Sage Journals: Discover world-class research

Abstract

People living in the Global North spend most of their time indoors in the built environment, especially their homes. Indoor air pollution has therefore become a major health concern, particularly in urban environments. Both exposure to poor-quality indoor environments, as well as vulnerability to adverse effects on health and well-being, have a unique geography, varying socially, spatially and temporally. Yet to date, the measurement of indoor air quality is relatively technical in focus, failing to account for the ways in which indoor environments are complex and varied, shaped by the physical environment, housing stock, policies, household dynamics, incomes and cultural norms. This paper aims to better understand the complex social and spatial drivers of vulnerability to poor indoor air quality. This is done by establishing a conceptual framing and building a new classification of vulnerability to indoor air pollution at the neighbourhood scale across England and Wales. First, the paper builds three separate indices utilising a spectrum of open-source data: environmental, structural and human-related. Second, it uses cluster analysis to generate indoor air quality profiles for neighbourhoods, identifying key patterns and drivers of vulnerability that typify different geographies. The findings allow for national, regional and local comparisons of vulnerability, which can be useful to a diverse range of stakeholders to assess potential exposure and guide intervention. Moreover, it highlights the dramatically different relationship between structural, environmental and human dimensions of vulnerability, encouraging indoor air quality exposure to be understood from multiple perspectives, not solely focused on low incomes.

Keywords

Indoor air quality spatial inequalities cluster analysis urban analytics ambient environment

Introduction

Indoor air pollution is a substantial global health burden increasingly recognised as an issue of environmental and social justice (Pillarisetti et al., 2022). Indoor air pollution refers to a variety of airborne pollutants, including damp and mould, chemicals in materials and sprays, or pollutants from the combustion of solid fuels and human activity. Both exposure to poor-quality ambient environments and vulnerability to adverse effects on health and well-being are unevenly distributed, socially and spatially (Boeing et al., 2023; Robinson and Williams, 2024).

While inequalities in the effect of indoor air quality (IAQ) are known globally and are mainly associated with low-income countries (Rana et al., 2021), they also arise in cooler climatic regions across the Global North, either where housing is not fit for purpose, outdoor air quality is poor, or where efforts to improve energy efficiency may have unintended consequences (Petrou et al., 2022). In the United Kingdom (UK), outdoor air pollution exposure is estimated to result in 28,000–36,000 premature deaths each year (Ferguson et al., 2021). Despite people in the Global North typically spending an estimated 80–90% of their time indoors, socio-spatial inequalities in indoor environmental quality are relatively poorly understood (Ferguson et al., 2021; Spiru and Simona, 2017).

The geographies of poor IAQ are complex and multi-faceted (Bouzarowski and Robinson, 2022; Shurety, 2025; Walker et al., 2022). Structural inequalities in housing and income mean that some people are unable to access high-quality housing or make necessary improvements (Hernandez and Swope, 2019). National regulations shape local building quality and indoor environmental conditions (Bonderup and Middlemiss, 2023). Vulnerability factors related to age, health or disability mean that some people are disproportionately vulnerable to the negative impacts of indoor air pollution (Makri and Stilianakis, 2008). Meanwhile, climate and poor outdoor air quality are highly localised, shaped by urban form (Munir et al., 2020). Whilst poor indoor environments have been shown to disproportionately impact low income, deprived and overcrowded populations (Ferguson et al., 2021), newly refurbished energy-efficient properties can also promote poor indoor environments without careful design (Recart and Sturts Dossick, 2022).

Yet to date, efforts to measure IAQ are overwhelmingly technical (Saad et al., 2017; Wagdi et al., 2018), lacking incorporation of socio-economic context, and is often spatially and temporally limited. Furthermore, current regulations are largely universal, overlooking specific implications for locales and individuals. This gap in understanding translates into policy with proposed legislation yet to be accepted (Mulcahy, 2023). Indeed, homes are complex, shaped by the physical environment, policies, household dynamics, social relations and cultural norms. Meaningfully and sensitively mapping indoor environments is challenging (Thomson et al., 2017).

In this paper, we aim to improve our understanding of the complex socio-spatial drivers of vulnerability to poor IAQ, focusing on cooler climates in the Global North. We build a series of IAQ vulnerability indices for neighbourhoods across England and Wales, identifying key drivers and their socio-spatial distribution through cluster analysis.

The paper is structured as follows. We critically review existing research about poor IAQ and its drivers from diverse national contexts, foregrounding social and spatial inequalities and vulnerabilities associated. Our conceptual framework divides drivers into three interrelated groups: environmental, structural and human-related. These are operationalised with open-source data, allowing us to map their socio-spatial distribution and used in score building and clustering exercises. We identify eight distinctive groups of neighbourhoods with a unique combination of IAQ indices. We show how a focus on poor IAQ reveals new geographies of urban inequality, but also highlight an urban-rural continuum in likely drivers. We conclude with reflections on how our framework could be applied to understand vulnerability at a neighbourhood scale. While we provide some figures in this publication, interactive maps can be viewed through the project website https://ambient-vulnerability.com.

Poor indoor air quality and its socio-spatial drivers

There is a long history situating the importance of air quality in relation to human health. The importance of IAQ, however, was not realised until the late 20th century, during which an energy crisis led to new building standards, increasing air tightness in properties, leading new buildings to be affected by poor ventilation (Sundell, 2004). Health issues associated with IAQ increased (see Seguel et al., 2017) and also gave rise to new health issues; upper-respiratory irritative symptoms, headaches, fatigue and even rashes, often labelled as a Sick Building Syndrome (Joshi, 2008).

