Академический Документы
Профессиональный Документы
Культура Документы
Contents
1. Introduction
2. Methods
3. Design and population
4. Data collection
5. Exposure variables
6. Blood pressure measurement and cut-points definitions
7. Covariate measurements and assessment
8. Sample size calculation
9. Statistical analysis
10. Results
11. Sociodemographic characteristics and hypertension
12. Hypertension and other risk factors
13. Air pollution exposure and hypertension
14. Stratified analysis
15. Multivariable analyses: global models
16. Multivariable analysis using individual pollution items
17. Discussion
18. Conclusion
19. Compliance with ethical standards
20. Conflict of interest
21. References
22. Citations
Full Text
Hypertension is a risk factor of several diseases, linked to high mortality and morbidity,
particularly in developing countries. Some studies have linked indoor and outdoor pollution
exposure items to hypertension, but results were inconsistent. Our objective was to assess
the association of living conditions related to air pollution to hypertension in Lebanon, a
Middle Eastern country. A national cross-sectional study was conducted all over Lebanon.
Blood pressure and its related medications were assessed to be able to classify participants as
hypertensive or not. Moreover, in addition to living conditions related to air pollution
exposure, we assessed potential predictors of hypertension, including sociodemographic
characteristics, self-reported health information and biological measurements. Furthermore,
we assessed dose-effect relationship of air pollution items in relation with hypertension.
Living conditions related to indoor and outdoor air pollution exposures were associated with
hypertension, with or without taking biological values into account. Moreover, we found a
dose-effect relationship of exposure with risk of disease (15% increase in risk of disease for
every additional pollution exposure item), after adjustment for sociodemographics and
biological characteristics (Ora = 1.15 [1.03-1.28]). Although additional studies would be
necessary to confirm these findings, interventions should start to sensitize the population
about the effect of air pollution on chronic diseases. The work on reducing pollution and
improving air quality should be implemented to decrease the disease burden on the
population and health system.
Introduction
In recent years, a series of epidemiological studies in developed countries have linked the
influence of environmental factors on the cardiovascular system, blood pressure in particular
(Brook et al. [6] ; Giorgini et al. [16] ).
For example, China being considered one of the most polluted countries in the world, has
conducted several studies regarding this issue. In 33 Chinese communities, long-term
exposure to ambient air pollution was more strongly associated with prehypertension than
with hypertension, especially among females and the elderly (Yang et al. [51] ). In another
Chinese study, long-term exposure to ambient PM2.5 was considered to be an important risk
factor of hypertension among elderly (Lin et al. [25] ). Similar results were obtained in a
national cross-sectional study in China (Liu et al. [26] ), and in other developing countries all
over the world (Ezejimofor et al. [9] ; Lee et al. [23] ; Ribeiro et al. [38] ; Stankovic and
Nikolic [43] ).
In the USA, residential proximity to major highways showed positive association with
hypertension among post-menopausal women (Kingsley et al. [20] ). In Canada, similar
results were found for emergency visits due to hypertension (Brook and Kousha [5] ).
Methods
The methodological details of the study presented in this paper have been previously
described elsewhere (Farah et al. [10] , [11] ; Zeidan et al. [52] ). The data for this study were
collected through a cross-sectional study, carried out between September 2013 and October
2014, using a multistage sample across Lebanon. Out of 2789 circumscriptions (villages or
communities, considered as clusters), 100 were randomly selected from the list in Lebanon
provided by the Lebanese Central Administration of Statistics, using an automatic random
number generator. From the list of dwellers provided by the local authorities in each cluster,
the selection of residents aged ≥ 40 years was done using a software program to ensure
randomness. Data were gathered from a face-to-face interview from participants who
provided an oral and written consent. The Institutional Review Board (IRB) of the Lebanese
University waived the need for ethical approval since this is an observational study with no
traceability of subjects. Individuals with learning disabilities or psychiatric disorders were
excluded.
