Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Oct 21.
Published in final edited form as: Environ Int. 2025 Jun 12;202:109607. doi: 10.1016/j.envint.2025.109607

Long-term exposure to air pollution and risk of dementia among older individuals of a Danish nationwide administrative cohort

Zorana J Andersen a, Youn-Hee Lim a, Jiawei Zhang a, Stéphane Tuffier a, Thomas Cole-Hunter a, Marie Bergmann a, Steffen Loft a, Laust H Mortensen b,c, Jie Chen d,m, Massimo Stafoggia e,f, Kees de Hoogh g,h, Klea Katsouyanni i,j, Danielle Vienneau g,h, Sophia Rodopoulou i, Evangelia Samoli i, Mariska Bauwelinck k, Jochem O Klompmaker l,m, Richard Atkinson n, Nicole AH Janssen l, Bente Oftedal o, Matteo Renzi e, Francesco Forastiere e,j, Maciek Strak d,l, Lau C Thygesen p, Paola Zaninotto q, Bert Brunekreef d, Gerard Hoek d, Rina So a,q,*
PMCID: PMC12536445  NIHMSID: NIHMS2105988  PMID: 40554225

Abstract

Background:

Increasing evidence associates air pollution with dementia, but some pollutants and susceptible groups are understudied.

Methods:

We followed all Danish residents aged ≥60 years as of 1–1-2000, without prior dementia, until 12–31-2018 for dementia incidence identified via hospital contact or prescription. We assessed annual mean levels of fine particulate matter (PM2⋅5), nitrogen dioxide (NO2), and black carbon (BC) in 2010 utilizing European-wide hybrid land-use regression models, at baseline (2000) residential addresses. We examined the associations between air pollution exposure and dementia incidence with Cox proportional hazard models, accounting for individual- and area-level socio-demographic covariates and whether the effects were modified by age, sex, income level, education attainment, employment status, and the presence of comorbid conditions, including cardio-metabolic, respiratory diseases, and depression.

Findings:

Among 934,792 individuals, 81,731 developed dementia over a mean follow-up of 11⋅6 years. Mean levels of PM2⋅5 and NO2, and BC were 12⋅5 and 20⋅6 µg/m3, and 1⋅0 × 10−5/m respectively. We detected strong associations between these pollutants and dementia incidence, with hazard ratios (HR) [95 % confidence intervals (CIs)] of 1⋅14 (1⋅12, 1⋅16), 1⋅25 (1⋅22, 1⋅28), and 1⋅23 (1⋅20, 1⋅26) per interquartile range increase of 1⋅9 μg/m3 for PM2⋅5, 10⋅2 μg/m3 for NO2, and 0⋅5 × 10−5/m for BC, respectively. Stronger associations were observed in elderly (≥75 years), those with stroke, the unemployed, and those with lower income or education levels than corresponding groups.

Discussion:

Even low levels of air pollution in Denmark were associated with dementia development, especially among certain susceptible groups, emphasizing the need for targeted intervention strategies.

Keywords: Dementia, Incidence, Administrative cohort, Air pollution, Long-term exposure, Denmark

1. Introduction

Along with aging of populations globally, there are growing concerns about an increasing burden of dementia: the numbers of affected individuals are projected to grow from 60 million in 2019 to 150 million in 2050 (GBD, 2022). Dementia is characterized by a progressive loss of memory and cognitive decline leading to functional limitations in language, problem solving, and thinking abilities, which can severely affect and limit daily life, with multiple pathophysiological mechanisms contributing to its development (Chin, 2023). Alzheimer’s disease is the most common type, followed by vascular dementia. In the lack of effective treatment, there has been increasing demand for research identifying modifiable factors that can be used in the prevention of dementia, including a growing interest in air pollution. (Livingston et al., 2020).

The pathophysiology of dementia is not fully understood, and deposits and plaques of amyloid beta and tau proteins, brain inflammation, and brain changes induced by vascular damage have been hypothesized to play a role (Jankowska-Kieltyka et al., 2021). Exposure to particulate matter (PM) leads to systemic- and neuro-inflammation, oxidative stress, increased mental load, respiratory and cardiovascular dysfunction, and sleep disturbances, all of which decrease the efficiency of glymphatic clearance, thereby elevating the risk of Alzheimer’s disease and dementia (Jankowska-Kieltyka et al., 2021). Air pollution has been shown to adversely affect cognitive function and accelerate cognitive decline, and lead to changes in brain structure that are linked to increased risk of developing dementia (Jankowska-Kieltyka et al., 2021). In line with these understandings of the underlying pathophysiology that can be related to air pollution, the recent meta-analyses summarizing epidemiological evidence on long-term exposure to air pollution and dementia show clear links with PM of diameter < 2⋅5 µm (PM2⋅5), but more limited and mixed evidence for nitrogen dioxide (NO2) (Wilker et al., 2023). Still, some of the most recent studies published since the meta-analyses from Wilker et al. (Wilker et al., 2023) failed to detect associations of air pollution with dementia incidence or mortality (Andersen et al., 2022; Blanco et al., 2024). Furthermore, there are only a few studies on dementia and black carbon (BC), an important traffic-related pollutant, with one showing associations (Shi et al., 2023), and others finding none (Blanco et al., 2024; Mortamais et al., 2021; Cerza et al., 2019; Yuchi et al., 2020). Finally, less is known regarding the characteristics of those most susceptible to air pollution with respect to dementia risk, with mixed and inconclusive results on sex, age, comorbidities, and socio-economic status (SES), which may influence dementia risk related to air pollution due to differences in biological responses, exposure levels, and access to preventive resources.

In this study, we aimed to examine whether long-term exposure to single air pollutants, including PM2⋅5, NO2, and BC, are associated with incidence of dementia as well as to identify the most vulnerable groups in terms of age, sex, SES, and comorbidities with major chronic cardio-metabolic and respiratory diseases.

2. Material and methods

2.1. Study population

We obtained information on 1,048,522 adults from the Danish nationwide administrative cohort who were 60 years or older and had lived in Denmark for a minimum of one year prior to January 1st, 2000 (details in the Supplementary Text S1). The collected information encompassed individual-level demographic details and SES (in 1999), including sex, age, country of origin, household income, levels of education, employment status, and marital status. Additionally, the data also included SES indicators at the area level, such as average household income and unemployment rate both at parish- and regional-levels (in 2001). Individuals were also linked to the Danish Civil Registration System (Pedersen, 2011) to obtain information on date of death and emigration outside Denmark, and historical information on residential addresses until December 31, 2018.

