Skip to main content

Environmental assessment of interventions to restrain the impact of industrial pollution using a quasi-experimental design: limitations of the interventions and recommendations for public health policy

Abstract

Background

In an industrial area, the asymmetry between the weights of the economic interests compared to the public-health needs can determine which interests are represented in decision-making processes. This might lead to partial interventions, whose impacts are not always evaluated. This study focuses on two interventions implemented in Taranto, Italy, a city hosting one of the largest steel plants in Europe. The first intervention deals with measures industrial plants must implement by law to reduce emissions during so called “wind days” in order to reduce PM10 and benzo [a] pyrene concentrations. The second one is a warning to the population with recommendations to aerate indoor spaces from 12 pm to 6 pm, when pollutant concentrations are believed to be lower.

Methods

To analyse the impact of the first intervention, we analysed monthly PM10 data in the period 2009–2016 from two monitoring stations and conducted an interrupted-time-series analysis. Coefficients of time-based covariates are estimated in the regression model. To minimise potential confounding, monthly concentrations of PM10 in a neighbourhood 13 km away from the steel plant were used as a control series. To evaluate the second intervention, hourly concentrations of PM10, SO2 and polycyclic-aromatic-hydrocarbons (PAHs) were analysed.

Results

PM10 concentrations in the intervention neighbourhood showed a peak just a few months before the introduction of the law. When compared to the control series, PM10 concentrations were constantly higher throughout the entire study period. After the intervention, there was a reduction in the difference between the two time-series (− 25.6%). During “wind days” results suggested no reduction in concentrations of air pollutants from 12 pm to 18 pm.

Conclusion

Results of our study suggest revising the warning to the population. Furthermore, they evidence that in complex highly industrialised areas, air quality interventions cannot focus on only a single pollutant, but rather should consider the complex relationships between the different contaminants. Environmental interventions should be reviewed periodically, particularly when they have implications for social constraints. While the results of our study can be related only to the specific situation reported in the article, the methodology applied might be useful for the environmental management in industrial areas with similar features.

Peer Review reports

Background

Addressing environmental issues is a complex process since these are produced by a large variety of factors and involve ecological, social, economic, and political dimensions, which are intrinsically correlated and interact with each other. The interests at stake, the different perspectives of the various social actors, and the intrinsic uncertainties and complexity of the systems make it particularly difficult to identify adequate technical-political solutions.

In terms of governance, an approach that focuses only on one aspect while neglecting the others is not able to sufficiently capture the variety of such complexity due to effects and feedbacks that are entirely unpredictable. This may be the case for interventions dealing with a single pollutant, neglecting the connections between it and other contaminants, or it could be a situation where the acceptability and social implications of an environmental measure are underestimated, resulting in an increased burden of distress on particular groups of people [1,2,3,4].

In an industrial area, the asymmetry between the weights of the economic interests compared to the public health aspects can determine which interests are represented in decision-making processes. As a consequence, this often leads to the implementation of partial environmental interventions, whose impacts are not always evident and relevant [5].

In this article, we discuss the specific case of Taranto, a southern Italian city (Fig. 1) where one of the largest steel-processing plants in Europe is situated. Several authors have reported negative health effects due to air pollution in this area [6,7,8,9,10]. In particular, an increased risk of mortality for all causes combined and specifically for lung cancer, respiratory diseases, and pleural mesothelioma was reported. Furthermore, excesses of cancer incidence have been observed in the youngest age classes for lymphoma and non-Hodgkin lymphoma, thyroid cancer, germ cell tumours, trophoblastic tumours, and gonad neoplasms [10].

Fig. 1
figure 1

a The area under study, including the neighbourhood targeted by the intervention (Tamburi), the surface of the steel plant, and the monitoring stations located in Tamburi and in the neighbourhood assumed as the control (Talsano). b Frequency of hourly wind measurements plotted by wind direction, with colour bands showing wind speed ranges. Taranto, 2006–2016. Map source: QGIS Development Team (YEAR). QGIS Geographic Information System. Open Source Geospatial Foundation Project. http://qgis.osgeo.org

Additionally, studies have shown associations between health outcomes and socio-economic deprivation, and identified an environmental justice issue whereas the most polluted neighbourhoods are also those with the highest deprivation [11, 12].

