The exposome constitutes a promising framework to improve understanding of the effects of environmental exposures on health by explicitly considering multiple testing and avoiding selective ...reporting. However, exposome studies are challenged by the simultaneous consideration of many correlated exposures.
We compared the performances of linear regression-based statistical methods in assessing exposome-health associations.
In a simulation study, we generated 237 exposure covariates with a realistic correlation structure and with a health outcome linearly related to 0 to 25 of these covariates. Statistical methods were compared primarily in terms of false discovery proportion (FDP) and sensitivity.
On average over all simulation settings, the elastic net and sparse partial least-squares regression showed a sensitivity of 76% and an FDP of 44%; Graphical Unit Evolutionary Stochastic Search (GUESS) and the deletion/substitution/addition (DSA) algorithm revealed a sensitivity of 81% and an FDP of 34%. The environment-wide association study (EWAS) underperformed these methods in terms of FDP (average FDP, 86%) despite a higher sensitivity. Performances decreased considerably when assuming an exposome exposure matrix with high levels of correlation between covariates.
Correlation between exposures is a challenge for exposome research, and the statistical methods investigated in this study were limited in their ability to efficiently differentiate true predictors from correlated covariates in a realistic exposome context. Although GUESS and DSA provided a marginally better balance between sensitivity and FDP, they did not outperform the other multivariate methods across all scenarios and properties examined, and computational complexity and flexibility should also be considered when choosing between these methods. Citation: Agier L, Portengen L, Chadeau-Hyam M, Basagaña X, Giorgis-Allemand L, Siroux V, Robinson O, Vlaanderen J, González JR, Nieuwenhuijsen MJ, Vineis P, Vrijheid M, Slama R, Vermeulen R. 2016. A systematic comparison of linear regression-based statistical methods to assess exposome-health associations. Environ Health Perspect 124:1848-1856; http://dx.doi.org/10.1289/EHP172.
•Annoyance was higher than predicted by the old EU standard curve.•Neither changing noise levels nor population characteristics explain this increase.•Non-acoustical factors are relevant in relation ...to aircraft noise annoyance.•Necessary to have a definition of HA derived substantially as recommended by ICBEN.
Since the 2000s, increased aircraft noise annoyance has been observed in the populations living near airports. The DEBATS-study compared the exposure–response relationship estimated among airports’ residents in France with old and new EU standard curves. It also examines whether non-acoustical factors may explain this annoyance. For 1244 adults living near three French airports, information about demographic and socio-economic factors as well as aircraft noise annoyance, situational, personal and attitudinal factors was collected with a face-to-face questionnaire. Outdoor aircraft noise exposure was estimated by linking home address to noise exposure maps. Logistic regression models were used to investigate the association between annoyance and a broad range of other variables in addition to the Lden. Severe noise annoyance was associated not only with increased aircraft noise levels, but also with non-acoustical factors. Annoyance was higher than predicted by the old EU standard curve when estimated with the model including non-acoustical factors in addition to the Lden. It was even higher when only noise exposure was considered. However, annoyance was lower in DEBATS than predicted by the new EU standard curve provided by WHO. The increase of noise annoyance does not seem to be explained by the factors already mentioned in the literature as possible explanations. However, it cannot be ruled out that methodological differences in the HA assessment may be the reason for changes in annoyance over the years. For this reason, we argue for a definition of HA derived substantially as recommended by ICBEN. The findings of the DEBATS study also confirm that taking into account non-acoustical factors such as situational, personal and attitudinal factors would improve annoyance predictions.
Display omitted
•Weather during critical windows in pregnancy may lead to lower term birthweight.•These critical windows appear to differ between male and female infants.•Temperature variability, not ...just mean temperature, may impact term birthweight.•Even apart from temperature, mean humidity may affect term birthweight.
Heat stress during pregnancy may limit fetal growth, with ramifications throughout the life course. However, critical exposure windows are unknown, and effects of meteorological variability have not been investigated.
We aimed to identify sensitive windows for the associations of mean and variability of temperature and humidity with term birthweight.
We analyzed data from two French mother–child cohorts, EDEN and PELAGIE (n = 4771), recruited in 2002–2006. Temperature exposure was assessed using a satellite-based model with daily 1-km2 resolution, and relative humidity exposure data were obtained from Météo France monitors. Distributed lag models were constructed using weekly means and standard deviation (SD, to quantify variability) from the first 37 gestational weeks. Analyses were then stratified by sex. Results for each exposure were adjusted for the other exposures, gestational age at birth, season and year of conception, cohort and recruitment center, and individual confounders.
