Does exposure to aircraft noise increase the mortality from cardiovascular disease

QUESTION

Does exposure to aircraft noise increase the mortalityfrom cardiovascular disease in the population livingin the vicinity of airports? Results of an ecologicalstudy in FranceAnne-Sophie Evrard, Liacine Bouaoun, Patricia Champelovier1, Jacques Lambert1,Bernard Laumon2Epidemiological Research and Surveillance Unit in Transport, Occupation and Environment (UMRESTTE), Université de Lyon, Lyon,Transport, Health and Safety Department of the French Institute of Science and Technology for Transport, Development and Networks(IFSTTAR), Bron, Université Lyon 1, UMRESTTE, Lyon, 1IFSTTAR, Planning, Mobilities and Environment Department, Transport andEnvironment Laboratory (LTE), Bron, 2IFSTTAR, Transport, Health and Safety Department, Bron, FranceAbstractThe impact of aircraft noise on health is of growing concern. We investigated the relationship between this exposureand mortality from cardiovascular disease, coronary heart disease, myocardial infarction, and stroke. We performedan ecological study on 161 communes (commune being the smallest administrative unit in France) close to the followingthree major French airports: Paris-Charles de Gaulle, Lyon Saint-Exupéry, and Toulouse-Blagnac. The mortalitydata were provided by the French Center on Medical Causes of Death for the period 2007-2010. Based on the dataprovided by the French Civil Aviation Authority, a weighted average exposure to aircraft noise (LdenAEI) was computedat the commune level. A Poisson regression model with commune-specific random intercepts, adjusted for potentialconfounding factors including air pollution, was used to investigate the association between mortality rates and LdenAEI.Positive associations were observed between LdenAEI and mortality from cardiovascular disease [adjusted mortalityrate ratio (MRR) per 10 dB(A) increase in LdenAEI = 1.18; 95% confidence interval (CI): 1.11-1.25], coronary heartdisease [MRR = 1.24 (1.12-1.36)], and myocardial infarction [MRR = 1.28 (1.11-1.46]. Stroke mortality was more weaklyassociated with LdenAEI [MRR = 1.08 (0.97-1.21]. These significant associations were not attenuated after the adjustmentfor air pollution. The present ecological study supports the hypothesis of an association between aircraft noise exposureand mortality from cardiovascular disease, coronary heart disease, and myocardial infarction. However, the potentialfor ecological bias and the possibility that this association could be due to residual confounding cannot be excluded.Keywords: Aircraft noise, environment, health, mortalityIntroductionThe impact of exposure to aircraft noise on health is ofgrowing concern[1] because of a steady rise in flights; people’sannoyance with this kind of noise also seems to be rising.[2]While many studies address the annoyance associated withaircraft noise[3-5] or report the adverse effects on the qualityAccess this article onlineQuick Response Code:Website:www.noiseandhealth.orgDOI:10.4103/1463-1741.165058PubMed ID:***© 2015 Noise & Health | Published by Wolters Kluwer – Medknowof sleep,[6-8] a fewer number of studies consider other healtheffects of this noise exposure such as cardiovascular disease.Noise is a psychosocial stressor that activates the sympatheticand endocrine system. According to the general stressmodel,[9] neuroendocrine arousal is associated with adverseThis is an open access article distributed under the terms of the CreativeCommons Attribution-NonCommercial-ShareAlike 3.0 License, whichallows others to remix, tweak, and build upon the work non-commercially,as long as the author is credited and the new creations are licensed underthe identical terms.For reprints contact: reprints@medknow.comHow to cite this article: Evrard AS, Bouaoun L, Champelovier P, LambertJ, Laumon B. Does exposure to aircraft noise increase the mortality fromcardiovascular disease in the population living in the vicinity of airports?Results of an ecological study in France. Noise Health 2015;17:328-36.328Evrard, et al.: Aircraft noise and cardiovascular mortality near airportsmetabolic outcomes that are well-known and established riskfactors for cardiovascular disease. Therefore, aircraft noiseexposure could increase the prevalence or incidence of thesediseases, ultimately increasing the risk of premature death.Several studies have shown an association between aircraftnoise exposure and hypertension.