We performed a literature review to investigate how epidemiological studies have been used to assess the health consequences of living in the vicinity of industries. 77 papers on the chronic effects of air pollution around major industrial areas were reviewed. Major health themes were cancers (27 studies), morbidity (25 studies), mortality (7 studies), and birth outcome (7 studies). Only 3 studies investigated mental health. While studies were available from many different countries, a majority of papers came from the United Kingdom, Italy, and Spain. Several studies were motivated by concerns from the population or by previous observations of an overincidence of cases. Geographical ecological designs were largely used for studying cancer and mortality, including statistical designs to quantify a relationship between health indicators and exposure. Morbidity was frequently investigated through cross-sectional surveys on the respiratory health of children. Few multicenter studies were performed. In a majority of papers, exposed areas were defined based on the distance to the industry and were located from <2 km to >20 km from the plants. Improving the exposure assessment would be an asset to future studies. Criteria to include industries in multicenter studies should be defined.
Industrial areas are characterized by a high density of industries, sharing common infrastructures, such as transport networks, waste water treatment plants, and waste incineration plants. These areas cluster at-risk activities and pollution sources. They have historically attracted, and may still attract, hundreds of employees who have settled in the vicinity of the plants. With extensive urbanization, industrial areas have been embedded in the urban landscape, increasing the nuisances and the exposure of the population. For instance, in the South of France, the industrial area of l’etang de Berre hosts 430 industries classified for the protection of the environment and more than 60% of the Seveso II (referring to the European directive 96/82/CE) plants of the region. About 16 towns representing more than 300,000 inhabitants are exposed to the plumes produced by these plants [
People living near major industrial areas are facing complex situations of exposure: occupational and environmental exposure, multiexposure to chemicals combined with exposure to noise, dusts, visual pollution, stress, and so forth The possible associated health risks are of the highest concern to the population.
Quantitative health risk assessments, based on the comparison of a hypothetical exposure (assessed through measured or modeled concentrations in different matrices combined with scenarios of exposure) to toxicological reference values or to regulatory values, have been extensively used for regulatory purposes. They can point out problems with specific pollutants or route of exposure. For instance, several risk assessments around large French industrial areas found that the levels of some compounds, including benzene, particulate matter (PM), and SO2, could be considered too high [
In this paper, we performed a literature review of the published studies investigating the health of population exposed to industrial air pollution around major industrial sites. The objectives were
The work focused on studies investigating the chronic effects of air pollution from large industrial areas and major complexes grouping several plans or multicenter studies involving similar types of industry that could be or not part of larger industrial complexes.
Papers published between 1980 and 2012 were searched based on the Scopus database that includes PubMed and other relevant literature database. As an initial research using key words referring to industry retrieved very few papers, we searched epidemiological studies on the impacts of air pollution around point sources. On a second step, papers dealing with industries were selected based on the title and abstracts.
The initial search equation was ((“Air Pollutants” [MeSH] OR “Air Pollution” [MeSH]) AND “epidemiology” [Subheading]) OR ((Air pollution [Title/Abstract] OR Air pollutants [Title/Abstract]) AND (epidemiology OR epidemio* OR “Case control study” OR cohort OR “cross sectional study” OR prevalence OR incidence OR Surveillance OR survey OR “Health risk” OR “Risk assessment” OR health OR “Health effects” OR Exposure OR “Health impact*” OR Mortality OR “Adverse effects”)) AND (industry OR industrial) AND (residents OR Residential OR inhabitants OR neighborhood* OR vicinity OR “living area” OR “living near” OR surrounding* OR populations).
Papers were analyzed focusing on the types of industries, the study design, the health indicators, and the exposure assessment. The objectives were to identify the methods but not to discuss the results reported by each paper. To do so, reviews focusing on specific industries would be more relevant.
From the initial search 230 papers in English or French were selected based on their title. Based on the abstracts, 155 papers were excluded (58 environmental studies only, 35 looking at exposure through water, soil, or food and not air directly, 36 using industrial areas as one source among other air pollution sources, 10 description of the health of a population without links to exposure, 8 on nuclear installations, 4 toxicological studies, 3 studies focusing on acute exposure after an accident, and 1 literature review). Two reports from the grey literature were added, but no specific search was performed to identify such reports on a larger scale.
77 papers were finally included in the review, published between 1989 and 2011. While papers were available from many different countries, a majority of papers came from 3 European countries: the United Kingdom, Italy, and Spain (Table
Summary of the papers in the literature review.
