The role of behaviour in vulnerability to air pollution—a framework and evidence review

1 Introduction

Poor urban air quality is a major public health and behavioural challenge (1). Outdoor ambient air pollution has been linked to approximately 40,000 deaths per year in the UK (2), and 4.2 million globally (3). Multiple toxicological and epidemiological studies have identified an association between exposure to air pollution and cardiovascular and respiratory diseases, leading to reduced life-expectancy [e.g., WHO (4) and COMEAP (5)].

Research also shows that the systemic effects of pollutants extend beyond the cardiovascular and respiratory systems, identifying other adverse health effects such as low birth weight and diabetes, as well as adverse impacts on subjective well-being, such as lower life satisfaction and happiness, higher anxiety, and depression (6–9). Beyond these health and psychological effects, a growing literature in economics and psychology highlights how air pollution impacts several individual, economic and social behaviours in a variety of settings from cognitive functioning (e.g., performance on standardised tests) (10–12) to dementia (13, 14), with other evidence linking pollution to labour productivity, traffic accidents, crime and migration (7).

Such adverse health and behavioural outcomes are often associated with factors such as socioeconomic status and proximity to pollutant sources and built infrastructure (7, 15–18). In addition, air pollution exposure is influenced by outdoor and indoor activities, occupational hazards, and travel patterns (19–22). Further, people may be more likely to avoid or protect themselves from air pollution exposure when they are aware of poor air quality and perceive it as a health risk (7). If people are constrained in their capacity to take adequate prevention and avoidance behaviours, pollution exposure might be exacerbated, e.g., (23). Taken together, this implies that there is differential vulnerability and adaptive capacity, i.e., certain groups of people are more vulnerable to air pollution risks than others due to their health, psychological and socio-economic characteristics as well as their behaviours.

Several previous papers have defined vulnerable subgroups, highlighted differential pollutant exposure that different groups face, and also highlighted the role of coping and adaptive behaviours (23–27). We build on and extend these vulnerability frameworks, conceptualising vulnerability as a function of three interrelated dimensions: (i) exposure, (ii) susceptibility and (iii) adaptation.

We advance the literature by decomposing vulnerability as an impact chain, which describes the determinants and outcomes. Doing this allows us to explicitly examine the behavioural components of vulnerability, particularly how behaviour can modify both exposure and susceptibility.

The rest of the paper is structured as follows. Section 2 presents the vulnerability framework, Section 3 assembles evidence on the determinants of exposure, susceptibility and adaptation, Section 4 provides a discussion on overlapping dimensions of vulnerability, and Section 5 concludes.

2 Air pollution vulnerability framework

Vulnerability to indoor and outdoor air pollution has been conceptualised in many ways in the literature, depending on the academic discipline and policy goals. A common aspect of these conceptualisations is that vulnerability refers to the extent to which an individual or group suffers greater adverse effects from air pollution than others. This term is sometimes used in a narrow sense, specifically focusing on health risks or factors relating to physical predisposition [e.g., Liao et al. (28)]. In such instances, the terms “vulnerability” and “susceptibility” are used interchangeably. Vulnerability can also have a broader definition, encompassing external factors, such as exposure to ambient pollutants (25, 29). This definition is often used by organisations such as the World Health Organisation (WHO), which has considered vulnerable groups to include children, the elderly, those with pre-existing cardiovascular and respiratory diseases, those exposed to toxic materials that interact with air pollution and the socio-economically deprived (30). Vulnerability can also encompass adaptation. For example, the IPCC define vulnerability as “the propensity or predisposition to be adversely affected, including sensitivity or susceptibility to harm and lack of capacity to cope and adapt (31).”

Our framework differs from existing syntheses in the following ways. First, the IPCC AR5 vulnerability framework is a broad, systemic perspective on climate risks and risk communication, whereas we focus on individual-level dimensions of vulnerability to air pollution. Second, while WHO (30) and Makri and Stilianakis (24) are more similar to ours in their focus on individual-level vulnerability, our framework allows us to explicitly identify interlinkages and feedback between exposure determinants, susceptibility and adaptation, while keeping in place the impact pathway from ambient pollution concentration to health and economic outcomes. For example, our framework allows us to explain how psychological factors such as perceptions of pollution exposure can modify indoor and outdoor behaviours which, in turn, determine individuals’ exposure levels. Therefore, behaviour is not just a supplementary factor in our framework; it represents a crucial component of vulnerability that can change individuals’ exposure and susceptibility to air pollution.

Our starting point is to consider the impact air pollution can have on well-being¹ through its impact on specific health and economic outcomes. Some individuals, due to their inherent characteristics, their choices and environmental factors within and beyond their control, suffer larger adverse effects than others for the same increase in ambient air pollution concentration. We refer to this differential effect as a person’s vulnerability to air pollution.

The above definition of air pollution vulnerability can be considered outcome-specific (i.e., relating to a specific health risk) or referring to an aggregation of multiple outcomes, such as overall well-being.

2.1 Components of vulnerability: exposure and susceptibility

Vulnerability can be decomposed into two components: “exposure” and “susceptibility.” Exposure is defined as a function of the concentration of pollutants in a microenvironment and the time spent by individuals in that microenvironment (32, 33). Exposure determinants are the factors that influence how much an individual’s exposure changes for a given change in exogenous air pollution concentration, i.e., the ambient air pollution concentration that is determined by conditions outside an individual’s direct control, such as emission sources and meteorological events. Exposure determinants are a function of where an individual lives, works, socialises, and how they travel between places. Susceptibility is the degree to which individual well-being is reduced as that individual’s pollution exposure increases.