[10-13] A multi-airport studyin the United States found that high exposure to aircraftnoise was significantly associated with hospitalization forcardiovascular disease among people above 65 years of ageliving near airports.[14] The evidence of an association ofaircraft noise with mortality is currently limited. In 2010,Huss et al.[15] reported an association between aircraft noiseand mortality from myocardial infarction in Switzerland witha dose-response relationship for the level and duration of noiseexposure but no association with stroke or cardiovascularmortality. A Danish study did not find any association ofaircraft noise with stroke mortality[16] and even a Canadianstudy did not find any association of aircraft noise withcoronary heart disease mortality.[17] More recently, a smallarea study near Heathrow airport in London, England showeda significant association between high levels of aircraft noiseand hospital admissions and mortality for stroke, coronaryheart disease, and cardiovascular disease.[18]We carried out an ecological study addressing the issue ofan association between weighted average exposure to aircraftnoise and mortality for some specific causes of interest suchas cardiovascular disease, coronary heart disease, myocardialinfarction, and stroke. Since air pollution has been foundto be associated with cardiovascular disease,[19-21] concernsfor disentangling the effects of noise and of air pollution oncardiovascular outcomes have been raised.[1] A secondaryaim of the present study was, therefore, to examine ifthe association between aircraft noise and mortality wasconfounded by air pollution.MethodsSpatial scaleWe carried out an ecological study based on 161 communes(a commune is the smallest administrative unit in France)spread over three geographical areas located in the vicinity ofthe following three major French airports [Figure 1]: ParisCharles de Gaulle (108 communes), Lyon Saint-Exupéry(31 communes), and Toulouse-Blagnac (22 communes);these areas are hereafter, referred to as the areas of Paris,Lyon, and Toulouse, respectively. In 2011, Paris-Charles deGaulle airport counted about 61 million passengers, LyonSaint-Exupéry airport about 8.5 million passengers, andToulouse-Blagnac airport about 7 million passengers.[22] Thestudy population corresponded to the population of these 161communes living in the vicinity of these three airports andthis was estimated to be 1.9 million inhabitants in 2009 (thatis, 3% of the total population of mainland France).329Figure 1: The three airports included in the present studyAircraft noise exposure and mortality data used in the presentstudy were obtained at the commune level.Aircraft noise exposure assessmentThe estimated exposure to aircraft noise was assessedby the French Civil Aviation Authority, which producesoutdoor noise exposure maps with the “Integrated NoiseModel”[23] (INM) for France’s largest airports. The INM isan internationally well-established computer model thatevaluates aircraft noise impact in the vicinity of airports.The INM outputs noise contours for an area. Aircraft noisecontours were available for the year 2008 for Paris-Charlesde Gaulle airport, 2003 for Lyon-Saint-Exupéry airport, and2004 for Toulouse-Blagnac airport. For Lyon-Saint-Exupéryand Toulouse-Blagnac airports, we used the most recentnoise exposure data available. These aircraft noise contourswere considered to be representative of the years precedingthe mortality assessment. The study area comprises allthe communes exposed to aircraft noise, defined as beingincluded in these noise contours and also the communessharing a common border with them. The inclusion of theseneighboring communes allowed us to increase the contrast inaircraft noise exposure.The noise indicator used in the present study is the dayevening-night equivalent level (Lden) in A-weighted decibels[dB(A)]. It is defined as a weighted average of soundpressure levels from daytime (6 AM-18 PM), evening (18PM-22 PM), and night (22 PM-6 AM). It is determined overthe year at the most exposed façade. In this calculation,evening and night sound pressure levels receive a penaltyof 5 dB(A) and 10 dB(A), respectively, to reflect people’ssensitivity to noise.