There was no evidence of association between term birthweight and mean temperature. We identified a critical window in weeks 6–20 for temperature variability (cumulative change in term birthweight of −54.2 g 95% CI: −102, −6 for a 1 °C increase in SD of temperature for each week in that window). Upon stratification by sex of the infant, the relationship remained for boys (weeks 1–21, cumulative change: −125 g 95% CI: −228, −21). For mean humidity, there was a critical window in weeks 26–37, with a cumulative change of −28 g (95% CI: −49, −7) associated with a 5% increase in humidity for each week. The critical window was longer and had a stronger association in boys (weeks 29–37; −37 g, 95% CI: −63, −11) than girls (week 14; −1.8 g, 95% CI: −3.6, −0.1).
Weekly temperature variability and mean humidity during critical exposure windows were associated with decreased term birthweight, especially in boys.
Studies of air pollution effects during pregnancy generally only consider exposure in the outdoor air at the home address. We aimed to compare exposure models differing in their ability to account ...for the spatial resolution of pollutants, space–time activity and indoor air pollution levels. We recruited 40 pregnant women in the Grenoble urban area, France, who carried a Global Positioning System (GPS) during up to 3weeks; in a subgroup, indoor measurements of fine particles (PM2.5) were conducted at home (n=9) and personal exposure to nitrogen dioxide (NO2) was assessed using passive air samplers (n=10). Outdoor concentrations of NO2, and PM2.5 were estimated from a dispersion model with a fine spatial resolution. Women spent on average 16h per day at home. Considering only outdoor levels, for estimates at the home address, the correlation between the estimate using the nearest background air monitoring station and the estimate from the dispersion model was high (r=0.93) for PM2.5 and moderate (r=0.67) for NO2. The model incorporating clean GPS data was less correlated with the estimate relying on raw GPS data (r=0.77) than the model ignoring space–time activity (r=0.93). PM2.5 outdoor levels were not to moderately correlated with estimates from the model incorporating indoor measurements and space–time activity (r=−0.10 to 0.47), while NO2 personal levels were not correlated with outdoor levels (r=−0.42 to 0.03). In this urban area, accounting for space–time activity little influenced exposure estimates; in a subgroup of subjects (n=9), incorporating indoor pollution levels seemed to strongly modify them.
•We developed 8 exposure models to assess air pollutants exposure in pregnant women.•We assessed space–time activity by GPS data and proposed an algorithm to clean them.•In this urban area integrating space–time activity little modified exposure estimates.•PM2.5 outdoor levels had low correlations with levels integrating indoor levels (n=10).•NO2 personal levels were not correlated with outdoor levels at the home address (n=9).
BACKGROUND:For chemicals with high within-subject temporal variability, assessing exposure biomarkers in a spot biospecimen poorly estimates average levels over long periods. The objective is to ...characterize the ability of within-subject pooling of biospecimens to reduce bias due to exposure misclassification when within-subject variability in biomarker concentrations is high.
METHODS:We considered chemicals with intraclass correlation coefficients of 0.6 and 0.2. In a simulation study, we hypothesized that the chemical urinary concentrations averaged over a given time period were associated with a health outcome and estimated the bias of studies assessing exposure that collected 1 to 50 random biospecimens per subject. We assumed a classical type error. We studied associations using a within-subject pooling approach and two measurement error models (simulation extrapolation and regression calibration), the latter requiring the assay of more than one biospecimen per subject.
RESULTS:For both continuous and binary outcomes, using one sample led to attenuation bias of 40% and 80% for compounds with intraclass correlation coefficients of 0.6 and 0.2, respectively. For a compound with an intraclass correlation coefficient of 0.6, the numbers of biospecimens required to limit bias to less than 10% were 6, 2, and 2 biospecimens with the pooling, simulation extrapolation, and regression calibration methods (these values were, respectively, 35, 8, and 2 for a compound with an intraclass correlation coefficient of 0.2). Compared with pooling, these methods did not improve power.
CONCLUSION:Within-subject pooling limits attenuation bias without increasing assay costs. Simulation extrapolation and regression calibration further limit bias, compared with the pooling approach, but increase assay costs.
