Intended for healthcare professionals

Rapid response to:

Practice Practice Pointer

Interpreting a covid-19 test result

BMJ 2020; 369 doi: (Published 12 May 2020) Cite this as: BMJ 2020;369:m1808

Read our latest coverage of the coronavirus pandemic

Rapid Response:

Re: Interpreting a covid-19 test result - On Specificity Again

Dear Editor

with respect to my previous Rapid Response [1], I first should like to correct a misprint in the figure for the victims of air pollution in Italy I had quoted from [2], which, for 2016, is 58,600 (not 58,200 as I wrote) from fine particulate matter (PM 2.5). Therefore the estimated overall mortality burden in Italy due to air pollution as caused by three main pollutants (NO 2, O3, PM 2.5) during 2016 according to the European Environment Agency was 76,200.

As regards the relationship between covid-19 and air pollution, in the last EEA report [3, p. 31] we are wisely reminded that "Spatial coincidence alone cannot be taken as causality.'' However chronic exposure to atmospheric pollution is not merely "spatial coincidence'', and it is both a priori and a posteriori clear that it aggravates "co-morbidities that could lead to fatal health outcomes of the virus infection'' [4].

As regards the specificity issue, I wish to add some update and remarks to strengthen the case that a thoughtful and cautious approach to SARS-CoV-2 testing is needed (this applies to a greater or lesser degree to all biomedical testing, of course, cf. "Test Bloater'' in [5]).

The WHO figure for the covid-19-attributed deaths in Italy up to 3 January is 74,985 (that is, less than the 2016 deaths due to air pollution for the mentioned three main pollutants), out of a total figure for confirmed cases of 2,141,201, giving a lethality of 3.5 % (the corresponding calculations for UK and Sweden give, respectively, 2.8 %, and 2.0 %) [6].

Based on these data, the prevalence of confirmed covid-19 in Italy should also be, up to the first decimal figure, 3.5 %. However, this estimate must be compared with the seroprevalence national inquiry, performed from 25 May to 15 July 2020, which estimated a national seroprevalence of 2.5%, although, unsurprisingly, with very big differences among regions (Lombardia – Northern Italy – was at 7.5%, while Campania – Southern Italy – was at 0.7%) [7]. So a 3.0 % prevalence may be closer to the truth.

It has been surmised by some readers that a 95% for specificity is too pessimistic. Now, whatever is guaranteed in laboratory studies, the most optimistic real-world value for specificity appears to have been 0.997 [8]. Under the assumption of a 3.0% prevalence in Italy, testing with sensitivity 0.7 and specificity 0.997 would give a false positive rate of 12.17 %. This means more than 260,000 Italian "cases" might have been false positives.

In my opinion this order of magnitude translates, in terms of restrictions on economic activity and sociality, and of psychological effects on healthy people mistakenly identified (and therefore both perceived and self-perceived ) as infected and infectious, into an iatrogenic national emergency.

Of course the possibility of arguing along these lines depends on the fact that, for the first time in the modern historical record, "cases'' have been allowed by WHO to be based on the outcome of a test alone [8]. As explained in my previous Rapid Response, at the beginning of the pandemic, testing was mostly limited to symptomatic individuals. But as soon as testing kits became more abundantly and widely available, screening and contact tracing got a bigger and bigger place among the causes for testing. To get an idea of the present SARS-CoV-2 testing policy, in Italy, for instance, in the two-week period 19 October-1 November 2020, out of 315,527 "cases'', only 35.1% have been tested because they had symptoms, and in the period 21 December 2020 - 3 January 2021, out of 188,732 "cases'', the corresponding percentage has been 34.2% [9, 10].

A welcome piece of news between these periods is that on last 14 December a "medical product alert'' has been published by WHO on "Nucleic acid testing (NAT) technologies that use real-time polymerase chain reaction (RT-PCR) for detection of SARS-CoV-2" [11], giving the following advice:

"4. Consider any positive result (SARS-CoV-2 detected) or negative results (SARS-CoV-2 not detected) in combination with specimen type, clinical observations, patient history, and epidemiological information.

5. Provide the Ct [cycle threshold] value in the report to the requesting healthcare provider."

I may be not alone in thinking that this alert could have been published a little earlier. It is also doubtful whether it has been readily and fully implemented.


[1] M. Mamone Capria: Re: Interpreting a covid-19 test result - Test Specificity is an Important Issue, 23 May 2020,

[2] "Air quality in Europe – 2019 report", European Environment Agency, October 16, 2019,

[3] "Air quality in Europe — 2020 report'', 23 November 2020,

[4] A. Pozzer et al., "Regional and global contributions of air pollution to risk of death from COVID-19'', Cardiovascular Research, 26 October 2020,

[5] M. Michael III, W.T. Boyce, A.J. Wilcox, Biomedical Bestiary: An Epidemiologic Guide to Flaws and Fallacies in the Medical Literature , Little Brown & Co (T), 1984.

[6] "Weekly epidemiological update - 5 January 2021’’,

[7] "ISTAT, Primi risultati dell'indagine di sieroprevalenza sul SARS-CoV-2’’, 3 August 2020,

[8] A.N. Cohen, B. Kessel, M.G. Milgroom, "Diagnosing COVID-19 infection: the danger of over-reliance on positive test results'', 28 September 2020,

[9] "Epidemia COVID-19 Aggiornamento nazionale 7 novembre 2020 – ore 11:00,'', 10 November 2020,

[10] "Epidemia COVID-19 Aggiornamento nazionale 5 gennaio 2021–ore 12,'' 8 January 2021,

[11] "WHO Information Notice for IVD Users’’, 14 December 2020,

Competing interests: No competing interests

11 January 2021
Marco Mamone Capria
University of Perugia, Italy
Department of Mathematics and Computer Science