Skip to Main content Skip to Navigation
Master Thesis

Modélisation de l’histoire d’une exposition sur le changement d’un marqueur au cours du temps

Abstract : In life-time epidemiology, several hypotheses of biological mechanisms of impact on health have been identified such as accumulation of exposure over time or increased exposure over a short exposure window. This complex time-dependent association issue is particularly present in many chronic diseases that are age related. We are mainly interested in the influence of the exposure process on the subsequent change of a disease marker. The principle is to assess the time-dependent association of exposure on marker change in order to find the mechanism that explains how the marker changes over time. In this work, we developed the statistical methodology to study the cumulative effect of intermittently measured noise exposure on the instantaneous slope of a disease marker using a joint modelling approach. Then the implementation of the method was done under R. The application was carried out on the data of the three cities (3C) by focusing on the centre of Bordeaux, the exposure considered is a modifiable factor, the Body Mass Index (BMI). The marker of cognitive evolution is the Isaacs Set Test of verbal fluency (STI). We considered an exposure window of ten years. The adjustment variables were age at start, presence of the APOE4 gene, gender and educational level. Despite convergence difficulties, some relevant and conclusive models could be analysed. First, a high BMI at the distance of the cognitive assessment has a deleterious effect on the instantaneous slope of the STI, which is consistent with adiposity being a long-term risk factor. Then a high BMI near the cognitive assessment has a protective effect on the decline of STI, probably due to reverse causality. These findings are consistent with observations from previous work. We also observed that for the same initial BMI level, a different BMI evolution significantly impacted the decline of STI, whereas different initial BMI levels with the same evolution did not seem to be associated with significantly different STI evolutions.
Document type :
Master Thesis
Complete list of metadata
Contributor : ISPED UB - Connect in order to contact the contributor
Submitted on : Wednesday, December 8, 2021 - 1:48:50 PM
Last modification on : Friday, December 17, 2021 - 3:47:07 AM
Long-term archiving on: : Wednesday, March 9, 2022 - 6:38:53 PM


Files produced by the author(s)


  • HAL Id : dumas-03470672, version 1



Ariane Bercu. Modélisation de l’histoire d’une exposition sur le changement d’un marqueur au cours du temps. Santé publique et épidémiologie. 2021. ⟨dumas-03470672⟩



Record views


Files downloads