The steel plant in Taranto was established in 1959 as a publicly owned plant and was downsized during the 80’s and sold to a private owner in 1995. It still produced nearly 8 million tons of steel (30% of the Italian output) in 2011, respectively around 0.06 and 75% of national and provincial GDP [13]. The plant covers a surface of 15 km2 including 200 km of railway tracks, 50 km of roads, 190 km of conveyor belts, and large open-air mineral stockyards (Fig. 1, panel a). The wind rose in the area for the period 2006–2016 shows that the most frequent and intense winds are from the NW. Another prevailing and persistent wind system is associated with southern winds, while winds from other sectors are more due to local circulations (Fig. 1, panel b) [14].

In 2010, the Court of Taranto had requested an epidemiological and environmental study [10, 15], the results of which found that outcomes were associated with increased levels of PM10 originating from the industrial site, particularly among the population living close to the industrial area [15]. In the environmental study, PM emissions from the steel plant were estimated to be more than 50 tons yearly [16]. Together with other PM emissions occurring at other points of the storage cycle, the yearly amount increased to 668 tons. Storage and handling of primary materials in the stockyards are a consistent source of particulate matter generated by wind erosion. Other emissions occur at several points in the storage cycle [17]: material loading onto and out from the pile and from the movement of trucks and loading equipment into the storage pile area. After some restrictions imposed during the legal trial in 2012, the steel production was roughly halved.

In 2014, the European Commission invited Italy to urgently address this severe pollution issue arguing that: “Italy is failing to ensure that ILVA (steel plant in Taranto) operates in conformity with EU legislation on industrial emissions, with potentially serious consequences for human health and the environment.” In addition, “The Commission has previously sent Italy two letters of formal notice, in September 2013 and April 2014, urging the Italian authorities to take measures in order to bring the operation of the ILVA plant into compliance with the Industrial Emissions Directive and other applicable EU environmental law” [18].

Following the increasing pressure from citizens’ organizations and legal initiatives of the local court, air quality and public health interventions have been implemented to reduce the impact of industrial emissions. These interventions were inspired by a Canadian experience of environmental management [19].

The aim of this article is to evaluate both a law introduced by the Apulia Regional Government to mitigate the impact of industrial emissions and a warning for the population introduced by local health authorities.

The first intervention we evaluate here is the “wind-days” law [20]. This was enforced in 2012 to protect the city, especially the high-density and deprived residential area (Tamburi) located less than 1 km from the steel plant and downwind of the north/west winds (Fig. 1, panel a). This law is a compendium of norms that the steel-producing plant and other industries must adhere to in order to reduce their emissions in specific weather conditions; previously, increased concentrations of pollutants had been documented in the neighbourhood targeted by the intervention [14]. Technical details of the law are discussed elsewhere (Mangia et al. 2020). For the purposes of this paper, it suffices to know that the regional environmental authorities defined “wind days” as those during where:

  1. i.

    The wind direction is in the range of 270°-360°,

  2. ii.

    The speed of the wind is predicted to be greater than 6.7 m/s for at least three consecutive hours.

Wind days are forecasted by a meteorological modelling system 72 h in advance and communicated to the industries 48 h before windy events might occur. Following a forecasting of a wind-day, industries must implement initiatives aimed at reducing the volume and the impact of industrial activities on the neighbourhood areas.

The second intervention was enforced in 2015 and consist of a warning to the residents in Tamburi to aerate indoor environments during winter, preferably between 12 pm and 6 pm in case a wind-day is forecasted by the Regional Environmental Agency [21].

Methods

For the years 2009–2019, PM10 and SO2 data recorded by the regional environmental authorities at the following two monitoring stations were analysed (Fig. 1):

  1. a)

    The first station, named “Machiavelli”, is located in Tamburi, the neighbourhood close to the industrial area and the mineral stockyards;

  2. b)

    The second is located in Talsano, about 13 km from the industrial area.

