OUP user menu

Social inequalities resulting from health risks related to ambient air quality—A European review

Séverine Deguen , Denis Zmirou-Navier
DOI: http://dx.doi.org/10.1093/eurpub/ckp220 27-35 First published online: 16 January 2010

Abstract

Background: Environmental nuisances, including ambient air pollution, are thought to contribute to social inequalities in health. There are two major mechanisms, which may act independently or synergistically, through which air pollution may play this role. Disadvantaged groups are recognized as being more often exposed to air pollution (differential exposure) and may also be more susceptible to the resultant health effects (differential susceptibility). Method: European research articles were obtained through a literature search in the Medline database using keywords ‘Socioeconomic Factors, Air Pollution, Health’ and synonymous expressions. Results: Some studies found that poorer people were more exposed to air pollution whereas the reverse was observed in other papers. A general pattern, however, is that, irrespective of exposure, subjects of low socio-economic status experience greater health effects of air pollution. So far as we are aware, no European study has explored this relationship among children. Conclusion: The housing market biases land use decisions and may explain why some subgroups suffer from both a low socio-economic status and high exposure to air pollution. Some data may be based on inaccurate exposure assessment. Cumulative exposures should be taken into account to explore health problems more accurately. The issue of exposure and health inequalities in relation to ambient air quality is complex and calls for global appraisal. There is no single pattern. Policies aimed at reducing the root causes of these inequalities could be based on urban multipolarity and diversity, two attributes that require long-term urban planning.

  • air pollution
  • environmental inequalities
  • health inequalities
  • social determinants

Introduction

There is now clear evidence of social inequalities in health in most industrialized countries:1 in general, socio-economically disadvantaged people are more strongly affected by various health problems2–4 than more affluent ones. Despite numerous factors already identified, some of these inequalities remain unexplained, leading to the hypothesis that environmental nuisances may also contribute to social health inequalities.5,6 Assessing how environmental exposure may partly explain such inequalities is a major subject of public health research.

According to the literature,5,6 there are two major mechanisms that may act independently or together, through which environmental exposure may contribute to social health inequalities. (i) Among the general population, disadvantaged groups are recognized as being more often exposed to sources of pollution (differential exposure), a situation that contradicts the principle of environmental equity, according to which no group of people should bear a disproportionate share of harmful environmental exposure. (ii) The general population may also be more likely to exhibit resultant health effects (differential susceptibility). To investigate this hypothesis, studies explored the assumption that exposure to environmental nuisances might give rise to greater health effects among socioeconomically disadvantaged groups; this issue of greater vulnerability is less well documented.

Many epidemiological studies, mostly in North America and in Europe, have demonstrated that both short- and long-term exposures are associated with several health events. In spite of the improvement of air quality during the recent decades, air pollution remains a major field for investigation and action in view to improving public health in Europe. In this context, this review deals with European studies that concern two issues: whether subjects or populations of poor socio-economical status (SES) live in areas with lower ambient air quality than richer ones; and whether the association between ambient air pollution and health is influenced by the SES assessed at an individual or ecological level.

Methods

European research articles were obtained through a literature search in the Medline database of the National Library of Medicine. Only articles written in English or in French were selected, up to the end of April 2009.

Three principal MeSH-terms were used for the literature search queries: ‘Europe AND socioeconomic factors AND air pollution’. Numerous synonymous expressions of these two keywords were also used, such as ‘social class, unemployment, income’ for socio-economic factors and ‘ozone, nitrogen dioxide, sulphur dioxide, carbon monoxide, particulate matter’ for air pollution. We have also included more general expressions, environmental justice and environmental inequity dealing with the socio-environmental disparities. Were excluded papers investigating only indoor air pollution and occupational or exposure to environmental tobacco smoke. Were also excluded papers in which air pollution exposure was measured using a proxy-indicator such as distance to high traffic roads or to industrial plants, and papers where no result was presented on either socio-economically based ‘differential exposure or differential susceptibility’.

Concerning the assessment of differences in response to exposure according to SES, were also excluded all papers which did not formally test this effect modification, either by a stratified analysis or through the introduction of an interaction term in some regression model. Studies where the SES was merely considered as a confounder were thus discarded.

The results section is structured according to the two mechanisms through which environmental exposure may contribute to social health inequalities, namely differential exposure and differential susceptibility. Papers are sorted according to the country where the study was conducted.