2.2. Definition of incident dementia

Dementia incidence was determined by the date of first-ever hospital contact including inpatient, outpatient, or emergency room with primary discharge diagnosis code of International Classification of Disease (ICD) version 8 (ICD-8) (used until 1994): 290.09–290.19; 293.09–19, and ICD-10 (used since 1994): F00-F03; G30; G31.8–9 or the date of the first redeemed prescription of anti-dementia drugs (Anatomical Therapeutic Chemical (ATC) code: N06D), whichever came first, between cohort baseline (January 1, 2000) and December 31, 2018. Individuals with the above mentioned hospital contact or prescription before baseline were considered as prevalence cases and excluded from the study.

We collected ICD codes and ATC codes through linking to the Danish National Patient Register (Schmidt et al., 2015), which has maintained all Danish hospital discharge diagnoses since 1977, and the Danish National Prescription Registry (Pottegård et al., 2017), which has recorded all prescriptions dispensed at Danish pharmacies since 1995, respectively.

2.3. Assessment of air pollution exposure

Annual mean exposure levels of PM2⋅5, NO2, and BC at the individuals’ geocoded residential addresses at baseline (31 December 1999) were estimated with European-wide hybrid land-use regression (LUR) models for the year 2010 with 100 × 100 m resolution, which were developed and validated within the framework of the ELAPSE (Effects of Low-level Air Pollution – A Study in Europe) project and a detailed description of this work is available elsewhere (de Hoogh et al., 2018). In summary, the European-wide hybrid LUR models were developed utilizing data from the European Environment Agency Air-Base routine monitoring of PM2⋅5 and NO2 in 2010 and data of BC monitoring from the European Study of Cohorts for Air Pollution Effects (ESCAPE) project. Annual average PM2⋅5 absorbance during 2009 and 2010, measured using filter reflectance, were treated as annual mean concentrations of BC for 2010. The LUR models included various predictors such as land cover and road length, satellite-derived air pollution data, and estimates from dispersion models. The validation revealed that the LUR models could explain 66 % of the spatial variations in the measured concentrations of PM2⋅5, 58 % for NO2, and 51 % for BC. The year 2010 was selected for the year for air pollution assessment modelling in the ELAPSE project, as this was the earliest year with sufficiently extensive PM2⋅5 monitoring data available across Europe. To maintain consistency, 2010 was also chosen for NO2.

For the sensitivity analysis where we allowed exposures to vary annually with accounting for residential mobility, annual average levels of PM2⋅5, NO2, and BC for each year of the following up period were also calculated using back- and forward- extrapolation of the 2010 models using data from the Danish Eulerian Hemispheric Model at 26 km × 26 km resolution to determine a global correction factor per year. Detailed descriptions for the method can be found in the Supplementary Text S2.

2.4. Statistical analyses

The methodology in this study closely followed the framework by the ELAPSE project (Brunekreef et al., 2021). We employed Cox proportional hazard models, using age as the underlying time scale, to investigate the association of long-term exposure to air pollution with incident dementia. The models included air pollution exposure and covariates as constant factors over time. The follow-up commenced on 1 January 2000 and ended at either the date of the incident dementia, or censored at the date of death, emigration, other forms of loss to follow-up, or on 31 December 2018, whichever happened first. All analyses (unless specified separately) considered data clustering within the same parish (the smallest administrative area in Denmark and available in the current database of this study) which adjusted the variance of the effect estimates. We assessed proportional hazard assumption for covariates and pollutants in the main model utilizing a statistical test for the scaled Schoenfeld residuals and the shapes of curves of the scaled Schoenfeld residuals against time.

We analysed the associations of air pollutants with dementia incidence with sequentially adjusted models with priori-selected confounders: Model 1 used age as underlying time scale and stratified by sex; Model 2 incorporated additional adjustments for individual-level demographic and SES factors, such as household income categorized into deciles, employment status [unemployed, employed, or having support/pension (sick or cash support, under education, pension, others)], origin (Danish, immigrants/descendants from Western, or those from non-Western countries), marital status (unmarried, divorced, widowed, or in a marriage/registered partnership), and the highest attained educational level (primary, upper secondary, vocation/qualifying, vocation bachelor/ short-cycle higher education, or levels higher than college); Model 3 (considered as our main model) further included adjustments for regional level of average household income and unemployment rate, and the absolute difference in these metrics between parish and region.

To visualize the association of long-term exposure to air pollution with incident dementia, we incorporated a natural cubic spline term for each pollutant with two degrees of freedom into the main model (Model 3) instead of the linear term, without the clustering term of parish due to computing time.

We examined whether the effects of the associations between air pollutants (modelled as linear) and incident dementia are modified by incorporating an interaction term for each air pollutant with a possible effect modifier in the main model, and then evaluated these using the Wald test. The interactions were assessed on the multiplicative scale. The potential effect modifiers included age at baseline grouped in tertile (<66⋅6 years, 66⋅6–75⋅1 years, or > 75⋅1 years), sex, household income categorized into quintiles, educational level (<college level or ≥ college level), occupational status (not employed or employed), comorbidity condition of each chronic diseases including stroke, ischemic heart disease (IHD), myocardial infarction (MI), asthma, chronic obstructive pulmonary diseases (COPD), diabetes, and depression (details for definition of these in supplementary Table S1). In addition, we estimated associations with most common subtypes of dementia, Alzheimer’s disease and vascular dementia based on our main model (model 3).

We performed several sensitivity analyses to check robustness of the results from the main models. 1) We used different definitions of dementia by varying sources: either hospitalization or medication, only on hospitalization, or only on medication. 2) To check the impact of health behaviour-related factors which is not captured in the nationwide administrative cohort, such as smoking status, body mass index (BMI), and levels of alcohol consumption, we applied an indirect adjustment method (Shin et al., 2014) by which hazard ratios (HRs) from Model 3 were mathematically recalibrated by using the data from the Danish National Health Survey (DNS) (Christensen et al., 2022) in 2010 and 2013. Details for the method is included in the Supplementary Text S3. Briefly, this method required two types of information: firstly, the associations between the unmeasured factors and the measured covariates in the main model; secondly, the risk of dementia associated with those unmeasured factors. The first set of information was derived from the DNS in 2013, and the second set from the DNS in 2010. The information on demographics, SES and air pollution exposure was also obtained and linked to the DNS as our main population dataset. 3) We additionally adjusted for another pollutant in the main model. 4) We relaxed the proportional hazard assumption by allowing the effects of continuous covariates (area-level SES) to vary a certain period of time (age) with time-varying coefficients and by applying different baseline rates across categorical variables (sex, household income in decile, occupational, and marital status, educational level) by including them as strata terms. 5) We compared HRs based on the 2010 exposure estimates with those back-extrapolated to the baseline year, and with those in a time varying analyses with back- and forward-extrapolated exposure estimates with 5-year calendar year strata term. 6) We examined the effect modification on the additive scale as well (The detailed method in the Supplementary Text S4). 7) We run the Fine-Gray models on a 5 % stratified sample, accounting for sex and dementia incidence proportions, to evaluate the impact of competing event for dementia (death). 8) We restricted analyses of main model (Model 3) to participants free from dementia at 2010 January 1 to examine the robustness of our finding to using the exposure models based on 2010 measurement. 9) We used time since baseline as time-scale instead of age and adjusted for age in our main model (Model 3).

