The spatial-temporal effect of air pollution on individuals’ reported health and its variation by ethnic groups in the United Kingdom: A multilevel longitudinal analysis

doi:10.21203/rs.3.rs-2145251/v1

Download PDF

Research Article

The spatial-temporal effect of air pollution on individuals’ reported health and its variation by ethnic groups in the United Kingdom: A multilevel longitudinal analysis

https://doi.org/10.21203/rs.3.rs-2145251/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 16 May, 2023

Read the published version in BMC Public Health →

You are reading this latest preprint version

Background

Air pollution affects the individuals’ health negatively; though it is unclear whether this effect is stronger for ethnic minorities compared to the rest of population. This study uses longitudinal data to investigate the spatial-temporal effect of air pollution on individuals’ reported health and its variation by ethnicity in the United-Kingdom (UK).

Methods

Longitudinal individual-level data from Understanding Society: the UK Household Longitudinal Study including 67,982 adult individuals with 404,264 repeated responses over 11years (2009–2019) were utilized and were linked to yearly concentrations of NO₂, SO₂, and particulate-matter (PM10, PM2.5) pollution once at the local authority and once at the census Lower Super Output Area (LSOA) of residence for each individual. This allows for analysis at two geographical scales over time. The association between air pollution and individuals’ health (Likert scale: 1–5, Excellent to poor) and its variation by ethnicity was assessed using three-levels mixed-effect linear models. The analysis distinguished between spatial (between areas) and temporal (across time within each area) effects of air pollution on health.

Results

Higher concentrations of NO₂, SO₂, PM10, and PM2.5 pollution were associated with poorer health. Decomposing air pollution into between (spatial: across local authorities or LSOAs) and within (temporal: across years within each local authority or LSOA) effects, showed a significant between effect for NO₂ and SO₂ pollutants at both geographical scales, while a significant between effect for PM10 and PM2.5 was shown only at the LSOAs level. No significant within effects were detected at either geographical level. Indian, Pakistani/Bangladeshi, Black/African/Caribbean and other ethnic groups and non-UK-born individuals reported poorer health with increasing concentrations of NO₂, SO₂, PM10, and PM2.5 pollutants in comparison to the British-white and UK-born individuals.

Conclusions

Using longitudinal data on individuals’ health linked with air pollution data at two geographical scales (coarse local authorities and detailed LSOAs), this study supports the presence of a spatial-temporal association between air pollution and poor self-reported health, which is stronger for ethnic minorities and foreign-born individuals in the UK, partly explained by location-specific differences. Air pollution mitigation is necessary to improve the individuals’ health, especially for ethnic minorities who are affected the most.

Air pollution

health

ethnicity

spatial-temporal

longitudinal

United Kingdom.

Recent global environmental debates have been focused on the issue of air pollution and its impact on human health (1, 2). Literature has shown an association between air pollution and elevated risks for mortality, clinical prescriptions, doctor visits, and hospital admissions for a range of acute and chronic health conditions, including cancer, cardiovascular and respiratory diseases (2–4). For example, in Belgium, a 3.5% increase in cardiovascular hospital admissions and a 4.5% increase in ischemic stroke hospital admissions were reported for every 10 µg/m³ increase in nitrogen dioxide (NO₂) pollution (5). In Montreal-Canada, poor self-rated health has been associated with increased exposure to Particulate matter with a diameter ≤ 2.5 µm (PM2.5) pollution (6). In the United-Kingdom (UK), the Committee on the Medical Effects of Air Pollution (COMEAP) has published a series of reports assessing the impacts of long-term exposure to air pollution on health and mortality. In those reports, PM2.5 was found to be associated with all-cause, cardiopulmonary, and lung cancer mortality (7).

The impact of air pollution on health is complex and is affected by a number of social, economic, individual, contextual, and environmental factors (2, 4). Gender, age, poverty, socioeconomic insecurity, educational attainment, marital status, household size and condition, occupation type and level, and income are among the socioeconomic factors affecting the association between air pollution and health (2, 4, 8–10). Pre-existing comorbidities, individual lifestyle habits (e.g., smoking, exercise, and alcohol consumption), contextual factors (e.g., neighbourhood condition, urbanicity, and population density), and environmental factors (e.g., the season, temperature, relative humidity, rainfall, and wind) also affect the individuals’ exposure to ambient air pollution and its associated illnesses (2, 4, 10–12). For example, the effect of air pollution on poor self-reported health in France (13) and in Germany (14) was exacerbated with increased socioeconomic insecurity, being unemployed, and living in more deprived areas.

Investigating the effect of air pollution on health by key population and socio-demographic subgroups would help in identifying the population sub-groups most affected by air pollution for specific intervention measures. In this context, literature has been focused on examining the effect of air pollution on health by gender, age, education, socioeconomic position, and deprivation. However, research on the effect of air pollution on health by ethnicity and migration (i.e. being a foreign-born individual) is still lacking in European countries and the UK as per a recent published systematic scoping literature review (4). Most of the research about the effect of air pollution on health by ethnicity was conducted in the U.S. (15–17), which is characterized by a different ethnic composition and structure than Europe; and to the best of our knowledge, only one study investigated the effect of air pollution on respiratory-asthma health in adolescents by ethnic groups in the UK (18). In this study, Astell-Burt et al. found that despite the higher concentrations of air pollution at the place of residence, ethnic minorities did not show lower lung function than the rest of the population. They even observed lower prevalence of asthma among some ethnic minority groups compared to the White UK group, even though ethnic minorities lived in more polluted regions (18).

Ethnicity forms an important topic in the health literature (19, 20) and might be an important effect modifier in the association between air pollution and health. Ethnic minorities often report poorer health compared to the rest of the population (20–24). Literature from the UK has shown that people from Pakistani and Bangladeshi origins tend to have the poorest reported health followed by people from African/Caribbean and Indian origins (20, 24–26). This is because ethnic minorities, in general, tend to occupy lower socioeconomic status and live in more deprived ethnic concentration communities with poor housing conditions (20, 27–29). In contrast, foreign-born individuals or migrants in the UK tend to have better health and lower rates of mortality compared to the native-born population which is linked to the “Healthy migrant effect” theory (30, 31). This theory indicates that healthier, more educated, wealthier, and better job market suited individuals are the ones who have the capability of migrating from their countries of origin to high-income countries such as the UK (25, 30).

In addition to the lack of studies on the effect of air pollution on health by ethnicity and country of birth, application of innovative study designs that can differentiate between spatial and temporal components are also lacking. Previous studies have examined the short-term effects of air pollution on health, mortality, and hospital admissions using time-series, case-crossover, or ecological designs (3, 4). The long-term effect of air pollution on health outcomes was also assessed in the literature using cohort designs (3, 4). However, to our knowledge, no study has used a between-within longitudinal study design to examine the spatial-temporal effect of long-term exposure to air pollution on health. Applying a between-within longitudinal design would involve a decomposition of the air pollution effect on health into between (computing the average air pollution concentration for each geographical area across the follow-up time) and within (computing the annual deviation in air pollution concentrations from the area-average between concentration for each unit of time within the follow-up period) (32). This approach would allow us to distinguish between the spatial (between) and the temporal (within) effects of air pollution on individuals’ health.

Finally, studies that link air pollution data to individual-level data at different geographical scales are lacking. Assessing the effect of air pollution on individuals’ health at two geographical scales (e.g., coarse local authorities and detailed census output areas) would allow for a comparison of the results between the two scales and a detailed exploration of the local-contextual patterns. Whilst analysis at a finer spatial scale using census output areas would add robustness to the results, analysis at a coarser scale using local authorities would better inform local mitigation approaches by providing overall area estimates for the local authority boards.

Given all these limitations, we will be using two geographical scales (coarse local authorities and detailed census output areas) to assess the spatial-temporal (between-within) effects of air pollution on individuals’ reported health and how this effect varies by ethnicity and country of birth (being a foreign-born versus not) in the UK. Thus, the objectives of this study are as follows:

To investigate the long-term (11 years) effect of exposure to NO₂, sulphur dioxide (SO₂), Particulate matter with diameter ≤ 10 µm (PM10), and PM2.5 air pollutants on individuals’ reported health – what we refer to as the overall pollution effect.

To investigate the between (spatial - average pollutant concentration across the follow-up time for each geographical area) and within (temporal - annual deviation in the pollutant concentration from the average area concentration for each time unit of the follow-up period) effects of air pollution on individuals’ reported health.

To include an interaction term between air pollution and ethnicity and between air pollution and country of birth to assess how the effect of air pollution on health varies by different ethnic groups and by being a foreign-born individual.

To support the two-scale analysis, the analysis to meet the above objectives will be performed twice, once on the dataset which includes the linkage of air pollution to the individual-level data at the local authority level and once on the dataset which includes the linkage of air pollution to the individual-level data at the census output areas level.

2.1. Study design and population

A longitudinal panel design was employed using individual-level data from “Understanding Society: The UK Household Longitudinal Study” (33). Understanding Society is a rich longitudinal dataset consisting of 10 data collection waves/panels that span from 2009 up to 2020 with around 40,000 households recruited at wave 1 from the four nations of the UK: England, Wales, Scotland, and Northern Ireland. It involves two main surveys: the youth survey which is filled by young people (aged 10 to 15) and the adult survey which is filled by individuals aged 16 and above (33).

The dataset includes information on the socio-demographic characteristics of individuals (e.g., age, gender, marital status, educational attainment, occupation, perceived financial situation, ethnicity, and country of birth) and on the individuals’ self-reported health, wellbeing, smoking status, as well as the local authority/council area and census Lower Super Output Areas (LSOAs) where households are located. Individuals recruited in the Understanding Society study are visited each year to collect information on changes to their household and individual circumstances (33).

The sample design of the Understanding Society main survey is made up of four components: 1) the large General Population Sample (around 26,000 recruited households at wave 1, 2009–2010); 2) the Ethnic Minority Boost Sample (around 4000 recruited households at wave 1, 2009–2010); 3) the former British Household Panel Survey sample (at wave 1, 2010); and 4) the Immigrant and Ethnic Minority Boost Sample (around 2,500 recruited households at wave 6, 2015) (33). Further information on the Understanding Society study design is described elsewhere (34, 35).