Within a geographical area, such as a neighbourhood, individuals’ exposure can be similar if they spend similar amounts of time within their neighbourhood or indoors. Many exposure determinants can be perceived to emerge from individuals’ environmental context (e.g., location of one’s home) rather than their sociodemographic attributes. However, individual susceptibilities can be different even with the same exposure, resulting in differences in their vulnerability. This is because susceptibility depends on characteristics that change slowly over time (e.g., age and health conditions).

Determinants of vulnerability such as age, underlying health conditions, dwelling characteristics, or access to health-care can then be classified with respect to: (i) the component they influence, and (ii) the difficulty of changing them. For example, house ventilation impacts exposure and changing it (e.g., opening windows), is relatively inexpensive. On the other hand, access to adequate health-care could impact susceptibility and changing this factor may require relatively higher economic, time, or psychological costs.

From this decomposition we can see how exposure and susceptibility are different components of vulnerability, yet can have a compounding effect.

2.2 Modifier of vulnerability: adaptation

We refer to air pollution adaptation as any actions that can reduce vulnerability by changing exposure or susceptibility. These factors can be conceptualised as antecedents to exposure and susceptibility, while recognising the possibility of some feedback as discussed in Section 4.

2.3 Graphical overview

Figure 1 presents the decomposition of vulnerability through an impact chain which describes the determinants and intermediate outcomes, drawing from the IPCC AR5 (31). Starting with the top-centre, one can think of some level of “pollution concentration” that is exogenous (i.e., the pollution that an individual has no control over). Vulnerability can then be decomposed into the determinants of an individual’s exposure and their underlying susceptibility, which can be moderated by behavioural adaptation “antecedents” and mediated by routine behaviours. Therefore, behaviour is a cross-cutting factor as opposed to an independent dimension of vulnerability. The factors noted in the left-hand column (psychological factors, resources to adapt and response to information and policy instruments) are adaptive behaviours that moderate the relationship between exogenous pollution concentration and individual-level outcomes. The factors noted in the top right (e.g., indoor, outdoor and occupational activities) denote routine behaviours or actions that mediate the relationship between exogenous pollution and individual-level outcomes.

In the column on the left, the determinants of adaptation which are moderators of vulnerability are described. These are comprised of (i) psychological factors, (ii) resources, and (iii) responsiveness to policy change and information campaigns. In the column on the right, we further break exposure determinants down to locational, based on factors relating to where people live and work, and behavioural, relating to their activities in these places which mediate the relationship between exogenous pollution and outcomes. Susceptibility is a function of demographic characteristics and underlying health conditions. Many of the health-related susceptibility factors are also considered as “outcomes” as air pollution exposure can exacerbate these conditions.

The next sections present a summary of recent evidence for each determinant of vulnerability shown in the figure. The discussion of outcomes is embedded within the discussion of each determinant, i.e., when describing exposure due to poor ventilation, we discuss associated outcomes such as higher incidence of respiratory illness.

3 Evidence for determinants of vulnerability

To identify relevant literature on vulnerability to air pollution, we used the proposed framework to guide structured searches in Web of Science and Google Scholar. Our search focused on academic and grey literature published in English from the year 2,000 onward, with emphasis on studies in high- and middle-income countries (HMICs). We acknowledge that findings from HMICs may not transfer directly to low-income countries, where pollution levels, occupational structures and adaptation options differ substantially. This is why we choose to focus on the former. Importantly, air pollution remains a significant public health burden in HMICs given the absence of a safe exposure threshold (34), making vulnerability to air pollution an important continued focus for research in these contexts. Moreover, integrating behavioural dimensions of vulnerability in HMICs is a specific gap that our paper addresses.

One further dimension on which we limit our scope is to be kept in mind. We acknowledge that different chemical pollutants (e.g., NO2, SO2, VOCs etc.) can have different impacts on health. However, the emphasis of our review is not on the differences in production and health impacts of these specific pollutants, but rather on how individuals can be affected by pollution in general, and how they can adapt to change their vulnerability. There is significant evidence suggesting that people do not tend to differentiate between pollution sources in their behaviour (35–37). In fact, communication of air pollution health risks to the public often takes the form of composite indices like the UK’s Daily Air Quality Index (DAQI), or the US and EU’s Air Quality Index (AQI) that comprise concentrations of multiple ambient pollutants and form the basis for initiating pollution alerts at specific days and times.

We cover literature spanning epidemiology, atmospheric science, environmental health, transport and urban planning, building engineering, economics, and psychology to reflect the interdisciplinary nature of work on this broad topic. We prioritized peer-reviewed empirical studies and systematic reviews, but also included high-quality grey literature and policy reports relevant to behavioural and social dimensions of air pollution. We excluded studies on pollution from ozone, volcanic eruptions, wildfires, and extreme weather events, as well as studies addressing adaptation behaviour unrelated to air pollution vulnerability.

While the review is not a formal systematic review, we aimed for representative coverage within each thematic area, preferring to cite recent reviews where evidence is relatively robust, such as on the health impacts of air pollution, and individual empirical papers in areas where the evidence is more nascent.

In what follows, we summarise the literature by the components of vulnerability identified in Section 2.

