Hygiene and Environmental Health Advances 4 (2022) 100019 Contents lists available at ScienceDirect Hygiene and Environmental Health Advances journal homepage: www.elsevier.com/locate/heha Industrial air pollutant emissions and mortality from Alzheimer’s disease in Canada Sabit Cakmak a , ∗ , Olaniyan Toyib b , Chris Hebbern c , Kimberly Mitchell a , Jasmine D. Cakmak d , Eric Lavigne a , Michael Tjepkema b , Naizhuo Zhao e a Population Studies Division, Environmental Health Science and Research Bureau, Health Canada, Ottawa, ON, Canada b Health Analysis, Statistics Canada/Government of Canada, Canada c Climate Change and Innovation Bureau, Health Canada, Ottawa, ON, Canada d Neuroscience, The University of Western Ontario, London, ON, Canada e Division of Clinical Epidemiology, McGill University Health Center, Montreal, QC, Canada a r t i c l e i n f o Keywords: Air pollution Alzheimer’s disease Mortality Industrial fine particulate matter (PM 2.5 ) Industrial nitrogen dioxide (NO 2 ) Industrial sulfur dioxide (SO 2 ) Canadian Census Health and Environment Cohort (CanCHEC) a b s t r a c t Background: There is increasing interest in the health effects of source-specific air pollution. However, the rela- tionship between industrial air pollutants and Alzheimer’s disease has received limited investigation. Objectives: To assess associations of industrial fine particulate matter (PM 2.5 ), nitrogen dioxide (NO 2 ), and sulfur dioxide (SO 2 ) exposures with mortality from Alzheimer’s disease. Methods: Approximately 3.2 million adults involved in the 2006 Canadian Census Health and Environment Cohort (CanCHEC) were followed from Census day (May 16, 2006) until death or December 31, 2016. Three-year moving- average industrial emissions with a one-year lag were assigned to the participants based on their residential postal codes. The neighborhood emission of each of the three industrial air pollutants for a postal code was estimated by considering weights of the air pollutant emissions from all industries within a 15 km buffer area, distances between the postal code area and the emitters, and percentages of time per year that the postal code area was downwind of the industrial emitters. Cox proportional hazards models were used to compute hazard ratios (HRs) for deaths from Alzheimer’s, adjusting for 15 socio-demographic and contextual covariates. Sensitivity analyses were conducted by adjusting for other industrial emissions, greenness, and comorbidity index, individually. Results: We identified 4500 deaths due to Alzheimer’s disease from 2006 to 2016 for a total of 32,909,200 person- years across the follow-up period. The adjusted HR for mortality from Alzheimer’s related to one interquartile range increase in industrial PM 2.5 , NO 2 , and SO 2 tonnes/meter per year are 1.006 (95% confidence intervals: 1.000-1.011), 0.994 (0.978-1.011), and 0.998 (0.996-1.001), respectively. Similar positive associations between industrial PM 2.5 and mortality from Alzheimer’s disease were observed, but there were no clear associations for NO 2 and SO 2 in sensitivity analyses. Conclusions: Exposure to industrial PM 2.5 increases the risk of mortality from Alzheimer’s disease. 1 p ( C a t c G l O s v 2 ( m 4 l d U a h R 2 ( . Introduction Air pollution is recognized as a global public health issue and articularly a risk factor for cardiovascular and respiratory diseases Dominici et al., 2006 ; Kampa and Castanas, 2008 ; Mills et al., 2009 ; akmak et al., 2018 ). Recent toxicological studies found that exposure to ir pollution can lead to inflammatory and oxidative reactions in lungs, rigger biochemical changes in brain tissues, and cause injury to the entral nervous systems of humans ( Babadjouni et al., 2017 ; González- uevara et al., 2014 ; Thomson, 2019 ). Thus, adverse effects of air pol- ution on neurological health outcomes including Parkinson’s disease, ∗ Corresponding author at: Population Studies Division, Environmental Health Sci ttawa, Ontario, Canada K1A 0K9, Canada. E-mail address: Sabit.Cakmak@canada.ca (S. Cakmak) . ttps://doi.org/10.1016/j.heha.2022.100019 eceived 23 July 2022; Received in revised form 16 August 2022; Accepted 17 Augu 773-0492/Crown Copyright © 2022 Published by Elsevier B.V. This is an open acce http://creativecommons.org/licenses/by-nc-nd/4.0/ ) troke, dementia, mild cognitive impairment, and total cerebral brain olume have been well documented (e.g. Béjot et al., 2018 ; Chen et al., 017a ; Chen et al., 2017b ; Zhao et al., 2021 ). Dementia is the second largest neurological cause of disability Carroll, 2019 ) and Alzheimer’s disease is the most common form of de- entia ( Moulton and Yang, 2012 ). In 2016, Alzheimer’s disease affected 3.8 million people worldwide ( Nichols et al., 2019 ), and this preva- ence is constantly increasing ( Brookmeyer et al., 2007 ). Alzheimer’s isease was officially listed as the sixth-leading cause of death in the nited States in 2019 and the seventh-leading cause of death in 2020 nd 2021 ( Alzheimer’s Association, 2022 ). A few studies, conducted in ence & Research Bureau, Health Canada, 251 Sir Frederick Banting Driveway, st 2022 ss article under the CC BY-NC-ND license https://doi.org/10.1016/j.heha.2022.100019 http://www.ScienceDirect.com http://www.elsevier.com/locate/heha http://crossmark.crossref.org/dialog/?doi=10.1016/j.heha.2022.100019&domain=pdf mailto:Sabit.Cakmak@canada.ca https://doi.org/10.1016/j.heha.2022.100019 http://creativecommons.org/licenses/by-nc-nd/4.0/ S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 s i s e a 2 i w ( t r p r m s r p s a e k o p s t N o a ( A C 2 2 E 2 ( d a e w a m i i t c t ( w ( c c i c l 2 ( h c a o t D G e a 2 e t w t c o ( o i f o v p o j e 𝐸 w y 1 ( f t f U e w “ c 7 B t C i t E w m s c 2 a m s W t i fl s ( everal countries including Canada, have demonstrated significant pos- tive associations between Alzheimer’s disease (or dementia) and expo- ures to air pollutants (e.g. fine particulate matter less than 2.5 microm- ters in diameter – PM 2.5 and nitrogen dioxide – NO 2 ); however, nearly ll focused on ambient air pollution ( Moulton and Yang, 2012 ; Shi et al., 020 ; Shou et al., 2019 ; Zhao et al., 2021 ). Therefore, there is limited nformation on the associations of mortality from Alzhemier’s disease ith industrial air pollution. Specific sources of air pollution may differentially influence health Oberdörster, 2000 ; Zhao et al., 2020 ), and the characteristics of pollu- ants can vary significantly by source ( EEA, 2022 ). In Canada, positive elationships of neurological disease incidence with traffic-related air ollution have been found by a few studies (e.g. Finkelstein and Jer- ett, 2007 ; Cakmak et al., 2019 ; Yuchi et al., 2020 ), but evaluations of ortality from neurological diseases and industrial air pollution are ab- ent. Finkelstein and Jerrett (2007) found an association between the isk of Parkinson’s disease and ambient levels of manganese in total sus- ended particles in the city of Hamilton, Ontario, a city with a large teelmaking sector. Yuchi et al. (2020) found that road proximity was ssociated with incidence of non-Alzheimer’s dementia, Parkinson’s dis- ase, Alzheimer’s disease and multiple sclerosis. It is thus important to now of the public health consequences of exposure to specific sources f air pollutants and whether