Association entre l'exposition à la pollution atmosphérique et la santé : utilisation des séries chronologiques

摘要: During the past twenty years, short-term relationships between air pollution and health have been largely investigated, mainly through time series studies. Their aim is to estimate association daily levels of an indicator numbers a event (death, hospital admission, etc.). In order get non-biased estimation this short term association, all factors which can modify relationship be taken into account (these are related events). The current approach consists using Poisson regression based on Generalized Additive Model (GAM). This model fits nonparametric functions allow for nonlinear effects provide better different variables. following included in : long-term trend seasonality, day week, holidays, vacations, temperature, influenza epidemics, pollen counts, pollution. A quasi-Poisson distribution allows taking frequently observed over dispersion data series. parameters smoothing function (penalized spline) used seasonality chosen minimize partial autocorrelation residuals. parameter pollutant multivariate relative risk. Repeating such study successive periods very useful real epidemiologic surveillance risks pollution, routinely registered data.