Daily mean PM10 concentrations have been computed using available hourly data. Following standard protocols, daily concentrations were discarded if more than five hourly values for one 24-h period were missing [7, 8].

In order to evaluate the “wind-day” law, average monthly PM10 concentrations were considered for the analysis for the period 2009–2016.

To evaluate the warning to the population, hourly concentrations of PM10, PAHs (polycyclic aromatic hydrocarbons) and SO2 measured in the period 2015–2019 at the Machiavelli monitoring station were used to compute the pollutants’ average daily concentration profiles. Then, each day was flagged as true positive (TP), true negative (TN), false positive (FP), or false negative (FN) according to the ex-post evaluations of forecasting performed by the regional environmental authority that was made periodically available on the internet [22]. After having checked the weather conditions, we changed the classification for just 2 days (Jan 16, 2016, and May 22, 2017), updating them to FN from TN.

In order to define the impact of the “wind-days” law, an interrupted time series study design was used [23, 24].

A segmented linear regression model was implemented to study the monthly concentrations of PM10 in Tamburi before and after the introduction of the “wind-day” law. This design permits the evaluation of whether the intervention produced a discontinuity in comparison with the underlying secular trend [25]. In its plainest form, three coefficients of time-based covariates are estimated in the regression model, which indicates the pre-intervention slope, the change in level at the intervention time, and the change in slope from pre-intervention to post-intervention [26]. In order to minimise potential confounding, due, for example, to a change in the meteorology such as the frequency of Saharan dust incursions [27], monthly concentrations of PM10 in a neighbourhood 13 km away from the steel plant (Talsano) were used as a control series [28]. For power purposes, an equal number of time points before and after the intervention were assumed [29].

Thus, the following model was used [30, 31]:

$$ {Y}_t={\beta}_0+{\beta}_1{T}_t+{\beta}_2{X}_t+{\beta}_3{X}_t{T}_t+{\beta}_4Z+{\beta}_5Z{T}_t+{\beta}_6{ZX}_t+{\beta}_7Z{X}_t{T}_t+{\varepsilon}_t $$

Whereby (Fig. 2):

Fig. 2
figure 2

Interrupted time series design with the use of a control series

Yt is the concentration of PM10 measured in the neighbourhood close to the industrial area (Tamburi) at each equally spaced time point t varying from 1 (January 2009) to 96 (December 2016).

β0 is baseline PM10 concentration in the control series (Talsano);

β1 is the slope of PM10 concentrations in the control series pre-intervention;

β2 is the change in the level of PM10 concentrations in the control series post-intervention;

β3 is the difference in the slopes of the PM10 concentrations in the control series pre-intervention and post-intervention;

Xt represents the intervention and is equal to 0 before the intervention and equal to 1 after;

Z is a dummy variable equal to 1 for the time series under study (Tamburi) and equal to zero for the control series (Talsano);

β4 is the difference in the level of PM10 concentrations between Tamburi and Talsano pre-intervention;

β5 is the difference in the slope of PM10 concentrations between Tamburi and Talsano pre-intervention;

β6 is the difference in the level of PM10 concentrations between Tamburi and Talsano immediately after the intervention;

β7 is the difference in the slope of PM10 concentrations between Tamburi and post-intervention (a difference-in-differences).

Tt represents the number of the months since the start of the study (1, 2, …, 96); Xt represents the intervention and is equal to 0 before the intervention and equal to 1 after;

Xt is a dummy variable. It represents the intervention and is equal to 0 before the intervention and equal to 1 after;

Z is a dummy variable. It is equal to 1 for the time series under study (Tamburi) and equal to zero for the control series (Talsano);

εt is the random error term.

Finally, we tested for autocorrelation using the Durbin-Watson test and adjusted regression standard errors for autocorrelation for the identified order [32]. 95% confidence intervals (95% CI) were calculated.

