key: cord-0859691-z3ns5y16
authors: Accarino, Gabriele; Lorenzetti, Stefano; Aloisio, Giovanni
title: Assessing correlations between short-term exposure to atmospheric pollutants and COVID-19 spread in all Italian territorial areas()
date: 2020-10-15
journal: Environ Pollut
DOI: 10.1016/j.envpol.2020.115714
sha: dbc93a835acd432d65ddea5249234375804b7c15
doc_id: 859691
cord_uid: z3ns5y16

The spread of SARS-CoV-2, the beta coronavirus responsible for the current pneumonia pandemic outbreak, has been speculated to be linked to short-term and long-term atmospheric pollutants exposure. The present work has been aimed at analyzing the atmospheric pollutants concentrations (PM(10), PM(2.5), NO(2)) and spatio-temporal distribution of cases and deaths (specifically incidence, mortality and lethality rates) across the whole Italian national territory, down to the level of each individual territorial area, with the goal of checking any potential short-term correlation between these two phenomena. The data analysis has been limited to the first quarter of 2020 to reduce the lockdown-dependent biased effects on the atmospheric pollutant levels as much as possible. The analysis looked at non-linear, monotonic correlations using the Spearman non-parametric correlation index. The statistical significance of the Spearman correlations has also been evaluated. The results of the statistical analysis suggest the hypothesis of a moderate-to-strong correlation between the number of days exceeding the annual regulatory limits of PM(10), PM(2.5) and NO(2) atmospheric pollutants and COVID-19 incidence, mortality and lethality rates for all the 107 territorial areas in Italy. A weak-to-moderate correlation seems to exist when considering the 36 territorial areas in four of the most affected regions (Lombardy, Piedmont, Emilia-Romagna and Veneto). Overall, PM(10) and PM(2.5) showed a higher non-linear correlation than NO(2) with incidence, mortality and lethality rates. As to particulate matters, PM(10) profile has been compared with the incidence rate variation that occurred in three of the most affected territorial areas in Northern Italy (i.e., Milan, Brescia, and Bergamo). All areas showed a similar PM(10) time trend but a different incidence rate variation, that was less severe in Milan compared with Brescia and Bergamo.