Data collection
Exposure variables
The major exposure variable was self-reported exposure to air pollution. In fact, self-reported
measures of air pollution or noise are considered similar the objective measures (Pitchika et
al. [36] ). In our case, it was composed of several items: living close to (< 100 m) a busy road
(defined as “packed with cars and trucks”) was used as a proxy measure of residential
exposure to traffic pollution (Kulick et al. [21] ; Weaver et al. [49] ). Similarly, other
questions were asked about living close to (< 100 m) an electric generator, and living or
working near a power plant. Questions about driving a car and having an air conditioner in
the car were also asked because these living conditions were considered protective against air
pollution exposure versus being a pedestrian and using public transportations (Vijayan et al.
[46] ). Furthermore, indoor air pollution was assessed by means of the heating of the house
(central, electric, air conditioner, butane gas, or wood) and cooking (butane, electricity or
wood); using wood in heating home and cooking was considered problematic because of
biomass burning. Passive smoking was assessed by the number of smokers at home who
smoked indoor, and exposure to passive smoking at work. Active smoking was considered as
a behavioral variable that directly affected the smoker himself; it was considered as a
potential confounding.
Finally, to summarize pollution items, a “pollution exposure index” was created by summing
up the abovementioned variables, taking into account the direction of the association with
hypertension: dichotomous variables were used and added. The higher the index, the higher
the number of pollution sources the participant is exposed to. A similar method using an
index for indoor and outdoor environmental risk factors was used by Moussa and
collaborators (Moussa et al. [30] ). Several reasons could account for this method: it is not
clear which components of air pollution are the most harmful, nor which time window of
exposure is most relevant; association between source specific particulate matter air pollution
and outcomes are sometimes mixed and generally weak (Stockfelt et al. [44] ). In addition,
the biological effects of particulate matter may vary in function of both the aerodynamic
diameter and the chemical composition. Other gaseous pollutants may also play an
independent role in disease genesis, suggesting the need to develop multi-pollutant preventive
approaches (Martinelli et al. [29] ). This method would allow us to assess a potential dose-
effect relationship. Moreover, the pollution index was dichotomized according to the median,
classifying individuals with low exposure to pollution (index ≤ 0) and high exposure to
pollution (index > 0).
Trained medical students performed the measurements. For blood pressure, we measured
SBP and DBP twice following a standardized protocol using an electronic automatic
validated device (Omron M6 Comfort; Omron, Kyoto, Japan) (O’Brien et al. [33] ): Blood
pressure was measured after the subject rested for 5 min, sitting with back support, legs
uncrossed, and the arm supported at heart level. The mean of 2 measurements was recorded.
All cut-points for uncontrolled blood pressure (BP), systolic blood pressure (SBP), and
diastolic blood pressure (DBP) were based on the Eighth Joint National Committee (JNC 8)
(James et al. [19] ) and the European Society of Hypertension (ESH) (Mancia et al. [28] ).
Patients with mean SBP ≥ 140 mmHg were considered to have high SBP and patients with
mean DBP ≥ 90 mmHg were considered to have high DBP. Hypertension was defined as
SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg; thus, participants were considered as
hypertensive in case they had measured hypertension or were under antihypertensive
medications (i.e., previously diagnosed with hypertension, even if their blood pressure
measurement was normal at the time of the study).
Anthropometric measurements included weight (kg), height (m), and waist circumference.
Body mass index (BMI) was calculated by dividing weight in kilograms by height in square
meters. BMI was categorized into three classes: normal weight (BMI < 25 kg/m2), overweight
(25 kg/m2 ≤ BMI < 30 kg/m2), and obese (BMI ≥ 30 kg/m2). For waist circumference, the
World Health Organization recommendations for measurement and cutoff section were
adopted (World Health Organization. Waist circumference and Waist-hip ratio: report of a
WHO expert consultation. Geneva 2008. Available on
http://apps.who.int/iris/bitstream/10665/44583/1/9789241501491%5feng.pdf).
