Air Pollution and Cognitive Impairment across the Life Course in Humans: A Systematic Review with Specific Focus on Income Level of Study Area

Cognitive function is a crucial determinant of human capital. The Lancet Commission (2020) has recognized air pollution as a risk factor for dementia. However, the scientific evidence on the impact of air pollution on cognitive outcomes across the life course and across different income settings, with varying levels of air pollution, needs further exploration. A systematic review was conducted, using Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) Guidelines to assess the association between air pollution and cognitive outcomes across the life course with a plan to analyze findings as per the income status of the study population. The PubMed search included keywords related to cognition and to pollution (in their titles) to identify studies on human participants published in English until 10 July 2020. The search yielded 84 relevant studies that described associations between exposure to air pollutants and an increased risk of lower cognitive function among children and adolescents, cognitive impairment and decline among adults, and dementia among older adults with supportive evidence of neuroimaging and inflammatory biomarkers. No study from low- and middle-income countries (LMICs)was identified despite high levels of air pollutants and high rates of dementia. To conclude, air pollution may impair cognitive function across the life-course, but a paucity of studies from reLMICs is a major lacuna in research.


Introduction
Ambient air pollution is a leading cause of the global disease burden according to the Global Burden of Diseases, Injuries, and Risk Factors Study 2015, especially in low-income and middle-income countries [1]. Recent estimates ascribe 8.9 million deaths per year to ambient particulate matter (PM) having a diameter less than 2.5 microns (PM2.5) [2]. The PM2.5 from fossil fuels alone were, in another study, estimated to be a contributing cause to 10.2 million global excess deaths in 2012, with 62% of deaths in China (3.9 million) and India (2.5 million) [3]. Western Pacific and South-East Asia have the largest burden of disease related to air pollution worldwide, contributed by heavy industry and air pollution hotspots in the developing nations therein [4]. However, lower-middle income countries (LMICs) as well as low-income countries (LICs) have often been left behind when it comes to conducting epidemiological studies on air pollution health effects [5,6]. Unfortunately, it is not certain that results from epidemiological studies in high-income countries (HICs) or upper-middle-income countries (UMICs) can be directly extrapolated to LMICs or LICs, because both the quantities and sources of air pollution often differ between HICs/HMICs and LMICs/LICs, making the chemical composition of exposure, and subsequent health effects, unique. For example, in India (an LMIC), pollution is worse than in China (a UMIC). There are 22 Indian cities on the global list of the 30 most polluted cities. Apart from urban sources of air pollution, the burning of agricultural stubble in nearby rural areas also contributes to the burden of air pollution in Indian cities. In addition, in India all indicators of air pollution greatly exceed WHO standards [7], and concentrations are increasing [8]. Furthermore, African PM emissions often originate from old diesel-powered vehicles, and poor household waste management, and households burning biomass are the predominant contributors to outdoor air pollution [9]. In order to reduce uncertainties in the estimates for LICs and LMICs, epidemiological studies in these countries are, thus, needed [6].
Cognitive function, a prominent determinant of human capital, health, and socioeconomic status, is impacted by cumulative biological, social, and environmental exposures across the life course. Cognitive disorders such as dementia entail great suffering and high societal costs, and the prevalence worldwide is increasing. The number of people living with dementia is 55 million and is estimated to reach 75 million worldwide by 2030, with the majority living in LMICs and LICs. Recent studies have reported a decline in the prevalence of dementia in high-income countries, suggesting that dementia may, at least partially, be preventable [10,11].
Emerging studies suggest that exposure to air pollution may be associated with cognitive impairment, with reported effects ranging from impaired neurocognitive development in infancy and childhood to higher rates of cognitive decline and dementia in later life [12][13][14][15][16][17]. The Lancet Commission (2020) has recognized air pollution as a risk factor for dementia [18].
The aim and objective of this paper is to systematically review the evidence base with respect to the relationship between air pollution and cognitive health outcomes including dementia across the life course and in diverse income settings. There is a special focus on income level of the country of the study areas since LICs and LMICs often previously have been left behind when it comes to epidemiological studies of air pollution health effects. The high burden of cognitive disorders in LICs and LMICs, combined with the high burden of disease due to air pollution in these countries, highlights the need to make an inventory of epidemiological studies on air pollution in association with cognitive disorders in these countries. The contextualizing of research findings in terms of income settings of the research studies is valuable as the countries with lower income levels are disproportionately affected by air pollution while being resource-constrained to address either air pollution or its health impact.

Materials and Methods
A systematic review was conducted to answer the research question on the impact of air pollution on cognitive health across the life course.
The review took place between January and October 2020 based on the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) [19] Statement using a defined protocol that is unpublished (Appendix A).

Search Strategy
A systematic search of the PubMed database was performed using PRISMA Guidelines with no time limit on the date of publication [19]. The Population, Investigated Exposure, Comparison, Outcome (PICO) Framework for this review is given in Table 1. The search string included keywords related to cognition and air pollution. Studies having any of the following keywords related to air pollution (exposure variable) and cognition (outcome variable) in their titles were identified. The keywords related to pollution utilized for PubMed search were Air Pollution, Pollutant (s), Particulate Matter, PM, Haze, Smog, Traffic-related air pollution (TRAP) and apportionment. The key words related to cognition utilized for PubMed search were Dementia, Cognitive, Memory, Attention, Cognition, Concentration, Orientation, Alertness, Alert, Intelligence, Emotion (s), Language, Reasoning, Planning, Decision making, Judgement, Recall, Learning, Emotion, Insight, Processing, Visuoconstructional, Coordination and Perception.