[24] Noise levels were estimated with a1-dB(A) resolution from a minimum of 50 dB(A) for theNoise & Health, September-October 2015, Volume 17Evrard, et al.: Aircraft noise and cardiovascular mortality near airportsParis area, and 45 dB(A) for both the Lyon and Toulouseareas. For each commune in the study area, the number ofinhabitants of the commune living within these noise levelsthat were at 1-dB (A) intervals was available based on the2009 French census.The following four underlying causes of death wereinvestigated in the present study: 1) cardiovascular disease(I00-I52), 2) coronary heart disease (I20-I25), 3) myocardialinfarction (I21-I22), and 4) stroke (I60-I64, excluding I63.6).Noise levels were aggregated to obtain an estimate ofcommune-level exposure to aircraft noise. A populationweighted average called average energetic index(LdenAEI)[25] was estimated by weighting for a givencommune, the Lden by the number of inhabitants livingwithin this noise level. For a given commune i, LdenAEIwas defined as follows:Correlations between aircraft noise and air pollutionexposure were assessed using Spearman’s rank correlationcoefficients.where, Lden,j is the noise level j (the difference between Lden,jand Lden,j+1 is 1 dB(A)), Pj the number of inhabitants in thecommune exposed to Lden,j dB(A), andStatistical analysisThe effects of aircraft noise on mortality rates were firstexamined with Poisson generalized additive mixed models(GAMMs)[26,27] including a smooth cubic spline functionin order to account for a potential nonlinear effect. As thesmoothed fit does not deviate from the linear fit for LdenAEI,associations with the continuous exposure variable werethen estimated and presented in the present paper. Wefitted Poisson GLMMs,[28] including a commune-specificrandom effect term, to account for overdispersion. Foreach outcome, the GLMMs model could be written asfollows:is the totalnumber of inhabitants in the commune.LdenAEI was then used as a measure of aircraft noise exposurein the statistical analyses presented in the present paper.Air pollution exposure assessmentInformation on exposure to air pollution, including bothnitrogen dioxide (NO2) and particulate matter of 10 μm orless (PM10), was available at the commune level but onlyfor the communes of the Paris and Lyon areas. Dispersionmodeling was used to estimate annual background airpollution concentration for each commune in the study area.Briefly, modeled concentrations were provided at a 50 m × 50m resolution by Airparif for the Paris area, and at a resolutionof 1,000 m × 1,000 m by Air Rhône-Alpes for the Lyon area.For both areas, modeled concentrations were validated bycomparison with concentrations measured by a monitoringstation network. The average air pollution exposure (for bothNO2 and PM10 indicators and expressed in µg/m3) for theyears 2008-2010 was used in the statistical analyses. It wasthen categorized into three categories corresponding to thetertiles of the distribution.Mortality dataThe mortality data were provided by the French Centeron Medical Causes of Death (CépiDc-Inserm) for theperiod 2007-2010. The tenth revision of the InternationalClassification of Diseases (ICD-10) was used to code andclassify the mortality data based on the death records. Thecommune of residence, which is systematically included inthe death record, was used as the spatial location.Noise & Health, September-October 2015, Volume 17where, i refers to the commune, Yi denotes the number ofdeaths observed in the commune i, Popi the