The Historical Trend of Air Pollution and Its Impact on Human Health in Campania Region (Italy)

: The Campania region covers an area of about 13,590 km 2 with 5.8 million residents. The area suffers from several environmental issues due to urbanization, the presence of industries, wastewater treatment, and solid waste management concerns. Air pollution is one of the most relevant environmental troubles in the Campania region, frequently exceeding the limit values established by European directives. In this paper, airborne pollutant concentration data measured by the regional air quality network from 2003 to 2019 are collected to individuate the historical trends of nitrogen dioxide (NO 2 ), coarse and ﬁne particulate matter with aerodynamic diameters smaller than 10 µ m (PM10) and 2.5 µ m (PM2.5), and ozone (O 3 ) through the analysis of the number of exceedances of limit values per year and the annual average concentration. Information on spatial variability and the effect of the receptor category is obtained by lumping together data belonging to the same province or category. To obtain information on the general air quality rather than on single pollutants, the European Air Quality Index (EU-AQI) is also evaluated. A special focus is dedicated to the effect of deep street canyons on air quality, since they are very common in the urban areas in Campania. Finally, the impact of air pollution from 2003 to 2019 on human health is also analyzed using the software AIRQ+.


Introduction
Air pollution is the major environmental risk in Europe [1-3] and one of the most critical in the world. In fact, the World Health Organization (WHO) estimates that air pollution, both environmental and domestic, causes 7 million deaths per year, 5.6 million deaths from non-communicable diseases, 1.5 million from pneumonia, and 3% from cardiorespiratory mortality. It is also estimated that 90% of premature deaths per year due to air pollution are concentrated in middle-and low-income countries.
The effects of air pollution on human health can be chronic (long-term) or acute (short-term). Chronic effects occur after prolonged exposure and include cardiovascular and pulmonary diseases, respiratory diseases, chronic changes in physiological functions, lung cancer, chronic cardiovascular diseases, and intrauterine growth restrictions. Acute effects due to short-term exposure (hours or days) to high concentrations of pollutants include respiratory and cardiovascular problems and physiological changes. Furthermore, climate change resulting from environmental pollution can also influence the geographic distribution of many infectious diseases [4][5][6][7]. Regarding age, epidemiological investigations of the effects of air pollution on health show that children and the elderly are the two groups most likely to be affected due to their relatively weak immune systems [8]. Maternal exposure to ambient air pollution is associated with adverse impacts on fertility, pregnancy, newborns, and children [9,10].
There is also emerging evidence that exposure to air pollution is associated with diabetes in adults and may be linked to obesity, systemic inflammation, Alzheimer's disease, and dementia [11,12].

1.
Analysis of the historical trend of atmospheric pollution in Campania (2003-2019).

2.
Analysis of the impact of pollution on mortality and morbidity.
In the first part, data from fixed air quality stations are analyzed to highlight the temporal trends of single pollutants (NO 2 , PM10, PM2.5, and O 3 ) and of the European air quality index (AQI).
In the second part, the impact of air pollution on human health in Campania is assessed using the AIRQ+ software, which estimates related mortality or morbidity in percentages.
The results of this work can be used as a support to better define intervention strategies and policies aimed at improving air quality and reducing the impact on public health in Campania.

The Area Studied
The Campania region covers an area of about 13,590 km 2 . From an administrative point of view, it is divided into five provinces. It is located in the south of Italy at the center of the Mediterranean Sea and is surrounded by the Apennine to the east, which occupies a big portion of the territory, and the Tyrrhenian Sea to the west. Its central position in the Mediterranean area is rooted in history.
According to Koppen [31], the Campania region has a Mediterranean climate with a cold winter and dry summer, with wetter summers near the coastline and more severe winters in the mountains [32]. The differences in climate between the high mountains in the interior and the coastal areas are quite large. Campania is the cradle of the "Mediterranean diet" similar to the eating habits of Spain and Greece, including high consumption of olive oil, legumes, cereals, fruits, and vegetables. Additionally, the urbanization of most of the cities in Campania is similar to other cities on the coasts of the Mediterranean Sea, characterized by a high density of buildings and very narrow streets. Now, about 5.8 million people live in Campania. Of them, 4 million live in the metropolitan area of Naples, and about 1 million are residents in the city of Naples, the capital of Campania.
Campania has the highest age-standardized death rate in Italy, equal to about 45 deaths per 10,000 inhabitants (a national average of about 33 deaths per 10,000 inhabitants) due to cardiovascular disease. Additionally, with regard to respiratory diseases, pneumonia, bronchitis, asthma, and chronic obstructive pulmonary disease (COPD), Campania is the leading region in Italy for deaths from chronic diseases of the lower respiratory tract with a mortality rate of 4.2 per 10,000 inhabitants. As for tumors, in Campania, the average survival rate for all tumors diagnosed in 2005-2009 is one of the worst in Italy, being less than 55%.