For this study, we utilized individual-level data on 67,982 individuals with 404,264 repeated responses (at least 2 repeated responses per individual) across 10 data collection waves over 11years (2009–2019) from the adult survey (age: 16+) of the Understanding Society data. It is worth noting that the initial adult survey of the Understanding Society data involved a total of 87,045 individuals with 444,181 repeated responses and that 39,917 observations were deleted due to the reasons summarised in Fig. 1.

2.2. Variables and measurements

2.2.1. Self-reported health

Individuals’ self-reported health which asks how individuals perceive their health in general is assessed on a 5-point Likert scale: 1 = excellent, 2 = very good, 3 = good, 4 = fair, 5 = poor. Out of the total 404,264 general health observations, 105 (0.03%) were missing and were filled out using another health indicator: satisfaction with health, which showed strong correlation (Pearson’s coefficient = 0.53) with the general health outcome. Satisfaction with health is measured on a 7-point Likert scale (completely satisfied, mostly satisfied, somewhat satisfied, neither satisfied nor dissatisfied, somewhat dissatisfied, mostly dissatisfied, and completely dissatisfied). Therefore, completely satisfied was coded to excellent health, mostly satisfied was coded to very good health, somewhat satisfied was coded to good health, neither satisfied nor dissatisfied and somewhat dissatisfied were coded to fair health, and mostly dissatisfied and completely dissatisfied were coded to poor health.

It should be noted that individuals’ self-reported health was chosen as the main outcome in this study due to its ability of capturing the health status from the perspective of the individual and it is considered as a reliable measure of health given the high observed correlations between self-reported health and objective health measures (e.g. mortality and hospital admissions) in the literature (36–38).

2.2.2. Air pollution

We downloaded yearly air pollution data that combine all sources of air pollution including road traffic and industrial/combustion processes for NO₂, SO₂, PM10, and PM2.5 pollutants from the “Department for Environment Food and Rural Affairs” online database (39). These are raster data of mean annual concentrations of pollutants measured in µg/m³ up to the year 2019, estimated using air dispersion models at a spatial resolution of 1x1 km and projected using the UK National Grid (39).

For each of the 391 local authorities/council areas in the UK, we computed the average concentration of NO₂, SO₂, PM10, and PM2.5 pollution from all the centroids of the 1x1 km raster cells that intersected/fell within the boundaries of the respective local authorities/council areas for each year from 2009 up to 2019. These average concentrations of air pollution were then linked to the “Understanding Society” data using the individuals’ local authority of residence for each year of observation per individual between 2009 and 2019, inclusive.

To minimize exposure bias and establish more robust results, we also linked the 1x1 km raster air pollution data to the Understanding Society data at the level of Lower Super Output Areas (LSOAs; data zones for Scotland and Super Output Areas for Northern Ireland), a finer geographical scale, for each individual and each year of follow-up (2009–2019). LSOAs are used to decompose England and Wales based on the population size into areas with a minimum population size of 1000 people and are the lowest level of geography offered by the Understanding Society dataset. The LSOAs in England and Wales are equivalent to data zones in Scotland and to Super Output Areas in Northern Ireland. For simplicity we refer to the joint LSOAs, data zones, and Super Output Areas as LSOAs. The linkage was done by computing the average air pollution concentration for each LSOA from all the 1x1 km raster squares that intersected with the respective LSOA. For example, if a certain LSOA intersected with only one raster square, it is given the value of air pollution of that square and if a certain LSOA intersected with two raster squares, it is given the average air pollution of those two squares, and so on so forth. Using these smaller spatial units, we ran our analysis at a smaller geographic scale than local authorities, which allowed us to explore local-contextual patterns of the effect of air pollution on health.

A map showing the local authorities in the UK (council areas in Scotland) and an enlarged subset of 20 local authorities in the southeast of the UK with an example of PM10 concentrations at 1x1 km grid for the year 2017 for Tower Hamlets local authority and its corresponding LSOAs was used to illustrate the process of air pollution linkages (Fig. 2).

2.2.3. Socio-demographic and lifestyle covariates

For this study, we selected a list of individual-level socio-demographic and lifestyle factors based on what is available in the Understanding Society data and based on the confounders considered by the air pollution-health literature (3, 4). These included age (coded as 16–18 and then in 5 years increments as 19–23; 24–28; 29–33; 34–38; 39–43; 44–48; 49–53; 54–58; 59–63; 59–63; 64–68; 69–73; 74–78; >78); gender (females versus males (Reference category)); ethnicity (Other-white, Pakistani/Bangladeshi, Indian, Black/African/Caribbean, mixed ethnicities, and other ethnicities versus British-white (Reference category)); country of birth (non-UK-born and no answer versus UK-born (Reference category)); marital status (living as a couple, single never married, divorced/separated, widowed, and no answer versus married (Reference category)); educational attainment (High school, lower education, other educational qualifications, and still a student versus university degree (Reference category)); occupation (Non-manual workers, manual workers, student/retired/not-working and no answer versus managers/professionals/employers (Reference category)); perceived financial situation (living difficultly and no answer versus living comfortably/doing alright (Reference category)); and smoking (smoker and no answer versus non-smoker (Reference category)) (42).

The question about smoking was not asked during wave 1 of data collection and during waves 3 and 4 for people above the age of 21 years old. Therefore, individual responses on smoking from wave 2 were used as a proxy for the smoking status in waves 1, 3, and 4 (42). This imputation is unlikely to deviate from the real smoking status scenario because the intraclass correlation coefficient (ICC) indicate a 97% similarity in the individual smoking responses across the data collection waves.

Finally, year dummies (calendar year: 2009–2019) were considered as a control for the time trend in our analysis following the approach of relevant studies (43, 44). Given that our study utilises yearly air pollution data, controlling for other temporal covariates considered by relevant literature such as seasonal trends (45, 46) was not possible.

2.3. Data analysis

Percentages were computed to describe the individuals’ socio-demographic and lifestyle factors for each wave (waves 1 to 10) of the Understanding Society sample. We also examined the correlation between NO₂, SO₂, PM10, and PM2.5 pollutants at the two geographical scales of local authorities and LSOAs using Pearson’s correlation coefficient. Given the high observed correlations between the pollutants (Pearson’s coefficient ≥ 0.7 (47); Table 2 and Table 3), the association of NO₂, SO₂, PM10, and PM2.5 pollutants with self-reported health was examined in separate regression models. However, a low to moderate correlation was observed between SO₂ and each of the other three pollutants which enabled the construction of bi-pollutant models adjusting the NO₂, PM10, and PM2.5 models for the SO₂ pollutant.

Intraclass correlation coefficients (ICCs) were computed to assess the homogeneity in the self-reported general health responses within individuals and household clusters. An ICC of more than 0.3 indicates the presence of fair to high homogeneity in the responses within the examined clusters across time (48). Given the presence of 65% homogeneity (ICC = 0.65; Table 4) within the responses of self-reported health for each individual across time, the mean of self-reported health was calculated from predictions of mixed-effects linear models which were adjusted for age in fixed effects and for the individual ID in random intercept.

To account for the study design which involves linkages of air pollution to individual data at two geographical scales (once at the LSOAs and once at the local authorities level) and to account for the repeated individuals’ responses across time, the association between self-reported health treated as a continuous Likert scale outcome and each of NO₂, SO₂, PM10, and PM2.5 pollutants was assessed using separate and bi-pollutant three-levels (repeated individual observations across time nested within local authorities or LSOAs) linear mixed effect models. These models were adjusted for the socio-demographic and lifestyle covariates and the year (2009–2019) dummies. In a supplementary analysis, we also demonstrate the association of self-reported health with each of the socio-demographic and lifestyle covariates (Additional file Table 1). It is worth noting that we did not account for the household clustering in the random intercept of the mixed effect models due to the low observed homogeneity in the self-reported health responses within each household cluster (ICC = 0.24; Table 4).

Table 1

Description of individual’s socio-demographic and lifestyle factors for each wave of the Understanding Society data (N = 404,264 surveys from 67,982 individuals)
		Wave1 (2009–2011)	Wave2 (2010–2012)	Wave3 (2011–2013)	Wave4 (2012–2014)	Wave5 (2013–2015)	Wave6 (2014–2016)	Wave7 (2015–2017)	Wave8 (2016–2018)	Wave9 (2017–2019)	Wave10 (2018–2019)
		N = 39,797	N = 50,728	N = 47,650	N = 45,236	N = 42,525	N = 37,116	N = 38,012	N = 36,017	N = 32,955	N = 30,505
		Percent	Percent	Percent	Percent	Percent	Percent	Percent	Percent	Percent	Percent
Gender	Male	45	45	46	46	46	46	46	46	45	45
Gender	Female	56	55	54	54	54	54	54	55	55	55
Age	Young (< 34)	27	27	26	26	26	26	25	24	23	21
	Middle age (34–58)	46	44	44	44	43	44	44	44	43	43
	Old (> 58)	27	29	30	30	31	30	31	33	34	35
Ethnicity	British-white	76	77	78	78	78	74	74	75	77	78
	Other-white	4	5	5	5	4	6	5	5	5	5
	Indian	4	3	3	3	3	4	4	4	4	4
	Pakistani/Bangladeshi	5	4	4	4	4	5	5	5	5	5
	Black/African/Caribbean	5	4	4	4	4	4	4	4	4	3
	Mixed ethnicities	2	1	1	2	2	2	2	2	2	2
	Other ethnicities	4	5	5	5	5	5	5	5	4	3
Country of birth	Born in the UK	83	67	67	67	68	67	67	67	68	69
	Not born in the UK	17	14	13	13	12	15	15	15	13	12
	No answer	0	20	21	20	20	18	18	18	19	19
Marital status	Married	53	53	53	52	52	53	53	53	54	55
	Living as a couple	11	11	11	11	11	11	11	10	10	10
	Widowed	6	6	6	6	6	6	6	6	6	6
	Divorced/separated	9	8	8	8	8	8	8	8	8	8
	Single never married	21	22	22	22	23	23	24	23	23	21
	no answer	0	0	0	0	0	0	0	0	0	0
Education	University degree	30	25	26	27	28	29	30	31	32	34
	High school degree	32	26	26	27	27	27	27	27	27	27
	Lower educational levels	1	1	1	1	1	1	1	1	1	1
	Other qualifications	30	40	40	38	37	36	35	35	34	33
	Still a student	7	7	7	7	7	7	7	6	6	5
Occupation	Managers/Professionals/employers	12	12	12	12	12	12	12	12	12	12
	Non-manual workers	26	26	26	26	26	27	27	27	26	26
	Manual workers	18	18	18	18	19	19	19	18	17	17
	Not applicable: Student/ retired/Not working	44	44	44	43	43	41	41	42	42	43
	No answer	0	0	1	1	0	1	1	1	2	3
Perceived financial situation	living comfortably/doing alright	56	57	57	59	60	65	67	70	69	70
	living difficultly	40	37	36	33	32	27	26	26	27	28
	no answer	4	6	8	8	8	8	7	5	3	3
Smoking status	non-smoker	70	74	69	68	75	71	79	82	84	85
	smoker	19	20	18	18	16	15	15	14	13	13
	no answer	11	6	13	14	8	14	7	4	3	2
Nation	England	84	77	76	77	77	79	79	79	78	79
	Wales	5	8	8	8	7	7	6	7	7	7
	Scotland	7	9	9	9	9	8	8	8	9	9
	Northern Ireland	4	7	7	7	6	6	6	6	7	6

