Over-the-Counter COVID-19 Antigen Home Test

Across about 650 symptomatic subjects, about 150 were positive by the RT-PCR comparator. The test showed a PPA of about 97% and an NPA of about 98%, with only a handful of false negatives and false positives. One test was invalid and gave a valid result on repeat testing.

By days post symptom onset, PPA was 100% at days 0, 1 and 3, about 98% at day 2 and about 87% at day 4. NPA was at least 94% at every day from 1 to 4, and about 88% at day 0.

The usability study had an overall success rate of about 91%, and most critical tasks reached 100%. The two tasks with lower success, hand washing and discarding the materials, were judged artefacts of the simulated home environment. In the readability study, lay users correctly read about 95% of mock devices: 100% of negatives, about 89% of 1.5X LoD positives and about 96% of 5X LoD positives. Incorrect reads occurred at random across age groups and among participants with and without impaired vision.

In the flex studies, false results appeared only with extraction buffer volumes below the labeled amount and with read times well outside the labeled window, and FDA concluded the test was robust under intended use conditions. FDA found the test substantially equivalent to its predicate and cleared it for over-the-counter home use, with a variant-monitoring plan in place.

The applicable special control under 21 CFR 866.3984 requires the lower bound of the two-sided 95% confidence interval for PPA to be greater than 70%. Here the lower bound sat well clear of that threshold, so the margin did not depend on borderline statistics. Because positives make up the PPA denominator, the about 150 comparator-positive subjects set its precision. The uncertainty at day 0 and day 4 reflects small strata, not a firm drop in performance. Even so, the lower day-4 estimate is consistent with the labeling's instruction to re-test after a negative result.

The NPA figure should be read alongside the discrepant-analysis footnotes. FDA reports the second RT-PCR results as footnotes, not folded into the primary 2x2 table, so the headline PPA and NPA reflect the prespecified comparator alone. The readability results are a separate measure. At 1.5X LoD, lay users read about 89% of devices correctly, against 100% for negatives, which shows how strongly line intensity drives home-test accuracy, independent of the chemistry.

The test is intended for non-prescription home use by individuals aged 14 years or older testing themselves, or by adults testing individuals aged 2 years or older, all with signs and symptoms of COVID-19. Negative results are presumptive, and symptomatic users with an initial negative result must re-test once between 48 and 72 hours later. The clinical evidence therefore had to come from lay users who collected their own anterior nasal swab and ran and read the test using only the labeling provided, not from trained staff.

Performance had to be measured against a highly sensitive 510(k)-cleared RT-PCR comparator on a separately collected swab. Results were needed across the claimed age range and by days post symptom onset, and the study ran while the Omicron variant was dominant, which the labeling must state. Because this test adds a sample port to the cassette, its swab-handling steps differ from those of the predicate. That made it necessary to show, in separate usability and readability studies, that lay users could follow the new workflow and correctly interpret results, including low-positive results.

Over-the-counter SARS-CoV-2 tests are classified under 21 CFR 866.3984, "Over-the-counter test to detect SARS-CoV-2 from clinical specimens," as Class II devices with special controls. The regulation was created through a De Novo classification granted for an over-the-counter molecular SARS-CoV-2 test, after which later home tests, antigen tests included, could reach the market through 510(k) rather than an Emergency Use Authorization. Over-the-counter COVID-19 antigen tests under this regulation are lateral flow immunochromatographic tests for the qualitative detection of SARS-CoV-2 nucleocapsid antigen in anterior nasal swabs from individuals 2 years and older.

The special controls set out the clinical evidence directly. FDA requires a prospective multisite clinical study, run in the intended use setting (at home or in a home-like environment), with participants who represent the intended use population across the clinical range of the analyte. Sites must be geographically diverse so that the circulating strains are represented, and results must be compared against a molecular comparator method FDA considers appropriate. The study must be large enough that the lower bound of the two-sided 95% confidence interval for positive percent agreement is greater than 70%, with added risk mitigations such as presumptive negative results and serial testing. Applicants must also submit usability, label comprehension and flex studies, a variant-monitoring protocol, and labeling that names the variant that was dominant during the study.

Home antigen tests are hard to validate because every step that a laboratory would control is in the hands of a lay user. That person collects the swab, adds the buffer, times the read and interprets a faint line. FDA's guidance on designing devices for home use treats the interaction between a lay user, an uncontrolled environment and the device as a distinct source of risk. Low-positive specimens and visual interpretation are where these tests usually fail.

• Running self-sampling and self-testing in a simulated home setting with labeling only met the special-control requirement that the study take place in the intended use setting. The performance estimate therefore already includes errors in collection and interpretation.
• Using study staff to collect the comparator swab, randomizing swab order and spacing collections 15 minutes apart controlled for the first swab depleting the specimen. It also kept the reference result independent of how well the lay user collected.
• Enrolling within 4 days of symptom onset and reporting PPA for each day from 0 to 4 supported the regulation's requirement to state the validated days-post-symptom-onset window in the labeling.
• Including adult-tested children aged 2-13, older adults and a wide range of education levels made the population representative of the intended use, which the special controls require.
• The readability panel deliberately included 1.5X LoD devices, and half the participants had impaired vision. This tested interpretation of faint lines, the step where visually read home tests are most likely to fail.