The first two authors equally contributed to this work 23 24 J o u r n a l P r e -p r o o f ABSTRACT 25 The spread of SARS-CoV-2, the beta coronavirus responsible for the current pneumonia 26 pandemic outbreak, has been speculated to be linked to short-term and long-term 27 atmospheric pollutants exposure. 28 The present work has been aimed at analyzing the atmospheric pollutants concentrations 29 (PM 10 , PM 2.5 , NO 2 ) and spatio-temporal distribution of cases and deaths (specifically 30 incidence, mortality and lethality rates) across the whole Italian national territory, down to the 31 level of each individual territorial area, with the goal of checking any potential short-term 32 correlation between these two phenomena. 33 The data analysis has been limited to the first quarter of 2020 to reduce the lockdown-34 dependent biased effects on the atmospheric pollutant levels as much as possible. The 35 analysis looked at non-linear, monotonic correlations using the Spearman non-parametric 36 correlation index. The statistical significance of the Spearman correlations has also been 37 evaluated. 38 The results of the statistical analysis suggest the hypothesis of a moderate-to-strong 39 correlation between the number of days exceeding the annual regulatory limits of PM 10 , PM 2.5 40 and NO 2 atmospheric pollutants and COVID-19 incidence, mortality and lethality rates for all 41 the 107 territorial areas in Italy. A weak-to-moderate correlation seems to exist when 42 considering the 36 territorial areas in four of the most affected regions (Lombardy, Piedmont, 43 Emilia-Romagna and Veneto). Overall, PM 10 and PM 2.5 showed a higher non-linear 44 correlation than NO 2 with incidence, mortality and lethality rates. 45 As to particulate matters, PM 10 profile has been compared with the incidence rate variation 46 that occurred in three of the most affected territorial areas in Northern Italy (i.e., Milan, 47 1. Introduction 54 The human fatal pneumonia that has currently spread worldwide, known as CoronaVIrus 55 Disease-2019 (COVID-19) , is caused by a very recently identified beta coronavirus named 56 Severe Acute Respiratory Syndrome CoronaVirus type 2 (SARS-CoV-2) (Sohrabi et al., 57 2020) . In less than 6 months, SARS-CoV-2 caused 6,194,508 confirmed cases of COVID-19 58 and 372,501 deaths (on May 31 st , 2020) (Dong et al., 2020) . Hence, the current SARS-CoV-2 59 pandemic is evidently much more widespread than the two previously reported coronavirus (Ahn et al., 2020) . At the beginning of 2020, exactly on January 30 th , the 63 World Health Organization (WHO) declared that the Chinese outbreak of COVID-19 had to be 64 considered a Public Health Emergency of International Concern posing a high risk to 65 countries with vulnerable health systems (Sohrabi et al., 2020) . On March 11 th , 2020, the 66 WHO (WHO Director, 2020) established that the international outbreak of a new SARS-CoV-2 67 coronavirus infection could be considered a pandemic, despite there being 77 countries and 68 territories with no reported cases at that time, and 55 countries and territories with no more 69 than 10 reported cases. 70 The first epidemic spread of SARS-CoV-2 occurred in Wuhan, a main city in the Chinese cases reported outside China as 125,000 COVID-19 cases from 118 countries and   75 territories were reported overall to the WHO itself. The first European confirmed cases of 76 COVID-19 were reported in France (on January 24 th , 2020) (Stoecklin et al., 2020) and as of 77 February 21 st , 2020, only 47 confirmed cases had been notified in 9 different countries of the 78 WHO European Region (Spiteri et al., 2020) . Among them, three cases were diagnosed in 79 Italy, including a Chinese couple of tourists at the end of January in Rome, which however did 80 not lead to a further spread of the virus. On February 20 th −21 st , 2020, the first detected cases 81 on Italian people were reported in two Italian towns: Vo' Euganeo (territorial area of Padua, 82 Veneto region) and Codogno (territorial area of Lodi, Lombardy region), the latter being 83 considered as "patient 1" (Grasselli et al., 2020; Onder et al., 2020; Romagnani et al., 2020) . 84 As a result of the pandemic spread all over the world, several studies have been focusing on 85 the potential role of atmospheric pollutants (mainly Particulate Matters -PMs) in the diffusion 86 of COVID-19 both in the short-and the long-term, as well as the impact of the virus on human 87 health (Bontempi, 2020a; Coccia, 2020; Fattorini and Regoli, 2020; Frontera et al., 2020a; 88 Frontera et al., 2020b; Liang et al., 2020; Martelletti and Martelletti, 2020; Onder et al., 2020; 89 Riccò et al., 2020; Wu et al., 2020; Zoran et al., 2020a; Zoran et al., 2020b) . 90 It is actually biologically plausible for people living in highly industrialized areas, and therefore 91 subject to higher pollution levels, to show more severe symptoms (Bontempi et al., 2020) . 92 Further studies have pointed out that atmospheric pollutants can indeed act as virus carriers 93 and boost pandemic diffusion (Del Buono et al., 2020; Sterpetti, 2020; Setti et al., 2020; 94 SIMA, 2020 ). Yet, as stated in a recent work (Bontempi, 2020a) , there is no scientific 95 evidence that PMs can act as a means of transport of the virus, thus promoting its diffusion. 96 J o u r n a l P r e -p r o o f Another study (Bontempi, 2020b) suggests that parameters other than environmental 97 pollution, such as trade-related aspects, should be assessed as a possible source of virus 98 diffusion at the onset of the pandemic in Italy. 99 A pandemic is actually a very complex phenomenon and several variables (confounding 100 factors) should be considered in the analysis to describe any possible correlation in a rigorous 101 way (Bontempi et al., 2020) . 102 As reported by Heederik et al. (Heederik et al., 2020) , scientific research has been adopting 103 different methodological approaches to this problem. The first approach relies on the use of weather, socioeconomic and behavioral variables (e.g. income, obesity, smoking habits), 111 days since the first reported case of COVID-19, population age distribution, and days since 112 the issuance of the stay-at-home order in each state. There are only few other studies based 113 on the use of confounders and the related sensitivity analysis (Isaifan, 2020; Liang et al., 114 2020; Zhu et al., 2020) . 115 Further works on this subject have merely pointed out the need to consider other variables 116 besides atmospheric pollution, such as temperature and relative humidity (Sajadi et al., 2020; 117 Scafetta, 2020) . Many studies conducted in Europe about the association between COVID-19 118 diffusion and long-term exposure to pollutants, did not consider the effects of confounding 119 factors in their analyses (Bontempi 2020a; Coccia 2020; Fattorini and Regoli, 2020; Frontera 120 et al., 2020a; Frontera et al., 2020b; Ogen, 2020; Travaglio et al., 2020; Zoran et al., 2020a; 121 J o u r n a l P r e -p r o o f Zoran et al., 2020b) . Moreover, several studies have also pointed out the connection among 122 climate change, atmospheric pollution and human health (Orru et al., 2017; Kinney, 2018;  123 Ravindra et al., 2019) . 124 Finally, further works have been trying to predict the behavior of some COVID-19 related 125 variables through the use of exponential models as well as dynamic SIR-based models (Chen 126 et al., 2020; Fanelli and Piazza, 2020; Remuzzi and Remuzzi, 2020; Reno et al., 2020) . Table 1 , data were only collected from reliable institutional and well-referenced 148 sources, in order to reduce as much as possible the uncertainty due to mixing different data 149 sources. The reference links to access raw data are also indicated. 150 Data collection for the short-term exposure to atmospheric pollutants and COVID-19 related 151 variables has been limited up to March 31 th , 2020 with the aim of minimizing the lockdown-152 dependent biased effects. Furthermore, COVID-19 incidence rate has only been examined up 153 to the end of the national lockdown on June 2 nd , 2020. about the epidemic spread in Italy on a regional scale and in the different territorial areas as 158 well. Several COVID-19 related variables are available at regional level, whereas only the 159 total cases are reported for the territorial areas. Therefore, COVID-19 deaths by territorial The CAMS dataset has a higher spatial resolution of 0. The variation of the incidence rate (ir_variation) for day t was calculated as: The number of days exceeding the annual regulatory limits (see Table 2 ) of particulate 263 matters (PM 2.5 and PM 10 ) and NO 2 (see Figure 1 , panels A, B and C) has been calculated to 264 determine the level of pollution across the Italian territorial areas in the period from January J o u r n a l P r e -p r o o f