Results

A total of 129 papers assessed inequalities in exposure in Europe according to some measure of socio-economic status, and 23 explored the modification of the relation between air pollution and some health event, often mortality, by the socio-economic status. They are described in tables 1 and 2 that provide information on the study design, how exposure and SES were assessed and key results. Additional information is given in table 2 on the health events and the methods used to assess effect modification.

View this table:
Table 1

European Studies investigating the environmental inequalities regarding air pollution exposure

AuthorsPopulation/countryStudy designaAir pollution variablesbGeographical level and SES variablesMain results
Brainard et al.8Birmingham, EnglandGeographicalAnnual average hourly CO and annual (hourly) NO2At a enumeration district scale (medium population of 496 residents): ethnicity, male unemployment, households without a car, homeowners, pensioners, social class, deprivation index (carstairs, Jarman and Townsend)The average CO and NO2 emissions for districts with deprived populations are higher than in affluent ones: 2331 vs. 2112 µg/m3 and 23.71 vs. 22.29 µg/m3, respectively. The averages of these pollutants were also higher among districts with high proportion of blacks than among more white districts: 2919 vs. 2276 µg/m3 for CO and 27.09 vs. 23.32 µg/m3 for NO2.
Briggs et al.13EnglandGeographicalAnnual average of NO2, PM10, O3 and SO2Three geographical levels of analysis: super output areas (SOAs, an average of 1500 persons), wards (aggregations of SOAs, an average of 6200 persons) and districts (an average of 139,000 persons). Several indicators of deprivation: index of multiple deprivation: income, employment, education and access to housing and servicesPositive correlations (varying around 0.3 and 0.2 at SOA and ward geographical scale) are found with all the air pollutants (except O3): a high level of air pollution was associated with a high level of deprivation (inverse relation for O3). Variation of the association strength was observed according to the geographical scale
Chaix et al.19Children aged 7–15 years, Malmo, Sweden (2001)MultilevelAnnual average of NO2 estimated for the points of the 100 metre grid that were the closest to the building of residence and school of attendanceAnnual mean of income of subjects aged ≥25 years in each residential building where children in the study lived in 2001 and in each neighbourhood of residence. The median number of people aged 25 years or older in buildings of residence was 2 and it was 1484 in neighbourhoods of residence.Children from low SES neighbourhoods were more exposed to NO2, both at their residence place (21.8 vs. 13.5 µg/m3 for the lowest and the highest income classes, respectively) and at school (19.7 vs. 13.7 µg/m3).
Forastiere et al.22Only residents of Rome aged 35 years and older (1998–2001)GeographicalPM, CO, NOx, BenzeneEstimation at census block scale (480 inhabitants on average) of a median per capita income index and a socio-economic index (SES, including educational level, occupational categories, working-age unemployment rate, family size, crowding and proportion of dwellings rented/owned)Concentrations increase with the average block income level for all traffic pollutants (PM: 16.7 vs. 21.7 µg/m3, for the low- and high income categories, respectively; CO: 10.4 vs. 24.3 µg/m3; NOx: 10.4 vs. 26.7 µg/m3; Benzene: 10.7 vs. 25.2 µg/m3). Environmental inequalities are stronger using the SES index (PM: 9.2 vs. 39.6 µg/m3, CO: 6.8 vs. 45.3 µg/m3, NOx: 11.2 vs. 41.6 µg/m3, Benzene: 7.5 vs. 46.2 µg/m3).
Havard et al.24Strasbourg, FranceGeographicalAnnual average of NO2At a French census block scale (2000 inhabitants in average): socio-economic index (including 19 socio-economic and demographic variables)There was an association between deprivation index and NO2 levels: the mid-level deprivation areas were the most exposed (39.6 µg/m3) whereas the most affluent areas were the least (30.6 µg/m3). Same relations were observed with SO2 and PM10, but inverse relationship with O3.
Kruize et al.20Rijnmond Region, NetherlandsSemi-IndividualAnnual average of modelled NO2 concentrations (25 × 25 m grid)IncomeThere is a significant association between income and NO2 level: the mean of NO2 are 37.7 and 38.2 µg/m3 for the higher and lower income categories, respectively.
McLeod et al.7England and WalesGeographicalNOx, PM10, SO2At local authority district scale and/or regional scale: social class index, population density and percentage of ethnic minorities.The higher social classes are more likely to be exposed to greater air pollution, whatever the pollutant, the socio-economic indicator and the model that was implemented.
