COVID: false negative PCR results over time

A recent study documented the pretty high incidence of false negative PCR test results for SARS-CoV-2 after the initial infection (see https://www.acpjournals.org/doi/10.7326/M20-1495)

 

Details:

-- 7 previously published studies provided data on PCR performance by time since symptom onset or SARS-CoV-2 exposure, from 1330 upper respiratory tract samples

    -- most studies did serial testing and required at least one positive PCR to consider a case confirmed

-- the FDA reports the specificity of PCR for SARS-CoV-2 to be 100%, though in this analysis the researchers assumed 90%, given that many studies were done outside of the US, and there might have been variability of test performance

 

Results:

-- false-negative rates by day of infection (see upper part of graph below):

    -- day 1 of infection: 100% (100%-100%)

    -- day 4 of infection: 67% (27%-94%)

    -- day 5 of infection (the usual day symptoms first appear): 38% (18%-65%)

    -- day 8 of infection (3 days after the beginning of symptoms): 20% (12%-30%)

    -- day 9 of infection: 21% (13%-31%)

    -- day 21 of infection: 66% (54%-77%)

-- per the lower graph below:

    -- a negative test result on day 3 would reduce the estimate of the relative probability that a case patient was infected by only 3% (going from 11.2% to 10.9%, from a large study of household contacts)

    -- a negative test on day 5 (1^st symptoms) reduced the probability that a case patient was infected by 60%

 

 

Commentary:

-- this study quantitated the incidence of false negative PCR results by the time after infection, showing a minimum rate of 20% on day 8 (3 days after symptom onset). The study clearly indicates that a negative test, though reassuring, is hardly definitive

-- as with blood tests in general, the higher the pretest probability of infection after a negative PCR, the higher the posttest probability of a real infection (ie: if the pretest probability is high, the patient likely had Covid-19)

-- this study suggests that people should not be tested until 3 to 5 days after exxposure, and that decisions regarding contact precautions or ending 14-day quarantine period prior to that may be inappropriate

   -- but, there still remains a 20% false-negative rate even in the best of circumstances, 3 days after symptom onset

-- the study raises many questions:

    -- what is the persistence of PCR positivity in patients who were and remain asymptomatic?

    -- how often should people be tested to determine infection?

        -- a French study presented a case report of a person with confirmed infection, both radiologically and with PCR of endotracheal aspirates, yet persistently negative PCR is by nasopharyngeal swabs

        -- or the Swiss cases presented in recent blog finding ?reinfection vs ?reactivation: http://gmodestmedblogs.blogspot.com/2020/05/covid-reinfection-cases-and-risk.html

   -- is the issue the reliability of the test itself or getting an adequate nasopharyngeal sample?

   -- is the transmissibility of infection different in patients with persistently negative nasopharyngeal swabs vs those with positive ones?

   -- would the individual patients with negative nasopharyngeal tests have been positive if they had been tested more often or at different intervals? what is the best sequence of testing in people to produce the most accurate results??

   -- in the longer-term, should we rely on PCR positivity after a few days of symptoms to determine Covid-19, and antibody testing later in the infection when we have reliable antibody tests?

 

Limitations:

-- there was significant heterogeneity in the designs of the studies used to develop this model; combining them in a larger analysis is therefore fraught (though they did sequentailly eliminate the individual studies and found no difference in outcome)

-- for most studies, they still used a positive PCR at some point to consider a case confirmed for Covid-19. A few studies did look at antibodies, though these are also suspect (http://gmodestmedblogs.blogspot.com/2020/05/covid-who-dont-trust-antibody-testing.html). 

-- the basis of the study was people who had a known, one-time exposure, not for people who have continuous load exposures such as health care workers or contacts of family members or others with infection

 

so, this study reinforces a few conclusions:

-- as with most tests, they are not perfect. In this case the PCR is pretty far from perfect

-- as with all imperfect tests, the actual posttest probability of a disease being present is dependent on the pretest probability (per Bayes theorem): ie a negative test in a low-probability patient is more likely to be accurate

-- we as clinicians, as well as the general public, need to understand that a negative test does not mean infection is not present, even when people are symptomatic (the test being most sensitive during the early symptomatic, especially in the 1^st 3-4 days or so after being symptomatic)

-- this all means that we should not be complacent with a negative test. Sort of similar to HIV, where we adopted the platform of universal precautions for everyone, effectively assuming that everyone is infected without a positive test result.

-- And, this also means that our tabulations of the numbers of infected people, at any stage of infection, are even more underestimated than we think, since they typically require PCR positivity....

geoff

 

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@uphams.org

 

to get access to all of the blogs:

1. go to http://gmodestmedblogs.blogspot.com/ to see them in reverse chronological order

2. click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

3. or you can just click on the magnifying glass on top right, then  type in a name in the search box and get all the blogs with that name in them

 

please feel free to circulate this to others. also, if you send me their emails, i can add them to the list