In further analysis, we decomposed the overall effect of air pollution (linked at the local authority or LSOAs level) on health into between (spatial) and within (temporal) effects. Between effects (Eq. 1) were used to determine the spatial effect of air pollution by computing the mean of air pollution across the 11 years of follow-up (2009–2019) for each local authority and each LSOA. On the contrary, within effects (Eq. 2) were used to determine the temporal effect of air pollution by calculating the yearly air pollution deviation from the 11 years mean for each local authority and LSOA. Therefore, the multilevel linear mixed effect models were used to examine the overall (Eq. 3) effect of air pollution as well as the between and within effects (Eq. 4) of air pollution on self-reported health at two geographical scales (coarse local authorities and detailed LSOAs).

Finally, we incorporated into the mixed effect models an interaction term between ethnicity and each of NO₂, SO₂, PM10, and PM2.5 pollutants and between country of birth and each of the four pollutants to assess whether the association between air pollution and health varies between ethnic groups and by country of birth. Interaction terms were incorporated into the overall pollutant models (equations 5 and 6) and into the between-within models, each at a time. Coefficient plots were used to visualize the interaction analysis coefficients.

$${\text{B}\text{e}\text{t}\text{w}\text{e}\text{e}\text{n} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{\text{t}\text{i}\text{j}}=\stackrel{-}{{\text{o}\text{v}\text{e}\text{r}\text{a}\text{l}\text{l} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{\text{j}}}$$

$${\text{W}\text{i}\text{t}\text{h}\text{i}\text{n} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{\text{t}\text{i}\text{j}}={|\text{o}\text{v}\text{e}\text{r}\text{a}\text{l}\text{l} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{\text{t}\text{j}}-\stackrel{-}{{ \text{o}\text{v}\text{e}\text{r}\text{a}\text{l}\text{l} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{\text{j}}}|$$

Where i is the individual; t is the time in years; and j is the local authority or LSOA.

$${Y}_{{tij}}={\beta }_{0} + {U}_{0ij}+{U}_{0j} + {\beta 1\text{o}\text{v}\text{e}\text{r}\text{a}\text{l}\text{l} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 2\text{A}\text{g}\text{e}}_{tij}+ {\beta 3\text{G}\text{e}\text{n}\text{d}\text{e}\text{r}}_{tij}+ {\beta 4\text{E}\text{t}\text{h}\text{n}\text{i}\text{c}\text{i}\text{t}\text{y}}_{tij}+ {\beta 5\text{C}\text{o}\text{u}\text{n}\text{t}\text{r}\text{y} \text{o}\text{f} \text{b}\text{i}\text{r}\text{t}\text{h}}_{tij}+ {\beta 6\text{M}\text{a}\text{r}\text{i}\text{t}\text{a}\text{l} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 7\text{E}\text{d}\text{u}\text{c}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+{\beta 8\text{O}\text{c}\text{c}\text{u}\text{p}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 9\text{P}\text{e}\text{r}\text{c}\text{e}\text{i}\text{v}\text{e}\text{d}\text{f}\text{i}\text{n}\text{a}\text{n}\text{c}\text{i}\text{a}\text{l} \text{s}\text{i}\text{t}\text{u}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 10\text{S}\text{m}\text{o}\text{k}\text{i}\text{n}\text{g} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 11\text{Y}\text{e}\text{a}\text{r} \text{d}\text{u}\text{m}\text{m}\text{i}\text{e}\text{s}}_{ij}+ {\epsilon }_{tij}$$

$${Y}_{{tij}}={\beta }_{0} + {U}_{0ij}+{U}_{0j} + {\beta 1\text{B}\text{e}\text{t}\text{w}\text{e}\text{e}\text{n} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 2\text{W}\text{i}\text{t}\text{h}\text{i}\text{n} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 3\text{A}\text{g}\text{e}}_{tij}+ {\beta 4\text{G}\text{e}\text{n}\text{d}\text{e}\text{r}}_{tij}+ {\beta 5\text{E}\text{t}\text{h}\text{n}\text{i}\text{c}\text{i}\text{t}\text{y}}_{tij}+ {\beta 6\text{C}\text{o}\text{u}\text{n}\text{t}\text{r}\text{y} \text{o}\text{f} \text{b}\text{i}\text{r}\text{t}\text{h}}_{tij}+ {\beta 7\text{M}\text{a}\text{r}\text{i}\text{t}\text{a}\text{l} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 8\text{E}\text{d}\text{u}\text{c}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 9\text{O}\text{c}\text{c}\text{u}\text{p}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 10\text{P}\text{e}\text{r}\text{c}\text{e}\text{i}\text{v}\text{e}\text{d} \text{f}\text{i}\text{n}\text{a}\text{n}\text{c}\text{i}\text{a}\text{l} \text{s}\text{i}\text{t}\text{u}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 11\text{S}\text{m}\text{o}\text{k}\text{i}\text{n}\text{g} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 12\text{Y}\text{e}\text{a}\text{r} \text{d}\text{u}\text{m}\text{m}\text{i}\text{e}\text{s}}_{ij} + {\epsilon }_{tij}$$

$${Y}_{{tij}}={\beta }_{0} + {U}_{0ij}+{U}_{0j} + \beta 1\text{o}\text{v}\text{e}\text{r}\text{a}\text{l}\text{l} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}\times {\text{E}\text{t}\text{h}\text{n}\text{i}\text{c}\text{i}\text{t}\text{y}}_{tij} + {\beta 4\text{A}\text{g}\text{e}}_{tij}+ {\beta 5\text{G}\text{e}\text{n}\text{d}\text{e}\text{r}}_{tij}+ {\beta 6\text{C}\text{o}\text{u}\text{n}\text{t}\text{r}\text{y} \text{o}\text{f} \text{b}\text{i}\text{r}\text{t}\text{h}}_{tij}+ {\beta 7\text{M}\text{a}\text{r}\text{i}\text{t}\text{a}\text{l} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 8\text{E}\text{d}\text{u}\text{c}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 9\text{O}\text{c}\text{c}\text{u}\text{p}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 10\text{P}\text{e}\text{r}\text{c}\text{e}\text{i}\text{v}\text{e}\text{d}\text{f}\text{i}\text{n}\text{a}\text{n}\text{c}\text{i}\text{a}\text{l} \text{s}\text{i}\text{t}\text{u}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 11\text{S}\text{m}\text{o}\text{k}\text{i}\text{n}\text{g} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 12\text{Y}\text{e}\text{a}\text{r} \text{d}\text{u}\text{m}\text{m}\text{i}\text{e}\text{s}}_{ij} + {\epsilon }_{tij}$$

$${Y}_{{tij}}={\beta }_{0} + {U}_{0ij}+{U}_{0j} + \beta 1\text{o}\text{v}\text{e}\text{r}\text{a}\text{l}\text{l} \text{p}\text{o}\text{l}\text{l}\text{u}\text{t}\text{a}\text{n}\text{t} \text{c}\text{o}\text{n}\text{c}\text{e}\text{n}\text{t}\text{r}\text{a}\text{t}\text{i}\text{o}\text{n}\times {\text{C}\text{o}\text{u}\text{n}\text{t}\text{r}\text{y} \text{o}\text{f} \text{b}\text{i}\text{r}\text{t}\text{h}}_{tij} + {\beta 4\text{A}\text{g}\text{e}}_{tij}+ {\beta 5\text{G}\text{e}\text{n}\text{d}\text{e}\text{r}}_{tij}+ {\beta 6\text{E}\text{t}\text{h}\text{n}\text{i}\text{c}\text{i}\text{t}\text{y}}_{tij}+ {\beta 7\text{M}\text{a}\text{r}\text{i}\text{t}\text{a}\text{l} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 8\text{E}\text{d}\text{u}\text{c}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 9\text{O}\text{c}\text{c}\text{u}\text{p}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 10\text{P}\text{e}\text{r}\text{c}\text{e}\text{i}\text{v}\text{e}\text{d}\text{f}\text{i}\text{n}\text{a}\text{n}\text{c}\text{i}\text{a}\text{l} \text{s}\text{i}\text{t}\text{u}\text{a}\text{t}\text{i}\text{o}\text{n}}_{tij}+ {\beta 11\text{S}\text{m}\text{o}\text{k}\text{i}\text{n}\text{g} \text{s}\text{t}\text{a}\text{t}\text{u}\text{s}}_{tij}+ {\beta 12\text{Y}\text{e}\text{a}\text{r} \text{d}\text{u}\text{m}\text{m}\text{i}\text{e}\text{s}}_{ij} + {\epsilon }_{tij}$$

Where Y_tij is the health outcome for individuals i, in local authority or LSOA j at year t; β₁, β₂ …. β₁₂ are the slopes of fixed effects; β₀ is the fixed intercept; U_0ij is level 2 random intercept of individuals nested in local authorities or LSOAs; U_0j is level 3 random intercept of local authorities or LSOAs; ε_tij are the model residuals.

In a sensitivity analysis, we performed the same multilevel mixed-effects linear models to examine the overall and the between-within effects of air pollution (linked at the level of local authority and LSOAs) on self-reported health and how this effect of air pollution on health varies by ethnic groups and country of birth only for individuals recruited in wave 1 of the Understanding Society data.

