[Forum SIS] Annuncio seminario - 29 settembre - Laura Pazzagli

[Elena Stanghellini]

Cari colleghi,

Vi comunico che venerdì 29 prossimo, la dr. Laura Pazzagli del Karolinska University Hospital terrà un seminario sui Marginal Structural Models in Epidemiologia. I dettagli qui sotto. Cari saluti, Elena S.

-----------------------------------------
Speaker: Laura Pazzagli, Ph.D.Centre for pharmacoepidemiology, Department of Medicine, Karolinska University Hospital, Stockholm (Sweden)

Luogo e data: Venerdì 29 settembre 2017, ore 14,00 Aula 23, Via A. Pascoli Perugia

Titolo: A longitudinal study to evaluate the joint effects of socio-economic exposures on end-stage renal disease in patients with Type I diabetes: the use of marginal structural models

Abstract: Lack of proper control for time-varying exposures can lead to substantial bias in estimates of treatment effect. Marginal structural models (MSMs) are mainly observational based methods and an important statistical tool proposed by Robins to adjust for time-varying confounders. These models work under an extension for longitudinal data, of the potential outcomes framework proposed by Rubin. One of the most common and intuitive ways to estimate the parameters of a MSM is using the inverse probability of treatment weighted estimator (IPTW). When the underlying assumptions hold the IPTW creates a pseudo-population in which the assignment of the treatment is like randomized, since controlling for the time-varying confounders ensures that they no longer affect the subsequent treatment levels.
A MSM is particularly useful in the case the time-varying confounder is also an intermediate between the treatment and the outcome, since it produces unbiased estimates of the effect of interest not conditioning on the intermediate variable. The main appeal of MSMs is the fact that they are a natural extension of the classical logistic and Cox models, i.e. a marginal structural logistic model is just a weighted logistic model, and similarly a marginal structural Cox model can be implemented. 
In this study we investigate the effect of the socioeconomic exposures unemployment and receiving income support on the development of end-stage renal disease in type 1 diabetes patients, using a marginal structural model in comparison with standard logistic regression.

*********************************************
Elena Stanghellini, Ph.D.
Full Professor of Statistics
University of Perugia
Rector's Delegate for International Relations

Department of Economics
Via Pascoli, 1 - 06100 Perugia - Italy
tel. +39 075 585 5228 fax. +39 075 585 5950
home page: www.stat.unipg.it/stanghellini
*********************************************