In order to define the impact of the warning to the population during winter, hourly concentrations of PM10, SO2 and PAHs were plotted. All days in the winter season during the period 2015–2019 were classified as follows:

  1. a.

    True positive days (TP), i.e. days forecasted from the regional authorities as wind-days and confirmed by later measures of meteorological variables;

  2. b.

    False positive (FP), i.e. days forecasted from the regional authorities as wind-days but not confirmed by later measures of meteorological variables;

  3. c.

    True negative days (TN), i.e. days forecasted from the regional authorities no to be wind-days and confirmed by later measures of meteorological variables;

  4. d.

    False negative days (FN), i.e. days forecasted from the regional authorities no to be wind-days but confirmed by later measures of meteorological variables.

Results

From 2009 to 2016, 303 wind-days were observed. In the post-intervention period, 44% fewer wind-days were registered (n = 109 days). The analysis of PM10 concentrations in Tamburi, during the observed period showed a peak just a few months before the introduction of the law (Fig. 3). When compared to the control series (Talsano), PM10 concentrations were constantly higher in Tamburi throughout the whole study period. After the intervention, there was a reduction in the difference between the two time-series (Fig. 3). PM10 concentrations in Talsano at the beginning of the observation period were equal to 24.1 μg/m3 (95% CI: 22.6–25.7), while in Tamburi they were 8.2 μg/m3 higher (95% CI: 5.6–10.8) (Table 1). Furthermore, in the period following the intervention, the difference in level was equal to 6.1 μg/m3 (− 11.2 – − 1.0), i.e. 2.1 μg/m3 less than in the previous period (− 25.6%).

Fig. 3
figure 3

PM10 concentrations (μg-2) measured by the monitoring stations located in the neighbourhood targeted by the intervention (Tamburi) (solid red line) and in the neighbourhood assumed as the control (Talsano) (black dashed line)

Table 1 Parameters estimate of the controlled interrupted time series model. Taranto. Years 2009-2016. The interruption is assumed as starting on the 1st November 2012

No difference in slopes pre- and post-intervention were observed (Table 1).

From 2015 to 2019, 159 days in the winter season were forecasted as wind-days. One hundred twenty-eight of them confirmed as wind-days by later measures of meteorological variables (true positive: 80.5%) (Table 2). One thousand five hundred seventy-three days were not forecasted as wind-days. However, 1531 of them were confirmed as not being wind-days (true negative: 97.3%) (Table 2).

Table 2 Classification of all days in the winter season during the period 2015-2019 as forecasted and/or confirmed to be wind days or not

Figure 4 shows the average daily profile of PM10 at the station for 2015–2019 for the months January to May and September to December, to which the warning to aerate indoor spaces from 12 pm to 6 pm, when PM10 concentrations are believed to be lower refers. The daily profile depends on meteorological conditions. During correctly predicted wind days (true positive), concentrations do not tend to decrease in the central hours of the day, i.e. the time period recommended to the population as ideal for indoor aeriation, but rather show an upward trend. Conversely, on true non-wind days (TN), there is a clear reduction in concentrations from 12:00 to 18:00. This different behaviour could be due to the prevalence of wind transport conditions with respect to the convective motion during highly windy conditions. The latter could inhibit the development of a boundary layer, with a consequent lower dilution of pollutants emitted from ground.

Fig. 4
figure 4

Average daily profile of PM10 for the years 2015–2019, from January to May and from September to December. Taranto, monitoring station Machiavelli (Tamburi)

It is interesting to note that in the unforeseen wind-days (false negative) there is a reduction of concentrations in the central hours. This could be related to the fact that wind conditions are not strong enough to be predicted adequately with a meteorological model and to influence the atmospheric boundary layer [27]. During false positive days (FP), the concentration profile is more flattened. This could also be due to some effect of the measures given to the companies or particular weather conditions to be further investigated with a larger dataset (Fig. 4).