industrial sources of air pollutants may ose higher risks than the overall PM 2.5 , NO 2 and SO 2 . If industrial ource air pollutants generate higher risks relative to their IQRs than otal PM 2.5 , NO 2 and SO 2 , this suggests that analyses using total PM 2.5 , O 2 and SO 2 may fail to identify important public health implications f exposure to air pollutants. Therefore, in this study we investigated the ssociations between emissions of three major industrial air pollutants i.e. PM 2.5 , NO 2 , and sulfur dioxide – SO 2 ) and mortality attributable to lzheimer’s disease using a large, nationally-representative cohort from anada. . Method .1. Study cohorts Our analyses were based on the 2006 Canadian Census Health and nvironment Cohort (CanCHEC) that was assembled by linking the 006 Census long-form questionnaire to tax and mortality databases Tjepkema et al., 2019 ). The 2006 Census long-form questionnaire is istributed to 20% of Canadian households (approximately 3.2 million dults aged 25 or older), with nearly 100% of households in remote ar- as and enumerated Indian reserves. At baseline, the Census contained a ide range of socio-demographic variables such as age, sex, educational ttainment, employment status, occupational level, visible minority, im- igration status, and marital status. The annual geographically-adjusted ncome for 2006 to 2016 obtained from the annual tax records was also ncluded in the CanCHEC. The respondents who completed the 2006 Census long-form ques- ionnaire were linked to income tax files to obtain annual postal ode history through standard deterministic and probabilistic linkage echniques using sex, date of birth, postal code, and marital status Tjepkema et al., 2019 ). Subsequently, tax-linked Census respondents ere deterministically linked to the Amalgamated Mortality Database AMDB) using social insurance numbers. The AMDB is a dataset that in- ludes death records from both the Canadian Mortality Database which ompiles provincial and territorial hospital death registries beginning n 1950, and deaths recorded in tax files. Deaths that occurred between ensus day (May 16, 2006) and December 31, 2016 were eligible for inkage. Follow-up extended from entry in the cohort (i.e., May 16, 006) until the date of death, reaching age 90, or the end of the study i.e., December 31, 2016). The postal code (as derived from tax files) as an increasingly low correspondence to where someone lives (be- ause the taxes are filed by a third party and their postal code is used nd not that of the person) as Canadians get into older age, especially 2 ver age of 90. Thus, we censored at age 90 in addition to being consis- ent with many other CanCHEC air pollution studies ( Zhao et al., 2021 ). eaths from Alzheimer’s disease were identified by the ICD-10 code of 30. Since the purpose of this analysis was to evaluate the long-term ffects of air pollution exposure, the study population was restricted to dults aged 25 to 90 years at enrollment. .2. Industrial air pollution emissions The wind and inverse-distance weighted (WIDW) industrial PM 2.5 mission was estimated as per Buteau et al. (2018) , and we improved heir estimates by including all (not only the closest major) emitters ithin a large geographic area. Specifically, we computed distances be- ween the participant’s residence and the nearby industries using the entroid of the six-digit residential postal code and the complete address f the industries as given in the National Population Release Inventory NPRI). The NPRI includes industrial emission data as well as addresses f industries that release PM 2.5 , NO 2 and SO 2 . We then estimated annual ndustrial PM 2.5 emissions by multiplying tons of the pollutant emitted rom all industrial emitters located within a 15 km radius (doubling the riginal 7.5 km radius) of a centroid of a postal code area by the in- erse of the distance between the centroid to the emitters, and by the ercentages of time per year that the postal code area was downwind f the industrial emitters. That is, the industrial PM 2.5 exposure for the th participant ( E j ) for a specific year was estimated by the following quation: 𝑗 = 𝑛 ∑ 𝑖 =1 𝐴 𝑖 × 𝑃 𝑖 ÷𝐷 𝑖 here E j is the industrial PM 2.5 exposure for jth participant for a specific ear with the unit of tons/meter, i represents an industrial source within 5 km to a postal code, A i denotes the amount of an industrial pollutant i.e. PM 2.5 ) emitted from the ith industrial source with the unit of tons or the year of interest, P i represents the percentage of hours per year hat the postal code was downwind of the source, and D i is the distance rom the ith industrial source to the postal code with the unit of meter. sing the same method, we estimated WIDW industrial NO 2 and SO 2 missions for each postal code for each year. Hourly wind direction data ere retrieved from the National Climatic Data Access Integration by the weathercan ” package ( LaZerte and Albers, 2018 ) in the R statistical omputing environment. Previous studies used radiuses of 2.5 km and .5 km from the residence to define the exposure ( Labelle et al., 2015 ; uteau et al., 2018 ). We extended it to 15 km to capture more indus- ries since air pollution can travel further distances ( National Research ouncil, 2010 ). The unit of the wind- and inverse-distance weighted ndustrial pollution to which a person is exposed yearly is defined as ons/meter per year. Following guidance from the Canadian Council of Ministers of the nvironment (2012) , we created a three-year moving-average exposure indow with a one-year lag for each of the follow-up years. These oving-averaged annual industrial air pollution emissions were as- igned to each participant based on their six-digit residential postal odes (obtained from the Historical Tax Summary File). .3. Environmental, geographical, and contextual covariates The presence of vegetation around people’s homes has been associ- ted with many benefits to health ( Lee and Maheswaran, 2011 ). Nor- alized Difference Vegetation Index (NDVI) is the most widely used atellite-derived indicator of the quantity of vegetation on the ground. e assigned decadal mean NDVI values for 2001 to 2010 to each par- icipant’s postal code and adjusted for NDVI in our models, includ- ng potential effect modification. The red and near-infrared bands re- ectance used to calculate NDVI was collected by Landsat 5 and Land- at 8 satellites during the vegetation growing season of May to August Crouse et al., 2017 ; CANUE, 2018 ). S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 a h t s r s C c m d c C ( M ( 4 i v a y T t u r k f i l p p u “ g s r c t a W c d D p 2 c f a 6 s t s g 9 o t W a a 3 t A M y 7 p i t s l r d s t i o d i ( ( N m f C a e e c r i m T w T t o b w f s a d m t 6 n 1 ( P p t N g N t e t We included airshed as a geographic covariate in our analyses. An irshed is an area where the movement of air, and air pollutants, can be indered by local geographical features, thus creating a region of rela- ively homogeneous air quality. These geographical features have been hown to correlate with regional differences in morbidity and mortality ates across Canada ( Pappin et al., 2019 ). Canada is divided into six air- hed regions (i.e., Western, Prairie, West Central, Southern Atlantic, East entral, and North) based on large-scale air