Air Pollution Data (Monitoring Network)
Data on air quality used in this paper are measured at fixed monitoring stations managed by Campania Regional Environmental Protection Agency (ARPAC). The air quality network was realized in the 1990s. Our analysis ranges from 2003 to 2019. In this period, the structure of the regional air quality network changed. In 2003, the number of fixed stations was 19 with 77 analyzers. Pollutants monitored were carbon monoxide (CO), NO 2 , NOx, O 3 , PM10, total suspended particles (TSP), and SO 2 . The configuration of the fixed-station network was improved in 2019, increasing the number of monitoring stations located with great density in sensitive areas in order to obtain optimal spatial coverage of the area. At present, the number of stations is 42, and the number of analyzers is 196. Table 1 reports a list of the fixed monitoring stations with the indication of the province and category. On the basis of previous studies [33,34], three different station categories are considered: traffic, industrial, and background. Traffic stations are located in urban or suburban areas near high-traffic roads; background stations are in urban or suburban areas far from emission sources; industrial stations are in industrial areas. Only two stations are sited in rural areas (San Gregorio Lago Matese and Ottati Alburni) and are considered background stations. The pollutants monitored are CO, NO 2 , O 3 , particulate matter (PM10 and PM2.5), SO 2 , and benzene. The pollutants measured in each station are not the same and depend mainly on the station category.
The criteria used for the localization of the stations, the measurement methods, and the quality standards align with European directives [35].
The spatial distribution of the monitoring stations in the region is reported in Figure 1.
The criteria used for the localization of the stations, the measurement methods, and the quality standards align with European directives [35].
The spatial distribution of the monitoring stations in the region is reported in Figure  1. Before analyzing air quality data, the monitoring efficiency per solar-year of each analyzer was verified. The solar-year efficiency was measured based on 1-h or 1-day timeaverage data as: for NO2 and O3, and for PM10 and PM2.5. The subscripts i, r, and y refer to the pollutant, the receptor, and the year, respectively. In some years, the monitoring efficiency was very low with large data gaps. Figure 2 shows the % of valid data depending on the fixed efficiency criteria. Imposing a highefficiency minimum criterion (>75%) would determine the loss of 35-45% of data, depending on the specific pollutant. For this reason, the efficiency criteria are fixed at 50%. Yearly datasets not respecting the fixed minimum efficiency criteria were not considered in the successive analysis. Before analyzing air quality data, the monitoring efficiency per solar-year of each analyzer was verified. The solar-year efficiency was measured based on 1-h or 1-day time-average data as: for PM10 and PM2.5. The subscripts i, r, and y refer to the pollutant, the receptor, and the year, respectively. In some years, the monitoring efficiency was very low with large data gaps. Figure 2 shows the % of valid data depending on the fixed efficiency criteria. Imposing a highefficiency minimum criterion (>75%) would determine the loss of 35-45% of data, depending on the specific pollutant. For this reason, the efficiency criteria are fixed at 50%. Yearly datasets not respecting the fixed minimum efficiency criteria were not considered in the successive analysis.