Country | Total number of papers | Health outcome (several health outcomes may be described in 1 paper) | |||||
---|---|---|---|---|---|---|---|
Cancer | Morbidity | Biomonitoring | Mortality | Birth outcome | Mental health | ||
United Kingdom | 15 | 5 | 5 | 2 | 4 | 0 | 1 |
Italy | 9 | 3 | 3 | 2 | 1 | 0 | 0 |
Spain | 8 | 7 | 0 | 1 | 1 | 0 | 0 |
Taiwan | 7 | 4 | 0 | 0 | 0 | 3 | 0 |
Israel | 6 | 1 | 3 | 0 | 8 | 1 | 0 |
United States | 6 | 1 | 0 | 1 | 0 | 2 | 2 |
Canada | 5 | 1 | 4 | 0 | 0 | 0 | 0 |
Sweden | 4 | 1 | 1 | 2 | 0 | 0 | 0 |
France | 2 | 2 | 1 | 0 | 0 | 0 | 0 |
Thailand | 2 | 0 | 2 | 0 | 0 | 0 | 0 |
| |||||||
Countries with 1 study only | Finland, Lithuania | Argentina, Australia, Brazil, India, Romania, South Africa | Korea | ||||
| |||||||
Total number of studies |
27 | 25 | 9 | 7 | 7 | 3 |
The 27 studies on cancer are detailed in Table
Studies investigating cancer.
Reference | Country | Industrial background | Health outcome | Epidemiological design | Exposure assessment |
---|---|---|---|---|---|
Zambon et al., 2007 [ |
Italy | Industrial waste incinerators, Municipal solid waste incinerators, Medical waste incinerators, thermal power plants, oil refinery industrial plants for the production of primary aluminium | Visceral and |
Case control (72 cases and 405 controls) | Dispersion modeling (Industrial Source Complex Model in long-term mode, version 3 (ISCLT3)) |
Biggeri et al., 1996 [ |
Italy | Shipyard, iron foundry, incinerator, and Trieste city center | Lung cancer (mortality) | Case-control study (755 case-control pairs) | Distance and angle from each subject location to each pollution source |
Yu et al., 2006 |
Taiwan | Oil refinery | Leukemia | Case control (171 cases and 410 controls) | Distance, based on previous studies (3 km radius from the geographic centroid of any of the four petrochemical complexes) |
Simonsen et al., 2010 [ |
United States | Petrochemical industries | Lung cancer (registry) | Case control (455 cases and 437 controls) | Distance (0.5 miles, 1 mile, and 2 miles) |
Edwards et al., 2006 [ |
United Kingdom | Iron and steel, chemical, and heavy engineering industries | Lung cancer (registry) | Case-control study (204 cases and 339 controls) | Distance, guided by a validation study using data from historical records |
Petrauskaite et al., 2002 [ |
Lithuania | Production of mineral fertilizers, aluminum fluoride, and sulfuric acid | Lung cancer (mortality) | Case-control study (410 cases 410 controls) | Distance, based on measurements of sulfuric acid and the prevailing wind (6 km) |
Lopez-Cima et al., 2011 [ |
Spain | 23 industrial installations reporting to the EPER | Lung cancer | Case-control study (626 case, 626 controls) | Distance |
Pascal et al., |
France | Oil refining, oil storage, petrochemical and organic chemical activities, chlorine chemistry, steel and metal working, chemical plants, waste incineration plant, port | All cancers, lung cancer, bladder cancer, breast cancer, multiple myeloma, malignant non-Hodgkin’s lymphoma, and acute leukemia (hospitalisations) | Standardised incidence ratio | Coupling of a dispersion model (ADMS4), a meteorological model and kriging to assess the SO2 levels |
Viel et al., 2011 [ |
France | 13 municipal solid waste incinerators | Non-Hodgkin's lymphomas (registry) | Standardised incidence ratio | Dispersion model (Atmospheric Dispersion Model System version 3—ADMS 3) for each category of pollutants (dioxins, metals, and dusts) |
Perceived exposure areas (criteria not | |||||
Bhopal et al., 1994 [ |
United Kingdom | Coke ovens (66 from 1980) | Cancer (registry) | Standardised incidence ratio | specified), modeled exposure (model not specified) 24-hour mean daily measures of SO2 and smoke over 56 months (1987–91) |
Wilkinson et al., 1999 [ |
United Kingdom | 11 oil refineries | Lymphohaematopoietic malignancy | Standardised incidence ratio | Distance (0–2 km, 0–7.5 km, and eight bands around refinery perimeters) |
Axelsson et al., 2010 [ |
Sweden | Industrial complex including a large cracker producing ethylene and propene | Leukemia, lymphoma, cancers of the lung, liver, and central nervous system, all cancers taken together (registry) | Standardised incidence ratio | Models (unspecified) of ethylene levels |
Eitan et al., 2010 [ |
Israel | Petroleum refineries, oil-fired power plant, and several large petrochemical, chemical, and agrochemical industries | Lung cancer, bladder cancer, and non-Hodgkin's lymphoma | Standardised incidence ratio | Spatial interpolation of SO2 and PM10 routine monitoring data |