populationnumber in the commune i (considered as an offset), Xi a vectorof explanatory covariates for adjustment, and ui representsthe corresponding random effect. βt denotes the regressioncoefficients corresponding to these covariates. As usual, thenonspatial random effect ui, also called heterogeneity, wasassumed to be normally distributed with a zero mean and aconstant variance.Data on potential confoundersThe models were adjusted for the following covariates, atthe commune level, considered to be a priori confoundingfactors: Gender, age, log-population density, lung cancermortality, and a deprivation index. The log-populationdensity was introduced instead of the population density inorder to take into account the fact that the density was greatlydifferent between the different communes.Lung cancer mortality (ICD-10 code: C34) was used atthe commune level as a proxy measure for commune-levelsmoking because data on individual smoking or smokingprevalence at the commune level were not available in France.As using the Townsend deprivation index[29] in France maybe not suitable for different reasons,[30,31] we preferred tointroduce the deprivation index proposed by Rey et al.[32]It was constructed at the commune level based on thefollowing four variables, each representing a dimension of330Evrard, et al.: Aircraft noise and cardiovascular mortality near airportsthe socioeconomic level:(1) The median household income,(2) The percentage of high school graduates in the populationaged 15 years and above,(3) The percentage of blue-collar workers in the activepopulation, and(4) The unemployment rate.These socioeconomic data were provided by the FrenchNational Institute for Statistics and Economic Studies(INSEE).Our deprivation index was defined as the first component ofa principal component analysis (PCA) of the four variables.This index accounted for 67% of the total variation ofthe model and was strongly correlated with each of theinitial variables (positively with the unemployment rateand the percentage of blue-collar workers and negativelywith income and the percentage of high school graduates).Positive values of the deprivation index correspond to thedeprived communes.Adjusted mortality rate ratios (MRRs) with their 95%confidence intervals (CIs) were computed for each covariateincluded in the models by taking the exponential of thecorresponding regression coefficient.Additional analyses were also carried out to examine theimpact of air pollution on the relationship between aircraftnoise exposure and mortality by adjusting the models on airpollution (NO2 and PM10 concentrations).Sensitivity analysesThe models were stratified on gender to test whether thepotential associations between aircraft noise exposure andmortality from the causes of interest remained similar forboth men and women.The Townsend deprivation index was introduced in the modelsinstead of the deprivation index obtained with the PCA.As aircraft noise levels were assumed to be much higher inthe Paris area compared to the other areas due to the largersize of the airport, the effect of the additional adjustment forthe study area was explored and a sensitivity analysis usingthe Paris data only was carried out.The version 10.1 of the ArcGIS software (Redlands,California)[33] was used to produce the maps. All datamanagement was carried out using SAS software (Cary NC,USA)[34] version 9.3 and statistical analyses were carried outusing R statistical software (Vienna, Austria)[35] version 3.0.2with the gam function of the mgcv package.