Description of the Air Quality Index Adopted
To have information on the historical trend of air quality instead of that of single pollutants, the EU-AQI was adopted with some minor modifications in the calculation procedure.
The EU-AQI is a daily air quality index that assumes numerical values from 1 to 5, corresponding with five air quality categories: very good, good, medium, poor, and very poor. The pollutants considered are five: O3, NO2, SO2, PM10, and PM2.5. The scheme to evaluate the EU-AQI is depicted in Table 2. First, a daily sub-index is calculated for each pollutant by linear interpolation (Equation (3)) of the measured pollutant concentrations between the concentration levels in μg/m 3 reported in Table 2.
where [Y] min and [Y] max are the minimum and maximum concentrations of the range to which the index refers, respectively. Then, the daily EU-AQI is assumed as the maximum of the calculated sub-indices (Equation (4)).
We introduce some modifications in the evaluation of the EU-AQI due to the absence of all the mandatory data but maintain the significance of the result: (i) SO2 concentration is generally low, and its sub-index is never the maximum. Therefore, it does not contribute to the EU-AQI. For this reason, SO2 was not considered in our calculation. (ii) In traffic stations, the absence of O3 data was not considered a limitation to calculating the EU-AQI;

Description of the Air Quality Index Adopted
To have information on the historical trend of air quality instead of that of single pollutants, the EU-AQI was adopted with some minor modifications in the calculation procedure.
The EU-AQI is a daily air quality index that assumes numerical values from 1 to 5, corresponding with five air quality categories: very good, good, medium, poor, and very poor. The pollutants considered are five: O 3 , NO 2 , SO 2 , PM10, and PM2.5. The scheme to evaluate the EU-AQI is depicted in Table 2. First, a daily sub-index is calculated for each pollutant by linear interpolation (Equation (3)) of the measured pollutant concentrations between the concentration levels in µg/m 3 reported in Table 2.
where [Y] min and [Y] max are the minimum and maximum concentrations of the range to which the index refers, respectively. Then, the daily EU-AQI is assumed as the maximum of the calculated sub-indices (Equation (4)).
AQI s,g = max(SI g,i ) We introduce some modifications in the evaluation of the EU-AQI due to the absence of all the mandatory data but maintain the significance of the result: (i) SO 2 concentration is generally low, and its sub-index is never the maximum. Therefore, it does not contribute to the EU-AQI. For this reason, SO 2 was not considered in our calculation. (ii) In traffic stations, the absence of O 3 data was not considered a limitation to calculating the EU-AQI; (iii) in the same way, if one set of data was absent between PM10 or PM2.5, the EU-AQI was evaluated anyway.

Method for Calculating the Impact on Human Health
AirQ+ is a software developed by the WHO to quantify the impact of air pollution on health due to PM2.5, PM10, NO 2 , O 3, and black carbon (BC). For each pollutant, the software allows us to develop two possible impact perspectives: acute effects related to short-term exposure, and chronic effects related to long-term exposure. AirQ+ is based on methodologies and functions that associate a certain health response to different levels of pollution, all based on epidemiological studies and the systematic review and meta-analysis of the data available up to 2013, mainly carried out by studies conducted in Western Europe and North America.
For inputs, the software requires (i) air quality data (the annual average concentration for long-term exposure effects or daily average concentration for short-term exposure); (ii) data of the population at risk (for example, the total number of adults aged ≥30 years); (iii) the health data of the resident population; (iv) a limit value to be taken into account (e.g., 10 µg/m 3 for PM2.5 as recommended by WHO) [36]; (v) population data and mortality data, both stratified by age, when using the life table analysis.
By default, the software provides relative risk values (RR) which estimate the extent of the association between exposure and a specific health endpoint. It estimates the probability with a 95% confidence interval of obtaining a certain effect on health in the exposed population compared to an unexposed population.
For the calculation of relative risk (RR), the software uses the following log-linear function: where β is the change in RR for a change of one unit of concentration; x is the concentration of the pollutant; and x 0 is a limit value (cut-off). Associating the exposed population (P) with the relative risk (RR), the attributable proportion (AP) is the attributed fraction of the health endpoint due to exposure in a given population and can be calculated using From the AP, the rate (or the number of cases per population unit) attributed to exposure (BE) in the population (B) is calculated as For a population of a given size N, this can be converted into the estimated number of cases attributed to exposure: Consequently, the frequency of the outcome in the exposure-free population can be estimated as In this work, only the chronic effects are calculated. Starting from the hourly data obtained from the ARPAC, the daily averages for each pollutant at each station are obtained. To get a valid 24-h average, an efficiency criterion >75% is fixed, i.e., at least 18 hourly average values out of 24 are necessary. Then, an annual average is evaluated if at least 50% of the 24-h valid average concentration data are available in the year.
The results obtained are the mortality (for all ages) for NO 2 , the incidence of chronic bronchitis in adults for PM10, and the incidence of mortality in the adult population (>30 years) for PM2.5. The "incidence" data refers to the basic incidence of a certain effect on the health of the exposed population. Since, in this case, we refer to mortality associated with all age groups, then we insert the mortality index of the population exposed to the risk, which is the total one.
The results obtained are expressed by the program in three equivalent forms: The number of cases compared to the exposed population; • The number of cases per 100,000 people exposed.
The result indicates, for a given health endpoint, the percentage of cases that could be avoided if the pollutant concentration did not exceed the threshold indicated by us in the cut-off value which coincides with WHO air quality guidelines.
AirQ+ is a useful software to determine the exposure to the air pollution and was developed on the basis of the scientific evidence evaluated within expert groups in determining the air pollutant-health outcome relationships. For this reason, AirQ+ has been often used in studying the health impacts of poor air quality. However, the tool needs some improvements to minimize its limitations, in particular, for some air pollutants such as NO2, BC, and ozone (long-term exposure), where the evidence of health outcome relationships is not as strong [29] and for evaluating the risk related to air pollution mixtures and not only to single pollutants [37].

