What accuracy figures mean

Sensitivity and specificity

FDA's statistical guidance on diagnostic tests defines sensitivity as the proportion of subjects with the target condition in whom the test is positive, and specificity as the proportion of subjects without the target condition in whom the test is negative. For a drug screen, "target condition" means the drug is really present at or above the cutoff, as shown by a reference method such as mass spectrometry.

Both figures only mean something together with the study design: which samples, how many, which reference method, and which cutoff. The 2022 study on our research page shows why: one device had 100% overall sensitivity but 50.0% overall specificity in that study.

Percent agreement

The same FDA guidance uses positive percent agreement, negative percent agreement and overall percent agreement when a test is compared with something other than a reference standard. It says that in that situation the terms sensitivity and specificity are not appropriate, because the estimates are "not of accuracy but of agreement".

Many drug-test studies report agreement with an expected result for prepared samples. In the FDA decision summary for K091588, lay users reading the test card agreed with the expected amphetamine result in 99.1% of readings in total, but in 94.4% at 75% and at 125% of the cutoff. Both numbers are true; neither is a real-world accuracy rate.

Cutoff

A cutoff is the concentration, usually in ng/mL, at or above which a test is designed to read positive. Below it, the test is designed to read negative even if some drug is present. Cutoffs differ between products for the same drug: amphetamine is 1,000 ng/mL in K091588 and 300 ng/mL in K091612, so their results cannot be compared directly.

Near-cutoff behaviour

Immunoassay screens are least reliable close to their cutoff. FDA decision summaries therefore report results at fixed fractions of the cutoff. The K091588 lay-user study used six levels (PDF page 6):

In that study every reading at 0%, 50%, 150% and 300% matched the expected result for every drug; all of the disagreements were at 75% and 125%. In the K091612 point-of-care study, between 9 and 13 of the 30 determinations at 75% or at 125% of the cutoff gave the "wrong-side" result, depending on drug and format (for example, 10 of 30 amphetamine dip-card determinations at 75% read positive). A total-agreement figure that mixes many easy samples with a few hard ones will always look high.

Cross-reactivity

A screen can respond to substances other than its target. Decision summaries report the concentration of a related compound that gives the same response as the cutoff. For the K091612 amphetamine strip (cutoff 300 ng/mL D-amphetamine), FDA's summary lists 850 ng/mL for D,L-amphetamine and 650 ng/mL for MDA, while ephedrine gave no equivalent response below 100,000 ng/mL (PDF page 7). FDA's home-test page also warns that some over-the-counter medicines can give the same result as amphetamines.

Sample type (matrix)

Figures apply only to the sample type that was studied. The K091588 summary says a matrix comparison was "not applicable" because the assay is intended for urine samples (PDF page 9). A urine result says nothing about how a saliva test performs, and the reverse.

Detection window

Even a perfect test misses a drug that is not in the sample yet, or no longer. FDA notes that it takes time for drugs to appear in urine, saliva or other samples, and publishes general detection-time guidelines for urine home tests that vary with how much and how long a drug was used and with each person's metabolism. See the FDA table on our how to choose a test kit page.

What a preliminary positive and a negative mean

Preliminary positive

A positive home screen is preliminary. FDA advises sending the sample to a laboratory for a second test and taking no serious action until the laboratory result is back. Foods, supplements, beverages and medicines can affect results, and a positive for a prescribed medicine does not on its own mean misuse.

Negative

A negative applies only to the substances tested, at the test's cutoffs, within their detection windows. FDA says no drug test of this type is 100% accurate; a negative can happen if the test looked for the wrong drug, the sample was collected too early or too late, or the test was stored incorrectly or used after its expiry date. A negative does not establish that no drug was used. More on confirmation testing.