Guillaume Chauvet

Dr. Guillaume Chauvet


Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 February 2022 in BMC Medical Research Methodology
Verified
Journal Article
Open access
PDF file
MMMarie-Astrid Metten
Marie-Astrid Metten
NCNathalie Costet
Nathalie Costet
LMLuc Multigner
Luc Multigner
JVJean-François Viel
Jean-François Viel
GCGuillaume Chauvet
Guillaume Chauvet
Background: Attrition in cohort studies challenges causal inference. Although inverse probability weighting (IPW) has been proposed to handle attrition in association analyses, its relevance has been little studied in this context. We aimed to investigate its ability to correct for selection bias in exposure-outcome estimation by addressing an important methodological issue: the specification of the response model.Methods: A simulation study compared the IPW method with complete-case analysis (CCA) for nine response-mechanism scenarios (3 missing at random – MAR and 6 missing not at random - MNAR). Eighteen response models differing by the type of variables included were assessed.Results: The IPW method was equivalent to CCA in terms of bias and consistently less efficient in all scenarios, regardless of the response model tested. The most effective response model included only the confounding factors of the association model.Conclusion: Our study questions the ability of the IPW method to correct for selection bias in situations of attrition leading to missing outcomes. If the method is to be used, we encourage including only the confounding variables of the association of interest in the response model.

6 May 2020 in Journal of the Royal Statistical Society Series B: Statistical Methodology
Verified
Journal Article
Closed access
PDF file
GCGuillaume Chauvet
Guillaume Chauvet
AVAudrey-Anne Vallée
Audrey-Anne Vallée

LDLaure Daniel
Laure Daniel
MMMarylou Michot
Marylou Michot
MEMaxime Esvan
Maxime Esvan
PGPauline Guérin
Pauline Guérin
GCGuillaume Chauvet
Guillaume Chauvet
FPFabienne PELE
Fabienne PELE
Indoor pollutants can have short- and long-term health effects, especially if exposure occurs during prenatal life or early childhood. This study describe the perceptions, knowledge, and practices of adults concerning indoor environmental pollution. Adults of 18 to 45 years of age were recruited in the department of Ille-et-Vilaine (Brittany-France) in 2019 through a stratified random draw in the waiting rooms of general practitioners (GPs) (n = 554) who completed a self-questionnaire. The 71% who had already heard of this type of pollution were older (p = 0.001), predominantly women (p = 0.007), not expecting a baby (p = 0.005), and had a higher knowledge score (p < 0.001). The average knowledge score was 6.6 ± 6.6 out of 11, which was higher for participants living in a couple and with a higher level of education (p < 0.001). Some practices were well implemented (>80% of participants) (aeration during renovation) whereas others were insufficiently practiced (<60% of participants) (paying attention to the composition of cosmetic products). Factors associated differed depending on the frequency of integration: living in a couple and having a child for well implemented practices and educational level, knowledge level, and perception for those under implemented. Knowledge must be improved to modify perceptions and certain practices, making sure not to increase social inequalities in health.

20 November 2023 in Statistics in Medicine
Verified
Journal Article
Open access
TVThanh Huan Vo
Thanh Huan Vo
VGValérie Garès
Valérie Garès
LZLi‐Chun Zhang
Li‐Chun Zhang
AHAndré Happe
André Happe
EOEmmanuel Oger
Emmanuel Oger
GCGuillaume Chauvet
Guillaume Chauvet
Record linkage is increasingly used, especially in medical studies, to combine data from different databases that refer to the same entities. The linked data can bring analysts novel and valuable knowledge that is impossible to obtain from a single database. However, linkage errors are usually unavoidable, regardless of record linkage methods, and ignoring these errors may lead to biased estimates. While different methods have been developed to deal with the linkage errors in the generalized linear model, there is not much interest on Cox regression model, although this is one of the most important statistical models in clinical and epidemiological research. In this work, we propose an adjusted estimating equation for secondary Cox regression analysis, where linked data have been prepared by a third‐party operator, and no information on matching variables is available to the analyst. Through a Monte Carlo simulation study, the proposed method is shown to lead to substantial bias reductions in the estimation of the parameters of the Cox model caused by false links. An asymptotically unbiased variance estimator for the adjusted estimators of Cox regression coefficients is also proposed. Finally, the proposed method is applied to a linked database from the Brest stroke registry in France.

20 May 2026 in Environmental and Ecological Statistics
Verified
Journal Article
Open access
PDF file
TDTrinh H. K. Duong
Trinh H. K. Duong
GCGuillaume Chauvet
Guillaume Chauvet
Olivier BouriaudOlivier Bouriaud
Olivier Bouriaud
National Forest Inventories (NFIs) are large-scale surveys that typically employ low sampling intensity, sufficient for national-level estimations. However, this low sampling intensity can make it difficult to produce reliable estimates for specific domains of interest due to limited sample sizes. NFIs use models (model-assisted or model-based approaches) for small area estimation to make estimations in the domain of interest with minimal or no sample. However, the reduced sample size can also be challenging for fitting models. Increasing the sampling intensity would resolve these issues. In this paper, we propose solutions to complement an existing NFI sample in order to improve estimation. We compare several sampling designs for intensification. This intensification poses the issue of integrating two dependent and non-overlapping samples with varying sampling intensities: the regular NFI sample and the intensified sample. We provide estimators of totals and ratios, and associated variance estimators for the domain of interest and the entire territory using a conditional approach. Our results show that intensification reduces the variance for an estimation at the level of both the domain of interest and the whole territory, and that the sampling design has a limited impact on the outcome of estimation, at least for the designs (simple random sampling, systematic sampling, Poisson sampling with equal probabilities, spatial systematic sampling) considered in the simulation study.

7 February 2025 in HAL
Verified
Preprint
Open access
PDF file
TDTrinh Duong
Trinh Duong
GCGuillaume Chauvet
Guillaume Chauvet
Olivier BouriaudOlivier Bouriaud
Olivier Bouriaud
National Forest Inventories (NFIs) are large-scale surveys that typically employ low sampling intensity, sufficient for national-level estimations. However, this low sampling intensity can make it difficult to produce reliable estimates for specific domains of interest under a designbased approach due to limited sample sizes. NFIs use models (model-assisted or model-based approaches) for small area estimation to make estimations in the domain of interest with minimal or no sample. However the reduced sample size can also be challenging for fitting models. Increasing the sampling intensity would represent resolve these issues. In this paper, we propose solutions to complement an existing NFI sample in order to improve estimation. We compare several sampling designs of intensification. This intensification poses the issue of integrating two dependent and non-overlapping samples with varying sampling intensities: the regular NFI sample and the intensified sample. We provide estimators of totals and ratios, and associated variance estimators for the domain of interest and the entire territory using a conditional approach. Our results show that intensification reduces the variance for an estimation at the level of both the domain of interest and the whole territory, that the choice of sampling designs considered has a limited impact on the estimation of the outcome.