Mitchell et al.9Leeds, UKGeographicalAnnual mean of NO2At a 200 m × 200 m cell level (3600 points spaced by 200 m intervals in a grid cell pattern throughout the 144 km2 inner box): Townsend deprivation indexA clear association between deprivation and NO2 level: in 2005, the mean of NO2 is around 18 µg/m3 for the most affluent areas vs. 22 µg/m3 for the least ones.
Namdeo et al.11Leeds, UKGeographicalAnnual mean of NO2At the Census Output Area level: cumulative deprivation indexDeprived population groups are disproportionately exposed to higher NO2 level as compared with the affluent group: a scenario gives for example, 20.5 µg/m3 vs. 19.2 µg/m3, respectively.
Naess et al.14General population aged 50–74 years residing in Oslo, Norway on 1 January 1992MultilevelAverage monthly concentrations of PM2.5 during period 1992–95Social deprivation at both individual and administrative neighbourhood levels: education, household income, occupational class, ownership status of dwelling, type of dwelling and crowded householdsThere is a gradual increase of PM2.5 when the proportion of subjects living in a flat increases across neighbourhoods (mean value of PM2.5 ranging from 12.1µg/m3 in the lowest category to 17.0 µg/m3 in the highest).
Rotko et al.16Population aged 25–55 years, Helsinki (Finland)Individual48-h exposure of NO2Occupational status, education level and employment statusThere is an association between personal exposure to NO2 and education level: less educated subjects have higher exposures than educated ones (mean of NO2 equal to 26.3 and 24.4 µg/m3, respectively). The same association is seen according to the employment status among men
Rotko et al.17Population aged 25–55 years, Helsinki (Finland)Individual48-h exposure of PM2.5Occupational status, education level and employment statusThere is an association between personal exposure to PM2.5 and education level: less educated subjects have higher exposures than educated ones (mean of PM2.5 equal to 18.98 and 13.41 µg/m3, respectively). There is also an association between PM2.5 and occupational status, with low exposures for white-collar employees compared to other categories (mean PM2.5 levels are 11.97 and 20.46 µg/m3, respectively). Stratification analysis by gender demonstrates that associations persist among men but not among women. For men, unemployment dramatically increases PM2.5 exposure (41.8 vs. 15.5 µg/m3).
Stroh et al.18Scania, SwedenAnnual average NO2 modelled with a 250 × 250 m grid resolutionIndividual data: country of birth, education levelStrength and direction of the association between NO2 and social categories varies within cities. In Malmö, subjects born in Sweden tend to live in areas with lower concentrations of NO2 than those born in other countries. Inverse conclusions are drawn in other cities. The association between NO2 and education ended show the same discrepancy between Malmö and the four other cities.
Schikowski et al.21Women aged 55 years at time of investigation, Ruhr, GermanySemi-IndividualPM10, NO2 and TSPEducation levelWomen with <10 years of school education are more exposed to PM10 than those with a higher education level. No association has been found with NO2.
Tonne et al.12London, EnglandGeographicalAnnual average NO2 and PM10At census ward scale: index of multiple deprivationThe mean of PM10 and NO2 increases from the less deprived neighbourhoods (C1, class 1) to the most ones (C5, Class 5): the mean for C1 and C5 are 38.1 and 46.7 µg/m3for NO2 and 25.7 and 27.5 µg/m3 for PM10, respectively.
Wheeler et al.10General population aged 16–79 years, EnglandSemi-individual (household)Index of air pollution combining annual average of NO2, PM10, NO2 and Benzene estimated at a ward geographical level. The air pollution index of each participant is equal to the level of their residential wardSocial class of head of householdEnvironmental inequity is observed among urban households: air quality is poorer among households of low social class. There is a suggestion of inverse relationship for rural and semi-rural households.
  • a: CO, carbon monoxide; NO2, nitrogen dioxide; O3, ozone; PM, particulate matter; PM10, particulate matter with an aerodynamic diameter of up to 10 mm; PM2.5, particulate matter with an aerodynamic diameter of up to 2.5 mm; SO2, sulphur dioxide; TSP, total suspended particulates.

  • b: Geographical: socio-economic status and air pollution exposure were both estimated at a same geographical level; semi-individual: socio-economic status and air pollution exposure were estimated at a individual and geographical level, respectively; individual: socio-economic status and air pollution exposure were both estimated at a individual level; multilevel: socio-economic status was estimated at both individual and geographical level whereas the air pollution exposure was estimated at geographical level.