3.1 Exposure

3.1.1 Proximity to pollution sources

The vast majority of evidence linking air pollution exposure and its health impacts in recent years are ecological, i.e., they use location of residence (with varying degrees of spatial resolution) as a proxy for assigned exposure to ambient (outdoor) air pollution (38–43). Proximity to high-density traffic and population density are common determinants of exposure in a number of these studies, which can also be linked to neighbourhood SES and other demographic characteristics. Other research has found associations between high air pollution and higher levels of neighbourhood deprivation, and that minority ethnic groups can be exposed to higher levels of pollution (38, 40, 43–45). These relationships are complex and geographically variable (15), with some notable exceptions where traffic related emissions are higher in high-SES areas (46).

While traffic is a significant contributor, urban pollution can originate from other sources. A global systematic review concluded that around 25% of urban ambient PM2.5² was contributed by traffic, 15% by industrial activities, 20% from domestic fuel burning, 22% from unspecified sources of human origin, and 18% from natural dust and sea salt (47).

A more recent global assessment across 204 countries estimates sectoral contribution to PM2.5 emissions and mortality attributable to emissions from these sectors (48). It finds that eliminating emissions from fossil-fuel combustion could avoid about 1.05 million deaths globally, or 27.3% of the total. Removing solid biofuel combustion, mainly from residential cooking and heating, could prevent another 0.77 million deaths, or 19.2% of the total. Transport, agriculture, and anthropogenic dust each contribute significant shares, varying between 6% and 9.3%. A similar source-apportionment study for 857 European cities, finds that for PM2.5-related mortality, the largest contributors were the residential sector at 22.7% and agriculture at 18.0%, followed by industry, transport, energy, and shipping. For NO2-related mortality, transport was by far the dominant source, contributing 48.5%, with smaller contributions from industry, energy, residential sources, and shipping (49).

The source apportionment studies referenced here focus primarily on measuring and modelling sources of outdoor air pollution. However, people in HMICs spend 60%–90% of their time in indoor environments (50), where the mix of pollution sources shift toward heating and cooking technologies and behaviours like smoking and use of aerosols. In the following section, we cover the evidence on indoor air pollution exposure in HMICs.

3.1.2 Indoor air pollution

Within buildings, individuals are exposed to a mix of pollution from ambient sources infiltrating inside, and pollution from indoor sources. These include emissions from building materials and furnishing, heating and cooking, the usage of cleaning products, and other occupant behaviours such as smoking (22, 23, 51).

The relationship between indoor air pollution and health has been well studied in developing countries due to the preponderance of open fires and use of biomass for heating and cooking and the associated health risks (51, 52). Fuel for residential heating and cooking in high-income countries tends to be electricity or gas, with most households having efficient stoves and flues that emit little pollution relative to the use of biomass. Yet, approximately 240 million people in Europe and North America still use solid fuels for heating purposes (53). Within Europe, the usage of solid fuel for heating is common along the Mediterranean coast (Portugal, Italy, Greece etc.) as well as Germany, Ireland and Scotland (54–56).

In the UK, recent policy incentives to encourage biomass burning for power and heating have significantly increased PM emissions from domestic biomass burning in urban areas in winter, rising from less than 5% in 1990 to over 17% in 2012 (57). To counteract this trend, recent legislation has restricted the sale of coal, wet wood and manufactured solid fuels for residential use (58).

A recent study of wood burning appliances (such as those in use in European countries) found that fireplace operation led to an increase in PM10 levels nearly 12 times that of ambient concentrations; while woodstoves were found to double PM10 concentrations (56). These results are in agreement with other recent studies that show increase in PM10 and an overall decrease in indoor air quality when indoor fireplace and wood stoves are operated (59–63).

Gas stoves, although less polluting than solid fuel burning, can also have a large impact on indoor air quality. Households with gas stoves have from 50% to 400% higher indoor concentrations of NO2 (depending on location, season and household characteristics) than households with electric cooking appliances (64). This is particularly the case when there is no dedicated ventilation system that vents outdoors (65). This higher exposure has a direct impact on people’s health, specially populations with higher susceptibility. This has long been known in the literature, with children exposed to a gas stove (a long-term increase of 30 μg/m3) having about a 20% higher change of developing respiratory illness (66). More recently, evidence has emerged on the effect of household gas appliances on lower cognitive function and inattention symptoms in preschooler children (67).

In a recent study on indoor air pollution exposure in London, Metcalfe and Roth (68) find that measured indoor PM2.5 in 258 households are at around 11 μg/m3 on average, which is higher than WHO standards. Activities such as smoking, cooking, and heating increase indoor PM2.5 to 14.6 μg/m3 during occupancy hours, and households are willing to pay £4.8 for a 1 μg/m3 reduction in indoor PM2.5. Maung et al. (69) conduct a systematic review on the health effects of indoor air pollution covering studies from Asia, Europe and North America finds that smoking, cooking, heating, candles, and insecticide use to be sources of indoor PM2.5. The review highlights the need for more research on the health effects of indoor air pollution where the evidence is relatively sparse compared to sources of indoor pollution. Another scoping review of indoor pollution focused on high-income countries finds that socio-economic inequalities in pollution exposure extend to the indoor environment, with low-SES individuals exposed to elevated levels of indoor air pollution driven by differences in building materials and indoor activities (70).

3.1.3 Ventilation

Studies have demonstrated that adequate ventilation is an important factor in reducing residential exposure to indoor sources of pollution such as cooking, radon and tobacco smoke. Poor ventilation can increase the incidence of damp and mould, and is also associated with a higher incidence of respiratory illness. Bentayeb et al. (71), for example, finds significant odds-ratios of 1.67 and 1.85 for the association between PM10 and breathlessness and coughing, respectively, in poorly ventilated nursing homes, while adequately ventilated nursing homes did not exhibit a significant relationship.

