In March 2020 we began looking into the phenomenon of false positive results in PCR tests for SARS-CoV-2. Remarkably, at that time the occurrence of false positives in these tests and the potential for serious impacts appeared to be universally ignored by public health authorities across the globe, even though the problem of false positives in PCR tests had long been recognized. We recommended that when prevalence is low, as in the testing of asymptomatic persons, positive results should be checked with further tests.
Over time a few national and international health authorities recognized the problem and recommended some limited retesting of positive results: Norway in late May 2020 (recommended retesting of positive results for asymptomatic persons with no known exposure), Australia in June 2020 (recommended retesting of weak positives in a low prevalence setting or a low risk asymptomatic population), and the WHO in September 2020 (appears to recommend retesting of weak positives, though it’s a little unclear).
Our investigation into false positive rates led to further insights: that the clinical false positive rate in PCR tests is highly variable, and — contrary to conventional wisdom — may vary with prevalence; and that the failure to incorporate false positives in analyses, along with other factors, produced overestimates of the asymptomatic ratio in SARS-CoV-2 infections.
Below are links to our work, press coverage, and some related papers.
Diagnosing SARS-CoV-2 Infection: The Danger of Over-reliance on Positive Test Results
Andrew N. Cohen, Bruce Kessel and Michael G. Milgroom
medRxiv 28 September 2020 (Version 1 posted 1 May 2020)
Summary: Contrary to the practice in previous epidemics, in addressing COVID-19 health authorities have treated a single positive result from a PCR-based test as confirmation of infection, irrespective of signs, symptoms and exposure. This is based on a widespread belief that positive results in these tests are highly reliable. However, evidence from external quality assessments and real-world data indicate enough a high enough false positive rate to make positive results highly unreliable over a broad range of scenarios. This has clinical and case management implications, and affects an array of epidemiological statistics, including the asymptomatic ratio, prevalence, and hospitalization and death rates, as well as epidemiologic models. Steps should be taken to raise awareness of false positives and reduce their frequency. The most important immediate action is to check positive results with additional tests, at least when prevalence is low.
Preprint and Supplemental Material
SARS CoV 2 Mass Testing Endangers Residents of Long-Term Care Facilities
Andrew N. Cohen and Bruce Kessel
SSRN 21 July 2020
Summary: Mass-testing for SARS-CoV-2 with RT-PCR has been widely implemented at nursing homes and assisted-living facilities in the U.S. and Canada. If prevalence is low, as is often the case in mass testing, and if false positive rates are similar to rates in similar RT-PCR tests, a large proportion of the positive results will be false. We here report numerous records of false positives at such facilities. Residents that test positive are routinely sequestered with other residents that tested positive. This exposes uninfected, false-positive-tested individuals to infected individuals, thereby increasing the risk that they will become infected with COVID-19. Testing meant to protect the frail and elderly may instead be endangering them. Requiring confirmation of positive results with a second test would eliminate most false positives. A few authorities have begun to implement this measure; all should.
Analysis Should Address Test Specificity/Sensitivity, and Adequate Assessment of Asymptomatic Status (posted 7 June 2020)
Problems with Review: False Positives; Inadequate Longitudinal Study; Overly Narrow Symptom Definitions; Poor Evidence of Asymptomatic Transmission (posted 2 July 2020)
Andrew N. Cohen and Bruce Kessel
Annals of Internal Medicine 174(2): 284-285 (February 2021)
Summary: We commented on a review that concluded that asymptomatic persons (meaning individuals that developed no symptoms over their entire infection period) account for 40-45% of SARS-CoV-2 infections, and that transmission by asymptomatic persons is a significant factor in the spread of COVID-19. We noted that the review and the studies it relied on failed to consider the potential for false positive test results, such that the asymptomatic SARS-CoV-2 carriers counted by the studies may not have actually been carriers. It also failed to address the necessity of screening individuals for all significant COVID-19-like symptoms and to screen them over a long-enough period, so that the asymptomatic carriers counted in the studies may not have been asymptomatic. Finally, the evidence of asymptomatic transmission reported by the review is much too weak to support its conclusion. We don’t claim that there are no asymptomatic carriers or asymptomatic transmission, only that the reported evidence is insufficient to show that a large proportion of infected individuals are asymptomatic or that asymptomatic transmission is a major factor in the spread of COVID-19.
Review Article, Online Comments, and Authors’ Response (posted 15 July 2020)
Published Comments, and Authors’ Response (February 2021)
Why Don’t Seniors Get the Same COVID-19 Tests as Athletes?
Original title: We protect professional athletes from testing errors. Why not the elderly?
Bruce Kessel and Andrew N. Cohen
Orlando Sentinel July 31, 2020
Summary: We protect professional athletes against testing errors, by checking positive results with further tests. Why don’t we provide the same protection to the elderly and the frail, for whom a false positive result can be life-threatening?
Single Coronavirus Pretest will not Protect Tourism Workers
Honolulu Star-Advertiser September 29, 2020
Summary: To protect Hawai’i against a problematic number of infected visitors, arriving travelers should be tested both before arrival and on arrival: pretest and retest.
Talk Ten Tuesdays
False positives in SARS-CoV-2 PCR tests are explained in three podcasts and accompanying articles.
1. False Positives in PCR Tests for COVID-19 November 9, 2020
2. Impacts of False-Positive Results in COVID-19 Tests November 16, 2020
3. Addressing the Problem of False-Positive PCR Results in COVID-19 Tests November 23, 2020
Why is a marine biologist investigating false positives in COVID-19 testing?
This followed a decade of investigating false positives in similar tests used in environmental monitoring: see What Mussels Can Teach Us, reported by Richard Harris on NPR’s Weekend Edition for June 14, 2020.
Did retesting positive results, as we recommended, keep the Tour de France on course?
Forewarned by some apparent false positive results in earlier races, which under the Tour rules could easily have disqualified entire teams, the organizers decided on the day before the Tour that they would retest all positive results in order to reduce the incidence of false positives.
When is a Positive Not a Positive? COVID-19 and the Tour de France. Cycling News, August 27, 2020
Tour de France to Carry Out Secondary COVID-19 Testing to Avoid False Positives. Cycling News, August 28, 2020
False positives and Parliament
When results from our analysis of the potential range of false positive rates were summarized in a UK government report and then cited in Parliament, the Science Editor at The London Times asked for clarification.
Coronavirus Analysis. The Times, September 21, 2020
Johann Holzmann asked about an inconsistency between the median false positive rate derived from our review of 43 external quality assessments of similar tests and the low positivity rate reported in certain countries. This argument, in various forms, is the sole substantive objection that has been raised regarding our analysis of SARS-CoV-2 false positive rates.
Read his question and Andrew Cohen’s answer
In March, Dr. Guihua Zhuang and eight colleagues published a paper in the Chinese Journal of Epidemiology that attempted to raise awareness about the potential for a large portion of positive results to be false, even if the false positive rate is low, when prevalence is also low. The paper was retracted soon after it was published. It was subsequently cited, incorrectly, by U.S. health officials. Since nothing in the paper appears to warrant its retraction, speculation that it was retracted for political rather than scientific reasons seems justified.
Here is the paper (in Chinese), my translation (made with the help of GoogleTranslate), and the English-language abstract as it appeared in PubMed in March.