Information and resources on the “false positive catastrophe” from Covid-19 testing
This post contains quick links to articles and resources on false positive tests examples and issues; understanding the false positives issue explains how badly pandemic policy backfired so badly
I’ve written extensively on how false positives have from the very beginning of the Covid-19 pandemic given the false impression of large numbers of Covid-19 cases, and as a result led to massive backfires in pandemic policy around the world. The false positives issue applies equally to cases, hospitalizations and Covid-19 deaths figures.
Understanding the degree to which “the false positive catastrophe” has led to an exaggeration of the pandemic is essential for ending this pandemic and avoiding the next one too.
It’s particularly important at times of lower prevalence because lower prevalence of the disease in the population renders almost all test positives false positives when testing is done without regard to symptoms (“asymptomatic testing” or “screening”). However, even at times of higher prevalence of the disease (“spikes”) the same issue renders most test positives false positives.
“Diagnostic testing,” which is testing after symptoms are present, gives rise to significantly less false positives than the screening or asymptomatic testing I focus on in this article.
This article presents a number of mainstream news articles and peer-reviewed scientific papers as references for understanding the false positive catastrophe. It also can be shared with people who are seeking to understand what’s happened the last couple of years, or those who question whether the false positives issue has been a serious issue, since we hear frequently that “false positives are rare.” This is a dangerously inaccurate statement and these resources below demonstrate why. False positives are extremely common, not rare.
First, a few examples of how badly false positives have impacted policy choices in specific areas:
Kimberly region in Australia in early 2022 saw two small villages with 55 false positive PCR tests get severe lockdowns, only to find out later that all positive test results were false positives probably due to contamination.
77 positive PCR test results from a number of NFL (American football) teams in August 2020 were all found to be false positives upon re-testing, again probably because of contamination.
In Hawaii where I live, all people arriving on the Big Island had to take an antigen test upon arrival. 93% of these test positives were found, upon re-testing, to be false positives.
Next, I’ll provide some more general warnings about false positive issues for both PCR and antigen tests:
The FDA warned in November 2020 how antigen tests, even if they are 98% accurate in lab testing, can produce 96% false positives with low disease prevalence (the 96% false positives occur when actual disease prevalence in the population is 0.1%, which is about where it has been during much of the pandemic).
This New York Times article from August 2020 quotes a number of experts about finding up to 90% false positive PCR test results due to sky-high “cycle thresholds,” which is the number of times the RNA is amplified in the lab. I wrote an article on this issue specifically and interviewed former Harvard professor Michael Mina about his statements to the New York Times. Anything over about 30 cycle threshold is very often just a false positive.
Harvard Medical School professor Westyn Branch-Elliman and two other academics described in a US News & World Report article from July 2021 how Covid-19 testing in schools could, with 0.1% actual disease prevalence and testing of asymptomatics, lead to 71 out of 72 test positives being false positives. They argue that schools should not engage in such testing because of this extremely high rate of false positives. This very high false positive rate arises due to widespread testing of asymptomatic people and highlights why we shouldn’t be testing people without Covid-19 symptoms.
A Guardian article from April 2021 explained why low background prevalence of the disease at issue will result in huge percentages of false positives, based on Bayesian logic. This issue arises with any kind of test for any disease, and is a widely known issue in epidemiology. But for some unknown reason this extremely important was mostly ignored in the Covid-19 pandemic.
A May 2022 paper in JAMA (Connor et al. 2022) found a 62% false positive rate for antigen test screening for Covid-19 in workplace testing. The actual false positive rate was very likely far higher because these researchers simply assumed that the PCR test used to verify the initial antigen tests, used in the workplace testing, is 100% accurate, but we’ve just seen how inaccurate this assumption can be. This high rate of false positives again arises due to widespread testing of asymptomatic people.
Last, I’ll offer a number of essays and papers written by me and my collaborators on the false positives issue:
An article written by me, which is a shorter version of an academic paper (preprint) by me, Dr. Blaine Williams (an ER doctor in Honolulu), and Dr. Daniel Howard (a Ph.D with training in epidemiology), explain the details of the “false positive catastrophe” in the Covid-19 context.
We also explained these issues in a “rapid response” essay, which is reviewed by the editors, for the British Medical Journal (BMJ) from June 2021.
I did the math on public data from China’s Covid-19 testing programs in Shanghai in early 2022, and found that it was very likely that fully 99.9% of Shanghai’s positive cases were false positives, again due to widespread testing of asymptomatic people. And yet China continues to pursue its insanely damaging “Covid-zero” policies, apparently for purely political reasons.
I and my coauthors Dr. Williams and Dr. Howard, also explained in this article how extensively the false positives issue led to large-scale exaggeration of cases, hospitalizations and Covid-19 deaths data. We estimate that all public Covid-19 stats should probably be reduced by 90% in order to yield a more accurate figure for cases, hospitalizations and deaths.