Pollution and Covid. Two vague clues don't make an evidence

Tempo di lettura: 15 mins

In these days, newspapers and television programs (and the web, of course) are giving space to a statement by the Italian Society of Environmental Medicine (SIMA) announcing important discoveries on the link between airborne particulate matter and Coronavirus, even describing them as important for the decisions to be taken in the coming weeks.

We are sorry to have to point out that, this time too, the "discoveries" described in the press releases are vague clues, completely preliminary, not yet subject to peer-review by experts in the field [1]; and the conclusions are a mix of confusion and wishful thinking.

The first vague clue: an “evident correlation” that doesn not exist

The SIMA position paper, published on SIMA's website in mid March [2], proposed the existence of a direct relationship between "the number of cases of COVID-19 and the PM10 pollution levels of the territories", claiming therefore that a high concentration of PM10 in the atmosphere could be an "amplifier" of the diffusion of the Coronavirus (and not of its effects, for example higher mortality). The evidence would be a "strong correlation" that the authors found between the number of exceedances of the daily limit of PM10 concentrations and the number of infected cases [3]. The position paper even goes so far as to hypothesize that while in the southern regions of Italy (less polluted) the prevalent pattern of viral transmission occurs by contact between people, in the regions of northern Italy most polluted, the infection goes mainly through contact with "infected particulate matter". In a way, exhuming the ancient theory of miasmas, used to explain the plague in pre-scientific times.

Beyond the phenomenological robustness of the theory proposed to explain the mechanism of transport of the Coronavirus by particulate matter, the theory has no data support, contrary to what has been stated. And some choices made by the authors in choosing the data used for the "correlation" are at least very questionable, as illustrated in the Appendix.

The hypothesis that atmospheric particulate matter could act as a carrier substrate for the transport of the virus, thus increasing the rate of contagion, has been also contested by 70 experts of the Italian Aerosol Society (IAS) [4], who expressed many concerns and recalled that the apparent correlation between conditions of poor atmospheric circulation, secondary aerosol formation, PM accumulation near the ground and virus spread, should not be mistaken for a cause-effect relationship [5]. Of similar opinion, the Position paper of the Italian Environment and Health Network that collects institutions of the National Health Service and the National System of Environmental Protection [6]. 

In conclusion: the position paper did not contain any evidence, but advanced a theory supporting that with very questionable calculations.

The vague clue is popular in the media

A preliminary study not yet published in scientific journals, and not subject to any peer review, usually has a limited circulation limited to the authors and small circles. In such a delicate moment of the epidemic, this rule should be respected even more. Instead, the authors have proposed their study in dozens of interviews with newspapers, radio and television stations. In these interviews they often said that these were preliminary results, but in fact, as was easy to expect, the message passed very often that - even if there is not all the evidence - the particulate matter carries the virus. The authors justified themselves by saying that they "took care to explain to journalists that it was a preliminary study" [7], so that the blame would be on the journalists; however, if all journalists misrepresent, it means that the way the news was communicated was inadequate.

Contrary to what happens with IPCC reports, in which the hypotheses of relevance for decision-makers are carefully anchored to a judgement on their uncertainty, with specific guidelines to distinguish between "probable", "very probable" and "extremely probable", the authors wrote generically "It should be noted that the specificity of the rate of increase in cases of contagion that has affected some areas of Northern Italy in particular could be linked to the conditions of air particulate pollution that exerted a carrier and boost action". The use of the term "could" is too general. There are many, too many things that “could be”. The work of scientists, particularly during an epidemic, should be to distinguish accurately speculative hypotheses, not yet proven, from facts: otherwise, you end up contributing to the confusion of communication, which inevitably exists in complex matters.

The second vague clue and the virus that is and is not there

The second clue was presented in a Press Release of April 24th "Presence of Coronavirus on the atmospheric particulate matter: possible early "indicator" of future recurrences of the COVID-19 epidemic", which is based, again, on unpublished work [8].

Here the same authors of the previous position paper announce to have found traces of SARS-CoV-2 viral RNA in atmospheric PM10 taken in some areas of Bergamo. They go so far as to hypothesize the use of this discovery as important for the management measures of phase 2 of the COVID-19 epidemic.