We also recoded in a sensitivity analysis the self-reported health outcome measured on a 5-point Likert scale into a binary outcome (0 = Excellent/very good/good health; 1 = Fair/poor health) and repeated the same previously described analysis but using three-levels mixed-effect logit models.

Statistical analysis was conducted using STATA software (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP) and spatial pre-processing was conducted using ArcGIS Pro software. Regression results were reported in terms of coefficients and 95% confidence intervals (CIs) per 10 µg/m³ increase in air pollution for the multilevel mixed-effect linear models and in terms of odd ratios (ORs) and 95% CIs per 10 µg/m³ increase in air pollution for the multilevel mixed-effect logit models in a sensitivity analysis. Statistical significance was considered at a P-value of less than 0.05.

3.1. Description of individuals’ socio-demographic and lifestyle factors

This study included a total of 67,982 adult individuals (aged 16+) with 404,264 repeated responses across 10 waves/panels spread over 11years (2009–2019) of individuals’ follow-up. The average number of observations per individual was 5.95 (SD = 2.86) with a minimum of 2 observations per individual and the mean follow-up time was 5.53 (SD = 3.00) years.

Table 1 summarises the descriptive statistics for the individuals’ socio-demographic and lifestyle factors for each of the 10 waves/panels of the Understanding Society sample. In all the waves, most of the individuals were females, aged between 34 and 58 years old, were married, were non-manual workers (if working), had a comfortable/alright financial situation, were non-smokers, and lived in England. As for educational attainment, individuals were equally distributed between high school, university, and other educational qualifications in all the waves (Table 1).

Most of the individuals were born in the UK (83% in wave 1) and belonged to British-white ethnicity (76% in wave 1). The share of ethnic minorities in wave 1 was as follows: Other-white (4%), Indians (4%), Pakistani/Bangladeshi (5%), Black/African/Caribbean (5%), mixed ethnicities (2%), and other ethnicities (4%) (Table 1).

3.2. Description of air pollution

3.2.1. Description of air pollution at the LSOAs level

The mean of NO₂, SO₂, PM10, and PM2.5 pollutants across the 42,619 LSOAs in the UK for each year from 2009 through 2019 is summarised in Fig. 3. Fluctuations in air pollution were observed from one year to another with lower levels of pollution noted in the last 5 years of the observation window in comparison to previous years for all the four pollutants with an exception for the year 2016 (Fig. 3). The yearly concentrations of NO₂ decreased from 17.6 µg/m³ in 2009 to 13.4 µg/m³ in 2019. Similarly, the concentrations of PM10 and PM2.5 decreased from 15.5 µg/m³ in 2009 to 13.5 µg/m³ in 2019 and from 10.4 µg/m³ in 2009 to 8.6 µg/m³ in 2019, respectively. A decline in SO₂ concentration from 2.7 µg/m³ in 2009 to 1.4 µg/m³ in 2019 was also observed (Fig. 3).

Table 2 summarises the average concentrations of air pollutants across the 42,619 LSOAs in the UK and their respective correlations. A high correlation (Pearson’s coefficient ≥ 0.7) was noted between NO₂, PM10, and PM2.5 pollutants, which could be explained by the chemical reactions between particulate matter and NO₂ pollutants in the atmosphere (Table 2).

Table 2

Exposure description and correlation matrix of air pollutants at the LSOAs level (N = 42,619 LSOAs)
	Pearson’s correlation coefficient
	NO₂ (µg/m³)	SO₂ (µg/m³)	PM10 (µg/m³)	PM2.5 (µg/m³)	Mean	SD	Median	Interquartile range
NO₂ (µg/m³)	1.00				15.76	7.43	14.91	9.63
SO₂ (µg/m³)	0.37	1.00			1.87	1.23	1.58	1.06
PM10 (µg/m³)	0.76	0.28	1.00		14.35	3.18	14.51	4.33
PM2.5 (µg/m³)	0.79	0.32	0.97	1.00	9.74	2.40	9.85	3.31
Strong correlations with correlation coefficient ≥ 0.70 are highlighted in bold

3.2.2. Description of air pollution at the Local authority level

Similar to the LSOAs, fluctuations in the mean of NO₂, SO₂, PM10, and PM2.5 concentrations across the 391 local authorities in the UK were observed from one year to another with lower levels of pollution noted in the last 5 years of the observation window (2015–2019) in comparison to previous years (2009–2014) with an exception for the year 2016 (Fig. 4).

A high correlation (Pearson’s coefficient ≥ 0.7) was also observed between NO₂, PM10, and PM2.5 pollutants at the local authority level (Table 3).

Table 3

Exposure description and correlation matrix of air pollutants at the local authority level (N = 391 local authorities)
	Pearson’s correlation coefficient
	NO₂ (µg/m³)	SO₂ (µg/m³)	PM10 (µg/m³)	PM2.5 (µg/m³)	Mean	SD	Median	Interquartile range
NO₂ (µg/m³)	1.00				12.99	7.08	11.71	8.18
SO₂ (µg/m³)	0.50	1.00			1.57	0.93	1.37	0.90
PM10 (µg/m³)	0.77	0.38	1.00		14.06	3.19	14.43	4.26
PM2.5 (µg/m³)	0.81	0.42	0.97	1.00	9.42	2.35	9.62	3.23
Strong correlations with correlation coefficient ≥ 0.70 are highlighted in bold

3.3. Description of individuals’ self-reported health

The mean of self-reported general health (1 to 5: excellent to poor) was 2.65 (SD = 0.36) indicating that most of the individuals report good health. Specifically, excellent/very good/good health was prevalent in 79% of the responses, while 21% of responses indicated fair/poor health. High homogeneity in the self-reported health responses was noted within the individual clusters over time (ICC = 0.65), while low homogeneity was observed within the household clusters (ICC = 0.24) (Table 4).

Table 4

Intraclass correlation coefficient for within individual and household clusters
		General Health scale (1 to 5: excellent to poor)
Individual ID	ICC [95%CI]	0.65 [0.64, 0.65]
	N of surveys	404,264
	N of individuals	67,982
	Mean^a (SD)	2.65 (0.36)
Household ID	ICC [95%CI]	0.24 [0.24, 0.24]
	N of surveys	404,264
	N of households	233,212

Strong ICCs > 0.3 are highlighted in italic-bold; Mean^a is based on predictions from mixed-effects linear models which are adjusted for age in fixed effects and for the individual ID in random intercept.

3.4. The spatial-temporal effect of air pollution on individuals’ health

3.4.1. The spatial-temporal effect of air pollution on individuals’ health at the LSOAs level

Results showed that poorer self-reported health (1–5 Likert scale: excellent to poor health) is associated with a 10 µg/m³ increase in NO₂ (coefficient = 0.05, 95%CI = 0.04, 0.06), SO₂ (coefficient = 0.12, 95%CI = 0.09, 0.14), PM10 (coefficient = 0.07, 95%CI = 0.05, 0.09), and PM2.5 (coefficient = 0.10, 95%CI = 0.08, 0.13) pollutants linked at the LSOAs level (Table 5). In bi-pollutant models adjusting each of NO₂, PM10 and PM2.5 models for SO₂ pollutant, similar results were observed of poorer reported health with every 10 µg/m³ increase in NO₂, PM10 and PM2.5 pollutants (Table 6).

Decomposing the overall effect of air pollution on health into between (spatial: across LSOAs) and within (temporal: across years within each LSOA) effects, showed significant positive associations with poorer health for the between effect for NO₂ (Coefficient = 0.05, 95%CI = 0.04, 0.06), SO₂ (Coefficient = 0.60, 95%CI = 0.52, 0.68), PM10 (Coefficient = 0.07, 95%CI = 0.05, 0.10), and PM2.5 (Coefficient = 0.12, 95%CI = 0.08, 0.15) pollutants. No significant within effects were observed for these four pollutants, although the sign of the coefficients is largely as expected (Table 5).

In a sensitivity analysis of the association between air pollution and self-reported health as a binary variable (fair/poor health versus excellent/very good/good health), results remained unchanged except for PM10 which does not show a significant between effect (OR = 1.08,95%CI = 0.98, 1.20) on fair/poor reported health, but rather it is showing significant within effect (OR = 1.28, 95%CI = 1.02, 1.61) (Additional file Table 2).

Likewise, sensitivity analysis for wave 1 recruited individuals showed similar results for the overall and for the between-within effects of the four pollutants on individuals’ health (Additional file Table 4).

Table 5

The association of self-reported general health with each of NO₂, SO₂, PM10, and PM2.5 air pollutants in separate models at the LSOAs level (N = 404,264 surveys from 67,982 individuals)
	Model 1	Model 2
	Coefficient [95%CI]	Coefficient [95%CI]
Overall pollution effect
NO₂ (µg/m³)	0.06 [0.05, 0.06]**	0.05 [0.04, 0.06]**
SO₂ (µg/m³)	0.13 [0.11, 0.16]**	0.12 [0.09, 0.14]**
PM10 (µg/m³)	0.07 [0.05, 0.10]**	0.07 [0.05, 0.09]**
PM2.5 (µg/m³)	0.11 [0.08, 0.13]**	0.10 [0.08, 0.13]**
Between pollution effect
NO₂ (µg/m³)	0.06 [0.05, 0.07]**	0.05 [0.04, 0.06]**
SO₂ (µg/m³)	0.87 [0.79, 0.96]**	0.60 [0.52, 0.68]**
PM10 (µg/m³)	0.08 [0.05, 0.10]**	0.07 [0.05, 0.10]**
PM2.5 (µg/m³)	0.12 [0.08, 0.15]**	0.12 [0.08, 0.15]**
Within pollution effect
NO₂ (µg/m³)	0.02 [-0.01, 0.04]	0.006 [-0.02, 0.03]
SO₂ (µg/m³)	0.01 [-0.04, 0.05]	0.01 [-0.03, 0.05]
PM10 (µg/m³)	0.03 [-0.01, 0.07]	0.02 [-0.02, 0.06]
PM2.5 (µg/m³)	-0.00 [-0.05, 0.05]	-0.01 [-0.06, 0.04]
*P-value < 0.01; P-value < 0.05;

Coefficients and 95%CIs are expressed in terms of 10 µg/m³ increase in the air pollutants;

Model 1: is adjusted for age, gender, and year dummies (2009–2019); Model 2 is adjusted for age, gender, ethnicity, country of birth, marital status, education, occupation, perceived financial situation, smoking status, and year dummies (2009–2019).