The highest values of polycyclic aromatic hydrocarbons (PAHs) are recorded on non-wind days – both true negative and false positive – especially during the first half of the morning. A possible explanation is that weather conditions such as calm wind may establish a short-range diffusion of pollutants. Since the control of benzo [a] pyrene (one chemical among the PAHs) was one reason to establish rules for minimizing the impact of industrial activities, this result supports the thesis that wind days are not a unique weather condition impacting the Tamburi neighbourhood (Fig. 5).

Fig. 5
figure 5

Average daily profile of PAHs for the years 2015–2019, from January to May and from September to December. Taranto, monitoring station Machiavelli (Tamburi)

Figure 6 shows the hourly trend of SO2 as a proxy of industrial combustion pollutants [14]. The profiles are very different from those observed for the PM10 in the four categories. During wind days, SO2 concentrations are higher than in all other weather conditions, with a marked increase in the central hours of the day, i.e. the time range that is the subject of the population warning.

Fig. 6
figure 6

Average daily profile of SO2 for the years 2015–2019, from January to May and from September to December. Taranto, monitoring station Machiavelli (Tamburi)

The difference in the behaviour of the PM10 and SO2 can be explained by the different emission sources. While a large contribution to the ground concentrations of PM10 comes from surface sources, SO2 concentrations are more related to industrial combustion processes and come from sources at different altitudes (in the case of Taranto from heights of 10 to 312 m). The rise of plumes from combustion sources is strongly affected by external weather conditions. Very strong wind conditions tend to inhibit such a rise, with a consequent decrease of the plume dilution and increase of the ground concentrations.

Discussion

In the general settings of highly industrial and highly polluted city, this study aimed to evaluate two of the most recent air-quality and public health interventions. In accordance with the original idea of improving air quality in the residential area close to the steel plant, after the introduction of the specific regional law, a strong reduction in PM10 concentration was observed (− 25.6%).

To evaluate the impact of the first public health intervention, an interrupted time series design was used. This study design is considered to be the most effective and powerful tool among quasi experimental designs, and an important tool for specific intervention evaluation [33]. In order to control for potential confounders, a control time-series was considered in the model. Some authors argue that the two series should not differ before the intervention i.e. they should be similar in terms of slope and level [31]. Here, this condition was met only for the slope. In fact, both series differed only in the baseline level of PM10 concentrations. However, this condition was planned not to be met in order to allow consideration of a neighbourhood (Talsano) 13 km away from the steel plant that is less massively affected by the industrial emissions. An assumption of the interrupted time series study design is the continuity assumption, i.e. the absence of co-interventions [34]. In our study, a considerable decrease of production might have acted as a confounding factor to the evaluation of the intervention. However, because this information was not available at a monthly level, this factor was not accounted for in the regression model. Therefore, the effect of the intervention is likely overestimated and the observed reduction in PM10 is partially due to the concomitant contraction of steel production. Seasonality might bias the results if not accounted for [33]. However, since we used a control series and considered an almost even distribution of winter and summer months before and after the intervention, results can be expected not to be biased [24]. The analysis of rainfall shows fluctuations in accumulated rainfall, decreasing in the post-intervention period by about 18% compared to the first period. In the interrupted time series analysis, we did not account for this potential confounder. However, rainfall is accounted for by design, while considering a control series. Thus, even if we cannot completely exclude that the amount and frequency of precipitations differ between the intervention- and the control-series, the confounding effect might be supposed to be residual.

For the second intervention, results suggested no reduction in concentrations of air pollutants from 12 to 18 pm. Thus, a revision of this warning to the population is needed.

One of the limitations of the study is the scarcity of monitoring stations and pollutant measurements in the area. The available measured pollutants, SO2 and PM10, are only partially representative of the pollutants targeted by the interventions. Specifically, there is no dimensional analysis of the dust emitted by the plant, nor is there a continuous measurement of benzo [a]pyrene. Furthermore, an analysis of PM2.5 concentrations was not feasible because data were not available in the period under study at the monitoring station used as control.

Conclusions

The results indicate the need to revise the warning given to the residents during wind-days. In fact, there is no evidence supporting it since under certain circumstances an increase in pollutant concentrations has been observed from 12 pm to 18 pm. Thus asking people to aerate indoor spaces in this time window might potentially harm their health.