Over time, it has become clear that poor IAQ represents a major risk to people’s health and well-being. It can have a systematic effect on the respiratory, cardiovascular, gastrointestinal, nervous, and immune systems, and have a carcinogenic effect on human reproduction (Berglund et al., 1992). Moreover, for vulnerable populations, the accumulation of these symptoms and prolonged exposure to air pollution can be lethal and increase the likelihood of premature death. Ella Addoo Kissi-Debrah, who passed away in 2013 after serious respiratory failure, was the first person to have ‘air pollution’ noted as an official cause of death in her death certificate in the UK (Loffhagen, 2022). Despite awareness of mortality and morbidity concerns, the complexity of poor IAQ and its drivers make it challenging to address.

Understanding of indoor air pollution from the physical sciences perspective is relatively comprehensive. Indoor air pollution results from increased concentrations of air pollutants in indoor spaces: mainly chemical compounds of an organic or inorganic nature, or radiation (radon, ozone and others). Amongst the most common inorganic pollutants are carbon dioxide, nitrogen dioxide and formaldehyde, but also plastic particles and other inorganic materials. These are mainly sourced from indoor materials, vanishes, paints, sprays, and combustion such as burning coal or biomass. Organic pollutants are mainly allergens, endotoxins, bacteria, fungal spores and other organic materials.

Despite knowledge of these diverse pollutants, the causal pathways between drivers and the level of pollutants can be complex. For example, Mannan and Al-Ghamdi (2021, Table 1) compiles evidence of indoor contaminants, their sources and health implications, where one source can produce multiple pollutants. To provide a somewhat clearer overview of likely drivers of vulnerability to poor IAQ, we build on Pourkiaei and Romain (2023) grouping IAQ drivers into three categories related to: outdoor environment; building characteristics; and indoor conditions. We find this generalisation useful and expand it in Figure 1, dividing IAQ drivers into three groups; environmental, structural and human. Environmental sources relate to the outdoor environment of the dwelling; structural are those from the structure of the dwelling, whether surfaces or the building itself; and human sources originate from people’s activities in the home. The remainder of this section highlights evidence of environmental, structural and human drivers of IAQ, with a particular focus on case studies in the context of the Global North.

Figure 1.

Break down of indoor air quality drivers.

Environmental drivers

Across the current literature, environmental drivers are often considered the most important for IAQ (Ferguson et al., 2020, 2021), acting as the default for the indoor environment. Sources of outdoor pollution can be naturally occurring or humanmade (Penard-Morand and Annesi-Maesano, 2004). Naturally occurring pollution is any pollution resulting from meteorological and seasonal conditions, such as thunderbolts or pollen, but also radon and methane. For human-made pollution, we can consider any pollution resulting from human activity, whether from the combustion of solid fuels (heating and traffic), methane resulting from farming practices, or many other chemicals emitted by different human activities.

While naturally occurring air pollutants are often highly concentrated at the pollution source, for example, pollen is tied to green spaces, the spatial distribution of human-made pollutants relates to human activities. For example, increased levels of outdoor pollution are strongly correlated with population density (Borck and Schrauth, 2021). As such, there is often a duality of outdoor pollution sources and a strong association with urban patterns and forms.

Structural drivers

Acknowledgement of the built environment and housing characteristics, and their role in IAQ, is likely the second most important factor (Ferguson et al., 2020; Mannan and Al-Ghamdi, 2021; Zhang and Srinivasan, 2020). The ability of buildings to exchange air with the outdoor environment, through ventilation and heating, is of the primary concern (Ferguson et al., 2020; Mannan and Al-Ghamdi, 2021). However, the relationships between ventilation, heating and air exchange are far from simple, and interventions in the form of energy efficiency retrofitting add to this complexity, preventing air from escaping completely (Petrou et al., 2022; Spiru and Simona, 2017).

Amongst other important structural drivers, indoor materials, their interaction with indoor air, and how they age, can also affect IAQ (Ferguson et al., 2020; Mannan and Al-Ghamdi, 2021). Studies show that the older the property is the higher the risk of mould and dampness (Howden-Chapman et al., 2005; Sharpe et al., 2014), and while improving the condition of the house generally results in less risk of mould occurrence, some evidence suggests otherwise (Recart and Sturts Dossick, 2022). Historically, homes in the UK were built for ventilation, owing to coal being the primary source of heating in a property (Rudge, 2012). Improved air tightness that does not allow the house to ‘breathe’ naturally due to efficiency interventions is a significant predictor of mould occurrence and the prevalence of asthma (Shrubsole et al., 2014). This is problematic not just in regards to mould but any other contaminants that may occur in the indoor space, for example, the increased use of synthetic materials in indoor spaces (Jones, 1999). Air pollution concentrations are often higher in newly built or refurbished houses as a result of new furnishings (Gonzalez-Martın et al., 2021).

From the other building-related factors, evidence suggests that risk of mould is higher in detached and terraced properties (Howden-Chapman et al., 2005; Sharpe et al., 2014; Moularat et al., 2011). While the absence of mechanical ventilation is by far the most likely predictor of mould growth, having mechanical ventilation is insufficient if the occupants cannot afford to ventilate because of heating costs (Ginestet et al., 2020). Lastly, the size and shape of indoor spaces matter. In small spaces or multioccupancy buildings, the spread of pollutants is hard to control as individuals have fewer opportunities to move between common spaces (Andargie et al., 2019) and are more at risk of pollution from surrounding units (Ferguson et al., 2020, 2021). Moreover, multistory buildings experience heat exchange on a seasonal basis, carrying pollution upwards (Andargie et al., 2019; Spiru and Simona, 2017).