Selection Criteria
The inclusion criteria for the studies were full-text articles published in English with no time limit on date of publication, including original studies, systematic reviews and meta-analysis from any country on human participants of any age or gender. Protocols, letters to the editor and grey literature were excluded.

Screening Strategy
The search strategy comprised a two-stage process, In the first stage, three reviewers (C.B.R., N.K., V.K.S.) independently screened identified studies for eligibility by screening the titles and abstracts for study inclusion and exclusion criteria. References and citations of included papers were also reviewed to include additional potential articles. Abstracts of conference proceedings were searched for any relevant papers and posters. The second stage comprised independently screening the full texts of studies for eligibility by the three aforesaid reviewers. If the reviewers did not agree, then a more experienced reviewer was consulted (M.C.).

Data Extraction and Analysis
Each included study was examined in detail. Data from the included studies were recorded for publication date, study setting, study type, duration of the study, study population, air pollution exposure and cognitive-outcome measures studied and reported. The results obtained were examined against the income level of the study area as per World Bank classification and contextualized. The extracted data were calibrated for consistency and completeness by independent reviewers (M.C. and K.C.). Due to heterogeneity in exposure and outcome variables and tools employed to assess these parameters, a statistical meta-analysis was not possible. Instead, the evidence was analyzed and presented based on nature of air pollutant, age profile, and gender of participants as well as income levels of study areas.

Quality Assessment
The quality of individual studies was evaluated using the Revised tool for Quality Assessment on Diagnostic Accuracy Studies (QUADAS-2) [21]. The QUADAS-2 assesses the quality of primary diagnostic accuracy studies based on four key domains covering patient selection, index test, reference standard, and flow of patients through the study and timing of the index test(s) and reference standard ("flow and timing"). Each domain is evaluated for the risk of bias (using signaling questions), and the first three are also assessed for applicability.
Two senior authors, not involved in the initial screening of studies for inclusion in the systematic review (vide supra), independently evaluated primary studies using QUADAS-2 (M.C., K.C.), and any difference in agreement was resolved by consensus with the third experienced author (M.K.), who was not involved in the initial screening of papers for eligibility criteria or quality evaluation. This ensured objectivity in the systematic review process.
Statistical methods for assessing publication bias using techniques such as funnel plots were considered but ruled out due to the heterogenous nature of available scientific literature on multiple exposures for multiple cognitive and neuroimaging outcomes across different age groups. We did not exclude any study based on quality. The results of analysis of publication bias are given in Table 2 and Figure 1.

3.1.Original research studies obtained on screening
A total of 1173 studies with a keyword related to cognition and a keyword related to pollution (in their titles) were obtained until 10 July 2020. The remaining 350 studies were manually reviewed for relevance (Rai, Sandhu, Kumari) to yield 53 original research studies ( Figure 2).

Original Research Studies Obtained on Screening
A total of 1173 studies with a keyword related to cognition and a keyword related to pollution (in their titles) were obtained until 10 July 2020. The remaining 350 studies were manually reviewed for relevance (Rai, Sandhu, Kumari) to yield 53 original research studies ( Figure 2).

Income Levels of Sites of Original Studies
The 53 original studies included 47 conducted in HICs and 6 conducted in LMICs. These are summarized in Table 3.

Income Levels of Sites of Original Studies
The 53 original studies included 47 conducted in HICs and 6 conducted in LMICs. These are summarized in Table 3.     High isophorone levels (>0.49 ng/m 3 ) and low HOME score were associated low math scale score. Decrement in math scale score was more than the additive effect of each exposure especially in male children.

High pollution versus low pollution areas
Frontal tau hyperphosphorylation with pre-tangle material amyloid-beta diffuse plaques Nearly 40% of highly exposed children and young adults had frontal tau hyperphosphorylation with pre-tangle material and 51% had amyloid-beta diffuse plaques compared versus 0% of controls living in low pollution areas. Did not report analysis stratified by sex Residence in high air pollution area associated with deficits in a combination of fluid and crystallized cognition tasks, high rates of prefrontal white matter hyperintense lesions Did not report analysis stratified by sex

Research from LMICs/LICs
Among the studies, 89% were conducted in HICs, while only 11% of studies were conducted in UMICs. There was no original research from either LMICs or LICs as per World Bank income levels, despite high levels of air pollution (Appendix B).

Gender-Based Analysis
A review of 53 original studies revealed that most studies did not report gender-based findings (n = 28), and an equal number (n = 7) reported significant gender differences and no gender differences. Six studies were conducted on female participants, while five studies researched only male participants. (Figure 3).

Research from LMICs/LICs
Among the studies, 89% were conducted in HICs, while only 11% of studies were conducted in UMICs. There was no original research from either LMICs or LICs as per World Bank income levels, despite high levels of air pollution (Appendix B).

Gender-Based Analysis
A review of 53 original studies revealed that most studies did not report genderbased findings (n = 28), and an equal number (n = 7) reported significant gender differences and no gender differences. Six studies were conducted on female participants, while five studies researched only male participants. (Figure 3).

Analysis Based on Life Course and Income Level of Study Areas
The results of the studies were analyzed based on the life-course stage (children and adolescents prenatal and postnatal exposures, adults and older adults) and on the income level of the study areas are presented below.

Studies Related to Air Pollution and Cognitive Impairment in Children and Adolescents Prenatal Exposure Studies
Prenatal exposure to different air pollutants has been analyzed for cognitive and other developmental indicators from infancy through adolescence using birth cohorts.