Improving building fabric is an important component of decarbonisation. However, an unintended consequence of energy efficiency and building insulation upgrades that improve the thermal envelope of a building may be a reduction in ventilation. A unit increase in household Standard Assessment Procedure (SAP) rating was associated with a 2% increased risk of having asthma in a survey-based assessment of higher energy efficiency homes in the UK (72). Other work has shown that an increase in pollutant concentrations is correlated with lower building air exchange rates after energy efficiency upgrades (73). Consequently, modelling studies have cautioned that improvements in energy efficiency must be accompanied by corresponding improvements in ventilation, to mitigate the risk of increasing indoor air pollution exposure for at risk groups such as those aged over 50 years, or suffering from cardio-respiratory diseases (74).

3.1.4 Transport mode and active travel

Multiple studies find differences in air pollution exposure based on transport mode. A systematic review reports the highest exposure for motorised transport users - car commuters followed by bus and motorcycle users (75). While exposure is higher for commuters using motorised transport, the inhaled dose is greater among active commuters due to larger inhalation rates and commuting time. However, the benefits of active commuting are larger than the associated risk as commuters using motorised transport were estimated to lose one additional year of life expectancy (YLE) compared to cyclists.

A quantitative review of European studies found that pedestrians are consistently less exposed than those using bus, bicycle or car modes (76), while active commuting has been found to result in small health benefits and only slightly increased risks (21).

A review of empirical and modelling evidence on a general adult population suggests that air pollution, specifically high pollution events, may diminish physical activity (77). For longer term outcomes, physical activity may have diminishing health benefits due to exposure to air pollution. A systematic review of the health benefits of modal shift to active transport, found that the benefits strongly outweighed detrimental effects of traffic incidents and air pollution exposure on health, regardless of geographic location (78).

3.1.5 Physical activity

The evidence on the impacts of air pollution exposure during physical activity finds that moderate to high levels of physical activity offset the short-term harmful effects of air pollution. A systematic review found that for healthy adults, exercise is generally beneficial, even in high pollution environments (79). However, for those with pre-existing conditions such as COPD, even low intensity activity may increase the adverse effects of air pollution. This finding is in line with research by Madureira et al. (80), who focus on the health effects of physical activity while exposed to traffic-related emissions and observe more susceptible populations are likely to suffer more intensified impacts. They also note that markers for long-term cardio-respiratory disease increase in individuals who exercise in environments with high traffic emissions.

3.1.6 Occupational activities

Increased air pollution exposure has been linked with some specific occupational activities, including driving, mail delivery, food preparation, construction. Frederickson et al. (81) shows that professional drivers are exposed to 1.72 μg/m3 (micrograms per squared meter) of PM2.5 more than a background monitoring station. Merlo et al. (82) undertake a historical mortality cohort study of bus drivers and maintenance workers exposed to air pollution in Genoa. The study finds evidence that exposure to urban air pollution is associated with mortality from Hodgkin’s lymphoma and lung cancer. In a meta-analysis on the relationship between traffic-related air pollution and lung cancer, Chen et al. (83) find that occupational exposure among professional drivers significantly increases the incidence and mortality of lung cancer. Two studies looking at the effect of air pollution exposure on mail carriers report positive associations with risk of respiratory disorders. Zuskin et al. (84) find that smoking habits also have an effect, while Karakatsani et al. (85) further report that exposure to particulate matter is associated with elevated risk for chronic cough, chronic phlegm and bronchial asthma.

Sofuoglu et al. (86) and Sjaastad and Svendsen (87) both report high exposure on particulate matter for workers in kitchens where deep frying occurs. Bigert et al. (88) report increased risk of myocardial infarction in female cooks in a Swedish cohort study of manual workers. In a pilot study, Phillips and Oh observe high concentrations of particulate matter emissions in food trucks, but note that pollution exposure may be manageable with good ventilation systems (89).

Sauni et al. (90) find that asthma causes work disability among construction workers, with 72% stating that symptoms worsened due to work. Torén et al. (19) observed high mortality rates in workers with any occupational exposure to vapours, gases, fumes, and dust due to COPD.

3.2 Susceptibility

3.2.1 Health factors

3.2.1.1 Physical health

There is a significant body of evidence on the health risk associated with exposure to air pollution. Air pollution has been linked to higher natural cause mortality rates; adverse cardiovascular health outcomes; respiratory diseases including chronic obstructive pulmonary disease (COPD), asthma, bronchitis symptoms, reduced lung function, and lung cancer. Air pollution has also been linked to adverse outcomes in pregnancy, diabetes and obesity, amongst others. It is also worth noting that different pollutants and dispersion pathways can have different impacts on health. It is beyond the scope of this review to cover this topic in detail and we would point the interested reader to review papers and reports by organisations such as the World Health Organisation (4), the Health Effects Institute (91) and UK Committee on the Medical Effects of Air Pollutants (5).

3.2.1.2 Cognitive function and psychological health

Lu (7) conducts a systematic review on the psychological effects of air pollution, finding negative associations between pollution and well-being, cognitive function and decision-making ability, along with other factors. The paper notes that it is unclear whether the effect of air pollution is more physiological or psychological, and that only a small percentage of studies have assessed both actual and perceived pollution levels. There is also evidence on negative effects of air pollution on cognitive performance Shehab and Pope (10), productivity (92, 93) and exam performance (11). Finally, there is also emerging evidence that air pollution increases road traffic accidents (94) as well as unethical and criminal behaviours, possibility through increased stress and anxiety (7).