Schechter et al., 1989 [ |
Canada | Two natural gas refineries | Cancer (registry) | Standardised incidence ratio | Unclear |
Monge-Corella et al., 2008 [ |
Spain | 18 EPER-registered paper, pulp, and board industries | Lung cancer (mortality) | Standardised incidence ratio | Distance (≤5 km from a paper, pulp, and board industry, ≤5 km from any other industrial installation, towns having no EPER-registered industry within 5 km of their municipal centroid (reference level)) |
Pless-Mulloli et al., 1998 [ |
United Kingdom | Teeside | Lung cancer (mortality) | Standardised mortality ratio | Distance (0.1–2.7 km, 1.5–4 km, and farther) |
García-Pérez et al., 2010 [ |
Spain | 118 integrated pollution prevention and control (IPPC) category 2 metal production and processing installations which report their emissions to the EPER | Leukemia (mortality) | Standardised mortality ratio | See Monge-Corella |
García-Pérez et al., 2009 [ |
Spain | 57 combustion installations which report their emissions to the EPER | Lung, larynx, and bladder cancer (mortality) | Standardised mortality ratio | See Monge-Corella |
García-Pérez et al., 2010 [ |
Spain | 118 integrated pollution prevention and control (IPPC) category 2 metal production and processing installations that reported their releases to air and water in 2001 | Tumours of the digestive system (mortality) | Standardised mortality ratio | See Monge-Corella |
Ramis et al., 2009 [ |
Spain | 452 industries reporting releases to air to the EPER, grouped by industrial sector | Non-Hodgkin's lymphomas (mortality) | Standardized mortality ratio | Distance (1, 1.5, and 2 km). |
Cambra et al., 2011 [ |
Spain | 284 industries declaring to the EPER emissions of pollutants | Lung cancer (mortality), haematological tumours (mortality) | Standardised mortality ratio | Distance (<2 km, >2 km) |
Michelozzi et al., 1998 [ |
Italy | A large waste disposal site (one of the largest in Europe), a waste incinerator, and a petrochemical refinery | All cancers, laryngeal cancer, lung cancer, liver cancer, kidney cancer, and lymphatic and haematopoietic cancers (mortality) | Standardised mortality ratio | Distance (3, 8, 10 km, 10 concentric circles with a radius increasing from 1 to 10 km to define nine bands) |
Pekkanen et al., 1995 [ |
Finland | Refinery | Leukemia, hematological cancers, all cancers (registries) | Standardised mortality ratio | Distance (4,4–7.9, 8–11.9, 12–15.9, and >16 km) |
Sans et al., 1995 [ |
United Kingdom | Petrochemical processing: alcohols, styrene, olefins, benzene, vinyl chloride monomer, and polyvinyl chloride (PVC) | Cancer incidence and mortality for all cancers, leukaemias, and cancer of the larynx | Standardised mortality ratio | Distance (0–3 km, 7–5 km, and eight bands between circles of radii 0.5, 1–0, 2–0, 3–0, 4–6, 5–7, 6-7, and 7–5 km) |
Yang et al., 2000 [ |
Taiwan | Kaohsiung oil refinery | Lung cancer (mortality) | Standardised mortality ratio | Distance |
Pan et al., 1994 [ |
Taiwan | Kaohsiung oil refinery | Cancer in children (mortality) | Standardised mortality ratio | Distance |
Tsai et al., 2009 [ |
Taiwan | Petrochemical industries | Bladder cancer (mortality) | Standardised mortality ratio | In each district, the number of employees of the industries divided by the population, in three clases |
The reasons for doing an epidemiological study on cancer near a major industrial area were frequently concerns from the population, explicitly quoted by 7 studies [
By contrast, multicenter studies refer to the literature and possible etiology in relation to the emissions to justify their choices [
Study areas vary from very rural areas with about 2,000 inhabitants [
Among the multicenter studies, industrial sites of different natures were involved in a study in Italy [
European registries of polluting industries were extensively used in Spain [
Most of the studies (20/27) used a geographical ecological design, based on standardized mortality or morbidity ratios, searching for a possible overincidence of the mortality or the morbidity. Poisson regression and similar statistical designs were used to assess a relationship between health indicators and exposure, taking into account confounding factors (mostly socioeconomic) (Table
Seven studies were case-control studies [
Lung cancer was the most commonly studied [