We presented the effects of air pollution exposure as HRs with 95 % confidence intervals (CI) per interquartile range (IQR) of each pollutant.

All statistical analyses and graphical presentations were conducted using R version 4.3.1 (R Project for Statistical Computing). We used coxph function from R package “survival” for using Cox model and ns function from “splines”. Analyses were conducted for the study population dataset with complete information on covariates in the main model and exposure to air pollutants.

3. Results

3.1. Analytical sample

From 1,048,522 individuals aged 60 years or older, we excluded: 9,636 with dementia diagnosis prior to cohort baseline on January 1st, 2000; 1,616 with missing on parish; 5 with missing on immigrant status; and 102,473 with missing on geocoding; resulting in a total of 934,792 participants for the final analysis (graphical presentation in Supplementary Fig. S1). The 113,730 excluded individuals were less likely to be older, more likely to be woman, be on welfare supports, be unmarried, divorced, or widowed, live in more deprived area with lower mean household income and higher unemployment rate, had lower household income and education, in comparison to those included in the analyses (N = 934,792) (Table S2 in the Supplementary material).

3.2. Descriptive statistics

Of 934,792 individuals, 81,731 developed dementia during a mean follow-up of 11⋅6 years. Participants who developed dementia were older, more likely to be women, unemployed or on welfare support or pension, of western country origin, married or widowed, highly educated, and have higher levels of air pollution at residence, than those who did not develop dementia by the end of the follow-up (Table 1). Mean levels of PM2⋅5, NO2, and BC were 12⋅5 and 20⋅6 µg/m3, and 1⋅0 × 10−5/m, respectively. Correlation between PM2⋅5 and NO2 was 0⋅58 (Table 2), and correlation of BC with PM2⋅5 and NO2 was 0⋅69 and 0⋅89, respectively.

Table 1.

Demographic and socio-economic characteristics for 934,792 individuals from the Danish nationwide administrative cohort, 60 years or older on January 1st, 2000, by dementia status at the end of follow-up on December 31, 2018.

Characteristics Total (n = 934,792) Without dementia (n = 853,061) With dementiaa (n = 81,731)
Age at baseline (years), mean ± SD 71.8 ± 8.3 71.8 ± 8.4 72.5 ± 7.0
Sex, n (%)
Men 408,985 (43⋅8) 378,025 (44⋅3) 30,960 (37⋅9)
Women 525,807 (56⋅2) 475,036 (55⋅7) 50,771 (62⋅1)
Household income (DKK), mean ± SD 140,388⋅79 ± 130,192⋅30 140,251⋅44 ± 131,404⋅66 141,822⋅35 ± 116,780⋅77
Occupational status, n (%)
Unemployed 5,442 (0⋅6) 5,139 (0⋅6) 303 (0⋅4)
Having support or pension 812,925 (87⋅0) 738,781 (86⋅6) 74,144 (90⋅7)
Employed 116,425 (12⋅5) 109,141 (12⋅8) 7,284 (8⋅9)
Immigrant status, n (%)
Danish origin 901,707 (96⋅5) 822,824 (96⋅5) 78,883 (96⋅5)
Western country of origin 22,249 (2⋅4) 20,113 (2⋅4) 2,136 (2⋅6)
Non-western country of origin 10,836 (1⋅2) 10,124 (1⋅2) 712 (0⋅9)
Marital status, n (%)
Unmarried 52,965 (5⋅7) 49,443 (5⋅8) 3,522 (4⋅3)
Divorced 90,546 (9⋅7) 83,152 (9⋅7) 7,394 (9⋅0)
Widowed 264,490 (28⋅3) 240,397 (28⋅2) 24,093 (29⋅5)
Married/registered partnership 526,791 (56⋅4) 480,069 (56⋅3) 46,722 (57⋅2)
Highest complete education level, n (%)
Primary 606,861 (64⋅9) 554,780 (65⋅0) 52,081 (63⋅7)
Upper secondary 7,456 (0⋅8) 6,754 (0⋅8) 702 (0⋅9)
Vocation/qualifying 220,614 (23⋅6) 201,085 (23⋅6) 19,529 (23⋅9)
Vocation bachelors/ short-cycle higher education 74,838 (8⋅0) 67,933 (8⋅0) 6,905 (8⋅4)
College level and over 25,023 (2⋅7) 22,509 (2⋅6) 2,514 (3⋅1)
Regional average household income/unemployment rate in 2001, n (%)
North Denmark: 155,994⋅2/2⋅55 104,161 (11⋅1) 97,058 (11⋅4) 7,103 (8⋅7)
South Denmark: 158,840⋅3/1⋅87 213,066 (22⋅8) 191,758 (22⋅5) 21,308 (26⋅1)
Central Denmark: 162,312⋅0/1⋅90 197,567 (21⋅1) 182,663 (21⋅4) 14,904 (18⋅2)
Zealand: 166,371⋅4/1⋅86 143,842 (15⋅4) 133,306 (15⋅6) 10,536 (12⋅9)
Capital region: 175,561⋅2/1⋅90 276,156 (29⋅5) 248,276 (29⋅1) 27,880 (34⋅1)
Parish level average 164,078⋅32 ± 164,035⋅24 ± 164,528⋅00 ±
income in 2001, mean ± SD 27,433⋅99 27,351⋅44 28,277⋅47
Parish level unemployment rate, mean ± SD 1⋅92 ± 0⋅67 1⋅93 ± 0⋅68 1⋅93 ± 0⋅65
PM2⋅5, mean ± SD (µg/m3) 12⋅46 ± 1⋅51 12⋅44 ± 1⋅51 12⋅65 ± 1⋅51
NO2, mean ± SD (µg/m3) 20⋅61 ± 7⋅85 20⋅51 ± 7⋅82 21⋅7 ± 8⋅07
BC, mean ± SD (10−5/m) 1⋅03 ± 0⋅36 1⋅02 ± 0⋅36 1⋅08 ± 0⋅37
Open in a new tab

Abbreviations: SD – standard deviation; DKK – Danish Krone; PM2⋅5 – Particulate matter aerodynamic diameter < 2⋅5 µm; NO2 – Nitrogen dioxide; BC – Black carbon.

a

Definition of having dementia were based on record of hospital contact or use of medication (see the Method for detail).

Table 2.