3.2.2 Demographic and socioeconomic factors

3.2.2.1 Age and sex

The young (<18) and the elderly (>65) have been identified as vulnerable populations, with studies pointing to increases in severity of asthma among young adults and children (42, 95). Among elderly populations, multiple studies report increased risk of respiratory and cardiovascular disease associated with higher exposure to particulate matter (71, 96–98). Some studies further reported increased risk of these diseases among older women (96, 99). Other review papers point to mixed results (100), and inconclusive evidence linking sex to air pollution related mortality (29).

3.2.2.2 Children

Children are consistently identified as one of the groups most vulnerable to air pollution. Their heightened susceptibility has biological foundations. Since they breathe more air relative to their body size, a given ambient concentration delivers a larger effective dose than it would to an adult, and their developing respiratory and immune systems are less able to withstand it (101). Exposure is associated with hindered lung development and greater severity of childhood asthma (42, 95, 102), while early-life exposure to indoor pollution sources such as gas appliances has been linked to lower cognitive function and inattention symptoms in preschoolers (67). Since these exposures coincide with sensitive developmental windows, their consequences can persist into later life. Ebenstein, Lavy, and Roth (11), for example, find that pollution during high-stakes examinations leads to worse exam performance, with measurable effects on subsequent economic outcomes. Children’s exposure is also shaped by where they spend their time and by behaviours largely outside their own control. Much of it accrues indoors, at home and at school, where they spend most of the day and indoor sources can dominate (69). Indoor NO2 from gas stoves, for example, has long been associated with an elevated risk of respiratory illness in children (66). We discuss how adaptation behaviour can reduce children’s exposure in Section 3.3.3, and how children are an important overlapping dimension of vulnerability in Section 4.

3.2.2.3 Income and deprivation

As socioeconomic status is a determinant of overall health, identifying the direction of causality between socioeconomic inequalities, increased exposure to air pollution and health is complicated. Conceptually, susceptibility factors and outcomes are interlinked since air pollution poses long-term health risks over individuals’ lifetimes, prompting some literature to treat susceptibility factors as “life course risks”, as opposed to a cause-effect relationship between pollution exposure and health outcomes (103). Evidence suggests that lower income and SES groups suffer from higher rates of cardiovascular disease (99, 104) and mortality (46), amongst other conditions, making them more susceptible to adverse impact of air pollution.

3.3 Adaptation

We define adaptation as comprising actions taken by individuals to reduce or avoid risks posed by air pollution. These include efforts to avoid air pollution exposure (e.g., changing travel route, staying indoors), defensive actions to mitigate or cope with exposure (e.g., buying face-masks or medical insurance), and efforts to reduce contributing to air pollution (e.g., cycling rather than driving). While higher exposure and higher susceptibility increase vulnerability to air pollution, greater adaptation reduces vulnerability. While adaptation operates at the individual level, a change in norms of adaptation behaviour can change overall exposure or susceptibility for a geographic region.

3.3.1 Psychological factors

A necessary antecedent to adaptation behaviour is being sufficiently aware and concerned about the risks posed by exposure to air pollution.

3.3.1.1 Perceived exposure

In addition to actual exposure to air pollution, perceptions of air pollution levels can also affect adaptation. For instance, in a survey of respondents in the Haifa region in Israel (n = 900), Zeidner and Shechter (105) found that the perceived level of pollution is a stronger predictor of emotional reactions towards air pollution and willingness to pay to reduce pollution than the objective level of pollution exposure. Some studies highlight the importance of place and location in influencing air pollution perception (106). Coi et al. (107) find that risk perception is higher in areas characterised by air pollution due to human activity (e.g., industry) in a survey (n = 282).

Other studies highlight that people significantly underestimate their own exposure which, in turn, implies fewer adaptive behaviours. Brody, Peck, and Highfield (108) surveyed residents of two cities in Texas (n = 378), finding that perceptions of air quality were not significantly correlated with air quality readings from monitoring stations. Instead, perceptions appear to be influenced by setting (urban vs. rural), state identification, access to information, and socioeconomic characteristics such as age, race, and political identification. They document a strong state-level “halo effect”, where respondents’ views of local air quality are influenced by their belief that pollution in other non-Texan cities across the United States is far worse.

Other studies also show that people tend to underestimate the personal risk of air pollution exposure, despite living in polluted areas. For instance, (109) conducted household surveys in two municipalities (n = 489) in south-central Chile, both highly polluted with smoke from wood-burning stoves, to find a strong “home halo effect”: individuals perceive their homes as less polluted than their cities, even though studies show indoor pollution in the municipalities studied might be equal to or higher than outdoor pollution. While they do not find perceptions are explained by age, gender or health status, they do find that income plays a role: families from the middle class perceived that their homes have, on average, better air quality than the city.

3.3.1.2 Perceived susceptibility

How people perceive air pollution risks is highly correlated with whether they will develop self-protective behaviours (110). Studies have therefore examined if perceived risks and worry about air pollution predict protective behaviours. Zeidner and Shechter (105) found that individuals with higher perceived risks are more prone to put in time and more willing to allocate financial resources towards pollution abatement. Shreedhar (111) found that self-reported risk aversion and worry positively predicts self-reported adaptive behaviours such as avoiding peak travel times and high-traffic routes, and wearing anti-pollution face-masks amongst cyclists in London (n = 181). Zhou et al. (112) also show that risk perception predicts self-reported intentions to avoid air pollution and face-mask use amongst young adults in Beijing (n = 164). Lu (7) found that perceived physical and mental health risk from air pollution related smog predicted skilled workers’ migration intention in the capital region of China (Beijing-Tianjin-Hebei).