Multiple Air Pollutants
Two European studies found that exposure to multiple air pollutants in the prenatal period and infancy (until 1 year of age) had an adverse impact on psychomotor development and cognitive performance, adaptive functioning, and behavioral indices.
Guxens et al. analyzed data from 9482 children from mother-infant pairs recruited between 1997 to 2008 in 6 European population-based birth cohorts, namely GENERA-TION R (Netherlands), DUISBURG (Germany), EDEN (France), GASPII (Italy), RHEA (Greece), and INMA (Spain) [54]. Prenatal exposures to NO2 and nitrogen oxides (NOx) for the exact pregnancy periods in all regions and PM (PM2.5, PM10, PM2.5-10, PM2.5 absorbance) in some regions were derived from background monitoring sites as well as land-use regression models for the period 2008 to 2011. Children were later assessed by a physician

Analysis Based on Life Course and Income Level of Study Areas
The results of the studies were analyzed based on the life-course stage (children and adolescents prenatal and postnatal exposures, adults and older adults) and on the income level of the study areas are presented below.

Studies Related to Air Pollution and Cognitive Impairment in Children and Adolescents Prenatal Exposure Studies
Prenatal exposure to different air pollutants has been analyzed for cognitive and other developmental indicators from infancy through adolescence using birth cohorts.

Multiple Air Pollutants
Two European studies found that exposure to multiple air pollutants in the prenatal period and infancy (until 1 year of age) had an adverse impact on psychomotor development and cognitive performance, adaptive functioning, and behavioral indices.
Guxens et al. analyzed data from 9482 children from mother-infant pairs recruited between 1997 to 2008 in 6 European population-based birth cohorts, namely GENERATION R (Netherlands), DUISBURG (Germany), EDEN (France), GASPII (Italy), RHEA (Greece), and INMA (Spain) [54]. Prenatal exposures to NO 2 and nitrogen oxides (NO x ) for the exact pregnancy periods in all regions and PM (PM 2.5 , PM 10 , PM 2.5-10 , PM 2.5 absorbance) in some regions were derived from background monitoring sites as well as land-use regression models for the period 2008 to 2011. Children were later assessed by a physician between 1 and 6 years of age. Prenatal air pollution exposure was associated with dose-dependent reduction in psychomotor development, where results of the different tests on average declined 0.68 points, with a 95% confidence interval (CI) of −1.25 to −0.11 per 10 µg/m 3 increase in NO 2 . No such associations were seen for cognitive impairment.
In another study from California, U.S., prenatal air pollutant exposure and first year of life exposure to NO 2 , PM 2.5 , PM 10 , ozone (O 3 ), and near-roadway air pollution was associated with impaired cognitive performance, adaptive functioning, and behavioral indices among children with Autism Spectrum Disorder (ASD) (n = 327), but not severity of ASD [29]. However, third trimester PM 10 exposure was paradoxically associated with improved behavioral performance in this study.
A Mexican study also reported diffuse neuroinflammation, damage to the neurovascular unit and production of autoantibodies to neural and tight-junction proteins among children chronically exposed to higher concentrations of ozone and PM 2.5 in Mexico. Such physiological responses enhance the risk of developing Alzheimer's disease later in life [68].

Traffic-Related Air Pollution (TRAP)
Similarly, prenatal exposure to TRAP was found to have an adverse impact on intelligence quotient (IQ) in two separate birth cohort studies in the U.S. and Italy.
Project Viva, a cohort of 1109 mother-child pairs in eastern Massachusetts, U.S.A, investigated the impact of prenatal and childhood exposure to TRAP, including black carbon (BC) and PM 2.5 based on proximity to roadways and traffic density, on childhood cognition [47]. At a mean age of 8 years, children with a residence located less than 50 m away from a major highway had lower nonverbal IQ (−7.5 points; 95% CI: −13.1, −1.9), verbal IQ (−3.8 points; 95% CI: −8.2, 0.6) and visual-motor abilities (−5.3 points; 95% CI: −11.0, 0.4). Findings for cross-sectional associations between major roadway proximity and (1) prenatal and childhood exposure to traffic density, (2) prenatal and childhood exposure to PM 2.5 , and (3) third trimester and childhood exposure to BC were equivocal. In a study in Rome, Italy, the cognition of children aged 7 years (n = 474) who belonged to a birth cohort of 719 newborns was investigated [43]. The authors found that both traffic intensity within a 100 m buffer around the home and incremental increases (per 10 µg/m 3 ) of intrauterine NO 2 exposure were associated with reduced verbal IQ and verbal comprehension IQ (on Wechsler Intelligence Scale for Children-III). Other pollutants also showed negative associations but with much wider confidence intervals.

3.
Particulate Matter (PM) Prenatal PM exposure was found to have a deleterious impact on brain volumes in a neuroimaging study, demonstrating the neurological basis of impaired cognition in children exposed prenatally to air pollutants in previous studies.
A dose-dependent thinning of cerebral cortex per 5 µg/m 3 increase in prenatal PM exposure in bilateral cerebral hemispheres was demonstrated in children living in Rotterdam, the Netherlands [14]. Reduced cerebral cortex in the precuneus and rostral middle frontal regions partially mediated the association between PM exposure and impaired inhibitory control. However, global brain volumes were not impacted.

Isophorone
Another cost-effective technique to assess exposure to multiple air pollutants is to study one indicator air pollutant, such as Isophorone (which is a common marker of industrial pollution), and its association with cognitive performance. Isophorone was investigated in a study on residential concentrations of 104 ambient air toxins from the National Air Toxics Assessment (2002). Machine learning algorithms were applied to the cognitive data of 6900 children from the Early Childhood Longitudinal Study Birth Cohort in the U.S.A. Here, a high Isophorone level (>0.49 ng/m 3 ) was associated with low performance on mathematics tasks among children from urban and highly populated urban areas, with their scores being reduced by 1.19 points (95% CI: −1.94, −0.44) [37].

