• Decrease font size
  • Return font size to normal
  • Increase font size
U.S. Department of Health and Human Services

Animal & Veterinary

  • Print
  • Share
  • E-mail

FDA Validates Rapid Screening Tests for Antibiotics in Milk

by Philip James Kijak, Team Leader, Analytical Methods Team, Office of Research
FDA Veterinarian Newsletter July/August 2004 Volume XIX, No IV

This article is based on a presentation the author made at the 2004 Mid-Atlantic States Conference for Bovine Practitioners sponsored by the Maryland Veterinary Medical Association on March 25, 2004.

The ability of regulators in the United States to test every load of milk sold for the presence of antibiotics is a complex regulatory and technical achievement, supported by the Food and Drug Administration’s (FDA) ability to test the milk screening test.

A national conference made up of State and Federal food regulators initiated the testing requirement in 1991. The regulators, along with representatives of the farmers, dairy industry and consumer groups, are organized as the National Conference on Interstate Milk Shipments (NCIMS). The purpose of the NCIMS is to develop regulations used by the States for Grade A milk and milk products in interstate commerce. The NCIMS developed the requirement that all tankers of milk in the United States be screened for residues of penicillin and other beta-lactam antibiotic drugs.

The only way to comply with that requirement, while at the same time not unduly delaying the delivery of milk, is through the use of highly accurate rapid screening tests. The United States had many commercial tests available, but did not have any program in place to determine whether these tests were suitable for use in a regulatory program. This is where FDA plays its role. NCIMS requested FDA to develop a program to validate these rapid screening tests for regulatory use.

FDA developed a validation program for test kits through a cooperative effort with the AOAC Research Institute. AOAC International, formerly known as the Association of Official Analytical Chemists, operates its Research Institute to provide independent certification on the performance of various commercial rapid screening tests. Under this program, FDA uses testing both by the kit manufacturer and an independent laboratory to determine the suitability of the kits for regulatory use.

In order to be considered for regulatory uses, a test kit must be able to detect—at or below the legal tolerance (safe level)—four of the six beta-lactam type antibiotic drugs commonly used in dairy cows. The six beta lactam drugs are ampicillin, amoxicillin, ceftiofur, cephapirin, cloxacillin and penicillin G.

In addition, all new tests must be capable of giving a printed record that includes the sample identification, date, time, operator, kit lot and result. When the program was first started, FDA would accept tests that required an operator to visually interpret the results and determine whether the milk was safe. Problems found with the use of the visually read tests led to the requirement in the late 1990s that all new test must be read by an instrument.

Sensitivity, selectivity

If a test can meet the preliminary requirements, then the primary focus of the validation is a test kit’s sensitivity and selectivity.
Sensitivity relates to the possibility of false negatives and selectivity to the possibility of false positives.

Sensitivity is the ability to detect a specific beta-lactam drug in milk. To calculate a test kit’s sensitivity, the independent laboratory tests a statistically significant number of samples of the test kit over a range of drug concentrations up to the tolerance level. The researchers are trying to determine the concentration of the drug at which the kit gives a positive result 90 percent of the time with 95 percent confidence (90/95). In other words, the test must be correct with 90 percent of the samples 95 percent of the time. The researchers run this test with each type of beta-lactam drug that the kit should be able to detect.

Selectivity is determined by the response of the kit to truly drug-free milk. To be acceptable, a kit must not give more than two positive readings for 60 known negative samples.

Researchers then do additional studies to be sure the test kits work properly when used in the field. One study is designed to determine the ruggedness of the kit. It evaluates the effect of slight changes to operating conditions, such as specified temperature, volumes and times, that would be expected under typical use of the test kit. Another study is designed to find out if the kits will give false positives or negatives when other veterinary drugs that might be used in dairy cows are in the milk. Additional studies test the performance of the kit when high levels of somatic cells or bacteria are present in the milk.


The kit manufacturer must include information that the researchers gather during the validation tests about sensitivity, selectivity, drug interference and other key findings in the instructions for use included with the kit, referred to as the kit label. This information makes the kit label the most important resource to the user when determining the appropriate kit to use.

The label includes both the calculated 90/95 concentration for each drug claimed and information on the response of the kit at specific drug concentrations. Both pieces of information are important in evaluating the sensitivity of a kit to a specific drug.
In the example in Figure 1, the concentration response curves to amoxicillin for two test kits are shown. The calculated 90/95 concentrations for the two kits are almost identical.