There is growing interest in examining the simultaneous effects of multiple exposures and, more generally, the effects of mixtures of exposures, as part of the exposome concept (being defined as the ...totality of human environmental exposures from conception onwards). Uncovering such combined effects is challenging owing to the large number of exposures, several of them being highly correlated. We performed a simulation study in an exposome context to compare the performance of several statistical methods that have been proposed to detect statistical interactions.
Simulations were based on an exposome including 237 exposures with a realistic correlation structure. We considered several statistical regression-based methods, including two-step Environment-Wide Association Study (EWAS
), the Deletion/Substitution/Addition (DSA) algorithm, the Least Absolute Shrinkage and Selection Operator (LASSO), Group-Lasso INTERaction-NET (GLINTERNET), a three-step method based on regression trees and finally Boosted Regression Trees (BRT). We assessed the performance of each method in terms of model size, predictive ability, sensitivity and false discovery rate.
GLINTERNET and DSA had better overall performance than the other methods, with GLINTERNET having better properties in terms of selecting the true predictors (sensitivity) and of predictive ability, while DSA had a lower number of false positives. In terms of ability to capture interaction terms, GLINTERNET and DSA had again the best performances, with the same trade-off between sensitivity and false discovery proportion. When GLINTERNET and DSA failed to select an exposure truly associated with the outcome, they tended to select a highly correlated one. When interactions were not present in the data, using variable selection methods that allowed for interactions had only slight costs in performance compared to methods that only searched for main effects.
GLINTERNET and DSA provided better performance in detecting two-way interactions, compared to other existing methods.
Background: Transportation noise seems to impair self-reported health status (SRHS). However, only a few studies have considered the role of noise annoyance and noise sensitivity in this deleterious ...effect. This study aims investigating mediator and moderator roles of noise annoyance and noise sensitivity. Methods: In 2013, the DEBATS longitudinal study included 1244 participants aged over 18 years and living around three French airports. These participants were followed up in 2015 and 2017. They self-reported their perceived health status, aircraft noise annoyance, and their noise sensitivity via a questionnaire during the three visits. Noise maps were used to estimate aircraft noise levels at the facade of participants' residence. Generalized linear mixed models with a random intercept at the participant level were used. Results: Aircraft noise levels were associated with severe annoyance. Severe annoyance tent to be associated with impaired SRHS. Aircraft noise levels were associated with impaired SRHS only in men (odds ratio OR = 1.47, 95% confidence interval CI = 1.02, 2.11, for a 10-dBA Lden increase in aircraft noise levels) with a weaker association adjusted for annoyance (OR = 1.36, 95% CI = 0.94, 1.98). The association was stronger in men who reported high noise sensitivity (OR = 1.84, 95% CI = 0.92, 3.70, versus OR = 1.39, 95% CI = 0.90, 2.14, for men who were not highly sensitive to noise). Conclusion: From our results, the deleterious effect of aircraft noise on SRHS could be mediated by noise annoyance and moderated by noise sensitivity. Further studies using causal inference methods are needed for identifying causal effect of exposure, mediator, and moderator.
Epidemiologic studies suggest an association between air pollution exposure and foetal growth. The possible underlying biological mechanisms have little been studied in humans, but animal studies ...suggest an impact of atmospheric pollutants on placental function.
Our aim was to investigate the association between exposure to atmospheric pollutants' levels during pregnancy and placental weight, birth weight and the placental to foetal weights ratio (PFR). For comparison purposes, the effects of active smoking on the same measures at birth have also been estimated.
The study relies on women from Eden mother–child cohort recruited in the middle-sized cities of Poitiers and Nancy (France). Nitrogen dioxide (NO2) and particulate matter with diameter <10μm (PM10) home address levels during pregnancy were assessed using ADMS-Urban dispersion model. We characterized associations of NO2, PM10 levels and active smoking with placental, birth weights and PFR by distinct linear regression models.
Air pollution levels were higher and had greater variability in Nancy (5th–95th centiles, 19.9–27.9μg/m3 for PM10) than in Poitiers (5th–95th centiles, 14.3–17.8μg/m3). Associations differed by study area: in Nancy (355 births), air pollution levels were associated with decreased placental weight and PFR, while in Poitiers (446 births), opposite or null associations were observed. Cigarette smoking was not associated with placental weight while it was associated with a decrease in birth weight and an increase in PFR.