But what have they really discovered? Even if the research has not yet passed the peer review and has not yet been published in a scientific journal, from the preliminary information it can be deduced that samples of ambient air have been taken: in these samples, which obviously contain particulate matter suspended in the atmosphere, the virus has not "been detected", but only traces of its RNA [9].

The presence of traces of viruses in particulate matter is by no means new, as traces of viruses from many millennia ago are also studied in ice cores [10]. But, of course, these are not active viruses, capable of being infectious. They are, in fact, fragments, traces that may indicate the presence of infectious viruses, but do not demonstrate it, as explained below.

Even if the presence of viruses still active on the particulate matter was confirmed in other studies, this would not lead to its infectivity. First of all, because after a few hours a virus, in the absence of a host to colonize and in whose cells it replicates, cannot continue to exist as a biological entity. Secondly, one or a few viruses are not enough to infect... but there must be a definite viral load under which there is no infection, as confirmed in a very convincing way on Nature by the Berlin Charité virologist Christian Drosten [11].

Let's try to clarify. These traces (or "the swab" we talk about daily) are detected thanks to a technology called polymerase chain reaction (PCR) (which won Kary Mullis the Nobel Prize for Chemistry in 1993), thanks to which a DNA fragment can be amplified to infinity. It is essentially the equivalent of an electron microscope that allows us to see what is invisible to the naked eye or with a powerful optical microscope. However, even for PCR, there is a threshold in the genome amplification cycles beyond which the signal is confused with the inevitable "background noise" so that the presence or absence of the searched genome becomes negative or indeterminate (if the value is very close to the threshold value). The values reported in the work are very close to this threshold and definitely below the one indicated by Drosten as the watershed of infectivity. In addition, no control sampling was performed in a region with low PM10 presence, e.g. in the mountains.

Returning to the presence of RNA traces on particulate matter, the authors go so far as to argue that these virus traces would still be useful for monitoring the evolution of the epidemic: "This first test opens up the possibility of testing the presence of the virus on atmospheric particulate matter in our cities in the coming months as an indicator for early detection of the reappearance of the coronavirus and taking appropriate preventive measures before the start of a new epidemic".

In fact, we read in the press release that the authors found the virus in 8 of the 22 days examined, while it is certain from the increase in positives that in the province of Bergamo in recent weeks the virus has been there continuously. Are we really sure that it is a good idea to adopt an "early detection method" that does not detect the virus in two-thirds of cases?

Another information disaster

If one reads the communiqué carefully, one understands that the authors are well aware that what they found does not provide any evidence to support the hypothesis of particulate matter as a booster of COVID-19 infections, but they write it ambiguously: "The evidence that SARS-CoV-2 RNA may be present on particulate matter in ambient air does not yet confirm with definitive certainty that there is a third infection route". The problem is not that there is no "definitive certainty": there is no minimum basis to demonstrate the presence of the virus on the dust particles suspended in the atmosphere and therefore, all the more so, to estimate if this represents a contagion route of some importance.

The consequence is that, again, the information conveyed by the media and social media is that the virus has been found in the particulate matter, and... this study confirms what the position paper suggested the previous month!

From the position paper, it seems that the authors are not yet able to evaluate the viability and especially the virulence of SARS-CoV-2, which, according to their position paper, would be particulate attached: "Further studies are underway to confirm these first tests on the possibility of considering PM as a 'carrier' of nuclei containing viral droplets, research that will have to go as far as evaluating the viability and especially the virulence of SARS-CoV-2 attached to particulate matter". In other interviews [12], the authors state that their research does not imply that the traces of SARS-CoV-2 found on the particulate matter may be virulent or cause infection, but it is not strange that this consideration was not given much weight. Also because the conclusion reached by the authors is "it is necessary to take into account the need to keep particulate emissions low in Phase 2 in order not to risk promoting the potential spread of the virus". And if there were no danger of infection, this conclusion would be meaningless.

Without having any new important scientific evidence in their hands, the authors go so far as to provide guidance to policy makers, for example by writing that the study "confirms the importance of widespread use of masks by the entire population".