Persistent Organic Pollutants (POPs)
Another set of toxic air pollutants are POPs. Prenatal exposure to POPs such as polychlorinated biphenyls (PCBs), was investigated in adolescents (13-15 years) from 2 Dutch birth cohorts in the Development at Adolescence and Chemical Exposure (DACE) study [26].
Maternal serum levels of PCBs including 3 hydroxylated variants were measured during pregnancy [26]. There was a significant association between the compound PCB-183 and lower total intelligence, hexabromocyclododecane (HBCDD) and lower performance intelligence, and polybrominated diphenyl ethers (PBDEs) with lower verbal memory. Several hydroxy PCBs (OH-PCBs) were associated with more optimal sustained attention and balance.
Among boys, poorer outcomes on fine motor skills and better outcomes on motor performance were reported, and prenatal DDE levels were associated with suboptimal motor performance, while a positive trend was seen between 4-OH-PCB-107 and fine motor skills. In girls, positive trends between OH-PCBs and verbal intelligence, and a negative trend between the compound BDE-153 and fine motor skills was observed. The associations persisted even after controlling for confounders such as maternal smoking, alcohol use, and education.

Polyaromatic Hydrocarbons (PAH)
Prenatal exposure to PAH is another set of pollutants implicated in lower IQ, neuroimaging markers, impaired cognition in children and behavioral conditions such as Attention Deficit Hyperactivity Disorder (ADHD) and Conduct Disorder in separate birth cohort studies and different ethnic populations.
A cohort involving prenatal personal ambient PAH monitoring was created with 665 urban Latina (Dominican) and African-American women, aged 18-35 years, residing in New York city [63]. The following inclusion criteria were applied: non-diabetic, non-hypertensive, not users of tobacco or illicit drugs, initial prenatal care by the 20th week of pregnancy, no known human immunodeficiency virus, low prenatal exposure to environmental tobacco smoke (cotinine levels < 15 µg/L in umbilical cord blood), and low exposure to pesticides (chlorpyrifos levels < 4.39 pg/g in umbilical cord blood). Results indicate that high prenatal PAH levels (>2.26 ng/m 3 ) were associated with lower IQ (p = 0.007) and verbal IQ (p = 0.003), with offspring experiencing a decrement of 4.31-4.67 IQ points at 5 years of age. Neuroimaging of 40 randomly selected children in high and low prenatal PAH exposure groups demonstrated an inverse dose-response relationship with reduction of the white matter surface in the left hemisphere of the brain in later childhood (7-9 years), which is associated with slower information processing speed during intelligence testing and more severe externalizing behavioral conditions such as ADHD and Conduct Disorder [48].
Similarly, a study in Krakow, Poland, monitored healthy, nonsmoking pregnant women's personal prenatal PAH levels (48 h) and examined blood samples and/or a cord blood samples at the time of delivery. After follow-up of the offspring (n = 214), the findings showed that high prenatal PAH exposure (>17.96 ng/m 3 ) was associated with reduced nonverbal reasoning ability on Raven's Coloured Progressive Matrices at 5 years of age, which corresponds to an average IQ decrease of 3.8 points [60].

Postnatal Exposure Studies
Most children continue to be exposed to air pollutants in postnatal periods as well. Several investigators have also explored the impact of postnatal exposure of the air pollutants known to have an adverse cognitive impact on prenatal exposure.

Multiple Air Pollutants
There has been original research on neuroimaging, neuropathological, and inflammatory markers as biomarkers of impaired cognition associated with postnatal exposure to multiple air pollutants.
Neuroimaging studies have also demonstrated the impact of high air-pollution exposure during the postnatal period on cognition. For instance, an MRI brain study found significant differences in white-matter volumes, especially in the right parietal and bilateral temporal areas, of children exposed to high levels of air pollution in Mexico City [69]. These same children had progressive deficits on the "Vocabulary" and "Digit Span" subtests of the Wechsler Intelligence Scale for Children-Revised (WISC-R).
Additionally, children with white-matter hyperintensities showed inflammatory markers, including evidence of inflammation resolution, immunoregulation, and tissue remodeling, while children without white-matter hyperintensities had raised levels of interleukin 12, pro-inflammatory cytokines, and low levels of neuroprotective cytokines and chemokines. The authors postulated that the air-pollution-induced neuroinflammation, characterized by complex modulation of cytokines and chemokines, causes structural and volumetric changes in the central nervous system (CNS) and, thereby, affects cognitive function.
Another study on urban children supports air-pollution-associated neuroinflammation and neuropathology, as nearly 40% of highly exposed children and young adults had frontal tau hyperphosphorylation with pre-tangle material, and 51% had amyloid-beta diffuse plaques [68]. Conversely, 0% of the samples/scans from similar controls living in lowpollution areas showed such effects.

2.
Traffic-Related Air Pollution (TRAP) Postnatal TRAP was found to be associated with impaired cognitive performance in four studies, though there is a contradictory study. All studies were from Spain.
Exposure to nitrogen dioxide (NO 2 ), PM 2.5 , and elemental carbon (EC) derived from traffic over 1 year was studied in relation to the attention-related cognitive domain among 2687 primary school children from 39 schools across Barcelona, Spain [38,50]. Results demonstrated that children's attention processes, assessed via a computerized, childfriendly Attention Network Test, were inversely associated with TRAP. Daily ambient levels of NO 2 and PM 2.5 contributed to an estimated equivalent of a 1.1-month (95% CI: 0.84, 1.37) delay in response speed according to the age-appropriate, natural developmental trajectory. Furthermore, boys demonstrated significantly worse cognitive performance on Hit Reaction Time following incremental increases of both EC and NO 2 .
Similarly, a study in Catalonia, Spain, explored the cognitive impact of TRAP exposure during the walking commute to school among children aged 7-10 years old (n = 1234) [12]. The authors found that exposure to PM 2.5 and BC was associated with a reduction in the developmental growth of working memory; however, no significant association was seen for NO.
A study conducted in Barcelona, Catalonia, Spain, also found an adverse impact of composite exposure to indoor and outdoor levels of TRAP at school, including EC, NO 2 , PM 2.5 and ultrafine particles (UFP), on the development of working memory in children over 3.5 years of age (n = 2897) [34].
Conversely, Freire et al. found no significant association between NO 2 and cognitive development in a population-based birth cohort of 210 children in Spain [61].