Human drivers

Human-related drivers are often the most difficult to identify, especially causal relationships. This is amplified by the difficulty of measuring air quality in individual households. Thus, the majority of the existing research is limited by spatial or temporal extent. Various human-related IAQ drivers are identified in the recent reviews, which can be grouped into three categories: human activity, demographic and socio-economic (Ferguson et al., 2020, 2021; Mannan and Al-Ghamdi, 2021; Pourkiaei and Romain, 2023; Zhang and Srinivasan, 2020) (Figure 1). Human activity refers to household practices that might generate pollutants including cooking, cleaning, smoking, but also simply breathing. We can also identify interactions between other dimensions and human activity. For example, heating or ventilation, part of the property and its structure, can be amplified by individual use. Such activities are shaped by cultural practices and norms tied to a geographic context (Bauer et al., 2021; Hitchings et al., 2015)

Population characteristics, such as age or preexisting health conditions, can act as a catalyst for indoor air pollution (Zhang and Srinivasan, 2020). There is plenty of evidence, however, to show that wider socio-economic characteristics are related to poor IAQ Ferguson et al. (2021). The systematic inequalities that deprived and low-income populations face as they are exposed to greater environmental hazards, mean they often have an increased susceptibility to poor health outcomes and less opportunity for health-promoting behaviours. An increase in the number of people in enclosed spaces causes increased CO₂ levels, especially without sufficient ventilation. Overcrowding also shapes indoor air pollution (Ferguson et al., 2020, 2021; Mannan and Al-Ghamdi, 2021). For example, in multi-family units, pollutants move from unit to unit, whilst residents have limited ability to make changes to the building itself (Spiru and Simona, 2017). Living in such a dwelling where there is also likely to be more occupants (Ferguson et al., 2020; Sharpe et al., 2014), where frequent washing and air drying are more common (Howden-Chapman et al., 2005), or occupants are not able to afford to use heating to the required level, means that occupants are less able to control pollutants.

Tenure has also been shown to shape indoor environmental quality. For example, mould is especially prevalent in private rentals, where residents lack housing rights and rely on landlords for repairs (Ambrose, 2015). In England, according to NHS (2022), 10% of privately rented dwellings and 5% of socially rented dwellings are affected by dampness or mould, compared to 2% of owner-occupied dwellings.

Data

To date, there is no publicly available dataset of socio-spatial variation to poor IAQ in households across England and Wales. Some data on household perceptions of indoor environmental quality are available from the English Housing Survey (2023); however, these are not sufficient for mapping or modelling IAQ spatially. A lack of IAQ data is understandable as such data collection would require extensive resources and a systematic approach applicable across time and space. Indoor environments are also often private and personal, complicating measurement (Robinson, 2019).

Nevertheless, other publicly available secondary data sets can be used to infer vulnerability to poor IAQ across England and Wales. This approach builds on well-established Indices of Multiple Deprivation (IMD) and social vulnerability (Cutter and Finch, 2008), where a collection of carefully selected influential variables are used to derive information about the likely distribution of deprivation or vulnerability at a neighbourhood scale.

We collect relevant secondary data for each dimension identified in our literature review: the outdoor environment, structure of the property, and human activity (Figure 1). We make use of public sources such as the National Atmospheric Emissions Inventory (NAEI, 2021), Energy Performance Certificates (EPC) (DLUHC, 2024), Census 2021 (ONS, 2021) and information on population smoking from the Office for National Statistics (ONS, 2022). Additionally, we combine Census information with EPCs to derive a measure of population floor-space density to represent indoor crowding.

We focus on Lower Super Output Areas (LSOA) to extract and aggregate the data, small-area geographies based on population distribution that comprise between 400 and 1200 households. A fairly granular scale representing neighbourhoods, LSOA provide enough geographic detail while keeping computational needs low. Although some datasets require minimal processing (e.g. the Census and ONS data exist in ready-made tables) others require more effort. Emissions for outdoor pollutants are provided in a 1 km² grid raster, for which we take the mean values of all raster cells touching and overlapping each LSOA. The 2021 smoking dataset is created by combining two randomised sample size surveys, only available at the Local Authority scale; thus, we disaggregate the data to LSOA. Although disaggregating data to smaller geographies can be problematic, as the aggregated value (most often mean) is representative for small geographies but can be misleading individually, this was the only data on smoking behaviour for the study area.

EPCs are published as property-scale records. First, we made sure to remove any duplicates and keep only the last available certificate for each property. Second, we excluded LSOA with less than 10 properties available so we could reasonably assume what an average property looks like. Lastly, we aggregated the EPC records to LSOA level using mean values, helping to overcome challenges around missing data (see Discussion and Conclusion).

In total, we collect or derive close to 80 variables across three dimensions, but only include 35 in the final score (see Supp Figure 2). Most are directly derived from the given dataset, except the variable capturing the floor-space population density, which EPC data and Census population estimates. EPC data includes the total floor area (FA) for each property, which we aggregated into an average (mean) for each LSOA. We then derived an average number of people in a household (NPH) from the Census (ONS, 2023). Thus, a simple formula for a floor-space density (FSP) in each area can be derived:

F S P = \frac{median F A}{median NPH}

(1)

Methods

Developing indices of vulnerability to poor indoor air quality

In comparison to small-area estimation methods, which apply modelled ground-truth information from the sample area to all other areas in a global setting, composite scores are often used where ground-truth data is unavailable. At their most basic, composite scores take the most relevant information (composite variables) available for all areas to infer information that is not available or immeasurable. A good example of the use of composite scores is the UK Consumer Price Index (CPI) (OECD, Organisation for Economic Co-Operation and Development, 2024) or the English or Welsh IMD (CDRC, 2019), used by the public as well as governmental institutions to make sense of, and interpret, complex processes at scale.