View this table:
Table 2

European Studies assessing the potential modification effect by the socio-economic status on the relation health and air pollution exposure

AuthorsPopulation/countryHealth variablesAir pollution variablesaGeographical level and SES variablesMethods to evaluate effect modificationMain results
Filleul et al.26Residents of Bordeaux (France), population older than 65 years (1988–97)Non-trauma and cardiorespiratory mortalityDaily mean of BSAt individual level: educational attainment (without primary school diploma, primary school diploma, secondary validated or higher) and previous occupation (never worked, white-collar, blue-collar)Stratified analysis and test for heterogeneityIncrease in mortality for a 10 mg/m3 increment in BS concentrations. Non-trauma mortality: only blue collars show a significant association: OR = 1.41 (1.05–1.90). Cardiorespiratory mortality: association is greater among subjects with high education: OR = 4.36 (1.15–16.54).
Filleul et al.27Residents of Bordeaux (France), population older than 65 years (1988–97)Non-trauma mortalityBS (above 90th percentile or below 10th percentile of observed ambient air concentrations)At individual level: educational level (no school, primary without diploma, primary with diploma) and previous occupation (domestic employees and women at home, blue-collar workers craftsmen and shopkeepers, other employees and intellectual occupations)Stratified analysis and test for heterogeneityNo effect modification according to socio-economic indicators.
Forastiere et al.22Residents of Rome (Italy) aged 35 years and older (1998–2001)MortalityDaily PM10Estimation at census block scale (480 inhabitants on average) of a median per capita income index and a socio-economic index (including educational level, occupational categories, working-age unemployment rate, family size, crowding and proportion of dwellings rented/owned)Interaction term in multivariate modelEffect modification of socio-economic status on the PM10-mortality association: the effect is stronger among people with lower income and SES (1.9 and 1.4% per 10 µg/m3, respectively) compared with those in the upper income and SES levels (0.0 and 0.1% per 10µg/m3, respectively).
Havard et al.28Residents of Strasbourg (France), population aged 35–74 years (2000–03)Myocardial infarction events24-h average PM10 concentrationsAt a French census block scale (2000 inhabitants on average): socio-economic index (including 19 socio-economic and demographic variables)Stratified analysis and test for heterogeneitySignificant influence of neighbourhood SES, with greater effect of PM10 observed among subjects living in the most deprived neighbourhoods (20.5% increase, 95%CI: 2.2–42.0).
Laurent et al.29Residents of Strasbourg (France), general population (2000–05)Astham attacksThe daily air pollution indicator considered for PM10, NO2, and SO2 was the 24-h average concentration. It was the maximum daily value of the 8-h moving average for the O3.At a French census block scale (2000 inhabitants in average): socio-economic index (including 19 socio-economic and demographic variables)Stratified analysis and test for heterogeneitySocio-economic deprivation had no influence on the association between air pollution and asthma attacks, whatever the pollutant.
Laurent et al.31Residents of Strasbourg (France), general population (2000–05)β-agonist sales for asthmaThe daily air pollution indicator considered for PM10, NO2, and SO2 was the 24-h average concentration. It was the maximum daily value of the 8-h moving average for the O3.At a French census block scale (2000 inhabitants on average): socio-economic index (including 19 socio-economic and demographic variables)Stratified analysis and test for heterogeneitySocio-economic deprivation had no influence on the association between air pollution and asthma attacks, whatever the pollutant.
Wojtyniak et al.25Two group of population (i) between 0 and 70 years and (ii) >70 years, residents of Cracow, Lodz, Poznan and Wroclaw (Poland)Non-trauma and cardiovascular mortalityBS, NO2 and SO2 (day of death or preceding day)EducationalStratified analysis and test for heterogeneityNon-trauma mortality: significant effect of BS among the less than secondary education group in both age groups. Significant effect of NO2 in the oldest age group and for those below secondary education only. Significant effect of SO2 in the oldest age group and those with less than a secondary education.Cardiovascular mortality: significant effect of BS only for those with less than a secondary education in both age groups. Significant effect of NO2 for secondary education and above, only in the oldest age group. Significant effect of SO2 only among subjects >70 years with below secondary education level.
  • a: BS, Black Smoke; NO2, nitrogen dioxide; O3, ozone; PM10, particulate matter with an aerodynamic diameter of up to 10 mm; SO2, sulphur dioxide.