Historical Trends of Pollutants
The historical trends are represented for each pollutant through several parameters: the number of times the limit value is exceeded and the annual average, or percentiles. Due to the large amount of data, data are spatially averaged per province. The number of exceedances is obtained by summing all the exceedances registered in the stations of the same province and dividing them by the number of stations active in that year. The annual average and percentiles are calculated as the average of all the receptors present in the same province. Figure 3a shows, for NO 2 , the number of exceedances of the 1-h average limit value of 200 µg/m 3 evaluated as the total number of exceedances registered in each province per year divided by the number of active receptors. Figure 3b reports, for each province, the annual NO 2 average of all the active NO 2 receptors in the year. Even though data of some years (2004, 2005, and 2013) are missing for all provinces because the monitoring efficiency was below the efficiency criteria adopted (>50% of data in the year), Figure 3a,b clearly show a decreasing trend of NO 2. In fact, since 2016, the average number of exceedances per province is near to zero, and the annual average is below the limit value of 40 µg/m 3 (Figure 3), while before 2010 the number of exceedances was often higher than 5, and the annual average higher than the limit value of 40 µg/m 3 .
Two statistical parameters are explored to analyze the historical trend of PM10, the number of exceedances of the 24-h limit value of 50 µg/m 3 , and the average annual concentration. In both cases, they are the average values in the province (Figure 4).
Additionally, in this case, there is a lack of data for 2004, 2005, 2010, and 2013. Therefore, the individuation of a trend is more difficult. In general, an increasing trend is observed from 2007 to 2009. However, from 2015 to 2019, a reduction in the concentrations of PM10 is registered. In Figure 4b, from 2015, the average annual concentration is always lower than the limit value of 40 µg/m 3 , but the guideline imposed by the WHO of 20 µg/m 3 is always exceeded.   The monitoring of PM2.5 in the Campania region started in 2007. Therefore, less data is available with respect to other pollutants. Considering the annual average, a reduction is evident starting in 2015 ( Figure 5). All the provinces are within the European limit value of 25 μg/m 3 , but the guideline of 10 μg/m 3 proposed by the WHO [38] is still exceeded ( Figure 5).  The monitoring of PM2.5 in the Campania region started in 2007. Therefore, less data is available with respect to other pollutants. Considering the annual average, a reduction is evident starting in 2015 ( Figure 5). All the provinces are within the European limit value of 25 µg/m 3 , but the guideline of 10 µg/m 3 proposed by the WHO [38] is still exceeded ( Figure 5). The historical trend of O3 is shown in Figure 6, where the average number of exceedances, per province, of the EU-limit value of 120 μg/m 3 of the maximum daily values of 8h mobile average is reported. Due to the scatter of data, it is not possible to clearly identify any trend. However, since 2016, an increasing trend is evident for two provinces (Avellino and Salerno). The highest average number of exceedances is observed in the province of Avellino, which has the highest average level on the sea.  The historical trend of O 3 is shown in Figure 6, where the average number of exceedances, per province, of the EU-limit value of 120 µg/m 3 of the maximum daily values of 8-h mobile average is reported. Due to the scatter of data, it is not possible to clearly identify any trend. However, since 2016, an increasing trend is evident for two provinces (Avellino and Salerno). The highest average number of exceedances is observed in the province of Avellino, which has the highest average level on the sea. The historical trend of O3 is shown in Figure 6, where the average number of exceedances, per province, of the EU-limit value of 120 μg/m 3 of the maximum daily values of 8h mobile average is reported. Due to the scatter of data, it is not possible to clearly identify any trend. However, since 2016, an increasing trend is evident for two provinces (Avellino and Salerno). The highest average number of exceedances is observed in the province of Avellino, which has the highest average level on the sea.