The latency of cancer was usually taken into account as the number of years of residence in the area before deaths. It varied from at least 1 year (e.g., [
Distance was used as the method to assess the exposure in 19 of the studies. The use of distance is seen as a way to overcome the lack of measurement data, but also to reduce the latency problem, as clearly stated by Pless-Mulloli et al.:
Several options were used for the distance (Table exposed group (“near”) ≤ 5 km from a metal production plant, intermediate ≤ 5 km from any industrial installation other than metal production and processing, unexposed group (“far”), consisting of towns having no EPER-registered industry within 5 km of their municipal centroid (reference level) [ distance: 0–2 km, 0–7.5 km, and eight bands around refinery perimeters with outer limits at 0.5, 1, 2, 3, 4.5, 5.6, 6.6, and 7.5 km [ three concentric circles with radii of 3, 8, and 10 km for descriptive purposes and 10 concentric circles with a radius increasing from 1 to 10 km to define nine bands [
Additional refinement may be added, taking into account, for instance, the residential history [
Another example of a complex exposure assessment initially relying on distance is given by Yu et al.: to account for the effects from monthly prevailing wind, they defined exposure wedges for each month by the monthly prevailing wind direction. Only addresses located within the exposure wedges were considered exposed during the particular month, and the exposure opportunity scores for these residences were assigned by the inverse of distance to the relevant petrochemical complexes [
Although reference sites are usually defined as the farthest to the plant, some studies include a further subclassification taking into account proximity to traffic, urban, semiurban, and rural areas. The definition of these areas may vary between studies. For instance, the industrial area can be defined based on the distance between the subject’s residence and an industrial installation (industrial distance), as the area defined by the first decile of industrial distance [
Models were used by only 5 studies. The Industrial Source Complex Model in long-term model was used by Zambon et al. [
The lack of emission data is a key limitation to modeling, acknowledged by some authors [
Measures alone were used by one study only, taking advantage of a relatively dense air quality monitoring network for SO2 and PM10 [
Studies on morbidity are detailed in Table
Studies investigating morbidity.
Reference | Country | Industrial background | Health outcome | Epidemiological design | Exposure assessment |
---|---|---|---|---|---|
Fung et al., 2007 [ |
Canada | Sarnia “Chemical Valley” | All hospital admissions, admissions with a primary diagnosis of respiratory diseases and cardiovascular diseases | Standardized admissions ratio | Comparison of three cities, annual averages of SO2, NO2, and O3 |
Pascal et al., 2011 [ |
France | Oil refining, oil storage, petrochemical and organic chemical activities, chlorine chemistry, steel and metal working, chemical plants, waste incineration plant, port | Hospitalisations for cardiovascular and respiratory diseases | Poisson regression models | Coupling of a dispersion model (ADMS4), a meteorological model and kriging to assess the SO2 levels |
Kosatsky et al., 2004 [ |
Canada | industrial area in Montreal | Hospitalisations for cardiovascular and respiratory diseases | Standardised admissions rates | O3, |
Bhopal et al., 1994 [ |
United Kingdom | Coke ovens (66 from 1980) | GPs activity: data on consultations, chronic conditions, hospital admissions, and current drug treatments. Lung function, Self-reported respiratory, and nonrespiratory health including asthma | Age and sex standardised rates and ratios, questionnaires (6399 adults, 1888 children) time series | Perceived exposure areas (criteria not specified), modeled exposure (model not specified) 24-hour mean daily measures of SO2 and smoke over 56 months (1987–91) |
Aylin et al., 2001 [ |
United Kingdom | Coke works | Hospital admissions for respiratory and cardiovascular diseases | Standardised admissions rates | Distance (7.5 km) |
| |||||
Patel et al., 2008 [ |
India | Vapi industrial area, dyes, chemical plants | Respiratory health, lung function | Questionnaires (2, 573 women) | Distance (<2 km, 2-3 km, 3-4 km, and farther) |
De Marco et al., 2010 [ |
