Ambient Air Pollution and Stroke

Background

Stroke is a leading cause of death in the US¹ and worldwide (www.who.int) and may lead to considerable neurological sequelae including aphasia, paraplegia and dementia. The estimated health care costs of stroke in the US exceed $36 billion per year¹. A large body of evidence supports the association between ambient air pollution exposure and increased cardiovascular mortality and morbidity², but only recently have several studies specifically demonstrated an association with increased stroke risk.

Major sources of air pollution include traffic, power plants and in developing countries, biomass combustion. Both particles and gases are emitted through combustion. Particulate matter with aerodynamic diameter <10 µm (PM₁₀) include ultrafine particles (PM_1.0), fine particles (PM_2.5) and coarse particles (PM_10–2.5). Ultrafine particles are emitted in fresh exhaust and coalesce into PM_2.5 within a short time frame. PM_2.5 includes both local sources from traffic emissions and domestic heating and regional sources from power plants, biogenic emissions and traffic whereas coarse particles are a heterogenous mixture that include road dust, endotoxins, and suspended crustal matter. Carbon monoxide (CO), nitrogen dioxide (NO₂), nitrogen oxides (NO_x), sulfur dioxide (SO₂) and ground-level ozone (O₃) are gaseous pollutants emitted as a result of combustion processes. CO is mainly attributed to mobile sources in urban environments and NO₂ and NO_x are rapidly formed in emissions from combustion sources such as traffic and power plants. The main source of SO₂ is from fossil fuel power plants. Ground-level O₃ is formed as a result of atmospheric reactions of NO₂ with hydrocarbons in the presence of sunlight and is a major constituent of photochemical smog. Several of the mentioned pollutants are regulated based on evidence of adverse health effects³. Possible mechanistic pathways including induction of oxidative stress, inflammation, atherosclerosis and autonomic dysregulation have been outlined in detail^2–4 and are beyond the scope of the current review.

This review aims to assess the current evidence regarding the association of air pollution exposure with incidence of ischemic and hemorrhagic stroke considering long-term and short-term exposure to ambient pollutants.

Long-Term Air Pollution Exposure

Most studies of long-term exposure to air pollution and stroke outcomes have used estimates of exposure at residential address in months to years as a proxy for long-term accumulated individual exposure. Exposure has then been assessed using residential distance to major roadways, measurements from closest available fixed monitor or advanced modeling of pollutants combining fixed monitoring measurements with land-use data, emissions databases, traffic density counts and meteorology incorporated into geographical information systems (GIS). These GIS models can also include population-based data such as average income level and average smoking prevalence.

Long-term air pollution exposure and stroke mortality

Studies considering long-term exposure to air pollution and stroke mortality have reported that living in areas with higher ambient pollution is associated with higher risk of stroke mortality (Table 1). Studies from the UK^{5, 6} and northwest Florida⁷ containing large administrative databases with cause of death, residence, sex and area-based data such as socioeconomic status, urbanization, smoking prevalence and greenness. Living near a main road⁵, traffic sources⁷, point sources of emissions⁷ or higher modeled exposure to PM₁₀, CO and NO_x⁶ were all associated with stroke mortality. Several cohort studies have also studied the association between long-term exposure to air pollution and stroke mortality^8–13. These studies have more detailed individual-level data that improves the ability to adjust for potential confounders that may influence the place of residence and the risk of stroke mortality. Strongest associations were reported in the prospective Womens’ Health Initiative cohort¹¹ that included well-validated outcome assessment. In the Californian residents of the American Cancer Society cohort study⁹, associations were reported for NO₂ and any stroke mortality and borderline significant associations for PM_2.5. In the California Teachers Study¹⁰, however, higher long-term PM₁₀ and PM_2.5 exposure were not associated with cerebrovascular mortality. In 232 rural districts of Japan¹² including 250 stroke deaths, higher long-term PM₁₀ exposure was not associated with stroke mortality. Specific characterization of stroke into types and subtypes was available in two studies, in Shizuoka, Japan¹³ and in Denmark⁸. Yorifuji et al¹³ reported associations between NO₂ and mortality from ischemic stroke and intracerebral hemorrhage but not subarachnoid hemorrhage. Andersen et al⁸ reported borderline significant associations between long-term NO₂ exposure and ischemic stroke but not for hemorrhagic strokes and did not further subtype hemorrhagic strokes.