Although valuable critiques are made of the indexification of poverty (Kiely and Strong, 2023), composite indicators such as the IMD have been helpful in assessing inequality in resources and redistributing resources (Deas et al., 2003). Two main methods are commonly used for composite score building; traditional composite scores and dimensionality reduction methods such as Principal Component Analysis (PCA) or Factor Analysis (FA). The two methodologies are diametrically different in their assumption about the phenomena they aim to capture. While the traditional method assumes the output score is a cumulative vector of all composite variables, PCA assumes the output score is a vector hidden within the correlations between all composite variables.

There is debate about which methodology is the most appropriate to which situation (Nardo et al., 2005). PCA and FA can be powerful and easy to use but rely on correlations between variables and their linearity (Greco et al., 2019). Although one could use other algorithms such as Isomap, which allows for the non-linearity in composite variables (see Wolf and Knaap, 2019), dependency on correlations can be hard to control and understand. Traditional composite scores are widely used for composite score building; however, they often suffer from double counting (Greco et al., 2019) where two correlated variables are incorporated independently despite common variations. Weighting variables has also been shown to generate more reliable estimates Kane and Case (2004), however, only when the reliability of variables is known. Ultimately, equal weight or unweighted composite scores remain the most commonly used method (Greco et al., 2019; Kara et al., 2022).

As there is no comprehensive data on IAQ, we remain loyal to the traditional method of the unweighted composite score given by equation (2), where composite score CS is the summation of composite variables CV¹…CVⁿ. Beforehand, each composite variable is normalised, standardised and divided into deciles to create consistent scoring. In this way, we consider the relative positions of each LSOA in England and Wales, rather than absolute, considering each variable equal.

C S = C V^{1} + C V^{2} + C V^{3} \dots C V^{n}

(2)

Instead of forcing all collected variables into one composite score, we create a score for each dimension of IAQ of our conceptual framework: environmental, structural and human (Figure 2(a)–(c) and (e)–(f)).

Figure 2.

Indices of indoor air quality in England and Wales and Bristol; environmental (a) and (d), structural (b) and (e) and human (c) and (f). Visualise the results using the online map here: [https://ambient-vulnerability.co.uk/maps/clusters-and-indices-of-indoor-air-quality-vulnerability/].

Identifying socio-spatial patterns in vulnerability to poor indoor air quality using cluster analysis

Analysing patterns of poor IAQ across three dimensions can be challenging. We identify areas with similar patterns of the three IAQ indices using cluster analysis. Cluster analysis is a common classification exercise described as an ‘art of finding groups in data’ (Kaufman and Rousseeuw, 2009). It has been applied extensively to understanding spatial relations between air pollution, health and socio-economic characteristics (Padilla et al., 2013).

In this study, we use an agglomerative hierarchical clustering algorithm to identify distinctive IAQ patterns. The indices present patterns often spatially related to the level of urbanity; therefore, we also incorporate the Urban/Rural classification (ONS, 2011) from the latest Census into the cluster analysis. We identify eight distinctive groups of areas (Figure 3), each with a distinctive combination of IAQ dimensions (see Supp Figure 1). Each cluster type is unique; however, they are related to differing degrees, reflected in our colour scheme.

Figure 3.

Description of each IAQ cluster.

Mapping the geographies of vulnerability to indoor air quality

We map each index for England and Wales, and a case study of the city of Bristol and the surrounding region 2. The city of Bristol is chosen as an example as it has some of the highest levels of poor air quality in the country outside London, as well as distinctive housing challenges associated with high prices and insecure tenancies. Distinctive geographies emerge for each dimension of vulnerability: where the similarity between human and environmental dimensions exists, structural dimensions often show an opposing pattern. High values for the environmental index (Figure 2(a) and (d)) correlate with urbanity, as outdoor pollution is associated with spatially intensive polluting human activities, especially transport (Boeing et al., 2023). Compared to the environmental index, structural and human indexes are geographically spread over neighbourhoods. At first glance, the structural index (Figure 2(b) and (e)) is primarily rural, although variation within cities exists. The human index is concentrated in major urban conurbations, coastal towns and some remote rural areas (Figure 2(c) and (f)).

Accounts of outdoor air pollution have typically focused on urban environments due to the multiplicity of flows that run through urban space (Adey, 2013; Graham, 2015; Walker et al., 2022). The importance of urban typologies for indoor air pollution is captured in four of our cluster types (Figure 3 and 4). The most overwhelmingly urban cluster type (99% of LSOA classified as urban) is the ‘Urban extreme of environment and human vulnerability’ (C5). These neighbourhoods have the highest levels of vulnerability across human, and to a lesser extent, environmental dimensions. Here, IAQ is likely to be poorer owing to a range of population-related vulnerability factors that have structural drivers, including overcrowding, renting or deprivation Ferguson et al. (2021). By comparison, the ‘Urban extreme of environment vulnerability’ (C7) cluster (95% of LSOA) has relatively low vulnerability based on structural and human dimensions, yet ranks highest for the environmental dimension. These dynamics can be explained in part by urbanity, concentrated in city centres where high levels of outdoor pollutants infiltrate indoors Boeing et al. (2023).