Italy | Largest chipboard industrial park | Respiratory and skin diseases | Questionnaires (ISAAC (1998), ECRHS (2002), SIDRIA, MM040NA and MM080 standardized questionnaires, 3854 children) | Distance (no wood factories <2 km from home and school (“unexposed” group) at least 1 low emission factory (but no chipboard industries) <2 km from home or school (group “at low exposure”), at least 1 chipboard industry <2 km from home or school (group “at high exposure”) |
Dubnov et al., 2007 |
Israel | Major coal-fired power station | Health status, pulmonary function tests (PFT), forced vital capacity (FVC) and forced expiratory volume during the first second (FEV1) | Questionnaires (ATS and National Heart and Lung Institute) (1492 children) |
|
Ginns and Gatrell, 1996 [ |
United Kingdom | Cement works | Respiratory health | Questionnaire (362 children) | Distance (near the industry versus area 9 to 19 km away) |
Halliday et al., 1993 |
Australia | Power stations | Asthma, general symptoms, measurement of lung function, bronchial reactivity, and skin test atopy was | Questionnaire (851 children) | Distance (near the industry versus area 40 km away) |
Peled et al., 2005 [ |
Israel | 2 power plants | Health status, lung function (peak expiratory flow) | Nested cohort study (285 children), questionnaire based on the American Thoracic Society’s (ATS) ATS-DLD-78 | PM10 and PM2.5 daily measurements at 6 stations |
Pignato et al., 2004 |
Italy | Petrochemical industries and oil refineries | Self-reported asthma, asthma-like symptoms, and allergic rhinitis | Questionnaires (1180 children) | Annual mean NO2 measurements |
Rusconi et al., 2011 [ |
Italy | Biggest high complexity refinery in the Mediterranean Sea and largest European liquid fuel gasification plant | Asthma, respiratory symptoms in children, FENO, and lung function measurements | Questionnaires (ISAAC) | Measurement of weekly concentrations of SO2, benzene, NO2, O3 |
Stenlund et al., 2009 [ |
Sweden | Steel industry | Self-reported health symptoms bronchitis- and asthma-like, and neurasthenic | Interventional, population-based questionnaire study (684 adults) | distance (two areas relatively close and relatively distant) |
De Moraes et al., 2010 [ |
Brazil | Petrochemical complex | Wheezing | Questionnaires (ISAAC) (209 children) | Cities in a 5-kilometer radius, communities established downwind of the petrochemical complex and thus, under greater influence of its dispersion plume (A, B, C), were classified as “exposed communities” (ECs) Those upwind of the plant and thus less exposed to its dispersion plume (D, E) were used as reference communities (RCs) |
Jadsri et al., 2006 [ |
Thailand | 50 chemical industries | Respiratory diseases | Spatial regression analysis | Dispersion of SO2, |
Câra et al., 2007 [ |
Romania | Iron, steel, and coke factory | Wheezing | Comparison of two periods before and after the closure of the factory (GPs information for 874 children) | Distance (near the industry and 10 km away) |
Pless-Mulloli et al., 2000 [ |
United Kingdom | Opencast coal mining sites | Respiratory illnesses | Questionnaires (3216 children) and GPs records (2442 records) | Distance (5 cities near industries and 5 referent cities further away) |
Smargiassi et al., 2009 [ |
Canada | Refinery | Emergency visits and hospital admissions for asthma in children | time stratified case-crossover | Distance (0.5–7.5 km) and daily SO2 measurements, at-home estimates of daily exposure based on a dispersion model (AERMOD) |
Howel et al., 2001 [ |
United Kingdom | Opencast coal mines | Respiratory health | GP data, respiratory events (2442) | Distance, PM10 measurements |
White et al., 2009 [ |
South Africa | Petrochemical refinery | Respiratory health | Questionnaire (ISAAC) (2361 children) | Distance, wind direction, and speed |
Wichmann et al., 2009 [ |
Argentina | Petrochemical industries | Respiratory health, lung function (standard spirometry) | Questionnaires (1191 children) | Distance, near petrochemical industries, near heavy roads, and 2 relatively nonpolluted areas, PM and VOCs measurements |
Yogev-Baggio et al., 2010[ |
Israel | Coal-fired power plant | Respiratory health, lung function (forced expiratory volume) | Questionnaires (1181 children) |
|
Aungudornpukdee et al., 2010 [ |
Thailand | 15 chemical industries | short-term memory dysfunction | Weschsler intelligence scale for children, questionnaires (2955 children) | Distance to major air pollution sources (industries, roads, etc.) |
