key: cord-0787950-om1a63u1 authors: Tripepi, Giovanni; Plebani, Mario; Iervasi, Giorgio; Gori, Mercedes; Leonardis, Daniela; D’Arrigo, Graziella; Fusaro, Maria title: Distance from the outbreak of infection, ozone pollution and public health consequences of SARS-CoV-2 epidemic date: 2020-11-24 journal: Eur J Public Health DOI: 10.1093/eurpub/ckaa221 sha: d3856fbced730aa3cf567085633af74e4f6851fe doc_id: 787950 cord_uid: om1a63u1 BACKGROUND: Italy was the second country in the world, after China, to be hit by SARS-CoV-2 pandemic. The Italy’s experience teaches that steps to limit people’s movement by imposing “red zones” need to be put in place early by carefully identifying the cities to be included within these areas of quarantine. The assessment of the relationship between the distance from an established outbreak of SARS-CoV-2 infection with transmission-linked cases and mortality observed in other sites could provide useful information to identify the optimal radius of red zones. METHODS: We investigated the relationship between SARS-CoV-2 cases and the distance of each Italian province from the first outbreak of SARS-CoV-2 epidemic in Italy (the city of Lodi placed in the Lombardia region). In 38 provinces of Lombardia and neighboring regions, we performed a breakpoint analysis to identify the radius of the red zone around Lodi minimizing epidemic spread and mortality in neighboring cities. RESULTS: In all Italian provinces a non-linear relationship was found between SARS-CoV-2 cases and distance from Lodi. In an analysis including the provinces of Lombardia and neighboring regions, SARS-CoV-2 cases and mortality increased when the distance from Lodi reduced below 92 km and 140 km, respectively, and such relationships were amplified by ozone (O(3)) pollution. CONCLUSIONS: The breakpoint analysis identifies the radius around the outbreak of Lodi minimizing the public health consequences of SARS-CoV-2 in neighboring cities. Such an approach can be useful to identify the red zones in future epidemics due to highly infective pathogens similar to SARS-CoV-2. The outbreak of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was firstly identified worldwide in Wuhan, China, in December 2019 (1) . In Europe, the first two imported cases of SARS-CoV-2 were found in Rome in January 2020 when two Chinese tourists were diagnosed to be affected by the virus whereas the first autochthonous case in Italy was a 38 years old man tested positive after returning back from Wuhan and identified in the province of Lodi (Codogno) in February 2020 (2) . Thus, the province of Lodi (placed in the Lombardia Region) is considered the original outbreak of infection in Italy (2) . In February 2020, 11 northern Italian cities, having in common a high social interactivity due to their industrialization and business activities, were identified as clusters of infection and placed under quarantine. By the beginning of March 2020, the virus had spread to all Italian territory resulting consistently worse, in term of severity and incidence, in Northern compared to South Italy (2). On March 8 th , 2020, the Italian Government expanded the quarantine to the Lombardia region and other 14 northern provinces. On March 11 th , 2020, the Italian Government prohibited nearly commercial activities except for supermarkets and pharmacies and on March 22 nd , 2020 a new decree closed all nonessential commercial activities and industries restricting people movement by imposing a national lockdown. Social distancing, together with the use of wearing masks and washing hands, is considered one the most effective practices against SARS-CoV-2 pandemic. Indeed, the distance between related infection cases is a peculiar property of communicable disease dispersal although the distance between a single case and their infector is rarely assessable. Moving from individuals to populations, the analysis of the relationship between the distance from an established outbreak of infection in a specific region within a country with transmission-linked cases and excess mortality rate observed in other sites could provide useful information to assess the optimal radius of red zone around an outbreak of infection to be used in future epidemics of the same type (i.e. with similar R 0 ). In this paper we tested the following hypotheses: a) Is there a relationship between the distance from Lodi (the original outbreak of COVID-19 infection in Italy) of each Italian province and the total SARS-CoV-2 cases in the same cities (n=107) at March 21 st , 2020 (the day before the publication of the decree which extended further restrictions at national level)? b) In the area around the outbreak of infection (i.e. the area including the provinces of Lombardia and neighboring regions, n=38 cities), is it possible to identify the optimal radius of the quarantine zone (i.e., the optimal distance from Lodi) which minimizes the spread of epidemic and excess mortality in neighboring cities? c) Does the ozone (O 3 ) pollution on the relationship between the distance from Lodi, SARS-CoV-2 cases and excess mortality in 38 Italian provinces of the same regions was investigated by dividing the study sample into two groups according to the median value of days with O 3 >120 μg/m 3 (the recommended limit fixed by the World Health Organization) during 2017 and 2019 (6) . The breakpoints of the regression lines between the distance from Lodi, SARS-CoV-2 cases and excess mortality in 38 Italian provinces of Lombardia, Emilia Romagna, Piemonte, Trentino Aldo Adige and Veneto was carried out by the SegReg software (available from: https://www.waterlog.info/segreg.htm). We called this metHod to derive the radius of 'red zOnes' to PrEvent the spread of future epidemics, as the HOPE method. In the breakpoint analysis, data were expressed as point estimate (the breakpoint) and 95% confidence block. The distance of each Italian province from the outbreak of infection was calculated by using the Google Mapping Technology, a system of recognized scientific validity (7) . The association between continuous variables was investigated by Pearson product moment correlation coefficient (r) and P value. Other calculations were performed by SPSS for Windows Version 22, Chicago, Illinois, USA. On March 21 st , 2020 the total SARS-CoV-2 cases (including active cases, recovered/discharge patients and deceased) in Italy were 53578 over a total population of