Differential exposure

The majority of European studies took place in the UK. In England and Wales, McLeod in 20007 investigated the relationship between PM10, NO2 and SO2, and socio-economic indicators. They found that higher social classes were more likely to be exposed to greater air pollution, whatever the pollutants and the socioeconomic indicators they used. In contrast, Brainard et al.8 found that the level of NO2 and CO in Birmingham was higher in communities with a greater proportion of coloured people and deprived classes. Several years later, in Leeds, Mitchell9 demonstrated social inequality in the distribution of NO2 according to the Townsend index. Comparing the trend of NO2 levels between 1993 and 2005, they demonstrated that the average difference between deprived and affluent communities declined from 10.6 µg/m3 in 1993 to 3.7 µg/m3 in 2005 as a result of city-wide improvements in air quality driven by fleet renewal. Wheeler and Ben-Shlomo,10 also found in 2005 that air quality is poorer among households of low social class. More recently, social inequalities in NO2 levels in Leeds were confirmed by Namdeo and Stringer11 at the detriment of poorer groups. In London, a comparison before and after the introduction of the Congestion Charging Zone showed that, although air pollution inequalities persisted, there was a greater reduction in air pollution in deprived areas than in the most affluent ones.12 Briggs et al.13 concluded that the strength of the association of the deprivation index with air pollution tended to be greater than for other environmental nuisances.

Two studies were conducted in Oslo, Norway. Irrespective of the socio-economic indicators they used, Naess et al.14 showed that the most deprived areas were exposed to higher PM2.5 levels and revealed a clear dose–response relationship between PM2.5 levels and the number of subjects living in flats. In contrast, no association between NO2 levels and education or occupation was found in the cohort of Norwegian men.15

Within the EXPOLIS study, environmental inequalities arising from personal exposure to NO2 and PM2.5 were explored in Helsinki, Finland.16,17 Personal levels of NO2 decreased with a higher level of education. Much greater contrasts in exposure were observed between socio-economic groups for men than for women, both for NO2 and PM2.5. While the occupational status was not correlated with PM2.5 globally, a stratified analysis by gender showed a strong association for men only: the mean PM2.5 exposure was ∼50% lower among white-collar workers than among the other occupational categories.

Two studies conducted in Sweden brought evidence of social inequalities related to NO2. Stroh et al.18 found that the strength and direction of the association between the socio-economic status and NO2 concentrations varied considerably between cities. In another study, children from areas with low neighbourhood socio-economic status were shown more exposed to NO2 both at home and at school.19

We found four other European studies that explored social inequalities related to air pollution. In Rijnmond (The Netherlands), according to Kruize et al.,20 lower income groups live in places with higher levels of NO2 than greater income groups. In a cohort of German women, Schikowski et al.21 revealed the existence of a social gradient with higher PM10 exposures among subjects with <10 years of school education than among those with higher education. Inversely, in Rome, Italy, the higher social class appeared to reside in areas with high traffic emissions; this disparity was even stronger when SES rather than income was considered.22 Using a French deprivation index and a fine census block resolution scale,23 Havard et al.24 found, in Strasbourg, France, that the mid-level deprivation areas were the most exposed to NO2, PM10 and CO.

Differential susceptibility

Few studies have been published on the role of SES in the relationship between air pollution and health in Europe. In Rome,22 social class clearly affected the relationship between PM10 and mortality: the upper social classes were not as affected by the harmful effects of air pollution as those in lower social classes. Since the former live in areas with higher air pollution, the authors interpreted their findings in terms of differential susceptibility. Supporting this hypothesis, they found a higher proportion of chronic diseases among the poor. They also argued that living in an area with a high level of air pollution, mainly in the city centre, did not necessarily result in greater exposure. Wealthier residents of Rome were said to spend less time in their homes than poorer social groups because they were more likely to have second residences outside the city.

In four Polish cities, Wojtyniak et al.25 showed a significant association between exposure to black smoke and either non-trauma or cardiovascular mortality among subjects who had not completed secondary education. Significant associations between SO2 or NO2 and cardiovascular mortality were also present more particularly among subjects aged >70 years with education below secondary school level.