Atari et al., 2009 [ |
Canada | Sarnia “Chemical Valley” | General health status, odour annoyance | Telephone interviews (804) | Land use regression (LUR) modeling based on SO2 and NO2 measurements |
Concern was a major motivation quoted by 12 studies [
A majority of the studies focused on the respiratory health of children (17 studies), using questionnaires specifically defined for the study or standardized questionnaires such as the ISAAC questionnaire from the International Study of Asthma and Allergies in Childhood [
Two studies were intervention studies. Câra et al. compared GPs information on the respiratory health of 874 children for two periods: when the industry was operating and after its closure [
Five studies used an ecological approach to study standard rates ratio based on hospital admissions or disease incidence. Two studies quantified the relationship between symptoms and measurements through a time-series analysis [
Participants of the cross-sectional surveys were selected based on their city of residence (or school), and distance was again the preferred method to define the exposed versus nonexposed cities. In most studies, a finer exposure assessment was performed for the participants, based on information collected through the questionnaires, modeling, or measurements. When measurements were available, they were not always used to assess exposure. For instance, Moraes et al. mentioned that concentrations were available for several pollutants (PM, NOx; SO2, O3, benzene, toluene, and xylenes) but used them for descriptive purposes only (in comparison to the World Health Organization air quality standards) [
White et al. reported that they did not have the budget for a model and that concentration and emissions data were missing. Therefore, they add that they rely on a meteorologically estimated exposure index based on wind direction and speed [
Fung et al. selected the participating cities based on the annual averages of SO2, NO2, and O3 and mentioned that the reference area
Pless-Mulloli et al. proposed two indicators to characterize the long-term versus short-term exposure: short-term exposure was assessed through PM10 measurements, and long-term exposure was defined as living near an active site [
One study compared the associations between emergency department visits and SO2 concentrations obtained from fixed monitors and from an air dispersion modeling and found some differences increasing with the distance [
Studies on mortality are detailed in Table
Studies investigating mortality.
Reference | Country | Industrial background | Health outcome | Epidemiological design | Exposure assessment |
---|---|---|---|---|---|
Hodgson et al., 2007 [ |
United Kingdom | Runcorn: chlor alkali plant, power stations | Mortality from renal diseases | Standardised mortality ratio | Dispersion of mercury (ADMS) |
Hodgson et al., 2004 [ |
United Kingdom | Runcorn: chlor alkali plant, power stations | Mortality, hospital admissions for kidney diseases | Standardised mortality ratio, standardized admissions rate | Distance |
Dolk et al., 1999 [ |
United Kingdom | Coke work | Mortality for cardiovascular and respiratory causes | Standardised mortality ratio | Distance (2 km, 7.5 km, bands of 0.5, 1, 2, 3, 4.6, 5.7, 6.7, and 7.5 km). |
Triolo et al., 2008 [ |
Italy | Industrial settlement | Mortality (all causes, cancers, cardiovascular, respiratory, diabetes, injuries, etc.) | Standardised mortality ratio | Distance: 3 concentric zones of 5 km around the industries, dispersion model (CMPM98) for SO2, O3, and SO2 measurements |
Cambra et al., 2011 [ |
Spain | 284 industries declaring the EPER emissions of pollutants | Mortality all causes, ischaemic heart disease, cerebrovascular diseases, chronic lower respiratory tract diseases | Standardised mortality ratio | Distance (<2 km, >2 km). |
Sarov et al., 2008 [ |
Israel | 17 plants: chemical, pharmacochemical, and heavy industry | Perinatal mortality | Standardised mortality ratio | Distance up to 20 km based on odors complaints |
Bhopal et al., 1994 [ |
United Kingdom | Coke ovens (66 from 1980) | Mortality | Age and sex standardised rates and ratios, Questionnaires (6399 adults, 1888 children) Time series | Perceived exposure areas (criteria not specified), modeled exposure (model not specified) 24 hour mean daily measures of SO2 and smoke over 56 months (1987–91) |
Studies are summarized in Table
Studies investigating birth outcome.