Individuals underwent random capillary blood glucose (RCBG) (Qiao et al. [37] ), using
Accu-Check Performa (Roche Diagnostics GmbH, Mannheim, Germany). Diabetes was
defined as RCBG > 200 mg/dL or self-reported medication use for glucose control (Howard
et al. [17] ). We defined current smokers as individuals who smoked tobacco in the previous
12 months, and we included those who had quit within the past year.
Dietary habits were assessed using the Lebanese Mediterranean Diet Score, a dietary intake
questionnaire for the quantitative estimation of adherence to a Mediterranean diet (MD)
adapted to the Lebanese context, ranging from 0 to 52 (maximal adherence) (Issa et al. [18] ).
Psychological distress was also assessed by the Beirut Distress Scale (BDS-22), a scale
developed and validated in Lebanon. Possible scores range from 0 to 66 (maximum
psychological distress) (Barbour et al. [3] ).
Sample size was initially calculated to measure the prevalence of cardiovascular RFs. We
used Epi Info™ (Center for Disease Control, Atlanta, GA, USA. Available from
http://wwwn.cdc.gov/epiinfo). Since no other figure was available, we used the prevalence of
hypertension (43.1% among individuals aged 50 years) as the reference (Sibai et al. [42] );
after taking into account, a worst acceptable result of ± 4% difference with the
aforementioned prevalence and a 95% confidence interval (CI) as well as the two-stage
sampling design, a minimal sample size of 1200 was required in a population over 40 years
of age.
Statistical analysis
All analyses were done using the Statistical Package for Social Sciences (SPSS), version 23·0
(IBM Corporation, Armonk, NY, USA). Cluster effect on sampling (that could lead to a false
narrowing of standard deviations) was taken into account, according to the method described
by Rumeau-Rouquette et al. (Rumeau-Rouquette et al. [39] ), and we used the SPSS features
that take into account this sampling design.
A p value < 0.05 was considered statistically significant. For normally distributed data,
description was done using the mean (M) and standard deviation (SD), while Student’s T test
and ANOVA test were used to compare means between two groups and more than two
groups, respectively. As for categorical variables, we used percentages with 95% CIs and the
Pearson’s chi-square test. P trends were calculated to measure the prevalence trends across
the different age groups and across pollution exposure index levels. A bivariate analysis
measuring the association of the different sociodemographic characteristics, and behavioral
and biological risk factors with the presence or absence of hypertension was done.
A stratified analysis was performed for the association between dichotomized pollution index
of exposure and hypertension, between subgroups of sociodemographic characteristics.
Homogeneity of OR among strata was assessed using Breslow-Day test; in addition, Mantel-
Haenszel test was performed to assess the statistical significance of adjusted OR. Moreover,
we assessed the interaction between the pollution index as a continuous variable and strata
variables in specific logistic regressions.
We then performed multivariate analyses using logistic regression models only including
selected variables, all of which had a p value < 0.2 in the bivariate analysis. The adjustment
of all associations found was based on the potential confounding variables: sociodemographic
variables (age, sex, marital status, working status, education, socioeconomic level) and other
potential cardiovascular risk factors (active smoking, Mediterranean diet adherence, alcohol
consumption, physical activity, diabetes, hyperlipidemia, and psychological distress). For all
multivariable models, we used the Hosmer-Lemeshow and the Pearson’s chi-square model
goodness of fit tests to ensure model adequacy.
Results
Overall, 1516 participants were interviewed; among them, the prevalence of hypertension
was found to be 47.7%. There was a higher prevalence among men versus women (50.6%
versus 45.1%; p = 0.031). Hypertension prevalence was significantly higher with older age
(p < 0.001), lowest education level (p < 0.001), widowed/divorced and married individuals
versus single ones (p < 0.001), extreme socioeconomic status individuals (p = 0.009) and
among those retired versus others (p < 0.001). Hypertension did not significantly differ
among participants living in different dwelling regions (Table 1).