Descriptive characteristics for annual mean exposure to air pollutants at individuals’ residential addresses at cohort baseline.

Pollutants Mean ± SD Interquartile range Min. Percentile
 Max. Spearman’s rank correlation coefficients (ρ)
5th 25th 50th 75th 95th PM2⋅5 NO2
PM2⋅5, µg/m3 12⋅5 ± 1⋅5 1⋅9 5⋅9 10⋅1 11⋅4 12⋅4 13⋅3 15⋅3 19⋅6
NO2, µg/m3 20⋅6 ± 7⋅9 10⋅2 4⋅0 9⋅7 15⋅0 19⋅1 25⋅2 35⋅6 72⋅2 0.58
BC, 10−5/m 1⋅0 ± 0⋅4 0⋅5 0⋅2 0⋅6 0⋅7 0⋅9 1⋅3 1⋅7 3⋅7 0⋅69 0⋅89
Open in a new tab

Abbreviations: SD – standard deviation; PM2⋅5 – Particulate matter aerodynamic diameter < 2⋅5 µm; NO2 – Nitrogen dioxide; BC – Black carbon.

The minimum and maximum values represent the averages of the lowest and highest five values, respectively, in accordance with the General Data Protection Regulation.

3.3. The association between air pollution and dementia incidence

We detected strong and statistically signification associations between each of the three pollutants and incidence of dementia, even after adjusting for individual and area-level socio-demographic characteristics in Model3, with HR of 1⋅14 (95 % CI: 1⋅12, 1⋅16) per 1⋅9 μg/m3 increase in PM2⋅5, 1⋅25 (95 % CI: 1⋅22, 1⋅28) per 10⋅2 μg/m3 increase in NO2, and 1⋅23 (95 % CI: 1⋅20, 1⋅26) per 0⋅5 × 10−5/m increase in BC in the Model3 (Table3). From Model 1 to Model 3, associations attenuated slightly for PM2⋅5, increased slightly for NO2, and were unchanged for BC.

Associations with PM2⋅5 and NO2 were linear, showing associations at the lowest levels and across the entire exposure range, while association with BC showed levelling off at around 2⋅0 × 10−5/m for BC (Fig. 1).

Fig. 1. Exposure-response curve for the association between long-term exposure to air pollution and incidence of dementia in the older population of the Danish administrative cohort (81,731 cases among 934,792 subjects).

Fig. 1.

Open in a new tab

Abbreviations: HR – Hazard ratio; CI – Confidence interval; PM2⋅5 – Particulate matter aerodynamic diameter < 2⋅5 µm; NO2 – Nitrogen dioxide; BC – Black carbon. Note: Natural cubic splines with two degrees of freedom were applied to air pollutants to assess the shape of the associations using the main model (Model 3) considering sex in a strata term and adjusted for household income, employment status, immigrant status, marital status, and highest level of education completed, regional average household income and unemployment rate, and the difference in average household income and unemployment rate between parish and region. Solid lines indicate hazard ratio values, and gray shaded areas indicate their 95 % confidence intervals. Dashed horizontal lines are the HRs equal to 1. Histograms with gray lines indicate the distribution of air pollutants.

3.4. Effect modification

We detected statistically significant effect modification of the association between all three pollutants and incident dementia by age, showing strongest associations in those older than 75 years [e.g., per 1⋅9 μg/m3 for PM2⋅5, HR: 1.21 (95 % CI, 1.18–1.25) for aged ≥ 75 years, HR: 1.13 (95 % CI, 1.11–1.15) for aged 66–75 year, and HR: 1.09 (95 % CI, 1.07–1.12) for aged < 66], but no difference in the estimated associations by sex (Fig. 2 and more numerical results in Supplementary Table S3). We also found that associations of all three pollutants with dementia were stronger in those with the lower household income [e.g., per 1⋅9 μg/m3 for PM2⋅5, HR: 1.17 (95 % CI, 1.14–1.21) for those with lowest income, and HR: 1.10 (95 % CI, 1.07–1.12) for those with highest income], lower education [e.g., per 1⋅9 μg/m3 for PM2⋅5, HR: 1.15 (95 % CI, 1.13–1.17) for those with < college level, and HR: 1.09 (95 % CI, 1.04–1.14) for those with ≥ college level], and those not employed [e.g., per 1⋅9 μg/m3 for PM2⋅5, HR: 1.15 (95 % CI, 1.13–1.17) for those not employed, and HR: 1.08 (95 % CI, 1.05–1.11) for those employed] at the cohort baseline. Stroke patients showed statistically significantly stronger associations of all three air pollutants with dementia than those who did not have stroke prior to cohort entry [e.g., per 1⋅9 μg/m3 for PM2⋅5, HR: 1.25 (95 % CI, 1.18–1.31) for those having stroke, and HR: 1.14 (95 % CI, 1.12–1.16) for those not], whereas there was no difference in association by prior IHD, MI, asthma, COPD, diabetes or depression.

Fig. 2. Effect modification of the association between long-term exposure to air pollution and risk of dementia by baseline characteristics in the older population of the Danish administrative cohort (81,731 cases among 934,792 subjects).

Fig. 2.

Open in a new tab

Abbreviations: HR – Hazard ratio; PM2⋅5 – Particulate matter aerodynamic diameter < 2⋅5 µm; NO2 – Nitrogen dioxide; BC – Black carbon; IHD – Ischemic heart disease; MI – Myocardial Infarction; COPD – Chronic obstructive pulmonary diseases. Note: Associations were derived from models where a product term between a potential effect modifier and a pollutant were included in the main model (Model 3) considering sex in strata term and parish level in cluster variable and adjusted for household income, employment status, immigrant status, marital status, and highest education level completed, regional average household income and unemployment rate, and the difference in average household income and unemployment rate, between parish and region. Black dots indicate hazard for the following interquartile range increments: 1⋅9 μg/m3 for PM2⋅5, 10⋅2 μg/m3 for NO2, and 0⋅5 × 10−5/m for black carbon. Whiskers from those points indicate the corresponding 95 % confidence intervals. The star sign (*) indicates the statistical significance of Wald test for the interaction term (p < 0⋅05). Numerical results are in the supplementary Table S3.