Reference | Country | Industries | Health outcome | Method | Exposure assessment |
---|---|---|---|---|---|
Bhopal et al., 1994 [ |
United Kingdom | Teeside | Sex ratio, birthweights, and stillbirths | Sex ratio | Perceived exposure areas (criteria not specified), modeled exposure (model not specified) |
Bentov et al., 2006 [ |
Israel | 17 chemical facilities | Major congenital malformations of the central nervous system | Standardized incidence ratio | Distance (exposed < 20 km), wind direction |
Brender et al., 2006 [ |
United States | 113 industries in the Texas National Priority Listing (NPL) sites | Oral clefts | Logistic regression | Distances (proximity ≤ 1 mile) |
Brender et al., 2008 [ |
United States | 113 industries in the Texas National Priority Listing (NPL) sites | Chromosomal anomalies | Case control (2099 cases, 4368 controls) | Distances (proximity ≤ 1 mile) |
Yang et al., 2000 [ |
Taiwan | Kaohsiung oil refineries | Sex ratios | Standardized sex ratio | Distance (all municipalities in the area) |
Yang et al., 2002 [ |
Taiwan | Kaohsiung oil refineries | Preterm delivery | Logistic regression model | Distance (at least 50% population or 50% area falling within a distance of 3 km from any one of the three complexes) |
Yang et al., 2004 [ |
Taiwan | Kaohsiung oil refineries | Preterm delivery | Logistic regression model | Distance (at least 50% population or 50% area falling within a distance of 3 km from any one of the three complexes) |
The health outcomes and the study design were various. Exposure assessment was poorly described compared to papers dealing with cancer or morbidity. Distance was the method used by all the studies but one [
Three studies investigated mental health, psychological distress [
The local background and concerns of the population were not the main motivation in the two studies in the United States based on industrial registries [
Two studies investigated the psychological distress of the population in relation to their proximity to industries registered in the Toxic Release Inventory through questionnaires. In these studies, the main assumption is not that an over-exposure to air pollutants can create adverse psychological effects, but that
Nine biomonitoring studies were reviewed. In none, even the one based in Teesside [
Biomonitoring studies.
Reference | Country | Industry | Biomarkers |
|
---|---|---|---|---|
Barregard et al., 2006 [ |
Italy and Sweden | Chlor alkali plants | Urinary mercury | 193 |
Rusconi et al., 2011 [ |
Italy | Biggest high complexity refinery in the Mediterranean Sea and largest European liquid fuel gasification plant | MDA-dG adducts | 54 |
Choi et al., 2000 [ |
Korea | Large-scale petrochemical industrial complex | Benzene in blood, metabolites of benzene in urine | 115 |
Pless-Mulloli et al., 2005 [ |
United Kingdom | Teesside | Polychlorinated dibenzo-p-dioxins, furans, and polychlorinated biphenyls in blood | 40 |
Thomas et al., 2009 [ |
United Kingdom | Large smelter lead/zinc smelter | Cadmium in urine | 180 |
Sala et al., 1999 [ |
Spain | Organochlorine compound factory | Organochloring in blood | 608 |
Stroh et al., 2009 [ |
Sweden | Lead smelters | Lead in blood | 3879 |
Williamson et al., 2006 [ |
United States | Six superfund sites | Serum Immunoglobulins | 3916 |
Thomas et al., 2009 [ |
United Kingdom | Large smelter lead/zinc smelter | Cadmium in urine | 180 |
Studies investigating mental health.
Reference | Country | Industries | Health Outcome | Method | Exposure assessment |
---|---|---|---|---|---|
Bush et al., 2001 [ |
United Kingdom | Teeside | Stigma | 5000 questionnaires + 41 interviews | Distance (three areas at 1.5, 7, and 8 km) |
Downey and Van Willigen, 2005 [ |
United States | Industries in the Toxic Release Inventory | Psychological distress (depression), perceived disorders | 1210 questionnaires | Distance, visual exposure |
Boardman et al., 2008 [ |
United States | Industries in the Toxic Release Inventory | psychological distress (K6 scale) | 1139 questionnaires | Distance, visual exposure |
Discussing the results of the studies was not the objective of this literature review. However, it was interesting to note that when studying cancer, very few results were statistically significant, although several studies concluded on a gradient of risk following the exposure gradient [
Morbidity, and especially less severe outcomes such as respiratory symptoms, eyes symptoms or consultations to the general practitioners tended to increase with exposure [
In the studies of declared health, complaints about odors or dust were correlated with the discomfort, in some cases positively [
Epidemiological studies investigating the impacts of air pollution produced by major industrial sources are scarce, as only 77 papers were found in this review. They correspond to a wide range of industrial activities. However, our search is likely to be incomplete, and the limits of this search are probably the largest on the biomonitoring studies and the mental health studies, as we did not included these as explicit key words in the search.