Table 1.

Studies of Long-term Air Pollution Exposure and Stroke Mortality

Study	Location	Study Design	Stroke Outcome	Relative risk (95% Confidence Intervals)	Exposure
Maheswaran 20035	England and Wales	Ecological	Any stroke	1.05 (1.04, 1.07)	living within 200m of main road compared to ≥1000m
Maheswaran 20056	Sheffield, UK	Ecological	Any stroke	1.37 (1.19, 1.57) PM10 1.26 (1.10, 1.46) CO 1.33 (1.14, 1.56) NOx	highest to lowest quintile of modelled pollutant
Hu 20087	Florida, USA	Ecological	Any stroke	1.09 (1.03, 1.15)*	per 10,000 vehicles/day within census tract
Andersen 20128	Denmark	Cohort	Any stroke Ischemic Hemorrhagic	1.22 (1.00, 1.50) 1.46 (0.90, 2.39) 1.00 (0.76, 1.31)	per interquartile range increase (43%) in mean modeled NO2 since 1971
Jerrett 20139	California, USA	Cohort	Any stroke	1.07 (0.99, 1.15) PM2.5 1.08 (1.02, 1.15) NO2 1.01 (0.92, 1.11) O3	per 5.3 µg/m3 PM2.5 per 4.12 ppb NO2 per 24.2 ppb O3
Lipsett 201110	California, USA	Cohort	Any stroke	0.99 (0.89, 1.09) PM10 1.16 (0.92, 1.46) PM2.5	per 10 µg/m3 mean PM10 1996– 2005, or mean PM2.5 1999–2005
Miller 200711	36 US cities	Cohort	Any stroke	1.83 (1.11, 3.00) PM2.5	annual mean in 2000 at closest monitor per 10 µg/m3
Ueda 201212	Japan	Cohort	Any stroke	0.86 (0.74, 1.01) PM10	per 10 µg/m3 annual mean at closest monitor
Yorifuji 201313	Shizuoka, Japan	Cohort	Any stroke Ischemic Hemorrhagic	1.19 (1.06, 1.34) 1.20 (1.04, 1.39) 1.28 (1.05, 1.57)	per 10 µg/m3 annual mean NO2

^*95% Credible interval from a Bayesian analysis

Abbreviatons: CO, Carbon monoxide. NO₂, Nitrogen dioxide. NO_x, Nitrogen oxides. O₃, Ozone. PM₁₀, Particles with aerodynamic diameter ≤ 10µm. PM_2.5, Fine particles with aerodynamic diameter≤ 2.5µm.

Long-term air pollution exposure and hospitalization for stroke

In studies of long-term exposure to air pollution and hospitalization for stroke, higher exposure at home addresses was also associated with higher risk of admission for stroke in some studies, but results were less consistent than for stroke mortality (Table 2). Most commonly reported pollutants included long-term exposure to PM₁₀^{6, 10, 14, 15}, PM_2.5^{10, 11, 16} and NO_x^{6, 17, 18} or NO₂^{8, 14, 15, 19, 20}. Many of the cohort studies reported positive associations^{8, 10, 11, 14, 21} whereas ecological studies^{6, 15, 19} and case-control studies^{17, 18, 21} showed mixed results. In a random-effects meta-analysis of 11 European cohorts¹⁶, long-term PM_2.5 was associated more strongly associated with stroke in subjects older than 60 years old, never-smokers and among subjects with exposure levels less than 25 µg/m³ (current annual mean air quality standard in Europe). Studies that compared long-term air pollution exposure and hospital admissions according to specific stroke type reported positive associations for NO₂, CO and traffic density and admissions for both ischemic and hemorrhagic stroke¹⁹ in Edmonton, Canada whereas NO₂^{8, 20} in Denmark or NO_x¹⁵ in London, UK demonstrated associations consistent with ischemic stroke but not hemorrhagic stroke. Two studies from Scania, Sweden^{17, 18} only including hospital admissions for ischemic stroke observed associations between higher long-term exposure to NO_x and higher risk of hospital admission for ischemic stroke in participants with diabetes but found no association in the overall population, in smokers, or in participants with hypertension or atrial fibrillation. A recent population-based cohort study in Denmark studying long-term NO₂ and traffic noise exposure and stroke incidence reported positive associations for ischemic stroke in separate analyses for both noise and NO₂ but in combined analyses NO₂ was only associated with fatal ischemic strokes²⁰.

