• Rhiannon Handcock

Milk Pregnancy Test: How accurate is it?

Updated: Feb 10, 2020

Have you tried the milk pregnancy testing of herd test samples yet? Or are you on the fence as you aren't sure how accurate they are?

This week we have summarised a study out of Kentucky that evaluated the IDEXX milk pregnancy test kit.

Key Points

  1. The accuracy of the test to detect pregnant and non-pregnant cows was 99% at first ultrasound check and 98% at second ultrasound check

  2. The earliest time point that pregnant cows were accurately classified as pregnant was 30 days after insemination

  3. The optimum time to check for pregnancy using this milk pregnancy test was in the 2-week period from 30 to 44 days after insemination

Although this study was done in the USA on USA cows, the test used was one that is available worldwide and is used in New Zealand. We couldn't find any published studies using the IDEXX kit in NZ, so this is the next best thing!

How does the milk pregnancy test work?

When a cow becomes pregnant the presence of the growing calf (fetus) causes a "highly specific marker of pregnancy" to be produced. This marker is called "pregnancy-associated glycoproteins" (PAG), and it is only produced when the cow is pregnant.

The only reason that a non-pregnant cow may have PAGs in her system is if she has suffered early embryonic loss and her body is yet to get rid of all the PAGs.

This makes tests that detect PAGs very effective at identifying pregnant cows.

At first PAGs were measured in blood samples, but in the last 5 or so years, a milk sample test was developed.

This milk pregnancy test categorises samples into 1 of 3 depending on their PAG levels:

  1. Not-pregnant

  2. Pregnant

  3. Recheck required

How accurate is the milk pregnancy test?

The first experiment reported in this study, looked at PAGs in milk for 119 cows at their first ultrasound pregnancy test between 33 and 52 days after insemination.

Over the 119 cows, 70 were confirmed pregnant by ultrasound (58.8%; blue in figure 1) and 49 were not pregnant (41.2%; purple in figure 1).

Figure 1. Proportion of pregnant and non-pregnant cows at the first ultrasound check

The ability of the milk pregnancy test to accurately determine if a cow was pregnant or not was compared with the ultrasound results as the "true" results.

Of the 49 non-pregnant cows (left side of Figure 2), 46 were correctly classed as non-pregnant by the milk PAG test (38.7% of 119; purple), 1 was wrongly classed as pregnant (0.8% of 119; blue).

Of the 70 pregnant cows (right side of Figure 2), 65 were correctly classed as pregnant by the milk PAG test (54.6% of 119; blue) and none were wrongly classed as non-pregnant (0%; purple).

There were 7 cows classed as a recheck (6% of all cows; grey); 5 were actually pregnant (4.2%) and 2 were actually not pregnant (1.7%).

Figure 2: The proportion of cows that the milk pregnancy test assigned as pregnant, non-pregnant or recheck that were actually pregnant or were actually not pregnant

The accuracy of the test was measured based on correctly identifying pregnant and non-pregnant cows (rechecks were not included).

Overall, the accuracy was 99% for this first check.

This was made up of the 46/47 correctly identified non-pregnant cows and the 65/65 correctly identified pregnant cows (111/112 = 99%).

The test was repeated at the second ultrasound check (60-74 days after insemination), with 60 cows included.

This time there were 55 pregnant cows and 5 non-pregnant cows.

All 5 non-pregnant cows were correctly identified by the milk pregnancy test, 45 of the 55 pregnant cows were correctly identified and 1 cow was incorrectly identified as being not pregnant when she actually was pregnant.

The proportion of cows requiring a recheck were higher than for the first scan, this time 15% (9/60). All of these 9 cows were actually pregnant.

Again, the accuracy was very high at 98% for this second check (remember rechecks were not included).

Timing of the test

A second experiment was conducted to pinpoint the earliest time period that the milk pregnancy test could identify pregnant cows reliably.

In the samples collected between insemination and 18 days after insemination, the levels of PAGs were well below the levels required to class cows as being pregnant or a recheck. In other words, within the first 18 days after insemination, pregnant cows would be classified as not pregnant, as their PAGs were not at a high enough level in their milk yet.

By 30 days of pregnancy, the milk pregnancy test was accurate in identifying pregnant cows.

Interestingly, between days 46 and 72 after insemination, there was a decline in PAG levels to an intermediate range (that would suggest a recheck) even though these cows were pregnant. After day 72, PAG levels rose back to a higher range that would suggest pregnancy.

This period of a decline in PAG levels was at a similar stage of pregnancy as the second check in the first experiment mentioned earlier. Which would explain why there were 15% of cows classed as recheck and all of which were actually pregnant.

From this second experiment, the authors concluded that the optimum time to check for pregnancy using this milk pregnancy test is in the 2-week period from 30 to 44 days after insemination.

What about the potential mixing of samples at herd-test?

One point that was not made in this paper was the potential of residual milk from the previous cow being mixed into the current cow's sample during herd testing.

Sometimes there are major differences between how things are done in experiments and how things are actually done on farm. In the study reported here, the samples were individually collected and not collected through herd testing procedures.

This means that this study did not check if there was an effect of residual milk on the accuracy of this test.

There are 2 ways that I can see this being an issue;

  1. A non-pregnant cow diluting a pregnant cow's sample

  2. A pregnant cow adding PAGs to a non-pregnant cow's sample

Both of these (in theory) would put cows into the "recheck" category so would require an ultrasound check to see if they were pregnant or not.

Overall, it appears that this milk pregnancy test is highly accurate at detecting pregnant and non-pregnant cows.

The next question is... would you use it?
Did you enjoy this research summary? Make sure to subscribe to our email list here (www.agrisciencer.com/subscribe) so you don't miss out on any future posts.

If you liked this post, you may be interested in...

Full Paper:

Lawson BC, Shahzad AH, Dolecheck KA, Martel EL, Velek KA, Ray DL, Lawrence JC, Silvia WJ. 2014. A pregnancy detection assay using milk samples: evaluation and considerations. Journal of Dairy Science. 97(10):6316-6325.

Further reading:



A platform for the discussion of agricultural science that is particularly relevant to the farming sector of New Zealand.  

The two editors, Isabel Vialoux and Rhiannon Handcock are PhD students/employees at Massey University.

This blog represents the views and opinions of Isabel and Rhiannon, not Massey University.

Never miss another post!

Contact us:

© 2020 The AgriSciencer

Proudly created with Wix.com