Table 6

The association of self-reported general health with each of NO₂, SO₂, PM10, and PM2.5 air pollutants in bi-pollutant models adjusted for SO₂ pollutant at the LSOAs level (N = 404,264 surveys from 67,982 individuals)
		Overall pollution effect
		Coefficient [95%CI]
NO₂ - SO₂ Model	NO₂ (µg/m³)	0.04 [0.03, 0.05]**
NO₂ - SO₂ Model	SO₂ (µg/m³)	0.09 [0.06, 0.11]**
PM10 - SO₂ Model	PM10 (µg/m³)	0.06 [0.04, 0.08]**
PM10 - SO₂ Model	SO₂ (µg/m³)	0.10 [0.08, 0.13]**
PM2.5 - SO₂ Model	PM2.5 (µg/m³)	0.08 [0.06, 0.11]**
PM2.5 - SO₂ Model	SO₂ (µg/m³)	0.10 [0.07, 0.13]**
*P-value < 0.01; P-value < 0.05;

Coefficients and 95%CIs are expressed in terms of 10 µg/m³ increase in the air pollutants;

Models are additionally adjusted for age, gender, ethnicity, country of birth, marital status, education, occupation, perceived financial situation, smoking status, and year dummies (2009–2019).

3.4.2. The spatial-temporal effect of air pollution on individuals’ health at the local authority level

At the local authority level, poorer self-reported health (1–5 Likert scale: excellent to poor health) was also associated with a 10 µg/m³ increase in NO₂ (coefficient = 0.03, 95%CI = 0.02, 0.05), SO₂ (coefficient = 0.08, 95%CI = 0.04, 0.13), PM10 (coefficient = 0.04, 95%CI = 0.008, 0.06), and PM2.5 (coefficient = 0.06, 95%CI = 0.03, 0.10) pollutants, suggesting that the higher the pollution levels in a local authority are, the poorer the health of individuals living there is (Table 7). Similar results were noted in the bi-pollutant models adjusting each of NO₂, PM10 and PM2.5 models for the SO₂ pollutant (Table 8).

Analysing the between-within effects revealed significant positive associations with poorer health for the between effect for NO₂ (Coefficient = 0.03, 95%CI = 0.01, 0.05) and SO₂ (Coefficient = 0.62, 95%CI = 0.47, 0.77) pollutants, while no significant within effects were noted for these two pollutants. Contrary to the LSOAs, both between and within effects were not present for PM10 and PM2.5 pollutants at the local authority level (Table 7).

In a sensitivity analysis of the association between air pollution and self-reported health as a binary variable (fair/poor health versus excellent/very good/good health), results remained unchanged except for PM10 and PM2.5 pollutants which do not show significant associations with fair/poor reported health (Additional file Table 3).

Sensitivity analysis for wave 1 recruited individuals showed similar results for the overall and for the between-within effects of the four pollutants (at the local authority level) on individuals’ health (Additional file Table 5).

Table 7

The association of self-reported general health with each of NO₂, SO₂, PM10, and PM2.5 air pollutants in separate models at the local authority level (N = 404,264 surveys from 67,982 individuals)
	Model 1	Model 2
	Coefficient [95%CI]	Coefficient [95%CI]
Overall pollution effect
NO₂ (µg/m³)	0.04 [0.02, 0.05]**	0.03 [0.02, 0.05]**
SO₂ (µg/m³)	0.08 [0.03, 0.12]**	0.08 [0.04, 0.13]**
PM10 (µg/m³)	0.04 [0.007, 0.07]*	0.04 [0.008, 0.06]*
PM2.5 (µg/m³)	0.07 [0.03, 0.11]**	0.06 [0.03, 0.10]**
Between pollution effect
NO₂ (µg/m³)	0.04 [0.01, 0.06]**	0.03 [0.01, 0.05]**
SO₂ (µg/m³)	0.86 [0.67, 1.05]**	0.62 [0.47, 0.77]**
PM10 (µg/m³)	0.01 [-0.04, 0.06]	0.02 [-0.02, 0.06]
PM2.5 (µg/m³)	0.03 [-0.05, 0.10]	0.04 [-0.01, 0.10]
Within pollution effect
NO₂ (µg/m³)	0.007 [-0.02, 0.04]	0.003 [-0.03, 0.03]
SO₂ (µg/m³)	0.02 [-0.04, 0.09]	0.02 [-0.05, 0.08]
PM10 (µg/m³)	0.01 [-0.04, 0.06]	0.008 [-0.04, 0.05]
PM2.5 (µg/m³)	-0.04 [-0.10, 0.01]	-0.04 [-0.10, 0.014]
*P-value < 0.01; P-value < 0.05;

Coefficients and 95%CIs are expressed in terms of 10 µg/m³ increase in the air pollutants;

Table 8

The association of self-reported general health with each of NO₂, SO₂, PM10, and PM2.5 air pollutants in bi-pollutant models adjusted for SO₂ pollutant at the local authority level (N = 404,264 surveys from 67,982 individuals)
		Overall pollution effect
		Coefficient [95%CI]
NO₂ - SO₂ Model	NO₂ (µg/m³)	0.03 [0.01, 0.04]**
NO₂ - SO₂ Model	SO₂ (µg/m³)	0.05 [0.006, 0.10]*
PM10 - SO₂ Model	PM10 (µg/m³)	0.03 [-0.003, 0.05]
PM10 - SO₂ Model	SO₂ (µg/m³)	0.07 [0.03, 0.12]**
PM2.5 - SO₂ Model	PM2.5 (µg/m³)	0.05 [0.01, 0.09]**
PM2.5 - SO₂ Model	SO₂ (µg/m³)	0.07 [0.02, 0.11]**
*P-value < 0.01; P-value < 0.05;

Coefficients and 95%CIs are expressed in terms of 10 µg/m³ increase in the air pollutants;

Models are additionally adjusted for age, gender, ethnicity, country of birth, marital status, education, occupation, perceived financial situation, smoking status, and year dummies (2009–2019).

3.5. The association of air pollution with individuals’ health by ethnicity and country of birth

3.5.1 The association of air pollution with individuals’ health by ethnicity and country of birth at the LSOAs level

Examining the association between ethnicity and individuals’ health revealed poorer self-reported health among Indian (coefficient = 0.062, 95%CI = 0.026, 0.098), Pakistani/Bangladeshi (coefficient = 0.174, 95%CI = 0.141, 0.208), mixed ethnicities (coefficient = 0.086, 95%CI = 0.041, 0.132), and other ethnicities (coefficient = 0.034, 95%CI = 0.006, 0.061) in comparison to the British-white (Additional file Table 1).

Analysis of the association between air pollution and individuals’ health by ethnicity and country of birth at the LSOAs level showed a stronger effect of air pollution on poor self-reported health among ethnic minorities. Specifically, individuals from an Indian, Pakistani/Bangladeshi, and other ethnicities origin reported poorer health with every 10 µg/m³ increase in SO₂, PM10, and PM2.5 pollutants compared to British-White. Non-UK-born individuals were also more likely to report poorer health than UK-born individuals with increasing concentrations of all the four pollutants (Fig. 5).

Similar associations were also noted when self-reported health was examined as a binary variable comparing fair/poor to excellent/very good/good health (Additional file Fig. 1) and in the sensitivity analysis for only wave 1 recruited individuals (Additional file Fig. 5).

Analysing the between-within (spatial-temporal) effects of air pollution on health by ethnicity and country of birth at the LSOAs level showed less consistent results than the overall effect of air pollution. Only borderline associations with poorer health were noted for the between effect of NO₂, PM10, and PM2.5 pollution among Black/African/Caribbean in comparison to British-white and among those not born in the UK in comparison to natives. In contrast, Indians and non-UK-born individuals showed better health with more temporal variation (within effect) in PM10 and PM2.5 pollutants (Fig. 6).

No significant associations were observed for the between or within pollution effect when health was analysed as a binary variable (Additional file Fig. 2).

Between-within analysis for individuals recruited at wave 1 of the Understanding Society study revealed similar results (Additional file Fig. 6).

3.5.2 The association of air pollution with individuals’ health by ethnicity and country of birth at the local authority level

Analysis of air pollution and health by ethnicity and country of birth at the local authority level also revealed a stronger effect of air pollution on poor self-reported health among ethnic minorities; yet with some noted differences than the analysis performed at the LSOAs level. At the local authority level, individuals from an Indian, Pakistani/Bangladeshi, Black/African/Caribbean, and other ethnicities origin reported poorer health than the British-white with every 10 µg/m³ increase in NO₂, SO₂, PM10, and PM2.5 pollution (Fig. 7). Similar to LSOAs, non-UK-born individuals were more likely to report poorer health than UK-born individuals with increasing concentrations of all the four pollutants linked at the local authority level (Fig. 7).

Similar associations were noted when self-reported health was examined as a binary variable comparing fair/poor to excellent/very good/good health; yet results did not differ between air pollution being linked at the local authority or the LSOAs level (Additional file Fig. 3).

Analysis for only wave 1 recruited individuals revealed poorer health among ethnic minorities in comparison to British-white and among non-UK-born individuals with increasing concentrations of the four pollutants (Additional file Fig. 7).

For the between-within (spatial-temporal) effects of air pollution at the local authority level, significant associations with poorer health whether measured as a Likert scale (Fig. 8) or as a binary variable (Additional file Fig. 4) were noted for the between effect of NO₂, PM10, and PM2.5 pollution among Black/African/Caribbean and other ethnicities in comparison to British-white and among those not born in the UK. In contrast, Indians, Black/African/Caribbean and non-UK-born individuals showed better health with more temporal variation (within effect) in PM10 and PM2.5 pollutants only when health was measured as a Likert scale (Fig. 8), while no significant associations were observed for the within pollution effect on health measured as a binary variable (Additional file Fig. 4). Between-within analysis for individuals recruited at wave 1 of the Understanding Society study revealed similar results (Additional file Fig. 8).