In bivariate analysis, no significant associations were found for family history of stroke or of
cardiovascular diseases; however, significant higher prevalence of hypertension was found
among individuals with diabetes mellitus (74.3 versus 40.8%), hypercholesterolemia (65.1
versus 38.7%), hypertriglyceridemia (66.4 versus 42.1%), diagnosed depression (58.0 versus
46.4%), previous smokers (55.6 versus 44-49%), those with no regular physical activity (50.3
versus 41.3%). An inverse association was found for passive smoking (44.4 versus 50.9%).
Individuals with hypertension had higher psychological distress (12.36 versus 10.83). No
significant association was found for alcohol consumption (Table 2).
Air pollution exposure and hypertension
Significant inverse associations were found for some pollution exposure items with
hypertension (Table 3): participants heating home electrically (43.4 versus 50.1%) and by
central heating (28.3 versus 49.2%) had lower prevalence of hypertension. Moreover, having
an air conditioning in the car (40.3%) or driving a car (39%) were associated with lower
hypertension. An inverse association was found for living or working near a power plant
(36.0 versus 48.4%). The association if hypertension with the rest of pollution items did not
reach statistical significance. Moreover, the majority of individuals declared cooking on
butane gas (95.1%), while very few cooked on electricity (4.4%) (bivariate results not
reported).
However, when adding pollution exposure items, there was a significant higher mean
exposure to pollution items among people with hypertension. The pollution index spanned
from − 3 (maximum protective factors exposure) to + 3 (maximum harmful factors exposure);
its mean was 0.45 (standard deviation = 1.40). The median was equal to 0. As for the
association with hypertension, the trend test was significant for a higher prevalence of
hypertension per pollution higher exposure index (p < 0.001; Fig. 1).Hypertension
prevalence per pollution exposure index level. The pollution index represents the sum of
pollution items to which a person is exposed: passive smoking + living close to a generator +
living close to a busy road + living close to a power plant + heating home with wood +
cooking on wood − heating home electrically − heating home centrally − driving a car −
having air conditioning in the car. p value for trend test < 0.001
Stratified analysis
In model 1 (Table 5), hypertension was found to be associated with several pollution items,
after adjustment over sociodemographic and behavioral characteristics of the population.
Heating home by wood (ORa = 1.88 [1.29-2.76]), by butane gas (ORa = 1.62 [1.14-2.30]),
and ever living close to a road full of cars/trucks (ORa = 1.78 [1.16-2.72]) and to a generator
(ORa = 1.53 [1.07-2.19]) were associated with hypertension, while heating home centrally
(ORa = 0.44 [0.24-0.79]) was inversely associated with hypertension.
After adding biological factors in model 2 (Table 5), heating home by wood (ORa = 1.59
[1.07-2.35]) and living close to a generator (ORa = 1.62 [1.14-2.31]) were still found to be
associated with hypertension, while heating home centrally was inversely associated with it
(ORa = 0.38 [0.20-0.72]). Ever living close to a road full of cars/trucks or to a power plant
gave positive associations with hypertension but of no statistical significance (p < 0.1), while
passive smoking gave an inverse and borderline result (p = 0.05).
Finally, in model 3 (Table 5), when pollution exposure items were replaced by the pollution
exposure index, a higher pollution exposure was associated with an increase in the odds of
hypertension (adjusted OR = 1.15 [1.03-1.28]): in other words, every point increment in the
pollution exposure index corresponds to an increase of hypertension odds by 15%. In
addition, to take into account the interaction effect obtained in Table 4, we performed the
same multivariate analysis of pollution index for females and males’ subgroups separately:
we obtained ORa = 1.22 [1.02;1.44] (p = 0.025) for females, and ORa = 1.08 [0.92;1.27] (p =
329) for males (Hosmer-Lemeshow test was 0.004).