Figure 4.

Indoor air quality vulnerability clusters for England and Wales.

The ‘Urban highs’ (C2) cluster also has a high level of urbanity (96% of LSOA), but vulnerability is relatively high across all three dimensions. Concentrated in dense parts of cities, the outdoor environment is likely to be polluting, but housing is also relatively old, enhancing vulnerability. These areas may also include homes that have secondary forms of heating including wood-burning stoves, increasingly popular amongst relatively affluent urban households (Brown et al., 2023). Growing levels of wood burning in cities – often for aesthetic reasons – have driven a rise in harmful PM2.5 and PM10 emissions (Wood et al., 2023).

Historically deprived areas on the outskirts of the cities or towns, including south Bristol, are part of the ‘Urban human vulnerability’ (C6). Based on human aspects, vulnerability is the second highest of any cluster type. However, here, housing acts as a protective factor to poor indoor air quality – neither very old nor very new – whilst modern heating systems lower risk of indoor pollution. Environmental vulnerability is also low, as areas are often peripheral to city centres.

Comparatively, for ‘Urban lows’ (C1), vulnerability is low across all dimensions. Still predominantly urban (94% of LSOA), these areas concentrate spatially are likely relatively affluent suburbs, made up of larger and owner-occupied properties, where outdoor pollutants are comparatively low.

Through evaluation of our three indices of vulnerability, distinctive dynamics also emerge in relatively rural areas. In many contexts in the Global North, rural vulnerability to poor IAQ risks being obscured by air pollution debates prioritising outdoor environments, and therefore dense urban environments (Hendryx et al., 2019). The most rural cluster ‘Urban-Rural extreme of housing vulnerability’ (C4) (51% of LSOA are urban) has the highest vulnerability based on housing of any cluster, but the lowest environment-based vulnerability. The cluster concentrates in rural areas where solid walled properties concentrate - often challenging to heat and prone to damp. Rural areas are more likely to be off the gas grid, necessitating solid fuels that promote indoor air pollution, including coal or wood (Roberts, 2020). However, the relative rurality of these areas means outdoor air pollution is low. For “Urban-rural house and human vulnerability” (C5), housing vulnerability is heightened, but to a lesser extent. The cluster spatially concentrates on the rural-urban fringe (54% of LSOA urban); therefore, environment-based vulnerability factors are slightly elevated.

Finally, the importance of human-related vulnerability for poor IAQ in some relatively rural communities should not be obscured. The “Urban-rural house and human vulnerability” (C5), cluster has a higher proportion of structural and human vulnerability. Where rural areas are classified as part of this cluster, remoteness, declining industries or peripherality from economic advantages of cities tends to enhance deprivation (Shucksmith et al., 2023), including remote parts of Cornwall, Northumberland and South Wales.

We also investigate the roles of specific variables in the clusters. In general the roles of variables in cluster composition are almost equal. This is mainly due to the chosen methodology which does not discriminate against any variables, and because most variables in each dimension are somewhat correlated. Nevertheless, there are notable differences. First, although the ‘Urban highs’ (C2) is high based on the environment dimension, and ‘Rural extreme of housing vulnerability’ (C4) is low, radon concentrations are generally low for the former and high for the latter. Second, where the structural dimension is high – ‘Rural extreme of housing vulnerability’ (C4) and ‘Urban-rural housing vulnerability’ (C5) – there is a higher concentration of older properties and properties with fireplaces.

Discussion and conclusion

We develop a conceptual framework for defining, analysing and mapping socio-spatial vulnerability to poor IAQ, using the case of neighbourhoods in England and Wales. First, through building three indices of vulnerability (environmental, structural and human) and second, through clustering analysis of indices to characterise IAQ challenges likely to typify different neighbourhood types. The results show uneven, complex, and often contradictory, spatial distributions across all indices. Although some patterns are strongly related to urbanity, interactions between individual IAQ indices suggest the relations between poor indoor environments and urban and social structures are more complex. For example, we identify distinct areas across England and Wales where all indices are high, but areas where only structural IAQ or only human IAQ is elevated.

Characterising areas based on IAQ indices and identifying areas with similar characteristics has three-fold benefits. First, it provides a simple and understandable way of classifying vulnerability to poor IAQ at a national and local level for policy-making and practice. Second, it offers a way to compare differences between neighbourhoods, Local Authorities and regions to underpin decision-making and resource allocation to address poor IAQ. Third, it offers a benchmark for further quantitative research on IAQ.

The study comes with its own limitations. Firstly, it is important to note the limitations of the data used. EPCs are issued for new buildings or buildings that have come on the market during the last 16 years. However, if a property was not sold or rented in that time period, and the owner did not request an EPC, data will not be available. Therefore, missingness exists in the housing stock and energy use variables, potentially affecting reliability. A short investigation of this missingness shows that across England and Wales, most LSOA have between 50% and 80% of properties with a valid EPC (Figure 5(a)). There is some urban and rural differentiation in missingness with higher percentages of properties with a valid EPC in cities (Figure 5(b)). Importantly, the proportion of valid EPCs positively correlates with the IMD (Figure 5), meaning missingness is higher in areas with lower deprivation. Although aggregation of property-scale EPC data to LSOA helps to overcome some limitations by finding ‘average’ property characteristics, it might be ineffective if missing data have similarities (Buyuklieva et al., 2024). Future work should investigate this further to determine if the imputation of missing data is possible (Basiri and Brunsdon, 2022).