Finally, in France, five studies investigated the impact of the socio-economic level on air pollution effects. In Bordeaux, Filleul et al.26 found a significant association between mortality among people aged >65 years and exposure to black smoke among blue-collar workers only. Also in Bordeaux, however, a cohort study27 comparing the characteristics of people who died on days when the highest and the lowest black smoke concentrations were observed, did not found modification of the effect of air pollution on mortality by the SES. In Strasbourg, two studies explored the air pollution effects on myocardial infarction events28 and on asthma attacks.29 Results from the former supported the hypothesis that neighbourhood SES may modify the acute effects of PM10 on the risk of MI: differential susceptibility was suggested as the more plausible explanation since these most deprived population did not live in the more polluted place.30 On the other hand, socio-economic deprivation did not modify the relation between emergency telephone calls for asthma and concentrations of PM10, SO2 and NO2;28 this finding was confirmed using the number of β-agonist sales for asthma.31

Discussion

This literature review bears on the still small number of papers that investigated exposure and/or susceptibility differentials in Europe according to the socio-economical status, a rather recent topic that is yet less documented than in the USA and Canada. The European studies yield mixed findings regarding exposure disparities: in some instances, the association between air pollution and SES translates into poorer populations or areas being at greater exposure. Inversely, richer populations have been reported at greater exposure in other studies. However, beyond these variations, the general pattern in terms of health consequences is that deprived populations, although not always more exposed, experience greater harmful effects of air pollution, because of vulnerability factors.

In contrast, more discrepant results are observed in the non-European literature.

For example, among recent papers, the study by Charafeddine and Boden32 in the USA found that subjects living in the most affluent counties with high particulate levels are significantly more likely to report fair or poor health, compared to those in poorer counties who experience exposure to the same air quality, whereas Zeka et al.,33 in 20 US cities, showed stronger associations between PM10 and mortality for the less educated subjects (although not statistically significant). Similarly, poorer education was associated with a greater impact of air pollution on mortality in Shangai,34 whereas the Chinese Longitudinal Health Longevity Survey35 showed that elderly subjects living in more privileged urban areas were more affected by air pollution than their counterparts in more deprived ones. By the same token, Gouvenia and Fletcher36 found in Sao Paulo, Brazil, a slightly increased risk of mortality associated with PM10 among elderly people living in the most privileged areas, while Martins et al.37 in the same city showed that poorer areas presented the strongest association between PM10 and mortality among the elderly. Generalization from these partial observations is clearly premature. Absence of consensus as to the methodology used when investigating environmental and social inequalities (geographic unit, methods of statistical analysis, exposure assessment procedures and definition of deprivation) renders most of the results non-comparable and might explain part of these discrepancies.38,39

Nonetheless, several pathways and mechanisms are discussed in the literature to explain these social differences. Inequalities in environmental conditions are often put forward. Residential segregation may be one major reason why communities differ in their exposures. In Europe, socio-demographic disparities, notably those related to racial segregation, are less marked than in the USA; here, social and economic resources are the main determinants of environmental disparities. The housing market biases land use decisions and might explain why some groups of people suffer from both a low socio-economic status and bad air quality at their place of residence. One reason is that the presence of pollution sources depresses the housing market and provides an opportunity for local authorities to construct council housing at low cost.40,41 Symmetrically, the presence of council housing in a given urban area tends to depress the price of land over time, encouraging the setting up of activities and facilities that generate pollution.

‘Differential exposure’ beyond ambient air quality might partly explain why health effects of air pollution might be different across social classes. Living in a residential area with high air pollution levels does not necessary cause greater overall exposure. Affluent people are likely to have second homes outside cities and they may, therefore, spend less time at their main residence. Not taking this into account could yield exposure misclassification in that, while more affluent social categories may tend to live in central, more expensive, areas with higher pollution in some cities, their true year long exposure is probably overestimated.22 Conversely, subjects in deprived areas live in old dilapidated homes with poor ventilation and insulation, factors which favour the concentration of indoor pollutants. Moreover, they may be more likely to spend time close to or in the traffic, for example, working on the street rather than inside office buildings, or doing long commuting in public transport. Hence, the true daily and long-term exposures of these groups are probably underestimated. It is well documented that poorer people are more likely to suffer from several types of environmental exposure. In the German study by Schikowski et al.21 the authors demonstrated that, in addition to the increase of PM10 levels with poorer education, the prevalence of occupational exposures and of current smoking followed the same gradient. Along the same line, Bell and Dominici41 suggested that factors other than ambient air exposure, such as residential or occupational exposures, might explain why areas with a high Afro-American population proportion and high unemployment might exhibit a greater impact of air pollution in US cities.

