Several studies have shown that there are already statistically significant associations between environmental noise and adverse health effects, whether in relation to short-term (maximum of one day) or long-term (over one year) exposure to environmental noise, and, consequently, also on both morbidity and mortality.
These studies have shown that this association occurs independently from confounding factors, mainly air pollutants. However, misalignment; that is, the fact that the locations where the exposure and health outcomes are measured have different spatial locations, is a problem that has not yet been addressed, and inferences can be seriously compromised without an explicit control of this misalignment.
The aim of our study was to evaluate the long-term effects of traffic noise on mortality in the city of Barcelona throughout the 2004-2007 period. To this end, we controlled for potential confounding factors, including both atmospheric pollutants and socioeconomic factors at a contextual level. In addition, we explicitly controlled the problem of spatial misalignment. To this end, we used a case-control design with individual data. We analyzed deaths caused by myocardial infarction, hypertension, or type II diabetes mellitus in Barcelona between 2004 and 2007. The controls were deaths (in Barcelona and throughout the same study period) due to AIDS or other external causes (e.g., accidental falls, accidental poisoning by psychotropic drugs, overdose by drugs of abuse, suicide and self-harm, or injuries resulting from motor vehicle accidents). The controls were matched to the cases by sex and age.
We used the equivalent weighted average sound pressure levels of three different time periods: daytime (7-21 h), evening (21-23 h), and nighttime (23-7 h); and controlled the analysis for the following confounding factors: air pollutants (NO2, PM10, and benzene), deprivation index (at a census section level), land use, and other spatial variables.
We explicitly controlled the heterogeneity (unequal distribution of both the outcome variable and the environmental exposure within the census section), spatial dependence (of the outcome variable observations), temporal trends (long-term behavior of the outcome variables), and spatial misalignment (between the outcome and the location of the environmental exposure). We used a (complete) Bayesian model estimated applying the INLA approach.
After adjusting for confounding factors, we found that traffic noise was associated with mortality secondary to myocardial infarction and type II diabetes mellitus in men and mortality secondary to hypertension in women. However, the association was only significant for all age groups considered in the case of hypertension in women.
It is also worth noting that this was the first study in Spain, and one of the first worldwide, to introduce the SPDE model of the INLA approach, within the Bayesian approach, in spatiotemporal modeling. In addition, this study presented an original method for dealing with the issue of spatial misalignment.