in Italy: the situation in the first quarter of 2020 296 The potential short-term correlation between atmospheric pollutants and COVID-19-related 297 variables has been assessed based on the number of days exceeding the annual regulatory 298 limits (see Table 2 Table 3 , rows 2A-2I). In particular, when considering incidence and mortality 319 rates, PM 2.5 and PM 10 concentrations exhibit a stronger correlation compared with NO 2 , 320 whereas when the lethality rate is considered, the correlation strength is almost the same 321 among pollutants. 322 Moreover, the statistical analysis of the short-term exposure to PM 2.5 , PM 10 , and NO 2 was 323 further targeted on the 36 territorial areas mostly affected by COVID-19 (Figure 3) . In this 324 case, the magnitude of the Spearman coefficients ranges from 0.32 to 0.49, thus highlighting 325 weak-to-moderate positive correlations ( 22 nd in all the considered territorial areas, which however corresponds to a decline in the 369 incidence rates on the same day or in the following days. 370 Moreover, by extending the incidence rate monitoring period (Figure 6 ) up to the last 371 lockdown day (June 2 nd , 2020), it can be observed that the outbreak dynamics are different in 372 the three territorial areas considered in Figure 5 . The cumulative incidence rates reported for 373 Milan, Brescia and Bergamo (Figure 6) showed that in Milan they were much lower than in 374 Brescia and Bergamo. In particular, as regards time trends, the cumulative incidence rates in 375 Brescia (Figure 6 , blue line) and Bergamo (Figure 6 , orange line) first showed an exponential 376 increase until the end of March, which was then followed by a more moderate, steady, linear 377 increase. In Milan (Figure 6 , green line), the initial increase of the incidence rate was at much (Fattorini and 427 Regoli, 2020), whereas only 66 administrative Italian regions were considered in (Ogen, 428 2020); further studies (Zoran et al., 2020a; Zoran et al., 2020b) merely focused on the 429 territorial area of Milan or performed the analysis at regional level only (Frontera et al., 2020) ; 430 finally, another study (Bontempi 2020a ) only considered 7 provinces in Lombardy and 6 431 provinces in Piedmont. All the aforementioned studies did not consider PM 2.5 , PM 10 and NO 2 432 at the same time, except for one (Fattorini and Regoli, 2020 J o u r n a l P r e -p r o o f (Bontempi, 2020b) , weather (Orru et al., 2017; Kinney, 2018; Ravindra et al., 2019) , socio-441 economic conditions (Cazzolla-Gatti et al., 2020; Wu et al., 2020) , and healthy-related 442 variables (Cazzolla-Gatti et al., 2020) . 443 The analysis of confounding factors is actually very challenging (Bontempi, 2020b; Cori and 444 Bianchi, 2020; Heederik et al., 2020) analyses have been separately performed, at both regional and territorial area level.

A new methodological approach is then required to determine the context and sort out the key 455 parameters to consider, as recently stated in (Bontempi, 2020b) . This approach should 456 promote international research and involve political authorities that have to rely on 457 multidisciplinary scientific committees. In this way, research efforts would be more effective, 458 and the correct dissemination of the relevant and suitable information would not only concern 459 the scientific community but also a wider audience, including the general population and 460 policymakers. 461 Therefore, due to the sensitivity of these issues, confounders were left out of the scope of the 462 analysis in this initial phase, adopting a methodological approach based on the "go slow to go 463 fast" paradigm, as recently suggested in (Heederik et al., 2020) . PMs with the incidence, mortality and lethality rates is more pronounced compared with NO 2

In the case of the 4-region subset (36 territorial areas). 490 This preliminary study, inspired by the "go slow to go fast" paradigm (Heederick et al. 2020), 491 helped assume the presence of potential correlations between short-term exposure (3 492 months) to particulate matters and several COVID-19-related variables in Italy.

