UNH COVID False Positives? Probably Not. Unreasonable Panic? Maybe.

[I've broken this out into a separate article because I slag my former employer for so many other reasons. And it's one of those rare occasions where I've come closer to Actual Journalism than my usual arrogant opinionizing.]

I was intrigued by this Greg Piper article at the College Fix: False-positive COVID scandal rocks Harvard, but student paper doesn’t ask for ‘positive’ threshold. Here's the problem:

The PCR tests most commonly used to test for the novel coronavirus are idiotically sensitive, to the point where they can catch dead virus or otherwise insignificant viral loads. This means even a “positive” test indicates a non-infectious person – someone who shouldn’t be forced to quarantine, much less wear a mask. (Reminder that people without symptoms are extremely unlikely to transmit virus even in the same home.)

I winced a little at Greg's "much less wear a mask" wording, implying that masking is a more drastic measure than quarantining. But I get the point: a low viral load means you're probably not going to infect someone, and there's a good chance you're immune yourself. At least for a while.

Greg points to this New York Times article (from August): Your Coronavirus Test Is Positive. Maybe It Shouldn’t Be. It details how the PCR test can, depending on the "cycle threshold" used, generate a positive result for very small, probably negligible, viral loads. See above: there's no need to panic in such cases. But (since only a yes/no result is returned), people panic anyway.

And then Greg points to a February 12 Harvard Crimson article: Updated Lab Protocols Invalidate Positive Covid-19 Test Results for More Than Two Dozen Harvard Affiliates. Indicating that the smart folks at Harvard may have made the exact blunder the New York Times detailed back in August.

So that's Harvard. But the University Near Here recently got its students back on campus. And near-immediately detected a spike in COVID cases, i.e., positive test results. Causing a shift to online classes only.

Which caused me to wonder if UNH was making the same mistake as Harvard. Hm.

UNH has an address for people to ask COVID questions. Even though I'm no longer affiliated other than my (now useless) employee-emeritus library card, I took a chance:

After reading a recent New York Times article (https://www.nytimes.com/2020/08/29/health/coronavirus-testing.html) I was wondering if UNH uses the PCR test for Covid, and if so, what "cycle threshold" is used to determine a positive result?

I didn't have a lot of hope on getting a response. But near-immediately:

The UNH Durham lab does use PCR testing. There are actually several different CT thresholds we use in a complicated method, so there isn't a single CT value.

OK… well, nice try. I thanked my correspondent for his answer:

Thanks very much for the response. As a retired UNH employee, I should have known it wouldn't be simple.

If there's publicly-available documentation for the complicated method, I wouldn't mind a pointer, but if there's not, no worries.

Best wishes!

And I really didn't expect a response after that. But I got one anyway:

I don't think there's anything publicly available, but I was able to find some more details for you:

The design of the UNH's testing strategy is unique in that a positive result is analyzed multiple times so false positives are very rare, included a pooled and unpooled test. In addition, if a positive is detected using our surveillance testing, that person is then called into to Health Services to get another swab, this time under CLIA regulations, specifically that the specimen is collected by a trained professional. That CLIA-certified swab is never pooled.

In summary this testing strategy of doing 3 different PCR runs for each positive and the fact that it relies on two separate swabs of a person results in a much lower false positive rate than most labs. Regarding Ct values, the lab analyzes both control and multiple regions of the viral genome. The Ct value is a relative measure of the concentration of target in the PCR reaction. The Ct value is not a lab specific number, nor is it absolute. The COVID method is qualitative (yes/no for presence of virus) not quantitively. So, since the experiment is not designed to be quantitative, Ct values are not clinically relevant and are not reported.

So, good news and bad: UNH understandably worries about people bungling their self-test in a way that causes false positives. But they don't seem too worried about the false positive concern expressed in the NYT article.

Good on UNH for being open about this. They seem confident that they're not getting false positives. Whether they are unnecessarily labelling people with negligible COVID viral loads as dangerous? I don't know.

There's also the possibility that UNH's assertion that "Ct values are not clinically relevant" is totally correct, according to current best practice, and the August NYT article is totally misguided.

I'll keep my non-virologist eyes open.


Last Modified 2021-02-19 7:39 AM EST