Persistent Organic Pollutants (POPs)
Similar to prenatal exposure, postnatal exposure to different POPs has been found to have an adverse cognitive impact.

Isophorone
Like prenatal exposure, postnatal Isophorone has also been found to be associated with lower performance on mathematical tasks in children.
Ambient Isophorone exposure was found to be associated with low math scores (−1.48; 95% CI: −2.79, −0.18), independent of the influence of the home-learning environment, on analyzing the 2002 U.S. National Air Toxics Assessment data. Despite no interaction effect, children with both high Isophorone exposure and a low score on Home Observation for Measurement of the Environment Inventory had a decrement in math-scale score beyond the additive effect of individual exposures, especially in males [35].

Studies Related to Air Pollution and Cognitive Impairment in Adults
Two studies from the U.S. and China reported impaired cognitive performance associated with exposure to air pollutants on cognitive performance in adults, which was contradicted by another study from the U.S.
Chen and Schwartz found consistent associations in their study of 1764 adults in the United States aged 37.5 (±10.9 years) using the Third National Health and Nutrition Examination Survey (1988-1991) [62]. The findings demonstrated that long-term exposure to ambient ozone reduced performance on both the symbol-digit substitution test and a serial-digit learning test, but not in simple reaction-time tests. Each 10 parts per billion (ppb) increase in annual ozone was associated with the equivalent of 3.5 and 5.3 years of aging-related decline in cognitive performance (depending on the outcome measure used). Similar associations were not seen for PM 10 . Another study, in China, on the effects of cumulative and transitory exposure to air pollution on cognitive performance reported an adverse cognitive impact in verbal and math tests with greater deficits on verbal tasks as people aged [67].
However, results from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) cohort found no consistent increase in the odds of cognitive impairment with every 10 µg/m 3 increase in PM 2.5 concentration (55). This was a geographically diverse, biracial U.S. cohort of both sexes (n = 20,150) aged ≥ 45 years with satellite-derived estimates of residential PM 2.5 concentrations.
Further studies are required to clarify these contradictory findings, given that cognitive reserve in adults can mask minimal cognitive impairments.

Studies Related to Air Pollution and Cognitive Dysfunction in the Older Adults
Like childhood studies, there is extensive literature on the impact of air pollutants on cognition in the elderly.

Air and Noise Pollution
Several research studies have focussed on combined exposure of air and noise pollution, both of which are potentially deletrious for cognition in the elderly.
A positive exposure-response relationship between dementia and PM 2.5 , NO 2 , but not ozone, was found in a cohort study in England (n = 30,978 aged 50-79 years) independent of noise levels [27]. Similarly, in the Betula project, a longitudinal study on health and ageing in Umeå, Sweden, the risk for incident dementia over a 22-year period between 1988 to 2010 was associated with higher residential exposure to NO x (n = 1469 persons aged 60 to 85 years at inclusion) [42]. The authors reported a hazard ratio (HR) of 1.43 (95% CI: 0.998, 2.05) for the highest exposure quartile compared to the lowest [42]. The risk for dementia was independent of noise exposure, even in a relatively low-exposure area (n = 1721 persons aged 55-85 years) [25].
However, an additional study investigating associations within the entire Betula dataset (n = 1469 aged  showed no association between NO x and episodic memory [36]. Research on a Heinz Nixdorf Recall population-based cohort (n = 4086 participants, aged 50-80 years from Bochum, Essen, and Mülheim in Germany) revealed that long-term residential exposure to size-fractioned PM, NO x , and traffic noise was negatively correlated with four neuropsychological subtests (verbal fluency, a labyrinth test, immediate verbal recall, and delayed verbal recall) as well as with global cognitive score (GCS) [44]. Here, the cognitive effects of dose-dependent increases in PM 2.5 were independent of noise exposure.
Tzivian et al., furthermore, found a positive dose-dependent association between longterm air pollution exposure and mild cognitive impairment (MCI), mainly amnestic subtype: odds ratio (OR) per interquartile range increase in PM 2.5 of 1.16 (95% CI: 1.05, 1.27) [44]. In a subsequent study, this was found to be amplified by noise exposure, supporting the synergistic negative effect of air pollution and road traffic noise [40]. Similarly, exposure to traffic-related fine PM was also determined to be a consistent and significant risk factor for MCI in another German study [64].