about 60 millions of inhabitants. In all Italian provinces (n=107), there was a non-linear relationship between SARS-CoV-2 cases and the distance from Lodi and the deviation from uniformity mostly concerned the provinces of Lombardia and those of neighboring regions (Piemonte, Emilia Romagna, Trentino Alto Adige and Veneto) (Figure 1 -upper panel) . In all Italian provinces (n=107), no relationship was found between SARS-CoV-2 cases and population density (see Extrafigure). In an analysis including the provinces of Lombardia and those of the four neighboring Regions (n=38 provinces), the burden of SARS-CoV-2 cases increased dramatically when the distance from Lodi reduced below 92 km (95% confidence block: 81-119 km) (Figure 1 -bottom panel) . When the same analysis was carried out according to the percent increase in all-cause mortality (Figure 1-bottom panel) the link between excess risk of mortality and distance from Lodi became steeper when the distance from the outbreak of Lodi reduced below 140 km (95% confidence block: 131-182 km). Thus, 92 km and 140 km emerged as the radiuses of red zone around Lodi (95% confidence block for both outcomes: 81-182 km) which minimize SARS-CoV-2 cases and excess mortality, respectively, in neighboring provinces during the first phase of epidemic in Italy (Figure 2) . We also found that the prevalent cases of SARS-CoV-2 infection (collected up to 21 st March, 2020) were directly and strongly related to the concomitant excess death rate observed in the same 38 provinces (r=0.90, r 2 =0.81, P<0.001), implying that 81% of the observed excess mortality is explained by SARS-CoV-2 cases (Figure 3) . On univariate analysis, the mean values of days with O 3 > 120 μg/m 3 (Figure 4) . Remarkably, the same effect modification was also found (P=0.005) for the excess mortality rate (Figure 4) . In fact, a 50-km reduction in the distance from Figure 1 (bottom panels) . This analysis showed that the number of SARS CoV-2 cases in provinces of Veneto did not materially differ from those found in other provinces of Piemonte, Emilia Romagna and Trentino Alto Adige (included Sondrio, which is located in Lombardia) when matched for the distance from Lodi. Similarly, in provinces of Veneto located >140 km far from Lodi, the excess mortality overlapped to that of provinces of Emilia Romagna and of two provinces of Piemonte, i.e. two regions in which the use of swabs was 50% and 77% lower, respectively, that those made in Veneto. further evidence that long term exposure to atmospheric contamination represents a favorable factor for the spread of SARS-CoV-2 epidemic from an outbreak to neighboring cities as well as for the excess mortality rate observed during the first phase of epidemic, this latter being a finding which was not previously described. The role of air pollution as an effect modifier is further supported by the notion that the unique N-terminal fragment within the spike protein which characterizes viral genome allows the attachment of the virus on air pollutants. This interpretation is germane to that of Coccia M (2) who suggests that the accelerated transmission of COVID-19 is mainly due to the mechanism of "air pollutionto-human transmission". Furthermore, there is also evidence that ozone per se induces lung inflammation through stimulation of the oxidative stress process thus exacerbating the health consequences of SARS-CoV-2 infection. Our hypothesis is that the coexistence between this mechanism(s) with the proximity Furthermore, our study is also the first one investigating, by using regions as an instrumental variable, whether a public health intervention contemplating also a wider use of swabs could reduce the spread of SARS-CoV-2 infection and excess mortality in a wide geographical area including five regions in North Italy totalizing about 25 millions of inhabitants. In our study, by using regions as an instrumental variable (10), we found that, at the beginning of the epidemic, the lower number of burden of SARS CoV-2 cases and the relatively lower excess mortality in the Veneto region in respect to the frequency of the same outcomes observed in the other provinces of Lombardia, Emilia Romagna, Trentino Alto Adige and Piemonte seems to be due to the higher distance from the original outbreak of infection (the city of Lodi) and lower O 3 pollution of the Veneto region rather than be due to the more frequent use of swabs in the same region. However, the hypothesis that the wider use of swabs could prevent the dispersal and the public health consequences of SARS CoV-2 cases needs to be specifically confirmed in larger studies worldwide. Our study presents some limitations. First, we did not test the external validity of the HOPE method in other countries. Thus, the generalizability of this method remains to be formally tested in future studies. Second, the data on O 3 pollution represent an aggregate of several days and therefore a granular analysis was not performed. Third, the public health utility of a wider use of swabs needs to be further investigated in specifically designed studies to definitely assess the utility of swabs for the containment of the epidemic. Fourth, given the ecological nature of our study, we cannot exclude the potential effect on the study results and interpretation of ecological fallacy, a type of bias that arises when an inference is made about an individual based on aggregate data. In conclusion, to the best of our knowledge, this is the first study describing a pragmatic method to define the radiuses of the red zone during the SARS-CoV-2 epidemic around an established outbreak of infection minimizing the dispersal of infection and excess mortality in neighboring cities. We also found that the distance from the outbreak of infection (Lodi, placed in the Lombardia Region) and O 3 pollution play a primary role to interpret the dispersal and the public health consequences (in terms of excess mortality) of SARS-CoV-2 in a wide geographical area in North Italy totalizing about 25 millions of inhabitants. Furthermore, given the fact that the district of Lodi is similar to other sites worldwide, the HOPE method we propose can be applied to other countries having similar characteristics in terms of industrialization, climate, and social interaction as Lodi and be useful to define the red zones in future epidemics due to highly infective pathogens with an R0 similar to that of SARS-CoV-2 (2