Almost similar results were found when introducing the pollution items one by one in the
models 1 and 2 (Table 6). In model 1 with adjustment over sociodemographic and behavioral
variables, the following items showed a significant association with hypertension: living
close to a busy road with cars and trucks (ORa = 1.35 [1.01;1.82]), living close to a generator
(ORa = 1.40 [1.04;1.89]), and heating home using wood (ORa = 1.66 [1.19;2.31]), while
heating home centrally (ORa = 0.31 [0.18;0.55]) and having an air conditioning in the car
(ORa = 0.69 [0.47;0.99]) were inversely associated with hypertension. In model 2, when
adding biological variables, sensibly similar results were obtained: living close to a generator
(ORa = 1.51 [1.08;2.13]) and heating home using wood (ORa = 1.45 [1.00;2.10]), while
heating home centrally (ORa = 0.34 [0.18;0.65]) was inversely associated with hypertension.
The items “living close to a busy road with cars and trucks” and “having an air conditioning
in the care” lost their statistical significance, although association remained unchanged
compared to model 1 (Table 6).
Discussion
In this study, we were able to show that declared pollution exposures were associated with
hypertension, with or without taking biological values into account. Moreover, we found a
dose-effect relationship of exposure with risk of disease, after adjustment for
sociodemographics and biological characteristics. Our results were particularly significant for
females (positive interaction of gender with pollution exposure index); for males, it was not
possible to conclude due to an inadequacy of the model to the data.
Living close to a busy road with cars and trucks (ORa = 1.35-1.78), close to a fuel
functioning power plant (ORa = 1.87), or to a power generator (ORa = 1.40-1.62) were
positively associated with prevalent hypertension. Similar ORs were obtained in previous
studies: ORa = 1.41-1.5 for residential proximity with high traffic (Fuks et al. [13] , [15] ),
while exposure to black carbon, a tracer of vehicular-traffic pollution, is known to be
associated to hypertension (Zhong et al. [54] ). For industry related pollution, a study
reported the impossibility to separate particulate matters related to industry from the global
pollution exposure; the association was however similar to the ones we found (K. B. Fuks et
al. [15] ). We note that although no studies have ever studied proximity to generators all over
the world because the use of local diesel power generators is a particularity of Lebanon; we
can however extrapolate from other diesel generator-related problems on blood vessels
(Wauters et al. [47] ).
Our results are also in line with those of meta-analyses regarding the effect of air pollution
components on hypertension, using objective measures of pollution exposure (Cai et al.
[8] ). Another review similarly shows that particulate pollutants cause significant increases in
BP parameters in relation to both short and long-term exposures, with robust evidence for
exposures to PM2.5 (Giorgini et al. [16] ). Recent epidemiological studies suggested a
positive association between residence within regions with higher levels of ambient PM and
an increased incidence and prevalence of overt hypertension. This is particularly true for
vulnerable subsets of individuals (pregnant women, elderly and high cardiovascular risk
individuals) (Giorgini et al. [16] ).
For passive smoking, we found an inverse and borderline relationship with hypertension.
Although meta-analyses indicate a clear positive association between passive smoking and
stroke and ischemic heart disease (Fischer and Kraemer [12] ), particularly in developing
countries (Olasky et al. [34] ), very few studies focused on passive smoking and
hypertension in non-pregnant adults (Li et al. [24] ; Seki et al. [41] ), with inconsistent
results between systolic and diastolic blood pressure when cotinine measurements were taken
(Alshaarawy et al. [1] ) and different effects between males and females (Mahmud and Feely
[27] ). This point remains to be studied in further dedicated research.