masses and meteorological haracteristics ( Crouse et al., 2016 ). The Canadian Marginalization Index (Can-Marg) is a census-based easure to reflect four dimensions of Canadian marginalization: resi- ential instability, material deprivation, dependency, and ethnic con- entration ( Matheson et al., 2012 ). Previous studies showed that an-Marg has been associated with differences in health outcomes Matheson et al., 2012 ; Pappin et al., 2019 ). In this study, we used Can- arg available at the dissemination area level. A dissemination area DA) is the smallest standard geographic area (each DA includes about 00-700 people on average) for which all census data are disseminated n Canada. Following Gordon and Janzen (2013) , we developed an urban form ariable to consider the possible effects of different built environments nd neighborhoods within communities on health for the 2006 Census ear and linked to individuals by their six-digit residential postal codes. he types of urban form were defined by population density and major ransportation modes in census tracts. In this study, the categories of the rban form variable included urban core, transit-reliant suburb, auto- eliant suburb, exurban, and rural. The Charlson Comorbidity Index (a measure of frailty) is a well- nown summary comorbidity index which is based on 17 conditions rom all annual episodes of hospitalizations ( Austin et al., 2015 ). The ndex is created by assigning a weighted integer of “one ” to the fol- owing ten conditions (myocardial infarction, congestive heart failure, eripheral vascular disease, cerebrovascular disease, dementia, chronic ulmonary disease, connective tissue disease-rheumatic disease, peptic lcer disease, mild liver disease, and diabetes without complications), two ” to the next four conditions (diabetes with complications, paraple- ia and hemiplegia, renal disease, and cancer), “three ” to moderate or evere liver disease, and “six ” to metastatic carcinoma, and HIV/AIDS epresenting the most severe morbidity. The summation of the weighted omorbidity scores results in a summary score ranging from (0-33). We hen assigned the 3-year lag index score to each participant yearly, to ccount for the lag-effect of those comorbidities in the survival model. e categorized the index score into no comorbidity (0 score), moderate omorbidity (1-3 score), and high comorbidity (4 + score). The Cana- ian Institute of Health Information Discharge Abstract Database (CIHI- AD) was used to obtain records of hospitalization during the study eriod. .4. Statistical methods Cox proportional hazards models were used to assess the asso- iations between industrial PM 2.5 /NO 2 /SO 2 emissions and mortality rom Alzheimer’s disease. The primary model was stratified by sex nd age group (from 25 to 54 years in a 10-year increment, 55- 9, and 70-90) and adjusted for educational attainment, employment tatus, income, occupational level, visible minority status, immigra- ion status, marital status, Can-Marg variables, urban form, and air- hed. The baseline hazard function was stratified by sex and age roup (from 25 to 54 years in a 10-year increment, 55-69, and 70- 0), and censored at 90 years of age. We evaluated the sensitivity f the primary model by adjusting for the other industrial air pollu- ants, greenness (i.e. NDVI), and comorbidity variables, individually. e further assessed possible effect modification across subgroups of sex, ge group, urban form, frailty (i.e. Charlson Comorbidity Index), and irshed. 3 . Results We identified 4500 deaths, of which 3700 deaths had full indus- rial air pollution emissions and covariate information available, from lzheimer’s disease, by following 3,184,500 adults (51.8% males) from ay 16, 2006 to December 31, 2016 for a total of 32,909,200 person- ears. Approximately 95.6% of Alzheimer’s disease deaths were in the 0 + age subgroup, while this subgroup only contained 25.6% partici- ants. Additionally, 26.8% of the deaths were in the decile of the lowest ncome with only 6.5% participants. The number of deaths in each of he other subgroups was generally proportional to the population in the ubgroups. Neighborhood industrial PM 2.5 emissions are more closely corre- ated with industrial NO 2 emissions (Pearson’s correlation coefficient, = 0.58, p < 0.01) than industrial SO 2 emissions ( r = 0.28, p < 0.01). In- ustrial NO 2 emissions are also correlated with industrial SO 2 emis- ions ( r = 0.27, p < 0.01). Detailed participant characteristics and indus- rial pollution emissions across subgroups of each covariate are shown n Table 1 . There were no significant differences in air pollutants by utcome status (not shown). We observed a significantly increased mortality risk of Alzheimer’s isease with every interquartile range (0.53 tonnes/meter per year) ncrement in industrial PM 2.5 emissions from the primary Cox model hazard ratio, HR: 1.006; 95% confidence interval, CI: 1.000–1.011) Table 2 ). The HR increased after additionally adjusting for industrial O 2 and industrial SO 2 either in a two-pollutant or a three-pollutant odel. The estimated risk remained consistent following the adjustment or greenness (NDVI) and decreased slightly with the addition of the harlson Comorbidity Index to the primary model. The adjusted HRs nd corresponding 95% CIs from primary and sensitivity analysis mod- ls did not suggest conclusive relationships between industrial NO 2 /SO 2 missions and mortality due to Alzheimer’s disease ( Table 2 ). When in- luded as potential confounders, there were no effects on the hazard atios in Table 2 from normalized difference vegetation index (NDVI), ncome, education, employment, or occupation, individually. Further- ore, these confounders were not significant factors for the models in able 2 and were thus not included in the stratified analysis. Effect modification of industrial PM 2.5 , NO 2 , and SO 2 emissions ith mortality of Alzheimer’s disease by many covariates is provided in able 3 . In the subgroup analyses, a significant positive association be- ween industrial PM 2.5 emissions and Alzheimer’s disease mortality was bserved in males but not in females. Similar significant associations etween industrial PM 2.5 emissions and Alzheimer’s disease mortality ere seen in the older age (70 + years), no comorbidities, and urban orm subgroups, while the risks were relatively higher in auto-reliant uburbs and the East Central airshed (Ontario and Quebec) ( Table 3 ). By age group, HRs (95%CI) were 0.905 (0.740 – 1.106) for under 65s nd 1.006 (1.000 – 1.013) for over 65s for PM 2.5 . While the effect of in- ustrial pollutants on AD were not statistically significant on age < 65, ostly due to a small number of AD deaths for this age group, the sta- istically significant associations were observed with PM 2.5 for the over 5 group. We found that none of the industrial pollutants have a sig- ificant impact on total non-accidental mortality (PM 2.5 = HR (95%CI): .001 (0.999 – 1.002); NO 2 = HR (95%CI: 0.995 (0.993 – 0.997); SO 2 95%CI = HR: 0.999 (0.998 – 1.000)), suggesting that industrial sourced M 2.5 may negatively impact AD and may be of significance in elderly opulations given the widespread exposure, or in individuals exposed o very high levels of PM 2.5 . We observed a significant positive association between industrial O 2 emissions and mortality