493 494

Current status of epidemiology, diagnosis, therapeutics, and vaccines for novel coronavirus disease 2019 (covid-19)

First data analysis about possible COVID-19 virus airborne diffusion due to air particulate matter (PM): The case of Lombardy (Italy)

Commercial exchanges instead of air pollution as possible origin of COVID-19 initial diffusion phase in Italy: More efforts are necessary to address interdisciplinary research

Understanding COVID-19 diffusion requires an interdisciplinary, multi-dimensional approach

Machine learning reveals that prolonged exposure to air pollution is associated with SARS-CoV-2 mortality and infectivity in Italy

A mathematical model for simulating the phase-based transmissibility of a novel coronavirus

Factors determining the diffusion of COVID-19 and suggested strategy to

The role of air pollution (PM and NO2) in COVID-19 spread and lethality: a systematic review

Covid-19 and air pollution: communicating the results of geographic correlation studies

The Italian outbreak of COVID-19: conditions, contributors, and concerns

An interactive web-based dashboard to track COVID-19 in real time

2020. Monitoring Covid-19 Impacts on Air Pollution. Dashboard Prod-ID: DAS-217-en

Analysis and forecast of COVID-19 spreading in China, Italy and France

Role of the chronic air pollution levels in the Covid-19 outbreak risk in Italy

Severe air pollution links to higher mortality in COVID-19 patients: The "double-hit" hypothesis

Regional air pollution persistence links to COVID-19 infection zoning

Critical care utilization for the covid-19 outbreak in Lombardy, Italy: early experience and forecast during an emergency response

Go slow to go fast: A plea for sustained scientific rigor in air pollution research during the COVID-19 pandemic

ICPD (Civil Protection Department GitHub data repository)

ICPD (Civil Protection Department of the Italian Government)

The dramatic impact of Coronavirus outbreak on air quality: has it saved as much as it has killed so far? Glob

Impact of the COVID-19 epidemic on the total mortality of the resident population in the first quarter

Popolazione residente comunale per sesso anno di nascita e stato civile

Statistical codes of territorial administrative units: municipalities, metropolitan cities, province and regions

Interactions of climate change, air pollution, and human health

Urban Air Pollution May Enhance COVID-19 Case-Fatality and Mortality Rates in the United States

Air pollution and the novel covid-19 disease: a putative disease risk factor. Sn compr

Assessing nitrogen dioxide (NO2) levels as a contributing factor to coronavirus (COVID-19) fatality

Case-fatality rate and characteristics of patients dying in relation to covid-19 in Italy

The interplay of climate change and air pollution on health

Generalized additive models: Building evidence of air pollution, climate change and human health

COVID-19 and Italy: what next?

Forecasting COVID-19-associated hospitalizations under different levels of social distancing in Lombardy and Emilia-Romagna, Northern Italy: Results from an extended SEIR compartmental model

SARS-CoV-2 infection and air pollutants: Correlation or causation?

The covid-19 infection: lessons from the italian experience

Temperature, humidity, and latitude analysis to estimate potential spread and seasonality of Coronavirus Disease 2019 (COVID-19)

Distribution of the SARS-CoV-2 pandemic and its monthly forecast based on seasonal climate patterns

Airborne transmission route of COVID-19: Why 2 meters/6 feet of inter-personal distance could not be enough

Position Paper Particulate Matter

World health organization declares global emergency: A review of the 2019 novel coronavirus (covid-19)

First cases of coronavirus disease 2019 (COVID-19) in the WHO European Region

ES.2020.25.9

Lessons learned during the COVID-19 virus pandemic

First cases of coronavirus disease 2019 (covid-19) in France: surveillance, investigations and control measures

Links between air pollution and COVID-19 in England

General's opening remarks at the media briefing on COVID-19 -11

Air quality guidelines for particulate matter, ozone, nitrogen dioxide and sulfur dioxide. Summary of risk assessment

Exposure to air pollution and COVID-19 mortality in the United States: A nationwide cross-sectional study

Association between short-term exposure to air pollution and COVID-19 infection: Evidence from China

Assessing the relationship between surface levels of PM2.5 and PM10 particulate matter impact on COVID-19 in Milan

Assessing the relationship between ground levels of ozone (O3) and nitrogen dioxide (NO2) with coronavirus (COVID-19) in Milan

Acknowledgments 495 The authors would like to thank Emanuele Taurino, Alessio Errico, Simone Miglietta, and