Several studies suggest that air pollution reduces subjective well-being measures like life satisfaction and happiness, and increases anxiety, annoyance, depression and even suicide (7). Evidence also suggests that people do not psychologically adapt to air pollution exposure over time. Menz and Welsch (113) use data from 48 countries over 1990–2006 to find that people’s life satisfaction does not habituate to air pollution.

3.3.1.3 Other factors

Studies have also shed light on factors that determine risk perception, which in turn can indirectly impact behaviour. These factors include age, sex, education, marital status, income, domain literacy (e.g., health or environmental), information behaviours (e.g., using internet, social media and news), value orientation, affect and emotions, as well as personal experience (107, 110, 114–116). For instance, those with higher level of education were more likely to report taking action to mitigate the environmental risk of climate change Lee et al. (117).

In one of the few papers studying this issue in the context of air pollution, Subiza-Pérez et al. (118) surveyed 338 non-experts and 33 experts from Basque Country in Spain, to evaluate environmental risk factors, including air pollution. They found that non-experts assigned a higher risk score to pollution than experts, although the ranking of relative risks were quite similar across both groups. The best predictors of air pollution exposure were being older, having environmental knowledge and having biospheric and egoistic environmental values, whereas educational level was less important. In another paper, Wei et al. (116) compared residents of two Chinese cities, namely Hefei and Chizhou (n = 700 and 400, respectively), which differ considerably in their smog experience. Surprisingly, differences in individuals’ smog information sources, risk perception, and protective behaviour were less stark. This suggests that smog experience and smog information increase risk perception which can, in turn, increase responsibility attribution and intentions to engage in protective behaviour.

3.3.2 Resources to adapt

Apart from knowing how and when to avoid air pollution, people can also buy goods and services to protect themselves. This can include, for example, purchase of air filtration equipment, face masks, or medical insurance. Several papers in this domain suggest that such expenditures increase when air pollution or income increases.

3.3.2.1 Buying protective measures

Liu et al. (119) find that elevated air pollution levels are positively associated with online searches for PM2.5 protective masks and air filters in China. Specifically, a 10 μg/m3 increase in ambient PM2.5 is associated with a 3.6%–8.4% increase of mobile queries for such masks, and a 1.1%–4.7% increase in searches for air filters. People in richer (and more polluted) cities tend to search more for masks and air filters than people in poorer (and cleaner) cities, suggesting that richer populations are more likely to adopt these protective behaviours.

Other studies find that individuals in China respond to contemporaneous increases in air pollution concentrations by buying more face-masks, with the magnitude of the effect ranging from 54.5% for all masks to 70.6% for specific PM protective masks, for a 100-point worsening of the AQI (120). Using data from air purifier markets in China, a study found that individuals are willing to pay $1.34 per year to remove 1 μg/m3 of PM10, with richer households having a higher willingness to pay (121).

More broadly, recent reviews suggest that air filters, purifiers and face-masks can mitigate air pollution exposure. However, how individuals actually use them in everyday scenarios remains a gap in existing research (122).

3.3.2.2 Buying medication and medical insurance

Another form of protective behaviour is defensive expenditure, i.e., individuals’ purchase of medication and medical insurance to treat (or reduce the risk of) pollution-related illnesses. Evidence on defensive expenditures in the US shows that a cap-and-trade program that reduced NOx emissions in 19 US states caused a 1.6% decrease in expenditure on medication and a 2.3% decrease in medication costs related to respiratory and cardiovascular diseases (123). Using data from a national survey of households in China, Zhao (124) shows that higher air pollution increases the probability of purchase of health insurance and expenditure on health insurance premiums. They also find that male-headed households are 11.7% less likely to purchase insurance than female-headed households, and that education level and income level have a positive association with household insurance purchases.

Chen and Chen (125) found that a rise in air pollution is associated with an increased probability of purchasing medical insurance using panel data from the China Health and Nutrition Survey from 2004 to 2015. They highlight that the positive effect is mainly observed on women, children, the elderly, and people with high income and educational levels, all of whom are more likely to take an avoidance behaviour either because they are more susceptible (e.g., due to age) and thus more motivated, or because they have a higher capability to adapt due to their resources (e.g., income) and knowledge (e.g., education). Other studies show there may be systematic differences by ethnicity. Using evidence from Los Angeles county in the US, Timmins, and O’Neil (126) finds that white people exhibit a higher marginal willingness to pay than Hispanics to avoid air pollution-induced cancer risk.

3.3.2.3 Moving to less polluted neighbourhoods

Air pollution can impact individuals’ choice of where to stay. People with greater capability (e.g., financial resources) and motivation (e.g., perceived risk and worry about air pollution), often choose to move out of more polluted areas. Capability and motivation can then interact with differential vulnerability by ethnicity and race, as identified earlier in Section 3.2.2. Using spatial data on toxic release inventory sites and decadal changes in the population of areas in California, a study from the US finds that areas with lower air quality see a reduction in their populations by between 10%–16% over ten years from 1990 to 2000 (127). The same authors find evidence that race-based residential sorting can increase even if differences in local air quality between neighbourhoods are reduced, possibly due to group-based sorting, i.e., living close to others of the same race effects dominating income-based sorting, i.e., moving to cleaner neighbourhoods as income increases (128).