only in the West Central airshed sub- roup, with 100 deaths. However, in the other subgroups, industrial O 2 /SO 2 emissions were not significantly associated with mortality due o Alzheimer’s disease ( Table 3 ). Compared to other findings in the literature, the median and IQR for stimated exposure levels of industrial pollutants of PM 2.5 and SO 2 for he province of Quebec were compared between this study and those S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 Table 1 Distribution of Alzheimer’s mortality and industrial pollutants across subgroups of covariates. Covariate Subgroup Person-years Number of deaths ∗ Median (IQR) – tonnes/meter per year PM 2.5 NO 2 SO 2 Sex Male 17,130,400 2,600 0.16 (0.53) 0.27 (3.79) 1.49 (5.31) Female 15,778,900 1,900 0.16 (0.53) 0.25 (3.67) 1.42 (5.25) Age < 65 24,773,200 100 0.22 (0.59) 1.90 (6.11) 0.49 (4.92) ≥ 65 8,136,000 4,400 0.20 (0.56) 1.80 (5.92) 0.42 (4.88) Educational attainment Not completed high school 6,476,700 2,000 0.12 (0.52) 0.12 (3.21) 1.03 (4.92) High school 9,008,747 1,200 0.15 (0.52) 0.22 (3.43) 1.33 (5.11) University 10,527,900 900 0.15 (0.51) 0.22 (3.43) 1.33 (5.12) Master or higher degree 6,895,700 400 0.23 (0.55) 0.47 (4.47) 2.18 (5.81) Employment status Employed 21,684,200 300 0.16 (0.52) 0.26 (3.62) 1.48 (5.20) Unemployed 1,212,700 < 100 0.15 (0.55) 0.18 (3.57) 1.30 (5.19) Not in labor force 10,012,300 4,200 0.16 (0.55) 0.25 (4.04) 1.43 (5.46) Income 1st decile - highest 3,625,900 500 0.15 (0.47) 0.23 (3.65) 1.39 (4.78) 2nd decile 3,559,900 200 0.15 (0.48) 0.22 (3.58) 1.30 (4.88) 3rd decile 3,489,500 200 0.15 (0.49) 0.22 (3.55) 1.31 (4.91) 4th decile 3,390,000 200 0.15 (0.51) 0.23 (3.61) 1.33 (5.02) 5th decile 3,293,300 300 0.16 (0.53) 0.25 (3.68) 1.39 (5.20) 6th decile 3,187,700 300 0.16 (0.54) 0.26 (3.77) 1.45 (5.33) 7th decile 3,070,300 400 0.17 (0.56) 0.28 (3.80) 1.51 (5.47) 8th decile 2,869,000 400 0.18 (0.57) 0.29 (3.86) 1.57 (5.59) 9th decile 2,592,100 500 0.19 (0.60) 0.33 (3.95) 1.73 (5.91) 10th decile - lowest 2,355,800 1,100 0.19 (0.61) 0.33 (4.19) 1.76 (6.05) Occupational level Management 2,641,700 < 100 0.16 (0.49) 0.24 (3.42) 1.50 (5.00) Professional 4,518,000 < 100 0.20 (0.51) 0.37 (4.21) 1.86 (5.43) Skilled, technical & supervisor 7,458,300 100 0.13 (0.49) 0.19 (3.35) 1.22 (4.82) Semi-skilled 7,206,200 100 0.16 (0.54) 0.25 (3.45) 1.42 (5.26) Unskilled 2,394,000 < 100 0.16 (0.57) 0.25 (3.63) 1.43 (5.58) No occupation 8,691,100 4,200 0.17 (0.57) 0.27 (4.26) 1.47 (5.56) Visible minority Not visible minority 27,271,600 4,300 0.14 (0.48) 0.23 (4.00) 1.20 (4.90) Visible minority 5,637,600 200 0.32 (0.68) 0.44 (2.94) 2.55 (6.46) Immigration status Non-immigrant 25,520,200 3,600 0.11 (0.44) 0.17 (3.63) 0.99 (4.54) Immigrant 7,389,000 900 0.36 (0.68) 0.70 (4.01) 2.89 (6.54) Marital status Single, never married 4,694,500 300 0.27 (0.63) 0.61 (5.55) 2.15 (6.58) Common - law 3,942,500 100 0.12 (0.48) 0.26 (4.71) 1.03 (4.82) Married 19,672,000 2,600 0.15 (0.50) 0.18 (2.97) 1.33 (4.86) Separated 914,600 < 100 0.19 (0.59) 0.35 (4.18) 1.73 (5.86) Divorced 2,075,300 200 0.20 (0.59) 0.48 (5.21) 1.82 (6.12) Widowed 1,610,400 1,200 0.18 (0.58) 0.30 (4.38) 1.56 (5.68) Can-Marg: Residential instability Q1 - lowest 7,737,800 700 0.10 (0.41) 0.08 (1.67) 0.92 (3.80) Q2 8,714,000 1,100 0.08 (0.36) 0.04 (2.19) 0.62 (3.55) Q3 6,541,500 900 0.17 (0.57) 0.27 (3.78) 1.53 (5.38) Q4 5,687,900 1,000 0.25 (0.63) 0.71 (5.90) 2.20 (6.11) Q5 - highest 4,228,000 900 0.38 (0.72) 1.80 (9.62) 3.67 (8.20) Can-Marg: Dependency Q1 - lowest 6,119,300 500 0.20 (0.51) 0.56 (4.39) 2.18 (5.66) Q2 5,480,400 500 0.19 (0.56) 0.37 (3.79) 1.72 (5.76) Q3 5,218,100 600 0.24 (0.62) 0.31 (4.14) 1.85 (5.90) Q4 6,549,600 1,000 0.16 (0.53) 0.28 (3.64) 1.36 (5.38) Q5 - highest 9,541,839 2,000 0.10 (0.43) 0.04 (3.05) 0.59 (3.95) Can-Marg: Material deprivation Q1 - lowest 7,040,200 800 0.14 (0.39) 0.22 (3.13) 1.43 (4.34) Q2 6,316,000 800 0.16 (0.46) 0.25 (3.00) 1.62 (5.07) Q3 6,509,800 900 0.14 (0.52) 0.16 (2.70) 1.25 (4.81) Q4 5,572,700 800 0.25 (0.67) 0.59 (5.17) 1.93 (6.64) Q5 - highest 7,470,400 1,200 0.15 (0.66) 0.24 (5.41) 1.10 (5.78) Can-Marg: Ethnic concentration Q1 - lowest 10,170,500 1,400 0.03 (0.22) 0.01 (1.78) 0.17 (2.23) Q2 8,085,800 1,400 0.11 (0.39) 0.12 (3.56) 0.74 (3.81) Q3 5,742,800 800 0.18 (0.49) 0.42 (4.41) 1.71 (5.00) Q4 4,529,900 500 0.33 (0.63) 1.09 (7.09) 2.95 (6.95) Q5 - highest 4,380,100 400 0.54 (0.68) 1.27 (5.87) 4.46 (6.99) Urban form Active urban core 2,796,000 500 0.39 (0.60) 1.77 (7.80) 4.20 (7.33) Transit-reliant suburb 2,287,700 300 0.51 (0.73) 1.85 (9.76) 4.22 (8.17) Car-reliant suburb 14,430,900 1,800 0.25 (0.59) 0.70 (5.61) 2.26 (6.28) Exurban 1,873,700 200 0.05 (0.22) 0.02 (1.84) 0.39 (2.19) Rural 11,520,900 1,600 0.01 (0.15) 0.01 (0.47) 0.07 (1.36) Airshed Western 4,109,600 600 0.16 (0.57) 0.11 (1.69) 1.07 (6.29) Prairie 4,431,700 400 0.12 (0.40) 0.22 (5.17) 1.97 (5.39) West Central 1,817,500 200 0.07 (0.19) 0.10 (1.41) 0.14 (0.56) Southern Atlantic 2,943,900 500 0.02 (0.29) 0.12 (4.72) 0.10 (2.72) East Central 19,317,500 2900 0.22 (0.59) 0.37 (5.34) 1.93 (5.53) Northern 289,000 < 100 0.01 (0.25) 0.01 (0.08) 0.38 (2.33) IQR: interquartile range, Can-Marg: the Canadian Marginalization Index. ∗ Numbers of deaths were rounded up to the nearest 100 due to confidentiality and may not add up to the total. 4 S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 Table 2 Hazard ratios (HRs) and 95% confidence intervals (CIs) from different Cox proportional hazards models for time to deaths due to Alzheimer’s, per interquartile range increase in industrial PM 2.5 (0.53 tonnes/meter per year), NO 2 (5.28 tonnes/meter per year), and SO 2 (3.73 tonnes/meter per year) emissions respectively. Model ∗∗ HR (95% CI) PM 2.5 NO 2 SO 2 Primary model ∗∗ 1.006 (1.000-1.011) 0.994 (0.978-1.010) 0.998 (0.996-1.001) Primary model + PM 2.5 - 0.984 (0.966-1.002) 0.997 (0.995-1.001) Primary model + NO 2 1.009 (1.004-1.015) - 0.998 (0.995-1.001) Primary model + SO 2 1.008 (1.003-1.012) 0.999 (0.982-1.017) - Primary model + PM 2.5 + NO 2 - - 0.998 (0.995-1.001) Primary model + PM 2.5 + SO 2 - 0.990 (0.971-1.009) - Primary model + NO 2 + SO 2 1.009 (1.004-1.015) - - Primary model + NDVI (greenness) 1.006 (1.000-1.011) 0.993 (0.976-1.008) 0.998 (0.996-1.001) Primary model + Charlson Comorbidity Index 1.005 (1.000-1.011) 0.992 (0.976-1.008) 0.998 (0.995-1.001) Note: Values in bold are statistically significant at p < 0.05. ∗∗ The primary model is stratified by sex, age, and adjusted for educational attainment, income adequacy quintile, employment status, occupational group, visible minority status, immigration status, marital status, Can-Marg variables (instability, ethnic concentration, material deprivation, and dependency), urban form, and airshed. Table 3 Effect modification of hazard ratios (HRs) and 95% confidence intervals (CIs) for industrial PM 2.5 , NO 2 , and SO 2 emissions with mortality of Alzheimer’s disease. Subgroup The number of deaths ┼ HR (95% CIs) PM 2.5 NO 2 SO 2 Sex Male 2100 1.014 (1.004-1.023) 0.993 (0.972-1.014) 0.998 (0.995-1.001) Female 1600 1.003 (0.992-1.013) 0.996 (0.973-1.019) 0.998 (0.995-1.002) Age group < 65 100 0.905 (0.740-1.106) 0.777 (0.585-1.031) 0.934 (0.845-1.031) ≥ 65 3600 1.006 (1.000-1.013) 0.990 (0.974-1.006) 0.997 (0.995-1.000) Urban form Urban core 400 1.001 (0.971-1.032) 