Figure 5.

Percentage of existing EPC given the number of households (from 2021 census) in and LSOA for England and Wales (a) and Bristol and its surroundings (b) accompanied by heatmap showing the relationship between percentage of existing EPCs and index of multiple deprivation (c).

Second, methodologically, we opt for a traditional composite score, prone to oversimplification and double counting underlying variation. This method is easier to interpret and can be used without any ground-truth data. If ground-truth data existed in future we could use PCA or Isomap to inspect the latent space of the collected variables. Ground-truth data would be also helpful in assessing the realities of air quality inequalities allowing us to better understand which variables or indices are important (Sarkar et al., 2024).

We have cross-tabulated each of the IAQ indices with the IMD (Figure 6) to see how the two indices relate to each other. It becomes clear that two of our IAQ indices, human and environment, have a positive association with underlying deprivation. This is not surprising for the human dimension as deprivation is incorporated into the index. However, the elevated environmental dimension suggests that the population in highly deprived areas is more likely to be exposed to high levels of pollution from the surroundings. On the other hand, the structural index shows a slightly negative relationship with the IMD; the population in areas that are more likely to be deprived are less likely to live in housing that would enhance indoor air pollution, either construction or heating type.

Figure 6.

Heatmaps showing the relationship between each IAQ index and index of multiple deprivation.

The disparity between the structural IAQ index and other dimensions is an important finding. It shows that alone, socio-economic status cannot be used for the prediction of poor IAQ. This resonates with previous studies (Deguen and Zmirou-Navier, 2010) highlighting the differential relationship between ambient air quality and factors such as deprivation, land use and urbanity. While deprivation can provide an initial indication of the vulnerability to poor IAQ in urban areas, the distribution of support and resources solely based on deprivation could overlook the socio-spatial complexity of IAQ challenges along the rural-urban continuum.

Supplemental Material

Supplemental Material - Indoor air quality vulnerability along the urban-rural continuum: A neighbourhood classification of exposure for England and Wales

Supplemental Material for Indoor air quality vulnerability along the urban-rural continuum: A neighbourhood classification of exposure for England and Wales by Lenka Hasova, Caitlin Robinson and Lin Zhang in Journal of Environment and Planning B: Urban Analytics and City Science.

Footnotes

Acknowledgements

We would like to thank Helen Stockton, Maya Fitchett and Niamh Storey from National Energy Action (NEA) for their valuable input during the writing of this paper. We would also like to thank Boyana Buyuklieva and Adam Dennett for their advice about the use of EPC data.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the United Kingdom Research and Innovation (UKRI) Future Leaders Fellowship grant Mapping Ambient Vulnerabilities MR/V021672/2.

ORCID iDs

Lenka Hasova

Caitlin Robinson

Data Availability Statement

All data is available under the Open Government License (OGL). Code and data to replicate the analysis is avauilable at: [].

Supplemental Material

Supplemental material for this article is available online.

Lenka Hasova is a postdoctoral researcher at the University of Bristol, where she recently completed her PhD in Human Geography as part of the Quantitative Spatial Science Research Group. Lenka researches and teaches widely across Geographic Data Science techniques. Her PhD research focuses on concepts and applications of spatial interaction models as networks.

Caitlin Robinson is an Senior Academic Fellow and Senior Proleptic Lecturer at the University of Bristol. Caitlin is a quantitative human geographer whose previous research has focused on understanding the causes and consequences of different forms of spatial inequality, especially related to energy and infrastructure. She is a Fellow of The Alan Turing Institute and co-lead of Bristol's Quantitative Spatial Science Research Group.

Lin Zhang is a postdoctoral researcher at The University of Manchester. Lin is interested in energy geographies and applying quantitative and qualitative techniques to understand the social and spatial distribution of energy poverty. Lin completed her PhD at the University of Leeds exploring energy poverty in the context of China.

References

Adey

(2013) Air/atmospheres of the megacity. Theory, Culture & Society 30(7–8): 291–308.

Ambrose

(2015) Improving energy efficiency in private rented housing: why don’t landlords act? Indoor and Built Environment 24(7): 913–924.

Andargie

Touchie

O’Brien

(2019) A review of factors affecting occupant comfort in multi-unit residential buildings. Building and Environment 160: 106182.

Basiri

Brunsdon

(2022) Missing data as data. Patterns 3(9): 100587.

Bauer

Moller

Gill

, et al. (2021) When energy efficiency goes out the window: how highly insulated buildings contribute to energy-intensive ventilation practices in Germany. Energy Research & Social Science 72: 101888.

Berglund

Brunekreef

Knoppe

, et al. (1992) Effects of indoor air pollution on human health. Indoor Air 2(1): 2–25.

Boeing

Pilgram

(2023) Local inequities in the relative production of and exposure to vehicular air pollution in Los Angeles. Urban Studies 60(12): 2351–2368.

Bonderup

Middlemiss

(2023) Mould or cold? Contrasting representations of unhealthy housing in Denmark and England and the relation to energy poverty. Energy Research & Social Science 102: 103176.

Borck

Schrauth

(2021) Population density and urban air quality. Regional Science and Urban Economics 86: 103596.

10.

Bouzarowski

Robinson

(2022) Injustices at the air–energy nexus. Environment and Planning F 1(2–4): 168–186.

11.

Brown

Huebner

Global Action Plan, and Impact on Urban Health (2023) Relight my Fire? Investigating the True Cost of Woodburning Stoves. London, UK: Impact on Urban Health.

12.

