Romain’s main research interests are three-fold:

1) Methodological research in Causal Inference

His work is heavily influenced by current subject-matter problems, practical considerations arising from analyses of real-life data, and the application of Causal Inference methodologies based on Marginal Structural Models (MSMs).

2) Software development for Causal Inference

Romain works on the development of an R package for MSM-based Causal Inference with longitudinal data. This package will implement three estimators of MSM parameters: the Inverse Probability of Treatment Weighted (IPTW), the G-computation and the Double Robust estimators. This package will rely on a recently developed data-adaptive estimator selection methodology in order to select models for nuisance parameters and/or the causal model.

3) Application of Causal Inference methodologies to the analyses of epidemiological data

Application of MSM-based Causal Inference methodologies to two large environmental studies conducted by Professor Ira Tager and funded by the California Air Resource Board:

• The Fresno Asthmatic Children’s Environment Study (FACES)

  A cohort of 250 clinically diagnosed asthmatic children (6-10 years of age) living in the Fresno/Clovis area has been enrolled and followed up since the end of 2000. The study design consists of a longitudinal and panel component. The broad goal is to determine the effects of exposure to particulate matter in combination with other pollutants on asthmatic children. Specifically, the study examines childhood asthma, how air pollution exposures affect the progression and severity of the disease, and the overall respiratory health of the children living in the Fresno-Clovis area. The study addresses the concern that repeated short-term responses to air pollution may translate into long-term health effects, including the worsening of asthma over time.

• A Pilot Study to Quantify Health Benefits of Incremental Improvements in Air Quality (Health Benefits Study):

  Over the past two decades, a large body of epidemiological data has been collected that indicate that ambient air pollution has considerable impact on human health, even at levels below those of prevailing air quality standards. Considerable effort has been expended to reduce ambient air pollution, with the expectation that such reductions would have an overall beneficial effect on human health. This hypothesis has never been tested directly. The overall goal of this study is precisely to identify the extent to which improvements in air quality in the South Coast Air Basin over the period 1980-2000 have or have not lead to improvements in health as measured by decreased mortality from specific cardiac, cardiovascular and respiratory mortality and decreases in morbidity from the same categories of diseases as measured by hospital discharges. This study has assembled and integrated the required exposure, demographic and health outcome data bases for the years 1980-2000. MSM-based Causal inference methodologies are applied to investigate causal relationships between decreases in the levels of individual and specific groups of pollutants and decreases in the rates of mortality and morbidity for diseases and groups of diseases which have been repeatedly associated with ambient levels of air pollution. Finally the objective of this study is also to estimate the economic value of any health benefits found to be associated with improved air quality over the period 1980-2000.