However, given that the papers we included in the review were written by different teams, in different areas and at different periods, we are still confident that it can give a good overview of the practices in the field. Yet, it has to be noted that several papers were produced by the same team and/or part larger initiatives on industrial pollution, which may limit the diversity of practices reported. We also included two reports from the grey literature in the review [
Several reasons may explain the low number of publications; few epidemiological studies may be performed because of the complexity of collecting health and exposure data or because quantitative risk assessment is extensively used to study industrial pollution. There may also be a publication bias, with studies showing no link between exposure and health not being published.
In many of the cases, the studies are justified by a concern from the population; that is, epidemiology is used to test the hypothesis made by the population that the industries impair their health. It is also used to investigate areas where an overincidence of a health outcome had been previously observed. There are few initiatives to identify the health effects of a given industry independently of the local context, and these initiatives are mostly multicenter studies based on industrial registries indeed, whatever the topic (cancer, mental health, etc).
In summary, the multicenter studies based on industrial registries are not taking into account the local context to select the areas under investigation, while mostly all others studies do. Therefore, there is likely to be a bias in site selection where to perform epidemiological studies, based on the existence of a local social mobilization. It would be interesting to understand why in some areas industries raised high concerns and lead to epidemiological studies, while in others there is such social mobilization, and if these reasons may result in biases in the result of the studies. On the other hand, it is essential to answer the population concerns, and, as stated by Ginns et al.,
Multicentric design is believed to be a solution to the local biases, as the influence of the confounding factors may decrease as the number of sites increases [
Independently of the health outcome and the statistical design used, the lack of information on the environmental and industrial background of the sites is striking in many papers. A major issue is raised by the exposure assessment. As industrial sites emit a complex mixture of pollutants, with plumes varying in composition and over time and space, epidemiologists have to rely on measurements and modeling of a subset of pollutants to assess an integrated exposure. Modeling is seen as the most efficient tool to avoid exposure misclassification. In Teesside, environmental data, land-use data, historical data, and data on the perception of air pollution and odors were analyzed to check that the distance to the site was an interesting proxy. Globally, measurements did not show large differences between exposed and nonexposed areas, but the dispersion models confirmed a gradient of pollution with distance [
This lack of environmental data is a major obstacle. It is striking to see that in many areas the population is highly concerned by the environmental pollution and its consequences, and that these concerns are answered through complex epidemiological studies, relying on poor environmental data. In short, there is a discrepancy between the expectancies of the population, the investment in collecting and analyzing health data, and the poor accessibility to key emissions and concentrations data.
When distance is the only possible choice, Hodgson et al. advised to integrate knowledge of the factors that drive exposure, for example relative emissions, and wind direction [
The bias in exposure assessment and the ecological bias are likely to limit the possibility of ecological studies to reveal low relative risks with statistically significant results, especially when studying cancer with a latency of several decades. Leukemia may be the only cancer for which the latency is a priori short enough to allow a good reconstruction of exposure based on present data.
A combination of multicentric studies and local studies could be efficient ways to increase knowledge on the health effects of industrial areas and answer the concerns from the population. As stated below, multicenter studies would limit local biases, and sites would not be selected based on an a priori population concern or over incidence. However, criteria to decide that sites are similar enough to be included in a multicenter study need to be defined. A focus on sites where the population requests more information could then be performed, with the support of social scientists.
These studies could be performed on several health issues and with several designs. An investigation of the mental health impacts would be highly relevant, as this issue seems to have been poorly taken into account by epidemiologists so far.
For the multicenter and the local studies, a better characterization of exposure would be an asset to improve our capacity to investigate the impacts of industrial pollution. It requires improving the availability of emission data and of monitoring data.
Finally, intervention studies documenting the possible improvements of the health status of the population after the closure of a plant, or a change in the industrial processes, would be highly informative to improve the knowledge and to help for management (a change in the industrial processes that have been shown to have positive effect in the environment and the health status could be reproduced elsewhere).