More recent work finds historical evidence for residential sorting dating back to the Industrial Revolution. Pollution from industrial coal use in large cities in England and Wales modified spatial organisation within these cities—with low-skilled workers moving to more polluted areas, usually in the Eastern regions of these cities (129).

3.3.3 Response to information and policy instruments

3.3.3.1 Impact of behavioural interventions

Characteristically, behavioural interventions are often implemented in a one-shot manner, and aim to provide information on pollution in a salient way. In the UK, Public Health England (1) find that overall, the evidence on the effectiveness of behavioural interventions is not strong and characterised by high uncertainty based on the number and quality of studies. However, the most effective interventions reviewed were (i) Clean Air Day or similar initiatives promoting awareness regarding air pollution problems, and (ii) eco-driving initiatives designed to improve driving behaviour and reduce engine idling time. Similarly, Riley et al. (130) considers 37 articles on communication and behaviour change campaigns and find that although most of these studies illustrated the potential for behaviour change, many of these have ambiguous results. Having said that, empirical evidence for behaviour change in response to alerts or other interventions exists across a number of domains.

3.3.3.2 Changing outdoor activities

The role of information in driving air pollution avoidance behaviour has often been studied in the context of how air pollution alerts impact outdoor activities. Many studies focus on how alerts on high pollution days influence time spent outdoors, exercise, park attendance, or cycling (130). A recent meta-analysis, finds a 1.1% decrease in physical activity associated with a one unit increase in ambient PM2.5 (μg/m3) (131). A study on air pollution and attendance at outdoor recreation facilities in southern California, USA, finds that individuals respond to smog alerts by reducing time spent at such facilities, specifically by reducing attendance at a zoo and an observatory in Los Angeles. The paper shows that populations with higher susceptibility such as children and the elderly are more likely to reduce attendance in response to the alerts (132).

While these studies suggest that alerts increase the capability of people to take immediate action, their long-run effectiveness and whether they help individuals build the motivation to adapt has been less studied. In this vein, another study using the same data finds that while individuals are responsive to alerts on the first day they are issued, they are less willing to change behaviour during subsequent consecutive days when alerts are issued. However, this drop-off in reduction of attendance at outdoor facilities is less pronounced among vulnerable groups (children and the elderly), suggesting that these groups perceive a higher risk and remain more sensitive to air pollution alerts, even if the alerts are issued in quick succession (133).

Using similar data on avoidance behaviour, namely attendance at Bristol Zoo in the UK, another paper finds that attendance of people with an annual membership reduced on days when there was an air pollution warning (134). In survey data from the United States (n = 10,000), some individuals, particularly those with respiratory disease, report changing their activities on days with poor air quality. The most commonly reported change in activity was spending less time outdoors and closing the windows of one’s house (135).

Using administrative data from Sydney, Australia, another paper finds changes in cycling behaviour in response to air pollution alerts. Cycling is reduced by between 14%–35% on days when an alert is issued (136). Further, the reduction in cycling activity is higher on weekends than on weekdays, suggesting that people are more likely to reduce leisure trips than work trips in response to air pollution alerts. This paper corroborates the “alert fatigue” effect found in earlier work, i.e., avoidance behaviour diminishes over successive days on which air pollution alerts are issued (133).

3.3.3.3 Changing transportation mode and route

There is also evidence that individuals may change their travel routes when presented with information on ambient air quality. A choice experiment and stated preference survey on urban cyclists in the US and Canada finds that cyclists prefer cleaner travel routes and are willing to spend a little more time travelling to reduce their exposure to pollution. Further, the paper found that cyclists are more sensitive to the level of pollution experienced on a regular basis rather than to instantaneous exposure to high levels of air pollution, which can occur while riding behind a transit bus or diesel truck (137).

A small sample (n = 22) study involving a trial of a pollution sensor and accompanying mobile phone application finds that the participants did not change their travel behaviour and attitudes toward pollution after engaging with the information provided through the application (138). Another study (n = 104) on parents and guardians of school going children in Antwerp, Belgium, finds that personalised information on ambient air pollution along their used and alternative travel routes encourage a substantial proportion of individuals to change to less polluted routes (139).

3.3.3.4 Impact of driving restriction policies

Several cities across the world have implemented ecologically-based policies aiming to reduce traffic emissions across set geographical areas, such as Low Emission Zones (LEZs). Recent papers have examined the effects of such policies on pollution concentrations and their consequent health implications (140–142). Congestion pricing is being rolled out in several cities, such as New York, which saw a 7.5% drop in total traffic (143), showing promise for the short-term effectiveness of such policies. However, there is relatively less evidence on changes in the modal mix of transport, i.e., the split between travel modes, in response to such policies. A paper on driving restriction policies in Chinese cities finds that public transport passenger volume increased in response to driving restriction policies (such as odd-even days and parking restrictions). Examining data specifically from Beijing, the paper did not find any change in the growth rate of the modal share of public transport in response to restrictions on private transport use in the city (144).