Buyuklieva

Oleron-Evans

Bailey

Dennett

, et al. (2024) Variations in domestic energy efficiency by property, neighbourhood and local authority type: where are the largest challenges for the net-zero transition of the UK’s residential stock? Frontiers in Sustainability 5: 1329034.

13.

CDRC (2019) English Indices of Deprivation (IMD) 2019 Data Pack. Leeds, UK: Consumer Data Research Centre.

14.

Cutter

Finch

(2008) Temporal and spatial changes in social vulnerability to natural hazards. Proceedings of the National Academy of Sciences of the United States of America 105(7): 2301–2306.

15.

Deas

Robson

Wong

, et al. (2003) Measuring neighbourhood deprivation: a critique of the index of multiple deprivation. Environment and Planning C: Government and Policy 21(6): 883–903.

16.

Deguen

Zmirou-Navier

(2010) Social inequalities resulting from health risks related to ambient air quality—a European review. European Journal of Public Health 20(1): 27–35.

17.

DLUHC (2024) Energy Performance of Buildings Data England and Wales. London, UK: Department for Levelling Up, Housing and Communities.

18.

English Housing Survey (2023)

Ministry of Housing, Communities and Local Government, Version Number. 4th Release.

19.

Ferguson

Taylor

Davies

, et al. (2020) Exposure to indoor air pollution across socio-economic groups in high-income countries: a scoping review of the literature and a modelling methodology. Environment International 143: 105748.

20.

Ferguson

Taylor

Zhou

, et al. (2021) Systemic inequalities in indoor air pollution exposure in London, UK. Buildings & cities 2(1): 425–448.

21.

Ginestet

Aschan-Leygonie

Bayeux

, et al. (2020) Mould in indoor environments: the role of heating, ventilation and fuel poverty. A French perspective. Building and Environment 169: 106577.

22.

Gonzalez-Martın

Kraakman

NJR

Perez

, et al. (2021) A state–of–the-art review on indoor air pollution and strategies for indoor air pollution control. Chemosphere 262: 128376.

23.

Graham

(2015) Life support: the political ecology of urban air. City 19(23): 192–215.

24.

Greco

Ishizaka

Tasiou

, et al. (2019) On the methodological framework of composite indices: a review of the issues of weighting, aggregation, and robustness. Social Indicators Research 141(1): 61–94.

25.

Hendryx

Higginbotham

Ewald

, et al. (2019) Air quality in association with rural coal mining and combustion in new South Wales Australia. Official Journal of the American Rural Health Association and the National Rural Health Care Association Health Association 35(4): 518–527.

26.

Hernandez

Swope

(2019) Housing as a platform for health and equity: evidence and future directions. American Journal of Public Health 109(10): 1363–1366.

27.

Hitchings

Waitt

Roggeveen

, et al. (2015) Winter cold in a summer place: perceived norms of seasonal adaptation and cultures of home heating in Australia. Energy Research & Social Science 8: 162–172.

28.

Howden-Chapman

Saville-Smith

Crane

, et al. (2005) Risk factors for mold in housing: a national survey. Indoor Air 15(6): 469–476.

29.

Jones

(1999) Indoor air quality and health. Atmospheric Environment 33: 4535–4564.

30.

Joshi

(2008) The sick building syndrome. Indian Journal of Occupational and Environmental Medicine 12(2): 61–64.

31.

Kane

Case

(2004) The reliability and validity of weighted composite scores. Applied Measurement in Education 17(3): 221–240.

32.

Kara

Valentin

Mainz

, et al. (2022) Composite measures of quality of health care: evidence mapping of methodology and reporting. PLoS One 17(5): e0268320.

33.

Kaufman

Rousseeuw

(2009) Finding Groups in Data: An Introduction to Cluster Analysis. Hoboken, NJ: John Wiley & Sons.

34.

Kiely

Strong

(2023) The indexification of poverty: the covert politics of small-area indices. Antipode 55(6): 1758–1780.

35.

Loffhagen

(2022) What is Ella’s law? Human rights bill combatting air pollution heads to commons. Evening Standard. Available at: https://www.standard.co.uk/news/uk/ellas-law-toxic-air-pollution-bill-human-rights-ella-kissi-debrah-environment-b1044488.html, In press.

36.

Makri

Stilianakis

(2008) Vulnerability to air pollution health effects. International Journal of Hygiene and Environmental Health 211(3): 326–336.

37.

Mannan

Al-Ghamdi

(2021) Indoor air quality in buildings: a comprehensive review on the factors influencing air pollution in residential and commercial structure. International Journal of Environmental Research and Public Health 18(6): 3276.

38.

Moularat

Hulin

Robine

, et al. (2011) Airborne fungal volatile organic compounds in rural and urban dwellings. The Science of the Total Environment 409(11): 2005–2009.

39.

Mulcahy

(2023) A bill before parliament for the right to breath clean air ProQuest. BMJ 380: 224.

40.

Munir

Mayfield

Coca

, et al. (2020) Analysis of air pollution in urban areas with airviro dispersion model—a case study in the city of Sheffield, United Kingdom. Atmosphere 11(3): 285.

41.

NAEI (2021) UK National Atmospheric Emissions Inventory for 2021. London, UK: NAEI.

42.

Nardo

Saisana

Saltelli

, et al. (2005) Tools for Composite Indicators Building. Brussels, Belgium: Joint Research Centre, European Commission.

43.

NHS (2022) Can Damp and Mould Affect my Health? London, UK: NHS.

44.