Table 2.

Studies of Long-term Air Pollution Exposure and Hospitalization for Stroke

Study	Location	Study Design	Stroke Outcome	Relative Risk (95% Confidence Intervals)	Exposure
Maheswaran 20056	Sheffield, UK	Ecological	Any stroke	1.13 (0.99, 1.29) PM10 1.11 (0.99, 1.25) CO 1.13 (1.04, 1.27) NOx	highest to lowest quintile of modelled pollutant
Andersen 20128	Denmark	Cohort	Any stroke Ischemic Hemorrhagic	1.05 (0.99, 1.11) 1.05 (0.95, 1.17 0.93 (0.81, 1.07)	per interquartile range increase (43%) in mean modeled NO2 since 1971
Lipsett 201110	California	Cohort	Any stroke	1.06 (1.00, 1.13) PM10 1.14 (0.99, 1.32) PM2.5	per 10 µg/m3 annual mean pollutant at closest monitor
Miller 200711	36 US cities	Cohort	Any stroke	1.28 (1.01, 1.61) PM2.5	per 10 µg/m3 mean of closest monitor during 2000
Maheswaran 201215	London, UK	Ecological	Ischemic Hemorrhagic	1.22 (0.77, 1.93) PM10 1.11 (0.93, 1.32) NO2 0.52 (0.20, 1.37) PM10 0.86 (0.60, 1.24) NO2	per 10 µg/m3 of modelled pollutant exposure
Atkinson 201314	England	Cohort	Any stroke	0.98 (0.95, 1.01) PM10 0.99 (0.95, 1.03) NO2 1.02 (1.00, 1.05) SO2 1.00 (0.97, 1.04) O3	per 3.0 µg/m3 PM10 per 10.7 µg/m3 NO2 per 2.2 µg/m3 SO2 per 3.0 µg/m3 O3 modelled annual mean
Stafoggia 201416	11 cohorts, Europe	Cohort	Any stroke	1.19 (0.88, 1.62)	per 5 µg/m3 annual mean PM2.5
Oudin 200917	Scania, Sweden	Case-control	Ischemic	0.95 (0.86, 1.06)	annual mean modelled NOx of 20–30 µg/m3 vs <10 µg/m3
Oudin 201118	Scania, Sweden	Case-control	Ischemic	In diabetics: 2.0 (1.2, 3.4) high NOx 1.3 (1.1, 1.6) low NOx	Modelled annual NOx High NOx ≥25 µg/m3 Low NOx <15 µg/m3 Reference: non-diabetics with low NOx
Johnson 201019	Edmonton, Canada	Ecological	Any stroke Nonhemorrhagic Hemorrhagic	1.29 (1.16, 1.43) 1.36 (1.19,1.56) 1.46 (1.19, 1.80)	highest (16.7–20.3 ppb) to lowest quintile (10.1–14.0 ppb) of NO2 exposure
Sørensen 201420	Denmark	Cohort	Any stroke Ischemic Hemorrhagic	1.08 (1.01, 1.16) 1.11 (1.03, 1.20) 1.00 (0.80, 1.24)	per 10 µg/m3 annual mean NO2
Johnson 201321	Edmonton, Canada	Case-control	Any stroke Ischemic TIA Hemorrhagic	1.01 (0.94, 1.08) 1.03 (0.94, 1.13) 0.95 (0.86, 1.05) 1.07 (0.92, 1.24)	per 5 ppb NO2

Abbreviatons: CO, Carbon monoxide. NO₂, Nitrogen dioxide. NO_x, Nitrogen oxides. O₃, Ozone. PM₁₀, Particles with aerodynamic diameter ≤ 10µm. PM_2.5, Fine particles with aerodynamic diameter≤ 2.5µm. SO₂, Sulfur dioxide.

Short-term Air Pollution Exposure

Day to day differences in air pollution exposure in the days preceding stroke are used to study possible triggering effects of air pollution on stroke. In time-series analyses, daily counts of stroke deaths or admissions are compared with air pollution levels on the same day or preceding days in a study region. In case-crossover analyses exposure levels preceding stroke mortality or hospitalization in an individual are contrasted with control periods within the same calendar month within each individual controlling for season and day of week and perfectly matching time-invariant patient characteristics by design.

A number of studies have investigated associations between short-term exposure to air pollutants including PM₁₀, PM_2.5, CO, NO₂, SO₂ and O₃ and stroke mortality or hospitalizations for stroke in many cities in North America, Europe and East Asia. Mean levels of pollutants varied considerably between study locations from low polluted cities like Dijon, France (daily mean PM₁₀ 20 µg/m³) to highly polluted cities like Wuhan, China (daily mean PM₁₀ 119 µg/m³).

Short-term air pollution exposure and stroke mortality

A majority of studies investigating short-term exposure to air pollution and stroke mortality have been time-series studies^22–36, the remainder used case-crossover design^37–41. A qualitative summary of the studies is provided in Table 3 (for detailed estimates please see Table I http://stroke.ahajournals.org). Most studies do not differentiate between ischemic and hemorrhagic stroke mortality. Several studies reported associations between short-term exposure to particle matter, including several size fractions, or gases and any stroke mortality. Only a few studies further characterized stroke into ischemic and hemorrhagic stroke mortality^{24, 33–35, 38}. Short-term exposure to particulate matter and gases were associated with both ischemic stroke and hemorrhagic stroke. In Tokyo³⁴ the risk increase for subarachnoid hemorrhage mortality per 10 µg/m³ PM_2.5 or NO₂ was roughly double the risk increase for ischemic or intracerebral hemorrhage mortality. It is possible that these hemorrhages may have more precise temporal relationship between air pollution exposure and the timing of stroke onset leading to less exposure misclassification and more precise estimation of the association⁴². Stronger associations between short-term air pollution exposure and stroke mortality were observed in elderly^{25, 30}, women²⁵ and individuals with a history of diabetes⁴¹ or cardiac disease³⁸ in some but not all studies.

Table 3.

Studies of Short-term Air Pollution Exposure and Stroke Mortality

Study	Location	Study Design	Stroke Outcome	Positive associations*	Null associations†
Chen 201322	8 Chinese cities	Time-series	Any stroke	PM10, NO2, SO2
Hoek 200123	Netherlands	Time-series	Any stroke	Black smoke, CO, SO2, and O3.	PM10 and NO2
Hong 200224	Seoul, Korea	Time-series	Ischemic Hemorrhagic	TSP, CO, NO2, SO2, O3 TSP	CO, NO2, SO2, O3
Hong 200225	Seoul, Korea	Time-series	Any stroke	PM10, CO, NO2, SO2, O3
Kan 200326	Shanghai, China	Time-series	Any stroke	PM10, NO2	SO2
Kettunen 200727	Helsinki, Finland	Time-series	Any stroke	PM2.5, CO in warm season	PM10, Coarse PM, PM0.1, NO2, O3 in warm season. No associations in cold season.
Li 201128	Tianjin, Taiwan	Time-series	Any stroke	PM10 on days with >20°C	PM10 on days with ≤ 20°C
Qian 200729	Wuhan, China	Time-series	Any stroke	PM10
Qian 200730	Wuhan, China	Time-series	Any stroke	NO2	SO2, O3
Qian 200831	Wuhan, China	Time-series	Any stroke	PM10 all days and NO2, SO2 on normal temperature days	O3 all days and NO2, SO2 on high temperature days
Qian 201032	Wuhan, China	Time-series	Any stroke	NO2 in spring. PM10, NO2, SO2 in winter.	PM10 and SO2 in spring. All pollutants summer or fall.
Turin 201233	Takashima, Japan	Time-series	Any stroke Ischemic Hemorrhagic	NO2	Suspended PM, NO2, SO2, O3 Suspended PM, SO2, O3 Suspended PM, NO2, SO2, O3
Yorifuji 201134	Tokyo, Japan	Time-series	Any stroke Ischemic Hemorrhagic	PM2.5, NO2 PM2.5, NO2	PM2.5, NO2
Yorifuji 201335	47 Japanese cities	Time-series	Any stroke Ischemic Hemorrhagic	PM10	PM10 PM10
Zanobetti 200936	112 US cities	Time-series	Any stroke	PM2.5, PMcoarse
Maynard 200737	Massachusetts, USA	Case- crossover	Any stroke	Black carbon	SO4
Qian 201338	Shanghai, China	Case- crossover	Any stroke Ischemic Hemorrhagic	PM10, NO2, SO2 PM10, NO2, SO2 NO2, SO2	PM10
Ren 201039	Massachusetts, USA	Case- crossover	Any stroke	O3
Zeka 200540	20 US cities	Case- crossover	Any stroke	PM10
Zeka 200641	20 US cities	Case- crossover	Any stroke	PM10 if pneumonia or ≥75 years old	PM10 if no pneumonia or ≤75 years old