[36]ResultsOverall, the average LdenAEI was estimated as 49.6 dB(A)[range: 42.0-64.1 dB(A)], as shown in Table 1. Half of thecommunes of the study area had an LdenAEI lower than 48.9dB(A). The highest average of LdenAEI was observed in theParis area [51.6 dB(A) compared to 45.3 dB(A) for the Lyonarea and to 45.7 dB(A) for the Toulouse area]. The communewith the highest LdenAEI [64.1 dB(A)] was located in the Parisarea. Moreover, LdenAEI varied more widely in the Paris area.The NO2 concentration was higher in the Paris area (mean:24.0 µg.m-3) than in the Lyon area (mean: 16.5 µg.m-3) andvaried more widely in the Paris area. The PM10 concentrationswere very similar in both the areas.Figure 2 shows the distribution of LdenAEI in the communesincluded in the study area according to the quartiles ofLdenAEI. A fairly increasing pattern of LdenAEI was observedon both the west and east sides of the Paris-Charles de Gaulleairport, whereas no specific geographical pattern of LdenAEIwas found either for the Lyon area or the Toulouse area.Adjusted MRRs derived from the models are presented inTable 2. Increased MRRs were observed with increasingage for mortality from all specific causes of interest. Thepopulation density was negatively associated with mortalityfrom all specific causes of interest except stroke. Thedeprivation index was associated with mortality from allspecific causes of interest, showing an increase in mortalityfor the most deprived communes. The lung cancer mortalitywas not associated with any specific cause of interest.Increasing LdenAEI was associated with mortality fromcardiovascular disease [MRR per 10 dB(A) increasein LdenAEI = 1.18 (1.11-1.25)], coronary heart disease[MRR = 1.24 (1.12-1.36)], and myocardial infarction[MRR = 1.28 (1.11-1.46)]. LdenAEI was more weaklyassociated with stroke mortality [MRR = 1.08 (0.97-1.21)].Table 1: Distribution of aircraft noise levels (LdenAEI) and of background air pollution concentrations (NO2 and PM10) for the161 communes of the study areaArea of study Numbers ofcommunesParisLyonToulouseTotala1083122161LdenAEI (dB(A))Mean51.645.345.749.6Median51.343.444.948.9Range45.0- 64.142.0- 55.142.0- 55.842.0- 64.1Mean24.016.5—22.3NO2 (µg.m-3)MedianRange23.415.9- 36.316.312.0- 21.9——21.212.0- 36.3Mean24.223.9—23.9PM10 (µg.m-3)MedianRange23.422.4- 27.124.022.3- 26.2——23.622.3- 27.1Only for 139 communes for NO2 and PM10 concentrationsa331Noise & Health, September-October 2015, Volume 17Evrard, et al.: Aircraft noise and cardiovascular mortality near airportsFigure 2: Distribution of LdenAEI in the communes included in the present studyTable 2: Adjusted MRRs estimated in models without air pollutionParametersLdenAEIaGenderAge (years)Log (density)Deprivation indexLung cancer mortalityCardiovascular diseaseMRR (95% CI)1.18 (1.11-1.25)1.04 (1.01-1.07)1.11 (1.09-1.13)0.93 (0.90-0.96)1.07 (1.05-1.10)1.01 (0.98-1.03)Coronary heart diseaseMRR (95% CI)1.24 (1.12-1.36)1.00 (0.96-1.05)1.10 (1.07-1.13)0.94 (0.89-0.98)1.07 (1.04-1.11)1.00 (0.96-1.04)Myocardial infarctionMRR (95% CI)1.28 (1.11-1.46)1.02 (0.95-1.08)1.08 (1.04-1.13)0.87 (0.81-0.93)1.10 (1.05-1.16)0.99 (0.93-1.04)StrokeMRR (95% CI)1.08 (0.97-1.21)1.01 (0.96-1.06)1.15 (1.11-1.18)0.96 (0.91-1.02)1.08 (1.04-1.13)1.02 (0.98-1.07)LdenAEI, gender, age, log-density, a deprivation index, and lung cancer mortality were simultaneously included in the models. aMRR per 10 dB(A) increases in LdenAEISupplementary results with adjustment for air pollutionAircraft noise levels (LdenAEI) were moderately correlatedwith NO2 concentrations (r = 0.45) while they were notcorrelated with PM10 concentrations (r = 0.06). The correlationbetween LdenAEI and NO2 concentration was lower for theParis area (r = 0.26). NO2 and PM10 concentrations werepositively correlated (r = 0.64).When NO2 concentration was taken into account in themodels including LdenAEI, the results did not change[Table 3].for mortality from coronary heart disease, and 1.37 (1.111.68) and 1.21 (0.94-1.55) for mortality from myocardialinfarction among men and women, respectively [Table 3].Introducing the Townsend deprivation index in the modelsdid not change the results [Table 3]. The Townsenddeprivation index was highly correlated with the deprivationindex obtained with PCA (r = 0.85).Introducing PM10 concentration in the model instead of NO2concentration did not change the results.The additional adjustment for the study area in the modelsdid not alter the results [Table 3]. Moreover, the associationsbetween aircraft noise exposure and mortality from all causesof interest remained similar when only the