3.5. Sensitivity analyses

Associations with dementia incidence defined based on hospital contacts only [HR: 1⋅15 (1⋅11, 1⋅19) for PM2⋅5; 1⋅33 (1⋅28, 1⋅39) for NO2; 1⋅31 (1⋅26, 1⋅37) for BC] were substantially stronger than those based on medication only [HR: 1⋅11 (1⋅09, 1⋅13) for PM2⋅5; 1⋅19 (1⋅17, 1⋅23) for NO2; 1⋅17 (1⋅15, 1⋅20) for BC] (Supplementary Table S4). In the 51,100 dementia cases defined based on hospital contact, 23,286 had Alzheimer’s disease and 5,457 had vascular dementia: strong and comparable associations were found with both types of dementia [HR for Alzheimer’s disease: 1⋅13 (1⋅07, 1⋅18) for PM2⋅5, 1⋅33 (1⋅26, 1⋅41) for NO2, 1⋅29 (1⋅22, 1⋅37) for BC; HR for vascular dementia: 1⋅12 (1⋅05, 1⋅20) for PM2⋅5, 1⋅27 (1⋅18, 1⋅37) for NO2, 1⋅24 (1⋅16, 1⋅33) for BC] (Supplementary Table S5). Associations with all three pollutants remained robust after additional indirect adjustment for lifestyle factors (smoking, obesity, and alcohol consumption), even for subtypes of dementia (Supplementary Table S6). Notably, in the DNS 2013 air pollution was positively associated with smoking and high risk alcohol drinking and negatively with overweight (Supplementary Table S7). However, smoking status and alcohol drinking were not associated with the risk of dementia, observed in the DNS 2010 (Supplementary Table S8). In two pollutant models, associations with PM2⋅5 were substantially attenuated after adjusting for NO2 or BC [1⋅05 (1⋅02, 1⋅07) and 1⋅05 (1⋅02, 1⋅07), respectively], but remained statistically significant, whereas associations with NO2 and BC were robust after adjustment for PM2⋅5 [1⋅21 (1⋅17, 1⋅25) and 1⋅18 (1⋅14, 1⋅21), respectively] (Supplementary Table S9). The proportional hazards assumption was not upheld for education levels, household income, marital status, occupational status, regional mean household income, regional unemployment rate (Data not shown), but the associations remained robust and proportional hazard assumption were met when we stratified or allowed the effect of these covariates to vary each year (Supplementary Table S10). We found that the associations remained positive though slightly weaker, particularly for PM2.5 and NO2 when using back-extrapolated baseline year exposure (Supplementary Table S11). The results also remain robust when using extrapolated time-varying exposures incorporating residential history (Supplementary Table S12). The observed effect modifications on the multiplicative scale were generally robust to those on additive scale, except for previous diabetes that only showed statistically significant effect modification on the additive scale (Supplementary Table S13). Lastly, the observed associations were robust to considering the competing event (Supplementary Table S14), to restricting analyses to participants free from dementia at 1 January 2010 (Supplementary Table S15), and to using time since baseline as time-scale in Cox model (Supplementary Table S16).

4. Discussion

In this nationwide analyses of the Danish population aged 60 years and above, we found strong associations between long-term exposure to air pollution, PM2⋅5 as well as with NO2 and BC, and incidence of dementia. We presented novel results that those in the lowest SES groups are especially susceptible to air pollution exposure-related risk of dementia, along with elderly (>75 years) subjects and stroke patients.

Our findings on PM2⋅5 agree well with the international literature (Wilker et al., 2023), although our detected associations were much stronger than those reported in the meta-analyses by Wilker et al. (Wilker et al., 2023) based on 14 studies: HR of 1⋅15 (95 % CI, 1⋅13–1⋅17) from our study and 1⋅04 (95 % CI, 0⋅99–1⋅09) from Wilker et al. per 2 µg/m3 (unit used in Wilker et al) for both. The strong associations in our cohort are well in line with the Wilker et al. observation of a stronger association of PM2⋅5 with dementia in European studies [HR: 1⋅21 (95 % CI, 0⋅90–1⋅63) per 2 µg/m3] than in North American studies [HR: 1⋅03 (95 % CI, 0⋅98–1⋅08)] (Wilker et al., 2023), and with the strong associations of PM2⋅5 with a number of health outcomes in Denmark, including all-cause mortality as shown earlier in this administrative cohort (So et al., 2022). The stronger association in our study and studies for other health outcomes in Denmark could be due to lower air pollution levels in Denmark, where steeper slopes were observed from supra-linear shapes of exposure–response curves for the association between long-term air pollution and health outcomes (Brunekreef et al., 2021). Notably, our estimate [1⋅15 (95 % CI, 1⋅13–1⋅17) per 2 µg/m3] were larger than the summarized HR from studies used passive case ascertainment based on ICD code in insurance or medication records [HR: 1⋅03 (95 % CI, 0⋅98–1⋅07)], but smaller than the HR summarizing results from studies based on active case ascertainment including in-person assessment [HR: 1⋅42 (95 % CI, 1⋅00–1⋅20)] (Wilker et al., 2023).

We found strong and significant positive associations with NO2 [HR: 1⋅25 (1⋅22, 1⋅28) per 10 µg/m3], in contrast to Wilker et al. who reported no statistically significant association based on nine studies [HR: 1⋅02 (95 % CI, 0⋅98–1⋅06) per 10 µg/m3]. In agreement with our results on NO2, we also showed the strong association of BC with dementia suggesting relevance of traffic-related pollution. Our study makes an important contribution to the limited evidence base on BC and dementia risk, which corroborates those from two large US studies using Medicare data of millions of participants aged 65 years and older (Shi et al., 2023; Li et al., 2022), but disagrees with five others that detected no association with dementia (Andersen et al., 2022; Blanco et al., 2024; Mortamais et al., 2021; Cerza et al., 2019; Yuchi et al., 2020). These studies with no association included a pooled European cohorts study of approximately 270,000 participants using mortality data for dementia definition (Andersen et al., 2022); the Seattle-based prospective cohort study of about 4,200 subjects aged 75 years and older (Blanco et al., 2024), a French population-based cohort of 7,066 subjects evaluating cognition function every two years using the ELAPSE exposure model for BC estimates (Mortamais et al., 2021), and a Canadian population-based cohort of about 678,000 individuals aged 45–84 years old with follow-up of 4 years (Yuchi et al., 2020).

We found that subjects with prior stroke were the most susceptible to dementia risk related to all three air pollutants (PM2⋅5, NO2, and BC). Our findings are supported by a Taiwanese study in stroke patients showing that air pollution exposure after stroke increased risk of dementia (Lee et al., 2023). Similarly, a Swedish study found that stroke patients had the highest risk of developing dementia, and that approximately 50 % of the association between air pollution and dementia risk was mediated by stroke (Grande et al., 2020). On the other hand, in a London-based study, Carey et al. found no difference in air pollution related risk in those with and without composite score of comorbidity with ischaemic heart disease (IHD), stroke, heart failure, and diabetes (Carey et al., 2018). Furthermore, a large Canadian study detected significantly stronger association of PM2⋅5 with dementia in those without prior stroke than those with, and no difference in estimated associations with and without comorbidity of stroke was found for NO2 (Chen et al., 2017). Similarly, and in agreement with our findings, a large US study in women found no difference in association of air pollution with dementia by comorbidities of diabetes, cardiovascular disease, hypercholesterolemia, or hypertension (Chen et al., 2017), and a Hong-King study, found no difference by comorbidities of diabetes, hypertension, or heart disease (Ran et al., 2021). Thus, evidence so far suggests that air pollution increases susceptibility towards developing dementia, especially in stroke patients, but not in cardio-metabolic or respiratory disease patients.