In conclusion, while the results of our study can be related only to the specific case reported in the article, the methodology applied might be useful for the environmental management in industrial areas with similar features.

Policy recommendations.

In the light of the current study, a number of recommendations arise:

  1. 1.

    In complex highly industrialised areas, air quality interventions cannot focus on only a single pollutant. One has to consider the complex relationships between the different contaminants and focus on a set of targeted pollutants.

  2. 2.

    If an intervention is planned to be implemented on specific days, data analysis focusing on those specific days is required. In fact, unspecific analyses can be misleading.

  3. 3.

    Environmental interventions should be implemented with regular planned evaluation points. In addition, scenarios with potential changes and/or adjustments have to be anticipated.

  4. 4.

    Environmental interventions should be reviewed periodically, particularly when they have implications for social constraints.

Availability of data and materials

Rough data used in our article is public and available at the following links:

Meteorological data: http://www.webgis.arpa.puglia.it/meteo/index.php

Data on air quality: http://old.arpa.puglia.it/web/guest/qualita_aria

The datasets used and/or analysed during the current study are also available from the corresponding author on reasonable request.

Abbreviations

EU:

European Union

FN:

False negative

FP:

False positive

GDP:

Gross Domestic Product

PAHs:

Polycyclic Aromatic Hydrocarbons

PM:

Particulate Matter

PM10 :

Particulate Matter with aerodynamic diameters smaller than 10 μm

SO2 :

Sulphur dioxide

TN:

True negative

TP:

True positive

References

  1. Kirschke S, Newig J. Addressing complexity in environmental management and governance. Sustainability. 2017;9(6):983.

    Article  Google Scholar 

  2. Biddle JC, Koontz TM. Goal specificity: a proxy measure for improvements in environmental outcomes in collaborative governance. J Environ Manag. 2014;145:268–76. https://doi.org/10.1016/j.jenvman.2014.06.029.

    Article  Google Scholar 

  3. Dewulf A, Craps M, Bouwen R, Taillieu T, Pahl-Wostl C. Integrated management of natural resources: dealing with ambiguous issues, multiple actors and diverging frames. Water Sci Technol. 2005;52(6):115–24. https://doi.org/10.2166/wst.2005.0159.

    CAS  Article  PubMed  Google Scholar 

  4. Macura B, Suskevics M, Garside R, Hannes K, Rees R, Rodela R. Systematic reviews of qualitative evidence for environmental policy and management: an overview of different methodological options. Environ Evid. 2019;8(1):1.

    Article  Google Scholar 

  5. Burns J, Boogaard H, Polus S, Pfadenhauer LM, Rohwer AC, van Erp AM, et al. Interventions to reduce ambient particulate matter air pollution and their effect on health. Cochrane Database Syst Rev. 2019;5:CD010919. https://doi.org/10.1002/14651858.CD010919.pub2.

    Article  PubMed  Google Scholar 

  6. Leogrande S, Alessandrini ER, Stafoggia M, Morabito A, Nocioni A, Ancona C, et al. Industrial air pollution and mortality in the Taranto area, Southern Italy: A difference-in-differences approach. Environ Int. 2019;132:105030. https://doi.org/10.1016/j.envint.2019.105030.

    CAS  Article  PubMed  Google Scholar 

  7. Alessandrini ER, Faustini A, Chiusolo M, Stafoggia M, Gandini M, Demaria M, et al. Air pollution and mortality in twenty-five Italian cities: results of the EpiAir2 project. Epidemiol Prev. 2013;37(4–5):220–9.

    PubMed  Google Scholar 

  8. Scarinzi C, Alessandrini ER, Chiusolo M, Galassi C, Baldini M, Serinelli M, et al. Air pollution and urgent hospital admissions in 25 Italian cities: results from the EpiAir2 project. Epidemiol Prev. 2013;37(4–5):230–41. https://doi.org/10.1289/isee.2013.O-3-23-01.