Historical Trends of the EU-AQI
The historical trend of air quality is reported in Figure 7 as the percentage of days in which the EU-AQI was greater than or equal to 3 (air quality from medium to very poor).
The EU-AQI annual for each province is evaluated by dividing the number of times that the EU-AQI is greater than or equal to 3 by the number of days in which the EU-AQI was evaluated for all stations in each province. Figure 7 shows a peak of air pollution in 2009 and an improvement in air quality since 2016. In fact, the percentage of the occurrences of EU-AQI ≥ 3 decreases from 2016 to 2019.

Historical Trends of the EU-AQI
The historical trend of air quality is reported in Figure 7 as the percentage of days in which the EU-AQI was greater than or equal to 3 (air quality from medium to very poor).
The EU-AQI annual for each province is evaluated by dividing the number of times that the EU-AQI is greater than or equal to 3 by the number of days in which the EU-AQI was evaluated for all stations in each province. Figure 7 shows a peak of air pollution in 2009 and an improvement in air quality since 2016. In fact, the percentage of the occurrences of EU-AQI ≥ 3 decreases from 2016 to 2019.

Historical Trends of the Impact on the Human Health
The impact of air pollution on human health in Campania was evaluated using the software AIRQ+, as described in the Methodology section, from 2003 to 2018. The results are reported in Figure 8.
The trend of the impact of NO2 measured as the percentage of mortality cases due to all natural causes attributable to NO2 concentrations is clearly decreasing (Figure 8 (up)), passing from a maximum of 10% in 2010 to values <4% in 2018. Generally, the highest percentages are observed in the province of Naples. A less-clear decreasing trend is also observed for PM10 (Figure 8 (middle)) since 2008. In this case, the effect is measured as the percentage of cases of chronic bronchitis and hospitalizations for respiratory diseases attributable to PM10. Finally, the bottom diagram of Figure 8 shows the percentage of cases of mortality from natural causes, such as COPD, lung cancer mortality, mortality from ischemia, and mortality from stroke related to adulthood (30+) attributable to PM2.5 concentrations. Additionally, for PM2.5, a decreasing trend is observed since 2010.
In general, for all three pollutants, the impact on human health decreases over time. This is consistent with the previously reported decreasing concentration trends.

Historical Trends of the Impact on the Human Health
The impact of air pollution on human health in Campania was evaluated using the software AIRQ+, as described in the Methodology section, from 2003 to 2018. The results are reported in Figure 8.
The trend of the impact of NO 2 measured as the percentage of mortality cases due to all natural causes attributable to NO 2 concentrations is clearly decreasing (Figure 8 (up)), passing from a maximum of 10% in 2010 to values <4% in 2018. Generally, the highest percentages are observed in the province of Naples. A less-clear decreasing trend is also observed for PM10 (Figure 8 (middle)) since 2008. In this case, the effect is measured as the percentage of cases of chronic bronchitis and hospitalizations for respiratory diseases attributable to PM10. Finally, the bottom diagram of Figure 8 shows the percentage of cases of mortality from natural causes, such as COPD, lung cancer mortality, mortality from ischemia, and mortality from stroke related to adulthood (30+) attributable to PM2.5 concentrations. Additionally, for PM2.5, a decreasing trend is observed since 2010.
In general, for all three pollutants, the impact on human health decreases over time. This is consistent with the previously reported decreasing concentration trends. Atmosphere 2021, 12, x FOR PEER REVIEW 22 of 23