This study reports the harmful effect of exposure to NO₂, SO₂, PM10, and PM2.5 air pollution (linked at coarse local authorities and detailed LSOAs geographical scales) on self-reported health in the UK for individuals followed from the year 2009 to 2019. These findings are corroborated by relevant literature whereby exposure to air pollution has been shown to be the leading cause for many respiratory (eg. asthma, bronchiolitis), cardiovascular (eg. chronic obstructive pulmonary disease, emphysema, myocardial infarction), cerebrovascular (eg. stroke), and cancer (eg. lung cancer) diseases (2, 4). This in turn contributes to increased rates of mortality (49–52), hospital admissions (3, 9, 53, 54), and poor self-reported health (6, 55).

Although the negative effect of air pollution on health is well-established in the literature, this study was novel in going a further step in an attempt to show the between-within effects of air pollution on health. Additionally, the analysis was carried out at two geographical scales, the coarse local authorities and more detailed LSOAs, which forms another novelty of this study. The between-within analysis is widely used in the fields of economics, behavioural finance, and strategic management (56). However, this type of analysis is rarely used in health research (57); and no previous study has assessed the between-within effects of air pollution on health. Through the application of the between-within analysis in this study, we observed significant between, but not within effects on poor self-reported health for NO₂ and SO₂ pollutants, at both the LSOAs and local authority levels. However, for PM10 and PM2.5 pollutants, significant between but not within effects were observed only when linked at the LSOAs level, but not at the local authority level. Therefore, individuals residing in local authorities or LSOAs with higher average concentrations of NO₂ and SO₂ pollution across the 11 years of follow-up exhibited poorer self-reported health in comparison to individuals residing in local authorities or LSOAs with lower pollution concentrations. For particulate matter pollution, only residing in more polluted LSOAs resulted in poorer health. Hence, analysis at the local authority level attenuated the spatial (between) effect of PM10 and PM2.5 pollution on individuals’ health in comparison to the analysis at the LSOAs level. This implies stronger associations at the LSOAs finer geographical scale compared to the coarser local authority level. However, conducting analysis at the coarser local authorities’ level was necessary in guiding local authority-specific decision-making regarding air pollution and health.

In all cases, our study shows strong evidence for the spatial rather than temporal effects of air pollution on health, whether linked at the coarse local authority level or at the finer LSOAs level. This could be explained by the low variation of yearly air pollution concentrations across the 11 years of follow-up, particularly for SO₂ pollutant as shown in Fig. 3 for LSOAs and Fig. 4 for local authorities. Hence, increasing the follow-up time to allow for more variation in air pollution might result in significant within effects. Additionally, air pollution exposure in this study was assessed at a yearly rather than monthly or daily basis which also limits the variation in air pollution across time, resulting in weaker temporal associations.

Despite the statistically insignificant within results, the coefficients indicated a positive association with poorer general health. This implies that the variation in air pollution over time within each local authority or LSOA can contribute to poorer health among individuals living in the respective local authority or LSOA. Hence, if the number of vehicles and/or industrial facilities increases over time in a respective local authority or LSOA, people may experience poorer health due to the increased air pollution exposure.

The observed between-within effects can be also explained by the emission source of the pollutants and their chemical reactivity in the atmosphere. The major source of NO₂ emissions is traffic exhaust (58) which varies across both local authorities/LSOAs (between: spatial) and time (within: temporal) depending on the number of vehicles and the movement of people. Yet, nitrogen oxides are highly reactive and seasonal pollutants (59), which makes it difficult to capture their temporal variation through yearly measurements. For instance, more NO₂ will be liberated into the atmosphere during warm seasons due to the chemical reactions between nitrogen oxides and ozone (59). Additionally, NO₂ is converted into nitric acid by several different reactions in the atmosphere (60). That’s why only spatial (between) but not temporal (within) effects for NO₂ pollutant were observed when taking the year as our time measuring unit. On the other hand, industrial processes and power plants are the major sources of SO₂ pollution (61), which is dominated by spatial (between) variation rather than temporal (within) variation as building a new factory requires much longer time than purchasing a motor vehicle. Particulate matter results from both traffic exhaust and industrial processes (62), and is considered a more stable pollutant that may stay suspended in the air for long periods of time (60). Thus, an overall effect of particulate matter on health is expected rather than a spatial or a temporal derived effect. Yet, the stable nature of particulate matter allows this pollutant to show a spatial effect when using high spatial resolution geographical scale such as LSOAs while this spatial effect will be attenuated when using low spatial resolution scale such as local authorities.

This study was also novel in analysing how the overall and the between and within effects of air pollution on self-reported health vary across six ethnic groups and by country of birth. Analysis revealed a stronger effect of air pollution on poor health among Pakistani/Bangladeshi, Indian, Black/African/Caribbean (only at the local authority level), and other ethnic minorities compared to British-White; and among non-UK-born individuals compared to natives. These findings are corroborated by similar research from the U.S. whereby non-Hispanic white individuals were 10% more likely to report hypertension and non-Hispanic blacks were 2 times more likely to report asthma with increasing concentrations of PM2.5 pollution (15, 16).

In contrast, the between-within analysis did not show consistent associations between air pollution and health across ethnic groups. Only individuals from Black/African/Caribbean origin and those not born in the UK reported poorer health with increasing concentrations of local authority and LSOAs-specific 11 years average NO₂, PM10, and PM2.5 pollution (between effects). Whilst better health was observed with more temporal variation in PM10 and PM2.5 pollutants (within effects) among Indians, Black/African/Caribbean (only at the local authority level), and non-UK-born individuals; yet these significant within effects disappeared when health was measured as a binary variable.

The observed ethnic differences in health in the context of air pollution can be explained by two concepts derived from relevant literature on ethnic inequalities in health. The first concept relates to the socioeconomic and lifestyle behavioural differences among ethnic groups. Research has shown that ethnic minorities often live in more disadvantaged communities and have lower socioeconomic status, lower healthcare coverage, and higher job/income insecurity which increases their risk of illness and lead to poor health (25, 27, 28). In fact, people from Pakistani and Bangladeshi origins tend to report the poorest health in the UK, followed by people from Indian and Caribbean origins (26). This was confirmed in our analysis whereby Pakistani/Bangladeshi, Indians, mixed, and other ethnicities individuals were more likely to report poor general health in comparison to British-white people (Additional file Table 1). However, our analysis accounted for major socioeconomic characteristics such as age, gender, marital status, education, occupation, and financial situation. Still, ethnic differences in the effect of air pollution on health persisted. Hence, those differences can be related to other socioeconomic and individual factors not captured in our analysis or to contextual location-specific factors which leads us to the second concept.

Contextual location-specific factors such as urbanisation, population density, neighbourhood, and housing conditions can help explain the observed ethnic differences in the effect of air pollution on health. Ethnic minorities and immigrants (foreign-born individuals) often reside in large cities and highly populated urbanised regions, near major roads and key transportation networks. This facilitates their movement and increases their chances of personal development, employment and business start-ups (63). In addition, ethnic minorities often live in low-priced social housing offered by local authorities, which is often situated in more deprived ethnic concentration neighbourhoods or close to major roads and industrial areas (29). In contrast, British-white and UK-born-individuals are at a greater advantage in terms of job security, financial means, and inheritance tenure to move away from metropolitan areas and highly polluted industrial regions. These location-specific factors would expose ethnic minorities and non-UK-born individuals into higher concentrations of air pollution related to traffic exhaust, industries, and burning of fossil fuels, which would manifest in greater health impacts compared to the rest of the population. In additional analysis through Chi2 tabulation, we show that a very high percentage of non-UK-born individuals (93.5%) and of ethnic minorities including Pakistani/Bangladeshi (99.6%), Indian (98.4%), Black/African/Caribbean (98.9%), mixed (94.4%) and other ethnicities (84.0%) live in urban areas, whereas this percentage is much lower for British-white (71.5%) and UK-born (74.7%) individuals (Additional file Table 6). In further analysis for individuals living in urban areas, we show that ethnic minorities and non-UK-born individuals live in more polluted local authorities especially for NO₂ pollutant with an average exposure exceeding 20 µg/m³ for individuals from Pakistani/Bangladeshi, Indian, Black/African/Caribbean, and mixed ethnicity origins compared to an average exposure of 14 µg/m³ for the British-white group (Additional file Table 7). Furthermore, the between-within (spatial-temporal) analysis revealed stronger between effects for NO_2, PM10, and PM2.5 pollution on poor self-reported health among Black/African/Caribbean and other ethnicities and among non-UK-born individuals, thus further confirming that residing in more polluted local authorities is a key explanation for the observed ethnic inequalities in health.

Despite the novelty of this study, it has some limitations. First, the study design involved linking individual-level data from the “Understanding society” panel survey to yearly air pollution contextual data at the local authority level. This can lead to exposure bias assuming that every individual within the same local authority within a specific year is exposed to the same level of air pollution. To address this problem, yearly air pollution data were also linked to individual-level data from the “Understanding Society” survey at the census Lower super output areas (LSOAs) level which reduces the exposure bias. Air pollution linked at both, the local authority and the LSOAs levels, revealed similar results with some noted differences. Thus, for future research, it is recommended to utilize another data source that allows for air pollution linkages at the level of postcodes, which exhibit area subdivisions with the highest spatial resolution in the UK. Second, our study examined the effect of air pollution on self-reported health rather than using more objective health measures such as mortality or hospital admissions. This could lead to social desirability or reporting bias whereby individuals overestimate or underestimate their general health. However, high correlations between self-reported health and objective health measures including mortality and hospital admissions were demonstrated by relevant literature, which increases the reliability of the self-reported health variable (36–38). Furthermore, research from the UK has shown an association between poorer self-rated health and greater morbidity within each ethnic group; hence, providing evidence that the use of self-rated health to measure health status in different ethnic groups in the UK is valid (64). Third, our study included all individuals recruited at different waves of the “Understanding society” data, that had at least two observations through the follow-up period (2009–2019). Therefore, some individuals were followed for the whole observation window of 11 years and started at wave 1 while others were recruited at later data collection waves and followed for a shorter period. Nevertheless, we performed sensitivity analysis only on individuals recruited in wave 1 to balance the cohort effect (Additional file Table 3, Table 5, Fig. 5, Fig. 6, Fig. 7, and Fig. 8) and results remained unchanged. Forth, the sample design of the “Understanding Society” survey involved ethnic minority boost samples at waves 1 and 6 of data collection to enable ethnicity-focused research; and thus, the survey included longitudinal weights that adjust for the overrepresentation of some ethnic groups. However, we could not adjust our analysis for the longitudinal weights as this requires that all individuals be followed until the last wave (wave 10) of the survey, which was not the case. Hence, our ethnicity analysis might not be generalizable to the whole UK population, but rather represent regions with dense ethnic minority concentrations. Finally, our study included individuals followed over 11 years (2009–2019) of time. However, the air pollution variation across these 11 years was low, which did not allow for the detection of significant temporal (within) effects of air pollution on health. For future research, we recommend using other datasets with longer follow up time to allow for more variation in air pollution which might result in significant temporal effects.