Paris data wereused [Table 3].Sensitivity analysesDiscussionWhen stratified by gender, MRRs were higher in men than inwomen for mortality from cardiovascular disease, coronaryheart disease, and myocardial infarction. After adjustment forNO2 concentration, MRRs per 10 dB(A) increase in LdenAEIwere 1.29 (1.17-1.42) and 1.12 (1.03-1.23) for mortality fromcardiovascular disease, 1.29 (1.12-1.49) and 1.15 (0.97-1.37)The present study is the first ecological study investigatingthe relationship between exposure to aircraft noise and themortality of the population living in the vicinity of the airportsin France. This study covers 161 communes of France with apopulation of 1.9 million people living close to Paris-Charles deNoise & Health, September-October 2015, Volume 17332Evrard, et al.: Aircraft noise and cardiovascular mortality near airportsTable 3: Adjusted MRRs* related to LdenAEI obtained in sensitivity analysesSensitivity analysesIncluding NO2aIncluding PM10aBy genderMalebFemalebIncluding the Townsenddeprivation indexcIncluding adjustment for thestudy areadIncluding data for the Parisarea onlydCardiovascular diseaseMRR (95% CI)1.18 (1.10-1.26)1.18 (1.10-1.25)Coronary heart diseaseMRR (95% CI)1.23 (1.10-1.38)1.20 (1.09-1.34)Myocardial infarctionMRR (95% CI)1.31 (1.12-1.53)1.26 (1.09-1.46)StrokeMRR (95% CI)1.06 (0.93-1.21)1.08 (0.95-1.22)1.29 (1.17-1.42)1.12 (1.03-1.23)1.19 (1.11-1.27)1.29 (1.12-1.49)1.15 (0.97-1.37)1.23 (1.10-1.38)1.37 (1.11-1.68)1.21 (0.94-1.55)1.31 (1.12-1.54)1.10 (0.90-1.33)1.00 (0.85-1.19)1.06 (0.93-1.21)1.18 (1.10-1.26)1.26 (1.12-1.41)1.35 (1.15-1.59)1.05 (0.92-1.20)1.11 (1.03-1.20)1.23 (1.09-1.40)1.27 (1.06-1.52)1.02 (0.88-1.18)*MRRs per 10 dB((A)) increase in LdenAEI. aLdenAEI, gender, age, log-density, a deprivation index, and lung cancer mortality were also included in the models. bLdenAEI, age, logdensity, a deprivation index, lung cancer mortality, and average NO2 concentration were also included in the models. cLdenAEI, gender, age, log-density, lung cancer mortality, andaverage NO2 concentration were also included in the models. dLdenAEI, gender, age, log-density, a deprivation index, lung cancer mortality, and average NO2 concentration werealso included in the modelsGaulle, Lyon-Saint-Exupéry, and Toulouse-Blagnac airports.Positive associations were reported between weighted averageexposure to aircraft noise and mortality from cardiovasculardisease, coronary heart disease, and myocardial infarction.Controlling the socioeconomic status of the commune(measured by a deprivation index), demographic factors of thecommune (such as age and gender of the inhabitants), and lungcancer mortality used as a proxy for smoking did not changethe results. When the models were stratified on the basis ofgender, the associations between exposure to aircraft noise andmortality from cardiovascular disease, coronary heart disease,and myocardial infarction remained significant with higherrisks among men than women.As aircraft noise levels were much higher in the Paris areathan in the other areas due to the larger size of the airport,the effect of the additional adjustment for the study area wasexplored and a sensitivity analysis using the Paris data onlywas conducted; however, the results remained similar.The present study seems to confirm the findings of recentstudies, suggesting that high levels of aircraft noise areassociated with mortality from cardiovascular disease andcoronary heart disease[18] and with mortality from myocardialinfarction.[15] Moreover, we observed a weak associationbetween aircraft noise and stroke mortality. These results arein accordance with the results of Huss et al.[15] and Sorensenet al.[16]The present study has attempted to take into account theissue of confounding air pollution. Accounting for NO2 orPM10 concentration did not change the results — air pollutiondoes not seem to be a confounding factor in the relationshipbetween aircraft noise and mortality from all causes ofinterest. These results are consistent with previous studies…

ANSWER:

REQUEST HELP FROM A TUTOR