Comparing NMR Results to Weekly Payment Test Results
From time to time questions are raised by dairy farmers regarding differences between NMR test results and weekly payment test results.
NMR undertakes regular checks between its milk recording laboratory and payment testing laboratories at Paisley and Wolverhampton that verify that when presented with identical samples there is no significant differences in results between the laboratories.
This note provides some explanations as to why there can appear to be discrepancies between testing laboratories. These explanations highlight the importance of ensuring that samples sent to different laboratories are directly comparable, since in the vast majority of instances this is not the case.
1. NMR vs NML: REASONS FOR DISCREPANCIES
1.1 Interpretation of NMR Records
i) Accounting for milk withdrawals
In some instances NMR customers compare the NMR Herd Average result with the payment test result. Whilst the NMR Herd Average is weighted, this average includes milk from all recorded cows, when in fact milk from some of these cows may not make it into the bulk tank.
For example if a herd has cows with high fat %ages, but these cows also suffer from high SCCs the farmer may choose to feed this milk to calves rather than add it to the bulk tank. In this scenario the NMR Herd Average fat result will be higher than the payment test fat result for the same day.
ii) Accounting for factoring
It is worth noting that the constituent measures for NMR factored herds are based on estimates for the missed recording (ie if taking am samples, the pm samples will be based on factors). The factors used in generating these estimates are not always going to be 100% accurate, so for these herds differences in the NMR Herd Average and the NML results will always occur
iii) Reference to the NMR bulk tank sample
Comparing payment test results with an NMR bulk tank result is better than comparing with an NMR Herd Average. However for the reasons described below re method of sampling, time of sampling, age of sample when tested etc etc, differences in the make up of the samples when they reach the two labs will occur.
1.2 Sampling method
The method used for taking the milk sample is a particularly important factor. NMR bulk tank samples are taken by using a dipper to draw a sample from the top of the bulk tank at the end of milking.
With milk tanks getting bigger and bigger, the effectiveness of agitation when the sample is taken is key. It is essential that the bulk milk tank is agitated for a full two minutes before the sample is taken. The nature of SCC's in particular means that they are difficult to distribute completely evenly throughout the milk, and even after careful agitation for an appropriate period there may still be a small variation between samples (particularly if the agitator was inefficient given the volume of milk in the tank).
In addition it should be noted that many milk buyers are now using in-line samplers on the back of their tankers. These generate a sample of milk by drawing a sample of milk gradually from the pipe as the milk is pumped from the farm bulk tank into the tanker. Clearly this method will theoretically result in a more representative sample than one taken manually from the top of the bulk tank, which will result in a difference between NMR and NML.
1.3 Time of sample collection
As the industry rationalises we are seeing fewer milk hauliers operating, but they are working more hours in the day. As a result many NMR customers are now having their milk collected at night rather than during the day.
Clearly the make up of milk does change over time, so comparing NMR results (sampled during the day) with payment test results (sampled at night) can be misleading.
1.4 Age of sample
It is worth noting that depending on conditions there can be some deterioration of cell counts over time. Equally BactoScans tend to rise over time.
NML samples are generally tested sooner than NMR samples. As a result the NMR results may show slightly lower SCC results. (If NMR tested bactoscans it is likely that these might be slightly higher).
It should be stressed that if the samples are stored correctly, pots are sealed properly etc etc, there shouldn't be any significant difference due to time lags unless the samples are well over three days old, but nevertheless the age of sample can provide an explanation for certain discrepancies.
2. THE ACCURACY OF NMR AND NML TESTING EQUIPMENT
It should be noted that the NMR Group undertakes regular ‘inter-lab' trials. These involve sending portions of milk from the same sample to each lab and comparing results. To date these trials have proven that all three labs are generating almost identical results for the same sample, giving us all the confidence we need that we will provide an accurate test of the milk provided in a sample pot at all of our labs.