^*Positive associations with confidence intervals not including the null.

^†Associations with confidence intervals including the null.

Abbreviatons: CO, Carbon monoxide. NO₂, Nitrogen dioxide. O₃, Ozone. PM₁₀, Particles with aerodynamic diameter ≤ 10µm. PM_2.5, Fine particles with aerodynamic diameter≤ 2.5µm. PM_coarse, Coarse particles with aerodynamic diameter between 2.5 and 10 µm in aerodynamic diameter. PM_0.1, Ultrafine particles with less than 0.1 µm aerodynamic diameter. SO₂, Sulfur dioxide. SO₄, Sulfate. TSP, Total suspended particles.

Short-term air pollution exposure and hospitalization for stroke

Studies of short-term air pollution exposure and hospitalization for any stroke have reported mixed results^43–63. However, in contrast to studies investigating short-term exposure to air pollution and stroke mortality that typically use death certificate data, some studies of associations with hospital admissions for stroke have had more data on stroke type. These studies have reported associations between PM₁₀^{45, 46, 64, 65}, PM_2.5^66–68, black carbon⁶⁸, CO^{51, 58, 64}, NO₂^{43, 58, 64, 68}, and O₃^{62, 69, 70} and ischemic stroke (Table 4, for detailed estimates please see Table II http://stroke.ahajournals.org). A majority did not observe associations between air pollutants and hemorrhagic stroke^{45, 58, 62, 65, 69} with a few exceptions^{56, 57, 63, 64, 71} including one study that specifically investigated days in Taiwan polluted by Asian dust storms originating from the Gobi desert⁶³. Of the studies with specific data on subtype of ischemic stroke, PM₁₀, PM_2.5 and O₃ were associated with strokes characterized as large-artery atherosclerotic strokes, small-vessel occlusions, lacunar strokes or TIAs rather than cardioembolic strokes^{46, 67–69}. Stronger associations were reported for recurrent ischemic strokes or history of stroke^{58, 70}, in individuals with diabetes or on diabetes medication^{67, 70}, and with one or more cardiovascular risk factors^{69, 70}. A few studies reported stronger associations between O₃ and ischemic stroke in men than women^{62, 69, 72}. Air pollution on warm days was more strongly associated with both hemorrhagic and ischemic stroke in Taiwan⁶⁴. Associations between air pollution and ischemic stroke were stronger in the warm season in Edmonton, Canada⁵⁸ and Dijon, France⁷⁰ in contrast to Wuhan⁶¹ where associations were stronger in the cold season. Differences may reflect better exposure classification due to time spent outdoors in climates like Edmonton, Canada but may also be due to seasonal interactions between pollutants.

Table 4.