This limited evidence on modal responses is important because area-based traffic and emissions policies may alter not only the volume of travel, but also where trips occur and which modes are used. More recent evidence suggests that some of the effects of these policies may reflect the displacement, rather than the elimination, of emissions-generating travel. In Madrid, Galdon-Sanchez et al. (145) estimate that an LEZ (Madrid Central) reduced congestion by 16% and pollution by 19% within the regulated area. Tassinari (146) similarly finds reductions in traffic within the zone, but also evidence of displacement to other parts of the city. The same study shows that the internal traffic effect faded after around seven months, as households adapted and some trips shifted toward public transport. Evidence from London points more directly to changes in travel mode. Xiao et al. (147) find that children living in the ULEZ area were more likely than those in to switch from inactive to active school travel, with an odds ratio of 3.51, and less likely to switch from active to inactive travel, with an odds ratio of 0.22, compared to children in Luton, a neighbouring city.

4 Overlapping dimensions of vulnerability

Having reviewed the literature on several dimensions of vulnerability to air pollution, we posit that more research is needed to understand the extent to which one factor can interact with others in specific contexts. For example, lower SES affects exposure through worse building characteristics or neighbourhood-level pollution, but can also be linked to susceptibility (adverse health outcomes identified in Fuller et al. (6)) and lack the resources for effective adaptation (defensive expenditure or relocate away from pollution) (126). Unpacking these interdependencies more systematically would help design effective policy to reduce vulnerability.

One such interdependency is apparent from several studies that have identified increased risk of respiratory and cardiovascular disease in older populations exposed to air pollution (71, 96). In addition, there is evidence that children can suffer from hindered lung development and increased severity of asthma as a result of exposure (42, 102). Compounding these risks, a recent review has suggested that for sensitive groups (elderly persons with pre-existing health conditions such as COPD), low intensity activity may exacerbate the negative effects of exposure to air pollution (79).

A second channel of interdependence is potential feedback from exposure and susceptibility to adaptation. For example, adverse health consequences of air pollution and consequent medical expenditures can increase the resource constraint for individuals to engage in defensive expenditure or change their occupational exposure.

Another significant factor contributing to the compounding effects of vulnerability is the occupational activity of individuals, particularly among low-skilled workers. Workers engaged in occupations that involve high exposure to air pollution (e.g., construction, food service, or delivery) face disproportionate health risks. Construction workers often work in environments where exposure to fine particulate matter (PM) and other hazardous pollutants is elevated. These workers can have a limited ability to adapt as transitioning to less hazardous occupations is difficult due to the need for specialised skills, job market barriers, or limited educational resources. Even when aware of the risks, their financial circumstances can prevent them from leaving these occupations or undertaking potentially costly defensive expenditures such as purchasing high-grade protective equipment. Occupational exposure thus amplifies health risks and limits adaptation behaviour to the extent that low-income individuals may be more likely to take on these “exposed” occupations.

5 Conclusion

We assemble and integrate recent evidence on population and behavioural risk factors associated with vulnerability to air pollution in high and middle income countries. Our main contributions are to formalise a conceptual framework that allows us, and future research, to consider the components of vulnerability systematically. In this paper, we use this framework to assemble recent evidence on air pollution vulnerability. The framework also leads us to consider how factors classified into susceptibility, exposure and adaptation can interact in systematic ways.

We document evidence on the relationship between population characteristics like socio-demographic attributes and adverse health outcomes, including cardiovascular and respiratory diseases. Building on this, we identify how these risks can be exacerbated by socioeconomic and racial inequalities in where people live, driven by residential sorting, proximity to traffic emissions and population density. Finally, we draw from a range of behavioural research to provide evidence for people’s ability to undertake adaptive behaviour and alter exposure. In the short run, people can shift their outdoor activity patterns, travel routes, choice of transport mode or defensive expenditure. Over the longer term, individuals may choose to move away from more to less polluted areas. There is evidence of certain groups undertaking evasive action—those of higher education, those engaged in active transport, and at-risk groups such as the children, the elderly and those with respiratory conditions.

This review has two limitations. First, given the wide range of literature covered, it is beyond the scope of the paper to undertake a systematic mapping of evidence or provide a quantitative comparison of the magnitude and strength of evidence regarding particular dimensions of vulnerability. Second, we do not formulate a Boolean search string to recover the exact same set of papers as would be the case in a systematic review or meta-analysis. These could be topics for further research using our conceptual framework to produce a systematic review or meta-analyses and/or build the evidence around our framework specifically for low-income countries.

Although pollutant emissions and their intensity relative to GDP have generally declined in HMICs (148), the absence of a safe exposure threshold means that vulnerability to air pollution remains relevant for research and policy in these countries. In our review, we find that the literature on adverse health impacts of air pollution is relatively large and robust. While there is growing evidence on policies like LEZs that can mitigate air pollution, we know relatively less about barriers to adaptation and behavioural responses pertaining to air pollution. In particular, while we have documented evidence that some groups of people may be more vulnerable to air pollution, we know less about the behavioural patterns (such as active travel or indoor ventilation) that can modify the vulnerability of these groups. Taken together, this suggests a need to better understand how to design, implement and evaluate health and behavioural policies to account for differential vulnerability.

Generative AI statement

The author(s) declared that Generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Gaetano Settimo, Istituto Superiore di Sanità, Italy

Reviewed by: Elena Bozzola, Bambino Gesù Children’s Hospital (IRCCS), Italy

Yangyang Wu, Southwest Jiaotong University, China

Footnotes

¹Well-being considered broadly, incorporating physical and mental health and economic well-being.

²Fine particulate matter, refers to a mixture of solid particles and liquid droplets found in the air. Its concentration is measured in the total mass in micrograms per meter squared (μg/m3) of particles smaller than a given diameter with PM2.5 indicating particles with a diameter of 2.5 micro meters (μm) and smaller.