Using a longitudinal panel design that involves linking individual-level to air pollution data at two geographical scales (coarse local authorities and detailed LSOAs), this study supports the presence of a spatial-temporal association between air pollution and individuals’ reported health in the UK. Though results showed stronger between (spatial) effects across local authorities/LSOAs rather than within (temporal) effects across time within each local authority/LSOA. Furthermore, this study demonstrates a stronger effect of air pollution on poor self-reported health among ethnic minorities and non-UK-born individuals, which is partly explained by location-specific differences. Our results are of importance for policymakers in the UK toward advancing legislations related to air pollution, health, time, and place with an emphasis on targeting the ethnic inequalities in health.

NO₂– Nitrogen dioxide

SO₂– Sulphur dioxide

PM10 – Particulate matter with diameter ≤ 10 µm

PM2.5 – Particulate matter with diameter ≤ 2.5 µm

UK – United-Kingdom

COMEAP – Committee on the Medical Effects of Air Pollution

LSOAs – Lower Super Output Areas

OR – Odd Ratio

CI – Confidence Interval

SD – Standard deviation

ICC – Intraclass correlation coefficient

Ethics approval and consent to participate: This paper is part of a PhD project that was granted ethical approval on the 14th of May 2020 by the School of Geography and Sustainable Development Ethics Committee, acting on behalf of the University Teaching and Research Ethics Committee (UTREC) of the University of St Andrews, Scotland, United Kingdom. This paper uses secondary data from the “Understanding Society: The UK Household Longitudinal Study” which was collected by the University of Essex and can be downloaded from the UK Data Archive. Thus, obtaining informed consent solely for this study is not applicable and was waived by the School of Geography and Sustainable Development Ethics Committee, acting on behalf of the University Teaching and Research Ethics Committee (UTREC) of the University of St Andrews, Scotland, United Kingdom. Informed consent was however asked from the individuals participating in the Understanding Society study during the primary data collection which was carried out by the University of Essex. The University of Essex Ethics Committee has approved all data collection on Understanding Society main study and innovation panel waves, including asking individuals’ consent for all data linkages except to health records. Requesting consent for health record linkage was approved at Wave 1 by the National Research Ethics Service (NRES) Oxfordshire REC A (08/H0604/124), at BHPS Wave 18 by the NRES Royal Free Hospital & Medical School (08/H0720/60) and at Wave 4 by NRES Southampton REC A (11/SC/0274). Approval for the collection of biosocial data by trained nurses in Waves 2 and 3 of the main survey was obtained from the National Research Ethics Service (Understanding Society - UK Household Longitudinal Study: A Biosocial Component, Oxfordshire A REC, Reference: 10/H0604/2).

The authors confirm that all methods were carried out in accordance with relevant guidelines and regulations in the Declaration of Helsinki.

Consent for publication: Not applicable

Availability of data and materials: We cannot make the data underlying our analysis publicly available due to ethical and legal restrictions. We are using the “Understanding Society: The UK Household Longitudinal Study” dataset which is an initiative funded by the Economic and Social Research Council and various Government Departments, with scientific leadership by the Institute for Social and Economic Research, University of Essex, and survey delivery by NatCen Social Research and Kantar Public. These data are protected by a copyright license and strictly distributed by the UK Data Service which is the largest digital repository for quantitative and qualitative social science and humanities research data in the UK. Therefore, data underlying our analysis can only be accessed through the UK Data Service for authorized researchers from the following URL: https://beta.ukdataservice.ac.uk/datacatalogue/series/series?id=2000053

Competing interests: The authors declare that they have no competing interests.

Funding: This paper is part of a PhD project that is funded by the St Leonard’s PhD Scholarship, School of Geography and Sustainable Development, and School of Medicine, University of St Andrews, Scotland, United Kingdom.

Authors' contributions: Mary Abed Al Ahad: Conceptualization, Investigation, Methodology, Data curation, Formal Analysis, Writing-Original Draft, Writing-Review and Editing, Visualization, Project administration. Hill Kulu, Urška Demšar and Frank Sullivan: Conceptualization, Funding acquisition, Supervision, Writing-Review and Editing.