For other indoor pollution items, our results on cooking and heating are similar to those found
in some developing countries: biomass fuel use was associated with a higher likelihood of
having hypertension and higher blood pressure in Peru (Peña et al. [35] ). Cook stoves in
rural China have been linked to hypertension through black carbon related to incomplete
combustion of biomass fuel, in addition to vehicles emissions exposure and proximity to
highways (Baumgartner et al. [4] ). In a review on this matter, biomass fuel was clearly
associated with hypertension, especially among women in low to middle income countries
(Burroughs Pena and Rollins [7] ). The inverse association we found with central heating
could be the reflection of a higher socioeconomic status and/or the fact that central heating
does not involve biomass or other fuel burning. In any case, our results go in line with the
World Health Organization (WHO) recommendations to use air cleaners and filters to
improve indoor air quality and decrease chronic diseases (Vijayan et al. [46] ).
As for the main affected subgroups, we found that females, lower age categories, those of
secondary education, of lowest socioeconomic status, those who are jobless, and those
divorced/widowed, are the ones mostly showing the association between pollution and
hypertension. These results are similar to those found by other researchers (Giorgini et al.
[16] ) and show us the most vulnerable populations to pollution effect on hypertension in
Lebanon. Although we found a significant interaction between female gender and pollution
exposure (females being more significantly affected than males), more in-depth studies are
suggested to further identify these subgroups and study interactions of pollutants composition
on sociodemographic and behavioral factors.
Particulate matter is mainly formed by fossil fuel combustion and is the main components of
exhaust emissions from motor vehicles. A review of several studies summarized the
pathways by which acute and chronic exposures to air pollutants might disrupt hemodynamic
balance favoring vasoconstriction, such as autonomic imbalance and augmented release of
various pro-oxidative, inflammatory and/or hemodynamically active mediators (Giorgini et
al. [16] ; Nogueira [32] ), showing the biological plausibility of the associations we found.
Moreover, ultrafine particles might also have direct actions on the heart and vasculature of
ultrafine particles translocated into the systemic circulation. The induction of oxidative stress
by these particles may be central to all putative pathways that trigger hypertension. In
chronic exposures, these alterations favor the development and progression of atherosclerosis
and possibly of hypertension in the long term (Nogueira [32] ).
Our study presents several limitations. As for all cross-sectional studies, there is a difficulty
in demonstrating temporality. The sample size we adopted is adequate to demonstrate
bivariate associations; however, it could be lower than necessary in some multivariable
analyses. There is also a possibility of selection bias during the sampling; although random
selection was adopted, a selection bias cannot be ruled out, due to the refusal of some
individuals to answer some questions, generating missing values in some analyses. Moreover,
although we performed a multivariable analysis to remove confounding effect, there is still a
possibility of residual confounding. There is also a possibility of over- (or under-) reporting
of exposures to air pollutants, since the study depended fully on subjective self-reporting.
However, this information bias is probably non-differential because it could equally apply to
people with or without the disease.
Finally, although the exposure index we used was useful, this method oversimplifies the
complexity of multiple toxics exposures, and it is only a way to roughly assess their
cumulative effect on the disease (Sarigiannis and Hansen [40] ). Its major drawback is that it
does not take varying chemical composition into account, since chemical mixtures could have
different effects on the human organism. Generating and validating scales for assessing
multiple exposures to pollution items is suggested in future work, particularly when
measurements are difficult to perform (such as in developing countries with limited
resources). This will constitute a useful tool to screen people at risk for hypertension or other
chronic diseases. Additional prospective studies taking into account these weak points would
be necessary to confirm our results, and to be able to compare our results to those of other
researchers as well.
Conclusion
Conflict of interest
References
Citations
1 Alshaarawy O, Xiao J, Shankar A, Association of serum cotinine levels and hypertension
in never smokers, Hypertension, 2013, 61, 2, 304, 308,
10.1161/HYPERTENSIONAHA.112.198218
~~~~~~~~
By Pascale Salameh; Mirna Chahine; Souheil Hallit; Rita Farah; Rouba Karen Zeidan;
Roland Asmar and Hassan Hosseiny
Environmental Science & Pollution Research is a copyright of Springer, 2018. All Rights Reserved.