0.962 (0.900-1.029) 0.988 (0.976-1.001) Transit-reliant suburb 300 0.925 (0.853-1.002) 0.951 (0.894-1.013) 0.990 (0.974-1.007) Auto-reliant suburb 1600 1.018 (1.005-1.030) 1.006 (0.980-1.032) 0.997 (0.993-1.002) Exurban 200 1.023 (0.982-1.067) 1.033 (0.981-1.088) 0.994 (0.977-1.012) Rural 1300 1.006 (1.000-1.012) 1.002 (0.978-1.026) 1.000 (0.999-1.002) Frailty (Charlson Comorbidity Index) No comorbidities (0 index) 2400 1.006 (1.000-1.012) 0.989 (0.968-1.010) 0.999 (0.996-1.001) Moderate (1-3 index) 1200 0.999 (0.975-1.024) 1.001 (0.977-1.025) 0.996 (0.992-1.001) High (4 + index) 100 0.989 (0.906-1.080) 0.946 (0.851-1.051) 0.959 (0.911-1.008) Airshed Western 400 1.006 (0.960-1.005) 0.935 (0.839-1.042) 0.998 (0.996-1.001) Prairie 300 0.979 (0.981-1.076) 1.021 (0.973-1.072) 1.001 (0.995-1.008) West Central 100 1.032 (0.981-1.086) 1.059 (1.000-1.121) 1.000 (0.995-1.002) Southern Atlantic 400 1.004 (0.994-1.015) 0.996 (0.961-1.032) 1.000 (0.991-1.009) East Central 2400 1.008 (1.000-1.016) 0.988 (0.967-1.009) 0.997 (0.993-1.001) Northern < 100 - - - Notes: Values in bold are statistically significant at p < 0.05. + Numbers of deaths were rounded up to the nearest 100 for the demonstration purpose due to confidentiality and may not add up to the total, while we used precise values of the deaths for the modelling analyses. The analysis was not performed if the number of deaths in the subgroup is less than 100. Frailty represents the Charleston Comorbidity Index. c m e S w t 4 i w t b ( p a 7 i p Y h e t ( d s t ( i i s i alculated by Buteau et al. (2018) ( Table 4 ). Our weighted estimates for edian and IQR for PM 2.5 and SO 2 are larger than Buteau’s weighted stimates. Similarly, our weighted estimate for the median and IQR of O 2 is larger than Buteau’s unweighted estimate. However, Buteau’s un- eighted estimate for the median of PM 2.5 is larger than ours. In addi- ion, the IQRs for our estimates were larger. . Discussion In this study, we observed significant positive associations between ndustrial PM 2.5 emissions and mortality from Alzheimer’s disease, hile the association between industrial NO 2 /SO 2 emissions and mor- ality was not clear. We used time-varying emission estimates to consider residential mo- ility of participants and potential calendar-year trends in the emissions Zhao et al., 2021 ). Compared with Buteau et al.’s (2018) study, we com- rehensively estimated industrial air pollutant emissions by considering ll, not only the closest major, emitters in a larger buffer area (15 km vs. 5 .5 km). These factors contributed to the higher emission level estimates n this study compared to Buteau et al. (2018) . Oxidative stress can be regarded as one factor that plays an im- ortant role in the development of Alzheimer’s disease ( Moulton and ang, 2012 ) and the appearance of senile plaques is the preceding allmark of the disease ( Markesbery, 1997 ). Increasing toxicological vidence shows that exposure to PM 2.5 is likely to induce respira- ory and systemic inflammation, resulting in chronic oxidative stress Migliore and Coppede, 2009 ; Ranft et al., 2009 ; Tsai et al., 2019 ). Ad- itionally, exposure to PM 2.5 intensifies generation of reactive oxygen pecies that impairs the blood-brain barrier and increases the produc- ion of amyloid-beta peptides, the principal component of senile plaques Mark, et al., 1996 ; Ranft et al., 2009 ). Thus, exposure to ambient PM 2.5 s treated as a risk factor for Alzheimer’s disease. The above toxicolog- cal plausibility is demonstrated by epidemiological studies, in which ignificant associations between ambient PM 2.5 exposure and morbid- ty/mortality from Alzheimer’s disease were observed ( Shi et al., 2020 ; S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 Table 4 Median and IQR for estimated PM 2.5 and SO 2 emissions in this study compared to Buteau et al. (2018) . Pollutant Estimated emissions in this study Estimated emissions in Buteau et al. (2018) Weighted (tons) 1 Weighted (tons) 2 Weighted (tons) 2 Unweighted (tons) 3 median IQR median IQR median IQR Median IQR PM 2.5 0.16 0.53 160 530 36 97 177 96 SO 2 1.45 3.73 1450 3730 71 255 299 1075 Note: Pollutants in this study are reported in tons/meter/year but are converted to tons/km/year here for comparison purposes (km = meter ∗ 1000). 1 Weighted by wind and inverse distance in meters (m). 2 Weighted by w i nd and inverse distance in kilometers (km). 3 Emissions by the nearest major emitter within 7.5 km of the residence (unweighted). T v fi a f a s s t t a l t p v h g t i s i o ( t t r a h p i i h o p t f a c A s t c I a o p l N a a o o o e i l t s t i i C r r s i p t b o s i 4 a t s m f a t p A m 2 g c t b s r s d r f hiankhaw et al., 2022 ) in different age groups. Differing from the pre- ious epidemiological studies that focused on ambient PM 2.5 , this is the rst study to observe an association between industrial PM 2.5 exposure nd mortality due to Alzheimer’s disease. In contrast to studies on PM 2.5 and alzheimer’s disease, there are ewer studies focusing on the associations between ambient NO 2 /SO 2 nd alzheimer’s disease, and findings in those studies do not show con- istent positive associations ( Fu and Yung, 2020 ). Ku et al. (2016) ob- erved injuries of neurobehavior induced by co-exposure (inhalation) o ambient PM 2.5 , NO 2 and SO 2 in mice brains. Similarly, deteriora- ion of spatial learning and memory, as well as the accumulation of myloid proteins, pathological abnormalities, and cognitive defects re- ated to alzheimer’s disease, were observed in mice following exposure o NO 2 inhalation ( Yan et al., 2016 ). In addition, levels of rat synapto- hysin, a crutial synaptic vesicle membrane protein involved in synaptic esicle trafficking and neurotransmitter release, were decreased in rat ippocampus after SO 2 inhalation ( Yun et al., 2013 ). These results sug- ested that SO 2 exposure may lead to synaptic depression and contribute o cognitive decline ( Yun et al., 2013 ). Although injuries of neurobehav- or induced by ambient NO 2 /SO 2 inhalation have been found in these tudies ( Ku et al., 2016 ; Yun et al., 2013 ), a clear association between ndustrial NO 2 /SO 2 emission and deaths from AD was not observed in ur study. Relatively low neighborhood industrial NO 2 /SO 2 emissions i.e., for many participants of this study, no industrial NO 2 /SO 2 emit- ers within the 15 km buffer of their residences) may have contributed o the null association. Significant associations were found for the East Central airshed. This egion contains the urbanized areas of Quebec and southwestern Ontario nd high levels of industrial pollution, while other airsheds may have ad low statistical power due to lower populations and lower levels of ollutants. It has been previously shown that airshed can significantly mpact health ( Zhao et al., 2018 ; Sheridan and Kalkstein, 2004 ) and s specifically a modifier for the effect of air polluants on neurological ealth ( Zhao et al., 2021 ). Additionally, we found a significant effect f age in those greater than 65. Neuro- inflammation is increasingly revalent in aged tissues ( Mumaw et al., 2016 ), which may explain why hose > 65 years of age are more vulnerable to its effects. Although hospital visits due to dementia has been used as one of the actors in computing the Charlson comorbidity index and