OECD, Organisation for Economic Co-Operation and Development (2024) Prices Inflation (CPI) - OECD Data. Paris, France: Organisation for Economic Co-Operation and Development.

45.

ONS (2011) 2011 Rural/Urban Classification. London, UK: Office for National Statistics.

46.

ONS (2021) Nomis; 2021 Census. London, UK: Office for National Statistics.

47.

ONS (2022) Adult Smoking Habits in the UK: 2021. London, UK: Office for National Statistics.

48.

ONS (2023) Ts001: Number of Usual Residents in Households and Communal Establishments: Census 2021. London, UK: Office for National Statistics.

49.

Padilla

Deguen

Lalloue

, et al. (2013) Cluster analysis of social and environment inequalities of infant mortality. A spatial study in small areas revealed by local disease mapping in France. The Science of the Total Environment 454: 433–441.

50.

Petrou

Hutchinson

Mavrogianni

, et al. (2022) Home energy efficiency under net zero: time to monitor UK indoor air. BMJ 377: e069435.

51.

Pillarisetti

Chowdhury

(2022) Indoor air pollution and health: bridging perspectives from developing and developed countries. Annual Review of Environment and Resources 47: 197–229.

52.

Pourkiaei

Romain

A-C

(2023) Scoping review of indoor air quality indexes: characterization and applications. Journal of Building Engineering 75: 106703.

53.

Penard-Morand

Annesi-Maesano

(2004) Air pollution: from sources of emissions to health effects. Breathe 1(2): 108–119.

54.

Rana

Islam

Khan

, et al. (2021) Association between household air pollution and child mortality in Myanmar using a multilevel mixed-effects poisson regression with robust variance. Scientific Reports 11: 12983.

55.

Recart

Sturts Dossick

(2022) Hygrothermal behavior of post-retrofit housing: a review of the impacts of the energy efficiency upgrade strategies. Energy and Buildings 262: 112001.

56.

Roberts

(2020) Warming with wood: exploring the everyday heating practices of rural off-gas households in Wales. Energy Policy 142: 111451.

57.

Robinson

(2019) Energy poverty and gender in England: a spatial perspective. Geoforum 104: 222–233.

58.

Robinson

Williams

(2024) Ambient Vulnerability. Global Environmental Change 84. Available at: https://doi.org/10.1016/j.gloenvcha.2024.102801

59.

Rudge

(2012) Coal fires, fresh air and the hardy British: a historical view of domestic energy efficiency and thermal comfort in Britain. Energy Policy 49: 6–11.

60.

Saad

Shakaff

AYM

Saad

ARM

, et al. (2017) Development of indoor environmental index: air quality index and thermal comfort index. AIP Conference Proceedings 1808(1): 020043.

61.

Sarkar

Cottineau-Mugadza

Wolf

(2024) Spatial inequalities and cities: a review. Environment and Planning B: Urban Analytics and City Science 51: 23998083241263422.

62.

Seguel

Merrill

Seguel

, et al. (2017) Indoor air quality. American Journal of Lifestyle Medicine 11(4): 284–295.

63.

Sharpe

Thornton

Osborne

(2014) Modifiable factors governing indoor fungal diversity and risk of asthma. Clinical and Experimental Allergy: Journal of the British Society for Allergy and Clinical Immunology 44(5): 631–641.

64.

Shrubsole

Macmillan

Davies

, et al. (2014) 100 unintended consequences of policies to improve the energy efficiency of the UK housing stock. Indoor and Built Environment 23(3): 340–352.

65.

Shucksmith

Glass

Chapman

, et al. (2023) Rural Poverty Today. Bristol, UK: Policy Press.

66.

Shurety

(2025) Moldy homes: Toxicity, race, and the geographies of domestic mold. Progress in Environmental Geography 4(1): 3–23. Available at: https://doi.org/10.1177/27539687241307956

67.

Spiru

Simona

(2017) A review on interactions between energy performance of the buildings, outdoor air pollution and the indoor air quality. Energy Procedia 128: 179–186.

68.

Sundell

(2004) On the history of indoor air quality and health. Indoor Air 14(s7): 51–58.

69.

Thomson

Bouzarovski

Snell

(2017) Rethinking the measurement of energy poverty in Europe: a critical analysis of indicators and data. Indoor + Built Environment: The Journal of the International Society of the Built Environment 26(7): 879–901.

70.

Wagdi

Tarabieh

Zeid

MNA

(2018) Indoor air quality index for preoccupancy assessment. Air Quality, Atmosphere & Health 11(4): 445–458.

71.

Walker

Booker

Young

(2022) Breathing in the polyrhythmic city: A spatiotemporal, rhythmanalytic account of urban air pollution and its inequalities. Environment and Planning C: Politics and Space 40(3): 572–591.

72.

Walker

Booker

Young

(2022) Breathing in the polyrhythmic city: a spatiotemporal, rhythmanalytic account of urban air pollution and its inequalities. Environment and Planning C: Politics and Space 40: 239965442094887.

73.

Wolf

Knaap

(2019) Learning geographical manifolds: a kernel trick for geographical machine learning. SocArXiv. Available at: https://osf.io/preprints/socarxiv/75s8v/, In preparation.

74.

Wood

Styring

, et al. (2023) Perceptions of accountability and trust in the regulatory governance of wood burning stove sustainability: survey evidence from the post-Brexit UK. Energy Policy 177: 113549.

75.

Zhang

Srinivasan

(2020) A systematic review of air quality sensors, guidelines, and measurement studies for indoor air quality management. Sustainability 12(21): 9045.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB