citation: Journal of the American Veterinary Medical Association 235, 3; 10.2460/javma.235.3.292

• Download Figure
• Download figure as PowerPoint slide

Thirty cattle were classified not pregnant by use of PPR and pregnant by use of the PSPB ELISA, and analysis of records revealed that the majority (27/30) received a PGF_2α analogue immediately following PPR. For the other 3 cows, 1 was classified by use of PPR as aborted at the 100-day pregnancy confirmation examination (the animal had been classified pregnant by use of PPR at 39 days after the last insemination), 1 was observed in estrus on the day of PPR and therefore not administered a PGF_2α analogue, and 1 was classified as not fit to breed and was therefore not administered a PGF_2α analogue. As number of days after the last insemination increased, discrepant results between the PPR and PSPB ELISA decreased (Table 2). Of the 30 cattle with discrepant results, 28 were identified at < 45 days after the last insemination.

Half (5/10) of the cattle classified pregnant by use of PPR and not pregnant by use of the PSPB ELISA were observed in estrus within 50 days after the testing date; these cattle were subsequently reexamined by use of PPR and consequently classified not pregnant. Of the other 5 cattle with discrepant results, 2 were reconfirmed as pregnant by use of PPR at the 100-day pregnancy confirmation and subsequently gave birth to a calf, 1 was confirmed pregnant at 80 days after the last insemination but was found not pregnant by use of PPR 1 week later, and 2 died or were sold before confirmation of pregnancy could be performed. Again, as the number of days after the last insemination increased, discrepant results between the PPR and PSPB ELISA decreased (Table 2). All discrepant results were identified at < 50 days after the last insemination.

The adjusted PSPB ELISA OD values of the cattle classified pregnant by use of both tests were compared with the PSPB ELISA OD values of the cattle classified pregnant by use of the PSPB ELISA and not pregnant by use of PPR. Of 470 cattle classified pregnant by use of the PSPB ELISA within 75 days after insemination, the adjusted OD was significantly (P = 0.01) higher for 441 cattle classified pregnant by use of PPR (least squares mean ± SD, 0.31 ± 0.01), compared with the adjusted OD for the 29 cattle classified not pregnant by use of PPR (mean, 0.22 ± 0.02; Figure 2). Results were similar when cattle < 90 days in lactation and > 30 days after the last insemination were excluded.

View Full Size

Box-and-whisker plots of the adjusted OD of the PSPB ELISA for 976 dairy cattle during phase 1 on the basis of PSPB ELISA and PPR results. The boxes indicate the interquartile range, the horizontal line within each box indicates the median, and the whiskers indicate the range of values. Mild (asterisk) and extreme (circle) outliers for each group are indicated. The high number of outliers in the P-P group is a result of the later stage for pregnancies in this group, which resulted in a higher concentration of PSPB and therefore higher OD values. NP-NP = Not pregnant by use of the PSPB ELISA and not pregnant by use of PPR. NP-P = Not pregnant by use of the PSPB ELISA and pregnant by use of PPR. P-NP = Pregnant by use of the PSPB ELISA and not pregnant by use of PPR. P-P = Pregnant by use of the PSPB ELISA and pregnant by use of PPR. PRX-NP = Pregnant and retest recommended by use of the PSPB ELISA and not pregnant by use of PPR. PRX-P = Pregnant and retest recommended by use of the PSPB ELISA and pregnant by use of PPR.

Citation: Journal of the American Veterinary Medical Association 235, 3; 10.2460/javma.235.3.292

• Download Figure
• Download figure as PowerPoint slide

Phase 2—Initially, 556 cows were enrolled, but incomplete or incorrect data and loss of cows to follow-up monitoring resulted in removal of 49 cows from the study. Thus, 507 cows were available for analysis. Of these, 206 (37%) were in their first lactation, and 350 (63%) were in their second lactation or greater. Median number of days in lactation was 144 (range, 71 to 660 days). Median number of days after the last insemination at time of collection of the blood sample was 32 (range, 28 to 60 days).

Results for the PPR and PSPB ELISA were compared (Table 3). Agreement between the 2 tests was N = 0.81 (95% confidence interval, 0.73 to 0.89). Forty-nine (10%) cows were classified pregnant, PRX, or NPRX by use of the PSPB ELISA and were subsequently classified pregnant by use of PPR 1 week later. Of these, 39 cows were injected with a PGF_2α analogue at the time of the nonpregnant classification by use of PPR, 3 had evidence of estrus activity prior to or on the day of PPR, 4 were sold or determined as unfit to breed, 2 received no PGF_2α and were subsequently found to be not pregnant, and 1 was enrolled in a resynchronization program. By use of the PSPB ELISA, 2 cows were classified not pregnant but were classified pregnant by use of the PPR 1 week later; 1 of these was found not pregnant at the 100-day pregnancy confirmation, and 1 was culled from the herd prior to examination.

The middle ELISA OD cutoff values in phase 2 ranged from 0.121 to 0.145 for 6 batches of samples; 0.128 was selected as the value used to standardize OD measurements (ie, C2_X). For 255 cows classified pregnant by use of the PSPB ELISA, the adjusted OD was significantly (P = 0.03) higher for 224 cows classified pregnant by use of PPR (least squares mean ± SD, 0.26 ± 0.01), compared with the adjusted OD for 30 cows classified not pregnant by use of PPR (mean, 0.21 ± 0.01; Figure 3). Results were similar when cows < 90 days in lactation or < 30 days after the last insemination were excluded from the analysis; however, the adjusted ODs for that analysis did not differ significantly (P = 0.09).

View Full Size

Box-and-whisker plots of the adjusted OD of the PSPB ELISA for 507 dairy cattle during phase 2 on the basis of PSPB ELISA and PPR results. The high number of outliers in the P-P group is a result of the later stage for pregnancies in this group, which resulted in a higher concentration of PSPB and therefore higher OD values. See Figure 2 for remainder of key.

Citation: Journal of the American Veterinary Medical Association 235, 3; 10.2460/javma.235.3.292

• Download Figure
• Download figure as PowerPoint slide

Discussion

The accuracy of pregnancy diagnosis is critical if a herd is to maintain reproductive efficiency. Furthermore, early identification and intervention in nonpregnant cows improves pregnancy rate and overall milk production, which generates more profit for the enterprise.^2,5 Although there was good agreement between the 2 tests in phase 1 (N = 0.82), the number of discrepancies was potentially economically important. The 2 tests disagreed for 90 of 976 (9%) cattle, including 50 classified as NPRX or PRX on the basis of PSPB ELISA results. Because the intervention for cattle was determined on the basis of the PPR result only, most cattle classified as not pregnant by use of PPR were administered PGF_2α; if the pregnancy diagnosis for these cattle was incorrect (ie, the cattle were actually pregnant), administration of PGF_2α would likely have resulted in luteolysis and pregnancy loss. When investigating the cows from phase 1 that were classified pregnant by use of PPR and not pregnant by use of the PSPB ELISA, 5 of 10 subsequently came into estrus soon after the PSPB ELISA was performed. Thus, the PSPB ELISA correctly identified the pregnancy status for these cows, whereas PPR did not. The reasons for this are likely to be attributable to induced or natural pregnancy loss in early gestation for which positive signs of pregnancy may persist for several days.^23,24 Alternatively, the palpator may have been biased because of concern about causing abortions by administration of PGF_2α. However, results of the PSPB ELISA misclassified at least 2 of these cows (both were pregnant at pregnancy confirmation approx 100 days after insemination). In addition, 50 of 976 (5%) cattle classified PRX or NPRX by use of the PSPB ELISA indicates a source of inefficiency: there would be a delay in determining pregnancy status because a second sample would have to be submitted for testing.

For cattle classified pregnant by use of the PSPB ELISA and PPR, the OD from phase 1 revealed that when the tests were in agreement, a significantly higher OD was evident than when the tests disagreed (ie, cattle classified not pregnant by use of PPR and pregnant by use of the PSPB ELISA). Because the concentration of PSPB in blood increases with progression of pregnancy, we restricted this analysis to cattle < 75 days after the last insemination. Furthermore, because of a long half-life, PSPB from the preceding pregnancy may remain in a cow's circulation for approximately 90 days^8,15; however, the difference between these 2 groups was significant when cows < 90 days in lactation were removed from the data set. Thus, results from phase 1 were consistent with the conclusion that PPR was able to detect nonpregnant cattle more accurately than the PSPB ELISA early in gestation. An explanation for this difference is that the PSPB ELISA continues to detect PSPB in blood after embryonic or fetal loss. If the OD cutoff values of the PSPB ELISA were adjusted to reduce the false-pregnant diagnoses, an increase in false-nonpregnant diagnoses would likely result, which would cause an unacceptable increase in pregnancy loss caused by PGF_2α administration. In 1 report,⁸ a false-pregnant diagnosis (determined by use of an RIA for PSPB) persisted for several weeks after embryonic loss. The discrepant results in phase 1 decreased with an increase in the number of days after the last insemination, which indicated a general increase in accuracy for both tests with increased numbers of days after last insemination (Table 2).

