A Kit ID Is Not a Patient Identifier Until Someone Links It

7 min read

In a clinic, a phlebotomist labels the tube in front of the patient. In an at-home test kit, the tube is labeled weeks earlier on a kit build line, before anyone knows whose blood, saliva or stool will go into it. That label carries a kit ID, not a patient identifier. It becomes one only when someone links it to a person, and every at-home program depends on that link holding.

Specimen identity in a home collection program is built in three places: the build line, the registration step and laboratory accessioning. Each fails in a different way, and the most expensive failure is the one nobody sees until the lab cannot match a tube.

What the laboratory rules actually ask for

Under CLIA, the testing laboratory must have written procedures for specimen labeling, “including patient name or unique patient identifier,” and written criteria for specimen acceptability and rejection (42 CFR 493.1242(a)(3) and (a)(7)). The test requisition must carry the patient’s name or a unique patient identifier and the date of collection (42 CFR 493.1241(c)(2) and (c)(6)). Accrediting bodies go further. The Joint Commission’s patient identification goal calls for at least two patient identifiers and for labeling specimen containers in the presence of the patient.

A home kit cannot meet the second part as written, because no trained collector is present. Programs work around it by treating a unique kit ID plus a registration record as the identifier, and often by having the participant write their name and date of birth on the tube as a second identifier. Whether that scheme satisfies a given laboratory’s labeling procedure is the laboratory director’s decision, not the kit manufacturer’s. Get it in writing before the first build.

Three links, three different failures

The distinction between systematic and random matters more than the rate. A random registration error costs one specimen. A label sequence error on the build line can cost thousands, and every kit it touches looks perfect until it reaches the lab.

Link one: the build line

Most kits carry the same ID in at least two places: the primary tube and something else, such as a requisition card, the kit box or the return bag. Those labels are usually printed as matched sets from one serial number. The failure is simple. A roll gets loaded one position off, a sheet is skipped, or two label stacks are pulled from adjacent serial ranges. From that kit forward, the tube says one number and the card says the next one.

The controls are equally simple, and they work on the mechanism rather than on inspection volume:

• Print paired labels in one pass from one record. If the tube label and the card label come off the same print job for the same serial, a roll offset cannot separate them.
• Verify the pair against each other, not against a list. A scan that confirms the tube ID equals the card ID catches a slip on the first bad kit. A scan that confirms each ID exists somewhere in the lot does not.
• Clear the line between serial ranges. Leftover labels from the last range are how two adjacent ranges end up in one kit.
• Reconcile at the end of the lot. Labels issued, labels applied, labels destroyed and kits built should close to zero. If they do not, find the gap before the lot ships. Label rolls and printed cards also run short more often than any other component, which is covered in why a 5,000-kit build needs more than 5,000 of each part.
• Send the lab the list. The lot record should produce a file of every kit ID built, so the registration system and the laboratory know which IDs exist before any specimen arrives.

Link two: registration

Registration is where the kit ID becomes a person. Usually the participant types the ID into a web page or app, or a clinic enters it into an ordering portal. That makes it a human transcription problem, and transcription errors are well studied. In the classic study of human copying errors summarized by Chen, Niemenmaa and Vinck, single-digit errors made up about 79 percent of all errors and adjacent transpositions (writing 47 as 74) about 10 percent.

A check digit is the inexpensive fix. The Luhn check digit, the same scheme used on payment card numbers, detects every single-digit error and every adjacent transposition except 09 and 90. Add one to the end of the kit ID and the registration page can refuse nearly nine in ten of the expected typing errors before they create a record. A QR code on the kit that opens the registration page with the ID already filled in removes most typing altogether.

The other registration failures are not typos:

• Never registered. The specimen arrives with no record. The lab holds it while someone tries to find the owner, and stability windows keep running.
• Registered twice, or registered by the wrong person. Common when a household receives two kits. The fix is a registration step that shows the participant the last four characters of the ID and asks them to confirm it against the tube.
• Collection date missing. CLIA requires it on the requisition. Capture it at registration, or print a write-in field on the tube label and the instructions.
• Ambiguous characters. If the ID is alphanumeric, drop O, I, L and similar characters that read as digits.

Link three: accessioning

Accessioning is where every upstream decision gets tested. The laboratory matches the tube to a requisition and checks it against written rejection criteria, as described in specimen return, from a home mailbox to laboratory accessioning. For identity, the questions to settle with the lab before launch are narrow: what happens to a specimen whose kit ID is not registered, how long it is held, who contacts the participant, and whether a handwritten name on the tube that disagrees with the registration record is a rejection.

Some context on rates. A College of American Pathologists Q-Probes study of more than 3.3 million specimen labels in 147 laboratories found labeling errors at 0.92 per 1,000 labels, with earlier reports ranging from 0.1 to 5 percent. Those specimens were collected by trained staff. At home there is no trained collector, so the kit and the registration flow have to carry more of the work.

A worked example

Illustrative numbers. A program builds a 10,000-kit lot. On the build line, a card roll is reloaded one position off at kit 6,201 and the pair check compares each label to the lot list instead of to each other. Kits 6,201 through 10,000 ship with a tube ID one higher than the card ID: 3,800 kits, all of them looking correct.

On the registration side, assume 2 percent of participants mistype their ID, which is 200 errors across the lot. Without a check digit, most of those create a record against the wrong kit or no kit. With a Luhn check digit, about 178 are refused at the keyboard and corrected on the spot, leaving roughly 22 for the lab to sort out.

The registration errors are a nuisance. The label slip is a recall-scale event. That is why the pair check belongs on the build line and the check digit belongs in the ID format, and why both have to be decided before the first label is printed.

What to settle before the first build

• Who owns the ID scheme. The laboratory, the program sponsor or the kit manufacturer. One owner, written down.
• The format. Length, character set, check digit, and whether it encodes anything (it usually should not).
• Where the ID appears. Tube, card, box, return bag. Every additional location is another pair to verify.
• The second identifier. Whether the participant writes name and date of birth on the tube, and what the lab does when it disagrees with registration.
• The data handoff. A kit ID file per lot to the registration system and the lab, and a reconciliation report back.
• The unregistered-specimen rule. Hold time, contact process and rejection point, in the lab’s written criteria.

Specimen identity in at-home testing is not a labeling problem or a software problem. It is three links built by three different parties, and it holds only if someone owns all three. Decide who that is before the label artwork goes to the printer.

Sources: 42 CFR 493.1242, Standard: Specimen submission, handling, and referral. 42 CFR 493.1241, Standard: Test request. The Joint Commission, National Patient Safety Goals, Laboratory Services, NPSG.01.01.01. Wagar EA et al., Specimen labeling errors: a Q-Probes analysis of 147 clinical laboratories, Arch Pathol Lab Med 2008;132(10):1617-1622. Chen Y, Niemenmaa M, Vinck AJH, A general check digit system based on finite groups, Des Codes Cryptogr 2015. University of New Brunswick, Secrets of the LUHN-10 algorithm.