Discussion of Results
It is now necessary to highlight the limitations of the study. The main limitation relates to the missing data in some years due to poor monitoring efficiency, and the fact that some datasets do not meet the assumed quality criteria. Moreover, the analysis is focused exclusively on NO2, PM10, PM2.5, and O3. Other pollutants were not considered for different reasons. Data on benzene and SO2 were available only in few monitoring stations. Moreover, SO2 concentrations are generally very low; polycyclic aromatic hydrocarbons (PAHs) were not monitored routinely. CO was excluded from the analysis because its concentration levels are largely below the legislative limits.
Another limitation is the spatial representativeness of the fixed stations. Most of the streets in the largest urban agglomerates in Campania are narrow, and fixed stations cannot be installed. They are generally deep street canyons with an aspect ratio (height-towidth ratio) H/W >2 [39]. Information on the air quality in these streets is largely incomplete and derived only from spot-monitoring campaigns. However, all of these indicate a

Discussion of Results
It is now necessary to highlight the limitations of the study. The main limitation relates to the missing data in some years due to poor monitoring efficiency, and the fact that some datasets do not meet the assumed quality criteria. Moreover, the analysis is focused exclusively on NO 2 , PM10, PM2.5, and O 3 . Other pollutants were not considered for different reasons. Data on benzene and SO 2 were available only in few monitoring stations. Moreover, SO 2 concentrations are generally very low; polycyclic aromatic hydrocarbons (PAHs) were not monitored routinely. CO was excluded from the analysis because its concentration levels are largely below the legislative limits.
Another limitation is the spatial representativeness of the fixed stations. Most of the streets in the largest urban agglomerates in Campania are narrow, and fixed stations cannot be installed. They are generally deep street canyons with an aspect ratio (heightto-width ratio) H/W >2 [39]. Information on the air quality in these streets is largely incomplete and derived only from spot-monitoring campaigns. However, all of these indicate a critical situation, with concentration levels of benzene, carbon oxide, and particle number concentrations higher than those measured at the same time at the nearest fixed stations of the regional air quality network [40][41][42].
In spite of these limitations, the results reported give a clear indication of a decreasing historical trend of air pollution with respect to NO 2 , and a less evident one for PM10 and PM2.5. These findings are in agreement with those reported in the annual report on air quality in urban areas in Italy [43].
With the aim to get some findings on the possible causes determining the improvement of air quality in the Campania region, data reported in Figures 3-6 were re-evaluated, lumping all stations into two categories: "background-industrial" and "traffic" (Figures 9  and 10).
Atmosphere 2021, 12, x FOR PEER REVIEW 22 of 23 critical situation, with concentration levels of benzene, carbon oxide, and particle number concentrations higher than those measured at the same time at the nearest fixed stations of the regional air quality network [40][41][42].
In spite of these limitations, the results reported give a clear indication of a decreasing historical trend of air pollution with respect to NO2, and a less evident one for PM10 and PM2.5. These findings are in agreement with those reported in the annual report on air quality in urban areas in Italy [43].
With the aim to get some findings on the possible causes determining the improvement of air quality in the Campania region, data reported in Figures 3-6 were re-evaluated, lumping all stations into two categories: "background-industrial" and "traffic" (Figures 9 and 10).
The historical trends reported are confirmed. Traffic stations have higher pollution levels with respect to NO2, PM10, and PM2.5, while ozone concentration is higher in background-industrial stations. However, no significant differences are observed between the trends of the two categories.    As reported in the Methodology section, the daily quality index EU-AQI is determined by the pollutant with the highest sub-index in the day. It is interesting to determine which pollutant most frequently determines the EU-AQI value. Figure 11 shows, for each pollutant, the percentage of occurrences of the maximum sub-index (the number of days with a maximum sub-index divided by the number of validated daily quality indexes in the year) from 2003 to 2019 across the whole region. It is evident that PM (PM10 and PM2.5) plays a dominant role from 2014 to 2019, with the maximum daily sub-index generally being met in more than 50% of traffic stations (Figure 11 (bottom)). It is also evident that, in the same period, ozone has an increasing trend in the industrial-background stations ( Figure 11 (up)). The historical trends reported are confirmed. Traffic stations have higher pollution levels with respect to NO 2 , PM10, and PM2.5, while ozone concentration is higher in background-industrial stations. However, no significant differences are observed between the trends of the two categories.
As reported in the Methodology section, the daily quality index EU-AQI is determined by the pollutant with the highest sub-index in the day. It is interesting to determine which pollutant most frequently determines the EU-AQI value. Figure 11 shows, for each pollutant, the percentage of occurrences of the maximum sub-index (the number of days with a maximum sub-index divided by the number of validated daily quality indexes in the year) from 2003 to 2019 across the whole region. It is evident that PM (PM10 and PM2.5) plays a dominant role from 2014 to 2019, with the maximum daily sub-index generally being met in more than 50% of traffic stations (Figure 11 (bottom)). It is also evident that, in the same period, ozone has an increasing trend in the industrial-background stations (Figure 11 (up)). Since vehicular traffic is one of the main sources of NOx, the historical trend of the four-wheel vehicular fleet composition is reported in Figure 12. A reduction of the percentage of the old EURO classes (EURO 0-2-3) can be observed, with an increase of the new EURO classes (EURO 4-5-6). The EURO 4-5-6 technologies represent those with the Since vehicular traffic is one of the main sources of NOx, the historical trend of the four-wheel vehicular fleet composition is reported in Figure 12. A reduction of the percentage of the old EURO classes (EURO 0-2-3) can be observed, with an increase of the new EURO classes (EURO 4-5-6). The EURO 4-5-6 technologies represent those with the highest reduction in NOx emissions, and their percentage from 2015 to 2019 increased from 45% to 60%. It is interesting to observe the different spatial distribution of NO2, PM10, and O3. Contour maps referring to annual averages in 2019 ( Figure 13) show how NO2 is quite high in urban areas (Naples and Salerno), while O3 is prevalent in suburban and rural areas. It is interesting to observe the different spatial distribution of NO 2, PM10, and O 3 . Contour maps referring to annual averages in 2019 ( Figure 13) show how NO 2 is quite high in urban areas (Naples and Salerno), while O 3 is prevalent in suburban and rural areas. It is interesting to observe the different spatial distribution of NO2, PM10, and O3. Contour maps referring to annual averages in 2019 ( Figure 13) show how NO2 is quite high in urban areas (Naples and Salerno), while O3 is prevalent in suburban and rural areas.