Acknowledgements: Not applicable

Ghorani-Azam A, Riahi-Zanjani B, Balali-Mood M. Effects of air pollution on human health and practical measures for prevention in Iran. J Res Med Sci. 2016;21:65-.
Manisalidis I, Stavropoulou E, Stavropoulos A, Bezirtzoglou E. Environmental and Health Impacts of Air Pollution: A Review. Front Public Health. 2020;8:14-.
Ab Manan N, Noor Aizuddin A, Hod R. Effect of Air Pollution and Hospital Admission: A Systematic Review. Ann Glob Health. 2018;84(4):670-8.
Abed Al Ahad M, Sullivan F, Demšar U, Melhem M, Kulu H. The effect of air-pollution and weather exposure on mortality and hospital admission and implications for further research: A systematic scoping review. PLoS One. 2020;15(10):e0241415.
Collart P, Dubourg D, Levêque A, Sierra NB, Coppieters Y. Short-term effects of nitrogen dioxide on hospital admissions for cardiovascular disease in Wallonia, Belgium. Int J Cardiol. 2018;255:231-6.
Goldberg MS, Wheeler AJ, Burnett RT, Mayo NE, Valois M-F, Brophy JM, et al. Physiological and perceived health effects from daily changes in air pollution and weather among persons with heart failure: A panel study. Journal of Exposure Science & Environmental Epidemiology. 2015;25(2):187-99.
COMEAP. Long-Term Exposure to Air Pollution: Effect on Mortality. UK Government website: Committee on the Medical Effects of Air Pollutants 2009.
Charafeddine R, Boden LI. Does income inequality modify the association between air pollution and health? Environmental Research. 2008;106(1):81-8.
Colais P, Faustini A, Stafoggia M, Berti G, Bisanti L, Cadum E, et al. Particulate Air Pollution and Hospital Admissions for Cardiac Diseases in Potentially Sensitive Subgroups. Epidemiology. 2012;23(3).
Faustini A, Stafoggia M, Renzi M, Cesaroni G, Alessandrini E, Davoli M, et al. Does chronic exposure to high levels of nitrogen dioxide exacerbate the short-term effects of airborne particles? Occupational and Environmental Medicine. 2016;73(11):772.
De Sario M, Katsouyanni K, Michelozzi P. Climate change, extreme weather events, air pollution and respiratory health in Europe. European Respiratory Journal. 2013;42(3):826.
Willers SM, Jonker MF, Klok L, Keuken MP, Odink J, van den Elshout S, et al. High resolution exposure modelling of heat and air pollution and the impact on mortality. Environment International. 2016;89-90:102-9.
Fleury-Bahi G, Préau M, Annabi-Attia T, Marcouyeux A, Wittenberg I. Perceived health and quality of life: the effect of exposure to atmospheric pollution. Journal of Risk Research. 2015;18(2):127-38.
Riedel N, Loerbroks A, Bolte G, Li J. Do perceived job insecurity and annoyance due to air and noise pollution predict incident self-rated poor health? A prospective analysis of independent and joint associations using a German national representative cohort study. BMJ Open. 2017;7(1):e012815.
Johnson D, Parker JD. Air pollution exposure and self-reported cardiovascular disease. Environmental Research. 2009;109(5):582-9.
Nachman KE, Parker JD. Exposures to fine particulate air pollution and respiratory outcomes in adults using two national datasets: a cross-sectional study. Environ Health. 2012;11:25.
Parker JD, Kravets N, Vaidyanathan A. Particulate Matter Air Pollution Exposure and Heart Disease Mortality Risks by Race and Ethnicity in the United States. Circulation. 2018;137(16):1688-97.
Astell-Burt T, Maynard MJ, Lenguerrand E, Whitrow MJ, Molaodi OR, Harding S. Effect of air pollution and racism on ethnic differences in respiratory health among adolescents living in an urban environment. Health & place. 2013;23:171-8.
Walker RJ, Strom Williams J, Egede LE. Influence of Race, Ethnicity and Social Determinants of Health on Diabetes Outcomes. Am J Med Sci. 2016;351(4):366-73.
Smith NR, Lewis DJ, Fahy A, Eldridge S, Taylor SJC, Moore DG, et al. Individual socio-demographic factors and perceptions of the environment as determinants of inequalities in adolescent physical and psychological health: the Olympic Regeneration in East London (ORiEL) study. BMC Public Health. 2015;15(1):150.
Assari S. Race, sense of control over life, and short-term risk of mortality among older adults in the United States. Arch Med Sci. 2017;13(5):1233-40.
Fischbacher CM, Cezard G, Bhopal RS, Pearce J, Bansal N, on behalf of the Scottish H, et al. Measures of socioeconomic position are not consistently associated with ethnic differences in cardiovascular disease in Scotland: methods from the Scottish Health and Ethnicity Linkage Study (SHELS). International Journal of Epidemiology. 2014;43(1):129-39.
Lanting LC, Joung IMA, Mackenbach JP, Lamberts SWJ, Bootsma AH. Ethnic Differences in Mortality, End-Stage Complications, and Quality of Care Among Diabetic Patients. Diabetes Care. 2005;28(9):2280.
Abed Al Ahad M, Demšar U, Sullivan F, Kulu H. Does Long-Term Air Pollution Exposure Affect Self-Reported Health and Limiting Long Term Illness Disproportionately for Ethnic Minorities in the UK? A Census-Based Individual Level Analysis. Applied Spatial Analysis and Policy. 2022.
Cézard G, Finney N, Kulu H, Marshall A. Ethnic differences in self-assessed health in Scotland: The role of socio-economic status and migrant generation. Population, Space and Place. 2020;n/a(n/a):e2403.
Nazroo JY. The Structuring of Ethnic Inequalities in Health: Economic Position, Racial Discrimination, and Racism. American Journal of Public Health. 2003;93(2):277-84.
Egede LE. Race, ethnicity, culture, and disparities in health care. Journal of General Internal Medicine. 2006;21(6):667.
Su JG, Jerrett M, de Nazelle A, Wolch J. Does exposure to air pollution in urban parks have socioeconomic, racial or ethnic gradients? Environmental Research. 2011;111(3):319-28.
van Ham M, Manley D. Social housing allocation, choice and neighbourhood ethnic mix in England. Journal of Housing and the Built Environment. 2009;24(4):407-22.
Wallace M, Kulu H. Migration and Health in England and Scotland: a Study of Migrant Selectivity and Salmon Bias. Population, Space and Place. 2014;20(8):694-708.
Wallace M, Kulu H. Mortality among immigrants in England and Wales by major causes of death, 1971–2012: A longitudinal analysis of register-based data. Social Science & Medicine. 2015;147:209-21.
Schmidt-Catran AW, Fairbrother M. The Random Effects in Multilevel Models: Getting Them Wrong and Getting Them Right. European Sociological Review. 2016;32(1):23-38.
UniversityofEssex. Understanding Society: Waves 1-10, 2009-2019 and Harmonised BHPS: Waves 1-18, 1991-2009. In: University of Essex IfSaER, NatCen Social Research, Kantar Public, editor. 13th Edition ed. UK Data Service2020.
Lynn P. Sample design for understanding society. Underst Soc Work Pap Ser. 2009;2009.
Lynn P, Nandi A, Parutis V, Platt L. Design and implementation of a high-quality probability sample of immigrants and ethnic minorities: Lessons learnt. Demographic Research. 2018;38:513-48.
Ul-Haq Z, Mackay DF, Pell JP. Association between self-reported general and mental health and adverse outcomes: a retrospective cohort study of 19,625 Scottish adults. PLoS One. 2014;9(4):e93857-e.
Williams G, Di Nardo F, Verma A. The relationship between self-reported health status and signs of psychological distress within European urban contexts. European Journal of Public Health. 2017;27(suppl_2):68-73.
Myint PK, Luben RN, Surtees PG, Wainwright NWJ, Welch AA, Bingham SA, et al. Relation Between Self-Reported Physical Functional Health and Chronic Disease Mortality in Men And Women in the European Prospective Investigation Into Cancer (EPIC–Norfolk): A Prospective Population Study. Annals of Epidemiology. 2006;16(6):492-500.
Modelled background pollution data [Internet]. 2020. Available from: https://uk-air.defra.gov.uk/data/pcm-data.
Local Authority Districts (December 2017) Full Clipped Boundaries in United Kingdom [Internet]. 2017. Available from: https://data.gov.uk/dataset/f8f57716-72b4-4c93-a50d-30b2a443c2c1/local-authority-districts-december-2017-full-clipped-boundaries-in-united-kingdom-wgs84.
Census Geography boundaries (Lower Layer Super Output Areas and Data Zones) [Internet]. 2011. Available from: https://statistics.ukdataservice.ac.uk/dataset/2011-census-geography-boundaries-lower-layer-super-output-areas-and-data-zones.
Abed Al Ahad M, Demšar U, Sullivan F, Kulu H. Air pollution and individuals’ mental well-being in the adult population in United Kingdom: A spatial-temporal longitudinal study and the moderating effect of ethnicity. PLoS One. 2022;17(3):e0264394.
Dimakopoulou K, Samoli E, Beelen R, Stafoggia M, Andersen ZJ, Hoffmann B, et al. Air pollution and nonmalignant respiratory mortality in 16 cohorts within the ESCAPE project. American journal of respiratory and critical care medicine. 2014;189(6):684-96.
Beelen R, Stafoggia M, Raaschou-Nielsen O, Andersen ZJ, Xun WW, Katsouyanni K, et al. Long-term exposure to air pollution and cardiovascular mortality: an analysis of 22 European cohorts. Epidemiology. 2014;25(3):368-78.
Carugno M, Consonni D, Randi G, Catelan D, Grisotto L, Bertazzi PA, et al. Air pollution exposure, cause-specific deaths and hospitalizations in a highly polluted Italian region. Environmental Research. 2016;147:415-24.
Hůnová I, Malý M, Řezáčová J, Braniš M. Association between ambient ozone and health outcomes in Prague. International Archives of Occupational and Environmental Health. 2013;86(1):89-97.
Schober P, Boer C, Schwarte LA. Correlation Coefficients: Appropriate Use and Interpretation. Anesth Analg. 2018;126(5):1763-8.
Lajos GJ, Haddad SM, Tedesco RP, Passini Jr R, Dias TZ, Nomura ML, et al. Intracluster correlation coefficients for the Brazilian Multicenter Study on Preterm Birth (EMIP): methodological and practical implications. BMC Medical Research Methodology. 2014;14(1):54.
Fischer PH, Marra M, Ameling CB, Hoek G, Beelen R, de Hoogh K, et al. Air Pollution and Mortality in Seven Million Adults: The Dutch Environmental Longitudinal Study (DUELS). Environmental Health Perspectives. 2015;123(7):697-704.
Hvidtfeldt UA, Sørensen M, Geels C, Ketzel M, Khan J, Tjønneland A, et al. Long-term residential exposure to PM2.5, PM10, black carbon, NO2, and ozone and mortality in a Danish cohort. Environment International. 2019;123:265-72.
Miller L, Xu X. Ambient PM2.5 Human Health Effects—Findings in China and Research Directions. Atmosphere. 2018;9(11).
Renzi M, Stafoggia M, Faustini A, Cesaroni G, Cattani G, Forastiere F. Analysis of Temporal Variability in the Short-term Effects of Ambient Air Pollutants on Nonaccidental Mortality in Rome, Italy (1998–2014). Environmental Health Perspectives. 2017;125(6):067019.
Almeida SM, Silva AV, Sarmento S. Effects of exposure to particles and ozone on hospital admissions for cardiorespiratory diseases in SetúBal, Portugal. J Toxicol Environ Health A. 2014;77(14-16):837-48.
Lanzinger S, Schneider A, Breitner S, Stafoggia M, Erzen I, Dostal M, et al. Ultrafine and Fine Particles and Hospital Admissions in Central Europe. Results from the UFIREG Study. American Journal of Respiratory and Critical Care Medicine. 2016;194(10):1233-41.
Shin J, Park JY, Choi J. Long-term exposure to ambient air pollutants and mental health status: A nationwide population-based cross-sectional study. PLoS One. 2018;13(4):e0195607.
McNeish D, Kelley K. Fixed effects models versus mixed effects models for clustered data: Reviewing the approaches, disentangling the differences, and making recommendations. Psychol Methods. 2019;24(1):20-35.
Dieleman JL, Templin T. Random-Effects, Fixed-Effects and the within-between Specification for Clustered Data in Observational Health Studies: A Simulation Study. PLoS One. 2014;9(10):e110257.
Grazuleviciene R, Maroziene L, Dulskiene V, Malinauskiene V, Azaraviciene A, Laurinaviciene D, et al. Exposure to urban nitrogen dioxide pollution and the risk of myocardial infarction. Scandinavian Journal of Work, Environment & Health. 2004;30(4):293-8.
Masiol M, Agostinelli C, Formenton G, Tarabotti E, Pavoni B. Thirteen years of air pollution hourly monitoring in a large city: Potential sources, trends, cycles and effects of car-free days. Science of The Total Environment. 2014;494-495:84-96.
Malek E, Davis T, Martin RS, Silva PJ. Meteorological and environmental aspects of one of the worst national air pollution episodes (January, 2004) in Logan, Cache Valley, Utah, USA. Atmospheric Research. 2006;79(2):108-22.
Vestreng V, Myhre G, Fagerli H, Reis S, Tarrasón L. Twenty-five years of continuous sulphur dioxide emission reduction in Europe. Atmos Chem Phys. 2007;7(13):3663-81.
Jyethi DS. Air Quality: Global and Regional Emissions of Particulate Matter, SOx, and NOx. In: Kulshrestha U, Saxena P, editors. Plant Responses to Air Pollution. Singapore: Springer Singapore; 2016. p. 5-19.
Lucas K, Stokes G, Bastiaanssen J, Burkinshaw J. Inequalities in Mobility and Access in the UK Transport System. Future of Mobility: Evidence Review, Government Office for Science. 2019.
Chandola T, Jenkinson C. Validating Self-rated Health in Different Ethnic Groups. Ethnicity & Health. 2000;5(2):151-9.

No competing interests reported.

Additionalfile.docx
Additional File: Sensitivity checks and additional analysis

Download PDF

Journal Publication

published 16 May, 2023

Read the published version in BMC Public Health →

Editorial decision: Major revision
13 Feb, 2023
Reviews received at journal
24 Jan, 2023
Reviewers agreed at journal
23 Jan, 2023
Reviewers agreed at journal
22 Jan, 2023
Reviewers invited by journal
21 Jan, 2023
Editor assigned by journal
03 Jan, 2023
Editor invited by journal
09 Nov, 2022
Submission checks completed at journal
09 Nov, 2022
First submitted to journal
08 Oct, 2022

You are reading this latest preprint version

The spatial-temporal effect of air pollution on individuals’ reported health and its variation by ethnic groups in the United Kingdom: A multilevel longitudinal analysis

Status:

Journal Publication

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

1. Introduction

2. Methods

2.1. Study design and population

2.2. Variables and measurements

2.2.1. Self-reported health

2.2.2. Air pollution

2.2.3. Socio-demographic and lifestyle covariates

2.3. Data analysis

3. Results

3.1. Description of individuals’ socio-demographic and lifestyle factors

3.2. Description of air pollution

3.2.1. Description of air pollution at the LSOAs level

3.2.2. Description of air pollution at the Local authority level

3.3. Description of individuals’ self-reported health

3.4. The spatial-temporal effect of air pollution on individuals’ health

3.4.1. The spatial-temporal effect of air pollution on individuals’ health at the LSOAs level

3.4.2. The spatial-temporal effect of air pollution on individuals’ health at the local authority level

3.5. The association of air pollution with individuals’ health by ethnicity and country of birth

4. Discussion

5. Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1