What does “generalized use” mean? Would it always seem, that is, every time you leave home as imposed in Lombardy? Instead in the press release it is written "If we all wear masks, the inter-personal distance of 2 meters is to be considered reasonably protective allowing people to resume a social life". But then, generalized or 2 meters?[13]


There are already too many reasons to reduce emissions of particulate matter, or climate-altering gases, and regardless of the COVID-19 outbreak, we must reduce air pollution and drastically combat global warming. There is no need for theories, which have not yet been proven, to show other possible dangers related to particulate matter, and which are only likely to contribute to confusion in the communication of complex issues, at a time when it would be more useful for scientists to speak clearly, responsibly and effectively. A useful recommendation in this epidemic is rather to stay in the open air, because it does not hurt, and if you are indoors you need to circulate air by opening windows [14].


1. The "PM10 air pollution status" is described in the position paper with the average number of exceedances of the threshold of 50 ug/m3 PM10. This is an arbitrary and unfounded choice: this threshold is only a regulatory reference and the average PM10 concentration is more likely to represent the presence of PM10 in the atmosphere, and its hypothetical capacity to carry viruses.

2. The choice of PM10 alone is also arbitrary. If the virus is carried by fine particles, it should be more meaningful to refer to PM2.5.

3. The authors propose a direct relationship between the average of the exceedances of the 50 ug/m3 threshold of PM10 in the period February 10th-29th and the Covid-19 positives of March 3rd, using the aggregated data of 5 classes of provinces. But they do not say how they defined the 5 categories of provinces.

4. The use of average PM10 data from as many ECUs in one province, and even more so from more provinces, makes little sense. For example, if we consider the province of Lodi, the data of Codogno, Bertonico and San Rocco al Porto (within or near the red hotbed zone) are necessarily considered together with those of Montanaso and Lodi (further away). The number of "infected" in Lodi on March 3rd was 38, compared, for example, to 97 in Codogno; this means that, per resident inhabitant, Codogno had 7 times the "infected" of Lodi. If we look at the exceedances, we can see that Codogno has 11 exceedances of 50 ug/m3 in the period 10-29 February, while Lodi-Vignati has 9, just because in two days (11 and 14 February) PM10 concentrations in Lodi were 44 ug/m3, and in Codogno 51 and 57 ug/m3. The modest difference between these two PM10 values, recorded 18 and 21 days before March 3rd, can be in "direct relation" with a 7 times higher "contagion" value? Not at all.

5. The average of exceedances is calculated by combining the data of many provinces (which exactly is not explained in the position paper), therefore considering also very different types of control units, both "traffic" (which have higher values) and "urban background" (which generally have lower values). Putting them together does not make much sense. For example Lodi-S. Alberto (urban background station) has practically the same annual average concentrations of PM10 as Bertonico and S.Rocco al Porto.

6. Figure 1 of the position paper is constructed by correlating the average number of exceedances in the period February 10th-29th with the logarithm of the total "COVID-19 infected" of March 3rd, by province. But nobody knows the number of "infected", only the number of those who tested positive to the buffer or other analysis systems is known. The number of swabs performed was very different, depending on the province. Especially in the first days of the epidemic, the number of swabs in the Lodigiana red zone was very high: they were performed not only on the symptomatic people requiring treatment (as now), but also on those who had been in contact with the positives. Then it stopped. This explains, in part, the high number of positives in Lodigiano in the first days, but it is not certain that in other areas there were not as many (e.g. Bergamo). For this reason, the number of infected people per province should be considered with great caution, particularly during the first weeks of the epidemic.

7. It is not surprising, therefore, that if one tries to replicate the methodology used in the position paper, applying it to 7 provinces in Lombardy (Bg, Bs, Cr, Lo, Mi, Mn, Pv, Va), one finds an evident absence of correlation. The following figure shows the clear absence of correlation between the exceedances of the threshold of 50ug/m3 for PM10 10-29 February and the positives of March, 3rd. These are provinces in which the "contagions" data are a little more homogeneous than those evaluated by the authors, as it is the symptomatic positives that have had access to treatment (however, the limit remains as to how much they can be truly representative of the total number of positives, which also include asymptomatic positives).