Multiple Air Pollutants and TRAP
Similar to other age groups, multiple air-pollutant exposure has been determined to have an adverse impact on cognitive impairment and its concomitants in the elderly.
Only one study has investigated the role of TRAP exposure in impairment of specific cognitive domains using a cohort-based approach. This study on older adults residing in Los Angeles reported an association between PM 2.5 exposure and lower verbal learning (β = −0.32 per 10 µg/m 3 ; 95% CI: −0.63, 0.00; p = 0.05). Moreover, NO 2 exposure over 20 ppb was associated with lower logical memory, and ozone exposure above 49 ppb was associated with lower executive function [53]. Still, none of these pollutants were significantly associated with global cognition.
Similarly, a study on the association between TRAP and cognitive function in 680 elderly participants (aged 71 ± 7 years) in the U.S. Department of Veterans Affairs (VA) Normative Aging Study demonstrated a significant relationship between BC exposure and cognitive impairment, assessed using the Mini Mental Status Examination (MMSE) scores. The multivariable-adjusted OR was 1.3 for a doubling in BC concentration (95% CI: 1.1, 1.6) [59]. TRAP exposure was found to adversely impact reasoning and memory, but not verbal fluency, in a study on 2867 individuals (mean age: 66 years) from the Whitehall II cohort in the United Kingdom. Higher PM 2.5 of 1.1 µg/m 3 (lag 4) was associated with a 0.03 (95% CI: −0.06, 0.002) 5-year decline in standardized memory score [55]. In addition, in a study of 3377 elderly participants from the National Social Life, Health, and Aging Project (NSHAP) in the U.S., air pollutant exposure was estimated using GIS-based spatio-temporal models for PM 2.5 and Environmental Protection Agency (EPA) monitors for NO 2 . Here, high PM 2.5 exposure was associated with a 0.22 (95% CI: −0.44, −0.01) and high NO 2 exposure with a 0.25 (95% CI: −0.43, −0.06) point decrease in Chicago Cognitive Function Measure test scores; these reductions are equivalent to aging 1.6 years and 1.9 years, respectively. The cognitive impact of PM 2.5 was found to be mediated by depression and modified by stroke, anxiety, and stress. On the other hand, the cognitive impact of NO 2 was mediated by stress, with impaired activities of daily living causing effect modification [39].
Another study by Molina-Sotomeyer et al. on 181 older women in Chile took into account the confounding effect of cardiorespiratory aerobic exercise for the impact of multiple air pollutants such as PM 10, PM 2.5 , NO 2, SO 2 and Ozone concentrations on cognition and cardiovascular markers. The authors used maximum oxygen uptake (VO 2max ), estimated by the Six-Minute Walk Test (6mWT), heart rate (HR), and oxygen saturation (SpO 2 ) as indicators of cardiorespiratory aerobic exercise. They marked significantly lower cognition performance amongst women residing in highly polluted areas as compared to those living in less-polluted areas. However, aerobic exercise was found to be a protective factor against adverse cognitive impact of air pollution, probably due to improvement in the mechanisms of oxygen transport [24].
Additionally, a large study on air pollution exposure and dementia incidence included all Canadian-born Ontario residents aged 55-85 years old and followed them from 2001 to 2013. The findings demonstrated that long-term mean residential exposure to air pollutants, including PM 2.5 , NO 2 and ozone, was associated with higher dementia incidence, even at the relatively low exposure levels throughout this study setting. The HR was 1.04 (95% CI: 1.03, 1.05) for every interquartile-range increase in PM 2.5 and 1.10 (95% CI: 1.08, 1.12) for every interquartile increase in NO 2 [32]. Lo et al. also reported a significant association between long-term exposure to PM 10 and O 3 and cognitive impairment in older adults, with ORs of 1.094 (95% CI: 1.020, 1.174) and 1.878 (95% CI: 1.363, 2.560), respectively [23]. A greater effect was seen for the combined exposure to PM 10 and O 3 (p < 0.001).
Further, research was conducted on the cross-sectional and cumulative (over 2.8 years) impact of outdoor air pollution exposure (PM and NO x ) on cognitive functions of 86,759 middle-to older-aged adults (mean age 56.86 ± 8.12 years) from the U.K. Biobank general population cohort. While there was weak association between air pollution exposure and cognitive performance (dose-dependent slower reaction time, higher error rate on a visuospatial memory test, and lower numeric memory scores) at baseline (after adjusting for confounders), no such associations were found at follow-up. The authors did not report analyses stratified by sex [28].
Particulate Matter (PM) PM has also been extensively investigated for a dose-response relationship with cognitive impairment in the elderly in several research projects in several high-income settings.
Oudin et al. used a dataset of 1806 participants from the Betula project in Sweden (enrolled between 1993 and 1995 and followed until 2010) to assess the impact of exposures of ambient PM 2.5 from residential wood burning and vehicle exhaust. Their results indicated a dose-dependent increased risk of incident dementia (Vascular Dementia and Alzheimer's Disease) per 1 µg/m 3 increase in PM 2.5 ; reported HRs were 1.55 (95% CI: 1.00, 2.41; p = 0.05) for local wood burning and 1.66 (95% CI: 1.16, 2.39; p = 0.006) for traffic exhaust [30].
Similarly, PM 2.5 exposure was associated with a 1.5 times greater cognitive error rate in a study on non-Hispanic, older White and Black U.S. adults (n = 780; age ≥ 55 years) from the 2001/2002 Americans' Changing Lives Study. The adverse cognitive impact of PM 2.5 , was found to be similar among men and women, and possibly mediated by neighborhood social stressors and environmental hazards [31].
A neuroimaging study examining the association between PM 2.5 and brain volumes in 1403 community-dwelling older women without dementia from the Women's Health Initiative Memory Study (1996)(1997)(1998) in Germany found that older women with greater PM 2.5 exposure had significantly smaller white matter (WM) volumes in the frontal and temporal lobes and corpus callosum, which is equivalent to 1-2 years of brain ageing [45]. Similar associations were not seen for gray matter (GM) volumes or hippocampal volumes.
Long-term exposure to high levels of PM 2.5-10 and PM 2.5 has been associated with faster dose-dependent cognitive decline in older women (aged 70 to 81 years) for every 10 µg/m 3 increment in long-term (2 years) exposure in the U.S. Nurses' Health Study Cognitive Cohort (n = 19409). The corresponding decrements on a global cognitive score following these PM 2.5-10 and PM 2.5 exposures were 0.020 (95% CI: −0.032, −0.008) and −0.018 (95% CI: −0.035, −0.002), respectively. The effect of 10 µg/m 3 higher PM exposure was cognitively equivalent to aging by approximately 2 years [58].
Similarly, Saenz et al. also found that exposure to indoor air pollution, due to the household's primary cooking fuel being wood or coal, was associated with poorer cognitive performance in older adults (over age 50) in Mexico [66].
Lee et al. reported that long-term exposure to a higher concentration of PM 2.5 was associated with increased hospitalizations for dementia, with an adjusted HR of 1.049 (95% CI: 1.048, 1.051) per 1 µg/m 3 increase in annual PM 2.5 . The hazard ratio for vascular dementia was somewhat higher at 1.086 (95% CI: 1.082, 1.090) [22]. Additionally, hospitalizations for dementia increased slightly as the level of urbanization increased: the HR for rural areas was 1.036 (95% CI: 1.031, 1.041) vis a vis 1.052 (95% CI: 1.050, 1.054) for the metropolitan areas.
The available literature strongly suggests that air pollution, especially PM, is associated with cognitive disorders in the elderly [70].

