An ecological study of PM2.5 and black carbon and acute hemorrhagic stroke: a long term effect study

Both PM2.5 and black carbon particle are associated with acute ischemic stroke. There is limited data on the correlation of long term effects of PM2.5 and black carbon on acute hemorrhagic stroke with a large sample size. This was an ecological study. Numbers of acute hemorrhagic stroke who were admitted and reimbursed from the national database of Thailand were collected. Data of PM2.5 and black carbon were collected. Poisson regression analysis adjusted by physical factors and regions was used to evaluate the association between PM2.5 and black carbon on acute hemorrhagic stroke. There were 82,389 patients diagnosed as acute hemorrhagic stroke throughout Thailand. The median of PM2.5 of all provinces was 29.19 microgram m−3, while black carbon had the median of 1.17 microgram m−3. PM 2.5 was significantly associated with numbers of acute hemorrhagic stroke with an adjusted coefficient of 0.019 (p < 0.001), while black carbon was also had significant adjusted coefficient of 0.199 (p < 0.001). Both PM2.5 and black carbon particle were associated with acute hemorrhagic stroke in a long term fashion by an ecological study of the national database.


Introduction
Acute hemorrhagic stroke is an emergency condition that requires prompt diagnosis and treatment.It accounted for 15%-20% of all strokes but has high morbidity and mortality [1,2].A mortality rate of acute hemorrhagic stroke may be high as 40% particularly in developing countries [3,4].A study from France reported that the 30-day mortality rate was decreasing from 40.9% in 1993 to 33.5% in 2002 [4].Additionally, 7.01% of patients with acute hemorrhagic stroke had stroke recurrence within five years based on Taiwan national database [5].Approximately one-third of patients with acute hemorrhagic stroke may have cognitive impairment at 90 days [6].
Even though hypertension, alcohol consumption, and psychosocial factors are the main risk factors for acute hemorrhagic stroke [1,7], environmental factors may be additional risk factors.Several studies reported that PM2.5 exposure significantly increases risk of acute ischemic stroke and its mortality by up to 0.34% and 1.4%, respectively [8][9][10][11].The risk for ischemic stroke was extremely high in patients with diabetes mellitus at 11% [12].There are limited studies of the association between PM2.5 and acute hemorrhagic stroke.
A study from Israel found that PM2.5 exposure was associated with an increased risk for ischemic stroke and TIA at 1.70% (p value = 0.030) and 2.74% (p value = 0.023).But, PM2.5 exposure decreased risk of hemorrhagic stroke by 4.43% (p value = 0.045) [13].In contrast, a study from Portugal reported that an increasing of PM2.5 by 10 microgram/l increased a risk of hemorrhagic stroke by 5.7% with 95% confidence interval of 1.020, 1.095 (p value = 0.002) [14].Both studies conducted on acute effects of PM2.5 on hemorrhagic stroke.Additionally, there are inconsistent data on the long term effects of PM2.5 on hemorrhagic stroke.Two previous studies found positive correlation between long term exposure of PM2.5 and hemorrhagic stroke with the highest odds ratio of 1.43 (95% confidence interval of 1.09, 1.88) [15,16], while the other three studies did not find the positive association [17][18][19].Note that three our of four studies had largest sample size of 3,540 patients with acute hemorrhagic stroke [16].Even though one study was a systematic review, it was published in 2019 and included only four studies.These data may indicate that long term effects of PM2.5 on acute hemorrhagic stroke are limited in literature.
Black carbon particle, a common particle from traffic and industry, also associated with acute ischemic stroke [20].An increasing of 1 μg m −3 of black carbon had a hazard ratio of 1.05 (95% confidence interval of 1.00, 1.10) for ischemic stroke.While, a study from the US did not find an association between black carbon on acute hemorrhagic stroke with a sample size of 577 patients [21].Similar to PM2.5, there is limited data of long term effects of black carbon on acute hemorrhagic stroke with limited sample size.
As there is limited data on long term effects and small sample size of PM2.5 and black carbon on acute hemorrhagic stroke, this study aimed to evaluate the effects of both PM2.5 and black carbon on acute hemorrhagic stroke in particular the long term effect with national database.

Materials and methods
This was an ecological study conducted in 77 provinces in Thailand.We collected data of numbers of acute hemorrhagic stroke who were admitted and reimbursed from the Universal Coverage Health Security Insurance Scheme Database, covering approximately 80% of Thailand's population.Diagnosis of acute hemorrhagic stroke was made and retrieved based on the codes of the Tenth Revision (ICD-10) International Classification of Disease; Cerebral Infarction (I61).Clinical data were also collected in terms of age (age group), sex, and comorbid diseases (hypertension, diabetes, dyslipidemia, atrial fibrillation, coronary artery disease, chronic kidney disease, previous stroke).These data were collected as number of patients with each factor by province between 1st Jan 2014 and 31st Dec 2016.The study protocol was approved by the institutional board review, Khon Kaen University, Thailand (HE621266).
Data of the PM2.5 were obtained from the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2) generated by NASA's Global Modeling and Assimilation Office (GMAO).NASA's updated MERRA-2 reanalysis data of the modern satellite measurements provide a set of data for air pollutants and using the GOCART model to calculate black carbon level [22].
MERRA-2 provides aerosol optical depth (AOD) to be calculated in the model comparing accuracy with ground monitoring station in Thailand by using an air quality data integration model.For optimal correlation, we used model V with fixed-effect model.We entered AOD into the model for estimating PM 2.5 using the loglinear model [23].Data of temperature and wind of each province were retrieved from the Meteorological Department.Regions in Thailand are divided into six parts as follows: central (9 provinces including Bangkok), north (17 provinces), northeast (20 provinces), east (9 provinces), west (8 provinces) and south (14 provinces).
The primary outcome of this study was numbers of patients with acute hemorrhagic stroke.The studied variables were shown in table 1 including main factors: PM 2.5 level, black particle level, physical factors, and regions; and additional factors: numbers of patients in various age groups, sex, and co-morbid diseases such as hypertension, diabetes, dyslipidemia, atrial fibrillation, coronary artery disease, chronic kidney disease, and previous stroke.Additionally, income and numbers of population were evaluated.
Statistical analyses.Descriptive statistics were used to summarize features of each factor.Results were presented as median (range) if data were not normally-distributed.We computed a prediction of numbers of patients with acute hemorrhagic stroke by PM 2.5 and black carbon particle by using a Poisson regression

Results
During the study period, there were 82,389 patients diagnosed as acute hemorrhagic stroke among 77 provinces of Thailand.Among studied variables, the most common age group for acute hemorrhagic stroke was between 46-70 years with a male: female ratio of 1.39:1 as shown in table 2. Hypertension was the most common comorbid disease with a median of 425 persons.The median of PM2.5 of all provinces was 29.19 microgram m −3 , while black carbon had the median of 1.17 microgram m −3 .The median temperature of all provinces was 27.06 °C with the median wind of 4.29 knot m −1 .PM 2.5 was significantly associated with numbers of acute hemorrhagic stroke with unadjusted and adjusted coefficients of 0.034 and 0.019, respectively (table 3 and figure 1), while black carbon was also had significant unadjusted and adjusted coefficient of 0.363 and 0.199, respectively (table 4 and figure 2).Temperature and wind positively associated with numbers of acute hemorrhagic stroke with adjusted coefficients of 0.079 in the PM2.5 model and 0.043 in the in the black carbon model (tables 3 and 4), while wind was also positively associated with numbers of acute hemorrhagic stroke with adjusted coefficients of 0.110 in the PM2.5 model and 0.165 in the black carbon model (tables 3 and 4).Regions had negative coefficients in both PM2.5 and black carbon model significantly (−0.058; p < 0.001).
As each region may have different PM2.5 and black carbon particle level, we did a subgroup analysis by regions of these two particles (table 5).The PM2.5 and was highest in the west region at 30.47 microgram m −3 , while black carbon particle was highest in the west region as well at 1.66 microgram m −3 .The PM2.5 was not significantly different among regions (p = 0.288), but the black carbon particle was significantly higher than the east (p = 029) and the south (p = 0.033) as shown in figures 3 and 4.