People with a low SES may be more sensitive to air pollution-related hazards because of the high prevalence of existing diseases, an attribute which refers to ‘differential susceptibility’. For example, Forastiere et al.22 raised this hypothesis to explain their results, having excluded the causal pathway of inequalities in environmental quality. They found a higher prevalence of chronic conditions such as diabetes, hypertensive diseases and heart failure in low than in high-income groups. The former may receive inferior medical treatment for their conditions.35 They may also have more limited access to good food, resulting in a reduced intake of antioxidant vitamins and polyunsaturated fatty acids that protect against adverse consequences of particle or ozone exposure. In the particular case of infant mortality, Romieu et al.42 suggested that both micronutrient deficiencies and concurrent illnesses might decrease the immune response and make children more vulnerable to the adverse effects of air pollution.

It has been suggested that the presence of competitive risk factors in poorer areas might explain why health risks associated with air pollution may in some instances be greater among wealthier groups.31,35 Some authors argue that poorer people are affected by many other risk factors that tend to increase mortality rates owing to other causes such as violence and drug abuse. As a consequence, wealthier people may artefactually appear more vulnerable to air pollution in relation with their baseline risk level since they are relatively protected from other risk factors that affect disadvantaged groups.

Policy considerations

The issue of exposure and health inequalities in relation to ambient air quality is complex and calls for a global appraisal. There is no single pattern nor, of course, single solution. However, urban planning policies that would look for ‘spatial multipolarity and social diversity’ might play at the very roots of these inequalities. Multipolarity refers to the structure of our large metropolitan areas. Currently, with some variation across and within countries, European cities tend to be laid out in a concentric pattern: historical and cultural areas concentrated in the centre, with also a high proportion of businesses and expensive housing, while low-cost residential areas are progressively expelled to the outskirts, where also industrial activities are located. In contrast to this concentric structure, ‘multipolarity’ calls for urban poles that provide a range of amenities (housing, workplaces, commercial, cultural or leisure sites) tending to reduce the need for long distance commuting in polluted environments. Diversity is a complementary principle of multipolarity, where each pole would provide the widest possible variety of activities and, most importantly, of housing profiles, places for the rich being intermingled with council residence. This diversity scheme would prevent the formation of peripheral clusters of poor housing, which is typically associated with lack of access to good education and other cultural amenities: the further they are from the city centres, the more likely they are to be let in a marginal status. As described above, this is how inequalities in exposure to ambient air interplay with inequalities in other environmental stressors and vulnerability factors.

Conclusion and perspectives

Few European studies investigated the effect modification of socio-economic factors on the association between air pollution and health and much is yet to be understood. However, the general pattern of the current evidence is that deprived populations, although not always more exposed, experience greater harmful air pollution effects, because of vulnerability factors. Two research directions seem particularly relevant. Comparative exposure studies that would aim to assess the relative contribution of outdoor air and of a variety of microenvironments (at home, at work, while commuting, during leisure activities) across different social categories would be very informative. These disparities may vary substantially across cities and countries. A European-wide study might help understand the core determinants of these inequalities. For such a study to be valuable, however, great efforts should be put on harmonization of methods and definitions. Further, very little data concern children. Now, poverty and deprivation in early childhood may have adverse health consequences throughout the entire life. Focused studies in children are needed to better understand mechanisms through which health inequalities could arise later in life, a call which is in line with the avenue proposed by the PINCHE project.43

Funding

This work was partly supported by the WHO Regional Office for Europe in the framework of preparatory work towards the 5th Ministerial Conference on Environment and Health.

Conflicts of interest: None declared.

Key points

  • Poor populations do not always live in areas with higher outdoor air pollution in Europe; results are country and city specific.

  • Few European studies investigated the effect modification of socio-economic factors on the relation between air pollution and health and much is yet to be understood.

  • Nevertheless, there is a general pattern: irrespective of the level of exposure to ambient air, the poor are more affected by effects associated with air pollutants.

  • Policies aimed at reducing the root causes of these inequalities could strive to foster urban multipolarity and diversity, which require long-term urban planning.

Acknowledgements

An extensive version of this review was provided as a background document to the WHO expert meeting on ‘Environment and Health risks: the influence and effects of social inequalities’ (Bonn, Germany, 9–10 September 2009) and will be published in the context of the Fifth Ministerial Conference on Environment and Health.

References

View Abstract