Yet, at low concentrations, the kit described by triangles gives a significantly greater percentage of positive samples.
The difference in sensitivity at low concentrations is documented in the label information. The respective sections of the label for each kit are shown in the second figure.

By reading and using the label information, the test kit users can make an informed decision about the suitability of a kit for their application. For example, if a milk producer wants to screen the bulk tank before pickup to ensure that it is not positive, the producer would want the most sensitive test possible.

Additional information required to be on the kit label includes a list of drugs known to cause either false positives or false negatives when present in the milk, information on the selectivity of the kit, and any other potential limitations to the performance of the kit that were discovered during the validation process.

Retest programs

The user of the test kit needs to be aware that the kits do have limitations.

Because the 90/95 concentration must be at or below the tolerance/safe level, the kits will always have the potential to give a false violative result, meaning that, although the drug is present and the kit indicated a violation, the concentrations is not at a level that would be in violation of the safety standards.

To minimize the consequence to the milk producer from a test kit’s false positive results, NCIMS calls for two retests of a sample before the milk is condemned. The initial retest is done using the same test as the one used for the initial screening. This retest is done in duplicate with the positive sample and a negative control sample. If either of the duplicate tests gives a positive result and the results for the positive and negative control are correct, the tanker load is a “presumptive positive.” Then a second retest is done.
This second retest is also done in duplicate along with positive and negative control samples. The second retest sample must be tested in a State or State certified laboratory, and may be done using a different test kit. If either of the second retests is positive, the result is called a screening test positive, and the milk considered to be adulterated with beta-lactam residues.

The effect of the retest program is to greatly decrease the likehood of false positive results. For example, if a kit had a false positive rate of 1 in 1,000, the probability of a negative sample being positive for both the initial screen in the first retest is 1 in 500,000. If the same test is used for the second retest, the chances are only 1 in 250 million that true negative milk will be declared screening test positive.

The retesting also decreases the probability that milk with a beta-lactam drug present will test positive when the antibiotic’s concentration is below the tolerance level. However, the effect is largely dependent on the concentration of the drug in the milk. At drug concentrations where the test kit usually gives a positive result, a negative retest is highly unlikely. But as drug concentration in the milk decreases to the point where the kit gives negative results, the probability that the retest will give a negative result also increases.

No information about amount of drug

Even though modern kits present printouts that typically include numeric results, they often do not offer information about the amount of drug present. This printout does not actually give a good idea of the level of drug present, because of the test-to-test variability of most test kits.
Figure 3 shows some results obtained of a test kit’s response to amoxicillin. The kit was functioning properly, and all negative results are well below the zero line. And at drug concentrations at tolerance and above, the result is always positive. However, there is a great range of actual readings obtained at any single concentration. For example, several of the high readings at 5 parts per billion (ppb) are well within the typical range of readings obtained at 12 ppb. If a user were to test milk with this test, and get a reading of 4 ppb, the drug concentration in the milk could be less than 5 ppb, greater than 12 ppb, or somewhere in between.
The sensitivity of the kit to each drug that the kit can detect for is different. This variability prevents the use of the test kit’s numeric readout to determine the drug concentration in the milk.


The screening tests are meant to be fast and accurate. They are not meant to supply complete information about the potential of antibiotics in milk.

For instance, most kits will not test for all six beta-lactam drugs. And, in most cases, the tests do not provide information on what drug caused the positive result.

Still, the screening tests fulfill their principal duty—keeping the milk supply in the United States safe—while not slowing down delivery of fresh milk to consumers.

Figure 1. Concentration response curves of two test kits to amoxicillin showing
the difference in the kits’ response. (Ref chart on PDF)

Figure 2. Examples of how the drug concentration response information shown in figure 1 is listed on the test kit label for both test kits. (Ref following table)


Drug Concentration(ppb)AmoxicillinDrug Concentration (ppb)Amoxicillin
10 47 
83 97 
10 100 10 100 
14  14  
20  20  
 Tolerance Safe Level (ppb)10 Tolerence Safe Level (ppb) 10 
90/95% Concentration (ppb) 7.5 90/95% Concentration (ppb)7.7

Figure 3. Results of test kit’s response to amoxicillin showing why screening tests cannot be used to show drug concentration. (Ref chart on PDF)