was used as measure to adjust for frailty in the study, we still cannot readily dis- ern whether industrial PM 2.5 emissions led to an increased incidence of lzheimer’s disease and consequently caused more deaths, or if the emis- ion caused more mortality in people who already had the disease. In he future, we will integrate administrative health data such as the Dis- harge Abstract Database (DAD) from the Canadian Institute of Health nformation (CIHI) with the CanCHEC to better understand the associ- tions with Alzheimer’s disease as an underlying vs. contributing cause f death. It is important to understand the public health consequences of ex- osure to specific sources of air pollutants. If industrial-source air pol- utants generate higher risks relative to their IQRs than total PM 2.5 , O and SO , this would suggest that analyses using total PM , NO 2 2 2.5 2 6 nd SO 2 may underestimate important public health implications of ir pollutant exposure. Our study suggests that controlling emissions f PM 2.5 from those industrial sources may curb the mortality burden f Alzheimer’s disease by controlling the overall environmental burden f PM 2.5 . The effect sizes observed in this study are small. However, the stimated contributions of industry sources to total PM in Canada s also small (14.2%) ( Meng et al., 2019 ). If this effect is calcu- ated/interpolated as a result of an increase equivalent to the IQR of otal PM 2.5 , we would expect an IQR increase in total PM 2.5 to be as- ociated with a 7% increase in AD mortality, which is similar in size o what Chen et al. (2017b) found between total PM 2.5 and dementia ncidence, with an HR of 1.06 (95% CI: 1.05–1.07) per IQR increase n PM 2.5 exposure after adjusting for age, sex, and region in Ontario, anada. Observational studies such as this one do not infer causality, but ather suggest and support hypotheses and suggest areas of further esearch. Causality is judged on several factors such as repeatability, trength of association, and biological plausibility. Multiple compar- sons increase the risk of false positives but the results need to be inter- reted based on other relevant evidence. Based on previous literature, he comparisons we made were reasonable and they could have either een made in one study or as part of several separate studies and the risk f false positives would be the same overall. Results from observational tudies are strengthened by similar findings from different populations n different countries done by different investigators. .1. Strengths and limitations Approximately 3.2 million participants living in not only urban but lso rural and remote, northern regions of Canada were included, and his large representative population is the principal strength of this tudy. Another noticeable strength is a relatively comprehensive adjust- ent for demographic, socioeconomic, environmental, and geographic actors. In particular, we included the Charlson Comorbidity Index to ssess the possible influence of frailty of individuals on the associa- ion between the exposures and Alzheimer’s disease mortality, since eople with comorbidities are likely dying from other causes but not lzheimer’s disease. This potential confounder was not considered in ost previous studies using the CanCHEC Cohort (e.g. Pappin et al., 019 ; Zhao et al., 2021 ). Additionally, we used a dynamic annual eographically-adjusted income to replace stationary unadjusted in- ome at the baseline. The dynamic annual income may reduce the po- ential error generated by ascertaining employment status as of baseline ecause we found a strong association between reported employment tatus as of census date and annual income through the follow-up pe- iod. Despite the improved estimation on industrial air pollution emis- ions, we acknowledge that diffusion of air pollutants is affected by ad- itional meteorological and geographic factors (e.g. air humidity, ter- ain and mixing height) besides wind ( Liu et al., 2017 , 2018 ). In the uture, we will consider using a complex atmospheric dispersion model S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 ( s p t i s l l o ( t f i a g t l n t m A d 5 s s s d u fl E d F D i t A c t R A A B B B B B B C C C C C C C C C D E E F F G G K K L L L L L M M M e.g. the California Puff) to estimate the industrial air pollutant emis- ions ( Exponent, 2018 ; Zhao et al., 2020 ), though at present, the com- uting time cost of using the atmospheric dispersion model to estimate he emissions at a relatively fine spatial resolution for the whole Canada s nearly unaffordable. Despite the adjustment for 15 covariates in our primary Cox model, ome potential confounders, typically smoking habits, were not col- ected in our cohort. However, based on our previous study using the ung-cancer mortality rate to indirectly adjust for smoking, the effects f smoking on HRs of mortality of neurological diseases are limited Zhao et al., 2021 ). It is important to note that emissions of the three industrial pollu- ants are correlated in space. Thus, the issue of collinearity may con- ound our findings when two or more emission variables were included n information loss ( Bennette and Vickers, 2012 ). Moreover, supposing linear and additive combination of exposures and covariates may also enerate errors ( Bobb et al., 2015 ), though most existing studies used he same supposition. Developing an efficient non-parametric machine earning method may be helpful to solve the above issues. Even though we did not observe significant associations between on-accidental mortality and air pollutants, we still can not ignore he possibility of underestimating the true risk. Since AD is developed ostly at later age there is a chance that the true effect of PM 2.5 on D risk might be underestimated due to possibility that participants are ying from other causes before they can develop dementia. . Conclusion We identified a positive association between industrial PM 2.5 emis- ions and mortality from Alzheimer’s disease. However, we did not ob- erve a clear association of mortality with industrial NO 2 /SO 2 emis- ions. Additional work is needed by using more accurate estimates of in- ustrial pollution emissions, better understanding of the disease course sing hospital records, and adopting novel statistical methods to more exibly model health effects of multiple exposures and covariates. thics Data were analyzed and maintained according to the principles of isclosure and confidentiality as per the Canadian Statistics Act. unding This research received no funding. eclaration of Competing Interests The authors declare that they have no known competing financial nterests or personal relationships that could have appeared to influence he work reported in this paper. cknowledgement The authors would like to thank Statistics Canada for help accessing ensus and socioeconomic data and Health Canada for the support for his publication. eferences ustin, S.R., Wong, Y.N., Uzzo, R.G., Beck, J.R., Egleston, B.L., 2015. Why summary co- morbidity measures such as the Charlson Comorbidity Index and Elixhauser score work. Med. Care 53, e65–e72 . lzheimer’s Association, 2022. 2022 Alzheimer’s Disease Facts and Figures. Avail- able at https://www.alz.org/media/documents/alzheimers-facts-and-figures.pdf . Ac- cessed July 2022. abadjouni, R.M., Hodis, D.M., Radwanski, R., Durazo, R., Patel, A., Liu, Q., Mack, W.L., 2017. Clinical effects of air pollution on the central nervous system; a review. J. Clin. Neurosci. 