Studies of Short-term Exposure to Air Pollution and Hospital Admissions for Stroke

Study	Location	Study Design	Stroke Outcome	Positive associations*	Null Associations†
Ballester 200143	Valencia, Spain	Time- series	Any stroke	NO2	CO, SO2, O3
Burnett 199944	Toronto, Canada	Time- series	Any stroke		PM10, CO, NO2, O3
Chan 200645	Taipei, Taiwan	Time- series	Any stroke Ischemic Hemorrhagic	PM10, PM2.5, O3	CO, NO2, SO2 PM10, PM2.5, CO, NO2, SO2, O3 PM10, PM2.5, CO, NO2, SO2, O3
Corea 201246	Mantua, Italy	Case- crossover	Any stroke Ischemic	PM10 PM10 in all ischemic, large vessel, small vessel, and lacunar	PM10 in cardioembolic. CO, NO2, SO2, O3.
Jalaludin 200647	Sydney, Australia	Time- series	Any stroke		PM10, PM2.5, CO, NO2, SO2, O3
Larrieu 200748	8 French cities	Time- series	Any stroke		PM10, NO2, O3
Le Tertre 200249	8 European cities	Time- series	Any stroke		PM10, black smoke
Linn 200050	Los Angeles, USA	Time-series	Any stroke	CO, NO2 in spring	PM10, O3
Moolgavkar 200051	Los Angeles, USA	Time- series	Any stroke	PM10, CO, NO2, SO2	PM2.5
Nascimento 201252	Sao Jose Campos, Brazil	Time- series	Any stroke	PM10, SO2	O3
Poloniecki 199753	London, UK	Time- series	Any stroke		Black smoke, CO, NO2, SO2, O3
Pönkä 199654	Helsinki, Finland	Time- series	Any stroke Ischemic	NO2 Total suspended particles
Sunyer 200355	7 European cities	Time- series	Any stroke		SO2
Turin 201256	Takashima, Japan	Time- series	Any stroke Ischemic Hemorrhagic	SO2	PM10,NO2, SO2, O3 PM10,NO2, SO2, O3 PM10,NO2, O3
Villeneuve 200657	Edmonton, Canada	Case- crossover	Any stroke Ischemic TIA Hemorrhagic	PM2.5 SO2	PM10, PM2.5, CO, NO2, SO2, O3 CO, NO2, SO2, O3 PM10, PM2.5, CO, NO2, SO2, O3 PM10, PM2.5, CO, NO2, O3
Villeneuve 201258	Edmonton, Canada	Case- crossover	Any stroke Ischemic Hemorrhagic	CO in warm season CO, NO2,O3 in warm season	PM2.5, NO2, SO2, O3. CO all year. PM2.5, NO2, SO2, O3. CO all year. PM2.5, CO, NO2, SO2, O3
Wong 199959	Hong Kong, China	Time- series	Any stroke		PM10, NO2, SO2, O3
Wordley 199760	Birmingham, UK	Time- series	Any stroke	PM10
Xiang 201361	Wuhan, China	Case- crossover	Any stroke	PM10, NO2 in cold season	PM10, NO2 SO2 all year and in subtypes. PM10, NO2 in warm season.
Xu 201362	Allegheny, USA	Case- crossover	Any stroke Ischemic Hemorrhagic	O3 O3	O3
Yang 200563	Taipei,Taiwan	Time- series	Any stroke Ischemic TIA Hemorrhagic	Asian dust and intracerebral	Asian dust Asian dust Asian dust and subarachnoidal
Tsai 200364	Kaoshiung, Taiwan	Case- crossover	Ischemic Hemorrhagic	PM10, NO2, SO2, O3 warm days, CO all days PM10, CO, NO2, O3 warm days	PM10, NO2, SO2, O3 cool days SO2 warm days, All pollutants cool days
Wellenius 200565	9 US cities	Case- crossover	Ischemic Hemorrhagic	PM10, CO, NO2, SO2	PM10, CO, NO2, SO2
Lisabeth 200866	Corpus Christi, USA	Time- series	Ischemic	PM2.5	O3
O’Donnell 201167	8 Canadian cities	Case- crossover	Ischemic	PM2.5 in diabetics and non- cardioembolic	PM2.5 in ischemic strokes overall
Wellenius 201268	Boston, USA	Case- crossover	Ischemic	PM2.5, black carbon, NO2. PM2.5 large and small vessel stroke	CO, SO4, O3. PM2.5cardioembolic stroke
Henrotin 200769	Dijon, France	Case- crossover	Ischemic Hemorrhagic	O3 in all ischemic, large vessel, and TIA	PM10, CO, NO2, SO2 PM10, CO, NO2, SO2, O3
Henrotin 201070	Dijon, France	Case- crossover	Ischemic	O3 in recurrent stroke	O3 in recurrent stroke
Yamazaki 200771	Japan	Case- crossover	Ischemic	PM7 2h before intracerebral hemorrhage	PM7, NO2, O3 in 24h averages
Bedada 201272	UK	Case- crossover	Minor stroke	NO	CO, NO2, SO2, O3