Polyaromatic Hydrocarbons (PAH)
There is only one study investigating the dose-dependent relationship between PAH and decline in cognitive performance. This study used the 2001-2002 National Health and Nutrition Examination Survey in the U.S. (n = 454) to demonstrate a dose-dependent association between PAH exposure per 1% increase (assessed by urinary 1-hydroxypyrene) and a decline in performance on digit symbol substitution tests by 1.8% among those aged ≥ 60 years [41].

Contribution of Genetic Factors
Cognitive impairment in the elderly may be associated with genetic factors such as polymorphisms in apolipoprotein E allele (ApoE4 variants) and hemochromatosis gene (HFE C282Y variant). Therefore, it is important that the confounding impact of such genetic factors are investigated to delineate the differential impact of air pollution and genetic loading, which has been investigated in some American and German studies.
In the Veterans Affairs Normative Aging Study from the U.S., Black Carbon (BC) was reported to be associated with lower cognition (n = 428 older men). Each doubling in BC levels was associated with 1.57 (95% CI: 1.20, 2.05) times higher odds of low MMSE scores in individuals with longer blood telomere length (OR = 3.23; 95% CI: 1.37, 7.59; p = 0.04 for BC-by-TL-interaction) [33], suggesting that genetic status may modify the effect of air pollution on cognitive health outcomes. The same study reported that older adults who lack an HFE C282Y variant (hemochromatosis gene polymorphisms) (n = 680; age 71 ± 7 years of age) are more susceptible to the adverse effects of TRAP exposure on cognitive function [58].

