Open in app

Sign in

Write

Sign in

Which High-Quality Covid Test Should You Buy?

Nicholas DiBella
6 min readNov 17, 2023

--

It is well-documented that “molecular” home Covid tests like Lucira and Cue are substantially more accurate than RATs (Rapid Antigen Tests). These molecular tests are also substantially more expensive. Recently, two additional molecular tests became available: Metrix and 3EO. With all these high-quality tests now on the market, which should you buy?

Below is a table that summarizes what I think are the most important factors to consider when deciding which to buy. I won’t make a personal recommendation, but hopefully the information here will help you to decide which is best for you.

Some notes:

1. I thought it was important to include the dates of the clinical studies since the evolution of variants over the years could affect clinical performance (as we’ve seen with the RATs since the emergence of Omicron). While each of the above companies claims that their test can detect all circulating variants, it’s another matter to say that their test can equally reliably detect those variants. That said, all of these companies say that they regularly monitor their primers to ensure that they cover new circulating variants, though it’s impossible to know whether their clinical sensitivities have changed since their respective trials were conducted. However, my understanding is that these tests are generally intended to target multiple SARS-CoV-2 gene targets that are unlikely to mutate.

2. I thought it was important to note whether the studies included asymptomatic infections, since asymptomatic infections might be more difficult to detect than symptomatic infections. For example, in the clinical study of the original Lucira “Check It” test (now discontinued), the sensitivity for symptomatic infections was 94% but only 90% for asymptomatic infections. That said, I have no doubt that the new Lucira Covid & Flu test still works for detecting asymptomatic infections, though it might be slightly less sensitive for that than the figure cited above. Similarly for Cue and 3EO. (Also, it’s worth noting that the difference in sensitivities for symptomatic infections vs asymptomatic infections for the original Lucira Check It test was not statistically significant.)

3. The numbers in the above table come from the most recent study each company submitted to the FDA (see links for the studies below). In particular, this includes an updated study Cue submitted to the FDA to get fully approved (through the “De Novo” classification) that included many more participants than the original study it submitted to obtain Emergency Use Authorization (EUA). The original Cue study (completed in early 2021) reported a sensitivity of 97.4% and included both symptomatic and asymptomatic infections; the updated Cue study reported a sensitivity of 92.9% and only included symptomatic infections. Most of the data from the updated Cue study come from the Nov 2021-Feb 2022 period and was therefore collected during the Omicron period. Cue is now (since June 2023) the only brand that has obtained full FDA approval; all of the other brands currently only have EUAs.

UPDATE (May 2024): The FDA has issued a warning against using Cue tests, and the company that makes the tests will be shutting down.

4. I left out the data for specificity (probability the test is negative if the person isn’t infected), as each test had a measured specificity between 98%-100%. Also, if you’re a Covid-cautious individual, then you probably care more about the probability of false negatives (more related to sensitivity) than the probability of false positives (more related to specificity).

5. The above intervals for sensitivity are 95% confidence intervals (95% CIs). I thought it was important to include the confidence intervals since there are no statistically significant differences among the measured sensitivities of the different brands of test (at least, not at the 5% level).

6. The sensitivity listed above for Metrix is the sensitivity when you use a nasal swab for their test. They also say you can use a saliva sample with their test (which none of the other companies say you can do, though of course many people do anyway), and officially they say the saliva sensitivity is 90.3% (95% CI: 80.5%-95.5%). However, this number is more accurately called “positive percent agreement (PPA) with a nasopharyngeal PCR” rather than “sensitivity”. PPA means the percentage of time that a person who tests positive on a nasopharyngeal PCR (i.e., the standard type of PCR test that goes up your nose) also tests positive using a saliva sample with Metrix. That said, the Metrix folks found that using a saliva sample with Metrix has a higher PPA — namely, 96.6% (95% CI: 88.3%-99.1%) — with a saliva-based PCR test (SalivaDirect). This number is not provided in the FDA report linked below but was kindly communicated to me for informational purposes when I contacted the company. I was also told that all positive samples detected by the saliva-based PCR were also detected by the nasopharyngeal PCR.

7. In fact, none of the above numbers are actually sensitivities. All of them are PPAs with a nasopharyngeal PCR. The FDA requires that the PPAs are determined by comparison with a high-sensitivity PCR test.

8. “Limit of detection” refers to the minimal amount of viral material that the test can reliably detect (at least 95% of the time) in the lab. For Lucira and Metrix, this is reported in genome copies per mL of viral transport medium (VTM) — i.e., the solution used to preserve a sample from a nasal swab. While 3EO reports a limit of detection in copies/mL, it doesn’t specify whether it is copies/mL in VTM or in some other solution. Also, in its most recent study, Cue reports its limit of detection in altogether different units — specifically, as 20 copies per wand — which makes it difficult to compare to the other tests. (The earlier Cue study also reports this limit of detection and additionally reports a limit of detection of 1300 copies/mL. However, it doesn’t specify if this is in copies/mL of VTM.) Each test is capable of detecting smaller amounts of the virus, but it does so less reliably. I thought including the limits of detection could help us get a more comprehensive picture of the different tests (especially since the differences in their measured sensitivities were not statistically significant).

9. I am not entirely sure about Lucira Covid & Flu’s limit of detection because the FDA report linked below only reported its value in copies/swab (which was 1090) rather than in copies/mL. However, the FDA report for the older Lucira Check It test indicated a limit of detection of 2700 copies/swab and specified this was equivalent to 900 copies/mL. So, I assumed the same conversion factor for the Lucira Covid & Flu test. (Thanks to Medium reader M for correcting a mistake I previously made on this point.)

10. I couldn’t find more precise clinical study dates for Lucira Covid & Flu.

11. Metrix has bundle discounts: 2 tests and a discounted reader for $85, 12 tests and 1 free reader for $300.

12. These are the prices as of January 7, 2024.

13. Links to test data:

14. Product links:

Please let me know if I’ve made any errors or if there’s anything else you think would be useful to add.

UPDATE (July 2024): I’ve corrected the limit of detection for Lucira (thanks to Medium reader M). I previously had it at 1092 copies/mL rather than 363 copies/mL.

UPDATE (May 2024): The FDA has issued a warning against using Cue tests, and the company that makes the tests will be shutting down.

Major Updates (late December 2023):

  • Included updated Cue data in table
  • Corrected limit of detection data in table
  • Included data about Metrix saliva test in point 6
Covid-19
Covid 19 Testing
Covid 19 Update

--

--

Nicholas DiBella

Written by Nicholas DiBella

5 Followers

Philosopher, ex-physicist

Help

Status

About

Careers

Press

Blog

Privacy

Terms

Text to speech

Teams