Discussion
This ecological study showed that both PM2.5 and black carbon constituent were associated with numbers of patients with acute hemorrhagic stroke significantly in a long term fashion based on a large sample size from the national database.Risk of acute hemorrhagic stroke by both acute and long term effects of PM2.5 are still controversial.Data of several studies conducted to evaluate acute effects of PM2.5 showed both positive and negative correlation [9,14,[24][25][26][27][28].This study supported the positive correlation of long term effects of PM2.5 and acute hemorrhagic stroke (table 3).The strength of this study was large sample size with 82,389 patients diagnosed as acute   hemorrhagic stroke by national database.Previous studies had the largest sample size of 3,540 patients with acute hemorrhagic stroke [16].The main mechanisms of PM2.5 related with acute hemorrhagic stroke in this study including microbleeding and high prevalence of hypertension.An exposure to PM2.5 for 12 months in hypertensive rats had significantly increased the incidence of cerebral microbleeds compared with those without PM2.5 exposures (37.5% versus 12.5%; p value < 0.001) [29].Hypertension, a known risk factor for acute hemorrhagic stroke, accounted up to 75% of patients with acute hemorrhagic stroke [30][31][32].In this study, hypertension was the most common condition with a three-year data collection (table 2).The results of this   study may imply that patients with hypertension who have a long term exposure of PM2.5 are at risk of hemorrhagic stroke over three-years period.
Several studies showed that black carbon is associated with hypertension, vascular damage, and kidney damage [33][34][35][36].Black carbon was associated with hypertension with an odds ratio of 1.07 (95% confidence interval of 1.04, 1.09), vascular injury with adjusted odds ratio of 1.081 (95% confidence interval of 1.056, 1.107), and lower creatinine clearance by 1.63 ml min −1 (95% confidence interval of −2.99, −0.28).Therefore, black carbon may be a potential factor for hemorrhagic stroke.Unlike this study, the study from the US did not find this association [21].This may be explained by the different study design.The US study evaluated the acute or same day effect of black carbon on hemorrhagic stroke, while this study had a pool data of three-year period.Additionally, the US study had the sample size of 577 patients with acute hemorrhagic stroke which was much smaller than this study.The small sample size may result in negative correlation of black carbon particle and acute hemorrhagic stroke.
There were other three significant factors associated with acute hemorrhagic stroke including high temperature, strong wind, and regions.A previous study from China found that high temperature in the past three days increased risk of hemorrhagic stroke admission by 42% (95% confidence interval 1.28, 1.58) [37].For strong wind speed, a study from Turkey found that wind speed associated with hemorrhagic stroke admission with odds ratio of 1.63 (95% confidence of 1.27, 2.09; p = 0.001) for lag 1.However, these associations are still inconsistent with other studies [38,39].Further studies are needed to confirm these associations.Geographic regions were another factor associated with acute hemorrhagic stroke (tables 3 and 4).We evaluated the level of PM2.5 and black carbon particle on six different regions of Thailand (table 5 and figures 3, 4).West region and northeast region were the top two regions with highest PM2.5 and black carbon level.Only black carbon level in the west region was significantly higher than the east and south regions.As PM2.5 sources are commonly from road dust and soil dust, the levels were not significantly different among regions in Thailand [40].For black carbon particle, west region has more industries with biomass burning causing significantly higher than the east and south where close to the ocean with strong wind [41].
There are some limitations.Specific personal factors such as hypertension, sleep apnea or diabetes were not included in the model as this study was an ecological study [42][43][44].Data of PM2.5 and black carbon were an average of each province per year.Finally, the predictive model indicated numbers of patients with acute hemorrhagic stroke, not the incidence.

Conclusions
Both PM2.5 and black carbon particle were associated with acute hemorrhagic stroke in a long term fashion by an ecological study of the national database.

Figure 2 .
Figure 2. Box plot showed mean black carbon particle levels and numbers of patients with intracerebral hemorrhage; coefficient of 0.199 (p < 0.001).

Figure 4 .
Figure 4. Black carbon particle levels categorized by regions of Thailand.Note.Star indicating significant different with the west region (p < 0.05).

Table 1 .
Studied variables for acute hemorrhagic stroke by an ecological study.
analysis.A multivariate Poisson regression analysis was also performed adjusted by temperature wind, and regions.Results were shown as coefficients, 95% confidence interval, and p values.Statistical analyses were performed by using STATA software (College Station, Texas, USA).

Table 2 .
Median and range of numbers of studied variables of 77 provinces in Thailand.

Table 3 .
A predictive model of numbers of patients with acute hemorrhagic stroke by PM2.5 using Poisson regression analysis.

Table 5 .
PM2.5 and black carbon particle levels categorized by regions of Thailand.