43, 16–24 . 7 ennette, C., Vickers, A., 2012. Against quantiles: categorization of continuous variables in epidemiologic research, and its discontents. BMC Med. Res. Method. 12, 21 . éjot, Y., Reis, J., Giroud, M., Feigin, V., 2018. A review of epidemiological research on stroke and dementia and exposure to air pollution. Int. J. Stroke 13, 687–695 . obb, J.F., Valeri, L., Henn, B.C., Christiani, D.C., Wright, R.O., Mazumdar, M., Godleski, J.J., Coull, B.A., 2015. Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Biostatistics 16, 493–508 . rookmeyer, R., Johnson, E., Ziegler-Graham, K., Arrighi, M.H., 2007. Forecasting the global burden of Alzheimer’s disease. Alzheimer’s Dementia 3, 186–191 . uteau, S., Doucet, M., Tétreault, L.F., Gamache, P., Fournier, M., Brand, A., Kosatsky, T., Smargiassi, A., 2018. A population-based birth cohort study of the association between childhood-onset asthma and exposure to industrial air pollutant emissions. Environ. Int. 121, 23–30 . akmak, S., Hebbern, C., Vanos, J., et al., 2019. Exposure to traffic and mortality risk in the 1991–2011 Canadian Census Health and Environment Cohort (CanCHEC). Environ. Int. 124, 16–24 . akmak, S., Hebbern, C., Pinault, L., et al., 2018. Associations between long-term PM 2.5 and ozone exposure and mortality in the Canadian Census Health and Environ- ment Cohort (CANCHEC), by spatial synoptic classification zone. Environ. Int. 111, 200–211 . anadian Council of Ministers of the Environment, 2012. Guidance document on achievement determination Canadian ambient air quality standards for fine partic- ulate matter and ozone. Available from: https://publications.gc.ca/collections/ collection_2013/ccme/En108-4-55-2012-eng.pdf , last accessed October 21, 2021. arroll, W.M., 2019. The global burden of neurological disorders. Lancet Neurol. 18, 418–419 . ANUE, 2018. The Canadian Urban Environmental Health Research Consortium (CANUE). Available from: https://canue.ca/wp-content/uploads/2018/11/CANUE- Metadata-NDVI-Landsat-Annual.pdf hen, C.Y., Hung, H.J., Chang, K.H., Hsu, C.Y., Muo, C.H., Tsai, C.H., Wu, T.N., 2017a. Long-term exposure to air pollution and the incidence of Parkinson’s disease: a nested case-control study. PLoS ONE 12 (8), e0182834 . hen, H., Kwong, J.C., Copes, R., Hystad, P., van Donkelaar, A., Tu, K., Jeffrey, R., Brook, J.R., Goldberg, M.S., Martin, R.V., Brian, J., Murray, B.J., Wilton, A.S., Kopp, A., Burnett, R.T., 2017b. Exposure to ambient air pollution and the incidence of dementia: a population-based cohort study. Environ. Int. 108, 271–277. doi: 10.1016/ j.envint.2017.08.020 . rouse, D.L., Philip, S., van Donkelaar, A., Martin, R.V., Jessiman, B., Peters, P.A., We- ichenthal, S., Brook, J.R., Hubbell, B., Burnett, R.T., 2016. A new method to jointly estimate the mortality risk of long-term exposure to fine particulate matter and its components. Sci. Rep. 6, 18916 . rouse, D.L., Pinault, L., Balram, A., Hystad, P., Peters, P.A., Chen, H., et al., 2017. Urban greenness and mortality in Canada’s largest cities: a national cohort study. Lancet Planet. Health 1, E289–E297 . ominici, F., Peng, R., Bell, M.L., Pham, L., McDermott, A., Zeger, S.L., Samet, J.M., 2006. Fine particulate air pollution and hospital admission for cardiovascular and respira- tory diseases. JAMA 295, 1127–1134 . EA, 2022. European Environment Agency. Available at https://www.eea.europa. eu/themes/air/air-pollution-sources-1 . Accessed July 2022. xponent, 2018. CALPUFF Modeling System. Available at http://www.src.com/ . Accessed January 20, 2022. inkelstein, M.M., Jerrett, M., 2007. A study of the relaitonships between Parkinson’s disease and markers of traffic-derived and environmental manganese air pollution in two Canadian cities. Environ. Res. 104, 420–432 . u, P., Yung, K.K.L., 2020. Air pollution and Alzheimer’s disease: a systematic review and meta-analysis. J. Alzheimer’s Dis. 77, 201–214 . ordon, D.L.A., Janzen, M., 2013. Estimating the size of Canada’s suburban population. J. Archit. Plan. Res. 30, 198–220 . onzález-Guevara, E., Martinez-Lazcano, J.C., Custodio, V., Hernandez-Ceron, M., Ru- bio, C., Paz, C., 2014. Exposure to ozone induces a systemic inflammatory response: possible source of the neurological alterations induced by this gas. Inhalation Toxicol. 26 (8), 485–491 . ampa, M., Castanas, E., 2008. Human health effects of air pollution. Environ. Pollut. 151, 362–367 . u, T., Ji, X., Zhang, Y., Li, G., Sang, N., 2016. PM2.5, SO2, and NO2 co-exposure impairs neurobehavior and induces mitochondrial injuries in the mouse brain. Chemosphere 163, 27–34 . abelle, R., et al., 2015. Hospitalizations for respiratory problems and exposure to indus- trial emissions in children. Environ. Pollut. 4, 77–85 . aZerte, S.E., Albers, S., 2018. Weathercan: Download and format weather data from En- vironment and Climate Change Canada. J. Open Source Softw. 3, 571 . ee, A.C.K., Maheswaran, R., 2011. The health benefits of urban green spaces: a review of the evidence. J. Public Health 33, 212–222 . iu, Y., Cao, G., Zhao, N., Mulligan, K., Ye, X., 2018. Improve ground-level PM 2.5 con- centration mapping using a random forests-based geostatistical approach. Environ. Pollut. 235, 272–282 . iu, Y., Zhao, N., Vanos, J.K., Cao, G., 2017. Effects of synoptic weather on ground-level PM2.5 concentrations in the United States. Atmos. Environ. 148, 297–305 . arkesbery, W.R., 1997. Oxidative stress hypothesis in Alzheimer’s disease. Free Radic. Biol. Med. 23, 134–147 . atheson, F.I., Dunn, J.R., Smith, K.L.W., Moineddin, R., Glazier, R.H., 2012. Develop- ment of the Canadian Marginalization index: a new tool for the study of inequality. Can. J. Public Health 103, S12–S16 . ark, R.J., Blanc, E.M., Mattson, M.P., 1996. Amyloid 𝛽-peptide and oxidative cellular injury in Alzheimer’s disease. Mol. Neurobiol. 12, 211–224 . http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0001 https://www.alz.org/media/documents/alzheimers-facts-and-figures.pdf http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0003 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0004 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0005 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0006 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0007 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0008 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0009 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0010 https://publications.gc.ca/collections/collection_2013/ccme/En108-4-55-2012-eng.pdf http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0013 https://canue.ca/wp-content/uploads/2018/11/CANUE-Metadata-NDVI-Landsat-Annual.pdf http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0014 https://doi.org/10.1016/\penalty -\@M j.envint.2017.08.020 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0016 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0017 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0018 https://www.eea.europa.eu/themes/air/air-pollution-sources-1 http://www.src.com/ http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0020 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0022 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0023 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0024 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0025 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0026 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0027 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0028 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0029 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0030 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0031 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0032 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0033 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0034 S. Cakmak, O. Toyib, C. Hebbern et al. Hygiene and Environmental Health Advances 4 (2022) 100019 M M M M M N N O P R S S S T T T T Y Y Y Z Z Z eng, J., Martin, R.V., Li, C., van Donkelaar, A., Tzompa-Sosa, Z.A., Yue, X., Xu, J.