    Article  PubMed  Google Scholar 

  9. Vigotti MA, Mataloni F, Bruni A, Minniti C, Gianicolo EA. Mortality analysis by neighbourhood in a city with high levels of industrial air pollution. Int J Public Health. 2014;59(4):645–53. https://doi.org/10.1007/s00038-014-0554-x.

    Article  PubMed  Google Scholar 

  10. Zona A, Iavarone I, Buzzoni C, Conti S, Santoro M, Fazzo L, et al. SENTIERI: Epidemiological study of residents in national priority contaminated sites. Fifth Report. Epidemiol Prev. 2019;43(2–3 Suppl 1):1–208.

    PubMed  Google Scholar 

  11. Mataloni F, Stafoggia M, Alessandrini E, Triassi M, Biggeri A, Forastiere F. A cohort study on mortality and morbidity in the area of Taranto, Southern Italy. Epidemiol Prev. 2012;36(5):237–52.

    PubMed  Google Scholar 

  12. Gianicolo EA, Mangia C, Cervino M. Investigating mortality heterogeneity among neighbourhoods of a highly industrialised Italian city: a meta-regression approach. Int J Public Health. 2016;61(7):777–85. https://doi.org/10.1007/s00038-016-0868-y.

    Article  PubMed  Google Scholar 

  13. Greco L, Bagnardi F. In the name of science and technology: the post-political environmental debate and the Taranto steel plant (Italy). Environ Value. 2018;27(5):489–512. https://doi.org/10.3197/096327118X15321668325939.

    Article  Google Scholar 

  14. Mangia C, Gianicolo EA, Bruni A, Vigotti MA, Cervino M. Spatial variability of air pollutants in the city of Taranto, Italy and its potential impact on exposure assessment. Environ Monit Assess. 2013;185(2):1719–35. https://doi.org/10.1007/s10661-012-2663-4.

    CAS  Article  PubMed  Google Scholar 

  15. Forastiere F, Biggeri A, Triassi M. Perizia Epidemiologica 2012 [Available from: available on Environment Ministry site http://aia.minambiente.it/DomandeAIADocumenti.aspx?id= in folder Varie].

    Google Scholar 

  16. Sanna M, Monguzzi R, Santilli N, Felici R. Perizia conferita il giorno 8 novembre 2010 dal Giudice per le Indagini Preliminari Dottoressa Patrizia Todisco e consegnata in Gennaio 2012. Taranto: Tribunale di Taranto, Ufficio del Giudice per le Indagini Preliminari; 2011.

  17. United States Environmental Protection Agency. AP-42: Compilation of Air Emissions Factors, 2006. www3.epa.gov/ttn/chief/ap42/ch13/index.html.

    Google Scholar 

  18. European Commission (page 1). European commission urges Italy to address severe pollution issues at Europe's biggest steel plant, press release IP/14/1151. 2014.

    Google Scholar 

  19. Al-Dabbagh O. Omar Al-DabbaghPage 1Hamilton’s AirQuality: status and expected. 2011. https://www.eng.mcmaster.ca/sites/default/files/uploads/omar_al-dabbagh.pdf.

    Google Scholar 

  20. Piano contenente le prime misure di intervento per il risanamento della Qualità dell'Aria nel quartiere Tamburi (TA). 2012. [Available from: https://trasparenza.regione.puglia.it/informazioni-ambientali/fattori-inquinanti/piano-di-risanamento-pm10-e-bap-comune-di-taranto-q-tamburi].

  21. Misure cautelative in occasione di possibili criticità dello stato di qualità dell’aria a Taranto. 2012. [Available from: https://www.comune.taranto.it/images/Misure_Cautelative_Wind_Day.pdf].

  22. Menegotto M. Report meteo di riscontro wind day. Bari: ARPA Puglia; 2019.

    Google Scholar 

  23. Gianicolo EAL, Eichler M, Muensterer O, Strauch K, Blettner M. Methods for evaluating causality in observational studies. Dtsch Arztebl Int. 2020;116(7):101–7. https://doi.org/10.3238/arztebl.2020.0101.