^*Positive associations with confidence intervals not including the null.

^†Associations with confidence interval sincluding the null.

Abbreviatons: BC, Black carbon. BS, Black smoke. CI, Confidence intervals. CO, Carbon monoxide. H, hour.Max, Daily maximum.NO, Nitric oxide. NO₂, Nitrogen dioxide. NS, Non-significant associations but estimates not provided in publication. O₃, Ozone. PM₁₀, Particles with aerodynamic diameter ≤ 10µm. PM_2.5, Fine particles with aerodynamic diameter ≤ 2.5µm. PM_coarse, Coarse particles with aerodynamic diameter between 2.5 and 10 µm in aerodynamic diameter. PM_0.1, Ultrafine particles with less than 0.1 µm aerodynamic diameter. SO₂, Sulfur dioxide. SO₄, Sulfate. TSP, Total suspended particles.

Summary

The current evidence suggests exposure to higher levels of air pollutants related to combustion increases the risk of stroke. Studies of both long-term and short-term air pollution exposure suggest consistent evidence of increased risk of ischemic stroke and moderately consistent evidence supporting an association with hemorrhagic stroke. A few studies exploring susceptible subgroups have indicated stronger associations in individuals with several cardiovascular risk factors, diabetes, previous stroke and of older age. A recently published meta-analysis focusing on short-term air pollution exposure and stroke incidence or mortality reported significant associations for PM_2.5, PM₁₀, SO₂, CO, NO₂, and O₃ for stroke with stronger associations for ischemic stroke⁷³.

Because much of the existing literature is based on linkage of administrative data, an important limitation of many available studies is limited ability to classify and validate specific stroke outcomes. Ischemic stroke and hemorrhagic stroke and their subtypes have in the majority of studies been analyzed as a combined outcome despite the clear possibility that air pollution may affect underlying pathophysiologic pathways differently. Only some have separately analyzed ischemic stroke and hemorrhagic stroke and a handful have considered subtypes of ischemic stroke or hemorrhagic stroke. Similarly only a handful used thorough chart reviews and/or adjudicated the diagnosis and onset time of stroke. This highlights the need for high-quality validated diagnostic characterization of stroke outcome in studies of air pollution. In a study of short-term air pollution exposure and stroke specifically investigating the bias introduced through misclassification of time of event of stroke found that incorrect temporal classification caused up to 66% bias towards the null⁴². This may be especially relevant in mortality studies where the date of death from death certificates is used while not accounting for the time between stroke onset and death. In studies of long-term exposure to air pollution, the ability to investigate associations with stroke is dependent on the validity and resolution of the spatial exposure assessment and the adequate control for confounders related to both air pollution at place of residence and the risk of stroke, in particular socioeconomic factors.

There is growing evidence to suggest that both accumulated exposure to higher air pollution over a period of years and higher mean levels over a period of days increase the risk of stroke. In addition to improving temporal classification of exposure by validating stroke onset time, future research efforts should be directed to careful characterization of stroke subtype because air pollution may variably affect the different pathophysiological pathways. Air pollution exposure and increased risk of stroke may represent a considerable public health problem and regulations have improved air quality in many countries in Europe and the US, resulting in greater life expectancy⁷⁴. Yet associations with stroke have been reported at levels in compliance with current standards^{16, 68} highlighting the continued importance of effective regulation and monitoring in high income countries as well as extending efforts to address regulation in low and middle income countries where levels of air pollution and prevalence of stroke are on the rise.