While a study with US Medicare population have found suggestive evidence of stronger association in the older people (aged ≥ 80 years) (Shi et al., 2020) which is in lines with stronger association in our study (for those aged ≥ 75 years), others have found no difference by age (Carey et al., 2018; Wang et al., 2022; Ran et al., 2021). The Difference in the association with air pollution by sex was also explored in a few studies but results remained inconclusive, with some suggesting that associations with dementia were limited to women (Shi et al., 2020) or slightly stronger in women (Chen et al., 2017), other showing slightly stronger associations in men (Carey et al., 2018), or no difference (Ran et al., 2021; Parra et al., 2022). Finally, we presented the novel finding of the highest susceptibility to air pollution in those with lowest SES, in terms of low income, low education, or unemployment. This has been rarely explored previously, only in two studies: a study in London found no difference in estimated associations between air pollution exposure and dementia by area-level deprivation index (Carey et al., 2018), whereas Wang et al. found no difference in the association of air pollution with dementia in US women by education (Wang et al., 2022). PM2⋅5 in Denmark originates primarily (more than 50 %) from long-range transported secondary particles from Eastern and Central Europe, followed by a smaller contribution of local sources. Of the local sources, the largest contribution is from biomass burning for residential heating, traffic, industrial activities, and energy production. Thus, our results from two-pollutant models, with most robust results for NO2 and BC, might suggest relevance of combustion sources, including traffic, for dementia development, but demand some caution due to correlation between PM2⋅5 and NO2 (ρ = 0⋅58).

Our analyses on subtypes of dementia based on hospital contacts only, as use of anti-dementia medication does not reliably differentiate between subtypes of dementia. Dementia diagnosis based on the 1st hospital contacts from the Danish National Patient Register are highly valid (Phung et al., 2007). Alzheimer’s disease defined based on the Danish Register had good validity even though it is underdiagnosed and underregistered, whereas validity of other subtypes, including vascular dementia, is relatively low, hence demanding some caution (Phung et al., 2007). We found comparable associations of air pollution with two most common subtypes of dementia (Alzheimer’s disease and vascular dementia), suggesting that air pollution pathophysiology is relevant for these major dementia subtypes (Supplementary Table S5), in agreement with a French study (Mortamais et al., 2021) and UK Biobank study (Parra et al., 2022), but in contrast to Cerza et al. who found associations only with vascular dementia, and none with overall of Alzheimer’s disease (Cerza et al., 2019). Findings of stronger associations with dementia incidence diagnosed by first hospital contacts than those based on first-ever dementia medication (Supplementary Table S4), are likely explained due to more precise and certain diagnosis determined by the neurologist at the hospital, with more sophisticated diagnostic tools available, as opposed to dementia cases defined by medication prescription only that included those prescribed by general practitioners. This result could also suggest stronger association of air pollution with more severe, or more advanced dementia cases that are more likely to be diagnosed by specialists at hospitals, or could indicate that air pollution may accelerate dementia progression.

The main strength of our study is the large population size of Danish adults 60 years or older in 2000, with up to 19 years of follow-up. Furthermore, we benefited from high-resolution air pollution data (100×100 m), validated definition of dementia incidence, and detailed data on SES and co-morbidities, from unique Danish nationwide registers on hospitalization, medication, education, and employment. Lastly, the findings of this study remain consistent across various sensitivity analyses, reinforcing the evidence for the association between air pollution and neurodegeneration.

Our study is limited to the lack of information on exposure to air pollution during any outdoor activity and related commuting. Additionally, we were unable to address possible confounding by road-traffic noise, which was suggested to be related to dementia (Cantuaria et al., 2021), whereas others found no association (Yuchi et al., 2020; Carey et al., 2018). Moreover, our main exposure assessment relied on models for the year 2010, which were applied to participants’ baseline addresses in 2000. This approach could potentially introduce exposure misclassification and might not fully capture the effect of long-term exposure to air pollution, particularly if there were spatiotemporal trends in pollutant levels. We chose this approach because of the unavailability of sufficient monitoring stations to develop models at fine-scale for PM2⋅5 in earlier years. de Hoogh et al. (2018) performed a sensitivity analysis whereby separate LUR models were developed for NO2 for the year 2000 and for PM2.5 for the year 2003. In Denmark the predictions from the 2010 model showed a high correlation with the 2000 model for NO2 (R2 > 84 %) and a bit lower with the 2013 model for PM2⋅5 (R2 = 49 %) (de Hoogh et al., 2018), suggesting that spatial contrasts remain stable over time, especially for NO2, and supporting the use of 2010 exposure in the analysis. In addition, sensitivity analyses, where we considered the temporal change or trend in pollution at a coarse spatial scale and residential mobility, showed that the associations remained positive though slightly weaker (Supplementary Table S11) and (Supplementary Table S12). Additionally, like many other cohort studies on dementia that rely on administrative databases, our study faced the limitation of not having individual-level data on lifestyle risk factors. Despite this, we managed to make indirect adjustment for these unmeasured factors, including smoking status, BMI, and alcohol consumption, showing minimal impact on the strength of the estimated association. However, it is still important to mention that the indirect adjustment method we used in this study is significantly influenced by how well the ancillary dataset (DNS 2013) reflects the main study dataset (Shin et al., 2014). The demographic and socioeconomic distribution of DNS 2013 has similar trend to those of the Danish nationwide administrative cohort in terms of age, sex, employment, immigrant, and marital status, and region of residence (Table S15 in supplementary material). However, participants in the DNS 2013 were more likely to have higher education and income as expected in survey participants. Consequently, this means the association between lifestyle factors and air pollution could be different in our main study population than that observed in the DNS 2013, which might introduce a bias in the results of the indirect adjustment. It has been suggested that lifestyle factors mediate the associations between SES and dementia (Deckers et al., 2019). We extensively adjusted for SES and showed minimal impact of lifestyle factors on dementia (Supplementary Table S8), so we expect minimal confounding by lifestyle factors. In addition, the follow-up in our study was limited to year 2018 based on availability of the hospital records, thus we could not incorporate the possibility of a difference in more recent trends on the association between air pollution and dementia incidence. Furthermore, we should note that our reported HRs are assumed to be constant over follow-up time (with age in our study as time-scale), potentially leading to misleading conclusions since the size of the HR can vary by the length of follow-up time. We included average household income and unemployment rate at regional level along with the deviation at parish levels, but this approach could overlook some other aspects of area-level deprivation, such as crime rate and housing conditions. Therefore, we cannot entirely rule out residual confounding. Furthermore, this approach also may not fully account for the spatial confounding. Additionally, we excluded about 11 % due to missing information, implying that the analysis may not fully reflect the Danish population as intended. Lastly, in our study, dementia cases were defined based on records from hospitalization and medication, so we might miss mild cases of dementia or cognitive decline which could lead to dementia, suggesting that our reported associations between air pollution and dementia might be underestimated.