    Article  PubMed  Google Scholar 

  24. Bernal JL, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol. 2017;46(1):348–55. https://doi.org/10.1093/ije/dyw098.

    Article  PubMed  Google Scholar 

  25. Ramsay CR, Matowe L, Grilli R, Grimshaw JM, Thomas RE. Interrupted time series designs in health technology assessment: lessons from two systematic reviews of behavior change strategies. Int J Technol Assess Health Care. 2003;19(4):613–23. https://doi.org/10.1017/S0266462303000576.

    Article  PubMed  Google Scholar 

  26. Kontopantelis E, Doran T, Springate DA, Buchan I, Reeves D. Regression based quasi-experimental approach when randomisation is not an option: interrupted time series analysis. BMJ. 2015;350(jun09 5):h2750. https://doi.org/10.1136/bmj.h2750.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Mangia C, Cervino M, Russo A, Singer S, Gianicolo EA. Effectiveness of an air quality intervention: an accountability study in a highly polluted industrial town. Air Qual Atmosphere Health. 2020;13:289–96.

    CAS  Article  Google Scholar 

  28. Lopez Bernal J, Cummins S, Gasparrini A. The use of controls in interrupted time series studies of public health interventions. Int J Epidemiol. 2018;47(6):2082–93. https://doi.org/10.1093/ije/dyy135.

    Article  PubMed  Google Scholar 

  29. Zhang F, Wagner AK, Ross-Degnan D. Simulation-based power calculation for designing interrupted time series analyses of health policy interventions. J Clin Epidemiol. 2011;64(11):1252–61. https://doi.org/10.1016/j.jclinepi.2011.02.007.

    Article  PubMed  Google Scholar 

  30. Linden A, Adams JL. Applying a propensity score-based weighting model to interrupted time series data: improving causal inference in programme evaluation. J Eval Clin Pract. 2011;17(6):1231–8. https://doi.org/10.1111/j.1365-2753.2010.01504.x.

    Article  PubMed  Google Scholar 

  31. Linden A. Conducting interrupted time-series analysis for single- and multiple-group comparisons. Stata J. 2015;15(2):480–500. https://doi.org/10.1177/1536867X1501500208.

    Article  Google Scholar 

  32. Newey WK, West KD. A simple, positive semi-definite, Heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica. 1987;55(3):703–8. https://doi.org/10.2307/1913610.

    Article  Google Scholar 

  33. Shadish W, Cook T, Campbell D. Experimental and quasi-experimental designs for generalized causal inference. Belmont: Wadsworth, Cengage LearningTM; 2002.

    Google Scholar 

  34. Bärnighausen T, Oldenburg C, Tugwell P, Bommer C, Ebert C, Barreto M, et al. Quasi-experimental study designs series-paper 7: assessing the assumptions. J Clin Epidemiol. 2017;89:53–66. https://doi.org/10.1016/j.jclinepi.2017.02.017.

    Article  PubMed  Google Scholar 

Download references

Acknowledgments

We wish to thank Katherine Taylor for helping revise and edit the English.

Funding

Funding information is not applicable. Open Access funding enabled and organized by Projekt DEAL.

Author information

Authors and Affiliations

Authors

Contributions

EG, MC and CM conceived work. AR, CM, MC and EG performed descriptive analysis. EG conducted the interrupted time series analysis. MB and SS supported in revising different versions of the manuscript and in discussing the results. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Emilio A. L. Gianicolo.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Gianicolo, E.A.L., Cervino, M., Russo, A. et al. Environmental assessment of interventions to restrain the impact of industrial pollution using a quasi-experimental design: limitations of the interventions and recommendations for public health policy. BMC Public Health 21, 1856 (2021). https://doi.org/10.1186/s12889-021-11832-3

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12889-021-11832-3

Keywords

  • Evaluation of interventions
  • Air quality
  • Steel industry
  • Taranto (southern Italy)
  • Interrupted time series