Results regarding air pollution estimated effects were not similar in both study areas and should therefore be taken with caution. The placental weight decrease observed with air pollutants in the more polluted area of Nancy is consistent with a recent epidemiological study. In this area, maternal active smoking and PM10 levels tended to have opposite effects on the PFR, suggesting different mechanisms of action of both pollutants on foetal growth.
► We studied the impact of atmospheric pollutants on placental weight in two urban areas. ► Particulate matter (PM10) was associated with a decrease in placental weight in the more polluted area. ► Maternal smoking was associated with a birth weight decrease, and an increase in placental to foetal weights ratio. ► Maternal smoking did not seem to influence placental weight.
Air pollution exposure represents a major health threat to the developing foetus. DNA methylation is one of the most well-known molecular determinants of the epigenetic status of cells. Blood DNA ...methylation has been proven sensitive to air pollutants, but the molecular impact of air pollution on new-borns has so far received little attention.
We investigated whether nitrogen dioxide (NO2), particulate matter (PM10), temperature and humidity during pregnancy are associated with differences in placental DNA methylation levels.
Whole-genome DNA-methylation was measured using the Illumina's Infinium HumanMethylation450 BeadChip in the placenta of 668 newborns from the EDEN cohort. We designed an original strategy using a priori biological information to focus on candidate genes with a specific expression pattern in placenta (active or silent) combined with an agnostic epigenome-wide association study (EWAS). We used robust linear regression to identify CpGs and differentially methylated regions (DMR) associated with each exposure during short- and long-term time-windows.
The candidate genes approach identified nine CpGs mapping to 9 genes associated with prenatal NO2 and PM10 exposure false discovery rate (FDR) p < 0.05. Among these, the methylation level of 2 CpGs located in ADORA2B remained significantly associated with NO2 exposure during the 2nd trimester and whole pregnancy in the EWAS (FDR p < 0.05). EWAS further revealed associations between the environmental exposures under study and variations of DNA methylation of 4 other CpGs. We further identified 27 DMRs significantly (FDR p < 0.05) associated with air pollutants exposure and 13 DMRs with meteorological conditions.
The methylation of ADORA2B, a gene whose expression was previously associated with hypoxia and pre-eclampsia, was consistently found here sensitive to atmospheric pollutants. In addition, air pollutants were associated to DMRs pointing towards genes previously implicated in preeclampsia, hypertensive and metabolic disorders. These findings demonstrate that air pollutants exposure at levels commonly experienced in the European population are associated with placental gene methylation and provide some mechanistic insight into some of the reported effects of air pollutants on preeclampsia.
•DNA methylation was measured in 668 placentas using the HumanMethylation450 BeadChip.•We used an original strategy combining a concept-driven and agnostic approaches.•Methylation of Alu or LINE-1 sequences was mainly not associated with air pollutants.•Pregnancy NO2 exposure was associated with lower ADORA2B methylation in multiple CpGs.•Highlighted genes were related to preeclampsia, hypertensive and metabolic disorders.
Although several cross-sectional studies have shown that aircraft noise exposure was associated with an increased risk of hypertension, a limited number of longitudinal studies have addressed this ...issue. This study is part of the DEBATS (Discussion on the health effect of aircraft noise) research programme and aimed to investigate the association between aircraft noise exposure and the incidence of hypertension.
In 2013, 1244 adults living near three major French airports were included in this longitudinal study. Systolic and diastolic blood pressure, as well as demographic and lifestyle factors, were collected at baseline and after 2 and 4 years of follow-up during face-to-face interviews. Exposure to aircraft noise was estimated for each participant's home address using noise maps. Statistical analyses were performed using mixed Poisson and linear regression models adjusted for potential confounding factors.
A 10 dB(A) increase in aircraft noise levels in terms of L
was associated with a higher incidence of hypertension (incidence rate ratio (IRR)=1.36, 95% CI 1.02 to 1.82). The association was also significant for L
(IRR 1.41, 95% CI 1.07; to 1.85) and L
(IRR 1.31, 95% CI 1.01 to 1.71). Systolic and diastolic blood pressure increased with all noise indicators.
These results strengthen those obtained from the cross-sectional analysis of the data collected at the time of inclusion in DEBATS, as well as those from previous studies conducted in other countries. Hence, they support the hypothesis that aircraft noise exposure may be considered as a risk factor for hypertension.