In conclusion, in an analysis of the Danish population aged 60 years or older followed between 2000 and 2018, at low levels of air pollution setting, we observed strong associations between long-term exposure to all major pollutants (PM2⋅5 as well as traffic-related NO2 and BC) and incidence of dementia. We found strong associations for both dementia subtypes, Alzheimer’s disease and vascular dementia. Furthermore, we found that subjects aged 75 years or older, those who had stroke, and those who were most socio-economically deprived were the groups most susceptible to adverse effects of air pollution with respect to dementia, and who may benefit from targeted information and prevention efforts.

Supplementary Material

1

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envint.2025.109607.

Table 3.

Associations between long-term exposure to air pollution and risk of dementia in the older population of the Danish administrative cohort.

Outcome Air pollutant Model1a
HR (95 % CI)		 Model 2b
HR (95 % CI)		 Model 3c
HR (95 % CI)
Dementiad (81,731 cases among 934,792 subjects) PM2⋅5 1⋅19 (1⋅17, 1⋅21) 1⋅18 (1⋅16, 1⋅20) 1⋅14 (1⋅12, 1⋅16)
NO2 1⋅20 (1⋅18, 1⋅23) 1⋅20 (1⋅18, 1⋅23) 1⋅25 (1⋅22, 1⋅28)
BC 1⋅23 (1⋅20, 1⋅25) 1⋅23 (1⋅20, 1⋅25) 1⋅23 (1⋅20, 1⋅26)
Open in a new tab

Abbreviations: HR – Hazard ratio; CI – Confidence interval; PM2⋅5 – Particulate matter aerodynamic diameter < 2⋅5 µm; NO2 – Nitrogen dioxide; BC – Black carbon. Note: Results are presented as hazard ratio and 95 % confidence interval for the following interquartile range increments: 1⋅9 μg/m3 for PM2⋅5, 10⋅2 μg/m3 for NO2, and 0⋅5 × 10−5/m for black carbon.

a

Model 1 accounted for sex (strata) and parish levels(cluster term).

b

Model 2: Model 1 further adjusted for household income, employment status, immigrant status, marital status, and the highest level of education completed.

c

Model 3: Model 2 further adjusted for regional levels of average household income and unemployment rate, and the disparities of average household income and unemployment rate between parish and region

d

Dementia was defined based on records from hospitalisation and medication (see the Method part for details).

Funding sources

This work was supported by funding from HEI Research Agreement (#4954-RFA14–3/16–5-3), jointly funded by the United States Environmental Protection Agency (EPA) (Assistance Award No. R-82811201) and certain motor vehicle and engine manufacturers, by the Novo Nordisk Foundation Challenge Programme: Harnessing the Power of Big Data to Address the Societal Challenge of Aging (#NNF17OC0027812), and by National Institutes of Health (NIH) for the project “Integrating Information about Aging Surveys: Novel Integration of Contextual Data to Study Late-Life Cognition and Alzheimer’s Disease and Related Dementia and Dementia Care” (3R01AG030153–17S1). The funders had no role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication.

Footnotes

Data statement

The authors do not have permission to share data.

CRediT authorship contribution statement

Zorana J. Andersen: Writing – original draft, Conceptualization. Youn-Hee Lim: Writing – review & editing, Data curation. Jiawei Zhang: Writing – review & editing, Data curation. Stéphane Tuffier: Writing – review & editing. Thomas Cole-Hunter: Writing – review & editing. Marie Bergmann: Writing – review & editing. Steffen Loft: Writing – review & editing. Laust H. Mortensen: Writing – review & editing. Jie Chen: Writing – review & editing, Methodology. Massimo Stafoggia: Writing – review & editing. Kees de Hoogh: Methodology. Klea Katsouyanni: Writing – review & editing, Methodology. Danielle Vienneau: Writing – review & editing, Methodology. Sophia Rodopoulou: Writing – review & editing, Methodology. Evangelia Samoli: Writing – review & editing, Methodology. Mariska Bauwelinck: Writing – review & editing. Jochem O. Klompmaker: Writing – review & editing. Richard Atkinson: Writing – review & editing. Nicole A.H. Janssen: Writing – review & editing. Bente Oftedal: Writing – review & editing. Matteo Renzi: Writing – review & editing. Francesco Forastiere: Writing – review & editing. Maciek Strak: Writing – review & editing. Lau C. Thygesen: Writing – review & editing. Paola Zaninotto: Writing – review & editing. Bert Brunekreef: Writing – review & editing, Funding acquisition. Gerard Hoek: Writing – review & editing, Funding acquisition. Rina So: Writing – original draft, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The authors do not have permission to share data.