Conclusions
The study analyzes the historical trends of air quality in the Campania region from 2003 to 2019. The trends for NO2, PM10, and PM2.5 are generally decreasing, in particular NO2. The decrease is less evident for PM10 and PM2.5, but in the former case, data are available only from 2007. Ozone shows an increasing trend from 2015 to 2019 in two provinces.
The trends measured through the EU-AQI confirm the results of single pollutants. PM10 and PM2.5 play a dominant role in determining the value of the EU-AQI, especially in traffic stations. However, in recent years, the prevalence of O3 is increasing, in particular at industrial-background stations.
Additionally, the impact of air pollution on human health expressed in terms of attributable fractions shows generally a decreasing trend.

Conclusions
The study analyzes the historical trends of air quality in the Campania region from 2003 to 2019. The trends for NO 2 , PM10, and PM2.5 are generally decreasing, in particular NO 2 . The decrease is less evident for PM10 and PM2.5, but in the former case, data are available only from 2007. Ozone shows an increasing trend from 2015 to 2019 in two provinces.
The trends measured through the EU-AQI confirm the results of single pollutants. PM10 and PM2.5 play a dominant role in determining the value of the EU-AQI, especially in traffic stations. However, in recent years, the prevalence of O 3 is increasing, in particular at industrial-background stations.
Additionally, the impact of air pollution on human health expressed in terms of attributable fractions shows generally a decreasing trend.
However, some limit values established by the European community are still not being respected, and the same goes for some WHO guidelines. Therefore, more efforts are necessary to reduce anthropogenic emissions.
The results of the study are limited due to missing data from years when the amount of data available does not meet the quality criteria. Another limitation is the incomplete representativeness of all the regional territory by the fixed stations. In fact, in the urban areas of Campania, many deep street canyons (H/W > 2) are present but are not monitored by fixed stations because they are very narrow. The results of several short-term monitoring campaigns performed inside the old center of Naples regarding benzene, CO, and particle number concentration suggest alarmingly poor air quality. At present, knowledge regarding the human health impacts of the air quality in the deep street canyons of Campania's urban agglomerates requires further study. The implementation of pervasive monitoring campaigns with the use of low-cost sensors [44] could be a possible solution for the effective air quality monitoring of these streets.