-W., Weagle, C.L., Burnett, R.T., 2019. Source contributions to ambient fine particulate matter for Canada. Environ. Sci. Technol. 53 (17), 10269–10278 . igliore, L., Coppede, F., 2009. Environmental-induced oxidative stress in neurodegener- ative disorders and aging. Mutat.Res./Genet. Toxicol. Environ. Mutagen. 674, 73–84 . ills, N., Donaldson, K., Hadoke, P.W., Boon, N.A., MacNee, W., Cassee, F.R., et al., 2009. Adverse cardiovascular effects of air pollution. Nat. Clin. Pract. Cardiovasc. Med. 6, 36–44 . oulton, P.V., Yang, W., 2012. Air pollution, oxidative stress, and Alzheimer’s disease. J. Environ. Public Health, 472751 2012 . umaw, C.L., Levesque, S., McGraw, C., Robertson, S., Lucas, S., Stafflinger, J.E., Campen, M.J., Hall, P., Norenberg, J.P., Anderson, T., Lund, A.K., McDonald, J.D., Ottens, A.K., Block, M.L., 2016. In: Microglial Priming through the Lung-Brain Axis: the Role of Air Pollution-Induced Circulating Factors, 5. Federation of American So- cieties for Experimental Biology, pp. 1880–1891 . ational Research Council, 2010. Global Sources of Local Pollution: an Assessment of Long-Range Transport of Key Air Pollutants to and from the United States. The Na- tional Academies Press, Washington, DC . ichols, E., Szoeke, C.E.I., Vollset, S.E., Abbasi, N., Abd-Allah, F., Abdela, J., et al., 2019. Global, regional, and national burden of Alzheimer’s disease and other dementias, 1990-2016: a systematic analysis for the Global Burden of Disease Study. Lancet Neu- rol. 18, 88–106 . berdörster, G., 2000. Pulmonary effects of inhaled ultrafine particles. Int. Arch. Occup. Environ. Health 74, 1–8 . appin, A.J., Christidis, T., Pinault, L.L., Crouse, D.L., Brook, J.R., Erickson, A., et al., 2019. Examining the shape of the association between low levels of fine particulate matter and mortality across three cycles of the Canadian census health and environment cohort. Environ. Health Perspect. 127, 107008 . anft, U., Schikowski, T., Sugiri, D., Krutmann, J., Kramer, U., 2009. Long-term exposure to traffic-related particulate matter impairs cognitive function in the elderly. Environ. Res. 109, 1004–1011 . heridan, S.C., Kalkstein, L.S, 2004. Progress in heat watch-warning system technology. Bull. Am. Meteorol. Soc. 85, 1931–1941 . hi, L., Wu, X., Yazdi, M.D., Braun, D., Awad, Y.A., Wei, Y., et al., 2020. Long-term effects of PM 2.5 on neurological disorders in the American Medicare population: a longitudi- nal cohort study. Lancet Planet. Health 4, e557–e565 . 8 hou, Y., Huang, Y., Zhu, X., Liu, C., Hu, Y., Wang, H., 2019. A review of the possible associations between ambient PM2.5 exposures and the development of Alzheimer’s disease. Ecotoxicol. Environ. Saf. 174, 344–352 . hiankhaw, K., Chattipakorn, N., Chattipakorn, S.C., 2022. PM2.5 exposure in associa- tion with AD-related neuropathology and cognitive outcomes. Environ. Pollut. 292, 118320 . homson, E.M., 2019. Air pollution, stress, and allostatic load: linking systemic and central nervous system impacts. J. Alzheimer’s Dis. 69, 597–614 . jepkema, M., Christidis, T., Bushnik, T., Pinault, L., 2019. Cohort profile: the Canadian census health and environment cohorts (CanCHECs). Health Rep. 30 (12), 18–26 . sai, T.-L., Lin, Y.-T., Hwang, B.-F., Nakayama, S.F., Tsai, C.-H., Sun, X.-L., Ma, C., Jung, C.-R., 2019. Fine particulate matter is a potential determinant of Alzheimer’s disease: a systemic review and meta-analysis. Environ. Res. 177, 108638 . an, W., Yun, Y., Ku, T., Li, G., Sang, N., 2016. NO 2 inhalation promotes Alzheimer’s disease-like progression: cyclooxygenase-2-derived prostaglandin E2 modulation and monoacylglycerol lipase inhibition-targeted medication. Sci. Rep. 6, 22429. doi: 10.1038/srep22429 . uchi, W., Sbihi, H., Davies, H., Tamburic, L., Brauer, M., 2020. Road proximity, air pollu- tion, noise, green space and neurologic disease incidence: a population-based cohort study. Environ. Health 19, 8 . un, Y., Yao, G., Yue, H., Guo, L., Qin, G., Li, G., Sang, N., 2013. SO2 inhalation causes synaptic injury in rat hippocampus via its derivatives in vivo. Chemosphere 93, 2426–2432 . hao, N., Cao, G., Vanos, J.K., Vecellio, D.J., 2018. The effects of synoptic weather on influenza infection incidences: a retrospective study utilizing digital disease surveil- lance. Int. J. Biometerol. 62, 69–84 . hao, N., Smargiassi, A., Hatzopoulou, M., Colmegna, I., Hudson, M., Fritzler, M.J., Awadalla, P., Bernatsky, S., 2020. Long-term exposure to a mixture of industrial SO 2 , NO 2 , and PM 2.5 and anti-citrullinated protein antibody positivity. Environ. Health 19, 86 . hao, N., Pinault, L., Toyib, O., Vanos, J., Tjepkema, M., Cakmak, S., 2021. Long-term ozone exposure and mortality from neurological diseases in Canada. Environ. Int. 157, 106817 . http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0035 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0036 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0037 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0038 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0039 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0040 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0041 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0042 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0043 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0044 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0045 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0046 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0047 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0048 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0049 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0050 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0051 https://doi.org/10.1038/srep22429 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0053 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0054 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0055 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0056 http://refhub.elsevier.com/S2773-0492(22)00019-8/sbref0057 Industrial air pollutant emissions and mortality from Alzheimer’s disease in Canada 1 Introduction 2 Method 2.1 Study cohorts 2.2 Industrial air pollution emissions 2.3 Environmental, geographical, and contextual covariates 2.4 Statistical methods 3 Results 4 Discussion 4.1 Strengths and limitations 5 Conclusion Ethics Funding Declaration of Competing Interests Acknowledgement References