References

  1. Andersen ZJ, Zhang J, Jørgensen JT, et al. , 2022. Long-term exposure to air pollution and mortality from dementia, psychiatric disorders, and suicide in a large pooled European cohort: ELAPSE study. Environ. Int 170, 107581. 10.1016/j.envint.2022.107581. [DOI] [PubMed] [Google Scholar]
  2. Blanco MN, Shaffer RM, Li G, et al. , 2024. Traffic-related air pollution and dementia incidence in the adult changes in thought study. Environ. Int 183, 108418. 10.1016/j.envint.2024.108418. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Brunekreef B, Strak M, Chen J, et al. , 2021. Mortality and morbidity effects of long-term exposure to low-level PM2.5, BC, NO2, and O3: an analysis of European cohorts in the ELAPSE project. Res. Rep. Health Eff. Inst 2021 (208), 1–127. [PMC free article] [PubMed] [Google Scholar]
  4. Cantuaria ML, Waldorff FB, Wermuth L, et al. , 2021. Residential exposure to transportation noise in Denmark and incidence of dementia: national cohort study. BMJ 374, n1954. 10.1136/bmj.n1954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Carey IM, Anderson HR, Atkinson RW, et al. , 2018. Are noise and air pollution related to the incidence of dementia? A cohort study in London, England. BMJ Open 8 (9), e022404. 10.1136/bmjopen-2018-022404. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cerza F, Renzi M, Gariazzo C, et al. , 2019. Long-term exposure to air pollution and hospitalization for dementia in the Rome longitudinal study. Environ. Health 18 (1), 72. 10.1186/s12940-019-0511-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Chen H, Kwong JC, Copes R, et al. , 2017. Exposure to ambient air pollution and the incidence of dementia: a population-based cohort study. Environ. Int 108, 271–277. 10.1016/j.envint.2017.08.020. [DOI] [PubMed] [Google Scholar]
  8. Chin KS, 2023. Pathophysiology of dementia. Aust. J. Gen. Pract 52 (8), 516–521. 10.31128/AJGP-02-23-6736. [DOI] [PubMed] [Google Scholar]
  9. Christensen AI, Lau CJ, Kristensen PL, et al. , 2022. The Danish National Health Survey: study design, response rate and respondent characteristics in 2010, 2013 and 2017. Scand. J. Public Health 50 (2), 180–188. 10.1177/1403494820966534. [DOI] [PubMed] [Google Scholar]
  10. de Hoogh K, Chen J, Gulliver J, et al. , 2018. Spatial PM2.5, NO2, O3 and BC models for Western Europe – evaluation of spatiotemporal stability. Environ. Int 120 (2), 81–92. 10.1016/j.envint.2018.07.036. [DOI] [PubMed] [Google Scholar]
  11. Deckers K, Cadar D, van Boxtel MPJ, Verhey FRJ, Steptoe A, Köhler S, 2019. Modifiable risk factors explain socioeconomic inequalities in dementia risk: evidence from a population-based prospective cohort study. J. Alzheimer’s Disease: JAD 71 (2), 549–557. 10.3233/JAD-190541. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. GBD, 2019. Dementia forecasting collaborators. Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. The Lancet Public health 7(2) (2022) e105–e125, doi: 10.1016/S2468-2667(21)00249-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Grande G, Ljungman PLS, Eneroth K, Bellander T, Rizzuto D, 2020. Association between cardiovascular disease and long-term exposure to air pollution with the risk of dementia. JAMA Neurol 77 (7), 801–809. 10.1001/jamaneurol.2019.4914. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Jankowska-Kieltyka M, Roman A, Nalepa I, 2021. The air we breathe: air pollution as a prevalent proinflammatory stimulus contributing to neurodegeneration. Front. Cell. Neurosci 15, 647643. 10.3389/fncel.2021.647643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Lee KW, Chung HW, Hsieh HM, et al. , 2023. Post-stroke dysphagia and ambient air pollution are associated with dementia. Front. Aging Neurosci 15, 1272213. 10.3389/fnagi.2023.1272213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Li J, Wang Y, Steenland K, et al. , 2022. Long-term effects of PM2.5 components on incident dementia in the northeastern United States. The Innovation 3 (2), 100208. 10.1016/j.xinn.2022.100208. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Livingston G, Huntley J, Sommerlad A, et al. , 2020. Dementia prevention, intervention, and care: 2020 report of the lancet commission. Lancet 396 (10248), 413–446. 10.1016/S0140-6736(20)30367-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Mortamais M, Gutierrez LA, de Hoogh K, et al. , 2021. Long-term exposure to ambient air pollution and risk of dementia: results of the prospective Three-City study. Environ. Int 148, 106376. 10.1016/j.envint.2020.106376. [DOI] [PubMed] [Google Scholar]
  19. Parra KL, Alexander GE, Raichlen DA, Klimentidis YC, Furlong MA, 2022. Exposure to air pollution and risk of incident dementia in the UK biobank. Environ. Res 209, 112895. 10.1016/j.envres.2022.112895. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Pedersen CB, 2011. The Danish civil registration system. Scand. J. Public Health 39 (7), 22–25. 10.1177/1403494810387965. [DOI] [PubMed] [Google Scholar]
  21. Phung TKT, Andersen BB, Høgh P, Kessing LV, Mortensen PB, Waldemar G, 2007. Validity of dementia diagnoses in the Danish hospital registers. Dement. Geriatr. Cogn. Disord 24 (3), 220–228. 10.1159/000107084. [DOI] [PubMed] [Google Scholar]
  22. Pottegård A, Schmidt SAJ, Wallach-Kildemoes H, Sørensen HT, Hallas J, Schmidt M, 2017. Data resource profile: the Danish national prescription registry. Int. J. Epidemiol 46 (3), 798. 10.1093/ije/dyw213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Ran J, Schooling CM, Han L, et al. , 2021. Long-term exposure to fine particulate matter and dementia incidence: a cohort study in Hong Kong. Environ. Pollut 271, 116303. 10.1016/j.envpol.2020.116303. [DOI] [PubMed] [Google Scholar]
  24. Schmidt M, Schmidt SAJ, Sandegaard JL, Ehrenstein V, Pedersen L, Sørensen HT, 2015. The Danish national patient registry: a review of content, data quality, and research potential. Clin. Epidemiol 7, 449–490. 10.2147/CLEP.S91125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Shi L, Wu X, Danesh Yazdi M, et al. , 2020. Long-term effects of PM(2⋅5) on neurological disorders in the American Medicare population: a longitudinal cohort study. Lancet Planet. Health. 4 (12), e557–e565. 10.1016/S2542-5196(20)30227-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Shi L, Zhu Q, Wang Y, et al. , 2023. Incident dementia and long-term exposure to constituents of fine particle air pollution: a national cohort study in the United States. PNAS 120 (1), e2211282119. 10.1073/pnas.2211282119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Shin HH, Cakmak S, Brion O, et al. , 2014. Indirect adjustment for multiple missing variables applicable to environmental epidemiology. Environ. Res 134, 482–487. 10.1016/j.envres.2014.05.016. [DOI] [PubMed] [Google Scholar]
  28. So R, Andersen ZJ, Chen J, et al. , 2022. Long-term exposure to air pollution and mortality in a Danish nationwide administrative cohort study: Beyond mortality from cardiopulmonary disease and lung cancer. Environ. Int 164, 107241. 10.1016/j.envint.2022.107241. [DOI] [PubMed] [Google Scholar]
  29. Wang X, Younan D, Millstein J, et al. Association of improved air quality with lower dementia risk in older women. In: Proceedings of the National Academy of Sciences of the United States of America. 2022;119(2). doi: 10.1073/pnas.2107833119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Wilker EH, Osman M, Weisskopf MG, 2023. Ambient air pollution and clinical dementia: systematic review and meta-analysis. BMJ 381, e071620. 10.1136/bmj-2022-071620. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Yuchi W, Sbihi H, Davies H, Tamburic L, Brauer M, 2020. Road proximity, air pollution, noise, green space and neurologic disease incidence: a population-based cohort study. Environ. Health 19 (1), 8. 10.1186/s12940-020-0565-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2105988-supplement-1.docx (121.5KB, docx)

Data Availability Statement

The authors do not have permission to share data.

RESOURCES