Discussion
Overall, the evidence from epidemiological studies points towards an association between exposure to pollutants and cognitive health effects across the life course. There is likely adverse cognitive impact of prenatal exposure to air pollutants, such as PAH, NO 2 , PM, POPs, with the evidence being strengthened by emerging neuroimaging research. Most studies also report an adverse cognitive impact of postnatal exposure to air pollutants in children, especially TRAP, POPs, and Isophorone. These associations have been validated by cognitive testing, neuroimaging studies and research on neuroinflammatory markers.
The largest body of research is on the cognitive impact of air pollution among the elderly population. There is robust research demonstrating dose-dependent relationships between air pollutants such as PM, black carbon, TRAP, NO 2 , ozone, PAH and cognitive performance, neuroimaging markers and incident dementia. Furthermore, research on possible aetio-pathological mechanisms indicate that some genetic factors, including APO E and HFE C28y allele status, may intensify the adverse effects of air pollution on cognition, although evidence is not conclusive.
The research on associations between air pollution and cognitive health outcomes has occurred primarily in the HICs and UMICs of Europe and North America to Asia and Latin America, whereas there is paucity of such research in the LMICs and LICs of Africa and Asia, including India. No original research has been conducted on the cognitive impact of air pollutants in LMICs or LICs, even though the air pollution is generally much higher than the World Health Organization guidelines in LMICs/LICs as compared to HICs and UMICs [71]. Even the composition of particulate ambient air pollutants differs substantially, as their sources (household, industrial, traffic, etc.) may vary by setting. For example, indoor exposure is generally higher in LICs and LMICs than in UMICs and HICs, as woodstoves are often used for cooking. The extrapolations of the doseresponse curves based on data from HICS and UMICs may not be applicable for LMICs and LICs. Therefore, generalizing results from HIC and UMICs to LICs and LMICs may be problematic due to varying levels, composition, and sources of air pollutants. The vast burden of both air pollution and neurocognitive disorders in LMICs, such as India with a 90% treatment gap [72], furthermore necessitates the synthesis of available evidence and systematic research to catalyze environmental policy change [73][74][75].
The proposed aetio-pathogenic pathways of air pollution's effect on cognitive decline include enhanced risk of hypertension, dyslipidemia, oxidative stress, insulin resistance, endothelial dysfunction, procoagulant states, and stroke [76,77]. The plausible aetio-pathogenic pathways furthermore include structural changes in brain, neurodegeneration, and neuroinflammation. Increased levels of PM 2.5 may be associated with smaller brain volumes, white matter lesions, higher rates of infarcts, and necrotic areas in the brain, all of which contribute to cognitive impairment [78]. Pathogenic mechanisms furthermore include hypoxia and direct neurotoxicity, which results from the breakdown of nasal, gut, lung epithelial, and blood-brain barriers, allowing an influx of airborne pollutants directly into the brain [79,80]. Secondary neurotoxicity can also occur when cytokines are transported from a lung injury site to the brain, resulting in neuroinflammation and neurodegeneration [79,80]. Air pollution, especially PM 2.5 , is also associated with the enhanced expression of pro-inflammatory mediators such as tumor necrosis factor (TNF-α) and interleukin-1β (IL-1β) as well as reactive oxygen species (ROS) [81]. The summary of proposed aetio-pathogenic impact of air pollution on cognition across the life course is given in Figure 4.
The quality of individual studies incorporated in this systematic review was evaluated using the Revised tool for Quality Assessment on Diagnostic Accuracy Studies (QUADAS-2) [21], and most research scored high on the quality of primary diagnostic accuracy. The risk of bias in the included studies on QUADAS-2 was low to moderate, considering that grey literature was not evaluated. The process of QUADAS-2 evaluation, by involving different researchers than the ones involved in primary screening, ensured objectivity. The studies scored low on risk of bias and applicability concerns for patient selection, index test, reference standard, flow of patients through the study, timing of the index test(s), and reference standard. This indicates that the diverse research on cognitive impact of air pollution across life course and across different income levels of sites of study is of good quality and can inform policy decisions. The quality of individual studies incorporated in this systematic review was evaluated using the Revised tool for Quality Assessment on Diagnostic Accuracy Studies (QUADAS-2) [21], and most research scored high on the quality of primary diagnostic accuracy. The risk of bias in the included studies on QUADAS-2 was low to moderate, considering that grey literature was not evaluated. The process of QUADAS-2 evaluation, by involving different researchers than the ones involved in primary screening, ensured Even though there is a growing evidence base for increasingly consistent results, dose-response relationships, and biological plausibility for air pollutants, particularly for exposure to PM 2.5 , the existing literature has certain limitations. The assessment of air pollution, although geocoded, may not reflect the true local exposure, as there is potential for variation even within a small geographical area. This can be further influenced by prevailing winds and seasonal patterns. Further, the individual exposure is only an estimate in these epidemiological studies and may not reflect the actual individual exposure that can be assessed in prospective studies using personal sensors. Moreover, it is a challenge to estimate and assess the effects of lifetime exposure to outdoor and indoor environmental pollutants in a mobile population, including occupational and transitory exposures, with complete data on exposure concentration, duration, pollutant type, and source apportionment. As the included studies utilized a variety of exposure assessment methods, such as dispersion models and land-use regression modelling, the data were too disparate to combine into a meta-analysis. Unlike cohort based epidemiological studies, case-finding bias may occur in hospital-based studies as participants may already have health concerns associated with exposure to air pollution (for example Bronchial Asthma and Chronic Obstructive Pulmonary Disease) necessitating their visit to the hospital. Adequate assessment of confounding factors, such as genetic factors (such as ApoE 4), cardiometabolic factors, individual and parental socioeconomic and educational status, and living conditions, is also required. Finally, there may be an emerging publication bias for studies reporting significant associations as this area expands.
This review has synthesized existing scientific evidence regarding the association between exposure to air pollution and cognitive health outcomes across the life course. However, the causal pathway of this relationship is still not fully understood, although evidence on aetio-pathogenic pathways is rapidly increasing from experimental studies. Establishing the causal effects of air pollution on cognitive health outcomes will likely require international, multicentric cohort-based research with harmonized protocols that periodically assess air pollutant exposures, cognitive function, systemic cardio-metabolic illnesses, neuroimaging, and inflammatory and genetic markers. In their 2015 systematic review of 13 studies, Peters et al. advocated for continued research on the association between air pollution and cognitive decline among older adults [81]. Additionally, two Lancet Commissions have supported emerging evidence for a causal role of PM in impaired cognition in midlife and later life, and have recommended further research [18,75]. As air pollution is a preventable risk factor, its regulation has potential health and fiscal benefits for individuals and society. Further scientific evidence (including life-course research, especially in LICs and LMICs) is required to bring about policy change for improved air quality and, thereby, protect cognition across the life course.

Conclusions
This systematic review of all published original studies on the cognitive impact of air pollution across the life course (using PRISMA guidelines) demonstrates that different air pollutants have an adverse cognitive impact across the life course. Most research is confined to HICs and UMICs with no original research in LMICs and LICS despite high rates of both air pollutants and cognitive impairment/dementia in these settings. Hence, the extrapolations of the dose-response curve based on data from HICS and UMICs may not be applicable for LMICs and LICs. As air pollution is a preventable risk factor, its regulation has potential health benefits and resultant cost-saving by potentially improving cognitive health and reducing the risk of dementia later in life. Further life-course research, especially in LICs and LMICs, is needed to establish aetio-pathogenic pathways for differential levels and cumulative exposure to air pollution, to promote policy change for air pollution and its health effects, including adverse cognitive impact.   The study inclusion and exclusion criteria are as follows: Inclusion criteria: • Articles will be eligible for inclusion in the review if they are full-text articles published in English with no time limit on date of publication; • Original studies, systematic reviews, and meta-analysis will all be eligible for inclusion; • The target population could be of any age; • The study could have been conducted in any country/countries. • Exclusion criteria: • The following publications will be excluded: Protocols, Letter to editor, No full text available; • Grey Literature will be excluded.
The country-level income will be determined by a nation's gross national income (GNI) per capita as High Income Countries (HICs), Upper and Middle Income Countries (UMICs), Lower and Middle Income countries (LMICS) and Lower Income Countries (LICs) as per the latest definitions given by the World Bank on its website.