8. Even if we consider the average of PM10 concentrations, instead of the number of exceedances, no correlation can be found.

9. The methodology adopted to search for correlation, therefore, appears very questionable, and it can be safely argued that the proposed 'direct relationship' of the position paper between PM10 presence and Coronavirus spread has no support in the data.

1 Please note that the article Setti et al (2020) Is there a Plausible Role for Particulate Matter in the spreading of COVID-19 in Northern Italy? published in the British Medical Journal Rapid Responses, is a peer-reviewed communication (sent, among other things, in response to an article dealing with an issue that is inconsistent with the subject of the communication)
2 Setti L, Passarini F, de Gennaro G et al. (2020) "Report on the effect of air particulate pollution and the spread of viruses in the population". Position paper Italian Society of Environmental Medicine, University of Bologna and University of Bari, minute 31.56
4 IAS contribution to the discussion on the relationship between air particulate pollution and Covid-19 diffusion (in Italian)
5 It is important to remember how it is not difficult, by properly processing the data, to find "spurious" correlations between different variables, which do not in any way mean a causal link, as has been shown very clearly and even ironically by some scholars. See for example
6 Italian Environment and Health Network, Air Pollution and Covid-19, Scienza in rete, April 13th, 2020 (in Italian)
7 Emanuele Bompan, The coronavirus claims another excellent victim: scientific communication. Here are the mistakes to avoid, Il Fatto Quotidiano, April 22nd, 2020
8 Setti L, Passarini F, de Gennaro G et al. (2020) SARS-Cov-2 RNA Found on Particulate Matter of Bergamo in Northern Italy: First Preliminary Evidencemed. Rxiv preprint. doi:
9. Let us suppose, instead, that large quantities of infectious particles were found (which they were not). We know that the virus, in the presence of infected and infectious individuals, is carried in droplets varying in size from a few tenths to a few microns in diameter. If we sample under these conditions, these droplets, which are suspended in the atmosphere, will certainly be sampled as well, because they are part of the atmospheric aerosol, as are the dust. Even under these extreme conditions, however, this does not show at all that there is a link between dust and droplets containing the virus; it only shows that these two categories can coexist in the atmosphere and are both collected when air sampling is carried out.
10 Zhong ZP, Solonenko NE, Li YF et al. (2020) Glacier ice archives fifteen-thousand-year-old viruses. bioRvx
11 Wölfel R, Corman VM, Guggemos W et al. Virological assessment of hospitalized patients with COVID-2019. Nature (2020). doi:
12 Interview with Radio Popolare on 24/4/2020, Gr 12.30.
13 If one reads the title of the sixth bibliographic reference one arrives at an article published by the authors themselves in Int. J. Environ. Res. Public Health: "Airborne Transmission Route of COVID-19: Why 2 Meters/6 Feet of Inter-Personal Distance Could Not Be Enough". It's all clear, isn't it?
14 Floriano Bonifazi and Francesco Forastiere, Opening windows in times of coronavirus, Scienza in rete, March 12th, 2020 (in Italian)

Aiuta Scienza in Rete a crescere. Il lavoro della redazione, soprattutto in questi momenti di emergenza, è enorme. Attualmente il giornale è interamente sostenuto dall'Editore Zadig, che non ricava alcun utile da questa attività, se non il piacere di fare giornalismo scientifico rigoroso, tempestivo e indipendente. Con il tuo contributo possiamo garantire un futuro a Scienza in Rete.

E' possibile inviare i contributi attraverso Paypal cliccando sul pulsante qui sopra. Questa forma di pagamento è garantita da Paypal.

Oppure attraverso bonifico bancario (IBAN: IT78X0311101614000000002939 intestato a Zadig srl - UBI SCPA - Agenzia di Milano, Piazzale Susa 2)

altri articoli

COVID-19 in Italy: lessons learnt and points for immediate discussion

These notes are strictly personal but express ideas which are shared by a vast community of researchers, including epidemiologists, clinicians, immunologists and lab scientists. This is meant to be a working document.