Diagnosis of pregnancy status by use of PPR or transrectal ultrasonography provides an opportunity for immediate intervention in cattle classified not pregnant. Currently, immediate results cannot be obtained for the PSPB ELISA because the process requires that samples be submitted to a laboratory where the test is performed, with a consequent delay of several days. This delay in the return of results would not allow interventions in cattle from which samples were obtained 1 week before PPR until those cattle were 34 or 35 days after insemination. Because 35 days after insemination is the earliest time at which PPR is commonly used, the PSPB ELISA does not appear to provide an advantage over PPR for earlier diagnosis. Development of a cowside or rapid on-farm PSPB ELISA with similar accuracy to the current test would decrease the time interval required for identification of nonpregnant cattle.

One method of resynchronization of cows is the administration of GnRH 7 days before a pregnancy test is performed. Then, on the day of the pregnancy test, if the cow is found to be not pregnant, PGF_2α is administered; 2 days later, another dose of GnRH is administered, and then the cow is inseminated approximately 16 hours later.^3,21 Because there is a delay of several days between collection of blood samples and receipt of the PSPB ELISA results, phase 2 was designed to incorporate the PSPB ELISA result into a resynchronization protocol. Therefore, on the first day of resynchronization, blood samples were collected and GnRH was administered. Consequently, the results of the PSPB ELISA would be available by day 7 of resynchronization (ie, day of PPR) and could be used in determining administration of PGF_2α. This schedule was efficient because it did not require an extra period of restraint of the cows for blood collection and the PSPB ELISA results were available before PGF_2α was administered. Results of the study reported here revealed good agreement between the 2 tests for the classification of nonpregnant cattle, with only 2 cows classified not pregnant by use of the PSPB ELISA and classified pregnant by use of PPR. However, there was more disagreement among cows classified pregnant by use of the PSPB ELISA because 31 of 255 (12%) of those cattle were classified not pregnant by use of PPR 1 week later. As mentioned previously, a likely reason for this discrepancy is that the PSPB ELISA detects PSPB concentrations in the circulation after embryonic or fetal death and therefore provides a false-pregnancy diagnosis. The 1 week interval between collection of blood samples and PGF_2α administration to nonpregnant cows is a period of relatively high embryonic loss.^5,21,25 Consequently, it would be expected that some proportion of cows pregnant on the day of blood collection would have embryonic or fetal death prior to the day of PGF_2α administration. More of these cattle may be identified by use of PPR, compared with those identified by use of the PSPB ELISA, and therefore this explains the results we obtained. Taken together, the errors from incorporating incorrect results of the PSPB ELISA into a resynchronization protocol will affect a substantial proportion of nonpregnant cattle and consequently decrease reproductive efficiency. Again, if a cow-side or rapid on-farm PSPB ELISA were developed with similar accuracy to the current test, there may be a considerable reduction in these errors.

To our knowledge, the information reported here represents the first independently conducted study to compare results for a commercial PSPB ELISA with results for another form of pregnancy detection. The 2 tests appeared to concur for most of the cattle tested for pregnancy status, although it also appeared that neither test was faultless. Discrepant OD results for cattle classified nonpregnant by use of PPR and pregnant by use of the PSPB ELISA provided evidence of false-positive results for pregnancy by use of the PSPB ELISA, most likely as a result of nonviable pregnancies or embryonic loss. This possible source of pregnancy diagnostic error and the delayed return of results for the PSPB ELISA results, compared with the diagnostic accuracy and immediacy of obtaining results for PPR, are potential weaknesses for use of the PSPB ELISA as a test for pregnancy status in dairy cattle.